MARKET ANALYSIS

FOR REAL ESTATE

Market Analysis for Real Estate is a comprehensive introduction to how real estate markets work and the
analytical tools and techniques that can be used to identify and interpret market signals. The markets for
space and varied property assets, including residential, office, retail, and industrial, are presented, analyzed,
and integrated into a complete understanding of the role of real estate markets within the workings of con-
temporary urban economies. Unlike other books on market analysis, the economic and financial theory
in this book is rigorous and well integrated with the specifics of the real estate market. Furthermore, it is
thoroughly explained as it assumes no previous coursework in economics or finance on the part of the
reader.The theoretical discussion is backed up with numerous real estate case study examples and problems,
which are presented throughout the text to assist both student and teacher.
Including discussion questions, exercises, several web links, and online slides, this textbook is suitable
for use on a variety of degree programs in real estate, finance, business, planning, and economics at under-
graduate and MSc/​MBA level. It is also a useful primer for professionals in these disciplines.

Dr.  Rena Mourouzi-​Sivitanidou was an award-​winning lecturer and researcher, teaching courses in
Real Estate Market Analysis, Real Estate Economics, and Real Estate Development. This book is a com-
pilation of her lecture notes and handouts for the very successful course “Market Analysis for Real Estate,”
which she taught for nine years in the Master in Real Estate Development program at the University of
Southern California.The effort of compiling her lectures into a book was initiated after she passed away in
2000, following the recommendation of her colleagues who recognized the value of the excellent teaching
material that she had created.The full responsibility for any omissions or mistakes that may be found in the
book lies exclusively with the editor.
 A note on the author
Dr. Rena Mourouzi-​Sivitanidou was born in Kyrenia, a very beautiful small town on the north coast of Cyprus,
and was raised by two very loving, and very well educated, parents who taught her the highest values and ideals
through the example of their life. Rena followed a bright and shining path since she was a little child. It was a
path that left her teachers impressed by her performance and captivated by her personality.Throughout elem-
entary and high school, she was consistently the best student in her class, winning awards and distinctions year
after year, graduating with perfect scores and a scholarship award for undergraduate studies in Athens, Greece.
After she finished her undergraduate studies at the National Technical University of Athens, Rena entered
the Planning program at the Georgia Institute of Technology in Atlanta, where she received her Master in City
Planning. She subsequently continued her graduate studies at the Department of Urban Studies and Planning
at the Massachusetts Institute of Technology where she received her Ph.D. During her doctoral studies she
focused on urban and regional economics, with an emphasis on urban, land, and real estate economics.Very
shortly after she graduated the Ph.D. program in early 1991, she was hired as an Assistant Professor at the Sol
Price School of Public Policy at the University of Southern California (USC), where she taught courses in
Real Estate Market Analysis, Real Estate Economics, and Real Estate Development for nine years.
Rena was an extremely hard-​working academician and quickly established herself as a dedicated and
distinguished teacher/​researcher and a leading scholar at the Lusk Center for Real Estate. As many of her
students wrote after she passed away, she was a wonderful teacher, one of the best there ever was; an out-
standing educator who went out of her way to help each and every one of her students, devoting herself to
them with love and patience; a compassionate, very knowledgeable, intelligent, energetic, motivated, skilled,
and admirable professor; an ideal teacher blending academic excellence with personal integrity.
Her unparalleled devotion to her work and her students won her eight consecutive awards for the best
professor of the year. As a researcher and scholar, she developed an equally impressive record filled with
many high-​quality publications in the top journals in her field, a chapter in a book and many presentations
at national and regional conferences. Her work focused on real estate and land pricing, capitalization rates,
real estate market behavior, and firm location choices.
This book is a compilation of the lecture notes and handouts that Rena used during the nine years
that she taught the very successful course “Market Analysis for Real Estate” in the Master in Real Estate
Development program at the University of Southern California. This effort was initiated after she passed
away in 2000, following the recommendation of her colleagues who recognized the value of the excellent
teaching material that she had created. The full responsibility for any omissions or mistakes that may be
found in the book lies exclusively with the editor.
 A note on the editor
Dr. Petros Sivitanides is Associate Professor of Real Estate and Director of the Real Estate Department
at Neapolis University Paphos, Cyprus. He has a Ph.D. from the Massachusetts Institute of Technology in
Real Estate Economics and a Master in City Planning from the Georgia Institute of Technology in the
United States. He has many years of experience in some of the leading real estate advisory companies
in the United States and the United Kingdom, such as CBRE Global Investors in Los Angeles, CBRE
Econometric Advisors in Boston, and AXA Real Estate in London. He has also published many articles
in popular international journals in Real Estate and won the Emerald Literati Award for the outstanding
paper that was published in the Journal of Property Investment and Finance in 2018.
MARKET ANALYSIS
FOR REAL ESTATE

Rena Mourouzi-​Sivitanidou

Edited by Petros Sivitanides

 

First published 2021

by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
52 Vanderbilt Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2021 Petros Sivitanides
The right of Petros Sivitanides to be identified as author of this work has been asserted by
him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or utilised
in any form or by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying and recording, or in any information
storage or retrieval system, without permission in writing from the publishers.
Trademark notice: Product or corporate names may be trademarks or registered trademarks,
and are used only for identification and explanation without intent to infringe.
British Library Cataloguing-​in-​Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-​in-​Publication Data
Names: Mourouzi-Sivitanidou, Rena, author. |
Sivitanides, Petros, editor. | Torto, Raymond G., editor.
Title: Market analysis for real estate / Rena Mourouzi-Sivitanidou;
edited by Petros Sivitanides, Raymond Torto.
Description: First Edition. | New York: Routledge, 2020. |
Includes bibliographical references and index.
Identifiers: LCCN 2019059036 (print) | LCCN 2019059037 (ebook) |
ISBN 9780367233471 (hardback) | ISBN 9780367233501 (paperback) |
ISBN 9780429279409 (ebook)
Subjects: LCSH: Real property–Valuation. | Real estate business.
Classification: LCC HD1387 .M68 2020 (print) |
LCC HD1387 (ebook) | DDC 333.33/2–dc23
LC record available at https://lccn.loc.gov/2019059036
LC ebook record available at https://lccn.loc.gov/2019059037
ISBN: 978-​0-​367-​23347-​1  (hbk)
ISBN: 978-​0-​367-​23350-​1  (pbk)
ISBN: 978-​0-​429-​27940-​9  (ebk)
Typeset in Bembo
by Newgen Publishing UK
 In memory of my very dear and beloved Rena,
an unconditionally loving and caring wife,
a most dedicated and extremely hard-​working teacher,
academician, and researcher
who left the imprints of her wonderful personality
in the hearts of so many students that attended her classes
and everyone who knew her.
May her soul rest in eternal joy, peace, and happiness.
Petros Sivitanides
 CONTENTS

Acknowledgments xxii

PART A
Introduction 1

1 Market analysis in perspective 3

Introduction 3
Book premise, focus, and objectives 4
Market analysis in perspective 5
Focus and contribution 5
General macroeconomic/​metropolitan growth forces 6
Macroeconomic forces that drive the demand for and supply of
specific property types 7
Location and project-​specific factors 7
Market analysis components 8
Metropolitan growth analysis 9
Macro analysis of real estate markets 11
Micro (site-specific) analysis of development projects 12
Market studies and their emphasis 12
Residential 13
Retail 13
Office 13
Industrial 13
Market studies in general: evaluation criteria 14
Book structure and content 15
References and additional readings 17

ix
 Contents

2 Real estate economics 19

Introduction 20
Real estate demand 20
Real estate demand concepts 20
Demand sensitivity to price/​rent changes: price elasticity of demand 20
Impact of actual price changes versus expected price changes 22
Exogenous determinants of real estate demand 22
Measuring changes in real estate demand: absorption concepts 24
The supply of real estate 26
Real estate supply concepts 26
The long-​run aggregate supply: is it relevant? 26
The short-​run aggregate supply 27
New construction 27
The determination of market expectations 33
Real estate price adjustments 34
Price determination mechanism 34
Long-​run versus short-​run changes 35
The simple stock-​flow model: an analytical tool 36
Assessment of the simple stock-​flow model 39
Assessing demand-​supply imbalances 40
Demand-​supply interactions: market inefficiencies 40
Assessing the extent of disequilibrium: popular/​simplistic measures 40
Construction minus net absorption (C–​AB) 41
Is C–​AB a good indicator of market strength? 41
Nominal vacancy rate (V) 43
Advanced methodologies 44
Nominal versus structural vacancy rate (V–​V*) 44
Prevailing rent versus implicit equilibrium rent (R–​R*) 48
Chapter summary 49
Questions 50
Demand 50
Supply 50
Price adjustments 50
Questions for discussion 50
References and additional readings 52

PART B
Metropolitan growth analysis 53

3 Metropolitan growth patterns 55

Introduction 55
Growth analysis in perspective 56
Defining metropolitan growth 56
Growth mechanisms and implications for real estate 58
Demand-​induced growth: causes and long-​run consequences 59

x
 Contents

Growth stimuli 59
Mechanism and long-​run consequences 60
Real estate market impacts 62
Supply-​induced growth: causes and long-​run consequences 64
Growth stimuli 64
Mechanism and long-​run consequences 64
Real estate market impacts 65
Education, technology, and metropolitan growth 65
Chapter summary 66
Questions 67
References and additional readings 67

4 Analyzing metropolitan economies 68

Introduction 68
Main analysis steps 69
Identification of the spatial unit of analysis 69
Analysis of basic growth indicators 69
Analysis of population and income trends 72
Analysis of sectoral employment, income, and industry structure trends 73
Analyzing employment and income growth patterns by industry 74
Analyzing absolute and relative employment shares 80
Synthesis: growth prospects 85
Selected sources of information for historical data and forecasts 86
Chapter summary 86
Questions 86
Project #1: Analyzing metropolitan growth patterns 87
Data to be provided by instructor 87
Main tasks 87
Introduction 87
Analyzing metropolitan growth patterns 87
Growth prospects and real estate markets 88
References and additional readings 88
Appendix 4A: About indices 88

PART C
Analyzing residential real estate markets 95

5 Residential real estate markets 97

Introduction 97
Overview of residential market analysis 98
Basic housing market economics 101
Housing demand: analysis elements 102
Demand for a particular housing type 102
Demand for tenure type 102
Demand for quality/​amenity levels 103

xi
 Contents

Demand for particular location characteristics 103

Households and demographic structures 103
Income, prices, and affordability 106
Mortgage rates and inflation 110
Expectations 112
New housing supply 112
New residential construction 113
The supply of resale homes 114
Chapter summary 115
Questions 116
References and additional readings 116

6 Macroeconomic analysis of residential real estate markets: accounting techniques 118

Introduction 118
Macroeconomic analysis framework 119
Market area definition 119
Define the primary market area by drawing a curve representing equal
commuting time from the site 119
Identify major employment nodes within the primary market area 121
Establish the competitive market area 121
Refine the definition of the competitive market area 121
Adjust boundaries to coincide with census tract boundaries 121
Time frame of the analysis 122
Assessing residential market strength 122
Accounting techniques: the aggregate approach 123
Accounting techniques: the disaggregate approach 128
Chapter summary 135
Questions 135
Project #2: Estimating effective residential demand using the accounting approach 136
The case 136
Development scenarios and price data 136
Mortgage requirements and affordability measures in 2023 136
Target demographic groups and preferences 136
Market area data 137
Questions 137
What to turn in 138
References and additional readings 138

7 Macroeconomic analysis of residential real estate markets: the basics of the

econometric approach 139
Introduction 139
The basics of regression analysis 140
Evaluating regression models 145
Prediction 145
Applications of regression models in market analysis 146
Macro analysis of real estate markets 146

xii
 Contents

Micro analysis of real estate projects 146

A numerical example of a simple linear regression 146
Multiple regression models 150
Chapter summary 150
Questions 151
References and additional readings 151
Appendix 7A: Further information on linear regression models 152

8 Macroeconomic analysis of residential real estate markets: applying the

econometric approach 155
Introduction 155
Applying the econometric approach to the Dallas apartment market 156
Step 1: Collect data on apartment demand, supply, rents, and their drivers 157
Step 2: Set up/​define a behavioral model 159
Specifying the statistical equations 159
Step 3: Estimate the model 164
Rent equation 164
Permit equation 165
Step 4: Develop forecasts and perform sensitivity analysis 167
Chapter summary 174
Questions 175
Project #3: Forecasting the Los Angeles apartment market 176
The data 176
Data analysis 178
References and additional readings 179
Appendix 8A: Calculating the speed of rental adjustment 179

9 Analyzing residential projects: a micro perspective 181

Introduction 182
Qualitative analysis of residential projects 184
Site and location analysis 184
The site and its immediate environment 184
Neighborhood and community attributes 186
Access and linkages 187
Analysis of local market conditions 188
Data sources 188
Consumer targeting and refinement of product design 189
Quantitative analysis of residential projects: estimating project rent and absorption rate 190
Analyzing the competition: the competitive differentials technique 191
Analysis steps 191
Numerical example 192
Critique 198
Hedonic valuation techniques 199
Analysis steps 199
Applications 203

xiii
 Contents

Estimating project price or rent given its amenity package 203

Determining and evaluating optimal project design 204
Numerical example 205
Critique of the hedonic valuation technique 207
Project absorption and gross revenue analysis 207
Numerical example of project absorption analysis 210
Econometric estimation of project absorption 213
Chapter summary 215
Questions 215
Project #4: On hedonic price models 216
Analysis tasks/​questions 216
References and additional readings 218
Appendix 9A: Determining optimal development density 218

10 Analysis of residential real estate markets: an example 221

Introduction 222
Residential development: an example 222
The project and relevant market and financial data 222
Time of entry 222
Residential acreage and zoning 222
Target markets 223
Target group demographics (1998) 223
Distribution of households across income groups (1998) 223
Residential product design 223
Hedonic price analysis 223
Non-​land development costs 224
The design type choice problem 225
Residential finance (permanent loan) 226
Residential preferences 226
Age Group 1 226
Age Group 2 226
Residential market demand, competing supply, and prices 227
Effective demand 227
Competing supply 227
Real market price index 227
Analysis 228
Determining optimal design and densities 228
Selecting the most profitable design type for each of the two target groups 228
Determining the optimal density for each target group 229
Determining how many units of each design type to include in the project 233
Estimating project absorption and revenue schedules 235
Estimate the nominal market price index 236
Determine the schedule of nominal house prices for the two design types to
be included in the project 236
Forecast effective housing market demand 237

xiv
 Contents

Estimate project absorption schedule 239

Estimate project revenue schedule 239
Evaluating residential market studies 239
Macroeconomic analysis of residential markets 239
Micro, site-​specific analysis of residential development projects 240

PART D
Analyzing the market for retail space 241

11 Retail markets and retail market studies 243

Introduction 243
Idiosyncrasies of the market for retail space 244
Idiosyncratic macroeconomic aspects 244
Idiosyncratic microeconomic aspects 244
Retail market fundamentals 245
The demand for retail space 245
Components of retail space demand 245
Purchasing power 247
Spending patterns 249
The supply of retail space 253
Retail market analysis objectives 256
Overview: the three phases of a retail market study 256
Phase I: Analyzing entry 256
Phase II: Location analysis/​evaluation 257
Phase III: Project evaluation 257
Chapter summary 259
Questions 260
References and additional readings 261

12 Analyzing the market for retail space 262

Introduction 263
Phase I: Assessing unrealized sales potential 263
Definition/​types of retail markets 264
Analysis techniques 264
Method 1: Estimating unrealized sales potential through accounting techniques 265
Step 1: Define the spatial unit of analysis 265
Step 2: Estimate the potential volume of sales, SE 265
Step 3: Estimate the sales volume supported by existing establishments, S 265
Step 4: Estimate the market’s unrealized sales potential 266
Example of the application of the accounting technique 266
Critique of the accounting technique 267
Method 2: Estimating unrealized sales potential through econometric techniques 268
Step 1: Define the spatial unit of analysis 268
Step 2: Model and predict sales/​capita 269
Step 3: Estimate the sales volume supported by existing establishments, S 269

xv
 Contents

Step 4: Estimate the market’s unrealized sales potential 269

Example of the application of the econometric approach 269
Critique of the econometric approach 270
Strategic implications of the analysis of market potential 270
Phase II: The spatial behavior of consumers 271
Location selection 274
Spatial demand analysis 275
Consumer 275
Macro-​access 275
Micro-​access 276
Micro-​environment 276
Supply analysis 276
Preferred location traits across center typologies 276
Phase III: Refining and evaluating the retail development scenario 277
Stage 1: Refining the development scenario 277
Stage 2: Retail trade area definition 277
Non-​stylized, multiple-​criteria methodologies for defining trade areas 280
Stylized methodologies for defining trade areas 281
Stage 3: Competition analysis 283
Identify and collect data on competing facilities 283
Evaluate and rate the planned development against competing developments
using the competitive differentials technique 284
Analyze the links between development traits and performance 284
Revise the development idea 285
Stage 4: Shopping pattern analysis 285
Stage 5: Evaluating the development plan 285
Step 1: Estimating the development’s likely capture rate 287
Step 2: Estimating sales potential by line of trade 292
Step 3: Estimating supportable square footage by line of trade 293
Step 4: Estimating achievable retail space rents 294
Step 5: Assessing the development scenario 298
Chapter summary 298
Questions 299
References and additional readings 301
Appendix 12A: Retail data sources 301
Appendix 12B: Further elaboration on the probabilistic approach used by
DiPasquale and Wheaton 303

13 Analyzing the market for retail space: synthesis and market studies 304
Introduction 305
Emphasis of retail market studies 305
Regional and super-​regional centers 305
Community centers 305
Neighborhood centers 305
Strip centers 306

xvi
 Contents

Discount and warehouse stores 306

Off-​price centers 306
Outlet centers 306
Specialty centers 306
Evaluating retail market studies 306
Broader evaluation criteria 306
Common flaws of retail market studies 306
Improper definition of the market/​trade area 307
Inaccurate data on spendable income and retail sales 307
Inappropriate assessment of supportable retail activity 307
Naive assumptions regarding future changes 307
Misinterpretation of “comparables” 307
Poor analysis of the risks pertaining to store and tenant failure 307
Use of very broad data sources 308
Retail development: a simple example 308
Project and market data 308
Time of entry 308
Retail GLA 308
Tenant clusters and sources of demand 308
Tenant data 309
Residential-​based demand and trade area households 309
Job-​based and visitor-​based demand 310
Retail rent index 310
Analysis 310
Questions 312
References and additional readings 314

PART E
Office market analysis 315

14 The market for office space 317

Introduction 317
Office market idiosyncrasies and determinants of project success 318
Office space demand and prospects 319
Office-​using sectors 320
On net absorption and its determinants 320
The supply of office space 322
Income-​related factors 322
Cost-​related factors and availability of capital 323
Overview of office market analysis 324
Analysis of office project viability from the market perspective 326
Chapter summary 326
Questions 326
References and additional readings 327

xvii
 Contents

15 Office market analysis: a macro perspective 329

Introduction 330
Analyzing office markets: a macro perspective 330
What market variables should be analyzed? 331
Office space demand 331
New office supply 331
Office market demand-​supply imbalances 331
Numerical example 332
Assessing office market strength: overview of office market gap estimation 335
Calculating new office space requirements, DS(t) 336
Estimating expected net absorption of office space, AB(t) 336
Simplistic view 336
“Informed” view 336
Simple accounting techniques for forecasting office-​using employment and
office space demand 339
Aggregate approaches: using office employment growth 339
Disaggregate approaches: using industry-​occupation matrices 339
Example 340
Office supply analysis using accounting approaches 343
Vacant stock 345
Completions 345
Critique 345
Econometric forecasting of office markets 346
Forecasting office space absorption, new construction, vacancies, and rents 346
Alternative rental adjustment models 347
Alternative 1: Traditional model with variable structural vacancy rate 350
Alternative 2: The equilibrium or desired rent model 350
Submarket analysis 351
Types of office nodes 352
Downtown district 352
Uptown district 352
Office parks 352
Special nodes 352
Shopping center office nodes 352
How to analyze office submarkets 352
Chapter summary 354
Questions 355
Project #5: Econometric forecasting of office markets 355
The case 355
The questions 356
The report 356
The data 356
References and additional readings 358

16 Micro analysis of office markets 359

Introduction 359

xviii
 Contents

Analyzing office locations 360

Productive or firm amenities 360
Worker/​residential amenities 362
Prestige and visibility 362
Location constraints 362
But, do office rents fully reflect location amenity differentials? 362
Analyzing office projects: estimating achievable rents and absorption schedules 364
Collection of data for competing office properties 365
Identify comparable office properties 365
Collect information on the identified comparables 366
Data analysis 368
Using the competitive differentials technique to assess project strength
and achievable rental rates 368
Critique 370
Using hedonic methodologies to estimate achievable office lease rates 370
Variables included in hedonic office rent models 370
Usefulness of hedonic rent analysis 372
Critique 372
Developing office-​project absorption and revenue schedules 373
Chapter summary 376
Questions 377
References and additional readings 378

17 Office market analysis: synthesis and market studies 380

Introduction 380
Example: multi-​tenant office development 381
The project 381
Time of entry 381
Structure typology 381
Analysis of hedonic lease rates, 1996 381
Market information 382
Analysis 382
Evaluating office market studies 384
Macro analysis of office markets 384
Micro, site-​, and project-​specific analysis of office projects 384
Project #6: Evaluating an office development scenario 388
What to turn in 388
Micro market data 388
Site and structure attributes 388
Office lease rates: hedonic analysis results 388
Macro market conditions 388
Forecast of metropolitan office market fundamentals 388
Space requirements per worker 390
Office market absorption equation 390
References and additional readings 390

xix
 Contents

PART F
Industrial market analysis 391

18 Industrial space market 393

Introduction 394
Industrial market idiosyncrasies 395
Functional heterogeneity 395
Large non-​speculative component 395
Industrial cycles parallel business, as opposed to real estate cycles 396
The demand for industrial space 396
Growth in industrial output or employment 396
Shifts in the real after-​tax cost of corporate capital 397
Replacement demand 397
Square feet per employee 398
The supply of industrial space 400
Industrial market analysis: an overview 401
Macroeconomic analysis of industrial markets 401
Accounting approaches 401
Econometric approaches 402
Microeconomic site-​specific analysis of industrial projects 402
Macroeconomic analysis of industrial markets 402
Estimating the industrial demand-​supply gap through accounting techniques 402
Accounting techniques for forecasting industrial space demand 403
Method 1: Demand analysis for general-​purpose industrial space 403
Method 2: Best-​fit industry analysis 404
Econometric modeling of the industrial market 409
Modeling investment in non-​speculative industrial buildings 410
Modeling the speculative industrial market 411
Site-​specific analysis of industrial projects 413
Step 1: Evaluation of site attributes 413
Site features 413
Site access and linkages 414
Traditional manufacturing (production) space 414
Warehouse and distribution space 414
Research and development space 415
Community constraints 415
Step 2: Identification of types of tenants most likely to be attracted to the site 416
Step 3: Analysis of the competition 416
Step 4: Estimation of project absorption schedule and achievable rents or land prices 417
On industrial market studies: evaluation criteria 419
Macro, non-​site-​specific analysis 419
Site-​specific analysis of development projects 420
Chapter summary 420
Questions 421
References and additional readings 422

xx
 Contents

PART G
Data sources 425

19 Data sources for real estate market analysis 427

Introduction 427
Sources for real estate market data 427
Sources for economic and demographic data 431
References 432

Index 433

xxi
 ACKNOWLEDGMENTS

The production of this book was a long and rather sentimental journey for me personally and would
not have reached publication were it not for the encouragement and support of a number of people
throughout the last 20 years. The beginning of this project dates back to 2000 and for that I have to thank
Rena’s colleagues at the University of Southern California, and especially Professors Peter Gordon and
Harry Richardson who advised me to put together in a book Rena’s excellent teaching notes so they
would not be lost after her passing.
I started working on this project after I had moved to Boston at the beginning of 2001. During that
period, Professor Karen Polenske (with whom both Rena and I had worked closely during our doctoral
studies at MIT) took it upon herself to meet with me from time to time in order to encourage me to finish
and publish the book. I would like to thank her for that, as she kept me focused on the project in a way that
I managed to finish the first draft of the whole book by the end of 2004, while still in Boston.
At the end of 2004, I left the United States and after a three-​year stop in Athens I returned to my home
country, Cyprus. During that time, I continued refining the book and it was practically finished by 2007.
Although I had finished the book then, I was very skeptical about publishing something with Rena’s name
as the author without her having seen it and approved it.This skepticism was based on my experience with
all the papers that she published. As she was a perfectionist, she would review the manuscripts very thor-
oughly and countless times before sending them to the editors.
The person who played the catalytic role in getting the book eventually to the publishers was Ray Torto.
I would like to express my gratitude to him for his support and encouragement to update and publish
the book, as the text was so well received by his students at the graduate studies program at the Harvard
Graduate School of Design. Ray’s strong encouragement to publish the book based on the very positive
feedback he received from his students was the key factor that helped me overcome my hesitations and
eventually decide to publish the book.
This happened in 2015, when the actual search for a publisher started. That effort did not come to fru-
ition until 2018, when I was approached by Routledge with a proposal for publishing the manuscript that
was used by Ray as the basic textbook for his course at the Harvard Graduate School of Design. For this
turn of events I have to thank Samuel Azasu for his very positive comments about the book and his rec-
ommendation to Ed Needle, the Routledge Editor for Construction and Real Estate, to publish the book.
After the contract was signed, the final manuscript production moved very smoothly and for that I have to
thank Ed Needle and Patrick Hetherington at Routledge for their help, guidance, and continuous support.

xxii
newgenprepdf

Acknowledgments

Last but not least, I would like to thank our mentor at MIT, Professor Bill Wheaton, who supervised
both of our doctoral dissertations and gave us an excellent foundation in the subject of real estate eco-
nomics, which is reflected to a large extent in the book.
In concluding, I would like to thank my family and Rena’s family for their support and encouragement
throughout the duration of this project, but also everybody else that has contributed in any way to the
realization of this project and has not been mentioned here.

xxiii
 PART A

Introduction
newgenprepdf

1
MARKET ANALYSIS IN PERSPECTIVE

Introduction 3
Book premise, focus, and objectives 4
Market analysis in perspective 5
Focus and contribution 5
General macroeconomic/​metropolitan growth forces 6
Macroeconomic forces that drive the demand for and supply of specific
property types 7
Location and project-​specific factors 7
Market analysis components 8
Metropolitan growth analysis 9
Macro analysis of real estate markets 11
Micro (site-​specific) analysis of development projects 12
Market studies and their emphasis 12
Residential 13
Retail 13
Office 13
Industrial 13
Market studies in general: evaluation criteria 14
Book structure and content 15
References and additional readings 17

Introduction
Real estate market research is a relatively new field of study, which builds on the accumulated stock of
knowledge from urban and real estate economics to address two major questions:

1. How do the various real estate markets function at the broader macroeconomic, or metropolitan level,
and at the more narrow microeconomic, or site level of analysis?
2. How exactly should one go about analyzing meaningfully real estate markets for development and/​or
investment purposes?

3
 Introduction

These two questions highlight two major areas of knowledge in the field of real estate market research
that are critically linked. In addressing the first question, this textbook will explore the basic economic
principles guiding the operation of real estate markets, placing special emphasis on their peculiarities and
idiosyncratic features. As it will become apparent, it is the knowledge of these features that helps one more
accurately evaluate specific real estate development or investment opportunities. In addressing the second
question, this textbook will explore both state-​of-​the-art as well as less sophisticated techniques that market
analysts can employ to proactively design and evaluate real estate development or investment programs.
The major premise underlying the structure and content of the book is that a better understanding of
how markets function coupled with a working knowledge of real estate market analysis tools will help real
estate professionals:

• Become sophisticated users of market studies.

• Communicate more effectively with colleagues in the real estate development and investment
community.
• Use skillfully available data to evaluate the prospects of a market and a specific project.

Book premise, focus, and objectives

The author rejects emphatically the “Field of Dreams” philosophy, “Build it and they will come...,” which
clearly dominated the real estate development industry during the 1980s. This philosophy suggests that
supply generates demand, a proposition that runs against fundamental economic principles. On the con-
trary, the book advances the premise that solid market research, based on a good understanding of the
peculiarities and idiosyncrasies of real estate markets, is a necessary and crucial component of the real estate
decision-​making process.
The operational and bottom-​line objective of market research is to help generate the knowledge and
information that relates to space market operations and is necessary for assessing whether a project is likely
to meet key development and investment objectives. For example, a typical key real estate development
objective involves the maximization of residual land value (RLV) or, alternatively, profit per area of land.
The two major factors involved in the calculation of RLV are property values or prices, and non-​land
development costs per square foot (see Figure 1.1). It is the strong link between property values and a
property’s income-​earning capacity that underscores the importance of market analysis in real estate devel-
opment and investment decision making.
The strong link between property values and a property’s income is embedded in the income approach
to value, the most relevant property valuation technique from an investment point of view.1 In its simplest
version, the income approach postulates that property value is a function of the property’s net operating
income (NOI) and the prevailing market capitalization rate.2
Movements in a property’s NOI are strongly influenced by movements in space or leasing markets. In
particular, a property’s NOI is a function of the rent it commands in the marketplace, its occupancy rate
and operating expenses. Property rents and occupancy rates are in turn shaped by market demand and
supply forces, as well as by idiosyncratic property and location characteristics. In particular, it has been
widely established in the real estate literature that movements in individual-​property rent and occupancy
levels are primarily driven by changes in market demand and supply conditions, while differences in rent and
occupancy levels across individual properties competing within the same marketplace are determined by
idiosyncratic property and location characteristics.
While capitalization rates are not typically the initial focus of market studies for development projects,
these rates are strongly influenced by movements in local market conditions. For example, empirical ana-
lysis of movements in office capitalization rates has provided evidence of the influence of market-​related
variables, such as absorption levels and changes in rents (Sivitanidou and Sivitanides, 1999).This underscores

4
 Market analysis in perspective

max RLV = (Price − Cost) FAR Floor Area Ratio

(needs to be
= [(NOI/Cap Rate) − Cost] FAR optimized)

Determined in = [(f(Rent, Occupancy)/Cap Rate) - Cost] FAR

the asset market

Determined in the user market

Net Operating Income Capitalization rate:

Income return required for holding an asset (driven by
expected rental income growth and market strength,
the opportunity cost of investment capital, risk factors, tax policy)

Note: Price, NOI, Rent, and Cost are in $/sq. ft.; FAR = sq. ft. of space / sq. ft. of land

Figure 1.1  Inputs in residual land value maximization

the additional importance of accurately evaluating real estate market prospects and the sources of property
income.
In sum, the role of market research is to analyze the market-​related determinants of project profitability
at the macro and micro level, help address questions of market entry and proactive project design, and
provide the bottom-​line rent and absorption figures that are necessary for the assessment of the financial
feasibility of a particular development.
Within this context, the focus of this book is on the forces and processes that drive broader space-​
market performance; the idiosyncratic project-​and location-​specific factors that shape deviations of indi-
vidual property performance from average market performance; and the techniques that analysts can use
to forecast market performance and assess the income-​earning prospects of specific development projects.
Hence the four major objectives of this book are:

1. To examine the major macroeconomic processes and factors that shape broader market performance
of the four major property types.
2. To present advanced as well as less sophisticated techniques for analyzing and forecasting broader
market performance for the four major property types.
3. To discuss project-​specific and location factors that influence individual property performance.
4. To present techniques for analyzing and forecasting individual project performance in a way that is
consistent with the analysis of the prospects of the broader market.

Market analysis in perspective

Focus and contribution
Generalizing along the lines discussed above, it can be argued that the major focus and contribu-
tion of market analysis for real estate is the assessment of the underlying determinants of a project’s

5
 Introduction

Project income-
earning
capacity
influences
a. General macro/metro c. Location and
growth forces property-specific
factors
b. Macro forces driving (amenities and
specific space markets
constraints)
(e.g., residential, office)

Fixed or time-variant
Time-variant

Effective Competing Location Competing

space space and space
demand supply product supply
demand

Average Consumer/tenant Project-specific

market prices/rents preferences prices/rents
market absorption and project absorption
ability
to pay

Determine the average performance Determine the competitive position and performance of a project
of a real estate market through time within its market at a given point in time

Figure 1.2  Influences on project income-​earning capacity

income-​earning capacity at two distinct levels: the macroeconomic (non-​site) level and the microeconomic or
project-​specific level. As indicated in Figure 1.2, a property’s income-​earning capacity is influenced by three
major sets of factors:

1 . General macroeconomic and metropolitan growth forces that may affect all property types.
2. Macroeconomic forces that drive or influence the demand for and supply of specific property types.
3. Location-​and project-​specific factors that may contribute to deviations of individual property per-
formance from market performance.

We elaborate on each of these factors in turn.

General macroeconomic/​metropolitan growth forces

These include forces that may affect the economy of a metropolitan area and all its real estate markets.
Examples of such forces include increasing global competition and technological advancements that may
undermine an area’s major industries; significant changes in exchange rates that may render foreign imports
more attractive at the expense of domestic industries; the economic well-​being of America’s major trading
partners, which can have a significant impact on export-​oriented metropolitan economies; or changes
in interest rates that may influence demand as well as the volume of new construction and, therefore,
the supply of all property types nationally (Barkham, 2013). These forces, being international, national

6
 Market analysis in perspective

or regional in scope, may very well impact a metropolitan area’s economy and local real estate market
(Hatemi et al., 2014). The effect of these forces may vary from metropolitan area to metropolitan area due
to differences in industrial structure and the sectoral composition of the local economy.

Macroeconomic forces that drive the demand for and supply of specific property types
These include macroeconomic factors such as population demographics, income and employment in
specific sectors of an area’s economy that affect demand for a particular type of property, or other macro-
economic drivers that affect new construction activity for a specific real estate sector. For example,
Wheaton and Nechayev (2008) have validated the effect of income and population on house prices
and, by extension, on housing demand. Similarly, a number of empirical studies (Rosen, 1984; Wheaton,
1987; Ibanez and Pennington-​Cross, 2013) have shown that the demand for office space is driven by
employment in specific sectors of the economy, referred to as “office-​using” employment. These include
primarily the FIRE (Finance, Insurance, and Real Estate) sector and Services, as defined in the North
American Industrial Classification System (NAICS).3 Thus, macroeconomic forces that affect specifically
these sectors, such as the health of the nation’s financial system, are bound to have an effect on an area’s
demand for office space. Similarly, Ibanez and Pennington-​Cross (2013) and Wheaton and Torto (1990)
have shown that demand for industrial space is driven primarily by manufacturing and wholesale trade
employment. Macroeconomic forces that affect an area’s employment in these sectors will have an impact
on the local industrial market.
Certainly in today’s world, the effect of technological changes on the demand for space is an important
consideration. For instance, the internet is affecting the size and location of retail and industrial space, while
new trends in office design and workplace strategies are changing the relationship between each new job
and the demand for office space, and even its location.
On the supply side, the major factors that drive new construction for a specific property type include
land costs, which vary across different land uses within the same metropolitan area, as well as across metro-
politan areas; labor costs that generally vary across metropolitan areas due to differences in labor market
conditions; and the cost of construction materials. All these macroeconomic factors, both broader and more
property-​type specific, jointly shape movements in effective space demand, and competing supply and,
hence, movements in absorption, vacancy rates, and average market rents/​prices through time.

Location and project-​specific factors

These include factors that affect the demand for and the competing supply of a particular real estate
product at a specific location. In particular, factors such as the quality and amenities of the project and
its immediate location, along with consumer/​tenant preferences and ability to pay, play a critical role in
determining location and product demand. On the supply side, factors such as local zoning constraints and
growth controls may create supply shortages and limit competition. Furthermore, the spatial distribution
of existing and new comparable properties plays a key role in determining the intensity of the competi-
tion faced by a specific property at a given location, especially in the case of retail. Overall, all these factors
shape the competitive position of a project within its market area, which plays a crucial role in determining
the level of rents/​prices and absorption it can achieve relative to the average market rent and total market
absorption, respectively.
To better understand the difference between the macroeconomic and microeconomic influences on
real estate markets, consider Figure 1.3, portraying movements in rents for two different projects/​locations
with different amenity levels. Although the time path of rents is similar for both project locations, there
are persistent differences in levels. Variations in performance through time are due to changes in macro-
economic forces that influence a metropolitan area’s economy and affect all locations/​projects similarly.

7
 Introduction

Rent/amenity continuum
Inter-temporal variations:
Variations in performance through time are due to
macro forces that affect similarly all projects/locations
within a competitive market area

Project/
Location A
Cross-section differences:
Differences in performance across projects/locations at the
same point in time are due to differences in project/location
amenities (and/or constraints)

Project/
Location B

Time t Time t+n Time

Note: Long-run shifts in cross-section differences may occur due to shifts in location amenities and/or constraints,
or the value placed on them.

Figure 1.3  Macroeconomic and microeconomic influences in real estate markets

Differences in performance across projects and locations at the same point in time (referred to also as cross-​
sectional differences) are due to differences in their amenity levels and/​or constraints.
Long-​run shifts in cross-​sectional rent differentials across projects/​locations within the same market are
more likely due to changes in amenity/​constraint differentials, or the way these differentials are valued. For
example, some have argued that dispersion of economic activity, induced by advances in information and
communication technologies, is narrowing locational differentials in rents and prices as the value of loca-
tional proximity has declined (Sivitanidou, 1997). However, recent trends in the urbanization of population,
especially of millennials, are having the opposite effect, raising prices and rents in some urban areas more
strongly than in non-​urbanized areas.

Market analysis components

A comprehensive and well-​integrated market study should appropriately examine all of the aforementioned
macroeconomic and microeconomic influences. As such, it should comprise three major components:

1 . Growth analysis of the metropolitan/​geographic area –​what are its economic engines?

2. Macroeconomic analysis of the real estate markets –​how is real estate performing within this geography?
3. Micro-​marketability analysis of the site or location –​what are the competitive advantages/​disadvantages
of the site or location?

The basic premise underlying metropolitan growth analysis is that real estate is durable and, therefore,
demand for new space is driven by growth at the margin. Economic justification of new real estate or new
buildings usually requires economic growth.

8
 Market analysis in perspective

Can there be any demand for real estate within a metropolitan area that is not experiencing any popu-
lation, employment, or income growth? Yes. Two sources of such demand can be identified:

1 . Replacement demand, which arises because of the depreciation of the existing stock.
2. Relocation demand, arising from intra-​metropolitan location adjustments, which is usually induced by
advances in information and communication technologies, rent or price differentials across markets or
submarkets, or other factors influencing the location decisions of firms.

It should be noted that depreciation rates are very low in the real estate sector because of its durability, and
therefore, replacement demand has historically been a very small percentage of existing stock. However,
new building technology and workplace design strategies have been contributing to higher depreciation
rates, especially for commercial property. In any case, by definition, these two sources of demand cannot
generate an increase in aggregate real estate demand at the local level.
The premise underlying the macro analysis component of market research is that macro/​market-​wide
determinants of real estate demand and supply change through time causing fluctuations in rents and
vacancy rates, thereby creating opportunities for profitable investments. Taking advantage of these oppor-
tunities requires identification of niches in terms of property types and target markets that offer the best
return prospects, as well as appropriate timing and execution of market entry and exit in order to profit
from future increases in rents and occupancy rates. Within this context, the purpose of macroeconomic
analysis is to assess, accurately, the most likely future movements in demand, supply, rents and occupancy
rates, and provide the basis for sound and profitable investment decisions.
The premise underlying the micro-​marketability analysis component is that proactive (and not reactive)
product design, matching consumer/​tenant preferences for location and product attributes, is necessary
for maximizing project value and developing a successful project. The underlying rationale of this premise
is that not only broader market prospects but also idiosyncratic project and location features are crucial
in determining project-​specific performance and achievable rental revenue levels. Thus, developers and
investors can maximize value and profits by carefully identifying the type of real estate products that are
valued by consumers in the marketplace.
Table  1.1 summarizes the basic characteristics of the three major components of real estate market
studies. The discussion below elaborates on these components in terms of the unit of analysis, the focus of
analysis, the bottom-​line outcome, and contribution.

Metropolitan growth analysis

GEOGRAPHIC ANALYSIS  The unit of analysis during this stage could be defined on the basis of the
geographical boundaries of the statistical area. Statistical areas are metropolitan and micropolitan statistical
areas (metro and micro areas), which are geographic entities delineated by the Office of Management and
Budget (OMB) in the U.S. for use by Federal statistical agencies in collecting, tabulating, and publishing
Federal statistics. The term “Core Based Statistical Area” (CBSA) is a collective term for both metro and
micro areas. A metro area contains a core urban area of 50,000 or more population, while a micro area
contains an urban core of at least 10,000 (but less than 50,000) population. Each metro or micro area
consists of one or more counties and includes the counties containing the core urban area, as well as any
adjacent counties that have a high degree of social and economic integration (as measured by commuting
to work) with the urban core.4 The rationale is dual. First, the metropolitan area represents an integrated
economy that shapes the exogenous economic forces that drive movements in aggregate demand and supply
for real estate and, hence, the time path of absorption, completions, vacancy rates, and rents. Second, the
metropolitan area represents an integrated real estate market because, from a real estate demand point of

9
 Introduction

Table 1.1  Market analysis components

a. Metropolitan growth b. Macro analysis of real estate c. Micro analysis of development

analysis markets projects
Unit of The metropolitan area The project’s competitive The project and its immediate and
analysis (economic market) of the market area: the geographic neighborhood environment
project area funneling effective
demand for a project and
encompassing the locations
of all competing projects
Focus of Analysis of the sources Analysis of historical and Analysis of location/​project
analysis and consequences of current indicators of market amenities and constraints,
metropolitan growth. In strength: effective market consumer preferences and
terms of sources, emphasis is demand, competing supply, ability to pay for location and
placed on industry demand prices/​rents, and vacancies project attributes, the project’s
shifts, industrial structure by product design type competitive position, and
and amenity-​induced and consumer segment, if implications regarding project
migration. In terms of possible absorption and prices
consequences, emphasis
is placed on population,
employment, and income
growth paths.
Bottom-​line Type and magnitude of Forecasts of the time paths of Location selection/​evaluation
outcome(s) growth the metropolitan effective market demand/​ and proactive design of a
area in question is net absorption, competing marketable project; forecasts of
experiencing and will likely supply, rents, and vacancies project-​specific rent/​price and
experience in the future; at the time of sale in the absorption schedules
and the set of opportunities case of for-​sale projects
and constraints such and during the expected
growth patterns present for holding period in the
real estate development/​ case of income-​producing
investment properties
Contribution Facilitates a broad assessment Facilitates the identification Provides inputs that are crucial
of demand at the margin for of product and consumer in evaluating the financial
real estate property types niches and the best timing feasibility and viability of the
for market entry; provides project
inputs for micro-​market
analysis and financial
feasibility analysis

view, each location is, to some extent, substitutable with any other location within the same metropolitan
area, but hardly with locations in other metropolitan areas.
For example, it would be inappropriate to perform an isolated macroeconomic analysis of a specific office
submarket and completely ignore the demand and supply conditions and prospects within the metropolitan
area. First, demand for a specific location and office building is shaped not only by growth in firms housed
in that submarket, but also by growth in firms in other submarkets and new firms deciding to locate within
the broader metropolitan area. Second, it is rather unreasonable to assume that an office building in one
submarket does not compete at all with similar buildings in other submarkets within the same metropolitan
area.5 Thus, an isolated analysis of the submarket could lead to underestimation of the competition the pro-
ject may be facing when completed and an erroneous assessment of its prospects and viability.

10
 Market analysis in perspective

FOCUS OF THE ANALYSIS  The focus of the analysis is the sources, mechanism and consequences of
metropolitan growth. In terms of sources, emphasis is placed on industry product demand shifts, industrial
structure, amenity-​induced migration and labor supply shifts. In terms of mechanism and consequences,
the focus is on the interaction between demand and supply forces in the product and labor markets, their
effects on population, employment and income growth, and their implications for real estate demand.

THE BOTTOM-​LINE  The goal of this analysis is to determine and assess the pattern of growth a metropolitan
area is experiencing and will likely experience (demand-​induced, or supply-​induced) in the years ahead;
the direction and magnitude of physical growth (employment and capital stock) and growth in wealth
associated with such a pattern; and the set of opportunities it presents for real estate development and
investment.

CONTRIBUTION  Metropolitan growth analysis sets the stage for a broad assessment of the demand for real
estate at the margin and provides some clues with respect to broader product type and consumer niches
that may experience strong demand growth in the future. It should be emphasized that this analysis never
reaches a go or no-​go decision for a specific project. Rather, it provides a context for making the project-​
specific decisions.

Macro analysis of real estate markets

This stage focuses on the real estate market within the project’s competitive market area. This is the geo-
graphic area channeling effective demand for the project and encompassing the locations of all competing
projects. With the exception of some retail property types, it is broadly accepted that any project competes
with all similar projects that are located within the same metropolitan area. Thus, the typical geographic
unit of macro analysis for any property type is the metropolitan area within which the project under
consideration is located. This definition serves also very well the increased data requirements for a more
sophisticated analysis of an area’s prospects.

FOCUS OF THE ANALYSIS  The analysis focuses on historical and current indicators of market strength, such
as effective market demand, competing supply, vacancy rates, and rents/​prices by product sub-​type and
consumer segment, to the extent that data availability allows it. Otherwise, the focus of the analysis will
be on the broader property type that is more relevant for the project under consideration. It should be
emphasized that the focus is not simply on reviewing and understanding the direction of recent trends, but
also on gaining a good understanding of the factors driving the movements in critical market indicators, in
a way that will allow the analyst to forecast future movements as accurately as possible.

BOTTOM-​LINE OUTCOME  The bottom-​line outcome of this analysis includes forecasts of effective market
demand (or net absorption), completions, total inventory, vacancy rates, and prices/​rents at the time of
sale in the case of for-​sale projects, or over the expected holding period, in the case of income-​producing
properties.

CONTRIBUTION  Macro analysis of real estate markets provides valuable guidance for time-​of-​entry decisions
and can facilitate the identification of product and consumer niches. It also provides crucial inputs for
micro-​market analysis and financial feasibility analysis. Such inputs include forecasts of market supply,
demand and prices/​rents. Market supply and demand forecasts are necessary for calculating the project’s
fair market share and absorption schedule, while market rent/​price forecasts can provide the basis for
forecasting project gross revenue and movements in project rents/​prices.

11
 Introduction

Micro (site-specific) analysis of development projects

UNIT OF ANALYSIS  This stage focuses on the specific site/​project under consideration and its immediate
environment. Major competing developments are also identified and more carefully evaluated during this
stage. In reality, most market research done by developers/​investors begins here when they are considering
a site. This is usually called the “bottom-​up” approach in contrast with the “top-​down” approach, which
is where economists start their analysis. We will demonstrate these approaches and their differences as we
proceed in this book.

FOCUS OF THE ANALYSIS  This stage focuses on the careful evaluation of location/​ project amenities and
constraints; evaluation of user preferences as well as ability and willingness to pay for specific location
and project attributes using hedonic valuation techniques and data on most recent market transactions;
assessment of the project’s competitive position within the local marketplace; and development of rent/​
price and absorption schedules using as a framework the forecasts for the broader market area developed
at the macro analysis stage. Schmitz and Brett (2009) argue that qualitative analytical approaches, such as
surveys of consumer preferences for a certain type of product, are becoming increasingly important at this
stage because of the increasing segmentation and specialization of real estate markets.

BOTTOM-​LINE OUTCOME  The bottom-​line outcome of this stage includes evaluation and/​or selection of
the site and location, proactive design adjustments to maximize the marketability and value of the project,
and forecasts of project-​specific rent/​price and absorption schedules.

CONTRIBUTION  Micro analysis of real estate markets provides inputs that are necessary for assessing the
financial feasibility of the project under consideration, such as the prices or rents it will command upon
completion and its most likely absorption schedule. The latter is not only necessary for formulating the
project’s revenue stream and evaluating its financial feasibility, but also for evaluating different project
phasing alternatives.
In sum, market analysis provides information necessary for:

• Assessing metropolitan growth prospects and their likely influence on the local real estate market.
• Assessing market strength and prospects for a particular property type (e.g., office industrial, retail or
apartment) or sub-​type.
• Evaluating site-​specific location benefits and constraints, proactively refining product design for max-
imum marketability, and developing forecasts of project rents/​prices, absorption rates, and revenue
streams.

It should be emphasized that market analysis differs from financial feasibility analysis. Market analysis
provides insights with respect to the aggregate demand for the real estate product, such as how much
absorption can be expected and at what rents/​prices, while financial feasibility analysis utilizes price/​rent
and absorption schedules derived from market research in combination with property operating and other
cost data to determine whether a project meets certain financial objectives.

Market studies and their emphasis

There are predispositions in the field regarding the relative importance of each of the three market ana-
lysis components. Most market research studies have each component but the macro analysis of specific
property markets and micro analysis of specific projects within these markets differ in emphasis. For
example:

12
 Market analysis in perspective

Residential
Residential macro analysis often places overwhelming emphasis on effective demand and affordability, while
downplaying the role of competing supply. This has some merit but due consideration should be given to
the relationship between the project, the competition, and location attributes. Residential micro analysis
does place strong emphasis on project design, as it should, since this is extremely important. The biggest
challenge for residential developers involves successful consumer targeting and identification of the most
appropriate amenity packages to provide to the selected target groups. We will discuss how analysts can
optimize amenity mix in residential product design in future chapters.

Retail
Retail market studies are typically better than the studies for other property types in terms of the level of
detail, sophistication, and substance.
Retail macro analysis tries to estimate the “Void,” or “Gap,” in the retail market.This is the unrealized retail
sales potential within a project’s trade area. The market void or unrealized sales potential essentially refers
to the additional sales that can be captured in a metropolitan area given its demographics and its household
spending patterns.
Retail micro analysis places great emphasis on (a) location with respect to consumer markets and sales poten-
tial, and (b) project design that focuses on issues of tenant mix and anchor tenants. A shopping center’s retail
sales are greatly dependent on consumer access. Given that retail activities are quite revenue-​sensitive, the focus
on consumer access is quite justifiable. The emphasis on tenant mix is also well founded given the positive
effect that co-​location of certain types of stores may have on sales (often referred to as shopping externalities).6

Office
Office macro analysis places emphasis on the analysis of competing supply. The emphasis of office market
studies on supply is not surprising given the extreme cyclical fluctuations of office construction due to the
time it usually takes to deliver office buildings. The office building boom of the 1980s has shown that it
is the lumpiness and long gestation-​period of office construction investments, combined with a myopic-​
expectations behavior on the part of investors/​developers that have largely driven such fluctuations.7
Because of this high volatility, the most important information that the macroeconomic analysis of office
markets can provide is the timing and cyclicality of rent and vacancy metrics.
Although the focus on supply is well founded, the analysis of office-​space demand prospects should not
be overlooked. Given the considerable variation in office-​employment growth rates through time, office-​
space demand prospects should be also thoroughly and carefully analyzed at the macroeconomic level.
Increased focus on office demand prospects is especially warranted in the years ahead given indications that
office employment will not continue to grow as quickly as it did in the past.
Office location is very important and should be more thoroughly evaluated because office tenants in
today’s world are both cost sensitive and attuned to attracting and retaining skilled workers. The best sites
are those providing for the minimization of tenant costs and a good pool of workers. A survey conducted
by the Urban Land Institute several years ago concluded that neighborhood quality matters more than
design. Nevertheless, design aspects should not be overlooked.

Industrial
Industrial macro analysis focuses mainly on industrial development potential at the metropolitan level. The
macro analysis of industrial markets is more complicated because industrial assets are highly heterogeneous,

13
 Introduction

comprising mainly production, warehouse/​distribution, and research/​development (R&D) space. These

different types and uses of space may be driven by different macro influences, but with the exception per-
haps of R&D, the degree of segmentation across product types is not very clear.8
Industrial micro analysis typically focuses on location accessibility, as well as site and space requirements.
The emphasis on location is well founded given the special access and site requirements of industrial
projects, as well as problems of community acceptance. More emphasis, however, should be placed on the
productive attributes of industrial sites because industrial firms are largely cost sensitive and their location
decisions are heavily influenced by the productive amenities that characterize different sites. The emphasis
on site and space requirements is also well founded in light of continuous technological and transportation
advances and the introduction of new manufacturing, warehouse, and distribution systems and processes.
For example, the latest distribution and supply chain practices emphasize flow-​through, material-​handling,
and vertical storage systems that require different design features for warehouse and distribution facilities.

Market studies in general: evaluation criteria

One of the objectives of this textbook is to help the reader become a sophisticated user of market studies.
It is presumed that users of this text will buy/​evaluate market research rather than undertake their own
study. Within this context, it is important to have in mind what constitutes a good market study in general.
Below we outline the major questions that need to be asked when evaluating a market study, as they relate
to its overall appeal, objectives, methodologies, data, and conclusions.

General evaluation

How persuasive are they? Why?

Objectives

Are the issues addressed by the study clearly articulated and explained?
Are the objectives of the study—​primary and secondary—​clearly spelled out?

Analysis elements

Are the peculiarities of the real estate market being studied clearly acknowledged and appropriately taken
into account?
Are the most important market-​related determinants of project profitability—​both at the macro and micro
level—​g iven due consideration?

Methodologies, assumptions, and data

Are the methodologies employed in the study clearly described and fully explained?
Are their pros and cons discussed? Are assumptions clearly spelled out?

14
 Market analysis in perspective

Are alternative methodologies used and are their outcomes compared to assess the reasonableness of the
final estimates?
Are future market conditions analyzed using reliable forecasting techniques, or do they merely represent
simple extrapolations of the most recent trends?
Are the data used—​primary and/​or secondary—​reliable? Are the data sourced? Are the methods of data
collection clearly listed and explained?
Are numerical variables assigned a range of values to account for potential forecast errors or uncertainty?
Is the impact of alternative numerical assumptions on market and project performance examined through
careful sensitivity analysis?
Are the data incorporated in the analysis in a coherent and integral fashion or is the study full of “boiler-
plate” information?

Conclusions

Are the conclusions of the study inherently linked to its objectives and the analysis outcomes?
Are they robust, that is, do they still hold for small changes in the assumed values of exogenous variables
employed in the study’s methodology?
Are conclusions based on subjective evaluation, intuition or gut feeling clearly distinguished from those
that are derived from careful analysis?
Are the results of sensitivity analysis regarding the performance of the project under alternative assumptions
integrated in the recommendations of the study?

Book structure and content

Given the main themes of market research and the objectives of this book, the material herein is structured
in six parts.

Part A: Introduction
The second chapter of this introductory section focuses on the basic principles of the economics of real
estate markets. Particular topics include real estate demand and the relevance of absorption concepts;
concepts of real estate supply, such as stock, completions, starts, and permits; and real estate cycles with
emphasis on rent and vacancy adjustments.

Part B: Metropolitan growth analysis

This section focuses on how to analyze meaningfully and how to interpret metro-​wide economic trends
and prospects, and how to make inferences regarding their effect on the demand for different property types.
In particular, Chapter 3 focuses on the sources, mechanisms, and different causes of metropolitan growth.

Part C: Analyzing residential real estate markets

Part C consists of six chapters. The first chapter, Chapter 5, focuses on the basic economics of housing
markets, while the subsequent three chapters elaborate on the macroeconomic analysis of residential
markets and on the techniques for assessing residential market strength and developing forecasts for demand,

15
 Introduction

supply, vacancy rates, and rents. In particular, Chapter 6 focuses on less sophisticated or non-​econometric
techniques for analyzing residential real estate markets at the macro level. Chapter 7 focuses on the basics of
regression analysis as a way of an introduction to Chapter 8, which elaborates on the specifics of the econo-
metric analysis of residential real estate markets through an application to the Dallas apartment market.
Chapter 9 focuses on the microeconomic analysis of residential projects, and particularly on the use of
accounting and econometric techniques for the estimation of project rents/​prices and absorption schedule,
as well as for refining project design in terms of what to build (housing type, development density, attribute
mix) and for whom. Finally, Chapter 10 synthesizes macroeconomic and microeconomic analysis elem-
ents in an example of a hypothetical residential development project and provides guidelines for evaluating
residential market studies.

Part D: Analyzing the market for retail space

Part D focuses on the market for retail space and consists of three chapters. In particular, Chapter 11
focuses on the macroeconomic determinants of the demand for retail space, consumer shopping behavior,
and the different retail formats that compose retail space supply. Chapter  12 elaborates on the three
major phases of a retail market study. The discussion of Phase I focuses on the assessment of a market’s
unrealized sales potential (either at the aggregate level or by product line), and the strategic implications
of such analysis. The discussion of Phases II and III focuses on the microeconomic analysis of retail
development projects and, specifically, the analysis and evaluation of retail sites, and the refinement and
evaluation of the final retail development concept. Specific micro issues discussed include location and
site requirements for the different types of retail developments; the identification of the most profitable
tenant mix; and the assessment of potential tenant sales and supportable square footage, as well as anchor
and non-​anchor rents.
Chapter 13, the last chapter on retail markets, presents a numerical example of the analysis for a retail
development project, outlines the points of emphasis of retail market studies for different retail formats, and
provides guidelines for evaluating retail market studies.

Part E: Office market analysis

Part E focuses on the analysis of office markets and consists of four chapters. Chapter 14 elaborates on
the fundamentals of office markets, and specifically on their cyclical nature and the drivers of demand
for and supply of office space. The chapter concludes with a brief overview of the office-​market analysis
process. Chapter 15 elaborates on the macroeconomic aspects of office market analysis and, specifically,
on econometric and less sophisticated techniques for analyzing and forecasting broader office market
behavior. Chapter  16 focuses on the microeconomic analysis of office projects. More specifically, this
chapter elaborates on location and site attributes that are relevant in the case of office developments and
discusses methodologies for estimating office project rent and absorption schedules. Finally, Chapter 17
synthesizes elements of macro and micro analysis for office projects in a simple numerical example, and
outlines broader evaluation criteria for office market studies.

Part F: Industrial market analysis

Part F, which consists of Chapter 18, focuses on industrial market analysis. Broader discussion topics include
idiosyncrasies of the industrial market, components of industrial space demand and supply, and a brief
overview of the industrial market-​analysis process. Macro-​analysis topics discussed include the assessment
of an area’s industrial development potential, and estimation of the industrial demand-​supply gap through
accounting techniques and econometric modeling approaches. Different versions of the latter are discussed

16
 Market analysis in perspective

depending on whether the emphasis is on the speculative or non-​speculative segments of the industrial
market.
The discussion of micro issues focuses on the site-​specific analysis of industrial development projects,
and specifically on the evaluation of site and location attributes for different types of industrial space (pro-
duction space, warehouse/​distribution and R&D), identification of the most likely tenants, analysis of com-
peting industrial projects, and the assessment of potential project/​land absorption and achievable rents.The
chapter concludes with a brief outline of evaluation criteria for industrial market studies.

Notes
1 The income approach, which focuses on the income-​producing capacity of a property, is one of the three major
approaches to value. The other two approaches include the sales comparison approach, which utilizes data on the
most current sales prices of properties comparable to the one under consideration, and the cost approach, which
focuses on the estimation of a building’s replacement cost.
2 The more advanced version of the income approach is represented by the discounted cash-​flow model in which a
property’s income-​earning capacity is again the prominent factor driving value. In this version, instead of using a
capitalization rate, which represents a required income return, a required total return (referred to as discount rate) is
used within the context of a multi-​year discounting formula. More on this later.
3 NAICS was developed under the auspices of the Office of Management and Budget (OMB), and adopted in
1997 to replace the Standard Industrial Classification (SIC) system. It was developed jointly by the U.S. Economic
Classification Policy Committee (ECPC), Statistics Canada, and Mexico’s Instituto Nacional de Estadistica y
Geografia to allow for a high level of comparability in business statistics among the North American countries.
4 There are variations of this definition in the New England States, which can be reviewed at www.census.gov/​
programs-​surveys/​metro-​micro.html. Other regions of the world have similar constructs for defining geographic  areas.
5 It is clear, however, that as the geographic footprint of the metropolitan market expands, the competitive sets can
become isolated from one another. For instance, a submarket north of an urban area may not be competing with a
submarket south of the urban area. Note that we said “may.”
6 Such synergies may exist between stores that sell complementary products, or stores selling products that are subject
to comparison shopping. Stores whose co-​location facilitates comparison shopping are those selling highly hetero-
geneous non-​standardized products.
7 In particular, empirical studies have shown that past increases in rents trigger office construction investments
(Sivitanidou and Sivitanides, 2000). This suggests that investors, acting myopically, assume that if rents are rising in
the present they will continue to do so in the future.
8 Production space, for example, is largely non-​speculative or alternatively build-​to-​suit. The market for such space is
driven by business investment decisions and industrial cycles parallel business cycles rather than cycles resembling
those of speculative markets.

References and additional readings

Barkham, R. 2013. Real Estate and the Great Financial Crisis. World Economics, 14:3, 5–​26.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988a. Defining the Market and Elements of the Study. Chap. 4 in
Real Estate Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice Hall.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988b. Real Estate Analyses that Focus on the Market. Chap. 1 in Real
Estate Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice Hall.
DiPasquale, D. and W.Wheaton. 1996a.The Operation of the Property Markets: A Macro and Micro Approach. Chap. 2
in Urban Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
DiPasquale, D. and W. Wheaton. 1996b. The Property and Capital Markets. Chap. 1 in Urban Economics and Real Estate
Markets. Englewood Cliffs, NJ: Prentice Hall.
Fanning, F. S. 2005. Market Analysis for Real Estate: Concepts and Applications in Valuation and Highest and Best Use. Chicago,
IL: Appraisal Institute.
Hatemi, A., E. Roca and A. Al-​Shayeb. 2014. How Integrated are Real Estate Markets with the World Market? Evidence
from Case-​Wise Boostrap Analysis. Economic Modelling, 37, February, 137–​142.
Ibanez, M. R. and A. Pennington-​Cross. 2013. Commercial Property Rent Dynamics in U.S. Metropolitan Areas: An
Examination of Office, Industrial, Flex and Retail Space. Journal of Real Estate Finance and Economics, 46:2, 232–​259.
Rosen, K. 1984. Toward a Model of the Office Building Sector. AREUEA Journal, 12:3, 261–​269.

17
 Introduction

Schmitz, A. and D. Brett. 2009. Understanding Real Estate Market Analysis. Chap. 1 in Real Estate Market Analysis: A
Case Study Approach. Washington, DC: Urban Land Institute.
Sivitanidou, R. 1997. Are Center Access Advantages Weakening? The Case of Office-​Commercial Markets. Journal of
Urban Economics, 42:1, 79–​97.
Sivitanidou, R. and P. Sivitanides. 1999. Office Capitalization Rates: Real Estate and Capital Market Influences. Journal
of Real Estate Finance and Economics, 18:3, 297–​322.
Sivitanidou, R. and P. Sivitanides. 2000. Does the Theory of Irreversible Investments Help Explain Movements in
Office-​Commercial Construction? Real Estate Economics, 28:4, 623–​661.
Wheaton, W. 1987. The Cyclic Behavior of the National Office Market. AREUEA Journal, 15:4, 289–​299.
Wheaton,W. and G. Nechayev. 2008.The 1998–​2005 Housing Bubble and the Current “Correction”: What’s Different
this Time? Journal of Real Estate Research, 30:1, 1–​26.
Wheaton, W. and R. Torto. 1990. An Investment Model of Demand and Supply for Industrial Real Estate. AREUEA
Journal, 18:4, 530–​547.

18
 2
REAL ESTATE ECONOMICS

Introduction 20
Real estate demand 20
Real estate demand concepts 20
Demand sensitivity to price/rent changes: price elasticity of demand 20
Impact of actual price changes versus expected price changes 22
Exogenous determinants of real estate demand 22
Measuring changes in real estate demand: absorption concepts 24
The supply of real estate 26
Real estate supply concepts 26
The long-run aggregate supply: is it relevant? 26
The short-run aggregate supply 27
New construction 27
The determination of market expectations 33
Real estate price adjustments 34
Price determination mechanism 34
Long-run versus short-run changes 35
The simple stock-flow model: an analytical tool 36
Assessment of the simple stock-flow model 39
Assessing demand-supply imbalances 40
Demand-supply interactions: market inefficiencies 40
Assessing the extent of disequilibrium: popular/simplistic measures 40
Construction minus net absorption (C–AB) 41
Is C–AB a good indicator of market strength? 41
Nominal vacancy rate (V) 43
Advanced methodologies 44
Nominal versus structural vacancy rate (V–V*) 44
Prevailing rent versus implicit equilibrium rent (R–R*) 48
Chapter summary 49

19
newgenprepdf

Introduction

Questions 50
Demand 50
Supply 50
Price adjustments 50
Questions for discussion 50
References and additional readings 52

Introduction
Real estate markets may be peculiar and idiosyncratic in a number of respects, but they still obey the basic
principles of demand and supply. This supply-​demand framework provides a basic analytical approach for
conceptualizing the workings of real estate markets. As one down-​to-​earth practitioner suggests, these
simple principles have been ignored by the real estate industry in favor of boilerplate analysis or simple
hunch and intuition (Featherstone, 1986). Hunch and intuition may be useful when they are based on a
solid understanding of how markets generate opportunities and constraints. However, such an approach
may be very misleading when it is based on a myopic interpretation of market conditions.This chapter first
reviews the fundamental concepts of demand, supply, prices, and price adjustments, then expands on how
they apply to real estate, and finally elaborates on their relevance to market analysis.

Real estate demand
The traditional economic definition of demand distinguishes among different demand concepts, such as,
effective demand, ex ante vs ex post demand, and pent-​up demand. After clarifying these concepts, we focus on
the price elasticity of demand, and elaborate on the difference between actual price effects and expected
price effects. After a discussion of the exogenous determinants of real estate demand, we conclude the
section with a review of the various absorption concepts that are commonly used to measure marginal
changes in real estate demand.

Real estate demand concepts

The demand for real estate space can be defined as the quantity of space or number of units demanded at
various prices. In this sense, it is appropriate to think of demand as a schedule as shown in Figure 2.1, rather
than as a single quantity. Figure 2.1 demonstrates the fundamental law of demand, which states that the
quantity demanded declines with price. In real estate terms, the law of demand would mean that a lower
amount of space or number of units is demanded at higher prices.
There are several nuances surrounding the demand definition. One is the concept of effective market
demand, that is, the demand that is backed up by purchasing power. There is also desired or ex-​ante demand.
This refers to the aggregate desired quantity of a good before consumers interact with the marketplace. After
interacting with the marketplace, however, realized or ex-​post demand may be different from the ex-​ante
demand for various reasons, such as supply constraints or price levels.The not-​yet-​realized demand is often
referred to as pent-​up demand.

Demand sensitivity to price/​rent changes: price elasticity of demand

An important trait of the demand curve is the sensitivity of quantity demanded to price changes. This sen-
sitivity is summarized by the concept of the price elasticity of demand εD.This is calculated as the ratio of the
percentage change in quantity demanded over the percentage change in prices. The price elasticity simply

20
 Real estate economics

(a) (b)

P P

P"
P"
P' P'

Q"Q' Q Q" Q' Q

Figure 2.1  Fundamental law of demand

shows by what percentage the quantity demanded will decrease in response to a percentage increase in
price. For example, a hypothetical estimate of the price elasticity of housing of –​0.5 would suggest that the
number of housing units demanded will decrease by 0.5% if the average price of housing increases by 1%.
In general, if the price elasticity is less than one, the demand curve is considered as inelastic. An inelastic
demand schedule implies that demand is not very sensitive to price increases, or that large price increases
induce relatively small decreases in the quantity demanded.

∆Q / Q [ percentage change in quantity demanded ]

εD = (2.1)
∆P / P [ percentage change in price ]

| εD | > 1 [demand is price elastic]

| εD | = 1 [demand is unit elastic]
| εD | < 1 [demand is price inelastic]

If the price elasticity is equal to one then demand is considered to be unit elastic, and refers to the case
in which a percentage increase in price induces exactly the same percentage decrease in the quantity
demanded. Finally, demand is considered to be elastic if its price elasticity is greater than one. An elastic
demand schedule implies that small increases in price induce large decreases in the amount of space or
number of units demanded.
The price elasticity of demand is determined by the availability of substitutes. For example, a product
with few substitutes should have a less elastic demand than a product with plenty of substitutes. Similarly,
the demand schedule for a submarket must be more price elastic than the demand schedule for the
whole metropolitan area since there are many substitutes for the former (other submarkets) but hardly any
substitutes for the latter. To better understand this argument, consider that most of the companies housed
in a metropolitan area serve the local population and businesses. Thus, while these firms can move from
one submarket to another submarket and still be able to serve their local clientele, they cannot do so if they
move to a different metropolitan area.
Why is the concept of the price elasticity of demand relevant for real estate analysis at the macro or
micro level? At the macro level, it can help gauge the impact of changes in market prices or rents on

21
 Introduction

demand and, more specifically, on the amount of space and/​or number of units demanded. At the micro
level, it can help investors and developers assess the impact of price increases on revenues.
Developers and investors would always prefer to face inelastic project demands because if prices/​rents
increase, revenues increase as well, as demand/​absorption does not decrease enough to eliminate the gains
from rent increases. In other words, if the price of real estate, P, goes up, the quantity demanded, Q, goes
down but, revenue, P*Q, still increases because the decrease in Q is considerably smaller than the increase
in P (Kau and Sirmans, 1985).

Impact of actual price changes versus expected price changes

In analyzing the effect of price changes, it is important to distinguish between actual price increases and
expected price increases. As discussed earlier, if actual prices increase, quantity demanded is impacted nega-
tively to a lesser or a greater extent, depending on the price elasticity of demand. In graphic terms, this
impact can be traced by moving along the demand curve since price, P, is an endogenous determinant of
demand (see Figure 2.1).
Are there any scenarios under which this fundamental law of demand may appear not to apply? For
example, some market analysts observing increasing housing demand during periods of rising prices may
be tempted to conclude that the law of demand is being violated. One could also make the same argument
in periods during which both office rents and absorption are increasing.
Although these phenomena appear to violate the law of demand, they are perfectly consistent with
economic theory. In the cases discussed above, increases in demand are not triggered by the actual price
increases but by the expectation of further increases in the future (assuming that no other changes that
would trigger an increase in demand are taking place in the marketplace). For example, consider a market
in which house prices rise initially due to massive immigration of households in the area and the resultant
increase in the demand for housing. These initial increases in house prices may ignite in the minds of
housing buyers expectations of further price increases in the future. Such a scenario is quite likely since
empirical studies have shown that real estate investors behave “myopically” or, in other words, tend to
extrapolate recent market developments and price movements into the future (Sivitanidou and Sivitanides,
1999). If that is the case, what will be the likely impact of these expectations for higher house prices on
single-​family housing demand? Would it be the same as the impact of actual price increases?
The answer to the above question is “no” for the following reason. In the case of the single-​family
market, while actual price increases may discourage some households from realizing their plans to buy a
house because they can no longer afford it, expectations of further price increases in the future may motivate
some other households to accelerate their decision to enter the market before prices climb at even higher
levels. Similarly, in the case of the office market, expected rent increases may motivate office firms to engage
in the so-​called “banking of office space,” that is, lease more space than they currently need for future
use. Therefore, under the assumption of reasonably behaving households and firms, expected price or rent
increases may result in an increase in demand for housing or office space, which is opposite of the effect
actual price increases would have. Such a behavior explains the phenomenon of increasing demand during
periods of increasing prices or rents. The effect of expectations for higher prices represents, therefore, a
shift of (and not movement along) the demand curve. In this fashion, expected price changes are exogenous
determinants of demand.

Exogenous determinants of real estate demand

So far, we have discussed the role of the endogenous determinants of real estate demand, that is, actual
prices and rents. As the previous discussion has indicated, however, quantity demanded does not depend
only on prices, but also on other non-​price or exogenous (as they are typically referred to) factors, that can

22
 Real estate economics

P D'
D

Figure 2.2  Demand shift

induce shifts of the demand schedule (see Figure  2.2).1 These exogenous determinants are of equal or
even greater importance to real estate analysts. Competent forecasts of these factors can be very helpful in
assessing real estate market prospects, evaluating project viability, and identifying real estate development
and investment opportunities. The exogenous drivers of the demand for real estate in the user market can
be classified into the following four categories:

• Market size (population, employment)

• Income/​wealth
• Prices of substitutes
• Expectations
• Interest rates and availability of credit

Market size variables that drive the demand for real estate include population, employment, or output,
depending on the property type under consideration. For example, in the case of housing and retail the
relevant exogenous determinant is the number of households or total employment, while in the case of
office space the most relevant market-​size variable is office employment. Wheaton and Nechayev (2008)
and Hwang and Quigley (2006) have validated the effect of total employment on housing demand through
its effect on house prices. In the case of industrial space demand, the relevant size variables include alter-
natively output, warehouse and distribution employment (Wheaton and Torto, 1990) or industrial employ-
ment (Ibanez and Pennington-​Cross, 2013).2 The effect of market size on real estate demand is positive,
that is, for the same price level and larger market size a greater quantity of real estate will be demanded in
terms of either square footage or number of units.
Income/​wealth affects directly the demand for retail and residential real estate in the sense that, keeping
prices constant, as income increases more households can afford to buy a house and a greater dollar
amount is available for retail spending. Therefore, increases in real income or wealth should be associated
with increases in the number of housing units and the square footage of retail space demanded. Wheaton
and Nechayev (2008) and Hwang and Quigley (2006) have also validated the effect of income on housing
demand through their effect on house prices.
Demand for office and industrial space may also be indirectly affected by income movements. For
example, as income increases, the demand for office services may increase to the point that local office
firms may need to hire more employees and expand their office space usage in order to accommodate this
increased demand. So eventually, income increases may lead to shifts in demand for office space through
their effect on office employment. Similarly, increased consumption of goods due to increases in income
may motivate wholesalers and retailers to increase their storage/​distribution space, thereby inducing shifts
in the demand for warehouse/​distribution space.

23
 Introduction

The price of substitutes could also induce shifts in the demand for real estate. For example, for a given
level of single-​family house prices, increases in apartment rents are likely to induce a shift of the demand
curve for single family-​housing to the right. Such a shift is likely to occur because as renting becomes more
expensive relative to owning a house, some renters may find homeownership more attractive. Similarly, in
the office market, as rents in the class A market rise, some firms may be forced to seek space in the class B
market where rents are more “affordable.” In such a case, the demand schedule for class B space will shift
to the right in order to reflect the greater amount of office space demanded in response to rent increases
in the class A market.
Finally, consumer or firm expectations may induce shifts in demand for the different types of real estate.
For example, as discussed earlier, expectations of higher prices or rents in the future may result in increases
in the number of housing units demanded or the amount of office space demanded. Similarly, growth
expectations on the part of firms may also induce shifts in the demand for commercial real estate. For
example, an office firm in a market that is growing rapidly may require a greater amount of space in antici-
pation of future expansion than an identical firm would require in a stable market that does not foresee
any expansion potential.

Measuring changes in real estate demand: absorption concepts

Given the durability of real estate, marginal shifts in demand are more important than aggregate demand
from a real estate development point of view. Real estate analysts use several proxies/​indicators of such
changes in demand, most of which are absorption measures. Properly used or misused proxies of marginal
changes in space demand include gross absorption, net absorption, and average or normal absorption.
Gross absorption is defined as the total amount of space involved in all leases signed during a particular
period. Note that physical occupancy of the space associated with a particular lease contract may take place
months after the contract is signed. Is gross absorption a good measure of marginal changes in real estate
demand? For example, if gross absorption in a market is going up does this mean that the market is healthy?
The answer is “no” because gross absorption does not account for space vacated.
In other words, gross absorption measures all leasing activity, which may simply represent movements
of tenants from one building to the other. As such, it does not really indicate anything about changes in
aggregate demand for real estate. In fact, if the space vacated were greater than the space leased, which
implies a decrease in the total amount of occupied space and possibly a weakening market, the positive
gross absorption measure would be very misleading. Therefore, real estate analysts should not pay too much
attention to gross absorption as net absorption is the appropriate measure of changes in real estate demand.
If net absorption is known, then gross absorption may be worth some consideration because it can provide
some information regarding the extent of turnover in the market.

Gross absorption (GAT)
Gross absorption is defined as the sum of all square footage (S) involved in all leases, n, signed during a particular
time period t:

n
GAt = ∑ Si (2.2)
i =1

Net absorption is defined as the change in a market’s occupied stock and is calculated using formula (2.3).
By definition, net absorption measures changes in aggregate demand for real estate and is definitely a much
better indicator than gross absorption because it accounts for vacated space. Thus, net absorption can take

24
 Real estate economics

negative values if the occupied stock of a market decreases or, in other words, if the space vacated during a
period is greater than the space leased.

Net absorption (ABT)
Net absorption is defined as the change in a market’s occupied stock during a particular time period. It can be
calculated using formula (2.3), where OS denotes occupied stock:
ABt = OSt –​ OSt–​1  (2.3)

Note that:
OSt = St (1–​Vt) and OSt–​1 = St–​1 (1–​Vt–​1)  (2.4)

where S is the market’s total stock (occupied plus vacant) and V is the vacancy rate.

Before evaluating net absorption, it is important to understand its determinants. Since it represents change
in demand, its determinants include prices/​rents, changes in market size (e.g., population, employment
etc.), changes in income/​wealth, and expectations for prices or employment growth. According to the
law of demand, prices/​rents should have a negative effect on net absorption while growth in market size,
income, and expectations of price increases or employment growth should have a positive effect.
Given the different factors that may boost net absorption this measure should be interpreted with
extreme caution. For example, increasing absorption may not necessarily reflect a rapidly growing employ-
ment base, but simply pent-​up demand, that is, demand from previous years that remained unrealized due
to supply constraints or high rents. If that is the case, developers should think twice before going ahead with
a new development project. Similarly, increasing absorption may simply be due to expectations of future
rent increases, which may induce firms to lease today more space than they currently need for future use.
In fact, if such “banking” of space is the major cause of increases in absorption during a period, subsequent
periods may see decreasing absorption, despite strong employment growth, because firms would have
already leased the space needed to accommodate additional employees.
The lesson that comes out of this discussion is that it is not sufficient to know whether net absorption is
strong or whether it is increasing or decreasing in order to accurately assess the strength of the market and
its prospects. It is more important to know why it is strong and why it is increasing or decreasing. Is it due
to changes in rents, employment, population or income growth? Or simply due to expectations?
Normal or average absorption is an estimate of the average net absorption over a historical period, usually a
long one, if the available data allow it. Some real estate analysts use average or normal absorption, but such
a measure could be extremely misleading when used for forecasting purposes. As historical data for office
and industrial net absorption show, this indicator moves along a wide spectrum of positive and negative
levels.Thus, net absorption over a given forecasting period may fluctuate a lot, depending on how its several
drivers will change.Therefore, an estimate of an average absorption over a number of years in the past is by
no means an indicator of the absorption levels that will be achieved in the years ahead.
To understand the inaccuracies that may result from using average absorption rates for developing
forecasts of future absorption rates, consider this example from the Los Angeles apartment market. Figure 2.3
depicts the six-​month new apartment absorption rates for the Los Angeles metropolitan area from the first
quarter of 2012 until the first quarter of 2017. Note that the absorption rate reported for each quarter is the
six-​month absorption rate of the new apartments that were completed during that quarter. If we estimate
the averages of these absorption rates for every eight quarters (two years) and particularly for 2012–​13,
2014–​15, and 2016–​17, we get clearly different averages and particularly, 67%, 82%, and 75%, respectively.

25
 Introduction

100%

90%

80%

70%

60%

50%

40%
2012Q1
2012Q2
2012Q3
2012Q4
2013Q1
2013Q2
2013Q3
2013Q4
2014Q1
2014Q2
2014Q3
2014Q4
2015Q1
2015Q2
2015Q3
2015Q4
2016Q1
2016Q2
2016Q3
2016Q4
2017Q1
2017Q2
2017Q3
2017Q4
Figure 2.3  Los Angeles metropolitan area: New apartment six-​month absorption rates

So if the average absorption rate of 2012–​13 was used as a basis for forecasting this rate for the following
two years (2014–​15), the analyst would underestimate the future absorption rate by 15 percentage points.
Similarly, using the average of 2014–​15 to forecast 2016–​17 would result in an overestimation of this rate
by seven percentage points.

The supply of real estate

In this section, we discuss first the different real estate supply concepts and then focus on the behavior of
new construction, which is the most important component of the supply side from a market analysis point
of view. In particular, we discuss the fundamental law of supply, the price elasticity of supply, and the various
factors that drive real estate development and investment decisions.

Real estate supply concepts

The term real estate supply refers in general to a schedule that describes the quantity of commercial space
or housing units supplied at various prices. Discussed in more detail in a subsequent section, the supply
curve is typically portrayed as an upward sloping curve reflecting the fundamental law of supply, which
states that greater quantity is supplied at higher prices. When dealing with real estate, it is useful to distin-
guish between three broader supply concepts: the long-​run aggregate supply, the short-​run aggregate supply, and
new construction. Although all three concepts are often mentioned in discussions of the supply side of real
estate markets, they are not all equally useful when it comes to producing period-​by-​period forecasts of
movements in a market’s inventory.

The long-​run aggregate supply: is it relevant?

The long-​run aggregate supply depicts the relationship between long-​run prices or rents and the total
number of units or square footage supplied over the long run (see Figure 2.4). The concept of long-​run
aggregate supply is difficult to operationalize but it is used in long-​run cross-​market analyses as well as in
theoretical studies focusing on the long-​r un behavior of real estate. A long-​run supply curve, as depicted

26
 Real estate economics

Q or S

Figure 2.4  Long-​run aggregate supply

Q or S

Figure 2.5  Short-​run aggregate supply

in Figure 2.4, would show that in the long run, the supply of real estate is not restricted by topograph-
ical, financial or time constraints. This would be in contrast to the supply curve depicted in Figure 2.5,
which shows no ability to increase the supply of real estate in the short run due to the construction lag
and other constraints.

The short-​run aggregate supply

The short-​run aggregate supply refers to a market’s total stock at a given point in time. In the short run, the
real estate stock is fixed and, therefore, the short-​run aggregate supply is represented by a vertical line, as
in Figure 2.5. The fixity of the real estate stock in the short run is due to the long time needed to plan
and develop a building and represents one of the most important behavioral characteristics of real estate in
understanding short-​run adjustments in property markets in response to demand shocks. Even if a building
permit is already in place, it could take from six months to a few years to deliver in the market a residential
or commercial building ready for occupancy. Given this construction lag, the short-​run supply of real estate
is insensitive to prices/​rent changes or, in economic terms, is completely price inelastic. For example, if a
20% increase in office rents takes place in a market tomorrow, the total office space stock will remain the
same. The rental increases will most likely lead to new construction, but this will take some time.

New construction
Because of the long life of real estate assets, new construction is by far the most important supply concept
when analyzing real estate markets. Most analysts need to forecast new construction or supply and do so
by way of the stock-​flow identity, which describes how a market’s total real estate stock, S, is determined at
any given point in time, t:

St = St–​1 (1–​d) + Ct  or  St = St–​1 (1–​d) + aPRMt–​n  (2.5)

27
 Introduction

where:
St : real estate stock at time t
d : depreciation rate
Ct : space completed at time t
PRM : space permitted at time t–​n
a : percentage of permits completed
n : time between permit issuance and project completion
In simple words, the stock-flow identity states that the real estate stock of a market at time t is equal to the
stock of the previous period, St–​1, minus the depreciated stock, dSt-​1, plus new completions during period
t, Ct. The depreciation rate, d, refers to three types of depreciation: physical, functional, and economic.

• Physical depreciation refers to the physical aging and deterioration of the building.
• Functional depreciation refers to the functional obsolescence of an existing building compared to new
buildings that provide new services or similar services more efficiently. Such efficiency advantages may
be due to better layout, design, technological infrastructure, equipment, etc.
• Economic depreciation refers to economic obsolescence due to external or environmental factors that
negatively affect the income-​earning capacity of the property.

An analysis of the depreciation of housing capital by Harding et al. (2007) over the period 1983–​2001
indicates that housing depreciates at roughly 2.5% per year without regular maintenance, while with such,
it depreciates at approximately 2% per year.
When a property becomes obsolete, the question of redevelopment may arise.The fundamental redevel-
opment rule is that the difference between the Residual Land Value (RLV) of the new building and the
RLV of the existing structure should be equal or greater than the redevelopment cost. This condition is
expressed in (2.6) first in a more general form and then more analytically:

RLVnew – RLVexisting ≥ Redevelopment Cost (2.6)

(Pnew –​ Cnew) * FARnew –​ Pexisting * FARexisting ≥ Redevelopment Cost (2.7)

Where:
P : price
C : construction cost
FAR : floor–​area ratio

As the stock-​flow identity indicates, the marginal change in a market’s stock at any period depends on the
amount of new construction less depreciation. New construction is the most critical and important supply
concept driving real estate markets and, certainly, the most important supply variable when undertaking
a market analysis study. The term new construction refers to completions—​the total square footage in all
new buildings that have been given a certificate of occupancy or passed the final inspection during the
period of analysis.
The new supply of real estate has three stages, commonly referred to as the “pipeline.” These are marked
by three project milestones:

1 . Permit: the development receives the permit or permission of the authorities based on approved plans.
2. Start or startup of construction, which usually means that “shovels are in the ground.”
3. Completion: the property is given occupancy certification.

28
 Real estate economics

In sum, from the conception of a development project to its completion there are at least three intervening
stages, each with possible “leaks.” For example, not all projects receive approval; also, not all projects that
get a building permit do actually start. Finally, not all projects that start are completed. The percentage of
permits that become starts and the percentage of starts that become completions varies under different
market or cyclical conditions. For example, in soft markets these percentages may tend to be lower, while
in tight markets they may tend to be higher. Given the considerable time that intervenes between a pro-
ject start and a project completion, new construction is reasonably forecastable in the short run by counting
projects and estimating delivery times. Data on starts and permits can be found in the Construction Report
Series, published by the U.S. Census Bureau. Data for local markets are usually followed by local transaction
service firms and developers.

The pipeline
Permits refer to the building permits issued based on approved plans. Not all permitted projects enter the
construction phase.

Starts account for the beginning of construction. They are identified by inspection records. Not all projects that
break ground are completed.

Completions A building is completed when it has a certificate of occupancy or passed final inspection under the
building permit.

The percentage of permits that become starts, and the percentage of starts that become completions, may vary
through time, depending on market conditions.

New construction follows the fundamental law of supply. All else being equal, the higher the property
prices are the larger the quantity of new space supplied in the market. Assuming a linear supply schedule,
this basic supply law is represented graphically in Figure 2.6 and described mathematically by the expres-
sion QS = c + dP.
As this figure shows, the new construction schedule is characterized by a minimum price level, Pmin,
representing the asset price threshold below which developers cannot make a reasonable profit, and hence
no space is supplied to the market. Think of Pmin as the minimum price per square foot or per unit at
which a developer can develop a commercial or residential project. When this minimum is above what is
considered affordable within a market there is an issue which is discussed in later chapters

Pmin

Q or C

Figure 2.6  New construction (completions)

29
 Introduction

The responsiveness of new construction to asset price or rent changes, whichever is measured on the
vertical axis, is called the elasticity of supply, εS. A price elasticity of new construction of 1.5, for example,
suggests that a 1% increase in property prices will induce a 1.5% increase in new construction. Thus, the
price elasticity of supply is very valuable in grasping the magnitude of the effect of changes in prices in
stimulating new construction. For example, assume that real estate analysts anticipate that a new govern-
ment policy will raise house prices by 10%. Then the elasticity figure of 1.5 would suggest that this new
policy would stimulate a 15% increase in new construction. New construction is on average very price
elastic, that is εS >1. Ihlanfeldt and Mayock (2014) report an average elasticity of housing construction of
approximately 2 in a study of 64 counties in the state of Florida. Elasticity of supply is largely determined
by the cost and availability of factors that go into the production of real estate. The more costly these
factors, the less price elastic is supply—​another topic for discussion in later chapters.

∆Q / Q [ percentage change in quantity supplied ]

εS = (2.8)
∆P / P [ percentage change in property price ]

An illustrative example of the high price elasticity of new construction can be drawn from the national
office market. Figure 2.7 depicts movements in office completions and rents over the period 1994–2014.3
A casual review of this figure suggests that huge increases in construction investment took place in the
U.S. due to sharp increases in rents and (presumably) climbing asset values during the periods 1997–​2001
and 2006–​2008.4 Additionally, construction activity during those periods was also encouraged by the avail-
ability of credit and positive investor expectations.
Office construction activity peaked in 2001, but was falling rapidly until 2004 when it bottomed out.
Fueled by low interest rates, the high availability of credit, and rapidly rising office rents and property

120,000 35
Office completions (×1000 sq. ft.)

100,000 30
Real office rents ($ / sq. ft.)

25
80,000
20
60,000
15
40,000
10

20,000 5

0 0
1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014

Office completions Real rents

Figure 2.7  Office completions and real rents

Source: CBRE Econometric Advisors

30
 Real estate economics

120,000 8%

6%
100,000
Office completions (×1000 sq. ft.)

Office employment growth (%)

4%
80,000

2%
60,000
0%

40,000
–2%

20,000 –4%

0 –6%
1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014
Office completions Office employment growth

Figure 2.8  Office completions and office employment growth

Source: CBRE Econometric Advisors

values, office construction activity was rising again in the period 2006–​2008, but this run was interrupted
by the global financial crisis in 2008/​9. Over the next three years office construction activity collapsed
from its 2008 peak—​dropping 88%. During the same period the real office rent dropped by only 14%
(from $29/​sq. ft. to $25/​sq. ft.).
Overall, during the period 1994–​2001 the volume of office construction consistently rose or fell when
real office rents were rising or falling, respectively.The only period during which real office rents and office
construction did not move clearly in parallel was the period 2012–​2014. Despite some recovery in office
rents during a fragile economic recovery, the heightened level of uncertainty on the part of real estate
investors led to caution with regard to development activity.
Although office rents represent a very important influence on office development decisions, office
employment growth—​reflective of the growth in demand for new space—​has also played an important
role in influencing developer and investor expectations and, in turn, office construction activity. For
instance, Figure 2.8 shows the relationship between the growth of office employment and office construc-
tion activity. The correlation is strong, except for periods when the market already has sufficient space to
absorb the growth in employment.
The relationship between the vacancy rate and office construction, portrayed in Figure 2.9, is not as
clear as the relationship between rents and office employment portrayed in the previous graphics, unless
the vacancy rate is lagged by one year or more. So it is clear in Figure 2.9 that the vacancy rate peaked one
or two years after office completions peaked. Vacancy rates are the result of the interaction of supply and
demand factors in the market. In other words, vacancy rate movements during a period are a function of
both changes in demand and changes in supply.

WHAT DETERMINES NEW CONSTRUCTION?  New construction is the sum of the square footage of all
individual projects completed within a period in a market. The major motivation for private real estate

31
 Introduction

120,000 25

100,000

Office completions (×1000 sq. ft.)

20

Office vacancy rate (%)

80,000
15

60,000

10
40,000

5
20,000

0 0
1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014
Office completions Vacancy rate

Figure 2.9  Office space completions and vacancy rate

Source: CBRE Econometric Advisors

development is to make money or profit. Thus, the major determinants of new construction in a market
are the factors that determine a project’s profitability and the uncertainty associated with it.The more profit
with little uncertainty, the more new construction. These factors include:

• Availability and cost of factors of production—​land, labor, materials, and capital.

• Expectations regarding future real estate demand and prices.
• Perceived market risk.

The cost of the factors of production affects negatively the amount of new space developed in a market. In
particular, the higher the cost of capital, labor, land, and building materials, the higher the cost of the pro-
ject, the smaller the profit, and the less the motivation of investors and developers to provide new space.
Thus, an increase in the cost of any of these factors should induce an upward and to the left shift of the
original construction schedule, since a smaller amount of space will be provided at the same price level.
This is illustrated in Figure 2.10, where we show the original supply schedule and the shift to the left. If
the costs of production rose, then every quantity of new real estate would be supplied only at a higher
price.The opposite is also shown: if the costs of production were to fall, the supply curve would shift down
and to the right, reflecting that for the same price, a higher quantity of real estate can be supplied in the
market. Of course, how much will be supplied eventually will depend on the interaction of supply with
the demand for real estate (reflected in pre-​leasing activity) and the other exogenous factors that may shift
the new supply schedule.
The cost of the factors required for the development of a property differ across real estate markets. For
example, labor costs in Boston maybe higher than in Los Angeles because of scarcity of labor in the former.
Similarly, land prices in denser metropolitan areas, such as Chicago or New York, are considerably higher
than land prices in less dense and smaller metropolitan areas, such as Tucson or Cleveland.

32
 Real estate economics

C
Figure 2.10  Effects of exogenous shifters on new construction

The determination of market expectations

On the revenue side, the most important inputs in assessing project viability are expectations regarding the
strength of demand and its implications for rents/prices. In theory, such expectations can be:

• Myopic, where P t +1 = Pt , if it is assumed that current demand/price levels or increases (whatever is

relevant from the investor’s or developer’s point of view) will continue into the future in the same
exact manner.
• Adaptive, where P t +1 = Pt + d , if developers and investors form their expectations by applying an adjust-
^
ment, d, to current values of the variables they are interested in based on past mistakes.
• Rational, where Pt+1 = Pt+1, if developers and investors based on all available relevant information can
predict correctly how the market responds to exogenous shocks once they occur.
(DiPasquale and Wheaton, 1996)

In reality, as many empirical studies of the real estate market suggest, real estate investors and developers
form their expectations myopically. Expectations of growth in demand, rents, and/​or prices should have a
positive effect on new construction (thereby inducing a downward shift of the supply curve), as they may
encourage developers and investors to build a greater number of housing units or square footage of com-
mercial space for the same price levels.
Every speculative investment, and especially those involving long gestation periods like real estate
development, involves some uncertainty, which the investor needs to factor in before he/​she commits
capital on a specific project. Within this context, new construction levels should also be affected by
market risk or the perceived uncertainty of a market in supporting profitable development of a particular
property type. According to conventional investment theory, investors facing higher risk require higher
returns, which in an efficient asset market should be reflected in lower prices. Since lower property
market prices would tend to discourage new development, risk and uncertainty should have a negative
effect on new construction.
Another approach to the relationship between uncertainty and new construction involves the theory
of irreversible investments. According to the “traditional” investment rule, an investment occurs when asset
price, P, is greater than or equal to investment cost or development cost, I. The “modern” investment rule
postulates that investment will occur when asset price, P, is greater than ωI, where ω is the option value
multiple and it is function of uncertainty, among others (see Dixit and Pindyck, 1994). This rule basic-
ally suggests that higher uncertainty increases the opportunity cost of investing now, thereby raising the
threshold asset price required by investors/​developers before making an irreversible investment, such as the

33
 Introduction

development of real estate. Thus, for a given asset price, an increase in risk should result in lower real estate
construction. Sivitanidou and Sivitanides (2000) tested this hypothesis in the case of office-​commercial
construction and found that the influence of uncertainty is statistically significant but weak compared to
the influences of the other determinants of new construction. In any case, one could argue that, all else
being equal, markets with more volatile economies should have less construction.
The concept of market risk is helpful in understanding part of the variations in construction levels
either across markets or within the same market through time. In cross-​market comparisons, this risk is
typically measured by the volatility of critical market indicators, such as demand or its major drivers, and
rents or prices. For example, an investor may evaluate the risk of an office market relative to another by
comparing the respective volatility indicators for such variables as local office employment, and/​or office
rental rates. Perceptions of risk either across markets or through time may be also shaped by indicators of
market strength. For example, a market may be considered more risky in periods during which the vacancy
rate is in the high teens compared to periods during which the vacancy rate is below 10%.

Real estate price adjustments

Rents, which are the payments for leased space, and prices, which are the payments for acquiring proper-
ties, are two of the most (if not the most) important figures that market analysis needs to determine and
provide for the assessment of the financial feasibility and viability of a development or investment project.
For this reason, it is extremely important to understand how market rents and prices are determined and
the mechanisms that drive their movements.

Price determination mechanism

As in any other market, real estate rents/​prices are determined through the interaction of supply and
demand or sellers and buyers in the marketplace. Figure  2.11 shows that the market rent/​price P* is
determined as the intersection of the demand and the supply curves. This intersection point represents the
rent/​price at which the number of willing tenants/​buyers equals the number of willing landlords/​sellers
(QD = QS). To understand why this price level should eventually prevail as the equilibrium market price,
consider the following:

• Suppose that the market price is at P1, which is below the equilibrium level P*. At this point the number
of units demanded, QD, is greater than the number of units supplied, QS. The excess demand will drive
prices up, which will force some buyers to drop out of the market and, at the same time, some additional
sellers, motivated by the higher prices, will enter the market. Once prices reach P* (at which QD = QS)
buyers will have no incentive to drive prices further up.

S
P2
P*
P1

QS QD’ Q D QS’

Figure 2.11  Market price determination

34
 Real estate economics

• If the market price is at P2, which is above the equilibrium level, the number of units demanded, QD’,
will be smaller than the number of units supplied, QS’, and sellers experiencing vacancies or low bids
will be motivated to reduce prices in order to attract tenants or buyers. Prices should continue falling
until QD’= QS’ at which point no seller will be motivated to reduce prices.

Long-​run versus short-​run changes

In reality, the market is always in flux and the simple supply and demand curves only provide a framework
for analyzing a market.To understand the basic price dynamics associated with this framework let’s look, for
example, at what will happen to the apartment market of an area that experiences significant immigration
of middle-​income households (Figure 2.12). In doing this it is extremely important for market analysis to
differentiate between the short-​and long-​run rental price impacts of such demand shifts.
A massive immigration of households will induce an upward shift of the demand curve as initially
more housing will be demanded at the same price level. The impact of this shift on prices in the short
run will be different from its long-​r un effect. In particular, house price increases or rent inflation in the
short run will be greater than in the long run. The reason is that in the short run, the existing residential
stock is fixed, as indicated by the vertical line in Figure 2.12(a), due to the construction lag or the time
it will take to build new housing. In the long run, however, developers will have the time to respond to
the increase in demand and the initial increase in prices by building new housing units. Thus, in the long
run the residential stock will rise to the level defined by the intersection of the demand curve, D', and
the long-​r un housing supply curve, S (see Figure 2.12(b)). At that point, demand will equal supply and
investors will have no motive to provide any more units unless prices or rents change again due to another
exogenous demand shock.
Figure 2.12 shows that the extent of the long-​run price increase as a result of this demand shift is
dependent on the slope or elasticity of the supply curve. The more responsive, or in economic terms, the
more elastic the supply curve is to price increases, the smaller the long-​run price increase due to an
increase in demand. If the supply is perfectly elastic (which implies a horizontal supply curve), then the
long-​r un price effect will be zero. Thus, as indicated by Figure 2.12, the short-​run price increase should
always be greater than the long-​run price increase unless the long-​run supply of real estate is perfectly
inelastic (vertical line), which is an unlikely scenario even in a dense city, as developers can always build
up or go higher!

(a) Short run (b) Long run

S P
P

D D' D D'
Q Q

Figure 2.12  Short-​run versus long-​run price changes

35
 Introduction

The simple stock-​flow model: an analytical tool

How does the market transition from the short-​run effect described by Figure 2.12(a) to the long-​run
equilibrium state described by Figure 2.12(b)? Supply and price/​rent adjustments in real estate markets
are very slow due to the long gestation period of development projects, the long tenant search processes,
and a host of other market inefficiencies. Thus, the demand shift in the apartment market in the previous
example will bring about a series of price and supply adjustments, which may take some time before the
market achieves its new equilibrium state.
Short-​run and medium-​term changes are of primary interest from a market-​analysis perspective, but
especially so given the medium-​term planning horizon of most real estate development projects. For this
reason, it is important to trace how the market moves over time—​from (a) to (b) in Figure 2.12—and
what are the specific time paths of short-​run changes in rents, new construction, and the apartment stock in
response to the original demand shock.
The tool market analysts use to trace the short-​run time paths of price, rent, and supply adjustments is
the stock-​flow model. The fundamental, simplest premise of the stock-flow model is that at any moment
in time, rents will/​must adjust so that demand equals the existing stock. This simple principle sets the stage
for identifying and forecasting time paths for not only rents or prices but also new construction.5
To demonstrate the basics of the stock-​flow model we will trace the adjustments in rents, supply, and
demand, which would follow from an initial surge in apartment demand within a market. Since new
apartment buildings cannot be built quickly, the increase in apartment demand will motivate landlords to
raise rents in order to dampen the excess demand. In response to these rent increases, developers would
plan to supply more units. As new space is completed and added to the existing stock, this would put some
pressure on rents to fall to equate demand with the new stock. As rents and prices decline, but still remain
above their long-​run equilibrium level, residential developers continue to plan new units but at a lower
rate because of reduced profits. As the new units are completed and the stock increases, rents and prices
have to fall further and so on. Figure 2.13 shows this process in steps and Figure 2.14 portrays this process
graphically.

Demand for
apartments
increases Landlords raise Rents must fall to
rents to dampen equate demand with
excess demand the higher stock

Acting on myopic Acting on myopic

New space is
expectations, expectations, dev-
completed and
developers plan to elopers supply a
added to the
supply additional smaller amount of
existing stock
space additional space

Less space is
added to the
existing stock
and so on

Figure 2.13 The simple stock-​flow model

36
 Real estate economics

R1
R2
R3
R**
D'
Rent

R*

S*=S1=Q1 S2 =Q2 S3=Q3 S**

Note: Based on the simple stock-flow model, during each period rents
must adjust so that demand equals the existing stock.

Figure 2.14 The stock-​flow model in a demand-​supply framework

To better comprehend the price dynamics implied by the simple stock-​flow model let us review its basic
equations and the short-​run and long-​run equilibrium conditions as they apply to the housing market. As
indicated in Box 2.1, the model is described by four equations and three equilibrium conditions.

Box 2.1  Stock-flow model equations

Dt = f [POPt, It, Rt] (2.9)

St = St–​1(1–​δ) + Ct (2.10)

Ct = f [Rt–​1, ct-​1, X t-​1] (2.11)

Rt = f [POPt, It, St–​1, ct–​1, X t–​1] or Rt = f [POPt, It, St] (2.12)

37
 Introduction

Equilibrium conditions
Dt = St (2.13)

∆St = St –​ St–​1 = 0 (2.14)

∆Rt = Rt –​ Rt–​1 = 0 (2.15)

Equation (2.9) describes apartment demand as a function of population (POP), income (I), and rents
(R). Equation (2.10) is the stock-flow equation that postulates that aggregate supply, or the stock in
each period, St, equals the stock of the previous period, St-​1, minus the depreciated stock, δSt–​1, plus that
period’s completions, Ct. Equation (2.11) describes completions as a function of rents, the cost of capital
(c), and other exogenous shifters (X). Finally, Equation (2.12) describes the market rent, which according
to the equilibrium condition (2.13) must be such so that at each period demand equals supply. As such,
it is derived by incorporating (2.11) in (2.10), equating the resulting equation with demand (2.9) and
solving for the rent that clears the market. Therefore, by definition the rent equation must include all the
exogenous variables included in (2.9) and (2.10). However, since completions at time t, Ct, and, therefore,
stock, St, are determined by lagged values of rents and other exogenous factors, St can directly enter the rent
equation, instead of its exogenous determinants. It should be emphasized that the rent equation represents
a transformation of the equilibrium condition (2.13) that accounts for the behavioral equations describing
demand and completions, as well as the stock-flow identity.
The short-​run movements in demand, supply, and rents are guided by the three equilibrium conditions
(2.13–​2.15). Equilibrium condition (2.13) describes the basic premise of the model: at each point in time,
rents adjust to equate real estate demand, Dt, with supply, St.This process is described in Figure 2.14, which
depicts short-​run movements in a market’s stock and rents.
To understand this process consider that the stock and rents are originally at their equilibrium levels, S*
and R*, respectively. In period 1 it is assumed that demand shifts upwardly from D to D’ due to a positive
exogenous shock. Given the construction lag, the stock cannot respond immediately to this change.Within
this framework, the stock in the short-​run is considered fixed. Thus, immediately after the demand shift
the quantity supplied is still S* = Q1 but R* is no longer the rent level that equates demand with supply
because on the shifted demand schedule Q1 is demanded at a higher rent level R1.Therefore, during period
1 market rents have to rise to R1 in order to equate demand with supply. However, as long as the long-​run
supply curve is not perfectly inelastic (i.e., is not a vertical line) or demand is not perfectly elastic (i.e., a
horizontal line), R1 will not be the new long-​run equilibrium rent, R**; in fact, it will always be greater
than R**.
By combining (2.10) and equilibrium condition (2.14), which states that ∆St = St –​ St-​1 should equal
zero, it can be shown that for (2.14) to hold, R** must be such that completions, Ct, equal the depreciated
stock, or equivalently, Ct = δ St.6 This would ensure that the change in stock is equal to zero.Thus, since R1
is greater than R**, completions in period 2 will be greater than depreciation and the stock will increase
to Q2. As a result, market rents need to decrease further to R2 in order to equate demand to the increased
supply. However, as R2 is still above R**, the area’s apartment stock should continue to increase in subse-
quent periods until it reaches its long-​run equilibrium level, S**. As a result, rents should continue to fall
until they reach, R**, which will equate the long-​run equilibrium stock, S**, with demand. At this point,
all three equilibrium conditions are satisfied. In particular, demand equals supply, the change in stock is zero
(as by definition R** must produce Ct = δSt), and rent change, Rt –​ Rt-​1, is zero since with stable stock and
demand there is no need for rents to change.

38
 Real estate economics

Assessment of the simple stock-​flow model

Two points need to be noted with respect to the simple stock-​flow model. The first relates to the question
of how real estate cycles are generated and in what way this model can help explain such cycles.The second
relates to the question of how good is the stock-​flow model in describing the realities of the real estate
markets.
Focusing on the first issue, if there is no over-​reaction or over-​shooting on the part of supply to the
initial price/​rent increases, then a one-​time shock cannot create a cyclical pattern within the context of the
stock-​flow model. As illustrated in Figure 2.14, in such a case, the stock will be gradually rising and prices/​
rents will be gradually falling until they stabilize at their new long-​run equilibrium level.
On the other hand, cycles can be generated by either frequent demand shocks in either direction (posi-
tive or negative) or a highly price-​elastic supply. For example, repeated positive demand shocks may gen-
erate a cyclical pattern in rents and construction as both variables would rise significantly, following the
demand increases, and then as the stock increases gradually in response to the price increases, supply will
outstrip demand and prices/​rents would fall. Further, if the supply curve is very price elastic, supply could
also generate a cyclical pattern, because it triggers excessive construction in response to the initial price/​
rent increases.
The specific time-​path of rents and construction in the case that supply overreacts can be explained
through the stock-​flow model as follows:

• The initial increase in demand triggers an increase in rents and prices, but now developers and investors
over-​react, initiating an excessive amount of new space.
• As a result, the new completions push the real estate stock above its long-​run equilibrium level which,
in turn, causes rents/​prices to decline below their long-​run level in order to equate the excessive stock
with existing demand.
• As rents and prices decline below their long-​run level, construction does also. In such a situation, the
long-​run completion rate should equal the depreciation rate, or if there is excessive stock, the comple-
tion rate needs to decrease below the depreciation rate. Given the durability of real estate, this makes
sense because that is the only way the existing stock can decrease.

Thus, depending on how much prices or rents fall, construction will become zero or less than depreciation
and the existing stock will start decreasing. As the stock decreases, rents and prices will rise to equate the
new lower stock level with existing demand. The stock will continue to decrease until rents and prices
reach the level at which new construction equals depreciation and aggregate demand equals aggregate
supply. At this rent/​price level, the stock will stabilize at its long-​run equilibrium level, eliminating the need
for any further rent and price adjustments.
The second issue regarding the simple stock-​flow model relates to the question of how good this
model is in describing the workings of the different real estate markets. As discussed earlier, the simple
stock-flow model views adjustments in real estate markets as a series of short-​run price equilibria, which
are realized through successive rent and price adjustments that equalize demand with supply. While such
a proposition is elegant and simple in its conception, it may not be applicable to all real estate markets
because of varying degrees of inefficiencies that prevent full or swift market adjustments. Such a con-
ceptual construct is best applicable to markets with relatively low and rather stable vacancy rates. Such
markets are relatively efficient, as they are characterized by shorter-​term leases that do not prevent rents
and demand from adjusting quickly to new equilibrating levels, not overly long construction lags, and
information efficiencies that facilitate quick adjustment of rents to changes in market conditions. Within
this context, the simple stock-​flow model may be more successful in tracking short-​run price adjustments
in residential markets.

39
 Introduction

The simple stock-​ flow model is less applicable to markets, such as the office market, which is
characterized by a high degree of inefficiencies, and, as a result, sustain high and volatile vacancy rates.7
These markets may be in constant disequilibrium, a situation that violates the fundamental assumption of
the simple stock-​flow model that rents at each period adjust to equalize demand with the existing stock.
The following section focuses on disequilibrium concepts and elaborates on two basic questions:

• How can we assess the extent of market disequilibrium?

• How should the stock-​flow model be modified to account for such a situation?

Assessing demand-​supply imbalances

Demand-​supply interactions: market inefficiencies
Generally speaking, any market would always be in equilibrium if supply could respond to upticks or
downticks in demand instantaneously. However, this is not the case, especially when it comes to real estate.
At any point in time, the real estate market may or may not be at a demand-​supply equilibrium because of
frequent exogenous shocks and a number of inefficiencies that prevent demand, supply, and rents or prices
to adjust quickly or completely to these shocks. These market inefficiencies include:

• Lack of information: Real estate is highly heterogeneous in terms of both quality and locational attributes.
Thus, timely market and project-​specific information required for the evaluation of specific transactions
is rarely readily available and its collection is time consuming and at times rather costly. These informa-
tion inefficiencies force tenants and buyers to engage in lengthy searches and prevent quick adjustment
of demand to price changes.
• Construction lags: Construction lags that last from several months to several years, depending on property
type, prevent speedy adjustment of supply to demand and price changes.
• Long-​term leases: Long-​term leases for some property types can range from 3 to 10+ years, which pre-
vent speedy adjustment of existing or in place contract rents to changes in supply and demand, and thus
hamper timely adjustments of space consumption to changes in market rates (as reflected in the latest
lease transactions). Only rents or rates associated with new lease transactions change in the market.

These inefficiencies characterize all property types, but at varying degrees. For example, information inef-
ficiencies are more severe in the retail and apartment market, construction lags are longer in the office and
retail market, and lease contracts are much shorter in the case of apartments, with hotels having the shortest
leases… one night!

Assessing the extent of disequilibrium: popular/​simplistic measures

The term disequilibrium refers to supply-​demand imbalances, or alternatively, excess demand or excess
supply. Within the context of the conventional definition of market equilibrium, the market is truly over-
supplied when rents or prices are declining and truly undersupplied when rents or prices are rising. There
are two popular and simplistic approaches often used for assessing the extent of supply-demand imbalances
characterizing real estate markets:

1 . Analysis of trends in the difference between completions, C, and net absorption, AB.

2. Analysis of trends in the nominal vacancy rate.

40
 Real estate economics

Construction minus net absorption (C–​AB)

The difference between completions C (additions to existing supply) and net absorption AB (the change in
occupied stock) reflects the change in a market’s vacant stock, on the assumption that depreciation is ignored
or very small. Thus, if C–AB is positive it implies that completions exceed net absorption and the market’s
vacant stock is rising. If it is negative it implies that net absorption exceeds completions and the market’s
vacant stock is declining.

Is C–​AB a good indicator of market strength?

Looking at some actual data on completions, absorption, and vacant stock for Los Angeles, we see that
the differences in C–AB portrayed in Figure 2.15 correlate well with the amount of vacant office space
shown in Figure 2.16. However, C–AB is not a very good measure of disequilibrium. First of all, it provides
indications about trends in vacant stock levels and not necessarily about vacancy rates, which are more
relevant (but not adequate) in evaluating the tightness of the market. Second, it does not set the stage for
looking at changes in rents, as the traditional rent adjustment model incorporates the vacancy rate and
not the vacant stock. Finally, this measure, or any measure, looked at in isolation, could be misleading. For
example, the vacant stock may be decreasing due to strong absorption, AB, which, in turn, may be pri-
marily due to decreasing rents.

6,000

5,000
Completion / absorption (×1000 sq. ft.)

4,000

3,000

2,000

1,000

–1,000

–2,000

–3,000

–4,000
1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014

Completions Absorption

Figure 2.15  Los Angeles MSA: Completions and net absorption

Source: CBRE Econometric Advisors

41
 Introduction

40,000

35,000

Vacant stock (×1000 sq. ft.)

30,000

25,000

20,000

15,000

10,000

5,000

0
1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014
Figure 2.16  Los Angeles MSA: Vacant office space stock
Source: CBRE Econometric Advisors

What does C–​AB really represent?

Ct –​ ABt  = Ct –​ (OSt –​ OSt–​1)                 [As ABt = (OSt –​ OSt–​1)]

[As OSt = (1–​Vt)St and
        = Ct –​ [(1–​Vt)St –​ (1–​Vt-​1)St-​1]               
OS t–​1 = (1–​Vt–​1)St–​1]

= Vt St –​ Vt–​1St–​1 + Ct –​ St + St–​1

= [Vt St –​Vt–​1St–​1] + dSt–​1    As, from the stock-​flow identity,

St = St–​1(1–​d)  + Ct

=>  Ct –​ St = –​St–​1(1–​d)

Ct –​ St +St–​1 = dSt–​1
Therefore:
Ct –​ ABt = [Vt St –​ Vt–​1St–​1] + dSt–​1 (2.16)

So, if dSt–​1 is small, construction minus net absorption simply represents changes in a market’s vacant stock.

If C–​AB is positive, the market’s vacant stock is rising

If C–​AB is negative, the market’s vacant stock is declining

42
 Real estate economics

Notation:
C : new construction
AB : net absorption
V : vacancy rate
S : stock (occupied plus vacant)
OS : occupied stock
d : depreciation rate

Nominal vacancy rate (V)

Real estate professionals often assess the strength of a real estate market by watching the trend in the nom-
inal vacancy rate (the ratio of the vacant stock over the total stock). However, this methodology is again
unclear because it does not provide a measure of the extent of disequilibrium. Solely looking at movements
in the vacancy rate (outside of the context of the structural vacancy rate theory, see below) cannot provide
sufficient information for making inferences about the future movements in rents.
Figure 2.17 portrays the historical vacancy rate movements and real rent changes in the Los Angeles
office market. From perusing this graphic, it is clear that decreasing vacancy rates do not necessarily imply
increasing rents, or vice versa. For example, in 2001 the office vacancy rate increased from 13% to 15.1%,
yet real rents rose considerably by 8.1%. Furthermore, in 2003 and 2004, the office vacancy rate was
declining, yet real rents kept decreasing.

25

20

15
Real rent change (%)
Vacancy rate (%)

10

–5

–10

–15
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014

Vacancy rate Real rent change

Figure 2.17  Los Angeles MSA: Real rent change and office vacancy rate
Source: CBRE Econometric Advisors

43
 Introduction

Advanced methodologies
Nominal versus structural vacancy rate (V–V*)
A conceptually more sound methodology for assessing the extent of market disequilibrium is not to view
vacancy rates in isolation, but to look at changes in the nominal vacancy rate, Vt, compared to the market’s
structural vacancy rate, V*. But what is the structural vacancy rate?
A number of analysts (Rosen and Smith, 1983) have argued that the structural vacancy rate is much like
the natural or frictional unemployment rate, which is defined as the minimum required rate to satisfy the
search needs of employers and job seekers. In a similar way, the structural vacancy rate can be thought of as
that portion of the stock of real estate that needs to remain vacant for two reasons. First, to satisfy landlords’
profit-​maximizing objectives that dictate that the marginal benefit of leaving a unit vacant is greater or
equal to its marginal cost. The marginal benefit of leaving a unit vacant is the potential rent increase that
may take place during the period the unit remains vacant, while the marginal cost is the forgone market
rent plus the interest that would be earned from this forgone rent. The second reason for which it would
be desirable to have a portion of the existing stock vacant is to facilitate tenant search. At any point in time,
there are firms or households looking for space to rent and their need to search the market can be best
accommodated if some of the stock is vacant.
And why is it important to look at the deviation of a market’s nominal vacancy rate from this rate?
It is imperative to look at the difference between the nominal and the structural vacancy rate because it
can reveal the extent and nature of the imbalance (oversupply or supply shortage) in the market, which in
turn will provide clues about the direction in which real estate rents will be moving in the future.8 The
expanded stock-​flow model provides an appropriate setting for illustrating the relationship between the
structural vacancy rate and rent changes in commercial real estate markets.

THE STOCK-​FLOW MODEL WITH VACANCY  The role of the structural vacancy rate in the functioning
of the real estate market can be better understood within the context of the stock-​flow model. The
simple model discussed earlier was not appropriate in describing the disequilibrium dynamics of real estate
markets, since by definition demand equals supply during each period of analysis. Nevertheless, it can easily
be modified to account for market disequilibrium by simply incorporating the vacancy rate in the model.
Figure 2.18 traces the effects of an exogenous increase in demand within the framework of an expanded
stock-​flow model. Given the fixed nature of the stock of real estate in the short run, the immediate effect
of this increase in demand will be a decrease in the vacancy rate.

SPACE RENTS
DEMANDED

VACANCY
RATE

STOCK COMPLETIONS

Figure 2.18 The stock-​flow model with vacancy

44
 Real estate economics

Vt
Rt

V*

vt <v* vt >v*

vt is decreasing vt is increasing vt is increasing v t is decreasing

below v*; below v*; above v*; above v*;
rents must be rents must still be rents must be rents must still be
increasing at an increasing, but at a declining at an decliniing, but at a
increasing rate decreasing rate increasing rate decreasing rate

Figure 2.19  Rent-​vacancy adjustments and the structural vacancy rate

However, the decrease in the vacancy rate will trigger an increase in rents only if the vacancy rate, Vt, is
below the market’s structural vacancy rate, V*, while rents will decrease if Vt is above V*. For example, let’s
consider two office markets A and B, where the nominal vacancy rate decreases to 12% and 8%, respect-
ively, as a result of office employment growth and positive net absorption. To demonstrate how the diffe-
rence between the nominal vacancy rate and the structural vacancy rate V* can help assess the degree of
disequilibrium, let’s assume that the structural vacancy rate in both markets is 10%.9 In that case [VA–​V*]
would be equal to 2% in market A, indicating that it is oversupplied and that a decrease in rents should be
expected. In the case of market B, [VB–​V*] would be equal to –​2%, indicating that there is excess demand
and that an increase in rents should be expected.Thus, although the vacancy rate decreased in both markets,
rents are likely to move in different directions because of the different position of each area’s vacancy rate
relative to the structural vacancy rate.
The dynamics of the rent-​vacancy adjustments within the context of the structural vacancy concept can
be better understood using a simplified graphic representation of the rent-​vacancy cycle as it is presented
in Figure 2.19. To begin with, let’s assume that the market is at its structural or equilibrium vacancy rate,
V*, and that demand increases as a result of a positive exogenous economic shock.

• Once demand increases, the vacancy rate, Vt, will start decreasing below its equilibrium level V*.
• As the vacancy rate, Vt, decreases below its equilibrium level, V*, rents start increasing at an increasing
rate.
• New construction responds to these rent increases and as a result the vacancy rate reaches a minimum
and starts increasing.

45
 Introduction

• As the vacancy rate, Vt, is increasing but is still below its equilibrium level, V*, rents will continue to increase
but at a slower rate. Rents reach their maximum level as the nominal vacancy rate rises back to its
equilibrium level.
• As the vacancy rate, Vt, continues to increase above its equilibrium level,V*—​most likely because of over-​
shooting of construction—​rents are declining at an increasing rate.
• As rents start declining, construction starts declining too. As a result the vacancy rate, Vt, reaches a max-
imum and starts decreasing but as long as it is still above its equilibrium level, V*, rents continue to decrease,
but at a decreasing rate.

This vacancy-​rent adjustment process is described by the traditional rent adjustment equation below:

∆R = a (V*–​Vt) (2.17)

where
∆R : percentage change in rental rates
a : speed of adjustment
Vt : nominal vacancy rate at time t
V* : equilibrium or structural vacancy rate

ANALYSIS OF THE STRUCTURAL VACANCY RATE  Does the natural vacancy rate remain constant or vary
through time? The answer to this question depends on the nature of the determinants of the structural or
equilibrium vacancy rate. If these determinants tend to vary considerably through time then the structural
or equilibrium vacancy rate should vary as well. The research literature postulates that the equilibrium
vacancy rate is determined by landlord and tenant search processes.Three studies (Wheaton and Torto, 1988;
Sivitanides, 1997, Hagen and Hansen, 2010) have presented evidence consistent with an intertemporally
variable natural vacancy rate.

LANDLORD’S PERSPECTIVE  From a landlord’s point of view, willingness to hold units vacant or expedite
lease-​up time should depend on two factors:

1. Expectations for demand and rental growth. Expectations of strong demand and rent growth in the future
would motivate landlords to hold vacant space in order to take advantage of the anticipated rent
increases. Such a behavior should contribute to a greater amount of vacant space in the market and,
therefore, to a higher equilibrium vacancy rate. Thus, the effect of such expectations on the structural
vacancy rate should be positive. Expectations for demand and rent growth should be driven by such
market strength indicators as recent absorption, completions, and/​or employment growth rates (see
Sivitanides, 1997).
2. Costs of holding vacant space. These include the forgone rental income, R*V, that would be earned if
the vacant units, V, were leased at the market rent, R, and the interest, i*R, that would be earned on
that income (Hendershott and Haurin, 1988).Thus, during periods of higher market rents and interest
rates, landlord’s costs of holding vacant space should be higher. Therefore, all else being equal, during
such periods the structural vacancy rate should be lower as landlords should be motivated to hold less
vacant space.

TENANT’S PERSPECTIVE  The structural vacancy rate is affected by two major features of tenant search
processes that may very well differ through time and across markets.These features and the factors that may
influence them can be summarized as follows (Sivitanidou, 2002):

46
 Real estate economics

1. The length of the search. Lengthier tenant searches should prolong average lease-​up time in the market,
thereby contributing to a higher structural vacancy rate. The length of tenant search depends on a
number of factors including:
• Idiosyncratic tastes/​space requirements. Tenants with idiosyncratic tastes/​space requirements may on
average engage in lengthier search, as a random visit to a unit is less likely to produce a match. In
this sense, markets with a more diverse or heterogeneous tenant base may be characterized by a higher
structural vacancy rate.
• Stock heterogeneity. The more heterogeneous the existing stock for a property type is in terms of
quality, the less likely it is that a random visit by a tenant will produce a match. Thus, this factor
should contribute to longer tenant search and higher structural vacancy rate.
• Spatial heterogeneity. Given that tenants/​buyers are not indifferent to the locational attributes of a
property considered for purchase or leasing, spatial heterogeneity should have the same positive
effect on the length of the search, and therefore, on the structural vacancy rate, as the stock het-
erogeneity. Spatial heterogeneity should be greater in more dispersed markets.
• Information inefficiencies. All else being equal, the more the information available on the properties
and the space available for renting or sale, the shorter a tenant’s search might be and the lower
the market’s structural vacancy rate. For example, the larger and most popular investment markets
have greater informational efficiencies than smaller ones that are rarely considered by institutional
investors.

Structural vacancy influences

Effect
Landlord’s perspective
Expectations for demand and rental growth +
Costs of holding vacant space -​
    Forgone market rent (R) -​
    Forgone interest earnings from forgone rent (i*R) -​
Tenant’s perspective
Length of the search
    Idiosyncratic tenant tastes/​tenant diversity +
   Stock heterogeneity +
   Spatial heterogeneity/​ dispersion +
   Information inefficiencies +
Cost of the search -​

2. The cost of the search. Search costs should be adversely related to the length of the search, and the struc-
tural vacancy rate, as they would motivate tenants to end their search sooner than later. In this sense,
markets or time periods with higher search costs should be characterized by a lower structural vacancy
rate. Stock heterogeneity and spatial heterogeneity/​dispersion should also contribute to higher search
costs per visit due to greater information needs, transportation costs, etc.

47
 Introduction

Prevailing rent versus implicit equilibrium rent (R–R*)

Another way to think about adjustments in the real estate market is by considering that real rents at any
point in time are moving towards an implicit equilibrium rent, R*, rather than the nominal vacancy rate
moving towards the structural vacancy rate V* (Wheaton and Torto, 1994). This proposition does not
contradict the concept of the natural or structural vacancy rate, since for any equilibrium value of rental
rates there is a corresponding equilibrium value for the vacancy rate.10
According to Wheaton and Torto (1994), the implicit equilibrium rent, R*, is determined by lease-​up
time, which in turn should be driven by tenant flows as reflected in space absorption, A, and the market
vacancy rate, V. Given the sluggishness by which real estate markets respond to exogenous shocks, the
rental adjustment process towards such an equilibrium rent is described by (2.18), where α represents the
speed by which prevailing rents move towards the equilibrium rent.

Rt –​ Rt-​1 = α [R*(At–​1,Vt–​1) –​ Rt–​1] (2.18)

The equilibrium rent should be higher with higher tenant flows and lower vacancies because they reduce
lease-​up time and motivate landlords to require higher rents. If historical data are available for observed
rental rates, absorption, and vacancy rates, the parameters of equation (2.18) can be estimated, thereby
allowing the calculation of R*.The estimated R* can provide valuable insights with respect to the disequi-
librium state of the market. For example, if the difference R–​R* is positive and of considerable magnitude,
it would suggest that prevailing rents are considerably above their equilibrium value and that considerable
decreases are very likely to take place in the future in order to restore equilibrium. On the contrary, if
R–​R* is negative it would signify a tight market in which prevailing rents would need to rise in order to
reach their implicit equilibrium values (Sivitanidou, 2002).
Drawing from Sivitanidou (2002), and Wheaton and Torto (1994), the influences on a market’s implicit
equilibrium rent can be identified within the context of their direct or indirect effect on lease-​up time.
When considering indirect influences on lease-​up time, the focus is on factors that may have an effect
on the matching rate between properties and searching tenants, as well as search costs. The more specific
influences on a market’s implicit equilibrium rent are summarized below:

1. Demand growth. For a given vacancy rate, stronger demand growth, due perhaps to stronger employ-
ment, income, population, or output growth should reduce lease-​up time and exert upward pressures
on a market’s equilibrium rent.
2. Idiosyncratic space requirements/​ Tenant diversity. All else equal, idiosyncratic tenant tastes/​ space
requirements may contribute to a lower matching rate and lower realized demand, as a random visit
to a unit is less likely to produce a match. Thus, greater tenant diversity should contribute to a lower
equilibrium rent. Sivitanidou (2002) presents empirical evidence that is consistent with a negative
effect of office tenant diversity on equilibrium rents.
3. Stock heterogeneity. The more heterogeneous a market’s existing stock is in terms of quality, the lower
the matching rate and the longer the lease-​up time, since a random visit by a tenant is less likely to
produce a match. At the same time, however, higher stock heterogeneity may increase the cost of
search (due to greater information requirements) and motivate tenants to be less demanding, thereby
facilitating the matching process. Because of the opposing influences of this factor on the matching
rate, its effect on the equilibrium rent could be either positive or negative, depending on which of the
two influences prevails.
4. Spatial heterogeneity. Spatial heterogeneity, reflected in the number and diversity of an area’s submarkets,
should have similar opposing influences on implicit equilibrium rents as stock heterogeneity.Thus, the
direction of its effect cannot be inferred a priori. Empirical evidence presented by Sivitanidou (2002)

48
 Real estate economics

is consistent with a positive effect of this factor on the equilibrium rent, suggesting that its effect on
search costs is greater than its direct effect on the matching rate.
5. Information inefficiencies. On one hand, all else being equal, greater information inefficiencies may
contribute to higher search costs, thereby motivating tenants to be less demanding and facilitating
the matching process. On the other hand, information inefficiencies may render search processes
by tenants and landlords less efficient, thereby contributing to a lower matching rate. Given these
opposing effects of information inefficiencies on the matching rate, its effect on a market’s implicit
equilibrium rent cannot be determined a priori.

Equilibrium rent influences

Effect
Expectations for demand growth +
Idiosyncratic tenant tastes/​tenant diversity -​
Stock heterogeneity -​/​+
Spatial heterogeneity/​dispersion -​/​+
Information inefficiencies -​/​+

Chapter summary
In this chapter, we examined the basic economic concepts of demand, supply, and price/​rent adjustments,
and discussed how they operate in the real estate market.We have also discussed concepts of disequilibrium
and ways of assessing the extent of disequilibrium.
The demand for real estate property or space does obey the fundamental laws of demand and its broader
drivers (shifters), which include:

• Market size (population, households, employment, or output, determined by metropolitan growth)

• Income/​wealth (determined by metropolitan growth)
• Relative prices
• Expectations regarding prices and growth.

The most important real estate supply concept from a market-​analysis point of view is new construction,
which also follows the fundamental law of supply. Its major drivers include:

• The cost and availability of production inputs (land, capital, labor, building materials)
• Expectations regarding demand/​rents/​prices
• Uncertainty and risk (volatility of the local economy and real estate market).

Real estate price adjustments are very slow due to a host of inefficiencies that stem largely from three major
factors: a) information inefficiencies, b) long-​term rental contracts that hinder swift rental and demand
adjustments, and c) long construction lags that force very slow supply adjustments. We introduced the
stock-​flow model that includes vacancies to understand and simulate short-​run movements of real estate
markets in response to exogenous shocks.
Due to frequent exogenous demand shocks and slow adjustment processes, real estate markets are more
than often in disequilibrium. Within this context, when analyzing real estate markets at a given point in
time, it is important to assess the degree of disequilibrium that may be prevailing in the market and its

49
 Introduction

implications regarding future movements in rents and vacancy rates. Simplistic methodologies used to gain
insights with respect to a market’s state of disequilibrium include examination of the difference between
construction and absorption, as well as analysis of vacancy rate trends, while more advanced methodologies
include comparison of the nominal vacancy rate to the market’s structural vacancy rate and/​or comparison
of the prevailing rent to the market’s equilibrium rent.

Questions
Demand
1 . What are the major determinants of the demand for real estate?
2. What will happen to the demand for single-​family housing if wages decrease?
3. What impact, if any, did the defense cuts that hit Southern California have on office space demand
and why?
4. Is gross absorption a good measure of market demand? Explain.
5. Is net absorption a better measure of market demand than gross absorption? Explain.
6. Why caution is needed when examining net absorption trends?
7. What are the broader factors that affect net absorption and what is the direction of their effect?
8. Explain the concept of the price elasticity of demand in the case of single-​family housing.

Supply
1. Refer to the various concepts of real estate supply and discuss which ones are more relevant for market
analysis and why.
2. Provide and discuss the stock-​flow identity and explain why it is useful when analyzing real estate
markets.
3. What are the major determinants of new construction and what is their expected effect?
4. Discuss the “pipeline effect,” which is often used when referring to real estate supply.

Price adjustments
1. Explain why short-​term price increases in real estate markets in response to positive demand shocks
should be greater than long-​run increases.
2. Discuss the workings of the simple and expanded stock-​flow model.
3. Are nominal vacancy rates good measures of market disequilibrium? Explain.
4. What is the structural vacancy rate?
5. Why is it important to look at the deviation of a market’s structural rate, V*, from its nominal vacancy
rate, V?
6. Is the structural vacancy rate constant through time? Explain.
7. Discuss factors that may contribute to variations in the structural vacancy rate through time or across
markets and their expected effect.
8. Discuss the more recently developed alternative approach to explaining rent movements in real estate
markets.

Questions for discussion

Consider a developer that specializes in the development of medium-​rise office-​commercial structures
that traditionally house small, client-​oriented professionals (e.g., doctors, lawyers, architects). Desperate to

50
 Real estate economics

re-​enter the development game after the real estate recession of the 2008, this developer closely monitors
trends in a small suburban office market with plenty of land available for development. This market has
seen a tremendous growth in employment and rental rates in the period 2003–​2008. As a result of myopic
behavior, however, completions skyrocketed, and the market soon became glutted. During the early 2010s,
this market glut was evident in vacancy rates well above the area’s structural vacancy rate. Recent estimates of
the latter during 2011–​2013 suggest it has stabilized at about 10%.
A recent market report presented the following series of selected data on the state of this office market
in 2018 and 2019:

2018.1 2018.2 2019.1 2019.2

Total stock (million sq. ft.) 20.500 20.700 20.750 20.900
Gross absorption (million sq. ft.) 0.280 0.730 0.650
Nominal vacancy rate (%) 17.2% 17.6% 16.5% 15.0%

Drawing from the data above, the report asserts that:

…rising gross absorption and falling vacancies are encouraging signs… better times lie ahead…
Development [emphasis added] opportunities in this market are good… prospective developers
may want to start securing land if they want to take advantage of the reversal of vacancy trends
in early 2019…

a ) Do you agree with this report’s conclusions? Explain clearly your answer.
b) Based on the data provided, compute an alternative absorption measure. Explain differences between
this measure and gross absorption.
c) How can you explain movements in this alternative measure during the time period in question?
Refer to additional data you may need to examine.

Notes
1 The strict meaning of the term “exogenous” from an econometric point of view is discussed in Chapter 7.
2 Ibanez and Pennington-​Cross (2013) measured industrial employment in their study as the sum of employment in
two sectors, Manufacturing and Warehousing and Storage, as defined by the North American Industry Classification
System.
3 Rents are measured in inflation-​adjusted terms, what economists call “real rents” as opposed to nominal rents.
4 The completion figures presented represent the sum of completions in 31 markets covered by CBRE Econometric
Advisors.
5 DiPasquale and Wheaton, 1996.
6 From (2.9) St –​ St–​1 = Ct –​ δSt–​1 and from (2.13) at equilibrium this should equal to zero, that is, Ct –​ δSt–​1 = 0. For
the latter condition to hold Ct must be equal to δSt–​1. Notice that at equilibrium St–​1 = St and therefore Ct = δ St
should hold for the inventory to remain constant.
7 The persistence of high vacancy rates is an indication that rents or prices do not clear the market reasonably fast
and it is inconsistent with the basic assumption of the simple stock-​flow model (DiPasquale and Wheaton, 1996).
8 Although a different modeling approach (discussed in the next section) has been recently presented by Wheaton
and Torto (1994), the concept of the structural vacancy rate still remains very appealing and is widely accepted in
the academic and professional real estate community.
9 Structural vacancy rates can vary across markets and over time. More on this later in this text.
10 A number of studies on imperfect markets have shown that such an equilibrium vacancy rate should be higher than
zero.This proposition is strongly supported by a wealth of historical data on vacancy rates for commercial real estate
markets that indicate that observed vacancy rates rarely fall below 5%.

51
 Introduction

References and additional readings

DiPasquale, D. and W. Wheaton. 1996. Econometric Analysis of Office and Industrial Markets. Chap.  12 in Urban
Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
Dixit, A. K. and R. S. Pindyck. 1994. Investment under Uncertainty. Princeton, NJ: Princeton University Press.
Featherstone, J. B. 1986. Approaching Market Analysis in a New Economic Environment. Journal of Real Estate
Development, 1:1, 5–​10.
Hagen, D. and Hansen, J. 2010. Rental Housing and the Natural Vacancy Rate. Journal of Real Estate Research, 32:4,
413–​433.
Harding, J. P., S. S. Rosenthalb, and C.F. Sirmans. 2007. Depreciation of Housing Capital, Maintenance, and House Price
Inflation: Estimates from a Repeat Sales Model. Journal of Urban Economics, 61:2, 193–​217.
Hendershott, H. and R. Haurin. 1988. Adjustments in the Real Estate Market. AREUEA Journal, 16:2, 343–​353.
Hwang, M. and J. Quigley. 2006. Economic Fundamentals in Local Housing Markets: Evidence from U.S. Metropolitan
Regions. Journal of Regional Science, 46:3, 425–​453.
Ibanez, M. R. and A. Pennington-​Cross. 2013. Commercial Property Rent Dynamics in U.S. Metropolitan Areas: An
Examination of Office, Industrial, Flex and Retail Space. Journal of Real Estate Finance and Economics, 46:2, 232–​259.
Ihlanfeldt, K. and T. Mayock. 2014. Housing Bubbles and Busts:  The Role of Supply Elasticity. Land Economics,
90:1,  79–​99.
Kau, J. and C. F. Sirmans. 1985. Real Estate. New York: McGraw-​Hill.
Rosen, K. and L. Smith. 1983. The Price Adjustment Process for Rental Housing and the Natural Vacancy Rate.
American Economic Review, 73:4, 779–​786.
Sivitanides, P. 1997. The Rent Adjustment Process and the Structural Vacancy Rate in the Commercial Real Estate
Market. Journal of Real Estate Research, 13:2, 195–​209.
Sivitanidou, R. 2002. Office Rent Processes:  The Case of U.S. Metropolitan Markets. Real Estate Economics, 30:2,
317–​344.
Sivitanidou, R. and P. Sivitanides. 1999. Office Capitalization Rates: Real Estate and Capital Market Influences. Journal
of Real Estate Finance and Economics, 18:3, 297–​322.
Sivitanidou, R. and P. Sivitanides. 2000. Does the Theory of Irreversible Investments Help Explain Movements in
Office-​Commercial Construction? Real Estate Economics, 28:4, 623–​661.
Wheaton,W. and G. Nechayev. 2008.The 1998–​2005 Housing Bubble and the Current “Correction”: What’s Different
this Time? Journal of Real Estate Research, 30:1, 1–​26.
Wheaton, W. and R. Torto. 1988.Vacancy Rates and the Future of Office Rents. AREUEA Journal, 16:4, 430–​436.
Wheaton,W. and R.Torto. 1990. An Investment Model of the Demand and Supply for Industrial Real Estate. AREUEA
Journal, 18:4, 530–​547.
Wheaton, W. and R. Torto. 1994. Office Rent Indices and their Behavior through Time. Journal of Urban Economics,
35:2, 121–​139.

52
 PART B

Metropolitan growth analysis

newgenprepdf

3
METROPOLITAN GROWTH
PATTERNS

Introduction 55
Growth analysis in perspective 56
Defining metropolitan growth 56
Growth mechanisms and implications for real estate 58
Demand-induced growth: causes and long-run consequences 59
Growth stimuli 59
Mechanism and long-run consequences 60
Real estate market impacts 62
Supply-induced growth: causes and long-run consequences 64
Growth stimuli 64
Mechanism and long-run consequences 64
Real estate market impacts 65
Education, technology, and metropolitan growth 65
Chapter summary 66
Questions 67
References and additional readings 67

Introduction
Metropolitan growth represents the most important source of increases in the (aggregate) demand
for commercial space and housing, and, as such, it merits special attention. We sometimes say that real
estate, especially commercial real estate, is “the economy in a box.” Most products are made in manu-
facturing spaces (or boxes), distributed in warehouse and retail spaces, and the services are provided
in office spaces or boxes. Simply, real estate is highly durable and, therefore, most of the demand for
new space comes from economic growth. For example, for the aggregate demand for office space to
increase in a metropolitan area, local employment and businesses need to grow so that the firms will
want to lease additional space in order to house the additional employees and equipment required for
producing the additional output. For this reason, metropolitan growth analysis leads the macro analysis
of specific property markets (office, industrial, retail, etc.) and site-​specific analysis.1 Within this context,

55
 Metropolitan growth analysis

this part of the textbook elaborates on metropolitan growth concepts and analytical processes. In par-
ticular, this chapter places metropolitan growth analysis into perspective and discusses the different
patterns of metropolitan growth, their causes, their mechanisms, and their implications for real estate
market analysis. Chapter 4 presents an example of a broad-​brush metropolitan-​g rowth analysis process
for a specific metropolitan area.

Growth analysis in perspective

Although national economic forces may affect to some extent most of the nation’s metropolitan areas,
each metropolitan economy has its own distinctive structure and pattern of growth. A good example is
the Riverside metropolitan area on the west coast of the U.S., which during the 1991 recession registered
positive employment growth while employment in the neighboring Los Angeles metropolitan area and
the nation was declining.
Whether and how exactly a metropolitan market grows is a very important question to address because
it has very critical implications regarding the strength of demand for additional real estate and the property
types that will be demanded in the area. For this reason, metropolitan growth analysis is the first important
component of a well-​integrated market study. This stage of analysis is used to address questions regarding
broader demand prospects within or across metropolitan areas, such as:

1. What are the longer-​

term demand prospects for product or property type X in metropolitan
area A?
2. Will metropolitan area A be more conducive to the development of product or property type X than
metropolitan area B?

Both of these and other related questions require analysis of the growth prospects of the overall metro-
politan economy, which is the first step in the three-​stage market-​analysis process. Metropolitan growth
analysis provides very important inputs into the second stage, that is, the macroeconomic analysis of
specific property types. In particular, the conclusions of metropolitan growth analysis will be very cru-
cial in determining the demand prospects for a specific property type during this stage. The macroeco-
nomic analysis of specific property types will provide, in turn, inputs into the site-​and project-​specific
analysis that focuses on location, project rents, and project absorption. This final stage of market analysis
will supply critical inputs that are necessary for evaluating the financial feasibility of the project under
consideration.

Defining metropolitan growth
What do we mean by growth? Metropolitan growth is a marginal concept and refers to three types of
changes in an economy:

• Physical growth
• Wealth growth
• Output growth.

Physical growth refers to population growth, employment growth, and growth in an area’s capital
stock; wealth growth denotes increases in metropolitan income and/​or wages; and output growth
refers to increases in the volume of goods and services produced within the area. Any of these changes

56
 Metropolitan growth patterns

will most likely generate new demand for real estate. For example, population and employment
growth will most likely generate additional demand for housing and commercial real estate (retail,
office, and industrial space); increases in income and wealth may generate additional demand for high-​
income housing or luxury retail; and output growth may generate additional demand for warehouse
and distribution space.

Sources of location-​specific space demand

Metropolitan growth

Physical Growth
Growth in population, employment, capital stock

Wealth Growth
Growth in incomes/wages, profits

Output Growth
Growth in goods and services produced

Location shifts within metropolitan boundaries

These shifts can generate demand for space at specific locations even when net demand does not grow at the
margin. Location shifts should never be ignored, but they are not the initial focus of traditional metropolitan
growth analysis.

Depreciation or obsolescence of existing stock

Residential stock 0.5–​1% per year
Office stock 1–​2% per year
Industrial stock 2–​4% per year

Can there be demand for additional space at specific locations even with zero growth in metropolitan
employment or income? Yes! Additional demand for real estate at specific locations within a metropolitan
area can come from two other sources:

• Location shifts due to changing intra-​metropolitan location preferences or substantial rent differentials
across submarkets leading tenants to look for cheaper space.
• Replacement demand, which is generated because of the depreciation and obsolescence of existing
stock.

Location shifts within a local economy or metropolitan area do not lead to a larger pie in terms of real
estate demand. Rather, they solely represent a shift of demand to a new location and a lowering of demand
in another location. Replacement demand is important in the long run, but it is usually not a factor in

57
 Metropolitan growth analysis

short-​run analysis. Generally, since real estate is a durable asset, it has a long life period. Estimates of real
estate depreciation are sketchy. Based on some anecdotal accounts, annual depreciation rates range from
0.5%–​1% in the case of the residential stock, from 1–​2% in the case of the office space stock, and from
2–​4% in the case of the industrial stock.

Growth mechanisms and implications for real estate

To better understand the relevance of the different metropolitan growth mechanisms in terms of real
estate demand, consider an investor planning to develop high-​end housing in one of the following three
metropolitan areas:

• Metropolitan Area 1
Area 1 is experiencing, and is likely to continue experiencing, amenity-​induced immigration of an
employable population. For instance, people might migrate to an area due to its nice climate and high
quality of life.
• Metropolitan Area 2
The area is experiencing and is likely to continue experiencing immigration of non-​employable popu-
lation (retirees).
• Metropolitan Area 3
The area is experiencing and is likely to continue experiencing shifts (increases) in the demand for the
exportable goods it produces.

These three metropolitan areas are experiencing different types of growth, as a result of the dissimilar eco-
nomic processes they are undergoing. Within this context, a developer of high-​end housing might ask two
questions:

1. What are the longer-​term prospects for growth in high-​end housing demand in each of these
metropolitan areas?
2. Which of these three areas will most likely be more conducive to high-​end housing development, if
we had to prioritize or just choose one place to develop?

To answer these questions we need to understand the different types of metropolitan growth taking place
in order to gain insights as to which area is more likely to experience significant growth in income. Income
growth is a crucial driver of the additional demand for high-​end housing. Within this context, the discus-
sion of metropolitan growth focuses on the following three broader questions:

a ) What are the causes of the different types of growth?

b) What are their mechanisms and their consequences?
c) What are the implications for real estate demand?

Most of the discussion that follows is based on the three-​sector model presented by DiPasquale and
Wheaton (1996). According to this model, which decomposes the regional economy into three markets—​
labor, output, and real estate—​there are two broader types of metropolitan growth processes:

1. Demand-​induced growth, which is driven by shifts in demand for an area’s main products, goods or ser-
vices (depicted in Figure 3.1).
2. Supply-​induced growth, which is driven most often by shifts in an area’s supply of labor (depicted in
Figure 3.2).

58
 Metropolitan growth patterns

Figure 3.1  Product market: Demand-​induced growth

Figure 3.2  Labor market: Supply-​induced growth

Understanding each of these processes can help foresee an area’s path of economic growth, assess its
implications for real estate demand, and better propose appropriate real estate development strategies. In
addition, when looking at historical data, such an understanding can help sort out the causes of an area’s
growth by identifying the type of growth that has taken place. Such analysis can help, in turn, evaluate
whether such causes will continue stimulating metropolitan growth and demand for certain real estate
products/​property types in the future.

Demand-​induced growth: causes and long-​run consequences

Growth stimuli
Demand-​induced metropolitan growth stems from factors triggering positive shifts (increases) in the
demand for a metropolitan area’s product. These include the following:

• National economic influences, such as cyclical upswings of the national economy and secular trends.
Associated with changes in consumer and producer expenditure, especially on durable goods, such
national forces may have differing effects on local metropolitan markets depending on those markets’
industrial structure. For example, increases in export demand for American cars may generate a
demand-​induced growth in Detroit (still today!), but not in San Jose, while increases in demand for
computer-​related products may generate a demand-​induced growth in the latter area, but not in the
former.
• Increases in an area’s comparative advantages in production relative to other metropolitan markets/​
regions producing similar products. Prices of competitive products elsewhere, for example, may rise
due to higher production costs, in which case demand for locally produced export goods may increase.
Non-​wage production cost differences across areas may be attributable to:
• Productivity-​enhancing amenities, such as tax or other incentives, more educated labor, quality infra-
structure, warm climate, etc.
• Industrial agglomeration, specialization, and innovation. Industrial clusters may foster innovation and,
as such, induce productivity increases and production cost savings. Clustering may be especially

59
 Metropolitan growth analysis

important during the early stages of product cycles, as it may facilitate labor market economies,
economies in the sharing of common inputs, and information/​communication economies and
innovation.
• Population growth. Increases in non-​employable population induce only product demand shifts; growth in
employable population, however, may induce both product demand and labor supply shifts.
• Increases in government expenditures or shifts in the inter-​area distribution of government contracts/​
spending (e.g., defense contracts). For example, the inflow of defense money into California (during
the period 1982–​1986) and Boston (during the 1980s) were a major stimulus of demand-​induced
growth in these areas in the past. A potential cutoff of this flow of capital to these areas may generate
some demand-​induced decline with negative consequences on employment, income, and real estate
demand.

Dudensing and Barkley (2010) present evidence of the positive effect of entrepreneurship and innovation
on metropolitan employment growth rates. Their analysis also confirmed the positive effect of human
capital on changes in per capita income. The impact of human capital on income growth was also verified
by the Abel and Gabe (2011) study that found that one percentage point increase in the proportion of
residents of a metropolitan area with a college degree is associated with about a 2% increase in GDP per
capita. Bartik and Erickcek (2008) present evidence of the positive effects of investments in higher edu-
cation and health care facilities on the average earnings of local residents, and therefore, on metropolitan
area wealth.

Mechanism and long-​run consequences

The long-​run consequences of demand-​driven growth will depend on whether it will result in growth in
an area’s job base, in its wealth, or both. The key to answering this critical question is the wage elasticity of
labor supply of the area under consideration. This is explained below as we trace step by step the impact of
a shift in an area’s product demand.

• Demand for product X rises in the product market due to any of the reasons discussed earlier (see 1 in
Figure 3.3).
• Industry output rises in response (see 2 in Figure 3.3).
• Demand for workers (and investment capital) increases in order to produce the higher level of output (see 3
in Figure 3.4). Note that in Figure 3.4 the vertical axis now measures wages (W), not prices (P).
• Wages rise in order to attract workers through increases in labor force participation (LFP) rates (increase
in labor but no increase in population) or immigration (increase in population due to movements
motivated by price/​wage factors). Areas characterized by wage-​inelastic labor supply, due to labor scarcity
for example, will experience considerable growth in wages but little growth in employment (see 4 in
Figure 3.4). Hence, the metro area will experience significant wealth growth but little growth in size, or
physical growth. It should be noted that according to DiPasquale and Wheaton (1996), the elasticity
of labor supply depends heavily on the extent of migration flows into the metropolitan area. Thus, the
desirability of an area as a place to live should have a bearing on the wage elasticity of its labor supply.
For example, less desirable places due to unfavorable climate, such as Anchorage, Alaska, may have a less
wage-​elastic labor supply, since higher wages and other incentives need to be offered by firms in order
to attract workers from other more amenable metropolitan areas.

In contrast, areas with wage-​elastic labor supply, such as Los Angeles, which are characterized by an attractive
climate, other amenities, and high historical birth rates, experience significant growth in employment but little

60
 Metropolitan growth patterns

Product market
P
1

2 Q

Figure 3.3  Shift in product demand in the product market

Inelastic labor supply

W 3

4 L

Figure 3.4  Shift in labor demand with inelastic labor supply

4'

4' L

Figure 3.5  Shift in labor demand with elastic labor supply

growth in wages (see 4' in Figure 3.5). This occurs because small increases in wages suffice to bring forth
the additional labor force required for the production of the higher levels of output (Figure 3.5). As a result,
most of the growth generated in such areas is growth in size, or physical growth.
It should be emphasized that in the case of the demand-​induced growth the size and the wealth effect
move in the same direction, but their relative magnitude depends on the wage elasticity of labor supply. In the
cases described above we have seen that increases in an area’s product demand induce both wealth (wage)
and size (employment) increases, but the former are large and the latter are small if labor supply is inelastic,
while the opposite is true if labor supply is elastic.
It should be also noted that the graphs present the long-​term impacts of demand shifts on employ-
ment and wages. Initial changes in these variables may be larger, but through time, they decrease to
their long-​r un equilibrium levels due to feedback effects. For example, initial increases in wages raise
production costs and product prices, which have a negative feedback effect on the demand for local
products. Production costs will also rise due to potential increases in the cost of capital. As indicated
earlier, increases in product demand will induce increases in the demand for not only labor but also

61
 Metropolitan growth analysis

capital. Assuming an upwardly sloping supply curve for investment capital, such a demand increase
should eventually raise its cost.

Real estate market impacts

What are the implications for real estate demand in the case of demand-​induced metropolitan growth?
In the case of the area with the inelastic labor supply, where the wealth effect should be rather sizable and
the employment effect small, there should be an increase in the demand for upgraded real estate, such as
higher-end owner-occupied housing, higher-end retail space, and higher-​end (class A) office space. In the
case of the area with elastic labor supply, where the wealth effect will most likely be small and the employ-
ment effect rather large, there should be an increase in the demand for average housing (initially rental), as
well as for basic retail services and, therefore, for low and middle-​end retail space. In addition, an increase in
the demand for class B office space—​for expanding basic professional services—​and for industrial space is
very likely to occur. Are these developments a signal to build any of these property types? Not necessarily.
Answering this question requires a more elaborate demand-​supply analysis for the property type under
consideration, which is the focus of the next stage of a typical market study.
The price impacts of increases in the demand for real estate, triggered by the aforementioned increases
in an area’s employment or wealth, will depend on the price elasticity of the supply of land and/​or real estate
in the market under consideration. The price elasticity of supply will again depend on the availability and
costs of factors of production used in real estate development, such as labor, land, and capital. For example,
if the supply of real estate is inelastic due to limited land availability and growth regulations, a shift in
demand for space will induce a large increase in space rents and real estate prices and a very small increase in
the area’s real estate stock (see 5 in Figure 3.6). On the contrary, if the supply of real estate is elastic due to
the abundance of developable land and the lack of institutional constraints on its development, a shift in
demand for real estate will induce a large increase in supply, but a very small increase in rents and prices
(see 5' in Figure 3.7).

5 Q

Figure 3.6  Shift in real estate demand with inelastic property supply

5'

Figure 3.7  Shift in real estate demand with elastic property supply

62
 Metropolitan growth patterns

Box 3.1 An example of demand-​induced growth

Boston in the 1980s
Growth cause
• During the 1980s, Boston’s high-​tech industries were experiencing increasing demand for their products. The
supply of workers, however, appeared to be quite inelastic because of the area’s low historical birth rates and
its unfavorable climate. As a result, real income per worker increased much faster than employment.
Consequences

• Rapid increase in wages and slow increase in employment.

Decade Annual
Change in population 6.1% 0.59%
Change in employment 10.9% 1.01%
Change in real income per worker 27.9% 2.49%

140
130
120
110
100
90
80
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

Total Employment Population Real Income per Worker

Sources: U.S. Census Bureau, U.S. Bureau of Economic Analysis

Direct effects
Income increase > increase in employment because of rapid increase in wages
Employment increase > increase in population because high wages may have attracted commuters, singles with
no children, and students, and may have induced increases in LFP.

Indirect effects
Higher wages  production costs up  non-​competitive environment for firms
               difficult to attract new firms

Household income up    homeownership demand up

           house prices up    affordability problems  hard to attract

  demand for services up

Population      

63
 Metropolitan growth analysis

Supply-​induced growth: causes and long-​run consequences

Growth stimuli
Supply-​induced growth is most often brought about by exogenous shifts in an area’s supply of labor. Labor-​
supply shifts may be triggered by the following factors:

• Immigration of employable population due to non-​ wage factors, such as area amenities, growth
expectations, or events occurring outside the metropolitan area’s boundaries. Amenity-​induced
immigration could take place in search of a better quality of life as signified by better climate, lower
crime rates, superior environmental quality, higher quality of education, etc. California, Arizona, and
certain southern states have certainly experienced this type of immigration that has produced such
labor supply shifts.
• Shifts in labor force participation due to increases in historical birth rates or due to other non-​wage factors,
such as increased participation in the labor force by women and the elderly, motivated by cultural
developments. Increased labor force participation was a major factor in labor force growth in the U.S. in
the 1970s and a falling labor force participation rate has been a factor in the 2010s with regard to the
pace of economic growth.
• Public policy programs, such as job training programs, education, and other favorable labor-​force
oriented programs and services, may help attract workers from other areas or motivate higher labor
force participation rates.

Mechanism and long-​run consequences

The important implication for real estate is that the impact of supply-​side growth on an area’s economy
and real estate market is different from the impact of demand-​induced growth. The major effect of a shift
in the supply of labor is growth in employment and decline in wages and wealth. The relative magnitude of
such effects largely depends on the price elasticity of product and labor demands. This is explained in more
detail below.

• Labor supply rises due to any of the aforementioned reasons (see 1 in Figure  3.8). As a result, more
workers will be available for the same wage.
• Such an increased supply of workers will exert a downward pressure on wages. As a result, employment
will rise and wages will fall. Thus, the area will grow in size but will become poorer on average: employ-
ment grows slightly but real wages decline a lot (see 2 in Figure 3.8).
• In markets characterized by elastic product and labor demands (due to highly competitive environ-
ment/​industries) employment grows a lot, but real wages decline a little (see 2' in Figure 3.9).

W 1

L
2

Figure 3.8  Supply-​induced growth with inelastic labor demand

64
 Metropolitan growth patterns

2'

2' L

Figure 3.9  Supply-​induced growth with elastic labor demand

The major conclusion regarding supply-​induced growth is that size and wealth effects move in an opposite dir-
ection.The relative magnitude of physical and wealth effects depends on the structure of the area’s economy.
As in the case of the demand-​induced growth, the scenarios present the long-​term impacts of labor supply
shifts on employment and wages. Initial changes in these variables are larger but, through time, they adjust
to their long-​run equilibrium levels due to feedback effects.
For example, initial decreases in wages (brought about by the shift in labor supply) reduce production
costs, which in turn have a positive feedback effect on the product market, inducing an increase in the
demand for the area’s output. This may generate, in turn, additional demand for labor, which may raise
wages, but not enough to elevate them back to their initial levels.

Real estate market impacts

Since labor supply shifts trigger decreases in income and increases in employment, they should impact the
area’s local real estate market. In particular, they should induce increases in the demand for developable
land and average or below-​average real estate (how much below average will depend on how severe the
negative wealth effects are).
In areas with elastic product and labor demands, where labor supply increases induce large increases in
size (employment) and small decreases in wages (average), total income will most likely increase, but average
income will decrease since wages decline. The increase in the number of workers with somewhat lower
income should generate demand for additional real estate of average or below-​average quality, such as rental
housing, low-​and middle-​end retailing, and office space for basic services.
An important consideration in evaluating an area’s capacity to considerably expand its real estate stock,
and especially its housing stock, is the ability of local governments to sufficiently expand infrastructure and
public services to appropriately accommodate growth. However, even if there is no issue in this respect,
large increases in employment will not necessarily translate into large increases of the area’s average-​or
near-​average-​quality housing stock unless the supply of this type of housing is price elastic. If it is not price
elastic, then demand increases will trigger large increases in lower-​quality-​house prices and only small
increases in the area’s housing stock.
In the case of markets with inelastic product and labor demands, where labor supply growth induces
little growth in employment size and large declines in wages, there should be little physical growth
accompanied by increases in demand for lower-​quality real estate, such as low-​end residential, retail, and
office space.

Education, technology, and metropolitan growth

A number of studies have confirmed the relationship between higher education and metropolitan
growth (Moretti, 2004; Abel and Gabe, 2011; MacKinnon, 2014). Furthermore, human capital increases

65
 Metropolitan growth analysis

productivity and idea generation (Becker, 1964) and has been found to lead to increases in long-​run
economic vitality (Glaeser, 2005). Abel and Gabe (2011, p. 1079) have found that “regional knowledge
stocks related to provision of producer services and information technology are important determinants
of economic vitality.”
In an analysis of the 30 largest MSAs as of the first quarter of 2014, MacKinnon (2014) detected a
strong positive correlation between real estate returns and total MSA employment growth. However, his
analysis indicates that this effect is entirely due to the MSA-​specific part of employment growth, which
is the employment growth in excess of the industry-​mix effect. The latter is defined as the employment
growth in the MSA that each sector would register, if it grew at the national growth rate of that sector.
MacKinnon actually found a near-​zero correlation between real estate returns and MSA growth due to
the industry-​mix effect. The study also provides evidence indicating that the change in the education level,
measured as the proportion of the population with a Bachelor’s Degree or higher university education,
had a positive effect on both the overall job growth rate and the MSA-​specific growth rate over the
period 2011–​2014.
As indicated earlier, one of the major causes of demand-​induced growth is the increase in product-
ivity through specialization, clustering, and innovation. With rapidly evolving technologies and rising
global competition, the role of innovation in employment growth is becoming increasingly important. In
response to these challenges, as well as the slow rate of job growth in the aftermath of the Great Recession
that began at the end of 2008, a new mega-​trend is emerging as a “key part of the new wave of local eco-
nomic development” (Katz and Wagner, 2014, p. 4). According to Katz and Wagner (2014), this new mega-​
trend is manifested in the creation of “innovation districts,” which are characterized by the clustering of
leading-​edge anchor institutions and companies, start-​ups, business incubators, and business accelerators in
a physically compact area that is transit-​accessible and technically wired and offers a mixture of residential
and commercial uses.

Chapter summary
This chapter has discussed the different types of metropolitan growth and their broader implications for
real estate demand. In general, we can distinguish two broader types of metropolitan growth processes that
have different real estate demand implications:

• Demand-​induced growth, which is caused by increases in demand for an area’s products and induces
increases in an area’s size (output, employment, stock) and wealth.
• Supply-​induced growth, which is caused by shifts in the supply of labor and induces decreases in wealth
and increases in employment.

The question of whether size effects will be greater than wealth effects in either type of growth will
depend on the elasticity of the labor supply in the case of the demand-​induced growth and the elasticity
of product and labor demand in the case of the supply-​induced growth.
Given the direction of the size and wealth effects of the two types of growth, demand-​induced growth
is more likely to boost demand for average or high-​end real estate, while supply-​induced growth is more
likely to boost demand for below-​average or low-​end real estate.
The overall conclusion from this chapter is that when looking at the longer-​term prospects of a metro-
politan market, it is not enough to look just at population, employment, and income changes. It is also
important to try and uncover the forces that generate these changes. Are they demand forces, supply forces
or public policy? Income/​wage changes may provide the analyst with some clues as to which of these
forces may be driving metropolitan growth or decline. Understanding the sources of economic growth will
help better assess the implications for the local economy and real estate market.

66
 Metropolitan growth patterns

Questions
1 . Discuss the factors that may trigger demand-​induced growth in a metropolitan area.
2. Discuss four policies that the mayor of your city should pursue in order to stimulate economic
growth as defined in this chapter. Explain what type of growth will result from each of the proposed
policies.
3. Describe how increases in an area’s product demand will affect the local economy.
4. Explain how demand-​induced metropolitan growth will affect the local real estate market.
5. Discuss the factors that may trigger supply-​induced growth in a metropolitan area and the prospects
of the metropolitan area in which you live in experiencing this type of growth.
6. Describe how a shift in an area’s labor supply will affect the local economy.
7. Explain how supply-​induced metropolitan growth will affect the local real estate market.

Note
1 Macro analysis is usually called a top-​down approach as opposed to the bottom-​up approach or micro analysis which
starts with the site analysis and its feasibility. The philosophy of this book is that the micro analysis is dependent on
the macro analysis for the important components of a feasibility study.

References and additional readings

Abel, J. R. and T. M. Gabe. 2011. Human Capital and Economic Activity in Urban America. Regional Studies, 45:8,
1079–​1090.
Bartik, T. J. and D. Erickcek . 2008. “Eds and Meds” and Metropolitan Economic Development. Chap. 2 in Urban and
Regional Policy and its Effects. Washington, DC: Brookings Institution Press.
Becker, G. (1964). Human Capital. New York: Columbia University Press.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Analyzing the Local Economic Environment. Chap. 3 in Real
Estate Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W. Wheaton. 1996. Economic Growth and Metropolitan Real Estate Markets. Chap. 7 in Urban
Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
Dudensing, R. M. and D. L. Barkley. 2010. Competitiveness of Southern Metropolitan Areas:  The Role of New
Economy Policies. The Review of Regional Studies, 40:2, 197–​226.
Glaeser, E. (2005). Reinventing Boston: 1640–​2003. Journal of Economic Geography, 5:2, 119–​153.
Katz, B. and J. Wagner. 2014. The Rise of Innovation Districts:  A New Geography of Innovation in America. Washington,
DC: Brookings.
MacKinnon, G. 2014. Education, Technology, Jobs and Real Estate. PREA Quarterly, Summer 2014, 44–​51.
Moretti, E. 2004. Estimating the Social Return to Higher Education:  Evidence from Longitudinal and Repeated
Cross-​Sectional Data. Journal of Econometrics, 121:1/​2, 175–​212.

67
newgenprepdf

4
ANALYZING METROPOLITAN
ECONOMIES

Introduction 68
Main analysis steps 69
Identification of the spatial unit of analysis 69
Analysis of basic growth indicators 69
Analysis of population and income trends 72
Analysis of sectoral employment, income, and industry structure trends 73
Analyzing employment and income growth patterns by industry 74
Analyzing absolute and relative employment shares 80
Synthesis: growth prospects 85
Selected sources of information for historical data and forecasts 86
Chapter summary 86
Questions 86
Project #1: Analyzing metropolitan growth patterns 87
Data to be provided by instructor 87
Main tasks 87
Introduction 87
Analyzing metropolitan growth patterns 87
Growth prospects and real estate markets 88
References and additional readings 88
Appendix 4A: About indices 88

Introduction
This chapter uses the case of Los Angeles to demonstrate a “broad-​brush” analysis framework that can be
applied to the analysis of metropolitan real estate markets. It is labeled as “broad-​brush” analysis because
it is based on broadly available data, and, as such, it is limited in terms of the level of detail at which the
different issues are explored. More detailed analysis would require the collection of additional information
depending on the growth regimes discussed earlier.
What should a section focusing on metropolitan-​growth analysis include? Simple descriptions of
trends? Forecasts? Simple analysis of what’s happening in the area’s employment, population, and income

68
 Analyzing metropolitan economies

or something more? The conclusion from the discussion in the previous chapter is that simple description
of trends is utterly insufficient in shedding light on real estate demand growth prospects. To better assess
such prospects, it is necessary to understand the reasons behind the historical and forecast movements in
the major urban growth indicators.

Main analysis steps
We look at metropolitan growth analysis as a series of methodological steps. The main analysis elements
include the following:

1 . Identification of the spatial unit of analysis

2. Analysis of basic growth indicators
3. Analysis of population and income trends
4. Analysis of sectoral employment, income, and industry-​structure trends
5. Synthesis: Growth prospects.

Identification of the spatial unit of analysis

The first step of metropolitan growth analysis involves the definition of the area of analysis. Metropolitan
growth analysis focuses on economically integrated spatial units, that is, areas within which economic
activity at all locations is strongly influenced by the same broader forces. Metropolitan Statistical Areas
(MSAs) represent integrated spatial units from both an economic and a real estate point of view. MSAs are
defined by the Office of Management and Budget according to the Standards for Defining Metropolitan
and Micropolitan Statistical Areas that are published in the Federal Register and updated every 10 years.
The latest definitions are published in the 2010 Federal register at the following link: www.govinfo.gov/​
content/​pkg/​FR-​2000-​12-​27/​pdf/​00-​32997.pdf. According to these standards, an MSA typically consists
of one central city, of at least 50,000 residents, and one or more contiguous counties that are metropolitan
in character, as determined by the percentage of labor force that is non-​agricultural and the amount of
commuting between city and county. As such, it represents an integrated labor market and a unified area
in terms of economic activity. Furthermore, it does represent an integrated real estate market, since com-
mercial and residential properties located within its boundaries may compete among themselves, but not
with similar properties in non-​adjacent metropolitan areas. According to the September 2018 delineation
files there are 1,251 Metropolitan Statistical Areas in the United States. These files can be found on the
U.S. Census Bureau website and specifically at this link: www.census.gov/​programs-​surveys/​metro-​micro/​
geographies/​geographic-​reference-​files.html.

Analysis of basic growth indicators

The second stage of metropolitan growth analysis focuses on the historical movements in basic growth
indicators, such as population, employment, and income. The objective of this stage is to identify mean-
ingful patterns in the movements of these three variables and potential associations among them in
order to better understand the underlying forces driving the local economy. This requires the following
steps:

1. Collect historical data and forecasts on population, employment, and income.

Historical employment and income data can be obtained from the U.S. Government and, specifically,
the Bureau of Economic Analysis (BEA). This information has been available on an annual basis since

69
 Metropolitan growth analysis

1969. Population data can be obtained from the Census Bureau. Under normal conditions, there is a
one to two-​year lag between the latest information published and the time of the release of the data.
More specifically, the employment and income data provided by BEA include:
Employment: Total employment, total non-​agricultural employment, and employment by major
industry
Income:  Total real personal income, total employee compensation, and per capita personal
income.
2. Identify periods with similar behavioral trends.
3 . Search for patterns of association.
• Compare population and employment movements.
• Examine labor force participation (LFP) movements.
• Compare employment and income movements.
• Examine movements in wages.
4 . Analyze/​identify the underlying forces behind observed trends using supplementary data.

In order to more productively examine historical trends and patterns of association, it is better at this
stage to deflate income figures (convert them into real dollars) and normalize them on a per-​worker basis.
Graphing historical and forecast movements in real income per worker, along with employment and popu-
lation, as in Figure 4.1, can provide a better understanding of their movements through time and facilitate
a more meaningful interpretation.
When dealing with variables that are expressed in different units or differ greatly in magnitude, such as
income per worker and employment, it is much easier to plot them on the same graph by converting them
into indices. Appendix 4A provides a detailed discussion of indices and how they should be constructed. It also
elaborates on the consumer price index (CPI) and demonstrates how it can be used to deflate income figures.

230

210

190

170

150

130

110

90
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017

Populaon Real personal income/worker Employment

Figure 4.1  Los Angeles MSA: Population, employment, and real personal income/​worker

Source: BEA, U.S. Census Bureau

70
 Analyzing metropolitan economies

Box 4.1 A broad-​brush analysis framework

The underlying premise of metropolitan growth analysis is that simple description of trends is utterly insuffi-
cient in shedding light on real estate demand growth prospects. Trends in population, employment, and income
should be analyzed and the reasons behind the historical and forecast movements in such growth indicators
should be identified and explained.

1. Identification of the spatial unit of analysis

Metropolitan growth analysis focuses on economically integrated spatial units. The typical unit of analysis
is the Metropolitan Statistical Area (MSA) within which the project/​property is located.

2. Analysis of basic growth indicators

Explore broad patterns of population, employment, and income growth, and try to identify meaningful
growth patterns and associations (see Figure 4.1).

3. Analysis of population trends
What are the primary components of population growth (internal growth, net migration) and their
characteristics (e.g., demographics and skill characteristics of migrants)?

Population Identity:
Popt = Popt–​1 (1 + Birth Rate –​Death Rate) + Net Migration
 Net Migration = Popt –​ Popt–​1 (1 + Birth Rate –​Death Rate)
What are the primary forces behind positive or negative net immigration?
What role do regional incomes and regional amenities play in shaping the area’s population growth?
What are the future prospects with respect to the influence of these factors?

4. Analysis of sectoral-​employment, income, and industry-​structure trends

How have the various industry sectors been performing in terms of employment and incomes, and how
are they expected to perform?
What are the area’s key employment sectors? What has been driving and what will likely drive the per-
formance of these sectors in the years ahead?
A first-​cut analysis involves simple examination of employment and income growth patterns (Figures 4.2–​
4.9), absolute and relative employment shares (Tables 4.1 and 4.2), and employment growth components
(Table 4.3).

5. Synthesis: Growth prospects
Develop a consistent story explaining past and future growth patterns and explore the implications of such
patterns for the long-​term prospects in the marginal demand for real estate product types.

Figure 4.1 presents historical trends for population, employment, and real income per worker in the Los
Angeles-​Long Beach-​Anaheim MSA from 1969 until 2017. All three variables are presented as indices that
take the value of 100 in 1969. As the figure shows, employment grew much faster than both population and
real income per worker on a cumulative basis. In particular, over the period 1969–​2017, employment grew

71
 Metropolitan growth analysis

by a cumulative 120%, while population grew by about 60% and real income per worker by only 23%.The
considerably faster employment growth, compared to population growth, suggests an increasing number of
commuters from neighboring counties and/​or increasing labor force participation rates. Furthermore, the
fact that real income per worker increased overall at a considerably slower pace than employment suggests
demand-​induced growth and an elastic labor supply. Such a growth pattern implies that Los Angeles grew
over the period 1969–2017 considerably faster in terms of physical size than in terms of wealth.
Focusing on employment, we can discern a slower growth rate in the 1990s and up to 2007 compared
to the previous 20 years. However, in this decade, as the economy returned to recovery in 2010 after the
global financial crisis, it seems that employment growth returned to the high rates that were registered
during the period 1970–​1990. The question looking ahead is whether this fast and uninterrupted
employment growth that has been taking place from 2010 until 2017 has continued or has slowed down.
Analysis of employment and income trends by industry in the next stage of analysis may shed light on
this question.
Historical population growth patterns seem to be clearly smoother and slower than employment growth
patterns. In particular, population seems to have registered higher cumulative growth rates in the 1970s and
1980s (in excess of 10%). In the 1990s the cumulative growth rate slowed down to 10%, while the global
financial crisis cut down the cumulative population growth rate of the first 10 years of the new millennium
to only 3.6%. Population growth in the Los Angeles MSA remained slow in the first seven years of this
decade as it has registered a cumulative rate of increase of only 4%.
In order to evaluate whether these trends will continue in the future, more information needs to be
collected in terms of the different sources of population growth in the Los Angeles area in the last five years.
Such information will help distinguish whether the slowdown in population growth was due to reduction
in birth rates or a considerable reduction in net migration, perhaps triggered by the legislation imposing
further restrictions on international migration. Understanding better the sources of population growth in
the Los Angeles area in recent years will also help more accurately identify patterns of association between
population growth and employment growth during this decade and better assess the prospects of those
patterns.
The long-​term trend conveyed by Figure 4.1 regarding real income per worker on a cumulative basis
in the Los Angeles area is that it has increased at a quite lower rate compared to employment. However,
we can distinguish two clearly different trends. In the period 1970–​1990 it remained practically flat (with
actually a small decrease during the 1970s), while in the period 1990–​2010 it was rising. In particular, in
the 1990s it registered a cumulative increase of 15% while in the 2000s it registered a total increase of
10.4%. In the first seven years of the 2010–​2020 decade it has practically remained flat on a cumulative
basis, reflecting in all likelihood the impact of the global financial crisis.

Analysis of population and income trends

At this stage, the analyst needs to examine the primary components of population growth and their
characteristics. In analyzing population trends, three important questions need to be addressed:

a) Why and how is the area’s population growing?

• Immigration?
• Natural growth?

The basic population identity below can help answer this question:

Popt = Popt–​1 (1 + Birth Rate – Death Rate) + Net Migration (4.1)

72
 Analyzing metropolitan economies

If a net migration figure is not available from the state government, the planning department, or other
local government agency, which is very likely, then it can be calculated as the residual using (4.2), if
last year’s birth rates and death rates are known or can be estimated:

Net Migration = Popt –​ Popt–​1 (1 + Birth Rate –​Death Rate) (4.2)

Obviously, in applying (4.2) the analyst needs data on population, births, and deaths. Information on
population can be obtained from the American Community Survey that provides population infor-
mation for county, county subdivisions, and census tracts. The information has about two years lag
and links to the data can be found on the following webpage: www.census.gov/​programs-​surveys/​
acs. Information on population may also be available from population reports/​statistics published by
the local government or regional/​state population centers. Data on births and deaths may be available
from public records or can be estimated using fertility rates and death or survival rates.

b) What are the primary forces behind positive or negative net migration?

This is a difficult question to answer without direct data. The analyst needs to check with local and
state government agencies, local and state university research centers, or other potential sources, if
there is any existing study on this issue. Examining what role regional incomes and regional amen-
ities play in shaping population growth may provide some useful insights into the question of what
the primary drivers of immigration in the area under consideration are. Furthermore, assessing the
future prospects of these factors will help form an opinion as to whether observed immigration
trends will continue in the future. If the data allow it, it may be useful to distinguish between inter-
national and domestic migration.

c) If supply shifts are due to immigration, what are the demographics of the migrants and what are their labor skills?
Do they match local industry requirements?

Again, this is a difficult question to answer but some anecdotal evidence may exist from university
reports, newspapers, planning departments, local chambers of commerce, and government associations,
such as the Southern California Association of Governments (SCAG). In the 1980s, for example,
low-​skilled workers were migrating to Los Angeles, and this is perhaps why supply effects were not
prevalent. Thus, addressing this question can help gain a better understanding of why a certain type of
growth did or did not occur. Since market analysis is carried out in order to gain a perspective about
what may happen in the future, the analyst needs to consider throughout this analysis the likelihood
of the continuation of the most recent population growth trends.

Analysis of sectoral employment, income, and industry structure trends

This stage of analysis can help identify which sectors contribute the most to the area’s economy, and which
ones have been growing faster recently in terms of employment and earnings. The following questions
need to be addressed in particular:

a) How have the various industry sectors been performing in terms of employment and incomes and
how are they expected to perform in the future?
b) What are the area’s key employment sectors and industries?
c) What has been driving and what will likely drive the performance of these sectors in the future?

73
 Metropolitan growth analysis

Answers to these questions will help further clarify what has been driving economic expansion in the area
and shed light on the type of growth that has been taking place. A first-​cut analysis involves a simple review
of several statistics by sector, such as employment and income growth patterns (Figures 4.2–​4.9), absolute
and relative employment shares (Tables 4.1 and 4.2), and employment growth components (Table 4.3).
Below we present such a simple analysis for the Los Angeles metropolitan area.

Analyzing employment and income growth patterns by industry

In order to answer the first question we examine long-​term historical trends for employment and real
compensation per employee for the major non-​agricultural sectors of the Los Angeles metropolitan
statistical area over the period 2001–​2017. These are presented in Figures  4.2 to 4.9 for Construction,
Manufacturing, Wholesale Trade, Retail Trade, Transportation, Warehousing, Utilities and Information,
FIRE (Finance, Insurance, and Real Estate), Services, and Government. Below we review these historical
trends in each sector.

CONSTRUCTION  As shown in Figure 4.2, construction employment has been very volatile, reflecting large
swings in real estate building activity. In particular, construction employment increased by a cumulative
20% during the real estate recovery that followed the 2001 recession, which had minimal impact on the Los
Angeles construction employment.This recovery period started in 2003 and ended in 2007, which marked
the beginning of the downturn of the U.S. real estate market, which in turn was the precursor of the 2008
global financial crisis. As a result of the latter, the U.S. real estate market and construction employment
entered into a steep downward path.This steep downturn lasted until 2010, with construction employment
plunging at a level more than 10% lower than its 2001 level. Since 2011, construction employment has
been on a steep upward path and by 2017 it was only 4% below its 2006 peak.
Real compensation/​employee in the Construction sector followed a different and definitely less vola-
tile path than employment with very small gains (a cumulative 5%) over the period 2003–​2006 when
Construction employment was rising rapidly. During the downturn, real wages were declining until 2010

130

120

110

100

90

80

70
2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

Employment Real compensaon/employee

Figure 4.2  Los Angeles MSA: Trends in Construction employment and real compensation/​employee

Source: BEA

74
 Analyzing metropolitan economies

160

150

140

130

120

110

100

90

80

70
2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017
Employment Real compensaon/employee

Figure 4.3  Los Angeles MSA: Trends in FIRE employment and real compensation/​employee

Source: BEA

and then remained stable until 2013. Since 2014, real compensation/​employee has been on a rising path
at varying growth rates and by 2017 it had surpassed its 2001 level. In sum, Construction employment
registered pronounced cyclical fluctuations over the period 2001–2017 and an overall gain of 15%, while
real compensation per employee registered less pronounced cyclical fluctuations with a minimal overall gain.

FINANCE, INSURANCE, AND REAL ESTATE (FIRE)  Although FIRE employment and Construction
employment tend to move together since a significant portion of employment in real estate is associated
with the development industry, this does not seem to be the case over the period 2001–​2017. As Figure 4.3
indicates, FIRE employment in the Los Angeles area during that period registered minimal fluctuations as
it has been on a rather stable upward path since 2001 with minimal drops during the 2007–​2010 recession.
As a result of these trends, FIRE employment registered a cumulative increase of 46.5% from 2001 until
2017. Real compensation/​employee though after a small rise in 2004 and 2005 plunged into a steep
downward path, which continued until 2009. By then real compensation per employee in the FIRE sector
had dropped by more than 20% below its 2001 level. Since 2010, real wages in this sector have been on a
slightly upward path but by 2017 they were still 17% below their 2001 level. In sum, FIRE employment
continued its secular upward trend over the last 18 years posting a cumulative increase of 46%, while real
compensation per employee posted a cumulative drop of 17%, which took place in the first decade of this
millennium.

MANUFACTURING  Continuing a long-​term secular trend, Los Angeles Manufacturing employment was
on a rather steep downward path over the period 2001–​2010, during which it registered a contraction of
over 30% (Figure 4.4). After 2010, it remained practically stable until 2017. Interestingly, real compensation
per manufacturing employee registered quite different trends. In particular, after posting an increase of
about 10% in 2003 and 2004, it remained practically stable until 2014. Since then it has been rising and by
2017 it was 19% higher than its 2001 level. The increasing real compensation per manufacturing employee
may be attributed to rising demand and to productivity effects.1 In sum, over the 18-​year period of analysis,
Manufacturing employment continued its secular downward trend, posting a cumulative decrease in excess

75
 Metropolitan growth analysis

130

120

110

100

90

80

70

60

50

40
2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017
Employment Real compensaon/employee

Figure 4.4  Los Angeles MSA: Trends in Manufacturing employment and real compensation/​employee

Source: BEA

150

140

130

120

110

100

90
2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

Employment Real compensaon/employee

Figure 4.5  Los Angeles MSA: Trends in Service employment and real compensation/​employee

Source: BEA

of 30%, while real compensation per employee registered a rather stable path with two periods of rather
rapid increases that resulted in a cumulative gain of 19%.

SERVICES  With the exception of 2008 and 2009, employment in the Services sector was rising rather
rapidly since 2001 (Figure  4.5). The pace of employment growth accelerated since 2010 and by 2017
Services employment in the Los Angeles area was about 40% higher than its 2001 level. This represents
a continuation of a secular trend of rising employment in the service sector, mostly at the expense of

76
 Analyzing metropolitan economies

the manufacturing sector. It should be noted that, with the exception of the 1991 and 2008 economic
downturn, Services employment has been growing at a rapid pace even when the national and the Los
Angeles economies were in a recession. Real compensation per worker in the Services sector followed a
more stable path. Following a small increase in 2003 and 2004, it remained practically stable until 2011. In
2012–​2013 it declined just below its 2001 level and since then it has been on a slightly upward path that
brought it by 2017 to about 3% above its 2001 level. In sum, Services employment continued its secular
upward trend over the last 18 years, posting a cumulative increase of 40%, while real compensation per
employee registered very small fluctuations and a minimal cumulative gain.

WHOLESALE TRADE  After a small decline in 2002 and 2003 as a result of the 2001 recession, Wholesale
employment in the Los Angeles MSA embarked on steadily rising path until 2007 (Figure 4.6). However,
during the 2008–​2010 recession it registered a rather sharp decline of 11%. From 2011 until 2015 it returned
to rapid growth, registering a cumulative increase of 15%. In 2016 and 2017 however it was declining again,
registering a fall of 7.5% in those two years. Real compensation/​employee in the wholesale sector followed
the same pattern with employment until 2011 but with less pronounced fluctuations. Over 2012–​2014 it
turned downwards, registering a 6% drop although employment continued to rise. Since 2015 it has been
on a rising path and by 2017 it was 5% above its 2001 level. In sum, although Wholesale Trade employment
and real compensation per employee registered opposite cyclical fluctuations in the second decade of this
millennium, overall they both registered a small rise of about 5% over the period of analysis.

RETAIL TRADE  Retail Trade employment followed mostly a similar pattern to employment in Wholesale
Trade (Figure 4.7). In particular, it was rising from 2001 to 2007 and declining from 2008 to 2010 as a
result of the recession. From 2011 until 2015 it was rising at a rather stable rate, reaching 5% above its 2001
level. In 2017 it was still at that level. Real compensation per employee, after registering a small increase in
2002 and 2003, plunged into a downward path that by 2009 took it 10% below its 2001 level. Since then
it has been fluctuating slightly around that level and, per the latest data, in 2017 it was still at that level. In
sum, Retail Trade employment despite the cyclical swings registered small gains of about 5% in the 18-​year
period covered by the data, while real compensation per employee registered a 10% drop.

115

110

105

100

95

90
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017

Employment Real compensaon/employee

Figure 4.6  Los Angeles MSA: Trends in Wholesale Trade employment and real compensation/​employee
Source: BEA

77
 Metropolitan growth analysis

110

105

100

95

90

85

80
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
Employment Real compensaon/employee

Figure 4.7  Los Angeles MSA: Trends in Retail Trade employment and real compensation/​employee
Source: BEA

125

120

115

110

105

100

95

90
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017

Employment Real compensaon/employee

Figure 4.8  Los Angeles MSA: Trends in Government employment and real compensation/​employee

Source: BEA

GOVERNMENT  Employment in the Government sector includes persons employed by the federal
government (including the military) as well as by state and local governments. As shown in Figure 4.8,
employment and real compensation per employee in this sector have moved in opposite directions for
most of the period under consideration. In particular, employment, barring a few small increases that took
place in 2002 and 2007–​2008, registered a clear downward trend and a cumulative decrease of 8% over the
period 2001–​2013. However, since 2014 it has been rising steadily and by 2017 it was only 1% below its
2001 level. Real compensation per Government employee, with the exception of a few short periods of
small drops (2008–​2009 and 2012), registered a clear long-​term upward trend and a cumulative increase
of 22% over the period 2001–​2015. In 2016 and 2017 it has been on a slightly downward path. In sum,

78
 Analyzing metropolitan economies

125
120
115
110
105
100
95
90
85
80
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
Employment Real compensaon/employee

Figure 4.9  Los Angeles MSA: Trends in Transportation and Warehousing, Utilities, and Information employment and
real compensation/​employee
Source: BEA

Government employment followed the cyclical fluctuations of the economy, registering a minimal overall
decrease over the last 18 years, while real compensation per employee registered a rather consistent upward
trend with cumulative gains in excess of 20%.

TRANSPORTATION AND WAREHOUSING, UTILITIES, AND INFORMATION (TWUI)  Employment in TWUI

followed the cyclical economy with a drop of about 9% over the recessionary period of 2001–​2003 and
a small rise of about 4% during the recovery period of 2004–​2007 (Figure 4.9). During the recessionary
period of 2008–​2010 it dropped by more than 10% but since 2011 it has been on a sharply rising path and
by 2017 the aggregate employment in these sectors was up by 21.5%. Compensation per employee in these
sectors registered small fluctuations around its 2001 level over the period 2001–​2011. However, since then
and up to 2016 it embarked on a downward path, falling by more than 12% below its 2001 level. In 2017
there was a reversal of this trend. In sum, despite registering a downward trend in the 2000s, employment
rebounded strongly in the 2010s, posting a cumulative increase of 21% over the last 18 years, while real
compensation per employee posted a cumulative decrease of 9%.
In sum, in terms of historical trends, all non-​agricultural sectors but Manufacturing and Government
registered some employment gains over the 18-​year period examined, but definitely at different rates. The
Service and FIRE employment kept growing at the fastest rates despite the cyclical swings of the economy
that were best reflected in the fluctuations of employment in Construction, Wholesale Trade, and Retail
Trade. Also, employment in Transportation, Warehousing, Utilities and Information followed mostly the
fluctuations of the overall economy but it has been increasing very rapidly since 2014. FIRE and Services
registered by far the largest employment increases (in excess of 40%) over 2001–​2017 with TWUI following
as the distant third with a cumulative increase of 21%. Manufacturing registered a decrease in excess of 30%.
Long-​term trends in real compensation per worker leave little doubt about the high elasticity of labor
supply in the Los Angeles area. In particular, real wages in all sectors but Government and Manufacturing
either decreased (FIRE, Retail Trade, and TWUI) or registered small gains (Construction, Services, and
Wholesale Trade). Only real compensation/​employee in Manufacturing and Government registered con-
sistent gains totaling 19% and 21%, respectively, over the period examined.

79
 Metropolitan growth analysis

Overall, this historical analysis shows that the Los Angeles metropolitan economy registered both types
of growth of the period 2001–​2017 across the different non-​agricultural sectors. In particular, FIRE, Retail
Trade, and TWUI, in which wages were decreasing while employment was rising, fit the supply-​induced
growth pattern, while Construction, Services, and Wholesale Trade, in which wages were stable or rising
while employment was rising, fit the demand-​induced growth pattern.

Analyzing absolute and relative employment shares

Analysis of the absolute and relative employment shares of the different economic sectors can help answer
the second question pertaining to the area’s key employment sectors. Such analysis can provide very useful
insights with respect to the diversity or specialization of an economy and the major drivers of local employ-
ment growth. As such, it can help better assess the prospects of the local economy and gain a perspective
of how it may perform in the future.
Absolute employment shares refer to the simple ratio of employment in a given sector over the area’s
total employment. The most common measure of relative employment composition is the so-​called location
quotient and refers to the ratio of a sector’s absolute share over the share of the same sector within a larger
economic area, such as a region or the nation.
We first focus on absolute employment shares in order to identify which sectors contribute currently
the most to the local economy, and whether any structural changes have taken place over the longer
term. Table  4.1 presents the shares for the major nine sectors as a percentage of total non-​agricultural
employment. As this table indicates, Services, with an employment share of 48.5%, was by far the sector
with the greatest contribution to the Los Angeles economy as of 2017. FIRE, Government, and Retail
Trade followed with a considerably smaller contribution ranging between 8% and 11%. Putting aside
Forestry, Fishing, and Mining, which have minimal contribution, the sector with the smallest employment
share (4.2%) was Construction. Looking at the employment shares of the different sectors in 2001, it is
obvious that the prominent transformation that has been taking place in the Los Angeles economy (and
most of the country for that matter) is the considerable increase of the share of Services and the consid-
erable decrease of the contribution of Manufacturing. In particular, the Manufacturing employment share
decreased steadily from 11.4% in 2001 to 6.4% in 2017, while the Service employment share increased
from 41.7% in 2001 to 48.5% in 2017.
The analysis of the absolute employment shares has provided some very interesting insights with respect
to the contribution of the major industries to the area’s employment base. The analysis of relative employ-
ment shares, more typically referred to as location quotients, can shed light on how the local economy

Table 4.1  Absolute employment shares in the Los Angeles economy

Sector 2001 2005 2010 2015 2017

Forestry, Fishing, and Mining 0.2% 0.2% 0.3% 0.3% 0.3%
Construction 4.3% 4.8% 3.8% 3.9% 4.2%
Manufacturing 11.4% 9.2% 7.7% 6.7% 6.4%
Wholesale Trade 4.9% 4.9% 4.8% 4.7% 4.3%
Retail Trade 9.6% 9.7% 9.2% 8.6% 8.3%
Transportation, Warehousing, Utilities, and Information 7.9% 7.2% 6.9% 7.4% 8.0%
FIRE 9.1% 10.6% 11.3% 10.9% 11.1%
Services 41.7% 43.0% 45.6% 48.6% 48.5%
Government and Government Enterprises 10.9% 10.3% 10.5% 9.0% 9.0%
Total 100.0% 100.0% 100.0% 100.0% 100.0%

80
 Analyzing metropolitan economies

compares to a larger economy, such as the broader region within which is located (the Western region, in
the case of Los Angeles) or the nation. Such a comparison may highlight important structural differences
and help identify industries that are more or less influential to the local economy than they are to the larger
economy. The rationale behind the concept of the location quotient and its potential interpretation are
discussed in more detail in the next section.

LOCATION QUOTIENT:  ORIGINS OF THE IDEA AND DEFINITION  The idea of the location quotient
originated in the traditional economic base theory. According to this theory, an area’s economy is structured
around two sectors:

1. Export or basic sector, which comprises industries exporting their products outside of metropolitan
boundaries
2. Local or non-​basic sector, which comprises industries supporting local demand

The prevalent thought was that the export sector is vital for the growth of the local economy in the sense
that increases in exports trigger increases in basic employment, which in turn stimulate increases in the
demand for local goods and increases in local-​sector employment. A multiplier effect is thus assumed to take
place, with increases in the demand for the products produced by the area’s export industries inducing a
series of economic events that enhance local employment and income.
Within this framework, a typical task encountered in the analysis of a metropolitan economy is the identi-
fication of its export industries. Hence, the concept of the location quotient was born.The location quotient
is a measure of the location concentration of an industry within a given area relative to a larger economy.The
formula for calculating a location quotient using the nation as the reference economy is as follows:

 Eimetro 
 E metro 
LQimetro =
 Eination  (4.3)
 E nation 

where:
LQ imetro : location quotient for industry i in metropolitan area m
Eimetro : employment in industry i in metropolitan area m
E metro : total employment in metropolitan area m
Eination : national employment in industry i
E nation : total national employment
Within the context of the economic base theory, the traditional interpretation of the location quotient is
as follows:
LQi > 1 signifies an export or basic industry, and
LQi < 1 signifies a local or non-​basic industry
Such an interpretation, however, may be inaccurate because local consumption and production patterns
may differ across markets, dictating different labor requirements.Thus, many scholars in the field, including
the author, accept location quotient not as a means of differentiating between basic and non-​basic
industries but as a measure of relative industry concentration or industry importance. A location quotient
greater than 1 simply means that the industry in the area under consideration has a greater concentration
than at the national level, and may signify its greater role in the local economy (depending on its share of
total local employment). Such concentration may be attributable to export demand due to competitive

81
 Metropolitan growth analysis

Table 4.2  Location quotients for the Los Angeles economy: 2001–​2017

Sector 2001 2005 2010 2015 2017

Forestry, fishing, and mining 0.20 0.21 0.26 0.21 0.20
Construction 0.71 0.75 0.73 0.75 0.76
Manufacturing 1.10 1.06 1.09 0.95 0.93
Wholesale trade 1.27 1.31 1.34 1.30 1.27
Retail trade 0.85 0.89 0.89 0.85 0.84
Transportation, warehousing, utilities, and information 1.27 1.26 1.27 1.28 1.33
FIRE 1.11 1.18 1.14 1.10 1.10
Services 1.07 1.07 1.06 1.10 1.10
Government and government enterprises 0.76 0.73 0.72 0.70 0.71

advantages, local consumption patterns shaped by the area’s demographics, and/​or local production patterns
shaped by the quality of the area’s labor force and industry structures.

RELATIVE EMPLOYMENT SHARES IN THE LOS ANGELES ECONOMY  Table 4.2 traces the longer-​term trends
in the location quotients for the major non-​agricultural sectors in the Los Angeles economy.These location
quotients have been calculated using the national economy as reference. It should be noted at the outset that
a declining location quotient does not necessarily imply a declining share of the sector under consideration
in the local economy. Such a trend, for example, could be the result of a constant sectoral share in the local
economy and an increasing share of the same sector in the national economy.
As Table 4.2 indicates, during the period 2001–​2017, the share of Wholesale Trade, TWUI, FIRE, and
Services in the Los Angeles employment base was higher than the respective share of the same sectors in
the national economy. The above one-​location quotients for these sectors, and especially, for Wholesale
Trade and TWUI, throughout the period of analysis may be indicative of some specialization of the local
economy in these industries. It is interesting to note the non-​negligible decrease of the Manufacturing
location quotient from 1.1 in 2001 to 0.9 in 2017. FIRE and Services registered small fluctuations around
the 1.1 mark.
Contrary to the aforementioned four sectors, whose importance in the Los Angeles employment
base was greater compared to the nation, the contribution of Construction, Government, and Retail
Trade to the local economy has been lower than their contribution to the national economy. The loca-
tion quotients of the first two sectors was ranging between 0.7 and 0.8 over the period 2001–​2017,
while the location quotient of Retail Trade was ranging between 0.8 and 0.9. In sum, as of 2017, it
seems that only Wholesale Trade and TWUI were noticeably over-​represented in the Los Angeles
economy compared to the national economy, while Government and Construction were considerably
under-​represented.
A note of caution should be added here. For a more accurate interpretation of what these location
quotients mean, one needs to look at them in conjunction with absolute employment shares. For example,
the fact that Wholesale Trade had a location quotient above 1 in 2017 does not mean that it is also one of
the major sectors of the Los Angeles economy, given that employment in this sector accounted for only
4.3% of the area’s total employment (see Table 4.1). On the other hand, the fact that the FIRE industry had
a location quotient considerably higher than 1 and its share of total employment is above 10% reinforces
the conclusion regarding its importance for the local economy.

SHIFT-​SHARE ANALYSIS  The review of historical developments in the Los Angeles economy over the
period 2001–​2017 has highlighted the cyclicality of most of the sectors and the persistence of some secular

82
 Analyzing metropolitan economies

trends that continued in spite of the cyclical movements of the economy. In reviewing these historical
growth patterns, some additional insights can be gained by addressing the following questions:

1. Is a sector in the local economy growing because the overall national economy is doing well?
2. Is it because specific sectors/​industries are doing well at the national level?
3. Or is it because of idiosyncratic features and productive advantages of the metropolitan economy?

We can address these questions using shift-​share analysis. It should be emphasized that this technique is
purely used for analysis of historical data, not forecasting. It only offers a glimpse at the localized or broader
nature of the forces that may drive the local economy. Thus, a more detailed analysis of historical trends in
the key industries and the specific factors that drive them is needed in order to shed more light on their
future prospects. Information provided by studies that may be available at local universities, local Chambers
of Commerce, and/​or Economic Development agencies may provide useful inputs for such an analysis.
Again, the ultimate objective is to gain a better understanding of the prospects for increases in an area’s
output, which, in combination with an understanding of local labor-​supply dynamics, may provide insights
regarding the type of growth that will be taking place in the area.
According to the shift-​share analysis framework, employment growth in a metropolitan area during a
given period can be decomposed into three components:

1. Share effect
Indicates by how much industry i would have grown if it grew at the same rate as all industries at the
national level.
2. Industry effect
Indicates the additional growth a metropolitan area is experiencing due to its particular industry mix.
For example, local growth rates should be higher than national growth rates if faster-​g rowing indus-
tries are over-​represented in the metropolitan economy compared to the national economy.
3. Competitive effect
Indicates the additional growth a metropolitan area’s industries may be experiencing due (perhaps) to
competitive metropolitan location factors. For example, specific sectors at the metropolitan level may
be growing faster than at the national level, due to the area’s competitive advantages.

Decomposing employment growth in this way may reveal more information about the strength of a
particular industry and its prospects for growth. For example, let’s look at the shift-​share analysis of the
employment growth that took place in the Los Angeles metropolitan area during the period 2010–2017
(see Table 4.3). During that period, national employment grew at a cumulative rate of 19.1%, while employ-
ment in the Los Angeles area increased 20%. So the Los Angeles economy performed slightly better than
the national economy.
Shift-​share analysis reveals that if every sector in the Los Angeles economy grew at the same rate as
the national economy did (19.1%), 1,382,279 new jobs would have been added in the area’s employment
base during the period 2010–​2017 (see share component in Table 4.3). Instead, more jobs were created
(1,447,883). As Table 4.3 indicates, all sectors but Manufacturing and Forestry, Fishing, and Mining gained
jobs during the period of analysis. Service jobs grew by 916,255, representing a 27.8% increase over their
2010 level, while TWUI jobs grew by 194,422, representing an increase of 38.7%.
The industry-​effect analysis indicates that an additional 272,268 jobs would be gained (on top of the
1,382,779) if each sector in the Los Angeles economy grew at the same rate as it grew at the national level.
The net gain in jobs over the share component suggests that some of the fastest-​g rowing sectors at the
national level, such as Services and FIRE, are over-​represented in the Los Angeles economy compared to
the nation.This is confirmed also by the above one-​location quotients for these two sectors (see Table 4.2).

83
 newgenrtpdf
Table 4.3  Los Angeles metropolitan area: Shift-​share analysis 2010–​2017

Sector Metropolitan Metropolitan Metropolitan Metropolitan National Share Shift components

employment employment employment employment employment component
Industry Competitive
2010 2017 change change(%) change(%)
effect effect
2010–​2017 2010–​2017 2010–​2017

Metropolitan growth analysis

Forestry, Fishing, and Mining 23,701 22,422 –​1,279 –​5.4% 51.4% 1,284 10,900 –​13,463
Construction 272,698 360,508 87,810 32.2% 8.3% 14,771 7,949 65,090
Manufacturing 560,599 553,550 –​7,049 –​1.3% –​21.4% 30,365 –​150,187 112,773
84

Wholesale Trade 344,112 369,242 25,130 7.3% 3.9% 18,639 –​5,086 11,577
Retail Trade 663,204 724,539 61,335 9.2% 5.3% 35,923 –​892 26,304
Transportation, Warehousing, 502,072 696,494 194,422 38.7% 15.5% 27,195 50,420 116,807
Utilities, and Information
FIRE 817,597 966,681 149,084 18.2% 46.9% 44,286 339,560 –​234,762
Services 3,296,412 4,212,667 916,255 27.8% 36.1% 178,554 1,010,393 –​272,692
Government and Government 756,421 778,596 22,175 2.9% 5.4% 40,972 0 –​18,797
Enterprises
Total 7,236,816 8,684,699 1,447,883 20.0% 19.1% 1,382,779 1,263,057 –​207,165
 Analyzing metropolitan economies

Finally, the competitive effect estimates convey good news for the Los Angeles economy.This effect is nega-
tive only for Forestry, Fishing, and Mining, FIRE, Services, and Government, suggesting that during the
period 2010–​2017 the Los Angeles economy was plagued by competitive disadvantages for only these four
sectors. A more detailed analysis is needed in order to identify the nature of these competitive disadvantages
and evaluate whether they will persist into the future.

∑ E i metro g i metro = [ ∑ i E i metro g nation ] + [ ∑ i E i metro (g i nation -g nation )] + [ ∑ i E i metro (g i metro -g i nation ) ]

Share Component Industry Effect Competitive Effect

Shift Components
where:
E i metro Employment in industry i in a given metropolitan area
g i metro Employment growth rate in industry i in this metropolitan area
g i nation Employment growth rate in industry i in the nation
nation
g Total Employment growth rate in the nation

Share Effect: Indicates by how much industry i would have grown if it grew at the same rate as
all industries at the national level

Industry Indicates the additional growth a metropolitan area is experiencing due to

Effect the nature of its industries

Competitive Indicates the additional growth a metropolitan area's sectors are experiencing possibly due to
Effect metropolitan location factors

Notes : Also see DiPasquale and Wheaton, Chapter 7

Synthesis: growth prospects
Overall, the historical analysis of the Los Angeles MSA economy by sector has shown that it has followed
very closely the fluctuations of the national economy, with only FIRE and Services defying or minimally
being affected by cyclical downturns.This conclusion was also confirmed by the shift-​share analysis, which
shows that the total employment growth rate over 2010–​2017 in the Los Angeles area presented a very
small deviation from the national employment growth rate. Most of the secondary economic sectors of
the Los Angeles MSA grew at rates faster than at the national level but the by far most important sector,
Services, grew at a slower rate (27.8% vs 36.1%).
Based on this analysis we can conclude that the Los Angeles MSA economy will continue to grow as
the national economy continues on the growth path. FIRE and Services are expected to continue to be
the fastest-​g rowing sectors in the medium term, with TWUI potentially continuing to be the third fastest
growing sector. The historical analysis has also confirmed the high elasticity of labor supply in the Los
Angeles area, which suggests that future growth will be mostly manifested in increases in employment with
small if any increases in real wages.This in turn suggests that there will be increasing demand for moderate-​
quality residential real estate as well as rising demand for commercial space by companies serving middle-​
income and lower-​middle-​income households.

85
 Metropolitan growth analysis

Selected sources of information for historical data and forecasts

Sources of metropolitan-​level historical data and/​or forecasts include:
BEA, Bureau of Economic Analysis (BEA), U.S. Department of Commerce
Additional potential sources for such information include studies/​research reports prepared at uni-
versities, local Chambers of Commerce, planning departments, and other government-​agency research
divisions. See other data sources in Clapp, J. M. (1987). Handbook for Real Estate Market Analyses. Englewood
Cliffs, NJ: Prentice Hall.

Chapter summary
In this chapter, we have discussed the basic steps of a broad-​brush metropolitan-​g rowth analysis process
and presented its application to the Los Angeles metropolitan area. This process includes five basic steps:

• Identification of the spatial unit of analysis

• Analysis of basic growth indicators
• Analysis of population and income trends
• Analysis of sectoral-​employment, income, and industry-​structure trends
• Synthesis.

The ultimate goal of this analysis is to understand the broader patterns of growth that the metropolitan area
under consideration will be experiencing in the years ahead, and their broader implications for real estate
demand. For example, broad-​brush analysis of the Los Angeles metropolitan area showed that its economy
has been following closely the swings of the national economy with the exception of Manufacturing,
FIRE, and Services that remained on the path of ongoing secular trends. Furthermore, the historical ana-
lysis has shown that the major sectors of the Los Angeles economy had experienced a demand-​induced
growth pattern with considerable gains in employment but very small gains in real wages. Within this
context it is expected that the medium-​term prospects of the Los Angeles economy will be very similar to
those of the national economy, with continuing growth in employment at rates comparable to those of the
nation but small gains in real wages. The secular trends of declining Manufacturing employment and the
fast rise of FIRE and Service employment are expected to continue in the next three to five years.

Questions
1. A company specializing in the development of speculative high-​end retail space is currently in search
of metropolitan areas whose growth patterns will likely be conducive to such a development. The
company has obtained information on two metropolitan areas, whose employment growth over the
next five years is expected to amount to 100,000 workers. The data suggest that the two areas’ current
average incomes (adjusted for living costs) are at about the same level; that their income-​distribution
profile is very similar; and that their existing supply of high-​end retail space is just sufficient to satisfy
current demand. In addition, the following information was gathered regarding each area’s economic
base, migration trends, and future prospects.

Metropolitan area A’s economy is largely based on a growing number of high-​technology, information-​
based industries (e.g., computer software). Information obtained from the local Chamber of Commerce
suggests that prospects for further industry growth are excellent, as foreign sales of those industries’
products are expected to increase substantially over the next five years. Such growth is likely to be
facilitated by local government support, including the provision of incubator facilities for new firms, a

86
 Analyzing metropolitan economies

host of cultural and other amenities, and the area’s high-​end residential neighborhoods likely to suit the
preferences of the location-​selective, highly skilled workforce employed by the area’s high-​tech indus-
tries. These high-​skill workers are drawn both from local universities and from adjacent metropolitan
areas with prestigious science and technology programs. Notably, amenity-​induced immigration of
workers in this area is rather small, due to its cold climate and perceptions of high living costs.
Metropolitan area B’s economy is based on trade, service, and entertainment sectors that are characterized
by relatively price-​elastic product and labor demands. Given the area’s strategic location and quality of
infrastructure, its growth is likely to continue being fueled by non-​negligible expansion in those sectors’
output. Furthermore, the area’s amenities and expectations for further growth are likely to continue
inducing sizable immigration of appropriately skilled workers. The latter are expected to continue being
attracted from two neighboring metropolitan areas whose growth outlook seems to be rather dim.

a) What type(s) of growth is each metropolitan area likely to experience in the coming years? What will
the dominant features of this growth be? Explain clearly your answer.
b) Which of the two metropolitan areas seems to have better prospects for developing speculative high-​
end retail space? Explain clearly your answer. If your answer is conditional, again, be sure to explain.

Project #1: Analyzing metropolitan growth patterns

The purpose of this exercise is to sharpen the skills of the students in describing broad patterns of metro-
politan growth, interpreting meaningfully such patterns, and exploring their implications for long-​term real
estate demand prospects.

Data to be provided by instructor

Historical and forecast data on population, employment, and earnings by one-​digit SIC sector, for at least
five metropolitan markets.

Main tasks
Preparation of a brief essay (5–​10 double-​spaced pages) on the results of a broad-​brush analysis of metro-
politan growth patterns in one of the metropolitan areas for which historical and forecast data have been
provided.The students are asked to follow the suggested structure and the accompanying instructions below.

Introduction
As an introduction to the essay, the importance and role of metropolitan growth analysis should be discussed
in one to two paragraphs. The introduction should also indicate which area was chosen and lay out the
structure of the report, presenting briefly the contents of its various sections.

Analyzing metropolitan growth patterns

Using the data provided by the instructor, the students should discuss the historical and forecast patterns of
growth in the metropolitan area they have chosen to analyze. More specifically:

a) First, use summary growth indicators (indices of population, total non-​agricultural employment, and
inflation-​adjusted non-​agricultural earnings per worker) to examine the overall historical and forecast

87
 Metropolitan growth analysis

growth patterns in the metropolitan area in question. Provide a brief but thoughtful discussion of such
patterns, supplemented with supporting summary data in tables and/​or graphs.
b) Second, examine growth indicators by one-​digit SIC sector (indices of non-​agricultural employment
and inflation-​adjusted earnings or compensation per worker by economic sector). Again, provide a brief
discussion of your analysis, supplemented with appropriate summary data.
c ) Third, using available sectoral data at the national level, provide a relatively more detailed analysis of
metropolitan sectoral employment. To this end, you can look (i) at the degree of concentration of
broad economic sectors using location quotient analysis and (ii) the main growth components of these
sectors in selected time periods using shift-​share analysis. Again, supplement your discussion with sum-
mary tables.
d) Given your analysis in (a)–​(c), what (preliminary) conclusions, if any, can you draw in terms of the
forces that underlie the observed and forecast metropolitan growth patterns? Are the forces that have
generated growth in the last two decades likely to continue operating in the next five years or so?
Explain.
Note that you are always welcome (in fact, encouraged) to refer to other, more detailed information you
consider important to collect in order to better address any of the questions posed above.

Growth prospects and real estate markets

Lastly, in the concluding section of your essay, explore the implications of your analysis for real estate
demand prospects and development opportunities in the metropolitan area under question over the next
five to ten years.

Note
1 It should be noted that output per worker is a much better indicator of productivity effects.

References and additional readings

Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Analyzing the Local Economic Environment. Chap. 3 in Real
Estate Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W. Wheaton. 1996. Economic Growth and Metropolitan Real Estate Markets. Chap. 7 in Urban
Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice-​Hall.
Schmitz, A. and D. Brett. 2015. Real Estate Market Analysis, 2nd Edition. Washington, DC: Urban Land Institute.

Appendix 4A: About indices

Sometimes it is very insightful to view several economic data series on the same graph. This often facilitates a more
meaningful interpretation of historical trends and forecasts. As indicated earlier, however, when different variables are
expressed in different units, or when they differ greatly in magnitude, it is very convenient to convert the data into
indices before displaying them on the same graph.

What’s an index?
An index is a numerical series that describes the movements of a variable through time.These movements are measured
with respect to a base period during which the index is set to 100. The example in Table 4A.1, presenting a population
index for Los Angeles, illustrates how indices in general should be interpreted. As this table demonstrates, percentage
changes are measured with respect to the base year, which typically is the first year of the series (1969 in this case).
For example, the 1972 index value of 103.4 implies that the population between 1969 and 1972 increased by 3.4%.
Similarly, the 1995 index value of 156.1 implies that over the period 1969–​1995, the population increased by 56.1%.

88
 Analyzing metropolitan economies

Table 4A.1  Los Angeles population index and interpretation

Year Population index Interpretation

1969 100.0
1970 101.6 1969–​1970 : 1.6% increase
1971 103.0 1969–​1971 : 3.0% increase
1972 103.4 1969–​1972 : 3.4% increase
.
.
.
1995 156.1 1969–​1995 : 56.1% increase

Table 4A.2  Calculating the Los Angeles population index

Year Population index Population (in thousands) Index calculation

1969 100.0 9,858.7 9,858.7*100
9,858.7
1970 101.6 10,014.8 10,014.8*100
9,858.7
1971 103.0 10,156.6 10,156.6*100
9,858.7
1972 103.4 10,192.3 10,192.3*100
9,858.7
. . . .
1995 156.1 15,392.3 15,392.3*100
9,858.7

How is the index calculated?

The index is calculated by simply dividing each of the numbers in the series with the base-​year number and then multi-
plying the result by 100.Table 4A.2 presents the formula and demonstrates how it was used to calculate the population
index presented in Table 4A.1.

Index comparisons
In order to provide an example of how indices can help plot variables of markedly different levels and/​or units on the
same graph, we plot the index for total employment and total real earnings for a hypothetical market in Figure 4A.1.
Table 4A.3 presents the numbers that were used to produce the graph. The graph provides immediately a very clear
picture of how movements in these two variables compare through time. In particular, it is obvious from the graph
that income in the market under consideration was rising faster than employment in 1970 and 1972. This suggests that
during those two years, earnings per worker were increasing, indicating a positive wealth effect, which is consistent with
a demand-​induced metropolitan growth pattern.

About the consumer price index (CPI) and income

Income figures are often provided in nominal terms. In such a case, they need to be adjusted to take into account gen-
eral price inflation, that is, converted into real income. Such an adjustment is needed because in numerous economic
applications in several fields, including real estate, it has been found that economic behavior is driven by movements
in real income, not nominal income. This is understandable since general price inflation diminishes purchasing power,
which is the underlying driver of consumer spending behavior. For example, a dollar in 1969 is not equivalent to a
dollar today in terms of purchasing power. A consumer could buy much more with a dollar in 1969 than a consumer

89
 Metropolitan growth analysis

Table 4A.3  Index comparisons 106

105
Year Employment Total real earnings
index index 104
103
1969 100.0 100.0 102
101

1970 99.8 100.3 100

99
98
1971 98.9 98.8
97
1969 1970 1971 1972

1972 102.6 105.1

Employment index Total real earnings index

Figure 4A.1  Index comparisons

can buy today with the same amount because the prices of all goods have increased considerably between 1969 and
today.Thus, nominal income figures have to be converted into figures reflecting “constant purchasing power,” i.e., con-
stant dollars, to allow for a more meaningful examination of their movements through time.The consumer price index
(CPI) is typically used for converting nominal income figures into constant dollars.

The consumer price index (CPI)

The CPI is a market price index measuring the price changes of a fixed market basket, which includes over 400 goods
and services representing average household purchases associated with day-​to-​day living. Income and social security
taxes are excluded, but sales taxes and user fees are included. There are CPI indices for two consumer groups:

• Urban consumers (the corresponding index is denoted as CPI-​U). This group excludes rural residents, military, and
institutionalized population.
• Urban wage earners (the corresponding index is denoted as CPI-​W). This group is a subset of urban consumers (32%)
and includes households with 50% or more of their income coming from clerical or wage occupations. In addition,
at least one household-​head earner must have been employed for 37 weeks during the previous 12 months.

These two indices are derived from detailed expenditure information provided by samples of families and individuals.
The CPI-​U is the most comprehensive index and the one most often reported by the media. Table 4A.4 presents two
CPI-​U series, one with 1982–​1984 as the base years and one with 1987 as the base year.
The CPI index is used to adjust incomes and prices for the effects of inflation or, alternatively, to convert current-​
dollar amounts into constant dollars of a base year. Table 4A.5 presents an income-​per-​worker series in current dollars
and its equivalent in 1982–​1984 dollars. It also presents the formula for converting current dollars to constant dollars
and its specific use for the calculation of the constant-​dollar equivalent of the 1990 income-​per-​worker.
Notice that, as demonstrated in the example portrayed in Table 4A.5, in order to convert the 1990 income to con-
stant 1982–​1984 dollars we need to divide it with the 1990 CPI index that has 1982–​1984 as its base years and then
multiply the result by 100. But what if we wanted to translate the 1990 income figure into constant dollars of a different
base year, such as 1970 for example? In that case, we would need a CPI series that has 1970 as its base year. This series
can be calculated from the series that has 1982–​1984 as its base years using the formula in (4A.1):

CPI (t )1982 −1984

CPI (t )1970 = * 100 (4A.1)
CPI (1970 )1982 −1984

90
 Analyzing metropolitan economies

Table 4A.4  Consumer price index series

Year CPI base years 1982–​19841,2 CPI base year 1987 Inflation
1970 38.8 34.2
1971 40.5 35.7 4.4%
1972 41.8 36.8 3.2%
1973 44.4 39.1 6.2%
1974 49.3 43.4 11.0%
1975 53.8 47.4 9.1%
1976 56.9 50.1 5.8%
1977 60.6 53.3 6.5%
1978 65.2 57.4 7.6%
1979 72.6 63.9 11.3%
1980 82.4 72.5 13.5%
1981 90.9 80.0 10.3%
1982 96.5 84.9 6.2%
1983 99.6 87.7 3.2%
1984 103.9 91.5 4.3%
1985 107.6 94.7 3.6%
1986 109.6 96.5 1.9%
1987 113.6 100.0 3.6%
1988 118.3 104.1 4.1%
1989 124.0 109.2 4.8%
1990 130.7 115.1 5.4%
1991 136.2 119.9 4.2%
1992 140.3 123.5 3.0%
1993 144.5 127.2 3.0%
1994 148.2 130.5 2.6%
1995 152.4 134.1 2.8%
1996 156.9 138.1 2.9%
1997 160.5 141.3 2.3%
1998 163.0 143.5 1.5%

The annual figures represent averages of the seasonally adjusted monthly CPI indexes.
1

Averaging the CPI indices for 1982, 1983, and 1984 yields 100.
2

Table 4A.6 presents three CPI series that have different base years (1982–​1984, 1970, and 1980). It also presents the
general formula for changing the base year of an index to a new year X and applies this formula to calculate the 1990
CPI figure that has 1970 as its base year.
Sometimes the analyst may be confronted with the case in which some data are provided in constant dollars of one
year, while some other data are provided in dollars of a different year.Table 4A.7, for example, presents data on personal
income per worker that are expressed in 1982–​1984 constant dollars from 1970 to 1975, and in 1980 constant dollars
for the period 1976–​1981. In such a case, any meaningful interpretation of the data requires that the two series be made
consistent.This can be done, for example, by converting the data from 1976 to 1981 to the same base year as the data for
the period 1970–​1975. In order to do that we need to apply to the former set of data the formula described by (4A.2):

INCOME1980
INCOME1982 −1984 = * CPI 1980 (4A.2)
CPI 1982 −1984

Table 4A.7 demonstrates also how this formula is applied specifically to convert the 1976 income-​per-​worker figures
to 1982–​1984 constant dollars.

91
 Metropolitan growth analysis

Table 4A.5  Converting income into constant 1982–​1984 dollars

Year (t) CPI Personal Personal

base: income/ income/
1982–1984 worker worker
(in 000s of (in 000s of
current $) 1982–1984 $)

1970 38.8 10.55 27.19 Income at Year t Current 47.09 22.40

1971 40.5 11.15 27.54 in Constant Dollars 49.15 22.69
1972 41.8 11.66 27.91 82-84 = * 100 50.73 23.00
1973 44.4 12.02 27.08 dollars CPI at t 53.88 22.31
1974 49.3 12.98 26.33 (82-84 base) 59.83 21.70
1975 53.8 14.22 26.43 65.29 21.78
1976 56.9 15.30 26.89 69.05 22.16
1977 60.6 16.23 26.78 So: 73.54 22.07
1978 65.2 17.45 26.76 79.13 22.05
1979 72.6 18.71 25.77 Income in 1990 88.11 21.24
1980 82.4 20.61 25.01 in Constant 35.42 100.00 20.61
1981 90.9 22.70 24.97 82-84 = * 100 110.32 20.58
1982 96.5 24.58 25.47 dollars 130.7 117.11 20.99
1983 99.6 25.58 25.68 120.87 21.16
1984 103.9 26.83 25.83 126.09 21.28
1985 107.6 27.98 26.00 = 27.10 130.58 21.43
1986 109.6 29.11 26.56 133.01 21.88
1987 113.6 30.47 26.83 137.86 22.10
1988 118.3 31.94 27.00 143.57 22.25
1989 124.0 33.62 27.11 150.49 22.34
1990 130.7 35.42 27.10 158.62 22.33

Table 4A.6  Changing the base year of an index to Year X

Year CPI CPI CPI Formula

(t) base year base year base year
1982–1984 1970 1980

1970 38.8 100.0 47.1

1971 40.5 104.4 49.2 Year t’s
1972 41.8 107.7 50.7 index at old
1973 44.4 114.4 53.9 Index in Year t base year
1974 49.3 127.1 59.8 at new = __________ * 100
1975 53.8 138.7 65.3 base Year X Year X’s index
1976 56.9 146.6 69.1 at old base year
1977 60.6 156.2 73.5
1978 65.2 168.0 79.1
1979 72.6 187.1 88.1
1980 82.4 212.4 100.0
1981 90.9 234.3 110.3 So:
1982 96.5 248.7 117.1
1983 99.6 256.7 120.9 Index in
1984 103.9 267.8 126.1 Year 1990
1985 107.6 277.3 130.6 at New Base = 130.7
__________ * 100
1986 109.6 282.5 133.0 Year 1970 38.8
1987 113.6 292.8 137.9
1988 118.3 304.9 143.6
1989 124.0 319.6 150.5 = 336.9
1990 130.7 336.9 158.6

92
 Analyzing metropolitan economies

Table 4A.7  Forming consistent time series when data are provided in constant dollars of different years

Year CPI Personal

(t) base income
1982–1984 per worker
in constant
1982–1984 $

1970 38.80 27.19

1971 40.50 27.54
1972 41.80 27.91 Year t’s Year t’s
1973 44.40 27.08 income in income
1974 49.30 26.33 constant in 1980 $ Year t’s
1975 53.80 26.43 1982-1984 $ = _________ * index
Year t’s at 1980
index at base
Year CPI CPI Personal Personal 1982-1984
(t) base base income income base
1980 1982–1984 per worker per worker
in constant in constant
1980 $ 1982–1984 $ Year 1976’s
income in 22.16
constant =
* 69.053
1976 69.05 56.90 22.16 26.89 1982-1984 $
1977 73.54 60.60 22.07 26.78 56.90
1978 79.13 65.20 22.05 26.76
1979 88.11 72.60 21.24 25.77 = 26.893
1980 100.00 82.40 20.61 25.01
1981 110.32 90.90 20.58 24.97

93
 PART C

Analyzing residential real estate markets

newgenprepdf

5
RESIDENTIAL REAL ESTATE
MARKETS

Introduction 97
Overview of residential market analysis 98
Basic housing market economics 101
Housing demand: analysis elements 102
Demand for a particular housing type 102
Demand for tenure type 102
Demand for quality/amenity levels 103
Demand for particular location characteristics 103
Households and demographic structures 103
Income, prices, and affordability 106
Mortgage rates and inflation 110
Expectations 112
New housing supply 112
New residential construction 113
The supply of resale homes 114
Chapter summary 115
Questions 116
References and additional readings 116

Introduction
In this part of the textbook, we will focus on urban housing markets and residential market analysis both
at the metropolitan-​area and the site-​specific level. Within this context, we first provide an overview of
the residential market-​analysis process and then discuss basic housing market economics. Subsequently we
review analytical frameworks and methodologies for assessing housing demand at the macroeconomic or
metropolitan-​area level.Then we focus on the micro-​analysis of residential real estate and, more specifically,
on issues regarding project amenities and product type design, as well as the quantitative analysis of housing
development scenarios.

97
 Analyzing residential real estate markets

In this chapter, we first outline the major components of a well-​integrated residential market study. We
then turn to focus on basic concepts of housing market economics and specifically on the major drivers of
housing demand and housing supply, including new and resale housing units. The purpose of the housing
economics section is to familiarize the reader and the real estate analyst with the fundamental forces that
drive the housing market and highlight the major analysis elements that a market study should focus on, at
least at the macroeconomic level.

Overview of residential market analysis

Before discussing methodological issues, it is important to highlight two major idiosyncrasies that distin-
guish the residential market from other real estate markets and have important analytical implications. In
particular, residential markets are:

1 . Highly heterogeneous, segmented by location, and stratified by various consumer types.

2. Largely driven by demographics and ability to pay/​affordability, as well as consumer and supplier
expectations.

The housing market is highly heterogeneous because housing units differ along many dimensions, including
tenure mode, lot size and topography, unit size, room typology, layout, design, amenities, construction
materials, and location (O’Sullivan, 2009). The heterogeneity of an area’s housing stock directly reflects the
diversity of its demand base that can be segmented primarily according to demographic characteristics and
ability to pay.
To better understand the bottom-​line questions a residential market study needs to address, consider
the following simple scenario. Suppose that based on population, income, and other data it has been
determined that metropolitan area X is very likely to experience increases in housing demand over the
coming years. Within this metropolitan area, a site suitable for residential use has been identified and its
potential for successful apartment development is evaluated by a market analysis firm. The bottom-​line
questions that the market study needs to answer are:

• When to build?
• What to build and for whom?

The first question refers to the appropriate timing of entering the local housing market so that project
value is maximized. For example, will the market be strong enough at the anticipated time of entry to
absorb the planned apartment units at a viable rate and prices, or should the development be delayed due
to weak market prospects? The second question refers to the definition of the housing project in terms of
density (proportion of built versus open space), project amenities (parking, pool, etc.), unit characteristics
(size, number of bedrooms, number of bathrooms, etc.), unit amenities (fireplace, balconies, etc.), and other
characteristics so that, again, project value is maximized. In order to answer these questions, the analyst
needs to carry out two major analysis steps (see Figure 5.1):

1. Assess broader market strength and most likely movements in market fundamentals over the project’s
planning horizon (macroeconomic analysis).
2. Refine/​re-​define the original development scenario to conform to the preferences of the household
types most likely to be attracted to the site, and assess strength of demand at the site and project level
(microeconomic, site-​specific, or project-​specific analysis).

98
 Residential real estate markets

Analysis framework
When to build? What to
build and for whom?

A. Assess market strength B. Refine/redefine the

at the nonsite, market/ development scenario and
submarket level assess market strength
(Identify niches in product at site and project levels
types, consumer targets, time) (What to build and for whom)

Q. How strong will the market for Q1. What type of development and
residential developments be product design will be most
during the expected absorption marketable at the given site?
and holding periods of a
project? Q2. What is the competitive strength,
income earning capacity, and
ANALYSIS achievable absorption rates of
FRAMEWORK such a development?
(Given a preliminary development scenario)
QUALITATIVE ANALYSIS (Q1)
Definition of the market area Analysis of the site and its environment

Definition of the analysis time frame Analysis of local market conditions

Assessment of market strength Refinement of product design and

using appropriate techniques consumer targeting

QUANTITATIVE ANALYSIS QUANTITATIVE ANALYSIS (Q1 & Q2)

Accounting analysis of Project price/rent analysis

residential demand and supply Focuses on project price/rent
Focuses on the estimation of estimation through competitive
demand-supply “GAPS” and differential or hedonic price analyses
effective housing market demand and determination of profitable
amenity packages (amenity mix
and development density)

Econometric analysis of Project absorption analysis

residential markets Focuses on the estimation of
Focuses on forecasting likely project absorption schedule
movements in housing completions, and evaluation of time-of-entry
prices/rental rates, and stock and phasing alternatives

Financial feasibility

Figure 5.1  Overview of residential market analysis

99
 Analyzing residential real estate markets

Macroeconomic analysis for residential real estate focuses on three major tasks:

1. The assessment of market strength and forecasting of most likely movements in housing demand, con-
struction, and prices/​rents in the broader residential market within which the project is expected to
compete upon completion and over the expected holding period.
2. The identification of broader niches, target consumer groups, and product types that match the
preferences of these consumers groups.
3. The determination of the most appropriate time of entry in the market.

In order to address these issues, the following specific questions need to be thoroughly investigated and
answered:

a ) What is the net or unmet demand for a particular residential product type?

b) What are the likely movements in apartment/​housing rents/​prices over the development/​holding
period?

Answering these questions requires three types of analyses:

1. Analysis of effective housing demand, taking into account demographic structures, ability to pay, and
preferences for tenure types and product design.
2. Analysis of available and expected supply, taking into account housing market dynamics and developers’
expectations.
3. Analysis of the supply-​demand gap and market price dynamics.

Within this context, the macroeconomic analysis of the residential market includes the following broader
methodological steps:

1 . Definition of the market area

2. Definition of the analysis time frame
3. Assessment of market strength through:
• Analysis of expected residential demand and supply and development of estimates of the demand-​
supply “gap” using simpler accounting techniques, and/or
• Econometric analysis of the behavioral relationships underlying the functioning of the local
housing market and development of forecasts of housing demand, supply, and price/​rental rate
movements.

The second major stage of residential market analysis focuses on the specific site and residential project
under consideration and, as such, it is referred to as microeconomic or site-​specific analysis. Emphasizing pro-
active (not reactive) design and consumer research (not marketing), such micro-​analysis aims at answering two
basic questions:

1 . What type of housing development and product design will be most marketable at the given site?
2. What is the competitive position of the proposed residential project, and what prices/​rents and absorp-
tion rates can be expected given such a position and forecast market movements?

Addressing these questions requires both qualitative and quantitative analysis of project attributes and the
local market environment. Qualitative analysis focuses on:

100
 Residential real estate markets

• The site and its environment

• Access and linkages
• Local market conditions
• Refinement of consumer targeting and modification of product design so as to optimize the match
between site/​project attributes and target group preferences.

Quantitative analysis of residential projects involves a more detailed analysis of the competition, which
provides the basis for further refinement of project design and evaluation of its potential performance. The
ultimate task of the quantitative analysis is the estimation of the bottom-​line figures needed to assess the
financial feasibility of the project. As such, it focuses on the:

1. Estimation of the project’s market rent/​price over its planning horizon. This involves two stages of analysis.
The first stage is the estimation of project rent/​price at the time of analysis through competitive differ-
ential techniques or hedonic valuation methodologies. The latter methodology allows the estimation
of price-​attribute relationships, which can provide the basis for the identification of the most profit-
able residential amenity mix and development density. The second stage involves the development of
forecasts of project rents/​prices at the time of completion and beyond, taking into account anticipated
changes in local housing market conditions over the project’s planning horizon.
2. Estimation of the project’s absorption schedule taking into account its competitive position (which is a
function of relative amenity packages and attributes), as well as the anticipated changes in market-​wide
demand-​supply conditions over its planning horizon. Such project absorption estimates can be used
in combination with rent estimates to calculate project revenues. The latter can in turn help evaluate
project viability from the market perspective. They can also provide the basis for assessing the impact
of alternative times of entry and phasing scenarios on project performance.

In sum, the two major stages of residential market analysis—​macroeconomic and site-​specific—​generate
the information that is necessary for evaluating project performance and deriving revenue projections.The
latter are crucial inputs in evaluating the financial feasibility of the project. In particular, feasibility studies
meticulously analyze project revenues and costs and evaluate whether the project is likely to meet key
financial and investment objectives.

Basic housing market economics

The macroeconomic analysis of residential real estate markets is not a simple task. Understanding the
elements of the analysis process and their relationships, from preliminary market definition and the non-​
site level of analysis to the site-​specific level, requires some basic knowledge of the economics of housing
markets. It is this background knowledge that this section is attempting to provide. Within this context, we
focus on the economic drivers of housing demand and housing supply.
The underlying premise of the macroeconomic analysis of residential real estate markets is that housing
market behavior differs across metropolitan areas. This differential behavior is driven not only by differences
in the levels and prospects of housing demand and supply drivers, but also by differences in the magnitude of the
effect that these drivers have on local housing demand and supply. For example, the price or income elas-
ticity of housing demand in one metropolitan area may not be the same as in another metropolitan area
due to different distribution of households across income and age groups. In addition, the price elasticity of
housing supply may also vary across metropolitan areas because of differences in home-​building-​industry
structures, land availability, and growth/​development restrictions. Such differences in income and price
elasticities of housing demand and supply across markets are bound to generate differential house price and
rent changes even in response to the same exogenous economic shock.

101
 Analyzing residential real estate markets

In general, there has been limited research on variations in income and price elasticities of housing
demand and supply across markets. Green et al. (2005) found substantial differences in price elasticities of
housing supply across metropolitan areas. Their analysis suggests that such differences can be attributed to
differences in population levels, population growth rates, density, house price levels, and regulatory envir-
onment. Also, Anundsen and Heebøll (2014) validated geographical variations in the long-​run effects of
changes in income, housing supply, and interest rates on house prices, as well as differences across locations
in the speed by which housing market equilibrium is restored. Changes in income have a direct effect
on housing demand and an indirect effect on supply through their effect on developer expectations.
Differences in the long-​run effect of income changes on house prices across metropolitan markets may
reflect differences in income elasticities of housing demand.

Housing demand: analysis elements

The basic unit of housing demand is the household. According to the U.S. Census Bureau definition, the
term “household” refers to one or more individuals living together in a family or a non-​family setting.
Therefore, by definition, the number of units demanded in an area should be equal to that area’s number of
households. Households differ in terms of a number of attributes and that is why it is important to empha-
size the multi-​dimensional nature of housing demand. Specifically, housing demand can be segmented along
four major dimensions:

Demand for a particular housing type

This is defined as the number of households that demand a particular housing type (i.e., single-​family
detached, townhouse, apartment, etc.) and have the ability to pay for it.

Demand for tenure type

Tenure mode is another major component of housing choice. When segmenting the market by tenure
mode, the analyst needs to quantify effective demand for either of two tenure types:

1. Renting
2. Owning.

DiPasquale and Wheaton (1996) point out that many households are prevented from owning a house because
their annual income is not adequate for securing financing or because they can’t afford the high transac-
tion costs (down payment and closing costs) associated with the purchase of a house. In addition, mobile
households may be discouraged from choosing ownership because of the problems associated with exiting
this tenure mode. Such problems include time-​consuming disposition processes and the risk of incurring
capital losses. Tenure type choice is also affected by federal policies that reduce the cost of homeownership,
such as the deductibility of interest payments on housing loans from taxable income. According to data
presented by DiPasquale and Wheaton (1996), homeownership rates rise with age and income.
Boehm and Schlottman (2011) confirmed also the effect of age and income on the transition from
renting to home ownership. In particular, they verified a significant effect of family income and net wealth
on the probability of transitioning from renting to homeownership. They also found that single males and
females, and older individuals are significantly less likely to make the initial transition to homeownership
from rental tenure. Furthermore, they confirmed that the out-​of-​pocket cost of homeownership (mort-
gage debt service, owner’s property taxes, home maintenance, insurance and owner’s cost of utilities) had a
significant negative impact on the likelihood of this transition consistently over the three decades (1970s,

102
 Residential real estate markets

1980s, and 1990s). On the contrary, house price appreciation was found to have a strong effect in the 1970s
but no statistically significant effect was confirmed in the 1990s.

Demand for quality/​amenity levels

Housing demand can also be segmented in terms of housing-​unit amenity levels and quality. For example,
different households have different needs and preferences in terms of number of bathrooms, number of
bedrooms, sizes of the different rooms, quality of the structure, and so on. Moreover, some households are
willing to pay for some amenities that are not mere necessities, such as fireplace, pool, or tennis court, while
others are not. Thus, demand for a particular housing type can be further segmented along the amenity
level and quality spectrum.

Demand for particular location characteristics

Location is another important dimension of housing demand. Each household, depending on its
circumstances, will have specific requirements with respect to the location of their house, such as proximity
to a good school, reasonable commuting distance from their workplace and other city destinations, quality
of the neighborhood, etc.
The multi-​dimensional segmentation of housing demand and heterogeneity of the housing stock is a
reflection of the diversity of an area’s households and their housing needs and preferences. Thus, demand
for specific housing types depends on the characteristics of an area’s households and a number of other
factors relating mainly to the cost of housing and its affordability. By all accounts, the key determinants of
housing demand at the urban area level are:

1 . Number of households and their demographic structure

2. Income, prices, and affordability
3. Mortgage rates and inflation
4. Expectations.

As discussed in more detail below, income and demographic structures are the most critical factors in
determining the allocation of an area’s aggregate housing demand to its major segments.

Households and demographic structures

Net household formation, that is, household growth, is of primary importance when trying to assess the
prospects of increases in demand for residential real estate in an area since the basic unit of housing demand
is the household.1 Net household formation is estimated simply as the change in the number of an area’s
households from one period to another. Positive net household formation would therefore indicate increases
in an area’s number of households. This in turn –​if it is not due to falling house prices or apartment rents –​
would imply an upward shift in the housing demand curve. Arestis and Gonzalez (2013) and Sivitanides
(2015, 2018) have verified the effect of population and household growth on house prices and, by extension,
on housing demand.Therefore, when analyzing the prospects of demand in a residential market, it is critical
to examine the prospects of the major non-​price drivers of net household formation. These include:

• Internal growth in the number of households, determined by:

• Demographics and especially expected changes in the size of the population in household-​forming
age cohorts.
• Social factors, such as divorce rates, marriage rates, etc.

103
 Analyzing residential real estate markets

• Net migration, driven by:

• Worker and residential amenities including affordable housing (in the case of supply-​driven metro-
politan growth).
• Job availability and wages (in the case of demand-​driven growth).

Demographics determine household formation to a significant extent as new households are formed at
different rates within the various age and income groups. Moreover, significant fluctuations in historical
birth rates can generate considerable variations in new household formation in the years ahead as baby-​
boom or baby-​bust generations age and enter household-​forming age groups. Death rates also affect the
net change in the number of an area’s households since they contribute to household losses.
Social factors, such as marriage rates and divorce rates, also affect the rate of household formation.
For example, higher marriage rates could have a negative effect on household formation if two separate
households merge into one. Higher divorce rates, however, may have a positive effect on household forma-
tion if one household splits into two separate households.
In the absence of survey-​based data, an area’s number of households can be estimated from population
and household-​size estimates or from population estimates and headship rates.2 A  recent study of age-​
specific headship rates in the UK by Duffy, Byrne and FitzGerald (2014) shows that headship rates rise
from about 38% for the age group of 25–​29, to 50% for the group of 30–​34 and then they gradually rise
to over 60% for the group of 50–​54. Given that both household size and headship rates vary by age and
income group, the formula for calculating net household formation is given by (5.1) or (5.2), depending
on whether household-​size estimates or headship rates are available.

HH = Σ (POPij /​ Sij) (5.1)

HH = Σ (POPij * Hij) (5.2)

where:
HH : number of households
POP : population
S : average household size
H : headship rate
i : age group (i=1, 2, 3,....)
j : income group (j=1, 2, 3,....)

As indicated by formula (5.1) –​and as will be shown in a sensitivity analysis example that follows in the
next ­chapter –​ household size is a very important consideration when used to convert a particular popula-
tion growth forecast to a household forecast in order to predict housing demand. Furthermore, household
size affects the type of units demanded since, all else being equal, larger households are more likely (but not
necessarily) to demand larger units. It appears that there is a secular trend of decreasing household size in
the U.S. In particular, according to U.S. Census Bureau data the average household size has decreased from
3.14 persons in 1970 to 2.53 in 2018. However, practically, it has remained stable since 2013, as it has been
fluctuating between 2.53 and 2.54.
It should be noted that although new households represent the source of new housing demand at the
aggregate market level, new demand for a specific residential development may come also from existing
households that decide to move within the project’s market area. For this reason, very often, housing
market studies examine housing mobility patterns in order to segment out specific age groups and types
of households that tend to move more frequently and could potentially seek housing at the development
under consideration. The American Housing Survey (AHS) published by the U.S. Census Bureau provides

104
 Residential real estate markets

information on housing mobility patterns within the nation’s largest metropolitan areas. The information
provided includes, among others, the reasons for leaving the previous house, reasons for choosing the new
neighborhood and home, and mobility by age of household head.
According to some data presented by DiPasquale and Wheaton (1996), younger households and renters
appear to be considerably more mobile than mature households and owners. Some of the main reasons
motivating such housing moves include dissatisfaction with the current unit or location, job changes, and
family or marital-​status changes. According to data from the latest American Housing Survey that took
place in 2017, the three top reasons for moving to a new house were dissatisfaction with the size and/or
quality of the current unit, dissatisfaction with the current neighborhood, and the creation of a new
household.
The demographic profile of new households plays an important role in shaping housing demand on the
margin. Since the impact of income is discussed separately, with the term demographics we refer here to
non-​income characteristics and primarily the age composition of an area’s population and households.The
composition of an area’s population in terms of life-​cycle stage and age does not only affect the number
of housing units demanded, because of differences in headship rates across age groups, but also the product
types demanded and tenure choice. For example, families with children tend to demand larger houses with
more bedrooms and bathrooms, while young adults with no children tend to demand smaller houses with
a smaller number of bedrooms and bathrooms. DiPasquale and Wheaton (1996) demonstrate the effect of
the age of household head on tenure choice. In particular, they present data indicating that homeownership
rates increase as the age of the household head increases.
Assessing the impact of demographic structures on demand for particular housing types requires analysis
of at least three critical demographic groups:

• 25–​34  year olds considered to include mostly young married households with no children, typically
demanding smaller and lower-​priced single-​family units or apartments.
• 35–​54 year olds considered to encompass launching and maturing (move-​up) families, typically demanding
larger and higher-​quality single-​family units depending on income.
• >55 year olds considered to include older households, referred to as empty nesters, typically demanding
smaller single-​family units, condominiums, or apartments depending on income.

Figure 5.2 presents the age distribution of the nation’s households by age of householder in the last five
decades and up to 2018. The figure shows the consistent increase in the percentage of older households
headed by persons aged 65 and older over the last 58 years. In particular, this percentage increased from
18% in 1960 to 26% in 2018, representing an addition of over 23 million households headed by persons
in this age group. The percentage of households headed by younger age cohorts of up to 29 years old,
who tend to demand smaller and affordable houses and apartments, has been decreasing consistently in
the last 38 years. In particular, the percentage of households headed by individuals younger than 25 years
old decreased from 8.1% in 1980 to 4.5% in 2018, while the percentage of households headed by individ-
uals aged 25–​29 years old decreased from 11.5% in 1980 to 7.5% in 2018. The percentage of households
headed by the age group of 35–​44, which represents mostly families with children at younger ages and
which demand single-​family houses, has decreased from 22.9% in 2000 to 16.9% in 2018. The percentage
of households headed by the age group of 45–​54, which represents mostly families with children at older
ages and also demand single-​family houses, has not registered a consistent long-​term trend as it has been
increasing from 1990 to 2010, but it decreased from 2010 to 2018. Finally, the percentage of households
headed by the age group of 55–​64, which represents the empty nest stage, has been rising consistently since
2000. In particular, it increased from 13% in 2000 to 18.8% in 2018. This increase reflects the continuing
entrance of late baby-​boomers in this age group.3

105
 Analyzing residential real estate markets

30%

25%

20%

15%

10%

5%

0%
Under 25 25 to 29 30 to 34 35 to 44 45 to 54 55 to 64 65 years
years years years years years and older
1960 1970 1980 1990 2000 2010 2018

Figure 5.2  Households by age of household head in the U.S.

Source: U.S. Census Bureau

Willingness to pay and, therefore, effective demand, for a specific housing type is also affected by tastes and
lifestyle preferences. Such preferences may be shaped by factors other than household income and the age of the
household head, such as education, occupation, household age composition, cultural orientation, and other
factors classified under the umbrella of consumer psychographics. It is very difficult, however, to incorporate
such factors in the macroeconomic analysis of a residential market and, especially, in a housing forecasting model.
The major problems in doing so are data availability and the difficulty in translating this psychographic informa-
tion in quantitative terms and in a way that would provide a consistent historical time-​series required to estimate
an econometric model. Psychographic information, which is typically gathered by surveying households that
are active in the local housing market, can be obtained from commercial data vendors and can certainly be very
useful in more specifically defining the character of a particular residential development and housing unit design.

Income, prices, and affordability

Bajari et al. (2013) verified the positive effect of household income on residential investment by both owners
and renters, while Sivitanides (2015, 2018) has confirmed statistically significant effects of income (proxied
by GDP) on house prices and by extension on housing demand. Household income is a primary deter-
minant of the tenure mode, quality, and type of unit effectively demanded by a household. Household
incomes, together with prices and mortgage rates, determine affordability. For this reason, the distribution
of an area’s households across the different income groups is one of the most important pieces of informa-
tion in determining effective demand for a given housing product of a certain price.
Assuming that homeownership is the preferred tenure type, all else being equal, increases in household
income should affect positively homeownership demand. Harmon (1988) verified the effect of income on
homeownership using data from the Panel Study of Income Dynamics. The estimates of the study point
to a short-​run income elasticity of 0.7 and a long-​run elasticity of about 1. Although there is no doubt
that the increase of an existing household’s income will make it less prone to rent, the effect of an increase
in average or median household income on an area’s aggregate demand for rental units is not as clear. The

106
 Residential real estate markets

reason is that income increases may also facilitate the formation of new households as they may allow adults
living with their parents or in non-​family households to move out and form their own households. The
majority of these newly formed households will most likely choose to rent rather than to own, thereby
contributing to increases in the demand for rental units.Therefore, the net effect on rental housing demand
will depend on how many renter households will shift to homeownership and how many new renter
households will be created from the decoupling of joint living arrangements.
In examining income trends, it is important to distinguish between increases in an area’s total income
and average or median household income because they could affect housing demand in different ways:

• Increases in an area’s total income do not necessarily imply increases in average household income because
they can be induced simply by increases in the number of households earning income. In fact, the
area’s average household income may decrease, even if total income increases, in the case where new
households with incomes below the average enter the area. Such developments would probably induce
increases in the demand for lower-​end owner-​occupied and rental housing.
• Increases in average household income, induced for example by real increases in wages, are likely to induce
changes in the distribution of housing demand across tenure types. Simply put, as the income of some
renters increases, they will afford to buy a house and they will likely do so, triggering an increase in the
demand for owner-​occupied housing. This should lead to an increase in the area’s homeownership rate,
if it is assumed that income increases do not contribute to new household formation. Otherwise, the
effect on the homeownership rate cannot be stated with certainty.
• Increases in average household income are likely to also induce increases in the demand for higher-​quality
units and decreases in demand for lower-​quality units within each tenure mode.

Given the several dimensions of housing demand that are affected by income, all income figures incorporated
in a residential market study should be examined and interpreted carefully. When dealing with income
statistics, it is often argued that the median is a better measure of central tendency than the average, because
the latter tends to overstate the income of an area’s typical household. This is attributed to the fact that
typical income distributions are characterized by a large number of households with low and medium
incomes and a small number of households with extremely large incomes. The median, which states the
income level that marks the middle of the area’s household income distribution, avoids this problem.4
When dealing with income figures the analyst needs to also distinguish not only between total income
and household income, but also between gross income and disposable income. Gross income refers to the total
income earned by a household, while disposable income refers to the income that is available to the house-
hold after federal and state income taxes, as well as social security taxes, are deducted. Carn et al. (1988)
point out that the latter is more relevant in assessing a household’s ability to pay for any desired commodity,
including housing. DiPasquale and Wheaton (1996) point out that a measure of a household’s longer-​term
income, referred to as permanent income, is even more relevant in gauging the demand for a durable good
such as housing. However, reliable estimates of such measures are difficult to obtain. For this reason, they
are not used in typical residential market studies.
Prices along with incomes and mortgage rates determine affordability. Affordability is the key indicator of
effective demand for housing and its measurement is typically linked to owner-​occupied housing. Effective
demand refers to the number of households that have the ability to pay for the desired housing unit given
their income and house prices. A  number of different affordability measures are typically available. For
households residing in California, for example, these would include:

1 . Mortgage lender’s affordability criterion

2. Affordability index of the National Association of Realtors (NAR)
3. Affordability index of the California Association of Realtors (CAR)

107
 Analyzing residential real estate markets

The mortgage lender’s affordability criterion is the typical measure used to evaluate how affordable a house
of a given price is to a household of a certain income level. This is the measure that a market analyst will use
to identify target income groups for a particular development of a given price range. Conversely, if spe-
cific income groups are targeted, then the analyst can use this criterion to determine first the prices that
will be affordable to these income groups and then the residential product types that are feasible at those
prices. The other two affordability measures are more aggregate measures, as they are based on the median
house price. Their purpose is to track broader affordability trends. Below, we discuss how each measure is
calculated.

(i) The mortgage lender’s affordability criterion is calculated as follows:

Affordability Criterion = Annual PITI /​Gross Income  ≤ 35% (varies)

where:

= Principal + Interest + Taxes + Insurance

PITI         

Principal + Interest    = Annual Mortgage Payment
         = Mortgage Constant × Loan Amount, or
         = Mortgage Constant × Loan-​To-​Value Ratio × House Price, where the mort-
gage constant can be calculated using (5.3)

(i / n )
Mortage Constant = (5.3)
1 − [1 / (1 + (i / n ))]mn

where:
i : mortgage rate
n : number of compounding periods per year (n=1 for annual payment calculation and n=12 for
monthly payment calculation)
m : term of the loan

Taxes      = Annual Property Tax Rate × House Price

Insurance  = Annual Rate × House Price
Note that in the case of renters the affordability criterion is calculated as:

Rental Expenditure /​Annual Gross Income ≤ 0.28 (greatly varies)

(ii) The National Association of Realtors (NAR) defines affordability in terms of the income needed to
qualify for the median-​priced home. For example, the NAR calculates the national homeownership
affordability index as the ratio of the median household income at the national level over the income
needed to qualify for the nation’s median-​priced home times 100. It is useful to track this index over
time and to compare affordability in the local market with affordability in the nation. All else being
equal, increasing affordability in the area could imply an increasing demand for owner-​occupied
housing, while decreasing affordability could be a signal for decreasing demand. In addition, the
area’s economic development agency may find it useful to know if local housing is more affordable
compared to the national average.

108
 Residential real estate markets

(iii) The California Association of Realtors also publishes its own affordability index, which states the per-
centage of all California residents who are able to afford the state’s existing median-​priced home. The
index is derived by juxtaposing the income distribution for all Californians (not just those who are in
the market for owner-​occupied housing) against the median home price. The affordability percentage
should be much higher among the income groups that are actually active in the market for owner-​
occupied housing.

Relative prices are also important determinants of the demand for a specific tenure mode and housing
product. For example, for many households, rental and owner-​occupied housing are close substitutes and
one factor that enters their decision to choose one over the other is the relative cost.Thus, for a given house
price level in the owner-​occupied market, an increase in apartment rents should motivate more households
to choose owner-​occupied housing, thereby inducing an upward shift in the demand curve for the latter
product type. On the contrary, all else being equal, an increase in house prices should induce an upward
shift in the demand for rental housing.
DiPasquale and Wheaton (1996) point out that in comparing the costs of renting versus the costs of
owning a house, quality-​controlled indices of rents and house prices need to be used.5 If time-​series data of
such indices is available when modeling econometrically either the owner-​occupied or the rental housing,
it would be prudent to incorporate the ratio of or the difference in these two indicators in the demand
equation. Similar relative price impacts should be in effect across quality levels and housing types that are
close substitutes (see example below).

An example of the effects of changing substitute prices

To better understand the effect of changing prices of substitutes on demand, consider the following three sub-
stitute types of housing:

New Single-​Family Detached (ND)

Used Single-​Family Detached (UD)
Townhouses (T)

Exogenous shock  ND prices up  UD prices up  T prices up

If due to zoning constraints (development moratoria, for example) prices of ND units go up, then some
households, which would otherwise purchase this type of unit, will no longer be able to afford it and they will
be demanding the closest and lower-​priced substitute, that is, UD units. As a result of this shift in demand in
the UD segment, prices for this type of unit will increase too. This in turn will shift demand from household
segments that will no longer be able to afford this type, to the closest and lower-​priced segment T, which eventu-
ally will also experience price increases.This filtering process across the different quality segments of the housing
market is well described in O’Sullivan (2009).

Conclusion: Trends in one segment filter through eventually to other segments along the quality/​amenity con-
tinuum. The implication of this argument for housing market analysis is that no narrowly defined segment
should be studied in isolation from the broader aggregate market.

109
 Analyzing residential real estate markets

Mortgage rates and inflation

MORTGAGE RATES  For most households the only venue for buying a house is through financing. Thus,
typically up to 80% of the house purchase price is financed through a loan. In financing terminology,
loans that are collateralized by real estate assets are called mortgages. The interest rate on these loans,
or mortgage rate, as it is typically referred to, can have a significant impact on the level of demand for
owner-​occupied housing since it determines the level of the required monthly payment. Obviously, lower
mortgage rates should encourage demand for owner-​occupied housing, since they allow for a lower
monthly mortgage payment, thus making homeownership more affordable. Conversely, rising mortgage
rates should discourage homeownership. Bajari et al. (2013) verified the negative effect of user cost of
capital and, therefore, mortgage rates, on housing investment by owners and renters.6 Apergis and Residis
(2003) also confirmed a negative effect of interest rates on house prices, pointing also to a negative effect
on housing demand.

INFLATION  Inflation can affect housing demand in several ways. Firstly, rising inflation, without
counterbalancing increases in household income, will reduce household purchasing power and have a
negative effect on demand for housing. Secondly, it induces higher nominal mortgage rates and, as such,
it discourages homeownership demand. Thirdly, to the extent that is perceived by housing buyers as an
indicator of future house price appreciation, it helps lower the perceived after-​tax user cost of capital, especially
for those in high-​income tax brackets and, as such, it encourages homeownership demand.7 Apergis and
Residis (2003) confirmed a positive effect of inflation on house prices, which points to such a positive
effect on housing demand. A simple summary measure of the after-​tax user cost of capital to homeowners
can be calculated using formula (5.4).

U = (i–​ti) –​ g = (1–​t)i –​ g (5.4)

where
i : nominal mortgage rate
t : marginal income tax rate
g : expected appreciation rate

DiPasquale and Wheaton (1996) present a more complete formula for the calculation of the user cost of
capital that accounts also for property tax payments by including in the equation the property tax rate, tp:

U = (i+tp) (1–​t) –​ g (5.5)

It should be noted that neither (5.4) nor (5.5) reflect the full cost of homeownership. According
to the  U.S. Department of Housing and Urban Development (2000), such a calculation should
account also for maintenance costs, forgone interest earnings on the equity invested in the house, and
depreciation.
Table 5.1 and Figure 5.3 present estimates of the real after-​tax user cost of capital for homeowners
over the period 2000–​2018 using formula (5.4). In applying this formula the figures in the inflation
column are used as proxies for expected house price appreciation in each period. Ideally, actual house
price appreciation figures should be used if available. As formula (5.4) implies, the value of the benefits
of homeownership are higher and, therefore, the real user cost of homeownership (as percentage of
the value of the house) is lower for households with higher incomes and higher marginal tax rates.
The user cost of capital also decreases with lower mortgage rates and higher expected house price
appreciation rates.

110
 Residential real estate markets

Table 5.1  Real after-​tax user cost of capital 2000–​2018

Year Mortgage rate1 Inflation User cost of capital for a marginal income tax rate
of 24%2
2000 7.9% 3.4% 2.6%
2001 6.9% 2.8% 2.5%
2002 6.4% 1.6% 3.3%
2003 5.7% 2.3% 2.0%
2004 5.7% 2.7% 1.6%
2005 5.9% 3.4% 1.1%
2006 6.5% 3.2% 1.7%
2007 6.4% 2.9% 2.0%
2008 6.1% 3.8% 0.8%
2009 5.1% -​0.3% 4.2%
2010 4.8% 1.6% 2.0%
2011 4.6% 3.1% 0.3%
2012 3.7% 2.1% 0.7%
2013 3.8% 1.5% 1.5%
2014 4.1% 1.6% 1.5%
2015 3.9% 0.1% 2.8%
2016 3.7% 1.3% 1.5%
2017 4.0% 2.1% 0.9%
2018 4.6% 2.4% 1.0%
1
National average contract interest rate on conventional single-​family mortgages.
2
Marginal income tax rate for income tranch of $82,501 to $157,500. The user cost of capital calculation is based
on formula 5.4.
Sources: FHFA, Census Bureau

4.5%

4.0%

3.5%

3.0%
User cost of capital

2.5%

2.0%

1.5%

1.0%

0.5%

0.0%
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018

Figure 5.3  User cost of capital in the U.S. 2000–2018

111
 Analyzing residential real estate markets

As it can be seen from Figure 5.3, from 2000 until 2008 the user cost of capital was mostly declining
primarily due to the declining mortgage rate. In 2009 there was a big spike pushing the user cost of cap-
ital above 4% due primarily to negative inflation during that year. However, in 2011 it registered its lowest
value at 0.3%, as a result of a relatively low interest rate (4.6%) and an inflation rate above 3%. After 2011
and until 2015 it has been rising due to a falling inflation rate and a relatively stable mortgage rate. With
the 2015 inflation rate dropping practically down to zero, the user cost of capital registered its third peak
in this decade at 2.8%. In 2016 and 2017, however, it decreased as inflation increased faster than mortgage
rates, and in 2018 it increased slightly to 1%.

Expectations
Besides demographics, prices, and incomes, expectations for prices and interest rates affect considerably
housing demand levels. Expectations of rising house prices and/​or rising interest rates may motivate some
households to buy a house sooner rather than later. Expectations regarding future house price movements
are largely myopic and, as such, they are shaped by past movements in house prices. Expectations regarding
interest rates are not necessarily myopic as they are shaped by the frequently announced intended monetary
policies of the Federal Reserve.
Consumer expectations can be revealed through the analysis of past behavior or through consumer
surveys. Case and Shiller (1989) have examined the behavior of homebuyers in boom and post-​boom
markets and found that past price movements were important in shaping their expectations. Anundsen
and Heebøll (2014) have also confirmed a statistically significant and geographically differentiated effect
of expectations (proxied by lagged house price appreciation) across U.S. metropolitan areas. In particular,
their findings point towards a greater effect of lagged house price appreciation in MSAs characterized by
constrained housing supply, larger population, and availability of non-​recourse lending. Using the simple
stock-​flow modeling framework, DiPasquale and Wheaton (1996) find that with myopic expectations,
house prices continue to rise after a positive demand shock, even as new residential construction enters
the market, because expectations of continuing house price appreciation trigger further demand increases.
As a result, construction overshoots and the process generates repeating cycles in house prices and
construction.

New housing supply
In the rental housing market, the term total supply refers to all rental units (occupied or vacant) in
the area under consideration. Likewise, in the owner-​occupied housing market it includes all existing
housing units that are occupied by their owners and those that are up for resale. However, when ana-
lyzing marginal demand for housing, the relevant supply concept is the one that comprises the units
that compete for the additional housing demand expected to be generated during the period of analysis.
For the lack of a better term, let’s refer to it as available supply. Thus, in the case of the rental housing
market, available supply during a given period encompasses the new rental units completed and the units
that are vacant during that period. Likewise, in the case of owner-​occupied housing, the available supply
encompasses the new units completed and the existing units that are up for resale during the period
under consideration.
In the discussion that follows, we will refer to new units as completions or new construction and
to housing units that are up for resale as resale units. It should be noted that in either market (rental or
owner-​occupied) a more refined calculation of available supply needs to take into account the structural
vacant stock. This issue is discussed in more detail in the section on accounting techniques in the next
chapter.

112
 Residential real estate markets

New residential construction

A careful assessment of new housing construction in the short and medium term is important because its
underestimation may lead to erroneous perceptions of the strength of the market and the potential per-
formance and viability of the project.

SHORT-​TERM NEW CONSTRUCTION  The number of new housing units likely to enter the market in
the short run can be estimated by surveying the residential projects that are under construction. Such
information can be obtained from commercial data vendors. Housing projects that have broken ground are
typically referred to as starts and represent the most reliable indicator of the number of units most likely
to enter the market in the short run. The Bureau of the Census provides quarterly data on single-​family
and apartment starts for the nation and the four major regions. Such data, however, are not available at the
metropolitan-​area level.The most readily available data for assessing short-​term new housing completions at
this geographical level are permit data provided by the Bureau of the Census for more than 100 metropolitan
markets. These data are provided monthly for structures with one, two, three, four, and five or more units.

LONGER-​TERM NEW CONSTRUCTION  In the longer term, planned residential projects may provide some
indications of new housing supply, but they can by no means provide a reliable forecast of the number
of units that will be eventually completed since many of these projects may be canceled, postponed or
never be built. Furthermore, many new housing projects may originate after the time of analysis and still
have enough time to be completed within the project’s planning horizon. The amount of land zoned for
residential uses may help the analyst establish the maximum possible number of units that can be built
in the area, but can in no way provide the basis for year-​by-​year forecasts of new housing supply, unless
availability of such land is extremely limited.
The best insights about the medium-​and long-​term prospects of new residential construction can be
obtained by understanding and forecasting likely movements in its determinants and by quantifying their
likely impact through econometric modeling. A major endogenous determinant of residential construction
is, of course, the price of housing, in the case of owner-​occupied housing, and rental rates, in the case of
rental apartments. For a given level of residential development costs, rising house prices and rents will
allow for higher profits and, thereby, trigger a greater level of residential development. The major exogenous
determinants of new residential construction, both owner-​occupied and rental, include:

• Availability and cost of factors of production

These include construction capital, land, labor, and construction materials. Obviously, keeping housing
rents and prices constant, rising residential land prices and other factor costs should reduce development
profits and discourage construction of new residential properties. Similarly, reduced availability of these
factors of production should also limit residential development activity in an urban area. Ihlanfeldt and
Mayock (2014) confirmed the effect of land availability on the price elasticity of housing supply.
  The availability of capital is an equally important determinant of new residential construction as the
availability of land. This is affected by the return on residential investments relative to other investment
vehicles, as well as the uncertainty and risk associated with such investments. For example, an anemic
performance of residential investments relative to investments in office, industrial, and retail properties
may reduce capital availability for new residential construction. Similarly, an increase in uncertainty
regarding residential property performance (due, for example, to demographic forces) relative to other
property types might also result in a decrease of the amount of capital available for new residential con-
struction. Studies on apartment cap rates suggest that investor perceptions of residential market risk vary
across the nation’s metropolitan areas. Similar analysis of capitalization rate movements through time

113
 Analyzing residential real estate markets

for other property types, such as office, indicate that investor perceptions of risk vary also through time,
depending on indicators of local market strength and the historical volatility of the market (Sivitanidou
and Sivitanides, 1999).
  After the collapse in real estate values during the 1991 recession, most lenders have adopted stricter
underwriting standards for real estate loans. As a result, the availability of development capital for com-
mercial real estate (including apartments) has been more restricted in the 1990s compared to the 1980s.
However, the low interest rates following the 2001 recession and the rapidly rising house prices in the
subsequent years up to 2007 fueled excessive lending, including a significant amount of sub-​prime loans
(high-​risk loans to individuals with questionable ability to repay them). The latter was one of the main
causes of the global financial crisis that brought down in the third quarter of 2008 Lehman Brothers,
one of the largest investment banks in the world.
• Expectations
Positive expectations about the economy (GDP growth), housing demand, vacancy rates, prices/​rents,
and the return on residential investments should encourage new construction. Positive expectations are
usually triggered by past movements in any of the aforementioned factors, as well as by growth-​inducing
government policies.
• Uncertainty or risk
Putting aside its effect on residential capital availability, uncertainty and risk should have a direct impact on
the behavior of the home-​building industry. In particular, greater uncertainty regarding the future health of
the local economy, the strength of future housing demand, and the prospects for house price/​rent increases
should discourage local and national homebuilders from engaging in new residential construction. Commonly
used indicators for gauging the risk and uncertainty characterizing residential markets include measures of
the volatility of the local economy and housing market. Other indicators of the short-​term risk of a resi-
dential market include the vacancy rate, and sales time or time-​on-​the market. According to DiPasquale and
Wheaton (1996), there is empirical evidence supporting a strong negative relationship between housing con-
struction and sales time, even after controlling for movements in interest rates and prices.
  Ihlanfeldt and Mayock (2014) found that the local fiscal and regulatory environment were one of the major
determinants of geographic variations in the elasticity of housing supply.As one would expect, in jurisdictions
with stricter growth controls and development constraints, housing supply was more price inelastic.

The supply of resale homes

As indicated before, the activity in the resale market is also relevant in assessing competing supply in the
market for owner-​occupied housing. Summarized in Table 5.2, the major determinants of the activity in
the resale market include:

• Mobility patterns, life cycle, and income changes

Mobility patterns affect the availability of resale homes through outmigration. For example, at any given
period, the greater the number of households migrating out of the metropolitan area, the greater the
number of owner-​occupied units entering the resale market (all else being equal). Life-​cycle changes
can also impact the availability of resale homes through changes in the tenure mode or the quality of
the housing unit demanded. For example, during a period with a relatively larger number of married
couples entering the full-​nest stage and moving to larger units, there will be an increase in the number
of resale units of smaller size. On the contrary, an increasing number of maturing families becoming
empty nesters would induce an increase in the number of resale units of larger size, since such a life-​
cycle change is typically associated with a move from larger units to smaller ones. Finally, income
increases can influence the availability of resale homes as they may motivate households to move from
lower-​quality homes to larger and higher-​quality units.

114
 Residential real estate markets

Table 5.2  Factors affecting the supply of resale units and potential effect

Factor Effect Type of units

Mobility patterns, life-​cycle changes
• Outmigration + Any type
• Transition of families to full-​nest stage + Smaller units
• Transition of families to empty-​nest stage + Larger units
Income increases + Smaller, lower-​quality units
Mortgage rate increases –​ Any type
Relative prices
•  Increases in prices of new units –​ Existing units
•  Increases in prices of larger units –​ Smaller units
Expectations of higher prices –​ Any type

• Mortgage rates and relative prices

Rising mortgage rates may reduce the supply of resale homes as they shape expectations of weaker
demand on the part of sellers and discourage them from putting their units in the market (Carn et al.,
1988). Furthermore, higher mortgage rates may prevent some homeowners from putting their existing
house on the market because simply they can no longer afford to buy the new house they were planning
to move into. Higher mortgage rates will also discourage demand for resale homes and owner-​occupied
housing units in general.
Higher prices of substitutes may also affect the availability of resale units because they reduce demand for
other units by sellers, who in turn refrain from putting their unit up for sale. For example, increases in the
prices of new owner-​occupied housing units may discourage several homeowners who were planning to
sell their existing home and buy a new one. At the same time, such development may boost the demand for
resale homes, as newly formed households, or renters that were planning to buy a new-​build house, may now
need to turn their attention to the resale market, as they may no longer be able to afford to buy a new unit.
• Expectations
Expectations for higher prices negatively affect resale activity, as homeowners may want to refrain from
putting their unit on the market until prices rise further.

Chapter summary
In this chapter, we first reviewed briefly the residential market analysis process by outlining the main
stages of its two major components, i.e., the macroeconomic analysis and the project-​specific analysis.
Subsequently, we focused on basic housing market economics and, specifically, on housing demand and
housing supply components and drivers.

• Housing demand can be segmented at least along three lines: tenure type, housing type, and unit
quality.
• There is substitutability between tenure types since if a household cannot afford to own a house it will
need to rent one.
• The critical factors affecting all three dimensions of housing demand (besides price) include house-
hold formation, demographic structures, income, mortgage rates, and expectations about prices and
interest rates.

115
 Analyzing residential real estate markets

• When analyzing income trends and their effect on housing demand, one needs to distinguish between
total income and household income, as well as between gross income and disposable income.
• An area’s available supply of rental housing consists of newly completed apartment units (i.e., completions)
and existing vacant units. Likewise, the available supply of owner-​occupied housing consists of newly
completed units and resale units.
• Short-​term completions of rental or owner-​occupied units can be determined from starts, that is,
projects that are already under construction, or permits issued.
• The medium-​and longer-​term completions of new rental and owner-​occupied units are driven by
factor costs, expectations regarding the economy, housing demand, and prices/​rents, as well as the avail-
ability of capital.
• The supply of resale housing units is driven by mobility patterns, life-​cycle stage, income changes, mort-
gage rates, relative prices, and expectations.

Questions
1. Describe briefly the major stages and sub-​stages of residential market analysis, outlining the usefulness
of each stage.
2. Discuss the three major dimensions of housing demand and their implications for housing market
analysis.
3. Discuss briefly the major factors driving housing demand. Comment on how each factor influences
demand for housing and the expected effect.
4. Discuss the prospects of each of the main factors that affect housing demand in your town/​city and
assess how total demand for housing in your city will behave in the next three years and why. Do you
expect it to remain stable or to be changing? If it will be changing, towards which direction will it be
moving (upwards or downwards) and at what rate (fast or slow)?
5. Discuss of the prospects of each of the main factors that affect the new supply of housing in your town/​
city and assess the most likely movements of total new supply in your city in the next three years.

Notes
1 Schmitz and Brett (2001) suggest that some analysts prefer to use employment growth projections instead of house-
hold growth projections in assessing residential demand prospects. The underlying rationale of this approach is that
new demand for housing comes mostly from workers moving into the area.
2 Headship rate is defined as the percentage of persons within an age group that are heading a household.
3 According to Fanning, Grissom, and Pearson (1994), the baby-​boom generation includes persons born between
1946 and 1966.
4 The interpretation of the median household income is that 50% of the area’s households have an income equal to
or below that level and 50% equal to or above that level.
5 Quality-​controlled price or rent indices ensure that the quoted price or rent refer to a constant-​quality house.
6 Bajari, Chan, Krueger, and Miller (2013) estimated the user cost of capital as the mortgage rate minus current house
price appreciation.
7 Interest rates rise with higher inflation in order to compensate for losses in purchasing power.

References and additional readings

Anundsen, A. K. and C. Heebøll. 2014. Regional US House Price Formation:  One Model Fits All? Norges Bank:
Working Papers, 2014:8, 155–​171.
Apergis, N. and Rezitis A. 2003. Housing Prices and Macroeconomic Factors in Greece: Prospects within the EMU.
Applied Economic Letters, 10:9, 561–​565.

116
 Residential real estate markets

Apgar, W. C. and G. Masnick. 1991. Some Simple Facts About the Demand for New Residential Construction in the
1990s. Journal of Real Estate Research, 6:3, 267–​292.
Arestis, P. and A. R. Gonzalez. 2013. Modelling the Housing Market in OECD Countries. International Review of Applied
Economics, 28:2, 131–​153.
Bajari, P., P. Chan, D. Krueger, and D. Miller. 2013. A Dynamic Model of Housing Demand: Estimation and Policy
Implications. International Economic Review, 54:2, 509–​442.
Boehm, T. P. and A. M. Schlottmann. 2011. Market Conditions and Housing Choices:  A Comparison of
Homeownership Across Three Decades. Real Estate Economics, 39:3, 547–​600.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Housing Market Economics. Chap. 7 in Real Estate Market
Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice Hall.
Case, C. and R. Shiller. 1989. The Behavior of Home Buyers in Boom and Post-​Boom Markets. New England Economic
Review, November/​December,  29–​46.
DiPasquale, D. and W. Wheaton. 1996. The Market for Housing Units. Chap.  8 in Urban Economics and Real Estate
Markets. Englewood Cliffs, NJ: Prentice Hall.
Duffy, D., D. Byrne, and J. FitzGerald. 2014. Alternative Scenarios for New Household Formation in Ireland. ESRI
Quarterly Economic Commentary. Spring, Dublin: The Economic and Social Research Institute.
Fanning, S., T. Grissom, and T. Pearson. 1994. The Market Analysis Process for an Existing Apartment Complex.
Chap. 14 in Market Analysis for Valuation Appraisals. Chicago, IL: Appraisal Institute.
Harmon, O. R. 1988. The Income Elasticity of Demand for Single-​Family Owner-​Occupied Housing: An Empirical
Reconciliation. Journal of Urban Economics, 24:2, 173–​185.
Green, R. K., S. Malpezzi, and S. K. Mayo. 2005. Metropolitan-​Specific Estimates of the Price Elasticity of Supply of
Housing, and their Sources. American Economic Review, 95:2, 334–​339.
Ihlanfeldt, K. and T. Mayock. 2014. Housing Bubbles and Busts:  The Role of Supply Elasticity. Land Economics,
90:1,79–​99.
Ihrke, D. 2014. Reason for Moving: 2012 to 2013. Suitland, MD: US Census Bureau, June.
O’Sullivan, A. 2009. Why is Housing Different? Chap. 13 in Urban Economics. New York: McGraw Hill
Schmitz, A. and D. Brett. 2001. Residential Development. Chap.  3 in Real Estate Market Analysis. Washington, DC:
Urban Land Institute.
Sivitanides, P. 2015. Macroeconomic Influences on Cyprus House Prices. Cyprus Economic Policy Review, 9:1, 3–​21.
Sivitanides, P. 2018. Macroeconomic Drivers of London House Prices. Journal of Property Investment and Finance, 36:6,
539–​551.
Sivitanidou, R. and P. Sivitanides. 1999. Office Capitalization Rates:  Real Estate and Capital Market Influences.
Journal of Real Estate Finance and Economics, 18:3, 297–​322.
U.S. Census Bureau. 1998. Historical Time Series HH6, Average Population per Household and Family:  1940 to Present.
Washington, DC: U.S. Census Bureau.
U.S. Department of Housing and Urban Development. 2000. U.S. Housing Market Conditions. Washington, DC: U.S.
Department of Housing and Urban Development.

117
newgenprepdf

6
MACROECONOMIC ANALYSIS
OF RESIDENTIAL REAL
ESTATE MARKETS
Accounting techniques

Introduction 118
Macroeconomic analysis framework 119
Market area definition 119
Define the primary market area by drawing a curve representing equal
commuting time from the site 119
Identify major employment nodes within the primary market area 121
Establish the competitive market area 121
Refine the definition of the competitive market area 121
Adjust boundaries to coincide with census tract boundaries 121
Time frame of the analysis 122
Assessing residential market strength 122
Accounting techniques: the aggregate approach 123
Accounting techniques: the disaggregate approach 128
Chapter summary 135
Questions 135
Project #2: Estimating effective residential demand using the accounting approach 136
The case 136
Development scenarios and price data 136
Mortgage requirements and affordability measures in 2023 136
Target demographic groups and preferences 136
Market area data 137
Questions 137
What to turn in 138
References and additional readings 138

Introduction
In the previous chapter, we outlined the two major components of residential market studies: a) the macro-​
analysis component that focuses on the state and strength of the broader market within which a housing

118
 Macro analysis: accounting techniques

project competes, and b) the micro-​analysis component focusing on the analysis of the competitive pos-
ition and expected performance of a specific residential development.
In this chapter, we focus on housing market analysis from the macro perspective. This type of analysis
is very important when contemplating residential development or investment projects because the per-
formance of individual housing units is significantly affected by the performance of the broader residen-
tial market within which they compete. Because of short rental contracts, rental housing is much more
exposed to market fluctuations compared to other types of commercial real estate that are characterized by
multi-​year leases. In the previous chapter, we discussed broad housing market indicators one should look at
in evaluating market strength. In this chapter, we continue by laying out a broad analytical framework for
assessing residential market strength.

Macroeconomic analysis framework

As indicated in Box 6.1, the macroeconomic-​analysis framework for residential development projects
incorporates three major steps: a) definition of the project’s market area, b) determination of the time frame
of analysis, and c) application of simple accounting and/​or more sophisticated econometric techniques for
assessing residential market strength. With the exception of econometric techniques, each of these steps
is discussed in the sections that follow. Due to the length of the discussion, the basics of the econometric
approach and its application to residential markets are presented separately in the next two chapters.

Market area definition

Define the primary market area by drawing a curve representing
equal commuting time from the site
If a residential site is given, the first step is to define the primary market area, that is, the area from which
the project is expected to primarily draw buyers, if it is owner-​occupied housing, or tenants, if it is rental
housing. Such an area can be identified by drawing a contour representing a maximum commuting time
from the site (see Figure 6.1).The underlying hypothesis is that there is a very strong link between place of
residence and place of work, and that an individual or household head will demand housing located within
a reasonable commuting distance from his/​her workplace. Research by Jones et al. (2012) has demonstrated
that housing market area delineations based on commuting patterns resulted in greater reduction of the
standard error of estimates of hedonic house price models compared to housing market area delineations
based on migration flows or local jurisdiction boundaries.
Within this context, the analyst needs to examine carefully worker commuting patterns in the area.
Local transportation surveys, which may be available from local planning departments or university trans-
portation centers, may provide insights with respect to actual commuting patterns to the area’s major
employment nodes. Examination of relocation patterns (i.e., who are the buyers at various locations in
the area and where they work) may provide some additional insights into the residence–​workplace rela-
tionship. More detailed information on actual commuting patterns at the metropolitan area level can be
also obtained from the decennial census, which provides journey-​to-​work statistics. Such statistics include
average commuting time for the whole area and by county of origin (central or suburban), as well as a
percentage breakdown of commuters by travel time. Of course, the more outdated this information is the
greater the caution that should be applied when using it. More recent commuting time estimates at the
metropolitan and micropolitan area level are provided by the American Community Survey. According to
the most recently released data by the Census Bureau regarding commuting times, the mean travel time
to work in 2017 at the national level was 26.4 minutes and ranged from 11.5 minutes in the Vernon, TX,
micropolitan statistical area to 38.6 minutes in East Stroudsburg, PA, metropolitan statistical area.

119
 Analyzing residential real estate markets

Box 6.1  Residential macroeconomic analysis framework

The methodological steps that need to be carried out when analyzing broader housing market strength include:

1. Define the market/​competitive area

a) Define the primary market area by drawing a contour representing equal commuting time from the site
The analysis of worker commuting patterns in the area using data from the latest census as well as the
latest American Community Survey will help better define the primary market area.
b) Identify major employment nodes within the primary market area
The analysis of the character and occupational structure of employment nodes located within the
primary market area will help better define potential demand groups and the development scenario.
Understanding the economic sectors represented by the companies located in these nodes may help
assess better the income level of the individuals that will demand housing at the particular site.
c) Establish the competitive market area by encompassing competing residential sites that are located within
commuting distance around the identified employment nodes
d) Refine the definition of the competitive market area taking into account socioeconomic constraints and political
boundaries
e) Adjust boundaries to coincide with census tract boundaries

2. Define the time frame of analysis

3. Assess housing market strength

a) Apply accounting approaches for the estimation of demand-​supply gaps, including:
i. Aggregate approach
ii. Disaggregate approach.
b) Apply econometric techniques for forecasting housing demand, supply, and rents/​prices

Once an in-​depth understanding of the area’s commuting patterns is gained, the primary market area of the
site can be delineated by first determining a maximum time for daily commuting and then drawing a con-
tour of equidistant locations, which correspond to this maximum driving time from the site. As indicated in
Figure 6.1, the shape of this area will be largely determined by the major transportation arteries (including
freeways, highways, and subway routes) that are conveniently accessible from both the site under consider-
ation and the major employment nodes located within the maximum commuting distance from it.

Site

Major employment
center

Freeway

Figure 6.1  Defining the market area

120
 Macro analysis: accounting techniques

Identify major employment nodes within the primary market area

Once the site’s area of influence has been delineated, the analyst needs to specifically identify the major
employment concentrations located within this area. Obviously, the employees working in these employ-
ment centers represent the project’s primary demand segment from which prospective buyers or renters
are more likely to be drawn to the site under consideration. According to Rabianski (2006), the spatial
relationship between the place of work and the place of residence represents the primary demand link for
households/​individuals demanding apartments. Thus, information on the character of these employment
nodes—​in terms of employment specialization and occupational structure—​may help better define target
demand groups in terms of income levels and housing preferences and refine the development idea.

Establish the competitive market area

The procedure just described can help establish a housing project’s market area from the demand side.
This is not enough, however, because the geographic definition of the market area needs to also take
into account the competing supply. In order to do this the analyst needs to take into account the full area
of influence of the employment centers that are located within the project’s primary market area. This
is necessary because all vacant or newly developed residential units within a reasonable commuting dis-
tance around these centers will be in essence competing with the units that will be developed at the site
under consideration. Within this context, vacant housing units, housing projects under construction, and
planned residential developments, located within these areas, should be of major interest to the analyst
when assessing competition.

Refine the definition of the competitive market area

The boundaries of the competitive market area need to be adjusted to take into account socioeconomic
constraints and political boundaries, such as jurisdictional and school-​district boundaries. Homogeneous
neighborhoods and jurisdictions with socioeconomic or other characteristics that are clearly incompatible
with the project’s target group can be excluded from the competitive market area. For example, if the pro-
ject involves housing for higher-​income households, areas with low-​income housing can be excluded from
the geographic delineation of the competitive area.

Adjust boundaries to coincide with census tract boundaries

Finally, in order to complete the geographic definition of the market area in a way that would make it feas-
ible to collect the data required for analytical purposes, the boundaries of the area need to be adjusted to
coincide with census tract defined boundaries, or political boundaries such as county boundaries.
It should be emphasized that the so-​defined market area is the project’s primary market area, on which a
greater focus needs to be applied when assessing demand, supply, and rents/​prices; however, this is not neces-
sarily the only area that macroeconomic residential analysis should focus on. For example, in the case of larger
urban areas, the market area defined by the process just described may be smaller than the metropolitan area
within which the site is located. In such cases, the analyst is strongly advised to examine first the broader
housing conditions and prospects within the metropolitan area and then try to understand how the metro-
politan area’s fundamentals relate to the project’s primary market area. Metropolitan area analysis will provide
a solid broader framework and perspective of the local housing market that will help better evaluate demand,
supply, and rent/​price data and, more importantly, make more accurate forecasts for the primary market area.
The use of the metropolitan area as the broader market area for a residential project is necessary if the
analyst wants to take advantage of the superior forecasting insights provided by econometric modeling.

121
 Analyzing residential real estate markets

Such a modeling approach requires sufficient historical information on fundamental housing market
variables, such as rents/​prices, stock, completions, and demand, which is rarely available for any geograph-
ical level smaller than the MSA level.
Performing macroeconomic analysis at the metropolitan-​area level, however, is not dictated by data
limitations only; it is also dictated by economic arguments that suggest that a housing project’s value is
linked to a significant extent to metropolitan-​area prospects. First of all, economic decline or growth does
not typically affect a single well-​defined location within a metropolitan area. Unless the decline or growth
in the metropolitan economy is due to a very localized incidence, which is rarely the case, the resulting job
losses or gains will not be confined within a single employment center but will be widely spread across
most of the area’s employment concentrations. This suggests that changes in the metropolitan economy
will affect housing demand at many locations, including the project’s primary market area.This argument is
completely consistent with the rationale underlying the designation of metropolitan statistical area bound-
aries by the U.S. Census Bureau. As DiPasquale and Wheaton (1996) indicate, such boundaries are defined
to encompass an integral labor market, as reflected in the area’s commuting patterns.
Second, it can be argued that within the urban continuum, a chain-​reaction location-​substitutability effect
may be at work, which, eventually, transfers price changes at any residential neighborhood to every other
residential location of similar quality within the metropolitan area. For example, house price changes at more
distant residential neighborhoods should influence prices in other similar-​quality residential neighborhoods
that are located at reasonable distances from them; similarly, changes in the latter will then affect prices in
other competing residential neighborhoods and so on. This argument is supported by empirical evidence
presented by Goetzmann and Spiegel (1997), which confirmed that the covariance of housing returns
between similar-​income neighborhoods in San Francisco was considerably higher (suggesting similar house
price trends) than the covariance of housing returns among neighborhoods with significantly different
median household income. In the longer term, any noticeable house price changes in one section of the
metropolitan area should eventually reverberate through a chain reaction to all other locations within it.1

Time frame of the analysis

The time frame of the analysis will depend on the nature of the residential development project under
consideration and specifically on:

1. The size of the planned project and the lease-​up/​sales period

Large projects that need to be phased over a number of years will definitely require a longer planning
horizon. The risk born by the investor and developer in this case is that market forecasts for years well
ahead of the time of analysis are more likely to be less accurate than forecasts for years closer to the
time during which the market study is carried out.
2. The type of planned project and investment strategy
In the case of owner-​occupied housing, the project’s time frame should be defined from the present up
to at least the anticipated time of completion. In the case of rental housing, the analysis time frame
should include an additional time interval to allow for leasing up of the building in order to maximize
sale value. A longer time horizon will need to be adopted if the investor desires to hold the property
for a longer period in order to take advantage of its income-​generating ability.

Assessing residential market strength

Starting from a preliminary residential product definition, macroeconomic analysis focuses on the question
of how strong the area’s housing market will be at the expected time of sale or during the holding
period. Obviously, regardless of how small the planned development or investment is, the focus is on the

122
 Macro analysis: accounting techniques

broader housing market, as broader market forces are likely to impact to a significant extent all residential
developments within the market area. The analyst can use two types of techniques for assessing broader
residential market strength:

1. Accounting techniques
2. Econometric techniques.

In residential market analysis, accounting techniques are typically used to estimate housing demand and
demand-​supply gaps based on simplistic assumptions. Econometric techniques are used to provide more
sophisticated forecasts of housing demand, supply, and rents/​prices through a scientific quantification of the
dynamic interrelationships and interactions among the key housing market variables. Below we discuss the
basic steps of the aggregate and disaggregate accounting techniques, as they apply to the residential market,
and point out their shortcomings.

Accounting techniques: the aggregate approach

In market studies, the term “gap” is often used to describe the difference between projected demand and
supply. As indicated in Figure 6.2, aggregate accounting approaches to estimating such a gap in the residen-
tial real estate market involve three basic steps:

1 . Estimate new housing demand from household growth.

2. Estimate expected available supply of residential units (owner-​occupied or rental).
3. Estimate the housing “gap” as the difference between expected household growth and available supply.

Below we discuss each of the steps using the example presented in Tables 6.1–​6.3.

ESTIMATE NEW DEMAND FROM HOUSEHOLD GROWTH  The first step is to estimate new demand from
expected household growth. When using the aggregate approach, the typical procedure is to obtain either
a reliable forecast of household growth or forecasts of population (POP) and household size (HS) for the
area under consideration. Such forecasts can be obtained from reputable econometric forecasting firms.2
In the example described in Table 6.1, the analyst has an estimate of the actual population in the metro-
politan area as of 2000 and a forecast for 2002. An estimate for the 2000 household size along with a fore-
cast for 2002 is also provided. From these figures, the analyst can calculate the expected household growth
(DH) over the period 2000–​2002 as:

DH = HH(02) –​  HH(00)
HH2000 = POP(00)/​HS(00) = 8,676,000/​2.69 = 3,225,279
HH2002 = POP(02)/​HS(02) = 8,936,280/​2.62 = 3,410,794
DH = 3,410,794 –​3,225,279 = 185,515

It should be emphasized that the household-​size forecasts used in the above calculation should be viewed
with great caution—​especially when the forecast horizon is longer. Special attention should be paid as to
whether such forecasts take into account the different household formation rates across various age and
income groups, as well as any anticipated changes in the demographic composition of the area’s population.

ESTIMATE EXPECTED AVAILABLE SUPPLY OF RESIDENTIAL UNITS  The formulas for carrying out this step
are very clearly presented in Table 6.2. A few comments are warranted, however, with respect to the factors
that enter these formulas. First, in calculating expected new supply two years ahead, the analyst needs an

123
 Analyzing residential real estate markets

TRADITIONAL
AGGREGATE APPROACH:
ESTIMATING HOUSING
MARKET GAP
(at end-of-year T+N)

EQUILIBRIUM CONDITION AT TIME T+N: S(T+N)(1- V*) = HH(T+N) 1

STEP 1 STEP 2

ESTIMATE NEW ESTIMATE

DEMAND FROM EXPECTED EXPECTED AVAILABLE
HOUSEHOLD GROWTH SUPPLY at T+N

= =

NEW COMPLETIONS
EXPECTED C(T+1)+...C(T+N)
MARGINAL
CHANGE IN NUMBER PLUS
OF HOUSEHOLDS (HH)
EXCESS VACANT STOCK
HH(T+N)-HH(T) V(T)S(T)-V*S(T)

MINUS

ADDITIONAL
STRUCTURAL VACANT
STOCK REQUIRED AT T+N
V*[S(T+N)-S(T)]

MINUS

EXPECTED STOCK
DEPRECIATION
dS(T)2

STEP 3
ESTIMATE
HOUSING GAP

NEW DEMAND MINUS AVAILABLE

FROM GROWTH SUPPLY

Notes: 1 V* denotes the market's structural or equilibrium vacancy rate

2
d is defined as the expected depreciation rate between years T and T+N

Figure 6.2  Aggregate approach to estimating the housing gap

124
 Macro analysis: accounting techniques

Table 6.1  Numerical example of the aggregate accounting approach

The case

A group of homebuilders contemplate developing housing in Metropolitan Area A by year-​end 2002.They thus decide to
engage in a superficial analysis of residential development prospects involving the estimation of the housing “Gap” as below:

Data requirements
• Population estimates for year-​end 2000 and forecasts for year-​end 2002 from an econometric forecasting firm.
• Average household size estimates from the same source.
• Year-​end 2000 housing stock, vacant units, and the area’s structural vacancy rate from a residential market
research group.
• Expected new supply of residential units, estimated from information on starts and permits.

Population, 2000; POP(00) 8,676,000

Population, 2002; POP(02) 8,936,280

Household size, 2000; HS(00) 2.69

Household size, 2002; HS(02) 2.62

Housing stock, 2000; S(00) 3,362,279

Vacant units, 2000; V(00)S(00) 137,000

Expected completions, 2001; C(01) 50,000

Expected completions, 2002; C(02) 70,000

Structural vacancy rate, V* 4.00%

Expected 2000–​2002 depreciation rate, d 1.00%

By definition at market equilibrium:

S(02) = Total housing units demanded in 2002/​(1 –​Structural vacancy rate)
= Total number of households in 2002/​(1 –​Structural vacancy rate)

Note: All numbers represent year-​end figures.

estimate of expected residential completions in each of the two years of the forecast (see discussion of
new construction in the previous chapter). Expected completions one to two years ahead can usually be
calculated with reasonable accuracy based on a survey of residential projects that have broken ground and/​
or permit information. An estimate of completions for more than two years ahead, however, can be more
accurately estimated through econometric forecasting, which is discussed in the next section.3
If there is a reliable survey of the number of existing vacant housing units (owner-​occupied units
available for sale, or vacant rental units) then the analyst can use that number as the estimate of the
market’s vacant stock. Alternatively, the vacant housing stock can be calculated using information on the
market’s vacancy rate and total housing inventory. Estimates of the market’s vacancy rate can be obtained

125
 Analyzing residential real estate markets

Table 6.2  Estimating the gap for the base-​case forecast

Step 1
Estimate new demand from household growth, 2000–​20021

Expected
Estimated growth in
number of demand
End-​of-​year Population Average household size households 2000–​2002
2000 8,676,000 2.69 3,225,279
2002 8,936,280 2.62 3,410,794 185,515

Step 2
Estimate available supply

Expected new supply, 2000–​2002 (excludes project under consideration) 120,000

= C(2001)+C(2002) : Sum of expected completions 120,000

Plus: Excess vacant stock, 2000 2,509

= V(00)*S(00) : Vacant stock, 2000 137,000
– V*S(2000) : Structural vacant stock, 2000 134,491

Minus: Additional structural vacant stock required in 2002 7,625

= V*S(02) : Structural vacant stock, 2002 142,116
–​ V*S(00) : Structural vacant stock, 2000 134,491

Minus: Expected depreciation, 2000–​2002 33,623

= d*S(00) : Depreciating stock, 2000–​2002 2
33,623

Equals: Available supply 81,261

Step 3
Estimate the housing gap

New demand from household growth 185,515

Minus: Available supply 81,261
Equals: Gap 104,254
1
All figures in bold are given.
2
Notice that d has been defined as the biannual depreciation rate. A more exact calculation would involve using an
annual depreciation rate, which would then be applied separately to the 2000 and 2001 stock.

126
 Macro analysis: accounting techniques

Table 6.3  Selective results of sensitivity analysis

Gap

•  Base case 100,084

•  What if population grows at 2% instead of 3% during 2000–​2002? 66,970
•  What if household size decreases at a slower rate down to 2.655? 55,121
•  What if new completions are 10% higher? 88,564

from brokers, housing research companies, or through surveys. The Census Bureau provides quarterly
vacancy rate estimates for the nation’s largest metropolitan areas, but not for counties and municipalities.
Information on a market’s total housing stock, however, is more difficult to obtain because simply it is
much more difficult and expensive to survey every housing unit in a large urban area.
Typically, the easiest way to estimate a market’s housing inventory is to use the housing-​unit count
provided by the most recent census as the starting level and then add permits up to the year of analysis
using the stock-flow identity. Note that the census provides separate inventory counts for structures with
different numbers of units. In order to use the stock-​flow identity, the analyst will need to make some
assumptions regarding the depreciation rate and the conversion factor to be applied to the stock and permit
figures of the previous period, respectively.
The concept behind the calculation of the excess vacant stock is that not all of the vacant housing
units are truly available for renting because the market always requires some structural vacancies in order
to allow for what is considered as “normal” search activity on the part of both tenants and landlords.
The concept of the structural (often referred to as “natural”) vacancy rate is parallel to the concept of
the structural or frictional unemployment rate and was first introduced in the analysis of rental housing
markets by Rosen and Smith (1983). Hagen and Hansen (2010) report empirical evidence that supports
the existence of a structural vacancy rate in the apartment market and its variability through time
and across geographic subareas. However, they did not find any significant variations across different
apartment types. In practice, the long-​term historical average vacancy can be used as an estimate of a
market’s structural vacancy rate.4
Estimates of the structural vacancy rate produced by simply averaging historical vacancy data should be
more accurate if such data cover a full cycle of the market at least. If similar data on apartment rents and/​
or house prices are available, however, a market’s structural vacancy rate can be estimated econometrically
from the rent or price adjustment equation. For a more detailed discussion of this issue, and how one can
estimate the structural vacancy rate econometrically, see the section on econometric forecasting models
and note 85 in Chapter 13. It should be noted that the structural vacancy rate varies across property types.
Looking at the vacancy ranges over the last 20 years for different property types one can see, for example,
that apartment vacancy rates are persistently lower than office vacancy rates.
Finally, in calculating an area’s available housing supply the depreciated stock needs to be taken into
account. The term “depreciated stock” refers to the housing units that are taken out of the active housing
market because of either physical deterioration, or conversion to other uses (economic obsolescence).
Carn et al. (1988) suggest that an estimate of the number of housing units expected to be taken out of the
active market can be generated on the basis of permit data for “intentional demolitions” and conversions,
and estimates of inventory losses due to “acts of nature” from actuarial tables. A less involved, and maybe
less accurate, technique for forecasting housing depreciation is to assume an annual depreciation rate. For
example, the U.S. Bureau of Economic Analysis, assuming a useful life of 80 years and a geometric decay
procedure, has estimated that the housing stock should depreciate at a fixed annual rate of 1.14% (U.S.
Department of Housing and Urban Development, 2000).

127
 Analyzing residential real estate markets

ESTIMATE THE HOUSING GAP AND PERFORM SENSITIVITY ANALYSIS  The housing “gap” estimate is in
essence an estimate of excess demand or excess supply depending on the sign of the derived number.
A  positive number indicates that the number of households looking for an apartment in 2002 will be
greater than the number of units available and, therefore, it would signify excess demand. A  negative
number would indicate that the available housing units to satisfy additional demand from new households
entering the market will exceed household growth and, therefore, it would signify excess supply.
Table 6.3 shows that sensitivity analysis is of vital importance because it can help the analyst understand
how sensitive the estimate is to the different assumptions entering the calculations. For example, if popula-
tion were to grow at a 100-​basis-​point slower rate, the housing gap would be about 30% less. Furthermore,
if average household size in 2002 were 2.655 instead of 2.62, the housing gap would be reduced almost by
half. On the contrary, if completions were 10% higher than expected, the housing gap would be reduced
by about the same percentage. This sensitivity analysis shows that the housing gap estimate is especially
sensitive to the assumptions regarding future household size and population growth and suggests that the
market analyst needs to pay special attention as to the soundness of such assumptions.

CRITIQUE  The aggregate accounting approach for estimating an area’s housing gap has several shortcomings.
First of all, it fails to account how changes in income and prices affect net household formation and, therefore,
growth in aggregate housing demand. In addition, the forecasts of housing market fundamentals produced
by this technique ignore the effect of prices/​rents on residential construction and the feedback effects
that housing demand, supply, and vacancy changes have on house prices and rents. As such, applications
of aggregate accounting techniques in estimating housing market prospects are more prone to error than
econometric approaches that consider explicitly the dynamic interactions among the key housing market
variables. Finally, due to the aggregate nature of the estimates produced by this technique, it can contribute
very little to the identification of proper target groups and niches in housing product types.

Accounting techniques: the disaggregate approach

Most market studies are done with a specific residential product type in mind. Analysis of the behavior of
the aggregate housing market can provide valuable insights regarding broader market trends that are likely
to reverberate to its different segments along the quality and size continuum, but it cannot provide conclu-
sive indications regarding the viability of a specific residential product. For this reason, a more disaggregate
approach that takes into account the characteristics and price of the specific residential development, as
well as consumer characteristics and preferences, is needed in assessing housing demand.

BASIC METHODOLOGICAL STEPS IN ESTIMATING HOUSING DEMAND  Applications of disaggregate

accounting approaches in housing-​gap analysis place emphasis on how to estimate effective market demand
across income and age groups for a specific housing type, taking into account income, house prices, and
affordability, as well as tenure and housing type preferences. Accounting for these factors is very important,
since, as it has been emphasized, they strongly influence the type of the specific housing unit a household
will eventually choose. Within this context, in estimating an area’s effective demand for a specific housing
type using the disaggregate approach, the following three broader methodological steps (or nine more
specific steps, as outlined in Figure 6.3) need to be carried out:

1 . Estimate household growth by age and income group (steps 1–​5 in Figure 6.3).
2. Estimate how many households would qualify for a specific housing unit of a given price taking into
account their incomes and affordability (steps 6–​7 in Figure 6.3).
3. Estimate how many of the qualifying households would prefer a given tenure mode and housing type
using survey-​based preference ratios (steps 8–​9 in Figure 6.3).

128
 Macro analysis: accounting techniques

Step 1 Obtain forecasts

of population by age

Step 2
Estimate population
growth across age groups

Step 3
Using headship rates and popu-
lation growth across age groups,
estimate household growth

Step 4
Obtain a matrix
distributing households
across income and age groups

Step 5
Using this matrix,
and household growth
by age group, estimate
household growth
by age and income

Step 6
Given a development
scenario, estimate
minimum annual
qualifying incomes

Step 7
Given the income levels
of target groups,
Estimate number of
households by age
and income qualifying
for given development

Step 8
Using survey-derived
preferences,
estimate effective
demand for
given tenure mode

Step 9
Using housing type
preference ratios,
estimate effective
demand for given
product

Figure 6.3  A disaggregate approach for estimating effective housing demand: the case of owner-​occupied housing

129
 Analyzing residential real estate markets

The more specific steps for carrying out the disaggregate approach are described in Figure 6.3. Below
we list these steps and provide the relevant mathematical formulas, where Ai denotes age group i, n the total
number of age groups, Ij income group j, k the total number of income groups, t the current year, and t+m
the planned completion year of the project.

1. Obtain forecasts of population by age for the year of the anticipated completion of the project
(A1,t+m, A2,t+m,.....An,t+m) and holding period (if it applies) from a competent forecasting firm.
2. Estimate population growth for each age group as Ai,t+m –​  Ai,t.
3. Estimate household growth in each age group using age-​specific headship rates, HAi, and the estimated
population growth from the previous step: (Ai,t+m –​ Ai,t) HAi.
4. Obtain a matrix with percentage allocation of households by age of household head and income
(RA1,I1,RA2,I1, ...,RAn,I1,RA1,I2,...,RAn,Ik). Such a matrix can be obtained from commercial data vendors
and typically represents projections based on information on an area’s age and income distribution as
provided by the latest decennial census. For a detailed discussion of how to project an area’s income
distribution from the latest census up to a current year and into the future, see Carn et al. (1988).
5. Using this matrix and household growth by age group, estimate household growth by age and income
as (Ai,t+m –​ Ai,t) HAi RAi,Ij.
6. Given a development scenario and expected unit sales prices, estimate the minimum required annual
income in order to qualify for a mortgage loan for purchasing a house at those price levels. To apply
this step, information on mortgage loan underwriting standards needs to be collected and used to cal-
culate the following items:

(a)  Loan amount =  Unit price × Loan-​to-​value ratio

(b)  Mortgage payment =  Loan amount × Mortgage constant
(c) Taxes =  Property tax rate × Unit price
(d) Insurance =  Homeowner insurance rate × Unit price
(e) PITI =  (b) + (c) + (d)
(f)  Qualifying gross income =  (PITI)/​Mortgage underwriting standard

It should be noted that for a mortgage providing for monthly payments, the term mortgage constant refers
to the total annual payment to amortize one dollar over a given loan term (Brueggeman and Fisher,
2018). PITI stands for principal, interest, taxes, and insurance. Finally, the mortgage-​underwriting
standard refers to the maximum acceptable ratio of PITI over household income required by lenders
when providing home loans.

7. Estimate the number of qualifying households, HHQ, within the target age groups based on the dis-
tribution of these age groups across income categories. In particular, add up the households in the
target age groups that fall in income categories that are equal to or greater than the estimated threshold.
Since income categories are defined as ranges, in most cases there will be an income group for which
the qualifying income will fall within this range instead of being clearly out of it. In such a case, the
number of qualifying households can be calculated by multiplying the number of households in that
income group with a factor calculated as below.

( Upper Income Range − Qualifying Income Level )

( Upper Income Range – Lower Income Range )

8. Using survey-​based preferences, estimate effective demand for homeownership,

HHH, as HHQ × Homeownership preference ratio.

130
 Macro analysis: accounting techniques

9. Using housing-​type preference ratios, estimate effective demand for the housing type under consider-
ation as: HHH × Housing type preference ratio.

Tables 6.4–​6.6 illustrate the application of each of the nine steps in a numerical example. Note that in this
example the preliminary development scenario calls for the development of patio homes to be priced at
$185,000. The target groups are move-​up households in the 35–​54 age category and the anticipated time of
completion is 2001. Thus, the numerical example focuses on the estimation of effective housing demand
in 2001.
Notice that in order to calculate the number of qualifying households in the selected target group
(13,077) one needs to look at Step 5 in Table 6.5 and estimate the number of households in the target age
groups that have income equal to or greater than the minimum qualifying income of $64,196.

WHO QUALIFIES?  Given that the estimated minimum qualifying income is $64,196, qualifying households
include:

1. A fraction of households in the $50,000 to $74,999 category. Through interpolation this fraction can
be calculated as:

( 74, 999 − 64,196)

= 0.4321
( 74, 499 − 50, 000 )

Therefore, the number of qualifying households in this income group (from both age groups) is:

(0.4321)*(4,863 + 7,847) = 5,492

2. All households within the two age groups with income above $75,000, that is:

2,445 + 5, 140 = 7,585

The sum of the qualifying households from these two income groups gives the total number of households
within the two target age groups—​13,077—​that are expected in 2001 to be looking for and be able to
afford to buy a patio home priced at $185,000.
It is obvious from this exercise that the information on the percentage distribution of an area’s
household by age of household head and income is the most critical information in determining the
demand levels for a given housing product using the disaggregate approach. For this reason, the analyst
needs to make every effort to obtain the most reliable estimate of such a distribution for the project’s
market area.

CRITIQUE OF THE DISAGGREGATE APPROACH  In contrast to the aggregate accounting approach, the
disaggregate approach does take into account the effect of prices, incomes, affordability, and preferences on
effective housing demand. As such, it can be used to assess niches in consumer groups and product types.
The major advantage of the disaggregate accounting approach is that it enables the analyst to generate
demand forecasts for the very specific housing product considered, taking into account its idiosyncratic design
features and price level. Furthermore, as household projections by age and income are provided by private
vendors for as small geographical areas as zip codes, this technique allows the analyst to focus on the
project’s primary market area or submarket.Thus, in contrast to the econometric approach, which typically
focuses on the aggregate housing market at the metropolitan area level (because of data constraints), the
disaggregate accounting approach enables market analysts to focus on both the housing type and geographical
area that is more relevant for the very specific project under consideration.

131
 Analyzing residential real estate markets

Table 6.4 The disaggregate approach: a numerical example

Step 1
Obtain forecasts of population by age

End-​of-​year figures
Age group 2000 2001 2002 2003 2004
15–​24 1,421,637 1,450,069 1,413,290 1,377,444 1,342,507
25–​34 1,374,356 1,311,842 1,331,398 1,351,245 1,371,389
35–​44 1,506,009 1,554,800 1,570,688 1,586,738 1,602,952
45–​54 1,137,291 1,205,528 1,271,833 1,341,783 1,415,581
55–​64 836,736 874,390 912,863 953,029 994,962
CAl>65 989,190 1,008,974 1,004,938 1,000,918 996,914
Total 7,380,987 7,520,011 7,459,079 7,398,640 7,338,691

Step 2
Estimate population growth across age groups

Age group 2000–2001 2001–2

​ 002 2002–​2003 2003–​2004 Population Calculated as: Population
Growth 2001–​Population 2000
2000-​2001
15–​24 28,433 (36,779) (35,846) (34,937)
25–​34 (62,513) 19,556 19,847 20,143
35–​44 48,791 15,888 16,050 16,214 48,791 = 1,554,800–​1,506,009
45–​54 68,237 66,304 69,951 73,798 68,237 = 1,205,528–​1,137,291
55–​64 37,653 38,473 40,166 41,933
>65 19,784 (4,036) (4,020) (4,004)
Total 139,024 (60,933) (60,439) (59,949)

Step 3
Estimate household growth across age groups

Age group Headship 2001 household Calculation of 2001

rate1 growth household growth
15–​24 5.00% 1,422
25–​34 49.05% (30,663)
35–​44 56.95% 27,787 = 48,791*0.5695
45–​54 57.50% 39,237 = 68,237*0.5750
55–​64 58.50% 22,027
>65 64.60% 12,780
Total 72,589

All figures are strictly hypothetical.

1

132
 Macro analysis: accounting techniques

Table 6.5  Numerical example continued

Step 4
Obtain a distribution of new households by income level1

Age <5,000 5,000 10,000 15,000 25,000 35,000 50,000 >75,000

groups – 9,999 –​14,999 –​24,999 –​34,999 –​49,999 –​74,999
15–​24 13.8% 16.6% 15.8% 25.6% 15.8% 8.1% 3.6% 0.7%
25–​34 5.7% 7.8% 9.6% 22.2% 20.3% 20.4% 10.9% 3.1%
35–​44 4.3% 6.1% 6.6% 15.7% 17.5% 23.5% 17.5% 8.8%
45–​54 4.9% 5.6% 6.6% 14.0% 15.3% 20.5% 20.0% 13.1%
55–​64 7.5% 9.9% 9.5% 18.8% 15.2% 17.0% 13.7% 8.4%
>65 9.9% 24.5% 17.8% 22.1% 11.5% 7.9% 4.1% 2.2%

Step 5
Estimate growth of new households (2000-​2001) by income level

Age 5,000 10,000 15,000 25,000 35,000 50,000 >75,000

groups –​9,999 –​14,999 –​24,999 –​34,999 –​49,999 –​74,999
15–​24 196 236 225 364 225 115 51 10
25–​34 (1,748) (2,392) (2,944) (6,807) (6,225) (6,255) (3,342) (951)
35–​44 1,1952 1,695 1,834 4,363 4,863 6,530 4,863 2,445
45–​54 1,9233 2,197 2,590 5,493 6,003 8,043 7,847 5,140
55–​64 1,652 2,181 2,093 4,141 3,348 3,745 3,018 1,850
>65 1,265 3,131 2,275 2,824 1,470 1,010 524 281
Total 4,483 7,048 6,072 10,378 9,684 13,188 12,961 8,776
1
All figures are strictly hypothetical.
2
This number was calculated as 1,195 = 0.043*27,787.
3
This number was calculated as 1,923 = 0.049*39,237.

The major shortcoming of the disaggregate accounting approach is that it does not allow the quan-
tification of dynamic relationships in residential markets, such as the impact of changes in demand and
supply on prices/​rents and feedback effects of the latter on the former. As such, its analytical power is very
limited when it comes to producing demand, supply and, most importantly, rent/​price forecasts that are
consistent with the conventional theory of supply-​demand dynamics. Furthermore, household projections
by age and income should be viewed with caution, especially when the geographical area of reference is
small, for two reasons. First, household formation is affected by house prices and rents, a factor that may
not be considered by private vendors producing household projections by age and income. Second, the
smaller the geographical area of reference, the more inaccurate the forecast of household growth may be, as
it becomes increasingly difficult to accurately forecast the micro-​location distribution of new households
by age group within an urban area.
Nevertheless, most of the time, the disaggregate accounting approach is the only technique available to
the analyst for assessing demand for a specific housing type and design. If econometric forecasts are avail-
able for the broader market (and the same tenure type) then the analyst can combine the two approaches
to generate a better demand forecast for a specific housing product that takes into account the price/​
rent dynamics of the broader market. This can be done by applying the econometrically derived price/​

133
 Analyzing residential real estate markets

Table 6.6  Numerical example continued

Steps 6, 7, 8, 9
Estimate minimum qualifying incomes and the number of qualifying households, number of
households preferring homeownership and given housing type

Affordability parameters
House price, 2001 $185,000
Annual interest rate, 2001 8.6%
Term of loan 30
Loan-​to-​value ratio 0.9
Property tax rate 2.5%
Homeowner insurance 0.5%
Mortgage underwriting standard 33.0%
Step 6
Calculating minimum qualifying incomes

Annual mortgage constant 0.094

Loan amount $166,500
Mortgage payments $15,635
PITI: $21,185
Qualifying gross annual income $64,196 = $21,185/​0.33

Step 7
Estimating the number of qualifying HHs in target age groups1

% Qualifying HHs in $50,000–​$75,000 income group 0.432

Qualifying HHs in $50,000–​$75,000 income group 5,492
Qualifying HHs in >$75,000 income group 7,585
Qualifying HHs in target age groups 13,077

Step 8
Estimating effective demand for homeownership

% Ownership preference in target groups 85%

Effective demand for ownership 11,115

Step 9
Estimating effective demand for given housing type

% Preference for patio homes: 60%

Effective market demand for patio homes 6,669

1
See text for a more detailed discussion of the calculations involved in this step.
Note: Numbers in bold are given. This index can be applied to year-​end 2020 prices to forecast prices at the
anticipated time of sale, that is, year-end 2023.

134
 Macro analysis: accounting techniques

rent change forecast to the price/​rent level reflecting the housing type and design under consideration.
Furthermore, absorption and new construction estimates at the urban area level, derived through econo-
metric techniques, can provide a framework of reference for demand and supply estimates for the specific
housing type and the primary market area under consideration.

Chapter summary
In this chapter, we have reviewed briefly the basic steps of macroeconomic analysis of residential markets
and elaborated on applications of accounting approaches. The three basic steps of residential macroeco-
nomic analysis include:

• Definition of the project’s primary market area, taking into account worker commuting patterns, major
employment centers at reasonable distances from the site, and the residential areas serving those centers.
• Determination of the time frame of analysis, taking into account the size and type of the project, as well
as the underlying investment strategy.
• Application of accounting or econometric techniques in order to estimate the residential demand-​
supply gap and evaluate housing market strength at the time of the completion of the project.

Accounting techniques can be applied in an aggregate or a disaggregate fashion. Aggregate techniques

can only provide rough estimates of total market demand and, typically, they ignore the income and
age composition of the area’s population, as well as the effect of prices on housing demand. On the
contrary, disaggregate analysis of housing demand by income and age group does take into account all
these factors and, in theory, it should provide more accurate estimates of demand for a given residential
product. The shortcoming of both the aggregate and disaggregate accounting techniques is that they
cannot account for the dynamic interactions between the economy, housing demand, supply, and rents/​
prices, that shape movements in local housing markets. The econometric approach is the only approach
that is sophisticated enough to account scientifically for these dynamics. Given the sophistication of this
technique, the next chapter discusses its basics, while Chapter 8 discusses how it applies specifically to
residential markets.

Questions
1 . Discuss the steps involved in defining a residential development’s primary market area.
2. Explain why analysis of the metropolitan housing market—​within which the planned residential pro-
ject is located—​is relevant even if the project’s primary market area is smaller.
3. Discuss the steps involved in applying the aggregate accounting approach for the estimation of an
area’s housing gap. Please explain the variables involved in each step.
4. Apply the aggregate accounting technique to assess the state of the housing market (excess demand,
equilibrium or excess supply) of the urban area within which your home is located for the next
two years.
5. Discuss the steps involved in applying the disaggregate accounting approach for the estimation of the
housing demand for a specific housing product.
6. Explain the advantages and shortcomings of accounting approaches (aggregate and disaggregate) to
housing market analysis vis-​à-​vis the econometric approach.

135
 Analyzing residential real estate markets

Project #2: Estimating effective residential demand using the accounting approach

The case
The objective of this project is to help students practice with the disaggregate approach for estimating
effective market demand for specific housing types within a competitive market area. The attached hypo-
thetical market data was presumably collected by a market analyst whose assigned task was to assess the
strength of effective residential demand for townhouses and patio homes within a given market.

Development scenarios and price data

The expected/​planned time of entry in the market is year-​end 2022 (or beginning of year 2023). This will
enable the project to capture part of the new demand during year 2023 (from year-​end 2022 to year-​end
2023). It is assumed that all units sold during year 2023 will be sold at year-​end 2023 prices (a conventional
assumption).
Two alternative development scenarios need to be analyzed. The first calls for the development of
townhouses while the second calls for the development of patio homes. Given current market conditions,
the location of the project, and the quality of construction, the planned townhouses could now (year-​end
2020) be sold for $200,000 each, while the patio homes could now be sold for $250,000 each. A leading
forecasting agency suggests that the prices of townhouses and patio homes are expected to follow the nom-
inal price indices below:

Nominal price index

Year-​end Townhomes Patios
2020 100.00 100.00
2021 103.50 101.00
2022 106.00 102.00
2023 107.00 105.00

Mortgage requirements and affordability measures in 2023

Expected annual interest rate 4.25%

Term of loan (years) 30
Loan-​to-​value ratio 0.95
Property tax rate 2.50%
Homeowner insurance 0.50%
Mortgage underwriting standard 35%

Target demographic groups and preferences

Target groups are households in the 35–​39 age category, referred to as Group 1, and households in the
40–​44 age category, referred to as Group 2.

136
 Macro analysis: accounting techniques

Preferences Age Group 1 Age Group 2

For homeownership: 75% of those qualifying 80% of those qualifying

For townhouses: 30% of those qualifying and 25% of those qualifying and
preferring to own preferring to own

For patio homes: 20% of those qualifying and 30% of those qualifying and
preferring to own preferring to own

Market area data
1.  FORECASTS FOR TARGET DEMOGRAPHIC GROUPS

Age groups Year-​end 2020 Year-​end 2023

35–​39 502,003 518,300
40–​44 379,097 402,800

2.  HEADSHIP RATES, 2023

Age groups Headship rates

35–​39 55.0%
40–​44 57.5%

3.  DISTRIBUTION OF HOUSEHOLDS BY INCOME, 2023

Income groups (USD)

Age 5,000–​ 10,000–​ 15,000–​ 25,000–​ 35,000–​ 50,000–​
groups <5,000 9,999 14,999 24,999 34,999 49,999 75,000 >75,000
35–​39 4.30% 6.10% 6.60% 15.70% 17.50% 23.50% 17.5% 8.8%
40–​44 4.90% 5.60% 6.60% 14.00% 15.30% 20.50% 20.0% 13.1%

4.  EXPECTED AVAILABLE SUPPLY WITHIN THE COMPETITIVE MARKET AREA, 2023

Townhouses 1,800 units
Patio homes 750 units

Questions
1 . What is the expected 2023 effective market demand for patio homes and townhouses?
2. Should the developer go ahead with patio homes or townhouses? Explain (think carefully before you
answer this question).

137
 Analyzing residential real estate markets

What to turn in
A couple of paragraphs of written text explaining the answers to the questions above and a hard copy of
the spreadsheet with the calculations.

Notes
1 This paragraph was added by the editor.
2 Carn et al. (1988) point out that household-​size estimates in areas with high percentage of population in group
living arrangements may be substantially inaccurate if such population is not excluded from the calculation.
3 Carn et al. (1988) suggest an alternative way for predicting future completions by interviewing local builders and
determining their minimum production levels (in terms of housing units) and inventory holdings.
4 Carn et al. (1988) argue that the “low-​point trend” may be a more accurate estimate than the historical average as
the latter may overstate the structural vacancy rate.

References and additional readings

Brueggeman, W. B. and J. Fisher. 2018. Real Estate Finance and Investments 16th Edition. New  York:  McGraw-​Hill
Education.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Empirical Residential Market Analysis. Chap. 8 in Real Estate
Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W. Wheaton. 1996. The Operation of Property Markets. Chap. 2 in Urban Economics and Real Estate
Markets. Englewood Cliffs, NJ: Prentice Hall.
Fanning, S., T. Grissom, and T. Pearson. 1994. The Market Analysis Process for an Existing Apartment Complex.
Chap. 14 in Market Analysis for Valuation Appraisals. Chicago, IL: Appraisal Institute.
Gabriel, S. A. and Nothaft, F. E. 1988. Rental Housing Markets and the Natural Vacancy Rate. AREUEA Journal, 16:4,
419–​429.
Goetzmann, W. and M. Spiegel. 1997. A Spatial Model of Housing Returns and Neighborhood Substitutability. Journal
of Real Estate Finance and Economics, 14:1/​2,  11–​31.
Hagen, D. and J. Hansen. 2010. Rental Housing and the Natural Vacancy Rate. Journal of Real Estate Research, 32:4,
413–​433.
Jones, C., M. Coombes, N. Dunse, D. Watkins, and C. Wymer. 2012. Tiered Housing Markets and their Relationship to
Labour Market Areas. Urban Studies, 49:12, 2633–​2650.
Rabianski, J. 2006. Apartment Market Area Delineation. Appraisal Journal, 74:1, 33–​42.
Rosen K. and L. Smith. 1983. The Price-​Adjustment Process for Rental Housing and the Natural Vacancy Rate.
American Economic Review, 73:4, 779–​786.
U.S. Department of Housing and Urban Development. 2000. Summary. U.S. Housing Market Conditions, Summer.

138
newgenprepdf

7
MACROECONOMIC ANALYSIS
OF RESIDENTIAL REAL
ESTATE MARKETS
The basics of the econometric approach

Introduction 139
The basics of regression analysis 140
Evaluating regression models 145
Prediction 145
Applications of regression models in market analysis 146
Macro analysis of real estate markets 146
Micro analysis of real estate projects 146
A numerical example of a simple linear regression 146
Multiple regression models 150
Chapter summary 150
Questions 151
References and additional readings 151
Appendix 7A: Further information on linear regression models 152

Introduction
So far, we have discussed simple accounting methodologies for assessing demand-​supply gaps in residential
real estate markets. Such methodologies can help assess niches in target groups and product types in the
shorter run, but are not sufficient for fully evaluating residential development potential because they ignore
the dynamics of the urban residential markets. In particular, they do not reflect the important behavioral
relationships between prices and the number of demanded units, between prices and the number of units
supplied, and the feedback effect of changes in demand and supply on prices and rents. The accounting
techniques are particularly shortsighted in their treatment of supply in that they usually examine short-​run
available supply without regard to future expected supply.
A more advanced approach, which can capture and simulate the dynamic relationships embedded in
residential market behavior, is the econometric approach. In particular, econometric techniques quantify
scientifically the effect that changes in the economy, population, and prices/​rents may have on residen-
tial demand and supply, as well as the effect that changes in the latter may have on residential prices and
rents. As such, they provide a solid basis for predicting residential demand, supply, and prices/​rents at the
anticipated time of project entry and during the lease-​up or holding period.

139
 Analyzing residential real estate markets

The forecasts produced by econometric models can help residential development planning in a number
of ways:

• Predictions of the time path of critical housing market indicators, such as rents/​prices and absorption,
can provide valuable guidance with respect to time-​of-​entry decisions and decisions regarding develop-
ment phasing.
• Forecasts of market price/​rent indices can be used to generate project-​specific price/​rent schedules,
which are necessary inputs for the evaluation of the feasibility and viability of the development.
• Forecasts of competing housing supply can be used, together with forecasts of effective housing demand and
other information, to determine a project’s absorption schedule, taking into account its competitive pos-
ition.This absorption schedule is also an important input in assessing the financial feasibility of the project.

The econometric approach to assessing housing market strength incorporates the simple stock-​flow model
discussed in an earlier chapter of this textbook. As such, it fully accounts for the dynamics of urban residen-
tial markets through time. The development of simple econometric models, however, requires the know-
ledge of basic regression techniques.Thus, this chapter focuses on the basics of regression analysis, the most
commonly used estimation technique in econometric analysis. It should be emphasized that the material
presented here is by no means sufficient for fully comprehending and applying regression analysis. Students
interested in specializing in the area of real estate market analysis are strongly advised to attend at least an
introductory course in econometrics.

The basics of regression analysis

Regression analysis studies the dependence of one variable, Y, referred to as the dependent variable, on other
variables, Xj(j=1,2,..,n), referred to as the independent or explanatory variables. For example, if Y represents
apartment rents and X the market vacancy rate, we can use regression analysis to examine whether
movements in the former depend on movements in the latter. Regression analysis techniques can in gen-
eral be used to:

• Identify the nature (positive or negative) of hypothesized relationships between the dependent variable
(i.e., Y) and the explanatory variables (i.e., the Xs).
• Evaluate the strength of the relationship between each of the Xs and Y.
• Predict Y, given expected values of the Xs.

In order to apply regression analysis techniques, we need data on the variable that needs to be predicted
and the variables that in theory should drive its movements. In general, two broader categories of data may
be available to the analyst:

• Cross-​section  data
• Time-​series  data.

Cross-​section data provide information for several units of observations (e.g., markets, properties) at the
same point in time. As such, they describe variations in one or more variables across different analysis units.
Consider, for example, that you have information about 290 condominium units that were sold in the
Boston Back Bay area in 2019 (see Tables 7.1 and 7.2). Available information includes the sales price, living
area, number of bedrooms, number of bathrooms, distance from river, and other characteristics for each
unit sold. With this data at hand, we can use regression analysis to examine whether and to what extent
location and other unit characteristics influence condominium sales prices. Thus, in this example, the
dependent variable (Y) is the sales price and the independent variables (Xi) include the unit characteristics.

140
 The basics of the econometric approach

Table 7.1  Examples of cross-​section and time-​series datasets

Cross-​section data Time-​series data

Notation Variable Notation Variable
Y Sales price Y Apartment rent index
X1 Living area X1 Number of households
X2 Number of bedrooms X2 Household size
X3 Number of bathrooms X3 User cost of capital
. . . .
. . . .
Xn Distance from river Xn Apartment stock
Unit of analysis: Sold condos Unit of analysis: Market
Number of sales in sample: 290 Number of years in sample: 20

Table 7.2  Sample cross-​section data: Condominium sales data

Condominimum Sales price, Floor Living area, Number of Number of Distance Distance
ID number $/​unit sq.ft./​room bathrooms bedrooms from river, ft. from park, ft.
1 300,289 3 325.0 2 2 150 16,650
2 109,721 2 222.0 1 1 1,200 9,400
3 194,610 1 297.5 2 2 270 19,100
4 219,442 4 299.3 2 2 320 20,000
5 155,919 6 255.7 2 1 450 20,000
6 291,627 1 360.5 3 2 450 1,400
7 195,765 3 282.8 2 2 450 1,400
8 230,991 4 289.0 2 2 450 1,400
9 173,244 3 334.0 1 1 520 1,500
10 236,766 1 289.0 2 3 640 2,000
11 289,894 4 312.6 3 3 640 2,000
12 242,541 5 273.3 2 2 640 2,000
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
285 269,105 6 283.0 2 2 150 24,500
286 164,04 5 224.0 2 2 150 26,000
287 109,721 3 234.3 1 1 720 200
288 196,343 3 322.7 1 1 720 200
289 240,474 2 335.0 2 2 720 200
290 161,694 2 274.3 1 1 720 200

Time-​series data provide information for the same unit over several time periods. Time-​series data,
therefore, describe movements in the characteristics of an analysis unit through time. Consider for example
that the analyst has information on rents and several demand and supply variables, such as the number of
households, average household size, user cost of capital, household income, average age, and apartment
stock, for the Los Angeles apartment market, over a 24-​year period (see Figure  7.1 and Table  7.3). In
such a case, regression analysis can be used to examine whether there is a systematic relationship between
movements in apartment rents and movements in these demand and supply variables.Thus in this example,
the dependent variable (Y) is the apartment rent and the independent variables (Xi) are the aforementioned
demand and supply variables.

141
 Analyzing residential real estate markets

Los Angeles
apartment rent index
130

120

110

100
Rent index

90

80

70

60

50
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Years

Figure 7.1  Sample time series data

Regression analysis techniques provide valuable insights into the relationship between the dependent
and the explanatory variables by quantifying the effect of the latter on the former. For example, using
the condo sales dataset we can use regression techniques to quantify the effect of each of the unit
characteristics on the sales price.This quantification is dictated by some underlying assumptions regarding
the mathematical relationships between the dependent variable and the explanatory variables. From a
mathematical point of view, these relationships may be linear or non-​linear. The simplest relationship is
the linear one.
In a linear regression model, the dependent variable, Y, is considered to be a linear function of one or
more independent variables, X, as described in the statistical equation (7.1):

Y = a + b1X1 + b2X2 + b3X3 + . . . + bnXn (7.1)

where:
Y : the dependent variable
a : the regression constant or intercept
X1, X2, . . ., Xn : a set of independent or explanatory variables
b1, b2, b3, . . . bn : regression coefficients, indicating the change in Y in response to a unit change in the
respective X, assuming all other Xs remain constant

The constant, a, and the coefficients, b1....bn, are referred to in statistical terminology as the parameters of
the model. It should be emphasized at this point that the regression coefficients do not say anything about
the relative importance of the independent variables. For example, if an independent variable has a larger
b than another independent variable it does not necessarily mean that it has also a greater relative effect.
To evaluate the relative importance of each of the independent variables, we need to calculate their beta
coefficients (the formula for calculating beta coefficients is provided in Appendix 7A).

142
 The basics of the econometric approach

Table 7.3  Sample time series data: The Los Angeles apartment market,Year 1–​Year 24

Year Real rent Households × Mean Household Household User cost Apartment Prime Inflation
index 1,000 population size income (in of housing stock/​ mortgage rate, %
age 87 $) capital, % households rate, %
[1]‌ [2]‌ [4]‌ [5]‌ [6]‌ [7]‌ [9]‌ [10] [11]
1 100.00 2442 28.98 2.89 39,716 1.28 0.39 8.80 5.06
2 100.25 2474 28.83 2.87 38,774 2.06 0.40 7.96 3.67
3 98.97 2484 28.85 2.84 40,091 2.27 0.42 7.61 3.21
4 96.37 2501 28.87 2.82 40,669 0.12 0.43 8.00 5.64
5 91.75 2535 28.87 2.80 40,086 –​3.97 0.43 8.79 10.29
6 87.37 2558 28.88 2.79 39,829 –​4.34 0.44 8.69 10.60
7 87.74 2611 29.14 2.76 40,722 –​0.20 0.43 8.89 6.60
8 89.41 2636 29.34 2.76 41,710 –​0.50 0.43 8.90 6.90
9 90.97 2691 29.52 2.73 43,422 −0.52 0.43 9.48 7.35
10 90.10 2713 29.68 2.73 44,695 –​3.71 0.43 9.90 10.84
11 87.28 2742 29.77 2.74 44,854 –​7.62 0.44 11.26 15.72
12 88.52 2774 29.78 2.75 45,681 –​0.16 0.44 13.31 9.75
13 91.60 2793 29.81 2.78 45,439 4.51 0.44 14.55 5.97
14 95.90 2800 29.85 2.83 46,016 7.00 0.43 12.19 1.77
15 98.41 2827 29.91 2.84 48,451 4.00 0.43 11.92 4.58
16 101.73 2855 29.98 2.87 49,786 3.36 0.43 11.07 4.61
17 105.51 2902 30.07 2.89 50,629 3.80 0.43 9.83 3.28
18 109.21 2926 30.18 2.92 51,529 4.72 0.44 9.07 1.81
19 109.61 2958 30.28 2.93 51,977 2.34 0.44 9.39 4.43
20 109.61 2995 30.37 2.94 51,691 2.35 0.44 9.80 4.71
21 108.50 2994 30.49 2.96 52,712 1.43 0.44 9.68 5.54
22 106.90 3010 30.79 2.97 51,090 2.08 0.45 9.02 4.41
23 104.64 3029 31.09 2.99 50,995 2.32 0.44 7.98 3.42
24 102.71 3032 31.25 3.01 50,384 2.47 0.44 7.04 2.60

Source: U.S. Department of Commerce; Bureau of Economic Analysis; Federal Reserve Bulletin; Torto Wheaton
Research.

Regression models with one explanatory variable are referred to as simple regression models, while
regression models with multiple explanatory variables are referred to as multiple regression models. Simple
regression models are in most cases inadequate in that rarely does one variable fully explain variations
in the dependent variable. For this reason, multiple regression models are more commonly used for
developing forecasting models. Note that the estimation of simple or multiple regression models involves
not only the estimation of the constant and the regression coefficients, but also several regression evalu-
ation statistics.
To better demonstrate the regression estimation procedure, Figure  7.2 portrays the mechanics of a
simple linear regression model with one explanatory variable. For simplicity of illustration, we assume that
only ten observations (ten years of history of a market or ten condominium sales) are used (in reality more
observations are needed in order to obtain reliable estimates). It is emphasized though that this number of
observations is barely sufficient for getting reliable coefficient estimates even for a simple regression model.
Since the regression model or statistical equation includes only one independent variable, the parameters
that need to be estimated include the constant, a, and the coefficient, b, of the independent variable. The
regression algorithm will determine a and b so that the line defined by these two parameters, as portrayed
in Figure 7.2, is the one that minimizes the sum of the squared distances of each observation from it.These

143
 Analyzing residential real estate markets

e10

Y=a+bX
e8
e6
e9
e7
e4
e5
e1
e3
e2
b

Figure 7.2  Best-​fit regression line minimizes the sum of squared residuals

distances, referred to in statistical terminology as errors or residuals, are noted in the figure as e1, e2, and so
on up to e10. Therefore, the parameters a and b will be estimated so that the expression (7.2) is minimized:

∑e 2
= e12 + e 22 + e 32 + .. + e10
2
(7.2)

This estimation technique is referred to in statistical books as the method of least squares, often
abbreviated to OLS (ordinary least squares). This estimation technique squares the errors before summing
them up in order to penalize unusual observations that are associated with considerably larger errors. As
Pindyck and Rubinfeld (1991) point out, the underlying rationale for this approach is that a prediction
involving two errors that have a magnitude of one unit each is preferable to a prediction involving a single
error that has a magnitude of two units.
Before proceeding, it is important to clarify the concepts of exogenous and endogenous variables,
which are very often mentioned in discussions of econometric and regression analyses. To better under-
stand these concepts consider that, typically, when trying to forecast a real estate market we will need to
estimate more than one regression model or statistical equation. Ideally, if data availability allows it, we will
need to estimate at least three statistical equations, one for rental rates or prices, one for demand (absorp-
tion, in the case of office for example) and one for supply (completions). In other words, we would estimate
three different regression models with rental rates, demand, and supply as the dependent variables.
In most cases, the three aforementioned equations are interrelated since the rental rate may be included
as an explanatory variable in the statistical equations of both demand and supply. Furthermore, demand
and supply shape the vacancy rate, which is usually included in the rental rate equation.Therefore, in order
to be able to forecast rental rates, one needs to estimate all three statistical equations. Whenever forecasting
involves the estimation of more than one interrelated statistical equation or regression model, we refer to
them as a system of equations.
When a system of equations is used, a distinction needs to be made between the endogenous and the
exogenous variables of the system. Endogenous variables are those that are determined within the system of

144
 The basics of the econometric approach

the statistical equations, or alternatively all variables that are used as dependent variables in any of the stat-
istical equations in the system. All other variables that appear as explanatory variables, but are not used as
dependent variables in any of the equations in the system, are referred to as exogenous variables. Obviously,
when only one statistical equation is used all explanatory variables are exogenous.

Evaluating regression models
Usually, the estimation of a regression model involves entering the data or importing them (if they are
already available in electronic form) in a spreadsheet or a statistical package. Once the data are entered or
imported, the analyst can use routines built in the software to estimate the desired regression model after
defining the dependent and the independent variables. It should be noted that statistical packages are more
convenient than spreadsheets for running regressions because it is much easier to experiment with different
model specifications and pick the one that works the best.
Whichever software is used, spreadsheet or statistical package, the estimation of a regression model
involves, as indicated earlier, not only the estimation of the constant and the coefficients but also other
statistics that are necessary for evaluating:

• The overall strength of the model or the extent to which movements in the set of independent variables
considered help explain statistically movements in the dependent variable
• The extent to which each independent variable exerts a statistically significant effect on the dependent
variable

The term “statistically significant effect” in plain language means that the data reveal a strong systematic
relationship between the dependent and the independent variable. The most typically used evaluation
measures are reviewed below in the discussion of a numerical example of a simple linear model. The for-
mulas underlying the calculation of these measures are presented in Appendix 7A of this chapter.

Prediction
As indicated earlier, one of the major uses of regression techniques in market analysis for real estate is pre-
diction. For example, a major objective of market studies is to predict movements in market rents over the
project’s planning horizon. If sufficiently long historical data are available, regression estimates can provide
the basis for generating such forecasts. In particular, one or more statistical equations need to be estimated,
which in turn will be used in combination with forecasts of the exogenous variables to generate forecasts
for market rents. If our forecasting model includes only one statistical equation the dependent variable
Y will be the market rent while the Xs may include relevant demand and supply variables for the prop-
erty type considered.1 The analyst can then utilize the estimated parameters of the statistical equation
to calculate the market rent for a given year. For example, a forecast of the market rent for year t+3 can
be produced by plugging in (7.3) the forecast values of the independent variables for that year, X1,t+3,....,
Xn,t+3:

Y = a + b1X 1,t + 3 + b 2X 2,t + 3 +  + b n X n ,t + 3 (7.3)

For a more detailed explanation of the use of regression estimates to develop real estate market forecasts,
please see the application of the econometric forecasting technique to the Dallas apartment market,
presented in the next chapter.

145
 Analyzing residential real estate markets

Applications of regression models in market analysis

Regression techniques can be very helpful in carrying out state-​of-​the art real estate market studies. In
particular, we can distinguish two broader levels of applications:

Macro analysis of real estate markets

At the macro level of analysis, time-​series econometric models, building on the stock-​flow framework, are
often used to:

1. Quantify the relationships between movements in exogenous economic variables and movements in
real estate demand and new construction.
2. Quantify the effect of prices/​rents on demand and supply, as well as the effect of the latter on prices/​
rents.
3. Use these relationships to produce sensible conditional forecasts. These are referred to as conditional
because they depend on the specific forecasts for the exogenous variables.

Micro analysis of real estate projects

At the micro level of analysis, cross-​section models can be used to:

1 . Refine project design by defining optimal amenity packages.

2. Derive estimates of the price or rent a specific project is more likely to command given its characteristics.
3. Determine residual land values and optimal development densities.

A numerical example of a simple linear regression

Table 7.4 presents a numerical example of a simple linear regression model where the dependent variable
(Y) is an apartment rent index (RENT) and the independent variable (X) is the number of households
(HH). The mathematical expression of this regression model is:

Y = a + bX or RENT = a + bHH  (7.4)

As indicated in Table  7.4, this statistical equation is estimated using 24 time-​series observations or
24 years of data. The OLS estimation procedure produces the following estimates for a and b (listed at the
bottom part of the table under the “coefficients” column and corresponding to the rows “intercept” and
“HH,” respectively):

a = 30.21 b = 0.0246 (7.5)

Given these estimates, the statistical equation can be written as:

RENT = 30.21 + 0.0246 HH   (7.6)

In evaluating the above regression results, four broader questions need to be addressed:

• Is the overall specification of the model economically sound?

This question aims at making sure that no obviously important factor is omitted from the set of explana-
tory variables included in the specification. The simple regression model presented in Table  7.4 is a

146
 The basics of the econometric approach

Table 7.4  Example of simple linear regression estimation results

Variables
Dependent:
RENT (Apartment rent index)
Independent:
HH (Number of households)

Typical Microsoft Excel output

Regression statistics
Multiple R 0.61
R square 0.37
Adjusted R square 0.35
Standard error 6.37
Observations 24
ANOVA
df SS MS F Significance F
Regression 1 533.59 533.59 13.13 0.00
Residual 22 893.86 40.63
Total 23 1427.45

Coefficients Standard t stat P-​value Lower 95% Upper 95%

error
Intercept 30.21 18.76 1.61 0.12 -​8.70 69.12
HH 0.0246 0.0068 3.62 0.00 0.0105 0.0386
Statistical model

RENT = 30.21 + 0.0246 HH
     (1.61)  (3.62)

perfect example of a non-​sound model specification because it does not account for a very important
influence on rents, the supply side. Besides not accounting fully for all potential demand influences,
the specification does not include any factor whatsoever that would capture movements in apartment
supply. This is a serious fallacy of the model since the interaction of demand and supply is at the core
of the price determination theory.
• How good is the estimated model in explaining movements in apartment rents through time;
The statistic mostly used to evaluate the overall explanatory power of a linear regression model is the
R2, referred to as R square or R squared. The R squared depicts the proportion of the total variation
in Y explained by movements in X, and therefore it takes values between zero and one. Obviously, the
higher the R squared the better the model in explaining variations in Y. The R squared in the example
presented in Table 7.4 is 0.37, suggesting that movements in the number of households explain only
37% of the variation in apartment rents over the period covered by the time-​series data used. Obviously,
the estimated model is not very good in explaining movements in apartment rents through time. The
adjusted R squared is a more accurate measure of the explanatory power of the model as it takes into
account the number of independent variables included in the model.

147
 Analyzing residential real estate markets

• Does the estimated coefficient of the independent variable, X, depict a relationship between Y and X that is consistent
with economic theory?
Economic theory suggests that if the market is at equilibrium, any increases in demand induce increases
in prices and/​or rents, assuming that supply remains constant. In our example, the number of households
is a driver of demand. An increase in the number of households should induce an increase in the number
of apartment units demanded and an increase in apartment rents for a given stock level. Thus, in theory
the effect of the number of households on rents should be positive. We can examine whether the
estimated regression model depicts such a relationship by examining the sign of the coefficient, b. As
Table 7.4 indicates, this coefficient has a positive sign indeed, which is consistent with economic theory.
If the coefficient estimate came out with a negative sign then the validity of the estimates and the statis-
tical model used would be highly questionable. However, getting the correct sign for the single variable
included in the model does not mean that the formulation is sound and acceptable. For example, as
indicated earlier, this particular model suffers from a significant shortcoming, that is, it does not include
any supply variable.
• How strong is the relationship between Y and X, or apartment rents and the number of households, as reflected in
the coefficient estimate, b?

The estimated parameters presented in (7.5) and (7.6) are merely the outcome of mathematical computations.
How do we know that the estimated average relationship between Y and X represents a strong relationship
that occurs most of the time as opposed to an average of a behavior with no systematic pattern? The math-
ematical computations will always derive a value for b independently of whether a systematic relationship
between Y and X is depicted by the data or not. Therefore, a statistic is needed that will enable the analyst
to evaluate whether the estimated coefficient reflects a systematic or, in statistical terminology, a “statistic-
ally significant” effect.
To better understand the rationale underlying a test of statistical significance for a regression coefficient,
consider that the estimate of b may not represent its true value. In theory, the true value of b is obtained
when the regression estimate uses all units of analysis that fall within the scope of the relationship examined
(Hamburg, 1979). This definition is more easily comprehensible in the case of cross-​section data, but it is
more difficult to operationalize in the case of time-​series analysis. A fairly reasonable interpretation as it
pertains in our example is that the true value of b would be obtained if the regression were estimated with
a historical series spanning much longer than 24 years.
Within this context, if we estimated regressions based on other historical series of similar or different
time-​span, we could get different average values for b. The Central Limit Theorem postulates that if we were
to plot the frequency distribution of these average values, we would obtain a bell-​shape curve referred
to as the normal distribution.2 This distribution is described in more detail in every statistical textbook.
For the purpose of this example, it is sufficient to mention that its basic characteristic is that its values are
distributed symmetrically around the average with most values being concentrated closer to the mean and
fewer values further away from the mean.
An important property of this distribution is that if we know its mean and its standard deviation, we
can compute the different values that it takes and the respective frequency or probability in which they
occur.3 For example, if the true value of b were zero, then based on the shape of the standard normal distri-
bution, we can argue that 95% of sample means would have values ranging from –​1.96 standard deviations
to +1.96 standard deviations. So if our estimate of b does not fall within this range we can state with 95%
confidence (referred to in statistical terminology as confidence level) that the true value of b is statistically
different from zero. This in turn would provide evidence that the average relationship between Y and X is
strong and statistically significant.
Within this context, evaluating the strength or statistical significance of any regression coefficient esti-
mate involves three basic steps:

148
 The basics of the econometric approach

1 . Obtain the t-​statistic from the regression estimate.

2. Determine the t-​critical from a t-​distribution table based on:
a) Degrees of freedom of the estimate calculated as

df = N–​K,   (7.7)

where N is the number of observations used to estimate the model and K is the number of
estimated parameters, that is, the number of explanatory variables plus one.
b) Confidence level, typically 95%
3. Compare the t-​statistic of the coefficient to the t-​critical:
If the absolute value of the t-​statistic ≥ t-​critical then the coefficient estimate can be deemed as stat-
istically significant.

The first step in evaluating the statistical significance of the relationship between Y and X in our
example is to obtain the t-​statistic associated with the estimate of the coefficient for HH. This statistic is
calculated as the ratio of the coefficient estimate over its standard error or standard deviation (listed in
Table 7.4 adjacent to the coefficient estimates). Thus, in actuality, the t-​statistic tells us how many standard
deviations this coefficient estimate is away from zero. To better understand this statement, consider the for-
mula used to calculate this statistic:

b − 0 0.02456 − 0
t= = = 3.62 (7.8)
sb 0.00677

To understand the concept of the standard error consider that the regression algorithm computes values
of a and b for every one of the 24 pairs of observations (a pair of RENT and HH per year). The reported
estimates for the constant and the coefficient in Table 7.4 are the averages of those 24 estimates.The standard
error describes in substance the range of these estimates around these averages. Notice that in order to esti-
mate the t-​statistic we use the sample estimate of the standard deviation, which may not be the same as the
standard deviation of the population, that is, the larger universe from which the sample has been drawn. It
has been shown that when the sample is small, as in our example, then the values of the ratio described by
(7.7) do not exactly follow the standard normal distribution (as is the case for the large samples) but the t
distribution. Hence the name t-​statistic.
The t-​probability distribution is symmetric like the normal distribution, but it has fatter tails because a
greater proportion of values are further away from the mean. As the degrees of freedom (explained below)
increase, the t-​distribution approaches the normal distribution.
Using the t-​distribution, we can make several inferences with respect to the range of values and the
statistical significance of the coefficient estimates presented in Table 7.4. For example, given the estimated
average value and standard error for coefficient b, we can conclude, based on the t-​distribution, that 95% of
its values fall between 0.0105 and 0.0386 (see Table 7.4 under the columns “Lower 95%” and ‘Upper 95%”).
The second step in evaluating the statistical significance of b is to determine the value of the t-​critical.This
is the t-​value that corresponds to the confidence level the analyst wants to use to reject the hypothesis that the
true value of b is zero. As indicated earlier, a 95% confidence level (implying that there is a 95% probability
that b is not zero) is typically used for this purpose.The t values that correspond to different confidence levels
or significance levels can be found from t-​distribution tables that are provided in any statistics textbook. The
significance level is calculated as 1-​confidence level, and is referred to in Table 7.4 as the P-​value. Notice
that in order to use a t-​distribution table one needs to know the degrees of freedom (df) associated with
the estimates. Applying (7.7) in our example we get df = 24 –​ 2 = 22. The t-​statistic that corresponds to
a P-​value of 5% for df = 22 is 2.074, which represents the t-​critical for this test. As the t-​distribution table

149
 Analyzing residential real estate markets

indicates, for df ≥ 120 the t-​value that corresponds to a P-​value of 5% is 1.96, which demonstrates that as the
sample becomes larger the t-​distribution approximates the normal standard distribution.
The third step in evaluating the regression coefficient is to compare the absolute value of b’s t-​statistic
to the t-​critical. Obviously, in our example, the former, which has a value of 3.62, is considerably greater
than the latter (2.07). Therefore, we can clearly conclude that the estimated coefficient is statistically sig-
nificant at the 95% confidence level. Actually, the P-​value listed adjacent to the t-​statistic in Table 7.4 is
0.0015, suggesting that the effect of HH on apartment rents is statistically significant at least at the 99%
confidence level.

Multiple regression models
Multiple regression models are typically used in real estate market analysis because most of the variables
we are interested in predicting are driven by more than one factor. For example, apartment rents are driven
not only by the number of households in a market but also by several other demand and supply factors,
such as age composition of the population, average household size, median household income, user cost of
homeownership, and the size of the existing apartment inventory.
In interpreting the coefficients of a multiple regression model, it is important to remember that each
coefficient measures the change in Y associated with a unit change in the independent variable it refers to,
assuming that the values of all other explanatory variables are held constant. That is why they are often referred to
as partial regression coefficients. As indicated earlier, the analyst should refrain from drawing any conclusions
regarding the relative importance of the independent variables based on the relative magnitude of the
­coefficient estimates. For this purpose, one needs to estimate the beta coefficient of the each of the explana-
tory variables included in the model (see Appendix 7A).
When dealing with a small number of observations, caution needs to be exercised to avoid including
an excessive number of explanatory variables. A general rule of thumb is that the number of explanatory
variables (including the intercept) should not exceed one-​fifth of the number of observations.
Finally, pairs of independent variables that are highly correlated should be avoided. If such two variables
are included in the model then the results will be distorted because of multi-​collinearity.

Chapter summary
This chapter has discussed the very basics of regression analysis and how it can be used, in general, as a
powerful analytical and forecasting tool in real estate market studies. Regression techniques study the
dependence of one variable (the dependent variable) on one or more variables (the independent variables)
that in theory drive its movements through time or influence variations in its levels across different units
of analysis. Regression analysis can help quantify scientifically the nature and strength of the relationships
between the dependent and the independent variables. This quantification can then provide the basis for
generating forecasts for the dependent variable conditioned on predictions for the exogenous variables that
drive the model (hence the term conditional forecasts).
Regression models can provide powerful analytical tools at both the macro and micro level of market
analysis for real estate:

• At the macro level, time-​series regression models can be used within the framework of the stock-​flow
model for generating forecasts of movements in real estate demand, supply, and prices/​rents by:
• quantifying the effect of movements in exogenous economic forces (such as employment and
income) and prices/​rents on real estate demand and supply
• quantifying the feedback effects of demand and supply on prices and rents

150
 The basics of the econometric approach

• combining the estimates of behavioral parameters with the forecasts of exogenous economic
drivers to simulate the behavior of the local real estate market over the project planning horizon.
• At the micro level of analysis, cross-​section models can be used to:
• estimate the value attached by consumers to different project attributes and use that as a basis for
defining optimal amenity packages and refining project design
• derive estimates of the price or rent a specific project is more likely to command in the marketplace,
given its characteristics
• determine residual land values and optimal development densities based on estimates of the prices
that alternative amenity packages will command in the market.

Questions
1 . Discuss what regression analysis is in general useful for.
2. Describe the two major types of data that can be used for regression analysis and how each type of
analysis could be useful in evaluating a particular project or property.
3. Write the equation for a linear regression model with house prices as the dependent variable and two
independent variables. Explain what each term of the equation represents.
4. Explain which variables are characterized as exogenous and which ones as endogenous.Why are these
terms relevant in real estate market analysis?
5. Write the statistical equation for the best hedonic price model in terms of functional form that can be
estimated from a dataset of 250 transactions that took place last year and include the following variables: sales
price, total square meters of covered space, number of bedrooms, number of bathrooms, and quality of con-
struction provided in three categories (moderate, good, and very good). How would you introduce in the
model the time of transaction if the data include the quarter in which each transaction occurred?
6. What are the four questions the analyst needs to ask when evaluating a regression model and what
statistics need to be examined in order to answer each question?
7. Explain the term “statistical significance” in plain language and discuss how one can evaluate the stat-
istical significance of an explanatory variable.

Notes
1 For alternative specifications of the rent equation, see the discussion in Chapter 15.
2 A frequency distribution depicts how often different values of a variable occur. For example, let’s assume that the
average values of b, estimated from ten different samples, range between 1 and 3. A very crude plot of the frequency
distribution of this variable could depict what percentage of these ten values falls between 1 and 2 and what per-
centage falls between 2 and 3. A more refined plot of the frequency distribution could depict what percentage of
the ten values fall within shorter value intervals, such as, 1–​1.2, 1.21–​1.4 and so on. Please see Hamburg (1979) for
a more elaborate discussion of frequency distributions.
3 The standard deviation is a statistical measure that reflects how spread the values of a variable are around its average.
Please see Hamburg (1979) for an elaborate discussion of measures of central tendency and dispersion.

References and additional readings

Brooks, C. and S. Tsolacos. 2010. Real Estate Modelling and Forecasting. Cambridge, UK: Cambridge University Press.
Clapp, J. 1987. Evaluating the Competition with Regression Analysis. Chap.  11 in Handbook for Real Estate Market
Analysis. Englewood Cliffs, NJ: Prentice Hall.
Hamburg, M. 1979. Basic Statistics, 2nd Edition. New York: Harcourt Brace Jovanovich.
Kachigan, S. 1986. Regression Analysis. Chap. 11 in Statistical Analysis. New York: Radius Press.
Pindyck, R. and D. Rubinfeld. 1991. Econometric Models and Economic Forecasts, 4th Edition. New York: McGraw-​Hill.

151
 Analyzing residential real estate markets

Appendix 7A: Further information on linear regression models

Description and estimation: The mechanics of the linear regression model
The simplest linear regression model that includes only one explanatory variable is described by the equation:

Yi = a + bXi + ei (7A.1)

or:

Yi = a + bXi (7A.2)

Estimation
Looking at the mechanics of a simple regression model with one explanatory variable can help to better understand
the underlying logic behind the estimation procedure. Figure 7A.1 portrays the basic metrics that underlie the estimates
of the coefficients a and b.
Decomposing the variation of Y around its mean

Yi = Y = (Yi − Y i ) + (Y i − Y ) (7A.3)

∑ (Y i − Y )2 = ∑ (Yi − Y i )2 + ∑ (Y i − Y )2 (7A.4)

TSS ESS RSS

Total Sum of Squares Error Sum of Squares Regression Sum of Squares
Total Variation of Y Unexplained Variation of Y Explained Variation of Y

Regression
line
minimizes ESS
(see below)

Figure 7.A1 The mechanics of the linear regression model

152
 The basics of the econometric approach

Equation (7A.3) decomposes the deviation of each sample observation, Yi, from the mean of Y (Y ) to the error
 i) and the difference between the predicted
(difference of actual value of each observation Yi from its predicted value Y
value of Yi (Y i) and the mean of Y (Y ). By squaring (7A.3) and using two properties of least-​squares residuals, we
can derive (7A.4), which states that the total variation of Y around its mean, referred to as Total Sum of Squares
(TSS), equals the sum of the unexplained or residual variation of Y, referred to as Error Sum of Squares (ESS) and
the explained variation of Y, referred to as Regression Sum of Squares (RSS). See Pindyck and Rubinfeld (1991,
p. 62) for a more detailed discussion of this derivation. In our example, these statistics are listed in the subsection of
Table 7.4 titled ANOVA (which stands for analysis of variance) under the column labeled “SS.” In particular, the first
number in this column is the RSS, the second number is the ESS, and the last number is the TSS. Given the low
explanatory power of the model, it is not surprising that the explained variation of Y, RSS, is smaller than its unex-
plained variation, ESS.

Evaluation of regression models: Simple evaluation measures

R-​squared  (R )2

R2 =
RSS
=
∑ (Y i − Y )2
(7A.5)
TSS ∑ (Y i − Y )2

The R-​squared measures the percentage variation in the dependent variable explained by the set of independent
variables included in the model. By definition, the R2 ranges between zero and one, and it is equal to the ratio
of RSS over TSS, explained above. The adjusted R-​squared, R2, shown below, with N representing the number
of observations and K the number of explanatory variables (Xs), adjusts for the model’s degrees of freedom. The
rationale for this adjustment is that the introduction of additional independent variables would result in loss of
degrees of freedom:

2 N −1
R = 1 − R2 (7A.6)
N − K −1

Therefore, when experimenting with specifications with different numbers of independent variables, it is more
appropriate to look at the adjusted R-​squared in order to get a sense of the relative explanatory power of the different
models.

Standard error of the regression estimate, s

∑ (Y i − Y i )2
s= (7A.7)
N −K

The standard error of the estimate measures the dispersion of the error term around the regression line.

Standard error of the coefficient, sb

s
sb = (7A.8)
∑ (X i − X )2

The standard error of the coefficient provides a measure of the dispersion of the coefficient estimate around its mean
and it is used in calculating the t-​statistic. As formula (7A.8) indicates, this statistic is calculated as the ratio of the
standard error of the estimate, s, over the square root of the sum of squared differences of each observation Xi of the
independent variable to which the coefficient refers from that variable’s mean X .

153
 Analyzing residential real estate markets

T-​statistic
b
tb = (7A.9)
sb

The t-​statistic provides a measure of the statistical significance of the coefficient estimates. As formula (7A.9) indicates,
this is calculated as the ratio of the estimated coefficient over its standard error. As indicated earlier, if the t-​statistic is
greater than its critical value (t-​critical), then the null hypothesis, H0, that b=0 can be rejected (see discussion below
about hypothesis testing). The critical value, t-​critical, can be identified from t-​distribution tables using:

• The regression equation’s degrees of freedom (number of observations minus number of independent variables
minus one)
• A confidence level (usually 95%).

Hypothesis testing: Evaluating the strength of variable effects

Evaluating the strength of the effects of the independent variables involves a process of hypothesis testing. Any
hypothesis-​testing procedure starts with the articulation of the hypothesis we would like to reject, referred to typically
in statistical textbooks as the null hypothesis and denoted as H0. In the example we discussed earlier, the null hypothesis
is that the coefficient of HH, bHH, is zero.The alternative hypothesis, that is, the one we favor (denoted as H1), is that the
coefficient of HH, bHH, is different from zero:

H0: bHH = 0

H1: bHH ≠ 0

The test of the statistical significance of bHH described earlier was actually a test of H0 as expressed above. To repeat
briefly, this process consists of defining a desired confidence level, which in turn is used as a basis for identifying
the t-​critical from t-​distribution tables taking also into account the degrees of freedom of the estimate. If the abso-
lute value of the estimated coefficient’s t-​statistic is greater than the t-​critical then we can reject H0 in favor of the
alternative H1.

Evaluating the relative importance of the independent variables in

multiple regression models
A typically used statistic for the evaluation of the relative importance of the different independent variables in multiple
regression models is the so-​called beta coefficient, which is calculated as:

sXi
βi = b i (7A.10)
sY

where
s X i : standard deviation of the independent variable under consideration
sY : standard deviation of the dependent variable

A beta coefficient indicates by how many standard deviations the dependent variable changes in response to a standard-​
deviation change in the independent variable.

154
newgenprepdf

8
MACROECONOMIC ANALYSIS
OF RESIDENTIAL REAL
ESTATE MARKETS
Applying the econometric approach

Introduction 155
Applying the econometric approach to the Dallas apartment market 156
Step 1: Collect data on apartment demand, supply, rents, and their drivers 157
Step 2: Set up/define a behavioral model 159
Specifying the statistical equations 159
Step 3: Estimate the model 164
Rent equation 164
Permit equation 165
Step 4: Develop forecasts and perform sensitivity analysis 167
Chapter summary 174
Questions 175
Project #3: Forecasting the Los Angeles apartment market 176
The data 176
Data analysis 178
References and additional readings 179
Appendix 8A: Calculating the speed of rental adjustment 179

Introduction
In the previous chapter, we discussed the advantages of econometric techniques in forecasting absorp-
tion, rent, and revenue schedules for a specific project, as well as the insights they can provide in guiding
entry decisions. But how do we specifically explore future real estate market outlooks using the econo-
metric approach? In this section we demonstrate how this can be done through a series of simple steps by
presenting a more elaborate and integrated application of this approach to the Dallas apartment market.
In particular, we examine the use of structural/​behavioral models in developing conditional forecasts of
apartment rent movements, new supply, and apartment stock over a five-​year period.
A structural model is a model that takes into account both demand and supply forces, as well as the price
determination mechanism. The form of these models is dictated by the underlying economic theory, not
by past trends, as in linear extrapolation models. Usually such models consist of a system of equations and
a system of identities that allow the analyst to produce conditional forecasts. The latter are referred to as

155
 Analyzing residential real estate markets

conditional because they are conditioned on the particular forecasts for the exogenous variables that enter
the model. Some of these models are simple, while some of them are quite complicated. Their level of
complexity does not depend so much on the level of accuracy that needs to be achieved, but rather on the
complexity of the market that is being modeled.

Applying the econometric approach to the Dallas apartment market

The discussion of the basics of regression analysis in the previous chapter focused on a simple two-​variable
model in order to keep the material as simple and understandable as possible.We now move to a more advanced
application of regression analysis, the development of a forecasting model for the Dallas apartment market,
which involves multiple regression analysis and more than one statistical equation. Setting up a forecasting
model is not an easy task, but it is quite doable if the analyst has a good knowledge of the functioning of the
market and a good understanding of the basics of statistical and econometric techniques. Below we review
the series of steps that need to be carried out in setting up and estimating the simplest possible structural/​
behavioral model for the apartment market in Dallas. These steps are summarized in Figure 8.1.

1 . Collect data on apartment demand, supply, rents, and their drivers.

2. Set up/​define a behavioral model.
3. Estimate the model.
4. Develop forecasts and perform sensitivity analysis.

Forecasting apartment rents,

new construction, and apartment stock

STEP 1
Collect time series data
on demand, supply,
and rents

STEP 2
Develop
a forecasting model

STEP 3
Estimate
the rent and permit
regression models

STEP 4
Forecast
rents, permits,
and apartment stock

Figure 8.1  Applying the econometric approach to the Dallas apartment market

156
 Applying the econometric approach

Step 1: Collect data on apartment demand, supply, rents, and their drivers
The primary requirement for setting up any econometric forecasting model is the availability of historical
data. Thus, the first step is to collect the longest time-​series possible on apartment demand, supply, rental
rates, and their drivers, for the Dallas metropolitan area. Table 8.1 presents such historical information for
30 years starting in 1967.This data will be used in order to demonstrate the application of the econometric
approach. Notice that no direct measure of apartment demand, such as the number of units that are occu-
pied each year, is available. Historical information on apartment completions is not available either, but
available permit information serves as a very good indicator of changes in the area’s apartment inventory
through time. The data and their sources are described in Table 8.1. Some further clarifications, however,
are needed.

• The information presented in the column labeled USER(T) represents the after-​tax user cost of capital
and has been calculated using the formula below:

USER = (1–​τ)r –​i   (8.1)

where:
τ: marginal tax rate, assumed 28% for this example
r: nominal mortgage rate (MORTG in Table 8.1)
i: inflation rate (INFL in Table 8.1), representing changes in the general consumer price index and
proxying house price appreciation expectations
Although this variable is a rough measure of the homeownership cost, it is very relevant in
assessing rental demand, since when USER is high one would expect more households to select
renting versus owning, while when USER is low one would expect demand to shift in favor of
homeownership.
• The apartment stock series has been constructed by combining the yearly permit data reported by the
U.S. Department of Commerce with the stock data reported by the decennial censuses. For example,
the first step in creating the annual stock series for the period 1970–​1980 was to add all the permits
issued during the period 1970–​1979. Assuming a one-​year lag between permit issuance and comple-
tion, this sum should reflect apartment completions over the period 1971–​1980. This figure was then
added to the 1970 Dallas stock of multi-​f amily units in structures with two or more units, as reported
by the 1970 census, in order to produce an estimate of the 1980 stock. The estimated stock number
was then compared to the stock level reported by the 1980 census and an annual depreciation rate
was derived based on the difference between the two numbers. The annual stock series for the period
1971–​1979 was then created by incorporating the estimated annual depreciation rate and annual
permit numbers into the standard stock-​flow equation and by applying the latter to the 1970 stock
level reported by the census. The same methodology was used to create the annual stock series for the
other decades.
• The rent data in the column labeled RRENT(T) represent the apartment rent component of
the CPI index for Dallas, in 1996 constant dollars.1 This information was obtained from the
Detailed CPI reports, which were published by the U.S. Department of Commerce but have been
discontinued since 2017. These data were collected through a survey that was carried out in 34
cities and repeatedly sampled the same units in each market. As such, they refer to a constant-​
quality sample and represent a true market rent index. Keeping the quality of the unit of reference
constant is very important in time-​series analysis because the analyst does not have to worry about
changes in the quality mix of the apartment stock, which would have an effect on changes in the
average rent through time.

157
 newgenrtpdf
Table 8.1  Step 1: Times-​series data on demand and supply

Dallas, MSA:
Forecasting apartment rents,
new construction, and apartment stock

Step 1: Collect time series data on demand, supply, and rents

YEAR INC (T) EMP(T) DEMP(T) IPE(T) USER(T) PRMT(T) PRMT STOCK(T) RRENT(T) RRI(T) DRRI(T) MORTG(T) INFL(T)
(billions) (×1,000) (×1,000) (×1,000) (%) (T–​1) ($/​unit) (%) (%) (%)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
1 25.32 554 45.7 1.77 9,644 60,958 658.63 100.00 6.6 3.0
2 27.59 596 42 46.3 0.43 21,388 9,644 70,602 644.36 97.83 –2.2 7.1 4.7
3 29.91 646 50 46.3 –0.28 16,607 21,388 91,990 627.08 95.21 –2.7 8.1 6.1
4 30.93 639 –7 48.4 0.84 18,414 16,607 1,08,597 617.34 93.73 –1.6 8.8 5.5

23 69.89 1,395 25 50.1 2.46 538 1399 3,96,354 477.36 72.48 –4.6 9.8 4.6
24 71.03 1,435 40 49.5 0.87 4,901 538 3,96,892 465.84 70.73 –2.4 9.7 6.1
25 73.00 1,423 -​12 51.3 3.49 3,124 4,901 4,01,793 471.44 71.58 1.2 9.0 3.0
26 76.27 1,439 16 53.0 2.85 1,862 3,124 4,04,917 467.41 70.97 –0.9 8.0 2.9
27 78.18 1,492 53 52.4 2.58 4,386 1,862 4,06,779 465.03 70.61 –0.5 7.3 2.7
28 81.85 1,562 70 52.4 3.41 10,156 4,386 4,11,165 470.10 71.37 1.1 8.4 2.6
29 85.86 1,617 55 53.1 3.22 11,475 10,156 4,21,321 475.57 72.21 1.2 8.0 2.5
30 89.55 1,696 79 52.8 2.32 10,612 11,475 4,32,796 474.43 72.03 –0.2 7.8 3.3

The non-​shaded variables have been constructed on the basis of the shaded variables
Data description:
INC: Total personal income (billions) (Bureau of Economic Analysis)
EMP: Total nonfarm employment (×1,000) (Bureau of Economic Analysis)
IPE: Total personal income per worker; in constant dollars (×1,000)
USER: User cost of homeownership (%)
PRMT: Authorized multi-​family permits (U.S. Department of Commerce, Construction Reports, Series C-​40)
STOCK: Stock of multi-​family units
RRENT: Residential rent component of CPI-​U (U.S. Department of Commerce in constant 1996 dollars)
RRI: Real rent index (Basis 1967)
DRRI: Percentage change in real rent index
INFL: Inflation rate (U.S. statistical abstract)
MORTG: Prime mortgage rate (FHLBB corporation, Washington, DC)
 Applying the econometric approach

• RRI(T) is a real rent index with 1967 as basis, constructed using RRENT(T).
• Table 8.1 includes also some variables constructed from the basic demand, supply, and rent data collected,
such as DEMP(T), PRMT(T-​1), and DRRI(T), which will be used as independent variables in the
forecasting equations presented later in this section. DEMP(T) represents absolute change in employ-
ment at time T calculated as EMP(T) minus EMP(T-​1), while DRRI(T) represents percentage change
in the real apartment rent index (RRI) at time T.

Step 2: Set up/​define a behavioral model

Once the data are collected, the next step is to develop a forecasting model. The simple version of the
stock-​flow model (that does not include the vacancy rate) can provide the basis for carrying out this task.
The fundamental dynamics of the model, outlined in Figure 8.2, can be described as follows:

• During each period, rents must adjust appropriately so that demand for apartment units equals the
existing apartment stock. In other words, in each period the rents adjust appropriately so as to restore
equilibrium in the market. That is the reason why this modeling framework does not include the
vacancy rate.
• At any period rents may rise (or fall) because of changes in exogenous demand drivers.
• Responding to increases in rents and increasing profits, residential developers and investors develop
plans and apply for building permits in order to produce more housing units.
• A portion of these permits becomes new construction and a new level of stock is established in the
market.
• Rents must again adjust so that the new apartment stock level equals demand.

Specifying the statistical equations

In order to develop a forecasting model for the Dallas apartment market, we need to translate the dynamics
described by Figure 8.2 into behavioral equations and identities.2 These dynamics describe in essence the
behavior of demand, rents, and new construction (permits). Given the lack of a direct measure of absorp-
tion of apartment units, an equation describing the behavior of demand cannot be estimated. It is possible,
however, to estimate regression models describing the behavior of rents and permits. Since the fundamental
premise of the model is that rents adjust during each period to equalize demand with supply, the rent
equation needs to account for both demand and supply influences.Thus, although the impact of exogenous
demand drivers on the market is not accounted by a separate demand equation, they are fully accounted in
the rent equation. A permit equation needs to be also developed in order to capture the impact of changes
in market rents and the local economy on the area’s apartment construction activity. Finally, the stock-​flow
identity can be used to translate apartment permits into new construction and new inventory levels, which
will in turn be shaping each period’s rent levels.
In sum, the forecasting model for the Dallas apartment market consists of two behavioral/​statistical
equations (the rent and permit equations) and one identity (the stock-​flow identity). All the equations of
the model are presented in Boxes 8.1 and 8.2. Box 8.1 presents the behavioral equations in general form,
that is, without specifying the mathematical relationship (linear or non-​linear) between the dependent
and the independent variables, while Box 8.2 presents the exact statistical specifications of the behavioral
equations.
It should be noted that two versions are presented for each of the two behavioral equations. The
first version reflects the assumption of a quick (single-​period) adjustment process, while the second

159
 Analyzing residential real estate markets

STEP 2: Develop a simple forecasting model

THE SIMPLE VERSION

OF THE STOCK-FLOW MODEL

THE LOGIC:

DEMAND At any point in time, rents must adjust appropriately

so that demand equals the existing apartment stock

At any point in time, rents may rise because of

RENT increases in exogenous demand drivers

Suppliers respond to increases in rents by deciding

to provide more space and applying for building permits

STOCK
A portion of permits becomes new construction and a
new level of stock is established in the market

PERMITS Rents must then adjust again so that the new level of
apartment stock equals demand

COMPLETIONS

Figure 8.2  Step 2: Developing a simple forecasting model

reflects the assumption of a slow (multi-​period) adjustment process. From a specification point of view,
the difference between the two versions is the addition of the lagged dependent variable R(T-​1) in (8.13)
and C(T-​1) in (8.15) on the right-​hand side of the formulation reflecting the slow adjustment process.
In general, the term “lagged dependent variable” is used when values of the dependent variable in pre-
vious periods are used as explanatory variables. As it is pointed out in many econometric textbooks,
such a specification is consistent with the assumption that the dependent variable adjusts slowly over
several periods.

160
 Applying the econometric approach

Box 8.1 Apartment model: Statistical equations and identities

1.  Rent equation
Quick rent adjustment process
D(T) = S(T) (8.2)
==>D [R(T), XD (T)] = S(T)   (8.3)

=>R*(T) = R [XD(T), S(T)]   (8.4)

Slow rent adjustment process

R(T) –​ R(T-​1) = gR [R* –​ R(T–​1)]   (8.5)
R(T) = (1–​gRR(T–​1) + gRR*   (8.6)

R* = R [R(T–​1), XD(T), S(T)]   (8.7)

2.  Completions equation

Quick completions adjustment process

C*(T) = C [XC(T–​m), R(T–​m)]   (8.8)

Slow completions adjustment process

C(T) –​ C(T–​1) = gC[C* –​ C(T–​1)]   (8.9)

C(T) = C [C(T–​1), XC(T–​m), R(T–​m)] (8.10)

3.  Stock-​flow identity

S(T) = S(T–​1)(1–​d) +C(T) (8.11)

Notation:
D Demand Xc Set of supply shifters
S Stock gR Speed of rent adjustment
R Prevailing market rent gC Speed of supply adjustment
R* Equilibrium (market-​clearing) rent d Depreciation rate
XD Set of demand shifters t Time indicator
C New construction (completions) m Lag indicator
C* Equilibrium (market-​clearing) completions

161
 Analyzing residential real estate markets

Box 8.2  Statistical specifications of the behavioral equations

1.  Rent equation
Quick rent adjustment process

R * (T ) = b0 + b1X 1D (T ) +  + bn X nD (T ) + bn +1S(T ) (8.12)

Slow rent adjustment process

R(T ) = b0 + b1X 1D (T ) +  + bn X nD (T ) + bn +1S(T ) + bn + 2R(T − 1) (8.13)

2.  Completions equation

Quick completions adjustment process

C * (T ) = c 0 + c1X 1c (T − m ) +  + c n X nc (T − m ) + c n +1R(T − m ) (8.14)

Slow completions adjustment process

C (T ) = c 0 + c1X 1c (T − m ) +  + c n X nc (T − m ) + c n +1R(T − m ) + c n + 2 (T − 1) (8.15)

3.  Stock-​flow identity

S(T ) = S(T − 1)(1 − d ) + C (T ) (8.16)

1
 It is important to note that in the statistical equations above all independent variables are entered with a posi-
tive sign simply by statistical convention. Therefore, this sign does not necessarily reflect the direction of the
expected effect for the explanatory variables included in the specification.

To better understand the difference between the quick and the slow rent-​adjustment mechanism let’s
assume that one or more of the demand drivers, XD, increases. As a result, the demand for apartments will
shift upward requiring rents to adjust to a new equilibrium level R**. The quick rent adjustment model
postulates that such an adjustment takes place rapidly within a single period, whereas the slow rent adjust-
ment model postulates that market rental rates will adjust gradually over several periods before they reach
the new equilibrium level R**.
Slow rent adjustment processes in real estate markets are forced by long rental contracts, market
imperfections, such as informational inefficiencies, product heterogeneity in terms of location, characteristics
and quality, and simply delayed responses of landlords to changes in market conditions. The heterogeneity
of an area’s housing stock may contribute to a slow rent adjustment process because it may force lengthy
search processes on the part of tenants that want to move. Long rental contracts, with terms ranging usually
from three to ten years, are prevalent in the office, industrial, and retail markets. Apartment rental contracts,
however, are typically much shorter (six months to a year), so in this sense they would not be the cause of
any slow adjustments in annual rental rates.
It is often argued that, because of the short rental contracts characterizing the apartment market, the
quick adjustment specification may depict better its rent dynamics. However, this may not be necessarily
the case because of the aforementioned market inefficiencies and, especially, slow landlord pricing responses
to changes in market tightness or softness.3 Such pricing responses are likely to be very different depending
on whether the landlord is dealing with a new tenant or an existing tenant. For example, landlords, in their
effort to attract new tenants for their vacant units, will be hard-​pressed to lower rates in light of soft market

162
 Applying the econometric approach

conditions. However, when it comes to existing tenants, landlords tend to avoid decreasing rates when the
market is soft, because they are very well aware of the reluctance of the former to move. Rents for existing
tenants stay constant or rise; they rarely decrease.
If landlord pricing behavior varies across existing tenants and new tenants, movements in rates charged
to these two market segments may differ considerably. In particular, rates charged to new tenants are more
likely to move more in sync with changes in market fundamentals, while rates charged to existing tenants
may move at a slower pace. Such a behavior suggests that a slow rent adjustment specification may better
describe the behavior of the apartment market, if the historical rent measure used by the analyst describes
largely rents for existing tenants. From an econometric point of view, the best strategy is to estimate both
a slow and a quick rent adjustment specification and use the one that best explains the behavioral pattern
of the apartment market under consideration.
It is important to understand that in the case of slow rent adjustment processes, at any given period,
but especially during periods following significant shifts in exogenous drivers, the prevailing market rent,
R(t), may be different than the market-​clearing rent, R*, implied by those periods’ exogenous demand and
supply shifters. The rent adjustment equation (8.5) states that under such conditions the change in market
rents, R(t) –​ R(t–​1), in each period is equal to a fraction, gR, of the difference between the prevailing market
rent in the previous period, R(t–​1), and the market-​clearing rent, R*.Thus, the coefficient gR represents the
percentage of the difference between actual and equilibrium rent that is eliminated in each period. As such,
it reflects the speed by which market rents adjust towards the market-​clearing rent R*, and it is referred to
as speed of adjustment. Obviously, gR (and gC in equation (8.9), which has a similar interpretation) can take
values only between zero and one. A value close to one would suggest a rapid adjustment, while a value
close to zero would suggest a very slow adjustment.
The same modeling logic is behind the two alternative completion equations.The slow-​adjustment spe-
cification of completions reflects the premise that development projects initiated in response to changes in
market conditions within a given period are completed over several periods. Such a premise makes sense
for a number of reasons. First of all, the size of apartment developments may vary dramatically ranging
from simple two to five unit structures to large apartment complexes with several hundred units. Although
both small apartment structures and large apartment complexes may be initiated during the same period,
not all projects will be completed within the same period, given the longer time needed for the larger
apartment complexes at all stages of the investment and development process. Second, even projects of the
same size may follow considerably different timetables, because of idiosyncratic project features that may
involve more or less lengthy financing, planning, and regulatory processes, as well as because of the varying
capabilities and agendas of the key actors involved.
It can be shown mathematically (see Appendix 8A) that in the case of slow adjustment processes, either
in rental rates or completions, the estimated coefficients of the lagged dependent variables bn+2 and cn+2 in
equations (8.13) and (8.15), respectively, are equal to (1 –​gR) and (1 –​gC), respectively, and represent the
so-​called persistence rate.This rate is the percentage of the deviation between the prevailing rent or comple-
tion levels from their equilibrium levels that persists into the next period (hence the term persistence rate).
This is an important point because we can use these estimates to calculate the speed of adjustment for rents
and completions as gR = (1–​bn+2) and gC = (1–​cn+2), respectively.
It should be noted that the explanatory variables XD in the demand and rent equations (8.3) and
(8.4) refer to the exogenous demand shifters discussed earlier, such as the user cost of homeownership,
market size, and income. Similarly, the explanatory variables, XC, in the supply (permit) equation (8.8),
refer to the exogenous drivers of apartment completions, such as expectations regarding the strength of the
local economy and the local apartment market, construction costs, cost of capital, zoning, etc. Given that
apartment construction investment decisions are made some time before the completion of a project, all
variables entering a typical completions equation are introduced with a lag, m. Note that m may vary across
the different explanatory variables included in the supply equation.

163
 Analyzing residential real estate markets

The supply equation includes also the rent and/​or the vacancy rate as high rents and low vacancies
may motivate builders to build more housing. Note however, that, quite often, a wrong sign or low t-​
statistic may be obtained for one or both of these variables when they are introduced together as explana-
tory variables, either contemporaneously or with different lags. The reason is that usually they are highly
correlated since rent movements are strongly influenced by the vacancy rate and vice-​versa. In such a case,
one of the two variables needs to be dropped out of the specification.

Step 3: Estimate the model

The estimated statistical equations include:

1. Rent equation

RRI (T ) = b0 + b1IPE(T ) + b2USER(T ) + b3STEMP (T ) + RRI (T − 1) (8.17)

2. Permit equation

PRMT (T ) = c 0 + c1DEMP (T ) + c 2DRRI (T ) + c 3STEMP (T − 1) + c 4PRMT (T − 1) (8.18)

Note:  Positive signs are used by convention; they do not necessarily reflect the direction of the
expected effect for the explanatory variables included in the specification.

Rent equation
Estimation of both a quick and a slow rent-​adjustment specification for the Dallas apartment market indicated
that the latter (depicted by (8.17)) explains much better historical real rent movements during the period of
analysis. This behavior is consistent with the fact that the rent index represents largely the rent charged to
existing tenants, which is more likely to adjust slowly in response to changes in market conditions.
It should be noted that, putting aside the inclusion of the lagged rent, (8.17) is exactly the same as the
price equation developed by DiPasquale and Wheaton (1996) for the single-​family housing market. The
demand variables, XD, incorporated in the rent equation include IPE (real income per employee) and USER
(user cost of capital). The specification of the demand variables is straightforward because their effect on
apartment rents is contemporaneous, and, therefore, there is no need for experimentation with different time
lags. The supply side is incorporated in STEMP, which is calculated as the ratio of stock over employment
(EMP) and represents stock per employee. Since this measure combines both demand and supply, it can be
thought of as an indicator of the tightness of the apartment market and as a substitute for market vacancy.
To better understand the specification of the rent equation described by (8.17), consider the derivation
of the rent equation, which starts from two fundamental conditions. First, that during each period demand,
D, equals supply, S; and second, that demand for apartments equals the number of households, HH times
the percentage of households that want to rent, HR.4 As (8.21) indicates, this percentage is considered to be
a function of IPE, USER, and RENT.

D = S (8.19)

D = HH HR (8.20)

HR = a0 + a1 IPE + a2 USER + a3 RENT (8.21)

164
 Applying the econometric approach

Substituting (8.20) and (8.21) in (8.19) and using employment E as a proxy for the number of households,
HH, we get:

E (a0 + a1 IPE + a2 USER + a3 RENT) = S (8.22)

a0 + a1 IPE + a2 USER + a3 RENT = S/​E =STEMP (8.23)

Solving (8.23) for RENT we obtain (8.25), which provides the basis for the specification of the rent
equation estimated in this application.

RENT = –​(a0 /​a3) –​ (a1/​a3)IPE –​ (a2/​a3) USER + (1/​a3)STEMP (8.24)

RENT = b0 + b1 IPE + b2 USER + b3 STEMP (8.25)

The estimated coefficients of (8.17), presented in Table  8.2, confirm the expected negative effect of
STEMP on rents, verifying that as the number of apartment units relative to the number of employees –​
which is used as proxy for number of households  –​rises, rents fall in order to equalize demand with
supply. Furthermore, USER and IPE have signs that make economic sense. In particular, USER has a
positive sign, which is consistent with the hypothesis that when the after-​tax user cost of homeownership
increases, demand for rental units increases, as some households will no longer be able to afford to buy
a house and by necessity they will turn to renting. IPE has a negative sign, indicating that the negative
effect of income increases because of demand shifts from rental towards homeownership is greater than
its potential positive effect on rental demand through new household formation that income increases
may encourage.

Permit equation
Historical data on apartment completions are not available at the metropolitan area level.The closest proxy
for which data are available at this geographical level is apartment permits. Thus, we estimate a forecasting
equation for permits, which are then converted to new apartment units by assuming a 95% comple-
tion rate. Coming up with the forecasting equation for apartment permits is not as straightforward as in
the case of rents. This is due to the potential time lag between the time that changes in the factors that
affect development decisions take place and the time that the respective building permits that result from
those development decisions are issued. Thus, some experimentation with different time lags and different
specifications of the independent variables (in terms of levels or changes) is required before choosing the
final forecasting equation for permits.
As indicated by (8.18), fluctuations in apartment permits are best explained by contemporaneous
changes in real rents [DRRI(T)] and employment [DEMP(T)], as well as the stock-​per-​employee ratio of
the previous period [STEMP(T–​1)], and the level of apartment permits issued during the previous period
[PRMT(T–​1)].5 The lack of any time lag in the effect of the first two variables is understandable given that
our analysis is using annual data. This means that the building permit is issued within 12 months from the
time the development decision has been taken, which does not seem unreasonable. A proxy for construc-
tion costs was not included due to the lack of historical data for Dallas going back to 1960. An attempt
to include a national construction cost indicator and an interest rate series as explanatory variables proved
unproductive, as the results showed no statistically significant relationship between these variables and
permit activity in Dallas.
Some comments are warranted with respect to the specific explanatory variables included in the permit
equation. First, the percentage change in the real rent index, DRRI(t), and not the rent index level, RRI(t),

165
Table 8.2  Step 3: Estimating the rent equation
A. Estimating the rent equation
Dependent variable Independent variables
RRI(T) RRI(T–​1) IPE(T) USER(T) STEMP(T)
100.0 45.7 1.77 0.11
97.8 100.0 46.3 0.43 0.12
95.2 97.8 46.3 –​0.28 0.14
93.7 95.2 48.4 0.84 0.17
92.0 93.7 48.4 2.53 0.19
89.1 92.0 48.3 2.08 0.20
83.1 89.1 46.6 –​2.94 0.20
76.2 83.1 46.2 –​5.97 0.20
74.6 76.2 46.5 –​0.64 0.21
75.2 74.6 47.4 1.60 0.21
76.5 75.2 47.8 –​0.29 0.20
77.8 76.5 47.9 –​2.17 0.20
76.2 77.8 46.3 –​6.07 0.21
75.0 76.2 45.5 –​4.39 0.21
76.0 75.0 45.5 0.68 0.21
79.3 76.0 46.9 6.68 0.22
82.8 79.3 47.1 5.08 0.22
83.6 82.8 47.6 4.68 0.25
83.8 83.6 48.3 4.27 0.26
85.8 83.8 48.7 6.08 0.28
81.3 85.8 49.9 2.13 0.29
76.0 81.3 50.2 2.36 0.29
72.5 76.0 50.1 2.46 0.28
70.7 72.5 49.5 0.87 0.28
71.6 70.7 51.3 3.49 0.28
71.0 71.6 53.0 2.85 0.28
70.6 71.0 52.4 2.58 0.27
71.4 70.6 52.4 3.41 0.26
72.2 71.4 53.1 3.22 0.26
72.0 72.2 52.8 2.32 0.26
Selected Excel regression output
Regression statistics
R square 0.97
Adjusted R square 0.97
Standard error 1.39
Observations 29
Coeff. Standard error t stat
Intercept 38.92 8.49 4.59
RRI(T–​1) 0.74 0.04 17.89
IPE(T) –​0.21 0.16 –​1.32
USER(T) 0.69 0.10 7.11
STEMP(T) –​41.28 9.87 –​4.18
Forecasting equation:
RRI(T) = 38.92 + 0.74 RRI(T-​1) –​0.21 IPE(T) + 0.69 USER(T) –​41.28 STEMP(T)
     (8.4) (17.9)       (–1.3)    (7.1)      (–4.2)
 Applying the econometric approach

appears in the forecasting equation because alternative estimates showed that the former captures better
than the latter the effect of rents on the level of apartment permits issued.This makes sense since permits—​
used as proxy for completions—​reflect also change, the change in an area’s inventory of apartment units.
Second, alternative estimates showed that the effect of changes in the local economy on permit activity is
better captured by the absolute change in the area’s employment, DEMP. Third, the stock-​per-​employee
ratio (STEMP) was introduced as a proxy of the vacancy rate or market tightness, as investors and developers
can be expected to be less reluctant to enter the market when it is softer. Finally, the inclusion of the lagged
dependent variable, PRMT(T-​1), reflects the assumption of slow adjustments in apartment permit and
completion levels due to the reasons discussed earlier.
Examination of the estimation results of the permit equation, presented in Table 8.3, verifies that all
coefficients, but DEMP(T), are statistically significant at the 95% confidence level and have signs that
are consistent with economic theory. The positive sign of the coefficients of DRRI(T) and DEMP(T)
make economic sense since one would reasonably expect that during periods of higher employment
growth and higher real rent growth, residential developers would be motivated to build a higher number
of apartment units. The negative sign of STEMP(T-​1) verifies the hypothesized adverse effects of softer
market conditions (as reflected in higher values of this ratio) on residential building activity.
As indicated earlier, the coefficient of the lagged dependent variable, PRMT(T–​1), represents the per-
sistence rate and suggests that 46% of the difference between the level of permits issued in each period and
the market-​clearing level of permits, PRMT*, persists into the next period. Said differently, during each
period 54% (the speed of adjustment, calculated as 1–​0.46) of the difference between the actual permit
level and the market-​clearing permit level is eliminated. This estimate suggests a relatively quick response
of supply to changes in market conditions.

Step 4: Develop forecasts and perform sensitivity analysis

The fourth step involves the development of forecasts and evaluation of their sensitivity to optimistic and
pessimistic scenarios. The sensitivity analysis has mainly two objectives. The first objective is to test the
robustness of the model. For example, small changes in the exogenous drivers should not generate entirely
different results. If this happens then the analyst needs to examine carefully the data and the regression
estimates and try to understand which of the equation or equations contribute to the instability of the
system and why. The second objective of the sensitivity analysis is to provide an insight into the sensitivity
of the local real estate market to changes in the macro economy and the potential risks and opportunities
associated with economic uncertainty in the years ahead.
In order to generate forecasts of rents, permits, and stock for the Dallas apartment market the analyst
can use the estimated rent and permit equations along with the stock-​flow identity in a recursive mode, as
indicated in Figure 8.3. Before using the stock-​flow identity, the depreciation rate and the percentage of
permits that is converted to completions need to be specified in numerical terms. For assigning reasonable
values to these parameters the analyst needs to examine relevant statistical data on depreciation and com-
pletion rates. However, for the purpose of this exercise, we assume a value of 0.005 for the depreciation rate
and 0.95 for the completion rate, respectively. In sum, the following three equations are used to develop
forecasts:

RRI(T) = 38.92 –​0.21 IPE(T) + 0.69 USER(T) –​ 41.28 STEMP(T) + 0.74 RRI(T–​1) (8.26)

PRMT(T) = 19,566.60 + 0.46 PRMT(T–​1) + 1,385.90 DRRI(T) + 52.55 DEMP(T)

–​ 58,601.75 STEMP(T–​1) (8.27)

STOCK(T) = STOCK(T–​1) (1–​0.005) + 0.95 PRMT(T–​1) (8.28)

167
Table 8.3  Step 3: Estimating the permit equation

B. Estimating the supply equation

Dependent variable Explanatory variables
PRMT(T) PRMT(T-​1) DRRI(T) DEMP(T) STEMP(T-​1)
9,644
21,388 9,644 –​2.2 42 0.11
16,607 21,388 –​2.7 50 0.12
18,414 16,607 –​1.6 –​7 0.14
13,634 18,414 –​1.9 26 0.17
10,748 13,634 –​3.1 52 0.19
7,773 10,748 –​6.8 54 0.20
8,001 7,773 –​8.3 10 0.20
2,889 8,001 –​2.2 14 0.20
6,844 2,889 0.9 28 0.21
14,731 6,844 1.7 45 0.21
15,741 14,731 1.7 75 0.20
12,140 15,741 -​2.1 62 0.20
10,402 12,140 -​1.5 51 0.21
10,690 10,402 1.4 53 0.21
27,475 10,690 4.4 2 0.21
51,725 27,475 4.3 84 0.22
30,902 51,725 1.0 108 0.22
27,920 30,902 0.3 51 0.25
13,665 27,920 2.3 -​9 0.26
2,664 13,665 –​5.3 7 0.28
1,399 2,664 –​6.5 18 0.29
538 1,399 –​4.6 25 0.29
4,901 538 –​2.4 40 0.28
3,124 4,901 1.2 –​12 0.28
1,862 3,124 –​0.9 16 0.28
4,386 1,862 –​0.5 53 0.28
10,156 4,386 1.1 70 0.27
11,475 10,156 1.2 55 0.26
10,612 11,475 –​0.2 79 0.26
Selected Excel regression output
Regression statistics
R square 0.70
Adjusted R square 0.65
Standard error 6,467.36
Observations 29
Coeff. Standard error t stat
Intercept 19,566.60 6,634.68 2.9
PRMT(T-​1) 0.46 0.13 3.5
DRRI(T) 1,385.90 413.66 3.4
DEMP(T) 52.55 44.88 1.2
STEMP(T-​1) –58,601.75 25,428.25 –​2.3
Forecasting equation
PRMT(T) = 19,566.60 + 0.46 PRMT(T–​1) + 1,385.90 DRRI(T) + 52.55 DEMP(T) –​58,601.75
STEMP(T–​1)
      (2.9)   (3.5)        (3.4)        (1.2)        (–2.3)
 Applying the econometric approach

...... ...... ...... ......

HISTORY STOCK(30) RRI(30) DRRI(30) PERMITS (30)

Step 1 Step 2 Step 3 Step 4

FORECAST STOCK(31) RRI(31) DRRI(31) PERMITS(31)

Step 5 Step 6 Step 7 Step 8

STOCK(32) RRI(32) DRRI(32) PERMITS(32)

Step 9 Step 10 Step 11 Step 12

STOCK(33) RRI(33) DRRI(33) PERMITS(33)

...... ...... ...... ......

Figure 8.3  A recursive modeling and forecasting system

As indicated in Figure 8.3, the first step in generating the forecast is to calculate the stock for the first fore-
cast period, which is period (31), using (8.28):

STOCK(31) = STOCK(30) (1–​0.005) + 0.95 PRMT(30) = 432,796 0.995 + 0.95 10,612

 STOCK(31) = 430,632 + 10,081 = 440,713 (8.29)

The estimate of the stock derived through (8.29) can be subsequently used along with the forecasts
of employment, income, mortgage rates, and inflation, presented in Table 8.4, to calculate the values of
STEMP, IPE, and USER for the 31st period. These estimates can then be used along with the coeffi-
cient estimates in (8.26) to generate an apartment rent index forecast for the 31st period. This step is
demonstrated using the optimistic forecasts for employment and income, presented in Tables 8.4 and 8.5:

RRI(31) = 38.92 –​0.21 IPE(31) + 0.69 USER(31) –​ 41.28 STEMP(31) + 0.74 RRI(30)

= 38.92 –​0.21 53.45 + 0.69 2.52 –​41.28 (441,553/​1,784,000) + 0.74 72.0

 RRI(31) = 72.5 (8.30)

Finally, the forecast of the market rent for the 31st period can be used to calculate the permits for that
period using equation (8.31) as indicated below:

PRMT(31) = 19,566.60 + 0.46 PRMT(30) + 1,385.90 DRRI(31) + 52.55 DEMP(31)

–​58,601.75 STEMP(30)
= 19,566.60+0.46 10,612 + 1,385.90 0.7 +52.55 88 –​58,601.75 (441,553/​1,696,000)

 PRMT(31) = 15,102 (8.31)

169
 newgenrtpdf
Table 8.4  Steps in generating forecasts of rents, permits, and apartment stock
Developing alternative, optimistic, and pessimistic forecasts
Forecasted values of exogenous variables: Optimistic employment and income forecasts

INC(T) EMP(T) INC EMP MORTG(T) INFL(T)

CHANGE CHANGE
YEAR(T) (Billions) (X1000) (%) (%) (%) (%)
30 89.55 1,696 4.3 4.9 7.8 3.3
31 95.4 1,784 6.5 5.2 7.8 3.1
32 99.4 1,847 4.2 3.5 7.4 2.8
33 102.3 1,876 2.9 1.6 7.2 2.7
34 104.9 1,917 2.6 2.2 7.3 2.9
35 108.2 1,953 3.2 1.9 6.9 2.8

YEAR(T) INC(T) EMP(T) DEMP(T) IPE(T) USER(T) PRMT(T) PRMT(T–​1) STOCK(T) RRI(T) DRRI(T) MORTG(T) INFL(T)
(billions) (×1,000) (×1,000) (×1000) (%) (%) (%) (%)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
1 25.3 554 45.70 1.77 9,644 60,958 100.0 6.6 3.0
2 27.6 596 42 46.30 0.43 21,388 9,644 70,602 97.8 –​2.2 7.1 4.7
. . . . . . . . . . . . .
30 89.5 1,696 79 52.80 2.32 10,612 11,475 432,796 72.0 –​4.6 7.8 3.3
31 95.4 1,784 88 53.45 2.52 Step 4 10,612 Step 1 Step 2 Step 3 7.8 3.1
32 99.4 1,847 62 53.81 2.53 Step 9 Step 5 Step 6 Step 7 Step 8 7.4 2.8
33 102.3 1,876 30 54.50 2.48 Step 10 etc. 7.2 2.7
34 104.9 1,917 41 54.70 2.36 7.3 2.9
35 108.2 1,953 36 55.41 2.17 6.9 2.8

The estimation process

First, complete the worksheet with the forecasted values of the exogenous variables. Then:
Step 1: Given the stock-​flow identity and starting from Year 30’s given values, forecast.Year 31’s apartment stock.
Step 2: Given Year 31’s stock, the forecasts of the relevant exogenous variables, and the estimated parameters of the rent equation, forecast Year 31’s real rent index (RRI).
Step 3: Given Year 30’s and Year 31’s real rent index, calculate its percentage change (DRRI).
Step 4: Given the estimated parameters of the permit equation,Year 30’s permits, DRRI, and the values of other exogenous variables, forecast Year 31’s permits.
Step 5: Complete column 8 with Year 31’s permits.
Step 6: Given the stock-​flow identity, and Year 31’s permits and stock, forecast Year 32’s apartment stock etc.
 newgenrtpdf
Table 8.5  Steps in generating forecasts of rents, permits, and apartment stock
Developing alternative, optimistic and pessimistic forecasts
Forecasted values of exogenous variables: optimistic employment and income forecasts
YEAR(T) INC(T) EMP(T) INC EMP MORTG(T) INFL(T) (%)
(billions) (×1,000) CHANGE CHANGE (%)
(%) (%)
30 89.55 1,696 4.3 4.9 7.8 3.3
31 95.4 1,784 6.5 5.2 7.8 3.1
32 99.4 1,847 4.2 3.5 7.4 2.8
33 102.3 1,876 2.9 1.6 7.2 2.7
34 104.9 1,917 2.6 2.2 7.3 2.9
35 108.2 1,953 3.2 1.9 6.9 2.8

YEAR(T) INC(T) EMP(T) DEMP(T) IPE(T) USER(T) PRMT(T) PRMT(T–​1) STOCK(T) RRI(T) DRRI(T) MORTG(T) INFL(T)
(billions) (×1,000) (×1,000) (×1,000) (%) (%) (%) (%)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
1 25.3 554 45.70 1.77 9,644 60,958 100.0 6.6 3.0
2 27.6 596 42 46.30 0.43 21,388 9,644 70,602 97.8 –​2.2 7.1 4.7
. . . . . . . . . . . . .
30 89.5 1,696 79 52.80 2.32 10,612 11,475 4,32,796 72.0 –​4.6 7.8 3.3
31 95.4 1,784 88 53.45 2.52 15,102 10,612 440,713 72.5 0.7 7.8 3.1
32 99.4 1,847 62 53.81 2.53 16,046 15,102 452,857 72.9 0.5 7.4 2.8
33 102.3 1,876 30 54.50 2.48 14,092 16,046 465,836 72.9 0.0 7.2 2.7
34 104.9 1,917 41 54.70 2.36 13,346 14,092 476,894 72.7 –0.2 7.3 2.9
35 108.2 1,953 36 55.41 2.17 12,212 13,346 487,188 72.3 –0.6 6.9 2.8

The estimation process

First, complete the worksheet with the forecasted values of the exogenous variables. Then:
Step 1: Given the stock-​flow identity and starting from Year 30’s given values, forecast Year 31’s apartment stock.
Step 2: Given Year 31’s stock, the forecasts of the relevant exogenous variables, and the estimated parameters of the rent equation, forecast Year 31’s real rent index (RRI).
Step 3: Given Year 30’s and Year 31’s real rent index, calculate its percentage change (DRRI).
Step 4: Given the estimated parameters of the permit equation,Year 30’s permits, DRRI, and the values of other exogenous variables, forecast Year 31’s permits.
Step 5: Complete column 8 with Year 31’s permits.
Step 6: Given the stock-flow identity, and Year 31’s permits and stock, forecast Year 32’s apartment stock etc.
 Analyzing residential real estate markets

With an estimate of the permits for the 31st period we can estimate the stock of the next forecast
period, which will then allow us to calculate the rent and permits for that period and so on. The applica-
tion of these recursive calculations is illustrated in Table 8.4. Notice that the number of steps in Table 8.4 is
not exactly the same as the number of steps in Figure 8.4 because the former describes the exact steps of
filling up the different cells in the columns in that table going forward. Since the table has separate columns
for PRMT(T) and PRMT(T-​1) it requires two steps to fill the cells in these two columns. In the process
described in Figure 8.4, however, these two steps are considered as one.
Table 8.5 presents the forecast permits, rent, and stock for the Dallas apartment market up to period 35,
assuming an optimistic employment and income forecast. As we can see from this table, under this scenario,
real apartment rents are predicted to register minor increases of 0.7% and 0.5% in the first two years of the
forecast. Based on the historical behavior of the market these increases should motivate homebuilders to
bring forth a higher number of new apartment units during the subsequent three years. As a result of these
developments, real rent growth is expected to slow down to 0% in the third year of the forecast and turn
slightly negative in the fourth and fifth years, registering decreases of 0.2% and 0.6%, respectively. Thus,
under the optimistic scenario, real rents in the Dallas apartment market are expected to remain practically
flat over the five-​year planning horizon.
The forecast of the Dallas apartment market under a pessimistic economic scenario is not drastically
different from the optimistic forecast just described, despite significant differences in the underlying eco-
nomic assumptions (see Table 8.6). Under this scenario, according to which employment and real income
bottom out in the second period of the forecast (year 32), real apartment rents are expected to register
an increase of 0.5% in the first forecast period and small declines in the subsequent four periods. Over
the five-​year horizon, real rents are expected to register a cumulative decrease of only 2%. This is a small
decrease given that the assumed pessimistic scenario was chosen to reflect the market’s worst performance
in terms of employment and income growth during the 30-​year period covered by the historical data.
In sum, this sensitivity analysis suggests that the Dallas apartment market  allows for minimal upside
in the case of an optimistic scenario and a minimal downside in the case of a pessimistic scenario. This
behavior is the result of two idiosyncratic functional characteristics of the Dallas apartment market. The
first is the slow adjustment in real apartment rents, as evidenced by the relatively high coefficient of 0.74
of RRI(T–​1). Because of this slow adjustment process, rents do not rise (fall) considerably when the initial
positive (negative) demand shock occurs. The second functional characteristic is the rather quick response
of completions to changes in market conditions, as evidenced by the relatively small coefficient of the
lagged permits, PRMT(T–​1). Thus, as demand and rents increase (decrease) after the initial demand shock,
completions rise (fall) relatively quickly, easing any upward (downward) pressures on rents.
The differences in real rent growth forecasts under the two alternative economic scenarios can be
easily seen in Figure 8.4. Given that these differences are by no means extreme, although differences in the
underlying economic assumptions are significant, we can conclude that this sensitivity analysis does not
indicate any instability in the model.To fully test, however, the robustness of the model the analyst needs to
examine also the sensitivity of the forecasts to changes in the other independent variables included in the
model, that is, mortgage rates and inflation.
It should be emphasized that in a proper sensitivity analysis, each scenario (optimistic or pessimistic)
should be consistent across all interrelated economic variables that enter the forecasting equations. For example,
in this application we assumed that mortgage rates and inflation remain constant in both the optimistic
and the pessimistic scenario, which is not very likely to be the case in the real world. Thus, when several
macroeconomic variables enter the forecasting equations, as is the case in our example, it may be better
to purchase forecasts for all exogenous variables involved in both scenarios from the same econometric
forecasting firm in order to ensure consistency both within and across scenarios. Besides ensuring consist-
ency, such scenarios will allow for a more complete test of the robustness of the model since changes in
several variables will be introduced simultaneously.

172
 newgenrtpdf
Table 8.6  Steps in generating forecasts of rents, permits, and apartment stock
Developing alternative, optimistic and pessimistic forecasts
Forecasted values of exogenous variables: Pessimistic employment and income forecasts
YEAR(T) INC(T) EMP(T) INC EMP MORTG(T) INFL(T) (%)
(billions) (×1,000) CHANGE CHANGE (%)
(%) (%)
30 89.55 1,696 4.3 4.9 7.8 3.3
31 94.4 1,762 5.4 3.9 7.8 3.1
32 94.6 1,743 0.2 –​1.1 7.4 2.8
33 97.4 1,752 3.0 0.5 7.2 2.7
34 99.3 1,775 1.9 1.3 7.3 2.9
35 100.9 1,808 1.6 1.8 6.9 2.8

YEAR(T) INC(T) EMP(T) DEMP(T) IPE(T) USER(T) PRMT(T) PRMT STOCK(T) RRI(T) DRRI (T) MORTG INFL (T)
(billions) (×1,000) (×1,000) (×1,000) (%) (T–​1) (%) (T) (%) (%)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
1 25.3 554 45.70 1.77 9,644 60,958 100.0 6.6 3.0
2 27.6 596 42 46.30 0.43 21,388 9,644 70,602 97.8 –​2.2 7.1 4.7
30 89.5 1,696 79 52.80 2.32 10,612 11,475 432,796 72.0 –​4.6 7.8 3.3
31 94.4 1,762 66 53.58 2.52 13,663 10,612 440,713 72.4 0.5 7.8 3.1
32 94.6 1,743 -​19 54.26 2.53 9,713 13,663 451,490 72.1 -​0.3 7.4 2.8
33 97.4 1,752 9 55.59 2.48 8,179 9,713 458,460 71.5 -​0.8 7.2 2.7
34 99.3 1,775 23 55.93 2.36 8,082 8,179 463,938 71.0 -​0.8 7.3 2.9
35 100.9 1,808 32 55.82 2.17 8,755 8,082 469,296 70.5 -​0.7 6.9 2.8

The estimation process

First, complete the worksheet with the forecasted values of the exogenous variables. Then:
Step 1: Given the stock-​flow identity and starting from Year 30’s given values, forecast Year 31’s apartment stock.
Step 2: Given Year 31’s stock, the forecasts of the relevant exogenous variables, and the estimated parameters of the rent equation, forecast Year 31’s real rent index (RRI).
Step 3: Given Year 30’s and Year 31’s real rent index, calculate its percentage change (DRRI).
Step 4: Given the estimated parameters of the permit equation,Year 30’s permits, DRRI, and the values of other exogenous variables, forecast Year 31’s permits.
Step 5: Complete column 8 with Year 31’s permits.
Step 6: Given the stock-​flow identity, and Year 31’s permits and stock, forecast Year 32’s apartment stock etc.
 Analyzing residential real estate markets

84

80
Real rent index

76

72

68

64
22 23 24 25 26 27 28 29 30 31 32 33 34 35

Optimistic Pessimistic

Figure 8.4  Optimistic and pessimistic apartment rent forecasts

The second best alternative to purchasing a pessimistic scenario from a vendor is to construct one by
applying to the employment and income forecasts the negative growth rates observed during previous
recessions. These variables exhibited considerably different rates of change during the 1982 and the 1991
recessions. The analyst may want to examine the sensitivity of the forecast to either set of assumptions.
Running a forecast using recessionary assumptions for the explanatory variables, as suggested by histor-
ical data, can be very helpful in gaining a perspective as to what could be a worst-​case scenario for the
apartment market under consideration. But again, every effort needs to be made to maintain consistency
across the predicted values of the different macroeconomic variables entering the forecasting equations in
each scenario.

Chapter summary
In this chapter, we have presented an application of the econometric approach, an analytical technique
clearly more sophisticated and advanced than the accounting approach, in forecasting the Dallas apartment
market. The application of this technique involves in general four basic steps:

1. Collect sufficiently long historical data on demand, supply, rents, and their drivers for the property
type and market under consideration. For example, in the case of the Dallas apartment market, we
used historical information on rents, permits, and stock, as well as historical information on local and
national economic variables, thought to influence demand for and supply of apartments in Dallas.
Such variables include local employment, local income, national mortgage rates, national inflation
rates, and a national construction cost index.
2. Set up and define a behavioral/​structural model for the local property market within the framework
of the stock-​flow model. Typically, the econometric model includes three behavioral equations: one
for demand, one for supply, and one for prices/​rents. In the application presented in this chapter, a
demand equation was not introduced due to the lack of historical data. All important demand drivers,
however, were accounted for in the rent formulation.

174
 Applying the econometric approach

The analyst at this stage needs to determine whether the modeling framework should reflect quick or
slow adjustment processes. Given the several inefficiencies of real estate markets, it is more likely that
a slow adjustment process will better fit the data, but the best course of action would be to estimate
both and evaluate which one better describes the market’s historical behavior. In the case of the Dallas
apartment market, for example, the slow-​adjustment specification proved indisputably more powerful
in explaining historical movements in real apartment rents. A similar specification worked better for
permits too, but the coefficient of the lagged permits, although clearly statistically significant, is smaller
than 0.5, suggesting a rather quick response of supply to market changes. Determining the final speci-
fication of behavioral equations depicting slow adjustment processes usually requires experimentation
with different lags to ensure that the best-​fit formulation is adopted.
3. Estimate the model’s equations using appropriate econometric techniques. Linear regression ana-
lysis is by far the most commonly used econometric technique for estimating the equations of real
estate market forecasting models. Estimation of the equations is necessary to quantify the idiosyncratic
influences of exogenous local and national economic drivers on real estate demand and supply, as
well as the effect of demand and supply on rents and vice-​versa. For example, estimation of the Dallas
apartment market equations helped quantify the effect of income per employee, user cost of capital,
and stock/​employee on market rents, as well as the effect of changes in rents and employment on
permit activity.
4 . Develop forecasts of demand, supply, and rents and perform sensitivity analysis. Forecasts can be
developed in a recursive fashion using the estimated coefficients of the equations and the forecast
values of the exogenous variables. For example, in the application presented, we estimated first the
apartment stock of the first forecast period using the stock and permits of the previous period that
were known. Then we used this stock forecast, along with forecasts of employment, income, and user
cost of capital, to forecast real rents for the first forecast period. Finally, we used this rent forecast, along
with forecasts of employment change, rent change, and stock per employee, to estimate permits for the
first forecast period. These steps were then applied in the same sequence in each period of analysis to
produce five-​year forecasts of permits, stock, and rents for the Dallas apartment market.

In sum, the application of the econometric approach to the Dallas apartment market allowed us to quan-
tify in a scientific way the key behavioral relationships between local/​national economic forces and the
local apartment market, as well as the internal dynamics of the latter. This quantification was made pos-
sible by using widely established econometric techniques to study historical movements in the market’s
key variables over a 30-year period. The estimated behavioral relationships then provided the basis for
forecasting the market’s performance under alternative economic outlooks.
Sensitivity analysis, which is made possible by the econometric approach, can provide useful insights
about the resilience of the market under a pessimistic scenario and the opportunities that may lie ahead in
the case of an optimistic scenario. In the case of the Dallas apartment market, for example, sensitivity ana-
lysis showed that rents would be quite resilient under a pessimistic scenario, but would also provide little
upside in the case of an optimistic scenario.

Questions
1 . Discuss the data requirements for estimating a forecasting model for an apartment market.
2. Discuss the components/​equations of an apartment-​forecasting model assuming that no data on
apartment vacancies and absorption are available and rents adjust to their equilibrium values in more
than one period after an exogenous shock. Be specific in terms of the variables to be included in each
equation and why.
3. Explain why rent adjustments in apartment markets may be slow, despite short rental contracts.

175
 Analyzing residential real estate markets

4. Describe how you would use the estimated regression results and any identities required to forecast
an apartment market. Specifically, describe the steps involved in generating forecasts of apartment
permits, stock, and rents.
5. Explain how sensitivity analysis can help evaluate an apartment market and guide residential develop-
ment decisions.

Project #3: Forecasting the Los Angeles apartment market

This exercise focuses on the market for multi-​family housing units within the Los Angeles PMSA (Primary
Metropolitan Statistical Area). Its purpose is to acquaint the students with the analysis of time-​series data
and their use in forecasting housing markets. The required historical data on apartment rents, stock, and
potential drivers of apartment demand and supply within the Los Angeles metropolis are provided in
Table 8.7. Given this set of data, the students are asked, first, to produce a series of conditional forecasts for
apartment rents, permits, and total number of multi-​family units for the period of 1999–2003 and, second,
to discuss the implications of these forecasts for apartment development or investment. A description of the
historical data and instructions on how to carry out this project follow.

The data
The database covers a period of 32 years—​from 1967 to 1998 and refers to the Los Angeles PMSA, which
is geographically defined as the Los Angeles County. These data are provided in Table  8.7 and briefly
described below:

YEAR Year of observation

RRENT Real rent index (in constant 82–​84 dollars), representing the real rent component of the
consumer price index; note that the basis year for this index is 1967, and that it refers to
a constant-​quality apartment sample (Data source: Bureau of Labor Statistics, Detailed CPI
Reports)
POP Total number of persons
(Data source: Bureau of Economic Analysis)
EMP Total non-​agricultural employment
(Data source: Bureau of Labor Statistics)
IPE Real personal income per worker (in constant 82–​84 dollars)
(Data source: Bureau of Labor Statistics)
PRMT Permits for units in structures with more than two units, authorized during the year of
observation
(Data source: Census Bureau, Construction Reports, Series C-​406)
STOCK: Total apartment stock representing the number of multi-​family units in structures with more
than two units; generated using benchmark figures from the 1960, 1970, 1980, and 1990
decennial censuses along with the stock-​flow identity and the annual permit data
MORTG: Interest rate on conventional fixed-​rate mortgages (%)
(Data source: FHLBB Series)
INFL: Inflation rate (%)
(Data source: Bureau of Labor Statistics)
COST: Inflation-​adjusted construction cost index
(Data source: R.S. Means Square Foot Costs, various years)
INT: Five-​year Treasury constant maturity (%)
(Data source: Federal Reserve Bulletin, various years)

176
 newgenrtpdf
Table 8.7 The Los Angeles apartment market: Relevant local and national historical data

YEAR RRENT(T) POP(T) EMP(T) IPE(T) MORTG(T) INFL(T) CPI(T) STOCK(T) PRMT(T) CCOST(T) INT(T)
1967 100.00 6,903,000 2,741,000 31,208 6.6 3.0 33.9 735,048 12,992 100.00 5.1
1968 97.60 6,946,000 2,835,000 31,055 7.1 4.7 35.5 747,413 19,550 100.04 5.7
1969 95.38 7,003,000 2,929,000 30,725 8.1 6.1 37.7 766,198 28,872 98.48 6.9
1970 95.53 7,053,000 2,807,000 31,995 8.8 5.5 39.8 794,109 38,026 103.15 7.4
1971 96.15 7,117,000 2,814,000 32,089 8.0 3.2 41.1 830,740 33,414 109.44 6.0
1972 94.75 7,067,000 2,915,000 32,254 7.6 3.4 42.5 862,821 41,860 115.07 6.0
1973 89.59 7,053,000 3,061,000 30,496 8.0 8.7 46.2 903,153 33,504 116.94 6.9
1974 83.72 7,095,000 30,63,000 29,859 8.8 12.3 51.9 935,265 14,139 119.14 7.8
1975 82.51 7,138,000 3,063,000 30,237 8.7 6.9 55.5 948,373 8,851 115.80 7.8
1976 84.15 7,224,000 3,139,000 31,313 8.9 4.8 58.2 956,288 15,007 118.12 7.2

Applying the econometric approach

1977 85.95 7,273,000 3,305,000 30,608 8.9 6.7 62.1 970,230 21,357 117.51 7.0
1978 86.04 7,367,000 3,501,000 30,376 9.5 9.0 67.7 990,383 26,376 116.68 8.3
1979 83.39 7,421,000 3,624,000 29,412 9.9 13.2 76.7 1,015,440 23,959 112.50 9.5
1980 83.12 7,528,000 3,613,000 29,113 11.3 12.5 86.3 1,038,107 20,231 110.96 11.5
1981 84.98 7,644,000 3,625,000 30,195 13.3 8.9 94.0 1,056,288 14,445 110.18 14.2
177

1982 89.71 7,809,000 3,491,000 31,494 14.6 3.8 97.6 1,068,955 9,879 117.66 13.0
1983 92.07 7,953,000 3,575,000 31,812 12.2 3.7 101.3 1,077,271 17,141 116.39 10.8
1984 95.07 8,071,000 3,704,000 32,895 11.9 3.9 105.3 1,092,478 23,407 115.01 12.2
1985 99.04 8,232,000 3,788,000 32,955 11.1 3.7 109.3 1,113,622 39,043 114.55 10.1
1986 104.92 8,441,000 3,893,000 33,994 9.8 1.0 110.5 1,149,599 52,691 115.86 7.3
1987 105.90 8,583,000 3,990,000 34,069 9.1 4.4 115.4 1,198,506 37,766 113.82 7.9
1988 106.31 8,699,000 4,067,000 34,444 9.4 4.4 120.5 1,233,185 32,313 111.87 8.5
1989 106.34 8,819,000 4,127,000 34,328 9.8 4.6 126.1 1,262,649 24,765 109.14 8.5
1990 104.70 8,890,000 4,084,000 35,182 9.7 6.1 133.8 1,284,913 16,090 106.11 8.4
1991 104.51 8,958,000 3,945,000 35,510 9.0 3.0 137.9 1,298,914 8,505 105.55 7.4
1992 102.82 9,051,000 3,770,000 37,401 8.0 2.9 141.9 1,305,695 5,038 104.29 6.2
1993 100.26 9,083,000 3,681,000 37,377 7.3 2.7 145.8 1,309,175 2,909 104.49 5.1
1994 97.84 9,065,000 3,715,000 36,985 8.4 2.6 149.7 1,310,630 2,876 103.39 6.7
1995 94.99 9,057,000 3,758,000 37,163 8.0 2.5 153.5 1,312,051 2,930 102.08 6.4
1996 92.88 9,098,000 3,818,000 37,154 7.8 3.3 158.6 13,13,523 3,032 100.58 6.2
1997 92.66 9,160,000 3,911,000 37,238 7.6 1.7 161.3 1,315,089 3,476 99.85 6.2
1998 93.70 9,249,000 4,006,000 37,330 6.9 1.6 163.9 1,317,077 4,803 100.06 5.1
 Analyzing residential real estate markets

Data analysis
Using the variant of the stock-​flow model discussed in this chapter and the historical data series provided
in Table 8.7, forecast apartment rents, apartment permits, and the apartment stock for the period 1999–​
2003 under alternative assumptions. To this end:

1. Estimate the following model equations or variants of them:

First, the Apartment Rent Model: The Apartment Rent Equation, for example, may be specified as:

RRENT(T) = a0 + a1 RRENT(T–​1) + a2 IPE(T) + a3 USER(T) + a4 (STOCK(T)/​EMP(T))

Note that the above model specification assumes that rents may not completely adjust to equate
apartment demand to the existing stock in one period.The speed of rental adjustment can be estimated
as (1–​a1); note that a1 is the regression coefficient of RRENT(T–​1). Also note that the real user cost
of capital (USER), can be computed using MORTG, INFL, and a marginal income tax rate of 28%
as: USER = [(1–​marginal tax rate) × MORTG] –​  INFL.
Second, the Permit Model. Although one will be suggested in class by the instructor, the students
should make an attempt to develop their own permit equation using the data provided. In specifying
the permit equation, the students should have in mind the following:
a) Permits may not adjust quickly towards the desired apartment stock, suggesting that lagged values
of the dependent variable be included in the set of explanatory variables.
b) Permits are influenced by rents/​prices, cost factors, and expectations. Be sure to explain the logic
underlying each and every explanatory variable incorporated in the model and discuss the direc-
tion of expected effects.
c) Due to the volatility of permits in Los Angeles, an R-​squared of 0.65 or greater is satisfactory.
2. Produce a base forecast using the following:
a) the estimated rent and permit equations (see (1));
b) the suggested stock-​flow identity (assuming an annual depreciation rate of 0.5% and a completion
rate of 90%)

STOCK(T) = STOCK(T–​1) * (1 –​0.005) + 0.90 * PRMT(T–​1); and

c) the base values of the exogenous variables for 1999–2003 provided by the instructor.The students
are welcome to replace those by their own series if they wish to do so.
Note that the stock-​flow identity provided above does not apply to the historical stock series
presented in Table 8.7. That series has been constructed using different depreciation rates for
each decade, as implied by the differences in stock levels reported by the decennial censuses
of 1960, 1970, 1980, and 1990, and the sum of yearly permits reported between those census
years.
3. Conduct sensitivity analysis, assuming alternative (optimistic and pessimistic) values for at least two of
the exogenous variables of the model. Explain the impact of these changes on the forecasts.

Discuss the analysis in five to seven pages. The report should be structured as follows:

Introduction: Discuss briefly the objectives of the project and the structure of the report.
Section 1: The model and estimated equations: Explain the logic of the model, and present, describe, and
discuss the estimated rent and permit equations.

178
 Applying the econometric approach

Section 2: The base and alternative forecasts: Present, describe, and explain the base and alternative forecasts.
Discuss the implications of these forecasts for apartment development and/​or investment in the
Los Angeles metropolitan area.
Conclusion: Offer concluding remarks, and explain what you have learned from this project.

Notes
1 If the owner-​occupied market were to be modeled, the analyst would need historical information on house prices.
The Federal Housing Finance Agency (FHFA) generates a monthly and quarterly transaction-​based constant-​quality
single-​family house price index (starting from January 1991) for the 100 largest metropolitan areas in the United
States, which provides a sufficiently long time-​series for developing econometric models for the housing markets
of these areas. The link to the FHFA webpage where this information is provided is the following: www.fhfa.gov/​
DataTools/​Downloads/​Pages/​House-​Price-​Index.aspx
2 Behavioral equations describe behavioral relationships that are estimated statistically while identities represent exact
mathematical relationships, not statistical equations.
3 Rosen and Smith (1983) present evidence supporting slow rental adjustment processes in the apartment market.
4 This setup is the same as the one used by DiPasquale and Wheaton (1996) for specifying the demand equation for
the owner-​occupied housing market (Chapter 10, page 259).
5 The explanatory variables included in the construction equation developed by DiPasquale and Wheaton (1996) for
Boston’s single-​family housing market includes house prices and lagged housing stock. The rationale for including
the latter is that as the area’s housing stock expands, land prices rise, reducing profits in the local homebuilding
industry and discouraging further housing development.
6 This report series has been discontinued, but the Census Bureau still provides data on the number of authorized
housing units for the largest metropolitan areas in the country via the Building Permits Survey with links to the data
at this webpage: www.census.gov/​construction/​bps.​

References and additional readings

Brooks, C. and S. Tsolacos. 2010. Real Estate Modelling and Forecasting. Cambridge, UK: Cambridge University Press.
DiPasquale, D. and W. Wheaton. 1996. The Cyclical Behavior of Metropolitan Housing Markets. Chap. 10 in Urban
Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
Rosen, K. and L. Smith. 1983. The Price Adjustment Process for Rental Housing and the Natural Vacancy Rate.
American Economic Review, 73, 779–​786.

Appendix 8A: Calculating the speed of rental adjustment

Let R(t) be the prevailing market rent, R* the market-​clearing rent and gR the speed of rental adjustment. The rent
adjustment equation can then be written as:

R(t ) − R(t − 1) = gR [ R * − R(t − 1)] (8A.1)

⇒ R(t ) = gR R * − gR R(t − 1) + R(t − 1) = gR R * +(1 − gR )R(t − 1) (8A.2)

Since R* is the market-​clearing rent implied by each period’s demand variables (XD) and stock (S), it can be written
in statistical form as in (8A.3):

R * = b0b1X 1D (t ) +  + bn X nD (t ) + bn +1S(t ) (8A.3)

Substituting (8A.3) into (8A.2) then we have:

R(t ) = gR [b0 + b1X 1D (t ) + ..... + bn X nD (t ) + bn +1S(t )] + (1 − gR )R(t − 1) (8A.4)

179
 Analyzing residential real estate markets

The above equation is equivalent to the statistical equation:

R(t ) = a0 + a 1X 1D (t ) +  + an X nD (t ) + an +1S(t ) + an + 2R(t − 1) (8A.5)

where

a0=gRb0, a1=gRb1, ......,an=gRbn, an+1=gRbn+1 and an+2=(1–​gR)

Therefore, from the estimated coefficients of the statistical model (8A.5) we can calculate the speed of rental adjust-
ment as gR = 1–​an+2.

180
newgenprepdf

9
ANALYZING RESIDENTIAL
PROJECTS
A micro perspective

Introduction 182
Qualitative analysis of residential projects 184
Site and location analysis 184
The site and its immediate environment 184
Neighborhood and community attributes 186
Access and linkages 187
Analysis of local market conditions 188
Data sources 188
Consumer targeting and refinement of product design 189
Quantitative analysis of residential projects: estimating project rent and absorption rate 190
Analyzing the competition: the competitive differentials technique 191
Analysis steps 191
Numerical example 192
Critique 198
Hedonic valuation techniques 199
Analysis steps 199
Applications 203
Estimating project price or rent given its amenity package 203
Determining and evaluating optimal project design 204
Numerical example 205
Critique of the hedonic valuation technique 207
Project absorption and gross revenue analysis 207
Numerical example of project absorption analysis 210
Econometric estimation of project absorption 213
Chapter summary 215
Questions 215
Project #4: On hedonic price models 216
Analysis tasks/questions 216
References and additional readings 218
Appendix 9A: Determining optimal development density 218

181
 Analyzing residential real estate markets

Introduction
In the previous three chapters, we focused on the macroeconomic analysis of residential real estate markets
and discussed techniques for assessing the strength of the broader housing market within which the pro-
ject under consideration is located. In particular, we discussed accounting techniques for assessing housing
demand-​supply gaps and the magnitude of effective market demand for specific housing types. We also
discussed how the econometric approach can be used to quantify in a more scientific way the effect of eco-
nomic, financial, and demographic forces on housing demand and supply, as well as the internal dynamics
of housing markets. Finally, we demonstrated how this approach can be utilized for generating forecasts of
future housing market demand, completions, and movements in residential prices/​rents.
While macroeconomic analysis must be an integral part of residential market analysis, it is not adequate
for making the final decision of going ahead with or abandoning a specific project. The investor or the
developer still needs to evaluate the merits and constraints of the specific site and residential development
considered and decide in a very definite manner what exactly to build and for whom. Within this context,
the purpose of the microeconomic/​marketability analysis is to refine the original development scenario,
assess market strength at the project and site-​specific level, and evaluate the income-​earning prospects of the
final product design, taking into account both its idiosyncratic attributes and the broader market outlook.
What questions should the microeconomic analysis address? As Figure 9.1 indicates, the major questions
that need to be addressed are:

1. What type of residential development and product design will be most marketable and profitable at
the given site?
2. What are the project’s competitive strengths and most likely achievable rents/​prices and absorption  rate?

In order to address these questions two broader types of analysis need to be carried out:

1. Qualitative analysis can help address some aspects of the first question relating to the site and its loca-
tion. This stage includes in particular:
• Analysis of the site and its environment
• Analysis of local market conditions
• Refinement of product design and consumer targeting.
2. Quantitative analysis can help address aspects of both the first and the second question, such as defining
a profitable amenity mix and assessing the project’s earning capacity at the time of completion and
beyond.The ultimate goal of quantitative analysis is the development of the project’s expected rev-
enue schedule. The latter is the most critical factor in evaluating the project’s financial feasibility and
assessing time-​of-​entry and/​or project-​phasing strategies. Within this context, quantitative analysis
focuses on: a) project price/​rent analysis, and b) project absorption analysis.

The focus of project price/​rent analysis is to assess as accurately as possible the market price or rent that a
specific residential development and product design will command in the local marketplace. Depending
on data availability, this analysis can be carried out using competitive differential techniques or hedonic
valuation techniques.The latter are clearly more sound than the former because they use widely established
econometric techniques. Furthermore, hedonic valuation techniques enable the analyst to determine the
most profitable development plan and product design by estimating the market rent/​price of alternative
residential amenity packages and development densities. Developing such alternative estimates through
competitive differential techniques is much more cumbersome and less accurate.
The focus of project absorption analysis is, similarly, the assessment of the share of market absorption that a
residential development is expected to capture given its specific design features and location attributes. Again,

182
 Analyzing residential projects

B. Re-define/refine the development scenario

and assess market strength at the site
and project levels
(What to build and for whom)

Q1. What type of development and product design will be most

marketable at the given site?

Q2. What is the competitive strength, income earning capacity,

and achievable absorption rate of such a development?

QUALITATIVE ANALYSIS (Q1)

Analysis of the site and its environment

Analysis of local market conditions

Refinement of product design and consumer targeting

QUANTITATIVE ANALYSIS (Q1 & Q2)

Project price/rent analysis

Focuses on the estimation of project price/rent through
competitive differential or hedonic price analyses and the
determination of profitable amenity packages (amenity mix
and development density)

Project absorption analysis

Focuses on the estimation of project absorption schedule
and the evaluation of time-of-entry and phasing alternatives

Figure 9.1  Stages of microeconomic analysis for residential projects

183
 Analyzing residential real estate markets

depending on data availability, simple accounting or econometric techniques can be used in estimating this share.
Once this share is estimated, it can provide the basis for developing project absorption and revenue schedules,
which in turn can provide valuable guidance in evaluating time-​of-​entry and project phasing alternatives.

Qualitative analysis of residential projects

Qualitative analysis of residential projects focuses on the opportunities and constraints the site and its
market present, and it should be carried out with the bottom-​line objectives in mind:

• Identification of target household groups with preferences that are compatible with site characteristics.
• Development of a residential product design that is most compatible with site characteristics and target
group preferences.

The ultimate goal is to determine the type of development and product design that will be most marketable
at the given site. To this aim, the analysis focuses 1) on site and location attributes, and 2) on local market
conditions and activity.The focus is on information that can shed light on the site’s advantages and constraints,
the geographic area within which the project will be operating and competing for customers, the consumer
groups most likely to be attracted to the site, and project and housing-​unit design features most likely to be
demanded by such groups. For example, if the site does not have convenient access to a good school, then the
developer may need to target young households without children, non-​family households, or empty-​nesters.
The conclusions derived from the qualitative analysis in terms of optimal product design features and their
contribution to the project’s market price or rent will be tested and refined through quantitative analysis.

Site and location analysis

The analysis of spatial markets includes analysis of the site and its immediate environment, as well as ana-
lysis of local market conditions. The rationale for this analysis is that that there are price differentials across
submarkets within an urban housing market attributable not only to differences in neighborhood quality
and other location advantages/​disadvantages but also to differences in locational demand for housing and
the mismatch between the available supply and neighborhood-​specific demand (Watkins, 2008). Jones et al.
(2003b) present evidence from the UK that supports the existence of spatial housing submarkets and their
stability over time.Within this context, site and location analysis, which is often referred to as “productivity
analysis” (Parli, 2008), needs to examine the following three categories of factors:

1 . Attributes of the site and its immediate environment

2. Neighborhood and community attributes
3. Access and linkages.

The specific factors that need to be examined under each category are listed in Figure 9.2 and briefly
discussed below.

The site and its immediate environment

The analyst needs to look first at the characteristics of the site, including physical characteristics, such
as size, shape, topography, geology, drainage, environmental conditions, and any other physical feature
that may present opportunities or constraints for residential development; the availability and capacity of
public utility lines; and legal restrictions, including the site’s zoning classification, other land-​use regulations
constraining its development, and private restrictions, such as deed restrictions, easements, etc.

184
 Analyzing residential projects

Analyze the attributes Analyze neighborhood Analyze access and

of the site and its and community linkages
immediate environment attributes

Physical characteristics Image Access to

• Size, shape • Neighborhood quality Employment centers
• Topography, geology • Crime rates • Commutable distances
Availability of utilities • Access to major streets
Legal limitations on use Development stage • Public transportation
• Zoning classification
• Private restrictions Demographic structure Access to higher-end
• Movers’ demographics services
Site environment
• Positive external Land use structure Access to shopping centers,
amenity effects • Predominant uses entertainment, cultural
• Negative external • Conformity activities
disamenity effects • Emerging trends

Local public goods and

Fiscal environment
• Local government finance
• Local government services
• The school system

Local government regulations

• Growth controls
• Minimum lot requirements
• FAR regulations
• Building codes
• Rent controls
• Environmental regulations
Consumer
targeting
ANALYSIS OF LOCAL REAL ESTATE MARKETS implications

Product
Consumer information Product information Local market conditions design
implications

Who is buying? What are consumers buying? General price appreciation

• Demographics • What is the product’s trends
• First vs second-time price range? Absorption rates?
buyers • What are the product’s Vacancy rates?
• Where do they design features and
come from? amenity levels?
Who is renting?

Figure 9.2  Site and location analysis

185
 Analyzing residential real estate markets

Second, the site’s immediate environment needs to be examined in order to identify positive and negative
external effects. These include effects external to the site, usually emanating from adjoining uses that may
negatively or positively affect the desirability, income-​earning capacity, and market value of the develop-
ment at the site. Examples of positive external effects include ocean or lake view, or frontage on a park or
open space. Studies have shown that the effects of such external factors may be substantial. Often-​cited
examples of negative external effects include effects from nearby traffic and noise, as well as effects from
nearby disruptive uses, such as industrial properties, run-​down houses or commercial developments. Studies
have shown that such negative external effects may be very localized. The identification of such effects is
very important because the location of the site is fixed and if serious negative externality effects are present,
there is very little that the developer can do to cure them in most cases.
It is important at this point to distinguish between “curable” and “incurable” negative external effects.
“Curable” effects are defined as those that can be mitigated at a marginal cost (MC) that is lower than the
marginal benefit (MB) their cure will bring about in terms of project value and revenues. The opposite
holds true for “incurable” effects. Although “curing” a negative externality is in most cases an infeasible task
for a variety of financial and non-​financial reasons, sometimes buying and re-​developing the neighboring
use, in addition to the target site, may prove to be profitable. The underlying principle of such an endeavor
is that the income from the improvements in the neighboring use combined with the enhancement
in value and earning ability of the target site will exceed the costs of improvements on both sites. It is
emphasized that very careful analysis is required to accurately assess the cost of “curing” the negative exter-
nality, or the cost of redevelopment, and the expected increase in value due to the planned improvements.
In principle, redevelopment will only make sense if the increased residual land value (marginal benefit) is
greater than the cost of redevelopment (marginal cost).

Neighborhood and community attributes

The second set of factors that need to be analyzed within the context of site and location analysis includes
neighborhood and community attributes. Examining such attributes is very important because many
empirical studies have shown that positive attributes are capitalized into higher rents and property values,
while negative attributes are capitalized into lower rents and property values. For example, Anderson and
West (2006) validated the effect of proximity to open space on housing values. In particular, their findings
suggest that this effect is higher in neighborhoods that have higher density, are near to the central business
district, have high income or a high crime rate. Since the investor/​developer has little control over these
attributes, they need to be analyzed and evaluated very carefully before any decision to go ahead with the
project is made.
The value of a residential project may be affected by several neighborhood and community attributes
that may influence locational demand and/​or local housing supply. These attributes can be classified under
the following broader categories:

• Image-​related attributes, such as overall neighborhood quality (high-​ end, middle-​

end, low-​end) and
crime rates.
• Development stage, referring to whether the neighborhood is a developing one, mature, or declining.
• Demographic structure, referring to the demographics of existing residents and people moving in the
neighborhood (young singles, single professionals, young couples with no children, full-​nesters, empty
nesters, retirees, etc.) and their income characteristics.
• Land-​use composition, referring to the land-​use mix of the site’s neighborhood, focusing on the types
of predominant uses, including natural features and open spaces, compatibility of uses, and emerging
trends, if any, that may be contributing to changes in the community’s land-​use mix.The land uses occu-
pying adjacent sites are very important in determining the broader type of the residential development

186
 Analyzing residential projects

most suitable for the site under consideration. For example, if all adjacent sites are developed with
affordable housing then the developer needs to think twice before developing the site with a residential
product that is quite different from affordable housing.
• Local public goods and fiscal environment, referring to the quality of public services, such as police and
fire protection, the public school system, etc., offered by the local government, as well as the finances
and ability of the latter to continue providing such services at an adequate, and competitive, level. The
quality and health of the community’s school system is of primary importance for families with school-​
age children.
• Local government regulations, such as growth controls, minimum lot size requirements, and FAR regulations,
are of primary importance in determining housing supply conditions and prospects within the commu-
nity within which the site is located.1 Such regulations may have a negative effect on land values if they
prevent the development of a site at its optimal density. If they are restrictive enough, however, to the
extent that they create supply shortages, they can have a positive effect on the value of existing housing
units. Jones et al. (2003b) present evidence of differential price trends in supply constrained housing
submarkets in the Glasgow urban area.

Some other local government regulations that can affect the project include building codes, rent controls,
and environmental regulations. Building codes refer to regulations that apply to the improvements on the
site and they set specification standards designating which construction materials can be used, or perform-
ance standards relating to fire protection and, overall, the safe and proper operation of the building (Dasso
and Ring, 1989; Peiser and Hamilton, 2012). These codes may differ across jurisdictions within the same
metropolitan area. Obviously, more demanding building codes increase the cost of residential projects.
Environmental regulations refer to a host of local, state, and federal laws aiming at preventing or
limiting adverse environmental impacts. Such regulations are especially relevant in the case of residential
developments in areas with a sensitive environment.
Finally, rent controls refer to government restrictions on the rent that can be charged for apartment
units. Such controls are of critical importance, if present in the community within which the site is located,
because they may limit the income-​earning ability of apartment developments to the point that they are
not financially feasible.

Access and linkages

The third component of site and location analysis for residential projects focuses on access and linkages.
Convenient access to services and destinations that households need to reach in satisfying their daily needs
provides competitive advantages to a residential site. In particular:

• Access to the area’s major employment centers is of critical importance since every typical household desires
to minimize the money and time cost, as well as the inconvenience, of daily commuting to work. Thus,
access to freeways, major streets, and public transportation stations, as well as commutable distances from
the particular site should be carefully examined.The negative impact of both the time and cost of travel
to work on residential location choice has been verified by several empirical studies (Kim et al., 2005). It
is also very important to look at the employment composition of the employment centers that are most
accessible from the site. Such information may provide important clues regarding the characteristics of
potential buyers or renters that can be drawn to the site.
• Access to adequate and high quality public services, such as fire and police protection, and good public
schools, should be valued by consumers when evaluating residential locations in different jurisdictions.
For example, Kim et al. (2005) verified that school quality has a statistically significant effect on residen-
tial values.

187
 Analyzing residential real estate markets

• Access to other services, such as restaurants and shopping, as well as cultural and entertainment facilities, is
also relevant and should be part of the microeconomic analysis of a particular site/​project since trips to
such destinations are part of a typical household’s life.

Analysis of local market conditions

Besides analyzing the attributes of the site and its location, the analyst should also examine local market
activity and conditions, because they could have important implications for product design. Analysis of the
local market focuses on:

• Consumer information
• Product information
• Local market conditions.

In collecting consumer information, the analyst needs to focus on the demographic profile of house buyers
or renters in the project’s market area,2 as well as whether they are first-​time or second-​time homebuyers
and where they come from. In gathering product information, the analyst needs to examine what spe-
cific types of houses or apartments the newcomers are buying or renting and what prices or rents they are
paying. Finally, in collecting local market information, the analyst needs to examine sales activity, current
price and rent levels, general price appreciation trends and/​or movements in apartment rents, the rate at
which houses available for sale or apartment units are absorbed, and vacancy rate levels and trends.
The median time on the market, expressed typically in months, is often used by real estate professionals to
gauge the pace at which housing units available for sale or rent are absorbed. Information on the median
time on the market for single-​family units at the national level can be obtained from research statistics
published by the National Association of Realtors. Obviously, a longer time on the market would reflect a
lower absorption rate. If accurate, data on the time on the market can help determine project capture rate in
the short term. For example, if the median time on the market is 4.4 months it would suggest that 50% of
the houses sold in that specific month were in the market for more than 4.4 months, while the remaining
50% were on the market for less than 4.4 months.There might be problems, however, in using this number
for assessing the absorption schedule of a project that will be completed 12–​24 months from the time of
analysis. What if the market conditions change during that time period? What about seasonal variations in
demand? These questions are very relevant since the economic forces underlying residential demand and
supply levels can certainly change considerably within a one-​or two-​year period.
Month’s supply, measured as the ratio of vacant houses over the number of houses sold last month, is
another indicator often used by the industry to assess market tightness. For example, if 375 houses are
vacant and available for sale in the market today and 54 houses were sold during the previous month,
then the implied “month’s supply” from these numbers is 375/​54=6.94. This number indicates how many
months it would take to absorb the existing 375 vacant units if the same number of units absorbed during
the last month were absorbed in each and every subsequent month. Such an indicator is usually calculated
using data from multiple listings or brokers who handle a large volume of sales and it can provide some
clues regarding short-​term market trends. For example, if “month’s supply” has been rising it would imply
that market conditions are deteriorating and that absorption rates going forward may need to be adjusted
downwards.

Data sources
Data for the site and its environment can be collected by inspecting the site and public records. Data on
community attributes can be obtained from local government agencies, such as local planning departments

188
 Analyzing residential projects

and economic development agencies. Information on demographics and other neighborhood and com-
munity attributes may be obtained from updated census data. Information relating to access and linkages
may also be obtained from local government, planning, and economic development offices, as well as from
economic censuses. County Business Patterns, provides employment information by SIC code for every
county in the country, but it is published with a lag of two years. The latest publication can be found
at this webpage: www.census.gov/​programs-​surveys/​cbp.html. Industrial directories may also help better
understand the composition of the area’s employment base. Finally, information on local housing market
conditions can be obtained by interviewing local brokers, by examining data from multiple listing services,
or purchasing data from housing research firms.

Consumer targeting and refinement of product design

The ultimate purpose of this stage is to refine consumer targeting and product design in a way that will
maximize the match between project attributes and consumer preferences. This, in turn, will provide the
basis for developing a sensible product design that will be tested and refined through the quantitative
techniques described in the following section. To this end, the analyst needs to synthesize the insights
and conclusions from the analyses described above and assess their implications with respect to project
attributes. For example, analysis of the site and its location, as well as local market activity, can help further
refine the original housing development idea and product design in the following ways:

• Analysis of the site and its environment will help better understand the types of uses that are feasible and
appropriate for the development and provide some initial clues as to the broader types of consumers
that may be attracted to the project.
• Analysis of neighborhood and community attributes will help better define the demographic and income
characteristics of the households that are more likely to value the site’s features and location. For example,
Smith and Olaru (2013) provide evidence that older families, older single occupants, and pensioner
couples place more value on locations with good access to urban amenities.
• Analysis of access and linkages to employment centers will provide additional information regarding
the demographics and income levels of employees that are likely to consider the given site in their search for
housing at a reasonable distance from their workplace.The growth prospects of such job concentrations
should be thoroughly assessed because incremental demand matters more than turnover.
• Analysis of local market conditions and trends will help define in more specific terms the consumer groups
that are more likely to be attracted to the site and the product types and design features that are more
likely to be sought by these groups.

Consumer surveys can help further refine the definition of target consumer groups. Furthermore, they can
help more accurately determine their preferences in terms of housing attributes, and how much they are
willing to pay for such attributes. Interviews with local brokerage offices, as to the specific design features
and project attributes sought by active buyers belonging to the targeted groups, can provide additional
information with respect to the most marketable product attributes.
A project evaluation matrix, such as the one demonstrated in Table 9.1, provides a means for systematic-
ally tabulating the information on potential consumer groups and their preferences. Such a matrix can help
better evaluate and understand what household types may be a better match for the project’s attributes.The
rows of this matrix refer to the different attributes of the project and the housing units and the columns
refer to the different consumer/​household types. The household types that appear to value most of the
project’s attributes are those most likely to be interested in buying or renting units within the development
under consideration.

189
 Analyzing residential real estate markets

Table 9.1  Project evaluation matrix

Project attributes Household types

Type 1 Type 2 ........... Type n
Attribute 1 X .... X
Attribute 2 X .... X
Attribute 3 X X .... X
Attribute 4 .... .... .... ....
................. .... .... .... ....
................. ....
Attribute k X ....

Quantitative analysis of residential projects: estimating

project rent and absorption rate
Upon entering the market, a project’s revenue will depend on the market price it will command and the
absorption rate it will achieve. Both of these factors will be determined by its specific characteristics, its
competitive position, and the market conditions that will prevail upon its completion. For this reason,
detailed analysis of the competition is one of the major tasks of quantitative analysis. A major objective
in analyzing the competition is to understand what product types/​design features sell best and at what
prices, as a means of optimizing project design and assessing the development’s prospects in the local
marketplace. The ultimate task of such an analysis is to quantify what the different project attributes
mean in terms of its pricing and absorption schedule relative to competing properties and the average
performance in the local marketplace. In general, two types of techniques are available to the analyst in
accomplishing this task beyond simple tabulations: 1) competitive differentials techniques and 2) hedonic
valuation techniques.
The competitive differentials methodology, which is outlined and explained in detail in Clapp (1987),
has smaller data requirements and uses non-​econometric techniques to derive a competitive position index
for the project.The latter can be used as the basis for estimating the project’s price/​rent and absorption rate
from the prices/​rents and absorption rates achieved by competing properties in the local marketplace. On
the contrary, hedonic price analysis uses widely established econometric techniques to estimate directly the
market price of each attribute of a residential project. This allows the estimation of the market price of the
bundle of attributes that describe the housing project under consideration. Hence, the considerably greater
data requirements for applying this methodology.

Red flag: The competitive differentials and hedonic valuation techniques discussed below produce rent/​price
estimates strictly and only for the time period during which the analysis is carried out. Unless the project is anticipated
to enter the market soon after the analysis is completed, estimates derived using these techniques need to
be carried forward in a way that reflects any changes in market conditions expected to take place during the period that
intervenes between the time of analysis and project market entry. The most appropriate methodology to project for-
ward the rent/​price estimates produced by competitive differential techniques or hedonic methodologies is to
apply market rent growth rates estimated econometrically. If econometric forecasts of market rent movements
are not available, supply-​demand equilibrium analysis needs to be carried out to provide the basis for assessing
how market rents/​prices may change by the time the project is completed, as well as during the lease-​up
period.

190
 Analyzing residential projects

Analyzing the competition: the competitive differentials technique

The whole discussion on the competitive differentials technique presented in this and the following
sections, including the numerical example presented below, are based on Clapp (1987). The competitive
differentials technique represents, in essence, a detailed analysis of all competing projects (existing, under
construction, or planned) within the development’s primary market area. Thus, this stage requires consid-
erable time and expense on the part of the analyst for the collection of primary data. If the primary market
area includes a huge number of competing properties, to the extent that is beyond the study’s time and
budget constraints, competitive analysis may focus on a smaller geographic area, typically referred to as the
project’s submarket.
It is important to emphasize that the competitive differentials technique has two distinctive goals:

1. To estimate the project’s competitive position index (CPI) in order to use it as a basis for calculating the
development’s absorption schedule at the anticipated time of its completion and during the lease-​up period.
2. To estimate the price/​rent the project would command given its characteristics and market prices at
the time of analysis.

These two distinctive objectives require the consideration of different sets of potentially competing prop-
erties. The first objective requires consideration of all potentially competing properties, including existing,
and those that are under construction or are in the planning stage, within the area that will be used as a
framework of reference for estimating the project’s capture rate.3 The second objective, however, needs to
focus only on competing properties for which market price or rent information is available at the time
of analysis. Therefore, these properties include only competing residential projects that have been already
completed and are operating in the local marketplace, or housing developments that realized sales of some
units before completion (off-​plan sales).
The premise of the competitive differentials technique is that a project’s advantages and disadvantages
relative to competing properties will determine how much better it will do in terms of both its market
absorption share and its price/​rent.Thus, if, for example, a housing project is assessed to be 10% better than
a competing residential development in terms of its amenity package, then it should command 10% higher
price or rent. The competitive differentials technique compares the attributes of the subject property to
those of competing properties and assigns scores to each of the attributes examined. These scores provide
the basis for constructing an overall measure of competitive strength of the subject property, which can be
used to adjust the project’s anticipated fair share of market absorption. These scores can also provide the
basis for constructing relative amenity indices that indicate how much better or worse the subject property
is from each competitive property included in the analysis. The latter is then used as an indicator of how
much lower or higher the project’s value or rent should be relative to the market prices or rents of the
comparable properties for which such information is available.

Analysis steps
The broader and more specific steps in estimating a housing project’s most likely market price or rent at the
time of analysis using the competitive differentials technique can be summarized as follows:

1. Identify competing residential developments among existing completed projects, projects under construc-
tion, projects that have been granted a permit, and planned projects that seem very likely to be completed
before or around the time the residential project under consideration is expected to be completed.
2. Collect and tabulate information on the attributes of the identified competing residential developments
along with the information on the housing project under consideration.

191
 Analyzing residential real estate markets

3. Evaluate competitive differentials across tabulated properties through amenity index techniques:
a) Convert information on each attribute into a score/​index that reflects the relative differences
across properties in terms of the desirability of that attribute.
b) Estimate total amenity indices for each project examined (including the subject property) by cal-
culating the weighted or unweighted sum of the scores of all its attributes.
c) Develop relative amenity indices for the residential project under consideration by dividing its
total amenity index by the total amenity index of each comparable.
4. Compute competitive market prices (or rents) for the project under consideration
a) Use the estimated relative amenity indices and the known market prices or rents of the respective
comparable properties to estimate alternative prices or rents for the subject property.
b) Estimate the price or rent the subject would most likely command as the weighted or unweighted
average of the estimates from step 4(a).

Numerical example
In demonstrating the different steps of the competitive differentials technique, we focus on a numerical
example based on Clapp (1987) in which the attributes of a subject property are compared to the attributes
of competing properties in order to estimate its price.

IDENTIFY COMPETING RESIDENTIAL PROJECTS  The first step in applying the competitive differential
technique is to carefully identify all competing residential developments within the project’s market
area or primary market, depending on budget constraints. It is important to identify not only existing
projects, but also projects under construction, projects that have received a permit but have not broken
ground yet, and planned projects that are very likely to be completed within the project’s planning
horizon. Such a task can be accomplished through field surveys, inspection of public records at local
government offices and planning departments, and interviews with local brokers, residential developers,
and planning officials. Such data may also be available from vendors as “pipeline data,” providing
information for all projects that are at different stages of development, from planning and design up to
the construction stage. However, it is unlikely that these data would provide detailed information about
the characteristics of the project and the units that will need to be collected for the application of the
competitive differentials technique.

COLLECT AND TABULATE INFORMATION ON THE ATTRIBUTES OF COMPETING PROJECTS  At this stage,
the analyst needs to focus not only on the site and location characteristics of the planned project but also
on its structural, functional, and qualitative features, as well as the amenities offered both within each unit
and within the development as a whole. Once the development’s, structure, site, and location attributes have
been tabulated carefully, the analyst needs to obtain similar information for competing (existing, under
construction, and planned) developments within the project’s competitive area. Such information can be
compiled through windshield surveys, by examining data from multiple listing services, by interviewing
local brokers, developers, and appraisers, and by reviewing environmental impact studies (EIS) for projects
not yet completed.
In applying the competitive differentials technique, the information on both the subject and the
comparables examined can be tabulated as in Table 9.2, which presents an example of specific data that
may be collected for such an analysis. For illustration purposes, this table presents information for only two
comparable developments. In an actual study, however, it should definitely include several more compar-
able properties.

192
Table 9.2  Step 1: Collect information on the attributes of competing projects

Attributes Subject Comps

Project Project 1 Project 2

Site and environment

Nearby traffic1 15 22 20
Nearby use2 110 100 105
Frontage2 120 100 105
Noise2 105 95 100
Pollution Similar for all properties
Neighborhood/​community
School system
Expenditure/​pupil 4000 3,700 3,500
Crime rates3 6 7 4
Property tax rates 0.025 0.025 0.02
Mass transit lines
Within walking distance 3 3 2
Neighborhood prestige/​incomes 37,500 37,500 35,000
Police services 2.3 2.1 3
Police officers/​thousand
Fire protection 0.7 0.6 0.8
Firemen/​thousand
Total expenditures/​person 1,300 1,000 1,500
Capital expenditures/​person 350 300 400
Access/​linkages:
Distances to:
School 1.25 5 3.25
Employment 10 6 8
Shopping4 7 15 12
Recreation Similar for all properties
Project amenities
Density
Units/​acre 4 5 3
Open space (acres) 2 1 1.5
Swimming pool Similar for all properties
Tennis court Similar for all properties
Health/​fitness Similar for all properties
Landscaping 105 105 110
Security 105 107 100
Parking space/​unit Similar for all properties
Structural unit amenities
Architecture 105 100 100
Age (months)5 100 99 98
Size (sq. ft.) 13,200 12,700 14,000
Bedrooms Similar for all properties
Kitchen (sq. ft.) 64 60 90
Bathrooms 1.5 2 1.5
Patios Similar for all properties
Fireplaces 2 2 1
Prices ? $135,550 $149,500

Notes:
1
Vehicles/​hour during peak hours.
2
100 indicates no effect on property values. An index higher (lower) than 100 indicates positive (negative) externality
impacts.
3
Violent crimes per thousand population.
4
Distance to closest regional mall.
5
100 indicates a new building. Assuming a 2% annual depreciation rate, the index declines by one point for every six
months of age.
Numerical example republished from Clapp (1987)
 Analyzing residential real estate markets

Several issues need to be noted regarding the data presented in Table 9.2. First, as indicated at the bottom
of this table, although the prices of competing projects are known, the price of the subject property is
in question and needs to be estimated. Second, notice that, besides site and location attributes, the table
includes several features/amenities of the property that is being evaluated labeled as “project amenities”
and “structural unit amenities.” Finally, it should be noted that the numbers reported for some factors that
are difficult to measure, such as nearby use, frontage, noise, landscaping, security, and architecture, do not
represent actual counts. Instead, these numbers represent indices the levels of which are chosen so that they
reflect the relative differences across projects as assessed by the analyst. Obviously, some judgment needs to
be applied by the analyst in determining index levels for these attributes, which makes their quantification
less objective. Please read the notes at the bottom of Table 9.2 for clarifications and definitions of how
some of the attributes listed in the table have been measured.
It should be emphasized that the factors presented in Table 9.2 do not represent an exhaustive list of all
the factors that may be relevant in evaluating a residential project against competing projects.This list needs
to be adjusted so that it includes all the features of the specific project considered that may have a bearing
on its value and marketability. Features that characterize competing projects, but are absent from the subject
property, should be definitely included in the tabulation so that the weaknesses of the latter vis a vis the
former are properly accounted for in assessing its competitive strength.

EVALUATE COMPETITIVE DIFFERENTIALS THROUGH AMENITY INDEX TECHNIQUES  At this stage the
analyst needs to translate the numbers describing the project’s various attributes into indices or scores that
reflect the relative position of the subject property vis-​à-vis the competition. These indices are very useful
because they allow the analyst, on the one hand, to add up the scores from all attributes and come up with
an overall amenity index for each project, and, on the other hand, to quantify how the subject property’s
score compares to the “average” score implied by the amenity indices of all the competitive projects under
consideration.
The “average” score of all the competitive projects under consideration can be thought of as the
market or submarket average amenity level (depending on the geographic focus of the competitive ana-
lysis) against which the project will have to compete. The competitive position of the project relative to
this “average” amenity level should be a key determinant of the share of market absorption that it will
eventually capture.
Table 9.3 presents the results of the application of the third step in the numerical example under con-
sideration. The main difference of this table from Table 9.2 is that all numbers are presented in the form of
indices/​scores. In creating these indices, one basic principle is used. This principle is to set the least favorable
measurement to 100. In identifying the least favorable measurement, the analyst needs to consider whether
a lower or a larger count is preferred for the attribute under consideration. For example, for factors such
as nearby traffic, crime rate, property tax rate, project density, and distances to different destinations, such
as schools, employment centers, and shopping, lower counts are more preferable than larger counts. On
the contrary for factors, such as school system expenditure per pupil, number of mass transit lines within
walking distance, neighborhood average income, number of police officers and firemen per thousand
population, total local government expenditures per person, total capital expenditures per person, acres of
open space, size of unit, number of bedrooms, size of kitchen, and number of fireplaces, higher counts are
viewed more favorably than lower counts by consumers.
As a result of the different way these two groups of factors are viewed by consumers, different formulas
should be used for converting the counts into indices. Box 9.1 presents the formulas that need to be used
in constructing indices for each of the two groups of amenities.

194
Table 9.3  Step 2: Evaluating competitive differential through amenity index techniques
Attributes Amenity indices

Subject Comps

Project Project 1 Project 2

Site and environment

Nearby traffic1 132 100 109
Nearby use 110 100 105
Frontage 120 100 105
Noise 105 95 100
Pollution Similar for all properties
Neighborhood/​community
School system
Expenditure/​pupil 114 106 100
Crime rates1 114 100 143
Property tax rates1 100 100 120
Mass transit lines within walking distance 150 150 100
Neighborhood prestige/​incomes 107 107 100
Police services
Police officers/​thousand 110 100 143
Fire protection
Fireman/​thousand 117 100 133
Total expenditures/​person 130 100 150
Capital expenditures/​person 117 100 133
Access/​linkages:
Distances to:
School1 175 100 135
Employment center1 100 140 120
Shopping1 153 100 120
Recreation Similar for all properties
Project amenities
Density
Units/​acre1 120 100 140
Open space (acres) 200 100 150
Swimming pool Similar for all properties
Tennis court Similar for all properties
Health/​fitness Similar for all properties
Landscaping 100 100 105
Security 105 107 100
Parking space/​unit Similar for all properties
Structural unit amenities
Architecture 105 100 100
Age (months) 102 101 100
Size (sq. ft.) 104 100 110
Bedrooms Similar for all properties
Kitchen (sq. ft.) 107 100 150
Bathrooms 100 133 100
Patios Similar for all properties
Fireplaces 200 200 100
Total unweighted amenity values2 3,196 2,839 3,071
Competitive position index3 1.03 (Calculations not shown here)

Notes:
1
This index can be estimated as follows:
(Largest Attribute Value –​Project I’s Attribute Value) /​(Largest Attribute Value) * 100 + 100
2
Weighted amenity values could be constructed by weighing the relative importance of each characteristic on project
value. Weights can be assigned to project characteristics with the aid of experts in the local market. Consumer surveys
may present a better alternative.
3
The ratio of the subject development’s amenity value over the average competitive amenity value gives the project’s
competitive position index.
Numerical example republished from Clapp (1987)
 Analyzing residential real estate markets

Box 9.1  Formulas for calculating amenity indices

For attributes for which lower counts are viewed more favorably by consumers (for example, distance to school),
formula (9.1) should be used:

Index = [[(Largest count -​Project i’s count)/Largest count]*100]+100   (9.1)

For attributes for which higher counts are viewed more favorably by consumers (for example, school system
expenditure per pupil), formula (9.2) should be used:

Index = (Project i’s count*100)/​Lowest count   (9.2)

Below we present an example of the use of these two formulas for the estimation of the subject property’s
index for two relevant attributes:  distance to school and school system expenditure/​pupil. The subject
property’s distance to school is 1.25 miles, while the longest distance indicated in Table  9.3 is 5 miles.
Since a lower count for this attribute is preferable, formula (9.1) is used to calculate the subject property’s
index as:

Index (distance to school) = [(5 –​ 1.25) /​ 5] * 100 + 100 = (0.75*100) + 100 = 175

The result seems to make sense and reflects that the subject property is 75% better than the comparable
property that is furthest away from the closest school in the area.
In the case of school-​system expenditure/​pupil, the project’s count is 4,000 and the lowest measure-
ment in Table 9.2 is 3,500. Since a larger value for this attribute is more desirable, formula (9.2) is used for
deriving the project’s index as:

Index (expenditure/​pupil) = (4000/​3500)*100 = 114

Once scores/​indices have been developed for all attributes examined, they can provide the basis for
developing a number of several useful measures for the projects included in the analysis:

• Unweighted or weighted total amenity index (TAIi)

Represents a property’s overall amenity score/​index calculated as the unweighted or weighted sum
of the amenity indices of all its attributes j (AIj), as indicated in (9.3), where Wj represents the weight
assigned to each attribute j:

TAIi = ΣAIj or TAIi = Σ Wj AIj (9.3)

• Relative amenity indices (RAISi)

Describes the competitive position of the subject property relative to a given comparable property and
is calculated as the ratio of the subject’s total amenity index (TAIs) over a comparable’s total amenity
index (TAIi):

RAISi = TAIS /​ TAIi   (9.4)

• Average amenity index (AAI)

Describes the average amenity level in the marketplace within which the project under consideration
will be competing and is calculated as the average of the total amenity indices of all the comparables

196
 Analyzing residential projects

evaluated. The calculation of a simple average in the case of n competing projects can be carried out
using the following formula:

AAI = Σ TAIi /​ n   (9.5)

In most cases, a calculation of the weighted average is more appropriate using as weights (Wi) the
number of units each comparable comprises in the case of residential or square footage in the case of
the other property types. In such a case the formula for calculating the average amenity index will be:

AAI = Σ Wi TAIi /​Σ Wi   (9.6)

• Subject property competitive position index (CPIs)

Describes the overall competitive position of the project in the marketplace and is calculated as the
subject’s total amenity index (TAIs) over the average amenity index:

CPIs = TAIs /​ AAI   (9.7)

Table 9.3 presents estimates of the amenity indices for the properties included in our example, as well the
project’s competitive position index. The total amenity indices for each property have been calculated by
summing up the scores/​indices for all amenities for each project without applying any weights whatsoever.
The competitive position index of the project (calculated as the project’s unweighted total amenity index
over the average amenity index) is 1.03, suggesting an amenity level 3% higher than the market-​average
amenity level.
Ideally, in summing up the scores of the different attributes in order to calculate total amenity indices
for a specific project, the analyst needs to assign weights since some structure, project, and location features
may weigh more heavily than others in the minds of consumers when they evaluate a property. If there is
information for a sufficient number of properties, the most accurate way to assess these weights is through
hedonic valuation techniques that will be discussed in the next section of this chapter. Otherwise, attribute
weights can be determined using judgment with the aid of experts in the local marketplace. Consumer
surveys, although more expensive and time consuming, represent a better alternative in terms of more
accurately determining such weights.

COMPUTE COMPETITIVE MARKET PRICES (OR RENTS) FOR THE PROJECT  The fourth major step of the
competitive differentials technique is the computation of the competitive price or rent the project under
consideration would command in the marketplace at the time of analysis, given its characteristics and its
advantages or disadvantages compared to competing projects. This task can be accomplished by calculating
relative amenity indices (using (9.4)), and applying them to the respective comparable’s market rent/​price per
the formula described by (9.8):

Project rent or price = Relative amenity index * Rent or price of respective comparable   (9.8)

Obviously, during this stage, the analysis focuses only on completed and operating properties for which
price or rent information is available. Competitive projects that are under construction or have received
permit, but not started yet, must be definitely included in the previous stages of analysis, and especially in
the calculation of the project’s CPI, but cannot be used for the calculation of the project’s price at the time
of analysis.
A very important word of caution regarding the application of (9.8).The prices of comparables entering
this formula can be expressed as either the total price or price per square foot. The analyst should be very
careful in selecting which of the two price measures to use. In particular, if the relative amenity indices do

197
 Analyzing residential real estate markets

Table 9.4  Step 3: Compute competitive market prices

a.  Calculate relative amenity indices

Relative amenity indices are given by the ratio of the total amenity index of the subject property over the total
amenity index of each comparable property. For example:
Using Project 1 Using Project 2
Subject’s relative amenity index 1.126 1.041

b.  Estimate achievable project price or rent as:

Relative (Un)weighted amenity index * Price of respective comparable property

For example:
Using Project 1 Using Project 2
Subject’s price: $152,601 $155,579
Given the inaccuracies inherent in the technique and imperfect pricing, the use of a number of comps will lead
to different sales price (or rent) estimates. Normally, a sensitivity analysis must be conducted using a range of
estimates.

capture differences in square footage between the subject and each of the competing projects, then the total
price and not price per square foot should be used. If the relative amenity indices do not reflect differences
in size across projects then the price per square foot should be definitely used when applying (9.8).
Table 9.4 presents some estimates of relative amenity indices and prices for the project. Obviously, using
(9.8) the analyst can generate as many price estimates as the number of competitive properties for which
price information has been collected.4 Given the lack of scientific estimates of how the different attributes
may weight in the minds of consumers, as well as imperfect market pricing, it is very likely that these alter-
native project value estimates will be different. A single final value estimate for the project can be derived
by a simple averaging of all the estimates. Alternatively, the analyst may calculate a weighted average, by
assigning greater weights to properties considered to better reflect the project under consideration. The
same formulas (adjusted to include rent instead of price) can be used to derive project rent estimates.
The analyst should be cautioned to avoid producing a single point price/​rent estimate, by multiplying
the project’s CPI times the average price of the different comparables, for two reasons. The first reason is
that such calculation is not mathematically equivalent with the methodology described in the previous
paragraph. The second reason is that the calculation of the project’s CPI may take into account projects
for which no market price or rent is available (projects that are under construction or received permit but
not started yet) and, which are not taken into account in the calculation of prices for the project using the
relative amenity indices.

Critique
The main objective of the competitive differential technique is to assess the relative strength of a pro-
ject through a detailed analysis of its competition. To this aim, the methodology evaluates and quantifies,
via non-​econometric techniques, how the subject property’s attributes measure up to the attributes of

198
 Analyzing residential projects

competitive properties. Relative amenity indices can be used to produce alternative estimates of project
prices/​rents, which in turn can provide the basis for deriving a range of project prices/​rents or a single
point estimate.The project’s overall strength relative to the average competing project in the marketplace is
quantified by estimating its competitive position index. This competitive position index can then provide
the basis for estimating project absorption. As Clapp (1987) points out, however, the technique suffers from
several shortcomings, especially when it comes to estimating the price/​rent the subject property is more
likely to command in the local marketplace. These shortcomings include the following:

• It is not always clear what constitutes “comparable and competitive” property.

• Often, equal weights are assigned to the various attributes, although it has been widely established in the
empirical housing literature that different attributes are indeed valued differently by households (Carroll
et al., 1997).
• Even when different weights are assigned to different attributes based simply on experience, such
weights are bound to be inaccurate. Given that each property is unique with multiple attributes, it is
very difficult to assess the average effect of every single attribute within a specific collection of proper-
ties in a way that will be consistent across all properties without using appropriate statistical procedures.
Therefore, it is very unlikely that differential attribute weights, assigned simply on the basis of experi-
ence, will accurately capture the average effect of each attribute on project value or rent as implied by
the pricing of the specific group of properties examined.

Hedonic valuation techniques

Given the inherent inability of competitive differential techniques to accurately assign weights for the
different attributes that influence house prices or rents, the most sound methodology for addressing
questions of project value is hedonic valuation analysis.
This approach solves the problems identified in the case of the competitive differentials technique
because the weights are not assigned subjectively, but estimated through advanced and widely established
statistical procedures that use information on market transactions and property-​specific data. Furthermore,
this technique does not require that the information collected should refer to properties that are exactly
the same or as similar as possible to the subject property. In fact, some variation in the characteristics of the
housing units for which information is collected is desirable in that it helps better estimate the effect of an
attribute on the prices of the properties that have that characteristic vs the properties that do not have that
feature. Since the basics of regression analysis have been discussed in Chapter 7, this section focuses only
on how this econometric technique is used for hedonic price analysis.
The premise of hedonic price analysis is that housing is a hedonic good, because it represents a bundle
of attributes (intrinsic and extrinsic) that contribute to consumer utility and, as such, are valued by the
consumer. Intrinsic attributes are those that characterize the project itself, and extrinsic are those that are
external to the project, such as location. Consumers pay one price for hedonic goods, but such a price can
be thought of as the “sum” of the implicit prices of all their attributes.

Analysis steps
Application of hedonic valuation techniques in residential project analysis involves the following three steps:

1. Collect and carefully inspect data on transaction prices or contract rents and characteristics of proper-
ties sold or rented in the local marketplace at the time of analysis.
2. Use econometric techniques to estimate alternative hedonic price or rent equations and identify the
specification that best fits the data.

199
 Analyzing residential real estate markets

3. Use the econometric estimates of the implicit prices of the different attributes to estimate the total
price of the bundle of characteristics that describes the housing project under consideration.

As in the case of competitive differential techniques, the application of hedonic valuation techniques
requires the collection of data on contract apartment rents or transaction prices (in the case of owner-​
occupied housing), property characteristics, and location attributes. Contrary to the competitive differentials
technique, however, the properties included in a sample for hedonic valuation purposes are not strictly
those that directly compete with the project under consideration; they typically are of similar type to the
subject property, but not necessarily with exactly the same structure or location characteristics. In fact, from
an econometric point of view, it is desirable to ensure considerable variation in attributes across the prop-
erties included in the sample. Since hedonic valuation techniques attempt to quantify statistically the effect
of such variations on market prices or rents, such a sample structure helps obtain more robust estimates of
these effects.
In gathering the data, it is important to keep in mind that hedonic price analysis can be very helpful in
estimating the density that will maximize the development’s residual land value, but only if information
on the density (units per acre or FAR) of the developments included in the sample is collected. A very
important step, after the data have been collected and processed, is their careful inspection to make sure that
there are no abnormal or extreme values for the different variables (prices or attributes). If such values are
identified then the analyst needs to examine whether they make any sense at all and whether they should
be removed from the sample.
The second step of hedonic valuation analysis involves the econometric estimation of equations that
express house prices (P) or rents(R) as a function of structure, project, and neighborhood attributes (Xi)
that are potentially valued by households. The simplest possible specification of a hedonic valuation model
is described by the linear model in (9.9) and graphed in Figure 9.3.

P = a0 + a1X1 + a2X2 + …… + anXn (9.9)

Coefficients b0 through bn in (9.9) represent the additional amount the consumer is willing to pay for
an extra unit of the characteristic the coefficient is referring to. In other words, if Xi changes by one
unit the market price, P, will change by an amount equal to ai. A problem of this linear specification is
that it assumes that this amount is constant regardless of the value of the attribute. For example, if X1
represents the number of fireplaces, the implicit assumption of (9.9) is that a1, or the implicit price of
an extra fireplace, is the same for the first, second, third, and nth fireplace. Yet the marginal consumer
utility associated with the second fireplace must be lower than the marginal consumer utility associated
with the first one. Consequently, the consumer should not be willing to pay as much more for the
second fireplace as he/​she is willing to pay for the first. This is referred to as the principle of diminishing
marginal utility.
The principle of diminishing marginal utility implies a non-​linear mathematical relationship between
price (P) and project attributes (Xs). Such a relationship is described by (9.10) and graphed in Figure 9.4,
where the diminishing effect of Xi on P as Xi increases is reflected in the diminishing slope of the curve.
The non-​linear relationship described by (9.10) can be transformed into a logarithmic formulation as
described in (9.11).5

P = e a0 X 1a1 X 2a2 ....X nan (9.10)

lnP = a0 + a1lnX1 + a2lnX2 + ……+anlnXn (9.11)

200
 Analyzing residential projects

Residential hedonic models:

Residential hedonic models express housing prices, P , as a function of hedonic housing
attributes, X:
P = f (X)
Linear hedonic models:

P=ao +a1 X1 +a2 X2 +a3 X3 + ... an Xn

where:
P : market price
X : hedonic attributes
ao : intercept
a1 , a 2 , a 3 , . . , a n : regression coefficients; can be interpreted as implicit attribute prices

a 1 = DP/DX 1
If X 1 changes by 1 unit
P changes by a1

P(X)

slope

a0

X1
Problem:
Implicit prices are constant, regardless of the value of the attribute. If X1, for example,
represents the number of fireplaces, the implicit price of the 1st, 2nd, 3rd, and nth fireplace is the
same. Yet, the marginal consumer utility associated with the second fireplace must be lower than the
marginal consumer utility associated with the first. Consequently, the consumer should not be
willing to pay for the 2nd fireplace as much as he/she would be willing to pay for the first.

Figure 9.3  Linear hedonic models

However, (9.10) can be re-​written as:

P ' = a0 + a1X 1' + a2X 2' +  + an X n' (9.12)

where

P ' = ln P , X 1' = ln X 1, X 2' = ln X 2 ,..., X n' = ln X n

201
 Analyzing residential real estate markets

Non-linear hedonic models (account for diminishing marginal utilities)

a a a
or P  e a0 X 1 1 X 2 2 X 3 3 ..... X nan

ln P  a0  a1 ln X 1  a2 ln X 2  a3 ln X 3  ....... an ln X n

a 1 = D lnP/D lnX 1 Implicit value of X 1 :

If X 1 changes by 1%, P(X 1 =m+1)-P(X 1 =m)
P increases by a 1 %

P(X)

a0

X1

Figure 9.4  Non-​linear hedonic models

The statistical equation described by (9.12) is linear in its parameters, which means that it can be estimated
with standard regression techniques using the natural logarithm of price as the dependent variable and
the natural logarithms of the values of the different attributes as independent variables. As the variables in
both sides of the model are in logarithms, this is typically referred to as the log-​log model as opposed to
the semi-​log model in which only the dependent variable is expressed in logarithmic form. The estimated
coefficients of the former specification represent the percentage increase in price as a result of a 1% increase
in the value of the attribute they refer to. For example, if the value of attribute X1 increases by 1% then
the house price will increase by a1%. In other terms, these coefficient estimates represent elasticities. Irwin
(2002) estimated alternatively log-​log, semi-​log, and simple linear house price hedonic models. Her results
show the clear superiority of log-​log and semi-​log models over the simple linear model with the log-​log
model results appearing slightly superior to the semi-​log model results.
At this point, a special note on dummy variables is warranted. Such variables often need to be included in
hedonic price specifications because there are some attributes which cannot be represented by continuous
or integer variables, such as the provision of assigned parking or the inclusion of a pool in the development
and many more. Such attributes can be represented numerically by constructing a so-​called dummy vari-
able that takes the value of one if the attribute is present and zero otherwise.

What is the natural logarithm (ln) of X?

It is the exponent to which we need to raise e = 2.71828 to obtain X. For example:
Ln X = a implies that X= ea or X= EXP(a)
where EXP denotes exponential function

Property attributes represented by dummy variables cannot be introduced in logarithmic form in hedonic
price equations.6 In fact, not only dummy variables, but also any other variable that takes zero or negative

202
 Analyzing residential projects

values cannot be introduced in logarithmic form because the logarithm of zero and negative numbers cannot be
calculated. A feature that may present such a specification problem is the fireplace, if the sample includes
both properties without fireplace and properties with one or more fireplaces. In such cases, the math-
ematical formulation and respective linear transformation of the hedonic price equation is described by
equations (9.13) and (9.14), where D1 represents a dummy variable and L1 a continuous or integer variable
that cannot be introduced in logarithmic form:

P = e a0 + a1D1 + a2 L1 X 1a3 X 2a4 ....X nan (9.13)

lnP = a0 + a1 D1 + a2 L1+ a3 lnX1 + a4 lnX2 + …… + an lnXn (9.14)

Zietz et al. (2008) in an empirical analysis of sales prices of houses sold in Orem/​Provo in Utah used quan-
tile hedonic regressions to quantify the effect of different housing characteristics on house prices. Their
findings suggest that the effect (regression coefficients) of some characteristics differ considerably across
different house price levels, suggesting that buyers of high-​price homes value more certain characteristics,
such as square footage, lot size, and bathrooms, than buyers of lower-​priced homes.
The third step of hedonic valuation analysis involves the use of the parameters of the estimated
equation to calculate the market price/​rent that the subject property should command given its specific
characteristics. We demonstrate specifically how such calculations can be carried out in the presentation of
a numerical example, after we discuss the different ways that hedonic valuation techniques can aid when
analyzing residential projects.

Applications
Hedonic techniques are very useful in residential microeconomic analysis due to the wide range of their
applications. They can be particularly useful in:

• Estimating project market price or rental rate given its attributes or amenity package
• Determining optimal project design
• Determining optimal development density.

Estimating project price or rent given its amenity package

One of the most crucial inputs in determining the feasibility and viability of a project is the market price
or rent it will command upon its completion. Hedonic valuation techniques can help estimate the most
likely project price or rent at the time of analysis, which will then provide the basis for forecasting the price
or rent the project will command upon its completion.The analyst can generate such forecasts by applying
to the estimated project price or rent the forecast growth rates for market prices or rents obtained from
the macroeconomic analysis of the local housing market. Such market-​level price/​rent growth rates should
not be applied blindly to the particular property examined. They should be adjusted appropriately based
on the analysis of the property and its location. The competitive position index, discussed earlier in the
competitive differentials technique, can provide the basis for adjusting the market-​wide price/​rent growth
rates in order to derive property-​specific price/​rent growth rates. In particular, the property-​specific price/​
rent growth rates can be derived by multiplying the market price/​rent growth rates by the property’s com-
petitive position index.
Hedonic regression techniques can be very helpful in estimating the most likely price or rent for the
project at the time of the analysis since they provide more reliable estimates of the implicit market prices

203
 Analyzing residential real estate markets

of each of the project attributes. Once a satisfactory econometric estimate of market price is obtained, the
analyst can calculate the price of the specific project by utilizing the estimated parameters of the regression
equation and the values of its different attributes.We demonstrate how this calculation is carried out in the
hedonic valuation example that follows this section.

Determining and evaluating optimal project design

Typically, the optimal design for a residential project is determined through a number of different types
of qualitative analyses, mostly including consumer surveys. The groups commonly surveyed include active
home buyers or renters in the local marketplace, residents of competing residential developments, targeted
household groups carefully selected by type and income, or employees at employment concentrations con-
veniently accessible from the site under consideration. Such surveys are often complemented by interviews
with sales agents and residential developers in the project’s market area, as well as collection of primary data
regarding the amenity mix, sales rate, and sales prices of competing projects. These surveys aim at better
defining the project’s target household groups, the particular housing design features mostly demanded by
these groups, and the prices they are willing to pay for such features.
It is emphasized that sound competitive analysis for optimizing project design should extend beyond
a simple listing of what design features and price ranges sell best at the time of analysis. The analyst
needs to also understand the economic and demographic forces underlying such preferences and market
activity. Such an understanding will be instrumental in evaluating how the desirability of popular design
features and prices may change if changes in their underlying economic and demographic determinants
are anticipated to take place over the forecast period.
Although consumer surveys and analysis of competitive projects provide the primary input in deter-
mining optimal residential project design, the results of hedonic price analysis can be helpful in evaluating
the economic and financial implications of such a design. First of all, they allow the analyst to estimate the
market price or rent for the specific bundle of attributes represented by a particular design. Second, it can
help estimate the achievable increase in a unit’s market price or rent from the addition of particular design
features.This analysis can provide the basis for first assessing the effect of such design modifications on pro-
ject revenue, and then evaluating the profitability of providing such features by comparing the additional
revenue they may generate to the cost of providing them.
In strict economic terms, the provision of an additional attribute is justified only if its marginal benefit
is greater or equal than its cost. For example, we can evaluate how much of a premium a second fireplace
will command in the marketplace by using the hedonic regression results to calculate the price of the pro-
ject assuming first the inclusion of one fireplace and then the inclusion of two fireplaces. As indicated by
(9.15) below, the difference of the latter from the former is the marginal benefit of including the second
fireplace.7

MB2nd fireplace = Price with 2 fireplaces –​Price with 1 fireplace (9.15)

Once the marginal benefit (MB2nd fireplace) is calculated, it can be compared to the cost of constructing an
additional fireplace (MC2nd fireplace) in order to evaluate whether it is economically feasible and profitable
adding this feature in the design. In particular, that would be the case if MB2nd fireplace ≥ MC2nd fireplace. It
should be emphasized, however, that if a certain amenity is considered critical for the successful marketing
of a project to its target groups, it may very well be provided even if MB<MC, as long as the project as a
whole is assessed to be reasonably profitable. This analysis could also be very helpful in guiding decisions
regarding the inclusion of amenities considered to be of secondary importance for the successful marketing
of the project.

204
 Analyzing residential projects

Numerical example
The objective of the regression analysis example (presented in Tables 9.5 and 9.6) is to demonstrate how
the results of hedonic price analysis can be used to estimate the price of condominium units with certain
characteristics. For this purpose, we are using a modified version of the house price equation estimated
by Webb (1982). Using a large sample of condominium sales, this analyst estimated the following hedonic
price equation:

lnP = b0 + b1 lnUNI + b2 lnBATH + b3 lnLOT + b4 lnRMS + b5 lnST + b6 AGE

+ b7 DATE + b8 (1/​TOM)
     (9.16)

The independent variables in (9.16) include the number of units in the building (UNI), number of
bathrooms in the unit (BATH), building lot size (LOT), number of rooms (RMS), number of stories
(ST), age of building (AGE), date the unit was sold (DATE), and time on the market (TOM). Note that
the dependent variable and most of the independent variables are in logarithmic form in order to reflect
the diminishing marginal effects of the different attributes. Also, note that the number of bedrooms is not
included because it is collinear with the number of bathrooms (BATH). AGE, DATE, and inverse TOM do
not enter the equation in logarithmic form, as the data do not support such specification.
Table 9.5 presents the coefficient estimates for (9.16). These coefficients are not exactly the same as the
ones reported by Webb (1982). The modifications were deemed necessary in order to allow for the mar-
ginal value analysis presented in Table 9.6. Table 9.6 demonstrates the usefulness of implicit value estimates

Table 9.5  Example on hedonic valuation techniques

Variable definitions
Dependent variable
SP Condominium sales price
Independent variables
UNI Number of units in the building
BATH Number of bathrooms in each unit
LOT Building lot size
RMS Number of rooms
ST Number of stories
AGE Age of building
DATE Date the unit was sold
TOM Time on the market (days)
Hedonic regression results

lnSP = 11.02 +0.038 lnUNI +0.042 lnBATH+ 0.033 lnLOT+ 0.274 lnRMS

+ 0.025 lnST –​0.011 AGE +0.00001 DATE + 0.0439 (1/​TOM)
R-​squared = 0.76

Modified from Webb, J. R. 1982.Valuation of Multifamily Residential Properties. In Sirmans, F. (ed), Research in
Real Estate,Volume 2. Greenwich, CT: JAI Press.

205
 Analyzing residential real estate markets

Table 9.6  Estimating market prices and marginal amenity values

Given the regression results in Table 9.5, the value of a new housing unit with given characteristics can be estimated. Consider
the Base Case below:

UNI 20
BATH 1.5
LOT 20,000
RMS 4
ST 5
AGE 0
DATE 1978
TOM 30

Then:
ln SP = 10.82 + 0.038 ln(20) + 0.042 ln(1.5) + 0.033 ln(20,000)
+ 0.274 ln(4) + 0.025 ln(5) –​0.011 (0) + 0.00001 1978 + 0.0439 (1/​30)
= 11.72
SP = Exp(11.72) = $122,885

Implicit values estimated from regression results are very useful in helping decide what quantity of an attribute
to provide. Using the regression results and the base case, marginal values for two-​unit attributes (i.e., rooms
and bathrooms) are estimated. As the tables below indicate, the marginal value of the fifth room is $6,692. If the
marginal cost of providing the fifth room is less than $6,692, then it would make economic sense to provide
it. The final decision, however, should be made taking into account the surveyed preferences of the specific
consumer groups most likely to be attracted to the development.

Attribute, unit, and marginal values

Rooms Baths

Attribute Unit value Marginal value Attribute Unit value Marginal

value
4 rooms 122,885 1.5 baths 122,885
5 rooms 130,633 7,748 2 baths 124,379 1,494
6 rooms 137,325 6,692 2.5 baths 125,550 1,171

in helping decide what quantity of an attribute to provide. In particular, the table presents marginal value
estimates for two different attributes, namely, rooms and bathrooms. These estimates were derived using
formula (9.15), as it applies to the different attributes. For example, the marginal value of $7,748 for the
fifth room was calculated by subtracting the estimated hedonic price of a unit with four rooms ($122,855)
from the hedonic price of a unit with five rooms ($130,633). If the marginal cost of providing the fifth
room were less than this amount then it would make economic sense to provide it.

206
 Analyzing residential projects

The analyst can experiment in the same way with different design features along the same lines. In
carrying out this type of experimentation, the analyst has to keep in mind three things. First, additional
design features increase price and, therefore, the analyst needs to ensure that the selected bundle of attributes
commands a price/​rent that is affordable to the project’s target group. Second, the quantities of attributes,
which are alternatively examined for the purpose of calculating their marginal values, have to be within
a reasonable range. For example, examining the marginal value of a third or a fourth bathroom in a two-​
bedroom unit makes little sense. Finally, the additional design features examined need to fall within the
range of the preferences of the consumer groups most likely to be attracted to the site.

Critique of the hedonic valuation technique

The hedonic valuation technique provides proper and scientifically sound estimates of the “weights” that
apply to the different project attributes when calculating its more likely market price. Furthermore, it allows
the use of the direct measurements of the values of the different attributes, while allowing for an objective
representation of attributes that cannot be measured (through the use of dummy variables). As such, it avoids
the potential inaccuracies and inconsistencies that may be involved in developing amenity “scores” or indices
for a number of diverse attributes, aiming at measuring the relative desirability of each attribute across projects.
A potential problem in using the results of hedonic analysis to calculate prices in the future is that
preferences and willingness to pay for different attributes may change through time. If that is the case, then
project price/​rent forecasts may not accurately reflect the price/​rent the property will command in the
future. The competitive differentials technique, however, suffers also from the same problem, and, perhaps,
even more so, since the estimate of the project’s current price is more inaccurate to start with.
Another potential problem is that values assigned to different attributes may differ across different
groups of consumers, a variable that is difficult to control for when running hedonic price regressions,
because of the lack of data. Again, this problem is also present when applying competitive differentials
techniques. Thus, despite these minor shortcomings, hedonic valuation techniques are undoubtedly more
sound than competitive differentials techniques for developing project price/​rent estimates.
According to Graevenitz and Panduro (2015), from an econometric point of view, the major concern
in using the hedonic model is the issue of omitted variables, as this modeling approach may not accurately
capture the spatial variations in prices due to misspecification, mismeasurement, or omission of inde-
pendent variables that have a spatial dimension.

Project absorption and gross revenue analysis

Once project price/​rent and density analysis has been completed and the development plan and product
design has been finalized, the analyst can proceed with the estimation of the project’s absorption and gross
revenue schedules.8 The focus of this analysis is the share of market demand that the residential project will
most likely capture given its competitive strength, as assessed through the competitive differentials tech-
nique. Since a project’s absorption is influenced by the overall demand-​supply conditions prevailing in the
local housing market at the time of its completion, the analyst needs to integrate and synthesize, at this stage,
the results of the macro and micro analysis, as indicated in Table 9.7.
The basic formula for calculating the demand or absorption for a specific housing project is given by
(9.17) and is based on the concept of fair market share:

(U p / CSM )* CPI S * EDM

Project absorption = Project’s fair Competitive Effective market (9.17)
market share position index demand

207
 Analyzing residential real estate markets

Table 9.7  Integrating macro and micro analysis results in project absorption and gross revenue schedules

Analysis stages Usefulness

Micro
Product type, price/​rent, and density analysis results Determine the number of units that will be built and
compete in the market for a share of local housing
demand

Macro
Market price/​rent index forecast
Micro Use jointly to develop forecasts of project price or rent
Project current price/​rent estimate over planning horizon

Macro
Forecasts of effective market demand and competing
supply Use jointly to estimate the project’s capture rate and
Micro absorption schedule. Use the latter along with price/​
Project available supply and competitive position index rent forecasts to generate gross revenue schedules.

where
Up : total number of units in the project, or number of units expected to be completed during the
period for which project absorption is calculated, if project phasing is planned
CSM : competing housing market supply
EDM : effective housing market demand
CPIS : project’s competitive position index

Formula (9.17) reflects the premise that a residential project, upon entering the market, will capture its
fair share of market demand, which is equal to the share of supply represented by the project, (Up /​ CSM),
adjusted by its competitive position index CPIS. The competitive position index can be calculated using
the competitive differentials technique discussed earlier in this chapter. Effective market demand can be
estimated using the accounting or econometric techniques discussed in Chapter 6.
The analyst is strongly advised at this point to ensure that CSM (competing housing market supply),
EDM (effective housing market demand), and CPIS (the project’s competitive position index) refer to the
same geographic area and housing type before applying (9.17). Depending on the type of technique used for
forecasting market demand and the time frame of analysis, potential inconsistencies may often need to be
reconciled in order to ensure that all the terms of (9.17) represent consistent numbers:

1. The analyst needs first of all to ensure that market supply forecasts entering (9.17) are consistent
with market demand forecasts in terms of both the geographical area and housing type they refer to.
If market analysis is conducted for the assessment of a development project, housing market demand
forecasts entering (9.17) need to refer to a specific residential product design of certain price. As
indicated, such forecasts are feasible using the disaggregate accounting approach that allows also
for focusing on the project’s primary market area. The latter may be smaller than the metropolitan

208
 Analyzing residential projects

area within which the project is located. The chances of carrying out an econometric analysis of
supply at this level of quality and geographical disaggregation are very slim, because of the lack of
historical data. Thus, generating a purely econometric forecast of competing supply for a very spe-
cific housing type and submarket is very difficult if not impossible. Although a short-​term forecast
(up to two years at maximum) of new competing supply can focus on the appropriate geographic
area and residential product type through a detailed pipeline analysis, longer-​term forecasts of com-
peting supply require econometric forecasting. Obviously, in such a case the analyst will need to
share down metropolitan supply forecasts by both submarket and housing type.This process requires the
following steps:
a) Generate an econometric forecast of aggregate housing supply comprising all housing types
within a given tenure type. Widely available residential market data typically allow distinction
between multi-​family and single-​family units as opposed to rental and owner-​occupied units.
b) Estimate the current shares of completions and existing inventory represented by the type of unit
under consideration and the project’s primary market area/​submarket.
c) Adjust estimated shares in (b) to reflect anticipated changes (if any) in the distribution of new
construction across housing types and submarkets over the forecast period.
d) Apply the adjusted shares to the econometric forecast of completions for the broader market in
order to forecast completions for the housing type represented by the project and its primary
market area or submarket.
e) Use forecast completions from step (d), existing inventory, and vacancy rate, along with demand
forecasts for the specific housing type and submarket under consideration to generate vacancy
rate forecasts.
f) Use forecast competing completions and vacant stock from step (d) to calculate competing avail-
able supply to enter in (9.17).
In carrying-​out step (b), the analyst can start by estimating the share of residential projects represented
by the housing type under consideration at the time of analysis. This share can be determined
through a thorough survey of the type and prices of residential projects under construction, those for
which building permits have been issued, but have not broken ground yet, and planned residential
developments for which no building permit has been issued yet. This analysis will help determine as
accurately as possible what percentage of residential completions in the first one to two years of the
forecast period will be competing with the housing type under consideration.
  The existing vacancy rate needed for applying step (e) can be assessed through a detailed survey
of vacant and total inventory of units representing the housing type/​price level considered within
the project’s submarket. In the lack of such data, the analyst will need to come up with estimates
of the shares of total and vacant stock represented by the housing type/​price level and submarket
considered. Local housing brokers may be able to provide estimates of housing-​type and submarket-​
specific vacancy rates based on their knowledge of the local market.
2. In applying formula (9.17), demand forecasts are also needed for the same geographic area for which
the competing new supply has been estimated. Housing market demand forecasts, especially those
derived through econometric techniques, may often refer to a larger geographic area than the one involved
in the analysis of the competition and the derivation of the project’s CPI. Such inconsistency may lead
to inaccuracies in estimating project absorption, especially if the residual area not covered by the com-
petitive analysis includes a significant number of competing properties, most of which are considerably
inferior or superior to the project under consideration. In such a case, the analyst needs to ensure that
EDM (effective market demand) in (9.17) refers to the same geographic area that CPIS (the project’s com-
petitive position index) refers to. Ideally, this area should represent the project’s primary market area, but
budget and time constraints may force the analyst to confine the analysis to a smaller submarket.

209
 Analyzing residential real estate markets

Within this context, ensuring consistency between EDM and CPIS may require estimation of the share
of market absorption most likely to be captured by the project’s primary market area or submarket.
Such a share is typically estimated based on the recent performance of the submarket or primary
market area relative to metropolitan market performance. Before extrapolating the submarket’s most
recent absorption share to the future, the analyst needs to evaluate whether such a share is likely to
remain stable, increase, or decrease in the years ahead. In assessing such prospects, the analyst may want
to look at a number of factors, such as potential changes in the relative attractiveness of the project’s
submarket, the potential distribution of new jobs and residential construction across submarkets, and
changes in the availability of housing for sale or rent within the project’s submarket relative to other
submarkets.
3. Residential market demand forecasts derived through econometric techniques typically refer to the
aggregate market comprising all housing types, while competitive analysis and project absorption refer
to housing units of specific design and price. In such a case, the analyst will also need to estimate the
share of aggregate housing demand expected to be captured by the housing type under consideration
in order to ensure consistency between EDM and CPIS. Recent historical shares, as well as expected
economic developments and demographic changes, can provide the basis for projecting the share of
aggregate housing demand most likely to be captured by the housing type under consideration. Such
adjustment is not needed if market demand has been estimated using the disaggregate accounting
approach, which takes into account the very specific design of the development and its most likely
market price.

Analysts may not use (9.17) for estimating project absorption. Often, absorption rate estimates for a planned
residential development are derived on the basis of the absorption rate or time-​on-​the-​market experience
of similar residential projects active in the marketplace at the time of analysis. In such a case, the analyst
needs to carefully consider differences between the planned project and the comparables used, in terms of
prices, location, and project attributes. Most importantly, the analyst will need to very carefully adjust such
rates to reflect any anticipated changes in local market conditions and financing environment that are expected to
take place during the period that intervenes between the time of analysis and project completion.

Numerical example of project absorption analysis

In the numerical example that follows, the underlying objective is to compute the absorption and gross
revenue schedule for a hypothetical residential project. It is assumed that the developer plans to maximize
the capacity of the site by building 200 units. The planned time of entry is the third year from the time of
analysis (Year 3) during which all units are expected to be completed. No project phasing is considered at
this point.
The absorption analysis for this hypothetical project is presented in Table 9.8. Column [2]‌in this table
presents the forecast of effective demand for Years 3–​6, which is typically calculated using the disaggregate
accounting approach, since it needs to refer to specific target demographic groups.9 Column [3] presents
the forecast competing supply, which as emphasized needs to refer to the same type of units and geographic area
that demand figures in column [2] refer to.
Given the data tabulated in Table 9.8, project absorption can be calculated as follows. Dividing 200 (the
number of units in the project) by the competing supply for Year 3 (in column [3]‌) provides the project’s
fair market share, which when multiplied by the competitive position index for that year (in column [6])
gives the project’s competitive market share, or market capture rate, shown in column [7]. Multiplying
the project’s market share by the forecast effective market demand for that year (in column [1]) provides
an estimate of project absorption, shown in column [8]. Notice that the competitive position index of
the project is assumed to decline through time. This reflects simply the fact that the project will be less

210
 Analyzing residential projects

Table 9.8  Residential project absorption analysis

One-phase example: Project size = 200 units, time of entry = Year 3

Year Forecast Competing Market Available Compe- Project Project Price/ Gross
[1] effective market price units titive market absorption unit potential
market supply index [5] position share [8] [9] income
demand [3] [4] index [7] [10]
[2] [6]

Year 1 100 $153,500

Year 2 101 $155,035
Year 3 5,147 5,870 100 200 1.030 0.035094 180 $153,500 $27,630,000
Year 4 3,507 3,518 104 20 1.030 0.005856 20 $159,640 $3,192,800
Year 5 3,072 3,495 110 0 1.025 0 0 $168,850 $0
Year 6 2,807 3,354 114 0 1.020 0 0 $174,990 $0

Available units/
Competing supply
x
Effective market demand
x
Project’s competitive position

Notes:
[1] Year of analysis.
[2] Figures are usually computed using the disaggregate accounting approach. The estimates must
refer to specific target groups. Aggregate econometric forecasts for metro area or submarket can
provide a broader framework for evaluating such demand forecasts.
[3] Number of housing units expected to compete for a share of effective demand. It is very
important that the figures refer to the same type of units and geographic area that effective
demand figures refer to. Since demand figures usually refer to specific types of units, competing
supply needs to be calculated using a combination of econometric and accounting techniques. See
the text for further discussion of this issue.
[4] Usually derived through econometric techniques.
[5] The number of available units in the Year 3 reflects the total number of units in the project.
The number of available units in Year 4 is calculated as: 200 - Year 3 Absorption (computed
in column [8]).
[6] Computed through competitive differential techniques.
[7] Project market share = [6] × ([5]/[3]), where [5]/[3] represents the project's “fair” market share.
[8] Project absorption = minimum ([5], [7] × [2]).
[9] Computed by applying the percentage changes implied by [4] to the first year price, estimated
through competitive differential or hedonic valuation techniques.
[10] Gross potential income = [9] × [8].

competitive compared to the new units that will be completed after it enters the market. This is consistent
with the widely established negative effect of aging on property values (all else being equal). As indicated,
the numbers in column [8] represent expected project absorption and are calculated using (9.18):

Project absorption = Project’s market share * Effective market demand (9.18)

211
 Analyzing residential real estate markets

Thus, expected project absorption in Year 3 is specifically calculated as:

Year 3 Project Absorption = 0.03509 * 5,147 = 180

This calculation shows that 180 units are expected to be absorbed upon completion of the project, given
the competing supply and market demand predicted for that year. Therefore, the number of available units
(column [5]‌) in Year 4 will be:

Available units (Year 4) = Available units (Year 3) –​Project absorption (Year 3)  (9.19)

Available units (Year 4) = 200 –​ 180  =20

Once the number of units expected to be available in the project in Year 4 is estimated, formula (9.17) can
be used to compute the number of units expected to be absorbed during that year. As Table 9.8 shows, all
20 units expected to be available during Year 4 should be absorbed, given that year’s predicted demand-​
supply conditions. Obviously, if the calculations indicated otherwise, the analyst would need to repeat the
series of calculations just described for Year 5 and so on.
Once an estimate of the number of units absorbed during each period is available, the analyst can proceed
with the calculation of the project’s gross potential income. This calculation requires forecasts of the prices at
which absorbed units will be sold or rented. Column [9]‌presents forecasts of such prices for Years 2 through
6. The price shown for Year 1 is the project price estimate derived through either the competitive differential technique
or the hedonic valuation technique at the time of analysis.The project price estimate for Year 2 is derived by applying
to the Year 1 price the forecast percentage changes in the market price index during that year as in (9.20):

Price (Year 2) = Price (Year 1) * Price index (Year 2) /​Price index (Year 1)  (9.20)

The price forecast for Year 2 is subsequently used in combination with the predicted market price indices
for Years 2 and 3 to calculate the price for Year 3 and so on. For example, the price estimate for Year 3 has
been calculated as:

Price (Year 3) = 155,035 * 100/​101 = 153,500

With the project price forecasts at hand, the analyst can estimate the gross potential income for Years 3 and
4, during which the units are expected to be absorbed. This is simply the product of the number of units
absorbed during each year (shown in column [8]‌) and the respective forecast unit prices (shown in column
[9]). These gross income calculations, shown in column [10] in Table 9.8, represent one of the most critical
inputs in evaluating project feasibility and viability.
The analysis presented in Table 9.8 assumes Year 3 as the year of entry. However, it may be worth exam-
ining whether the project will be more profitable if the time of entry is Year 4 instead.The project’s absorption
schedule and gross revenue schedule, under this scenario, can be derived using the formulas just discussed.
The rule of thumb in choosing among the two alternative scenarios is to select the one that has the
highest net present value (NPV), calculated as the difference between the present value of project revenues
(PVR) and the present value of project costs (PVC). In principle, entering the market later is more profit-
able (or results in greater NPV) only if expected price/​revenue increases are sufficiently high to offset the
effect of the discount rate.
Table 9.9 presents absorption schedule and gross potential income estimates, assuming that the time of
entry is Year 4. As the previous analysis has shown, when the project enters the market in Year 3 then 180
units are absorbed during that year and the remaining 20 are absorbed in Year 4. As Table 9.9 indicates,
when the project enters the market in Year 4, all 200 units are absorbed during that year.

212
 Analyzing residential projects

Table 9.9  Analyzing alternative times of market entry

Time of entry: Year 4
Year Forecast Competing Market Competitive Project Project Price/​unit Gross potential
effective market market price position market absorption income
demand supply index share
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌ [6]‌ [7]‌ [8]‌ [9]‌ [10]
Year 1 100 $153,500
Year 2 101 $155,035
Year 3 5,147 5,870 100 0 1.030 0 0 $153,500 $0
Year 4 3,507 3,518 104 200 1.030 0.05856 200 $159,640 $31,928,000
Year 5 3,072 3,495 110 0 1.025 0 0 $168,850 $0
Year 6 2,807 3,354 114 0 1.020 0 0 $174,990 $0

Table 9.10 presents the net present value (NPV) analysis for the two alternatives. The basic idea of the
present value analysis is that income streams received at different periods in the future are not directly
comparable, and thus they need to be discounted to the present time (hence the term present value). The
formula for calculating the present value (PV) of an amount expected to be received in the future after n
periods (FV), as well as the formula for calculating NPV, are provided in Table 9.10.
The calculations, presented in Table 9.10, point to a positive NPV for the project, when the time of
entry is Year 3, and a negative NPV, when the time of entry is Year 4. The reason for this result is twofold.
First, the income received one year later (Year 4 instead of Year3) has actually lower present value, PVR,
because the anticipated increase in house prices (4%) is quite a lot lower than the investor’s discount rate
or return requirement of 15%. Second the delay of the project for one year contributes to increases in
the cost of development, PVC, due, for example, to higher land holding costs. Therefore, the NPV analysis
suggests that, given the investor’s high return requirements, the project will be financially viable only if the
time of entry is Year 3.

Econometric estimation of project absorption

Can we use hedonic regression techniques to estimate project absorption just as we use them to estimate
project price? There are two issues in using hedonic regression techniques to estimate project absorption.
The major issue, of course, is data availability, because it is extremely difficult to find data on absorption rates
by project for as many properties as needed to provide a sufficient sample for estimating hedonic equations.
The second issue is an econometric one and has to do with the potential simultaneity between absorption
and prices that may be present in a structural-​model setting. Such a simultaneity issue may arise because
project absorption is a function of its price, while at the same time project price may also be shaped by its
absorption rate. These relationships are described by (9.21) and (9.22), which represent a simultaneous-​
equation system.

Absorption = f[Price, Amenities] (9.21)

Price = f[Absorption, Amenities] (9.22)

In (9.21), price should have a negative effect on absorption, while amenities should have a positive effect. In
(9.22), both absorption and amenities should be positively associated with price.10 Simultaneous-​equation
systems, such as the one described by (9.21) and (9.22), can be estimated through more advanced statistical
techniques that account for simultaneity.11

213
 Analyzing residential real estate markets

Table 9.10  Evaluating alternative development scenarios

n
Present Value Formula PV = FV/ (1+d)
PV = Present value
FV = Future value
d = Discount rate
n = Compounding periods

Net Present Value NPV = PVR - PVC

Formula
PVR = Present value of revenues
PVC = Present value of costs

FV3 = $27,630,000

Alternative 1
Revenue Schedules PVR = FV3 / (1+d)3+ FV4 / (1+d)4 =$19,992,667
Year of Entry: Year 3
d = 15%
FV4 = $3,192,800

0 1 2 3 4 5 6

NPV = PVR-PVC
PVC = $18,960,000 = $1,032,667

FV4 = $31,928,000

Alternative 2 PVR = FV4 / (1+d)4 = $18,254,938

Revenue Schedules
Year of Entry: Year 4
d =15%

0 1 2 3 4 5 6

NPV = PVR-PVC
PVC = $19,528,800
=-$1,273,862

214
 Analyzing residential projects

Chapter summary
The chapter has focused on the microeconomic aspects of residential market analysis.The major objectives
of such an analysis include defining the type of housing development and product design that will be most
marketable and profitable at a given site, as well as quantifying the competitive strength, achievable prices/​
rents, absorption rates, and income-​earning capacity of such a development. In pursuing these objectives,
the analyst needs to carry out both qualitative and quantitative analyses.

• Qualitative analysis focuses on site and location attributes, as well as local housing market conditions.
Site and location analysis typically focus on physical, neighborhood, and community characteristics, as
well as access and linkages. Analysis of the local market focuses on housing consumer characteristics,
housing product information, and local market conditions. Information on all these factors is thor-
oughly examined and evaluated in terms of its implications for product design and consumer targeting.
• Quantitative analysis at the project-​specific level focuses primarily on estimating project price/​rent and
absorption schedule using competitive differential techniques or hedonic valuation techniques:
• Competitive differential techniques are used to evaluate the strength of the project vis à vis the
competition and the price or rent it should command given its characteristics. The major problem
of competitive differential techniques is that they either do not assign weights to the different pro-
ject and location attributes (in which case they fail to account for their differential effect on project
value), or if they do assign such weights, these are not based on any rigorous statistical analysis, but
rather on experience and judgment.
• Hedonic valuation techniques provide a more accurate methodology for assessing a specific project’s
price given its attributes, because the effect of each attribute on market price is estimated through
sound statistical techniques. Hedonic price analysis can be very useful in calculating the market value
of alternative amenity packages and density designs.These estimates can in turn provide the basis for
identifying the bundle of development and product attributes that is more likely to maximize project
profitability and residual land value. Forecasts of project prices or rents can be produced by applying
econometrically derived market price or rent growth rates (from the macroeconomic analysis) to the
estimated hedonic price or rent for the project at the time of analysis.
• Competitive differential techniques can set the stage for calculating the project’s competitive pos-
ition index, which can be used in combination with forecasts of effective demand and competing
supply to estimate project absorption upon its completion. Project absorption estimates can be used
along with project price or rent forecasts to estimate expected gross income. The latter is one of the
most critical inputs in determining the financial feasibility and viability of the residential develop-
ment under consideration.

Questions
1. Discuss the major objectives of the microeconomic analysis for residential projects and the different
methodologies that a market analyst can use to attain those objectives.
2. Discuss the three sets of factors a thorough site and location analysis should focus on.
3. What are the factors of interest when analyzing the project’s local market?
4. Describe the usefulness and major steps of applying the competitive differentials technique in residen-
tial project analysis.
5. Discuss the formulas for constructing amenity indices for the different types of attributes that describe
a specific residential development.
6. What is the major criticism of the competitive differential technique and how can the analyst over-
come its shortcomings?

215
 Analyzing residential real estate markets

7. Discuss the major steps in applying hedonic valuation techniques to residential project analysis.
8. Discuss alternative formulations of hedonic price equations and the underlying relationships between
price and project attributes implied by each formulation. What is the interpretation of the estimated
coefficients in each case?
9. How can hedonic valuation techniques help optimize residential project design?
10. Describe how one should go about estimating optimal density for a residential project?
11. What is the problem in using absorption rates of competitive projects at the time of analysis as the basis
for forecasting project absorption at the anticipated time of its completion?
12. What is the formula for estimating project absorption and what potential inconsistencies may need to
be resolved before using it?

Project #4: On hedonic price models

For the completion of this project, the instructor needs to provide the students with a set of data on
housing units sold, including information on sales prices, structure characteristics, and location attributes.
An effort needs to be made to include in the dataset critical access characteristics and community attributes
that may have an impact on house prices. The latter set of factors needs to be accounted only if the sample
includes transactions associated with properties located within different communities.

Analysis tasks/​questions
1. Using the sales data provided, estimate a hedonic price model. Be sure to:
a) Explain what functional form you have assumed.
b) Comment on the explanatory power of the model.
c) Comment on the signs and significance of the model’s explanatory variables.

2. Explain the usefulness of such a model for market analysis purposes and provide examples:
a) Estimate the market price of the base unit (with features as provided by the instructor).
b) Using the base unit, determine marginal amenity values for the second and third bedroom or
bathroom, as well as proximity to a desirable residential amenity, such as a park, school, or the
coastline.
c) Graph and comment on the price profile of such a unit at different distances from a destination,
such as a school, or downtown, or the ocean, which, according to the results of hedonic price
analysis, is valued by housing consumers.

Write a two-​to three-​page report incorporating the answers to the questions asked. Make sure you attach
all of your analytical results.

Notes
1 Fischel (1990) indicates that growth controls include tightening of traditional zoning laws, as well as moratoria on
the extension of water and sewer lines and development permit issuance conditional on the provision of new public
facilities. FAR (Floor Area Ratio) regulations impose a maximum on the ratio of the total square footage of the
building over the lot’s total area. Minimum lot requirements impose a minimum lot size for subdivisions involving
residential land.
2 According to Rabianski (2006, p. 37), the project’s market area is defined as the geographic area from which the
subject property can draw tenants or buyers and should be distinguished from its trade area, which according to the
same author is the more narrowly defined geographic area “in which the subject property competes for tenants with
its direct and most proximate competition.”

216
 Analyzing residential projects

3 We elaborate on this in a subsequent section.

4 The analyst will not have any market rent/​price information for projects under construction (with the exception of
off plan sales) or in the planning stage.
5 Equation (9.10) is derived from (9.9) based on the three mathematical principles below:
lnAb = b lnA, lnAB = lnA + lnB and lne =1
6 DiPasquale and Wheaton (1996) estimate a log-​log hedonic model in which dummy variables are represented using
values of either 1 or 2 instead of 0 and 1, respectively, so they can be introduced in a logarithmic form.
7 Formula (9.14) can be used to estimate the marginal benefit from increasing the quantity of any attribute from A
to A+k, independently of whether that attribute is introduced linearly or in logarithmic form in the hedonic price
equation. Alternative formulas can be used for the calculation of this marginal benefit, but those do vary depending
on whether the attribute is introduced linearly or in logarithmic form and, in the former case, whether the attribute
is expressed as a continuous or a dummy variable. In order to articulate these formulas, we assume that the numerical
nature of the attributes included in the hedonic price equation are such that its underlying non-​linear specification
is described by (9.13), which is repeated here for the ease of discussion that follows.

P = e a0 + a1D1 + a2 L1 X 1a3 X 2a4 ....X nan (9.13)

where
D1 : project attribute expressed as a dummy variable
L1 : project attribute that can be expressed as a continuous or integer variable, but takes zero and/​or negative values
Xi : project attributes that can be introduced in logarithmic form
Within this context the formula for calculating the marginal benefit for an attribute introduced as a dummy
variable (represented by D1 in (9.12)) is

MBD = e a0 + a2 L1 X 1a3 X 2a4 ....X nan (e a1 − 1)

The formula for calculating the marginal benefit from increasing the quantity of L1 from A to A+k is:

MBL1 = e a0 + a1D1 + a2 A X 1a3 X 2a4 ....X nan (e a2k − 1)

Finally, the formula for calculating the marginal benefit from increasing the value of an attribute that can be
introduced in logarithmic form, such as X1, from A to A+k is:

MBX1 = e a0 + a1D1 + a2 L1 X 2a4 ....X nan [( A + k )a3 − Aa3 ]

8 A residential project’s absorption represents the number of its units sold or rented within a given period.
9 Historical data needed for applying the econometric approach are rarely available for a specific product type and
price range. Aggregate econometric forecasts referring to all housing types, however, may provide the basis for
sharing down demand for a specific housing type.
10 Multi-​collinearity may be another econometric issue associated with specifications (9.22) and (9.23), since pro-
ject price and amenities, appearing as explanatory variables in (9.22), and project absorption rates and amenities,
appearing as explanatory variables in (9.23), should in theory be highly correlated.
11 Two-​stage least squares is a commonly used estimation technique when simultaneity is present. See Chapter 11 in
Pindyck and Rubinfeld (1991), for a more detailed discussion of this issue.
12 FAR stands for floor area ratio.
13 DiPasquale and Wheaton (1996) define residual land value as the “residual profits to be obtained from land, after
construction costs.” It should be noted that the methodology and formula used here is a simplistic one and very
different than the residual valuation technique, which is based on a similar broader concept but is a much more
meticulous and sophisticated technique as it uses the discounted cash flow model.
14 DiPasquale and Wheaton (1996), assuming that P = α –​ βFAR and C= μ + τFAR, show that FAR* and RVL* can
be calculated as:

FAR * =
(α − µ ) and RLV * =
(α − µ ) FAR *
2 (β + τ ) 2

217
 Analyzing residential real estate markets

where
P : house price per square foot of floor area
α : the sum of the values of all other attributes included in the hedonic price equation (including the constant)
β : the coefficient of FAR in the hedonic price equation
C : construction cost per square foot of floor area
μ : the constant of the construction cost equation
τ : the incremental change in cost as FAR rises

References and additional readings

Anderson, S. T. and S. E. West. 2006. Open Space, Residential Property Values, and Spatial Context. Regional Science and
Urban Economics, 36:6, 773–​789.
Carroll, T., T. Clauretie and H. Neill. 1997. Effect of Foreclosure Status on Residential Selling Prices. Journal of Real
Estate Research, 13:1, 95–​102.
Clap, J. 1987. Evaluating the Competition with Regression Analysis. Chap.  11 in Handbook for Real Estate Market
Analysis. Englewood Cliffs, NJ: Prentice Hall.
Dasso, J. and A. Ring. 1989. Real Estate, Eleventh Edition. Englewood Cliffs, NJ: Prentice Hall.
DiPasquale, D. and W.Wheaton. 1996.The Urban Housing Market: Structural Attributes and Density. Chap. 4 in Urban
Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
Fischel, W. 1990. Do Growth Controls Matter? Cambridge, MA: Lincoln Institute of Land Policy.
Graevenitz, K. and T. E. Panduro. 2015. An Alternative to the Standard Spatial Econometric Approaches in Hedonic
House Price Models. Land Economics, 91:2, 386–​409.
Irwin, E. G. 2002. The Effects of Open Space on Residential Property Values. Land Economics, 78:4, 465–​480.
Jones, C., C. Leishman and C. Watkins. 2003a. Housing Market Processes, Urban Housing Submarkets and Planning
Policy. Town Planning Review, 76:2, 215–​233.
Jones, C, C. Leishman and C. Watkins. 2003b. Structural Change in a Local Housing Market. Environment and Planning
A, 35:7, 1315–​1326.
Kim, J. H., F. Pagliara and J. Preston. 2005. The Intention to Move and Residential Location Choice Behaviour. Urban
Studies, 42:9, 1621–​1636.
Parli, R. 2008. Apartment Market Analysis. Appraisal Journal, Winter,  60–​72.
Peiser, R. and D. Hamilton. 2012. Professional Real Estate Development:  The ULI Guide to the Business. Washington,
DC: Independent Publishers Group.
Pindyck, R. and D. Rubinfeld. 1991. Econometric Models and Economic Forecasts. New York: McGraw-​Hill.
Rabianski, J. 2006. Apartment Market Area Delineation. Appraisal Journal, 74:1, 33–​41.
Smith, B. and D. Olaru. 2013. Lifecycle Stages and Residential Location Choice in the Presence of Latent Preference
Heterogeneity. Environment and Planning A, 45:10, 2495–​2514.
Watkins, G. 2008. Microeconomic Perspectives on the Structure and Operation of Local Housing Markets. Housing
Studies, 23:2, 163–​177.
Webb, J. 1982. Valuation of Multifamily Residential Properties. In Sirmans F. (ed.), Research in Real Estate, 2, 159–​183.
Greenwich, CT: JAI Press.
Zietz, J., E. N. Zietz and G. S. Sirmans. 2008. Determinants of House Prices: A Quantile Regression Approach. Journal
of Real Estate Finance and Economics, 37:4, 317–​333.

Appendix 9A: Determining optimal development density

Hedonic price analysis can help determine not only the optimal bundle of project and unit attributes but also optimal
development densities. However, this is feasible only if project density—​in the form of units per area of land or square
feet of floor area per square feet of land (FAR)—​is included among the explanatory variables of the estimated hedonic
rent/​price equations.12 The results of such equations can help estimate a housing project’s market rent/​price not only
for different amenity packages, but also for different development densities. These price estimates can then be used in
combination with development cost estimates to calculate the project’s residual land value (RLV) under alternative
development densities and housing unit designs.13 With such estimates at hand, the analyst can identify the density and
product design that maximizes RLV.
The RLV of a given development plan can be calculated using two alternative formulas:

RLV = [P(D) –​ C(D)]*D or RLV = [P(FAR) –​ C(FAR)]*FAR (9A.1)

218
 Analyzing residential projects

RLV, P,C
RLV = (P(D) - C(D))*D
where:
RLV : Residual land value ($/area of land)
max RLV
P(D) : Unit price ($/housing unit)
C(D) : Non-land cost ($/housing unit) RLV
D : Development density
(housing units per area of land)

C(FAR)
RLV = (P(FAR) - C(FAR))*FAR
where: P(FAR)
RLV : Residual land value ($/square foot of land)
P(FAR) : Unit price ($/square foot of floor area)
C(FAR) : Non-land cost ($/square foot of floor area)
FAR : Development density (square feet of floor
area/square feet of land)

1 FAR* 1 FAR
Optimal development density is FAR*

Figure 9.A1  Determining residual land prices for different amenity packages

Source: Modified from DiPasquale and Wheaton (1996).

where:
P(D) : unit price ($/​housing unit) for a given density level
C(D) : unit non-​land development cost ($/​housing unit) for a given density level
D : development density (units/​acre)
P(FAR) : unit price ($/​square foot of floor area) for a given FAR level
C(FAR) : unit non-​land development cost ($/​square foot of floor area) for a given FAR level
FAR : square feet of floor area/​square feet of land

Consider the following numerical application of (9A.1). Assume that, based on a hedonic price equation that includes
density (expressed in units/​acre) as one of the explanatory variables, unit prices are estimated at P=$150,000 for a pro-
ject density of five units/​acre. Let’s also assume that non-​land development costs are estimated at $120,000/​unit. In such
a case then the RVL for the project will be $150,000, as indicated by the calculation below:

RLV = (150,000 –​ 120,000) * 5/​acre = 150,000/​acre

As (9A.1) indicates, optimal density analysis requires not only an estimate of a hedonic price equation that includes
density as an explanatory variable, but also an estimate of a residential construction cost equation that also includes density as
an explanatory variable. The parameters (coefficients) of such an equation can be estimated either by assimilating data
on unit construction costs and attributes, including project density, for a sufficient number of residential developments,
or by consulting local housing builders and developers.
It should be noted that, even in the case of single-​family residential developments—​where alternative density ratios
could affect only the number of units built but not necessarily the design and construction cost of each unit per se—​the
per unit cost for the development as a whole may still vary because of potentially different landscaping requirements,
and potential economies or diseconomies of scale in construction.
According to DiPasquale and Wheaton (1996), RLV should be a concave function of FAR because of the way
price and cost move with FAR (see Figure 9.A1). In particular, all else being equal, the market price of a housing unit
should be a declining function of density (FAR), since lower FAR allows for more open space that should be valued by
consumers. Development costs, on the other hand, should rise with increasing FAR because of additional foundational,

219
 Analyzing residential real estate markets

structural, and, possibly, elevator requirements (DiPasquale and Wheaton, 1996). As a result of these dynamics, the dis-
crepancy between price and cost is the largest when the number of units and FAR are the smallest. As the number
of units and FAR increase, RLV increases too, but at some point, as the discrepancy between price and cost becomes
smaller and smaller, it reaches a maximum (RLV*) and then starts decreasing. RLV becomes zero when price and cost
become equal. The FAR that corresponds to the maximum RLV value represents the most profitable density for the
development (FAR*).
Maximization of the project’s RLV can be carried out by first calculating the price of a residential development
under alternative density configurations and then by using those price estimates in combination with corresponding
cost estimates for each development plan to calculate RLVs. The optimal project density and design can be identi-
fied as the one that produces the maximum RLV (see Figure 9.A1). DiPasquale and Wheaton (1996), assuming linear
relationships between FAR and both house prices and construction costs, derived the mathematical expressions for
calculating FAR* and RLV*.14

220
 10
ANALYSIS OF RESIDENTIAL
REAL ESTATE MARKETS
An example

Introduction 222
Residential development: an example 222
The project and relevant market and financial data 222
Time of entry 222
Residential acreage and zoning 222
Target markets 223
Target group demographics (1998) 223
Distribution of households across income groups (1998) 223
Residential product design 223
Hedonic price analysis 223
Non-land development costs 224
The design type choice problem 225
Residential finance (permanent loan) 226
Residential preferences 226
Age Group 1 226
Age Group 2 226
Residential market demand, competing supply, and prices 227
Effective demand 227
Competing supply 227
Real market price index 227
Analysis 228
Determining optimal design and densities 228
Selecting the most profitable design type for each of the two target groups 228
Determining the optimal density for each target group 229
Determining how many units of each design type to include in the project 233
Estimating project absorption and revenue schedules 235
Estimate the nominal market price index 236
Determine the schedule of nominal house prices for the two design types to be
included in the project 236

221
newgenprepdf

Analyzing residential real estate markets

Forecast effective housing market demand 237

Estimate project absorption schedule 239
Estimate project revenue schedule 239
Evaluating residential market studies 239
Macroeconomic analysis of residential markets 239
Micro, site-specific analysis of residential development projects 240

Introduction
In this chapter, we present an example that demonstrates how the microeconomic analysis of a residen-
tial project (based on hedonic valuation techniques) and the macroeconomic analysis of the local housing
market can be synthesized and integrated to derive the information needed to estimate project price,
absorption, and revenue schedules. A short review of the basic evaluation criteria for residential market
studies is also presented.

Residential development: an example

The example presented below represents an adaptation of the case originally prepared for the Market
Analysis section of the 1996 Comprehensive Exam taken by the students in the Master in Real Estate
Development (MRED) Program at the School of Policy, Planning, and Development at the University of
Southern California. All analysis data provided in this example are hypothetical.
It is emphasized at the outset that the major objectives of this example are simply to demonstrate how
hedonic price analysis can be utilized for optimizing project design, and how microeconomic analysis can be
integrated with macroeconomic analysis in order to assess the project’s income-​earning capacity. As such, this
example by no means describes the complete market analysis process for a residential development project.
The presentation of this example is structured as follows. The first section presents crucial information
about the project, as well as relevant market and financial data. The second section presents the numerical
analysis in two stages. The first stage focuses on the determination of optimal project design and density,
while the second focuses on forecasting project price, absorption, and revenue schedules at the anticipated
time of completion and beyond.

The project and relevant market and financial data

The project is conceived as a development of for-​sale housing on 8.35 acres of residential land. Given
available data, market analysis needs to be conducted not only to help determine what to build (residential
design types and their densities), but also to help develop sensible sales and revenue schedules.

Time of entry
The earliest possible time of entry is year-​end 1997 or, to put it differently, the beginning of the year 1998.
Entering the market at that point will allow the development to compete for a share of 1998’s effective
residential market demand.

Residential acreage and zoning

Total residential acreage is 8.35 (acres).The zoning ordinance allows a maximum average density of 9.6 resi-
dential units per acre. As long as this average density limit for the whole development is maintained, this
restriction does not prevent the development of different residential types at their optimal densities.

222
 Example analysis of residential markets

Target markets
Preliminary analysis of residential development prospects in the area indicates that the appropriate target
market encompasses households with expected 1998 incomes of $50,000 or more. Furthermore, two spe-
cific age cohorts have been identified as the major target groups for the development under consideration:

• Age Group 1, including households in the 35–​44 age category.

• Age Group 2, including households in the 45–​59 age category.

Target group demographics (1998)

The expected household growth within the site’s market area in 1998 is provided in Table 10.1.

Distribution of households across income groups (1998)

The simple matrix in Table 10.2 indicates how new households in the target groups are expected to be
distributed across more detailed income categories in 1998.

Residential product design
Four potential residential design types of average competitive position are being considered. The analyst needs
to determine the most profitable design type for each of the target groups.The differentiated features of these
design types (described in Table 10.3) include the number of bedrooms and bathrooms, as well as the size
category of living space (including living room and kitchen area), represented by dummy variables Living
1, Living 2, Living 3, and Living 4. The analyst’s objective is to determine which package(s) of attributes
will maximize the developer’s profit.

Hedonic price analysis

The hedonic price equation results in Table 10.4 have been derived using detailed information on 400
residential sales transactions that took place in 1996 and involve 200 buyers from each of the two target age

Table 10.1 Target group demographics

Age group Expected household growth, 1998

Age Group 1 4,500
Age Group 2 7,000

Table 10.2  Income distribution

Age group Income categories

$50,000–​$59,999 $60,000–​$74,999 $75,000–​$120,000
Age Group 1 55% 30% 15%
Age Group 2 40% 35% 20%

223
 Analyzing residential real estate markets

Table 10.3  Alternative design types

Type 1(Base case) 1 Type 2 Type 3 Type 4

2 bedrooms 3 bedrooms 4 bedrooms 5 bedrooms
2 bathrooms 2.5 bathrooms 3 bathrooms 3.5 bathrooms
Living 1=1 Living 2=1 Living 3=1 Living 4=1

Marginal
cost 2 $17,541 $14,629 $12,500

1
Other attribute values of the base case are presented in Table 10.6.
2
The marginal cost of $17,541 represents the additional cost of providing a Type 2 instead
of a Type 1 unit. Similarly, the additional cost of providing a Type 3 instead of a Type 2
unit is $14,629, while the extra cost of providing a Type 4 instead of a Type 3 unit is
$12,500.

Table 10.4  Hedonic analysis results, 1996

(Dependent and non-​dummy independent variables are in logarithmic form)

Variables Age Group 1 Age Group 2

Coefficient t-​stat Coefficient t-​stat
Constant 10.8900 2.68 10.9171 3.15
Parking spaces 0.3207 3.16 0.3300 2.79
Development density –​0.2918 –​4.15 –​0.3309 –​6.01
Structure quality {1,0} 0.3408 5.19 0.3408 7.32
Living 1 {1,0} 0.0400 4.47 0.0450 5.43
Living 2 {1,0} 0.0800 4.31 0.0830 4.98
Living 3 {1,0} 0.0900 3.98 0.1030 3.65
Living 4 {1,0} 0.0950 2.78 0.1020 4.95
Bedrooms 0.1800 6.19 0.1900 3.61
Bathrooms 0.1000 3.89 0.1020 2.78
Fireplace {1,0} 0.0570 3.95 0.0300 4.25
School proximity {1,0} 0.3508 5.97 0.3604 4.89
Adjacent open space {1,0} 0.3405 4.46 0.3602 5.01

R-​squared 0.89
Adjusted R-​squared 0.88

groups. The definition of the independent variables included in this equation and the relevant attributes of
the base unit (Type 1) are provided in Tables 10.5 and 10.6, respectively.

Non-​land development costs
Estimates of non-​land development costs for each design with and without fireplace are provided in
Table 10.7.

224
 Example analysis of residential markets

Table 10.5  Hedonic equation variable explanations, continued

Living 1 {1,0} Dummy; for a Type 1 unit, Living 1=1; else =0

Living 2 {1,0} Dummy; for a Type 2 unit, Living 2=1; else =0
Living 3 {1,0} Dummy; for a Type 3 unit, Living 3=1; else =0
Living 4 {1,0} Dummy; for a Type 4 unit, Living 4=1; else=0
(Thus, for determining the price of a Type 1 unit, use Living 1=1, Living 2=0,
Living 3=0, Living 4=0. Similarly, for determining the price of a Type 2 unit,
use Living 1=0, Living 2=1, Living 3=0, Living 4=0)
Bedrooms, bathrooms Numeric and self-​explanatory. The value of these variables depends on the type
of unit under consideration (see design type description presented earlier).
Fireplace {1,0} Dummy taking the value of 1 if a fireplace is provided and the value of 0 otherwise.
The analyst needs to decide whether a fireplace should be provided. Note that in
the case of each design type, the cost of providing a fireplace is $1,200.
Proximity to :
Schools {1,0} Dummy taking the value of 1 if the site is located within reasonable distance
from quality schools and the value of 0 otherwise. In the case of the
residential development under consideration, this variable takes the value of 1.
Adjacent open/​g reen space {1,0} Dummy taking the value of 1 if a property is developed adjacent to open/​
green space. In the case of the residential development under consideration,
this variable takes the value of 1.

Table 10.6  Base unit (Type 1) attribute values

Parking spaces/​unit 2.5 per unit

Development density 9.6 units per acre
Structure quality 1 (very good)
Living 1 1
Living 2 0
Living 3 0
Living 4 0
Bedrooms 2
Bathrooms 2
Fireplace 0
School proximity 1 (high school within a reasonable distance)
Adjacent open/​g reen Space 1

The design type choice problem

Given the data provided, the design type choice problem has four dimensions:

a) Select the design type to be provided for each target group. To address this issue use the base density and
assume no fireplace. Is it worth, for example, providing Type 2 vs Type 1 for Age Group 1? Or Type 3
vs Type 2? Or Type 4 vs Type 3? (The same questions apply to Age Group 2).
b) Decide whether to include a fireplace in the chosen design type for each target group. Continue using base
density in analyzing this problem.
c) Determine the optimal density for the design type chosen for each target group.
d) Decide how many units of each type to provide, taking into account the acreage available and the allowable
average density.

225
 Analyzing residential real estate markets

Table 10.7  Non-​land development costs per unit, 19961

With fireplace Without fireplace2

Type 1 88,500 87,300
Type 2 106,041 104,841
Type 3 120,670 119,470
Type 4 133,170 131,970
1
Note that the above non-​land development costs are provided in 1996 dollars, as prices to be estimated using
the hedonic analysis results are. Non-​land development costs for any other year (if needed) should be assumed to
follow the same price index as house prices do.
2
The difference between non-​land development costs with and without fireplace reflects the marginal cost of
adding a fireplace in the unit ($1,200).

Table 10.8  Financial data

Annual prime mortgage rate 8.25%

Term of the permanent loan 360 months (30 years)
Loan-​to-​value ratio 0.95
Property tax rate 2.50%
Homeowner’s insurance 0.50%
Mortgage underwriting standard 35%

Residential finance (permanent loan)

Table 10.8 provides the prime mortgage rate for permanent residential loans and other relevant financial
data. Note that the information provided refers only to the year 1998.

Residential preferences
Age Group 1
Of those households in Age Group 1 qualifying to own the most profitable design type (as determined
using the hedonic analysis results pertaining to Age Group 1), only 48% prefer to do so. Some households
in Age Group 1 may also be able to afford to own the most profitable design type for Age Group 2, as
dictated by the hedonic analysis results. Consumer surveys have shown that only 10% of these households
prefer to own this particular design type.

Age Group 2
Of those households in Age Group 2 qualifying to own the most profitable design type (as determined
using the hedonic analysis results pertaining to Age Group 2), only 64% prefer to do so. Some households
in Age Group 2 may also be able to afford to own the most profitable design type for Age Group 1, as
dictated by the hedonic analysis results. Consumer surveys have shown that only 12% of these households
prefer to own this particular design type.

226
 Example analysis of residential markets

Residential market demand, competing supply, and prices

Effective demand
Effective market demand in 1998 needs to be determined from information already made available.
Effective market demand is expected to be growing annually by 5.75% during the period 1999–​2000 and
by 1.2% during the period 2001–​2004. Note that these demand growth figures apply to all design types.

Competing supply
The number of residential units expected to compete with the residential development under consid-
eration for a share of effective market demand is provided in Table 10.9. For example, if Type 1 units
are developed on the site, the relevant competing supply figure in 1998 is 2,800. If Type 3 units are
developed, then the relevant competing supply figure in 1998 is 2,350. Since one design type for each
target age group will most likely be developed, only two of the relevant competing supply series need
to be utilized in the analysis.

Real market price index

Econometric analysis of house price movements in the site’s broader housing market resulted in the
following equation for the real market price index (t-​statistics in parenthesis below each coefficient):

P(t) = 50 + 0.77 P(t–​1) –​ 67.0 USER (t) + 0.00011 INC (t) –​ 70.00 SHH (t)
 (3.14) (7.15)  (–​
5.89)     (5.78)     (–​
3.65)
R2 = 0.98

where:
P(t) : real price index at time t
USER(t) : user cost of homeownership at time t
INC(t) : household income at time t
SHH(t) : housing stock/​household at time t

Table 10.10 provides the information needed to forecast the market price index. Note that the 1996 market
index is assumed to be 100 and that general price inflation during the forecast period is expected to be 3%.

Table 10.9  Competing supply for different design types

Year Type 1 Type 2 Type 3 Type 4

1998 2,800 2,600 2,350 2,876
1999 1,750 1,980 1,709 2,325
2000 2,430 2,287 1,974 2,152
2001 2,570 2,395 2,067 1,831
2002 2,200 2,426 2,475 2,926
2003 2,100 2,138 2,181 2,567
2004 2,150 2,195 2,239 2,122

227
 Analyzing residential real estate markets

Table 10.10  Residential market variables

Year Nominal mortage rate INC SHH Marginal tax rate Expected inflation
1997 8.00% 30,000 0.35 28% 3%
1998 8.25% 30,750 0.342 28% 3%
1999 8.80% 31,519 0.347 28% 3%
2000 8.00% 32,307 0.36 28% 3%
2001 7.70% 33,114 0.38 28% 3%
2002 8.00% 33,942 0.35 28% 3%
2003 8.80% 34,719 0.34 28% 3%
2004 8.70% 35,670 0.34 28% 3%

Analysis
The analysis involves two major parts. The first part is clearly microeconomic in nature and focuses on the
use of hedonic price equations for the estimation of alternative project prices, which are then used as the
basis for determining the optimal product design and density. The second part involves the integration of
macroeconomic and microeconomic analysis for the estimation of project prices and absorption schedule,
which in turn provide the basis for the calculation of project revenues over the forecast period.

Determining optimal design and densities

This section demonstrates the methodological steps involved in applying hedonic price analysis for opti-
mizing residential product design and densities. It is emphasized that in reality the analyst will not rely
merely on hedonic analysis for optimizing design. Expressed consumer preferences regarding product
design features, identified through carefully targeted consumer surveys, may play a more important role in
determining the most marketable design features. Hedonic price analysis, however, can help quantify the
economic and financial implications of incorporating such features in the development, thereby providing
important insights for more financially robust decisions. In the final analysis, however, some attributes may
be included independently of what a marginal cost/​benefit analysis may suggest, simply because the targeted
consumer surveys indicate strongly that they are vital for the marketability of the project. The conclusions
of the two types of analysis may not be entirely consistent because hedonic analysis may describe a broader
consumer segment or market area, while targeted consumer surveys may focus on narrower consumer
segments and geographies deemed to be directly relevant to the project under consideration.
The analysis presented below involves the following steps:

1. Determine the most profitable design type to be provided for each target group and whether to include a
fireplace or not, using the base density in hedonic price calculations.
2 . Determine the optimal density for the design type chosen for each target group.
3. Given the optimal densities for each group, determine how many units of each design type to provide, taking into
account the acreage available and the allowable average density.

Selecting the most profitable design type for each of the two target groups
The most profitable design type for Group 1, which includes households in the 35–​44 age category, needs
to be determined on the basis of the group’s willingness to pay for design attributes and the cost of pro-
viding these attributes. Table 10.11 provides an example of the calculations of the prices that consumers
in Group 1 are willing to pay for the attribute package provided by Design Type 1 with and without a
fireplace.

228
 Example analysis of residential markets

Table 10.11  Estimating Target Group 1’s willingness to pay for alternative Type 1 designs

Design Type 1 variable1 Coefficient Attribute [2]‌*ln[3] or [2]*[3]1

[1]‌ [2]‌ [3]‌
Constant 10.8900 1 10.8900
Parking 0.3207 2.5 0.2939
Density -​0.2918 9.6 -​0.6599
Quality 0.3408 1 0.3408
Living 1 0.0400 1 0.0400
Living 2 0.0800 0 0.0000
Living 3 0.0900 0 0.0000
Living 4 0.0950 0 0.0000
Bedrooms 0.1800 2 0.1248
Bathrooms 0.1000 2 0.0693
Fireplace 0.0570 0 0.0000
School access 0.3508 1 0.3508
Open space 0.3405 1 0.3405
ln(Price) 11.79016626
Unit price $131,948
Type 1
Is it worth providing a fireplace? Yes
Unit price w/​o fireplace $131,948
Unit price w/​fireplace $139,688
Marginal value $7,740
Marginal cost $1,200

The formula [2]‌*ln[3] applies to non-​dummy variables, which are parking, density, bedrooms, and bathrooms. The
1

formula [2]*[3] applies to the remaining variables.

Notice that the unit price for Design Type 1 without fireplace has been calculated as the exponent of
11.790, which is the natural logarithm of the price estimate [ln(Price)] derived using the results of the
hedonic equation. The price of Design Type 1 with fireplace is derived in a similar manner, after taking into
account in the calculation of ln(Price) that the fireplace variable takes the value of 1. These estimates show
that the marginal value of adding a fireplace in the unit is clearly greater than its cost, suggesting that it
would be profitable to incorporate this feature in Design Type 1.
Similar calculations were carried out for the other three design types, as they pertain to Target Group 1.
The price estimates for the different design types are summarized and compared with respective unit
cost estimates in Table 10.12. Comparison of the marginal value estimates for the different design types
with their marginal costs reveals that Design Type 2 with fireplace allows for the highest profit per unit (i.e.,
159,926 –​106,041 = $53,885). In the case of Design Types 3 and 4, marginal cost exceeds marginal benefit,
signaling declining profit.
The same series of calculations were carried out for Target Group 2. The results of these calculations,
presented in Table  10.13, indicate that the most profitable attribute package for this group is the one
represented by Design Type 3 with fireplace.

Determining the optimal density for each target group

Once the most profitable design for each target group is determined, the next step is to identify the density
(units per acre) for each design type that will maximize residual land value. Notice that density is one of

229
 Analyzing residential real estate markets

Table 10.12  Design type analysis at base density for Target Group 1

Price with Price without Unit cost with Unit cost without
fireplace fireplace fireplace fireplace
Type 1 139,688 131,948 88,500 87,300
Type 2 159,926 151,065 106,041 104,841
Type 3 173,248 163,649 120,670 119,470
Type 4 184,068 173,869 133,170 131,970

Marginal value Marginal value Marginal cost Marginal cost

with fireplace without fireplace with fireplace without fireplace
Type 2–​Type 1 20,238 19,116 17,541 17,541
Type 3–​Type 2 13,322 12,584 14,629 14,629
Type 4–​Type 3 10,820 10,220 12,500 12,500
Develop Design Type 2
Price of units with fireplace at base density $159,926

Table 10.13  Design type analysis at base density for Target Group 2

Price with Price without Unit cost with Unit cost without
fireplace fireplace fireplace fireplace
Design Type 1 134,577 130,599 88,500 87,300
Design Type 2 154,460 149,895 106,041 104,841
Design Type 3 169,557 164,546 120,670 119,470
Design Type 4 179,524 174,219 133,170 131,970

Marginal value Marginal value Marginal cost Marginal cost

with fireplace without fireplace with fireplace without fireplace
Type 2–​Type 1 19,882 19,295 17,541 17,541
Type 3–​Type 2 15,097 14,651 14,629 14,629
Type 4–​Type 3 9,967 9,673 12,500 12,500
Develop Design Type 3
Price of units with fireplace at base density $169,557

the attributes included in the hedonic equation and that the hedonic price estimates used in the preceding
analysis were made under the assumption of a base density.Thus, in order to determine the optimal density
at which each design type should be developed the analyst needs to:

1. Estimate the prices for different possible density values (from 1 to 15, for example), using the hedonic
equation results for each target group.
2. Derive estimates of residual land value ($/​acre) for each alternative density, using the price estimates
from (1) and formula (9.15).

Once these calculations are carried out, optimal development densities for each design type can be
determined by inspecting the results and identifying the density values that generate the maximum
Residual Land Value (RLV)1 estimate. Table 10.14 demonstrates an example of the hedonic price calcula-
tion for Design Type 2 (the most profitable design for Target Group 1) utilizing a density value of 12 units/​

230
 Example analysis of residential markets

Table 10.14  Example of an estimate of Design Type 2 market price

Variable Coefficient Attribute value [2]‌*ln[3] or [2]*[3]1

[1]‌ [2]‌ [3]‌
Constant 10.8900 1 10.8900
Parking 0.3207 2.5 0.2939
Density –​0.2918 12 –​0.7250
Quality 0.3408 1 0.3408
Living 1 0.0400 0 0.0000
Living 2 0.0800 1 0.0800
Living 3 0.0900 0 0.0000
Living 4 0.0950 0 0.0000
Bedrooms 0.1800 3 0.1978
Bathrooms 0.1000 2.5 0.0916
Fireplace 0.0570 1 0.0570
School access 0.3508 1 0.3508
Open space 0.3405 1 0.3405
ln(Price) 11.91736221
Unit price 149,846

The formula [2]‌*ln[3] applies to non-​dummy variables, which are parking, density, bedrooms, and bathrooms. The
1

formula [2]*[3] applies to the remaining variables.

Table 10.15  Identifying optimal development density for Type 2 units

Unit price, P Unit cost, C Units/​acre, D RLV ($/​acre)

(P–​C)*D
252,738 106,041 2 293,394
206,464 106,041 4 401,693
183,430 106,041 6 464,333
168,663 106,041 8 500,974
158,032 106,041 10 519,913
149,846 106,041 12 525,658
146,387 106,041 13 524,499
143,256 106,041 14 521,010
140,401 106,041 15 515,405
Conclusion: Design type Unit price Units/​acre
Target Group 1 Type 2 $149,846 12

acre. This value was chosen on purpose. As indicated in Table 10.15, which presents the price and RLV
calculations for different density values, this is the density value that maximizes RLV for Target Group 1.
Figure 10.1 demonstrates graphically how RLV varies with density and confirms the conclusion that
RLV peaks when development density reaches 12 units/​acre. It also demonstrates that development density
can make a big difference in the profit/​acre the investor can extract from the project.
Now that a preliminary determination of an optimal density has been made, we need to examine
whether the price that Target Group 1 is willing to pay for Design Type 2, within the optimal density
setting of 12 units per acre, still provides higher profit per unit than any other design type under the same

231
 Analyzing residential real estate markets

Residual land value per acre

Optimal density = 12 units/acre
600,000

500,000
Residual land value/acre

400,000

300,000

200,000

100,000

0
0 2 4 6 8 10 12 13 14 15
Units per acre

Figure 10.1  Density vs residual land value/​acre for Design Type 2 units

Table 10.16  Design type analysis at optimal density (Design Type 2 remains the optimal design type)

Price with Price without Unit cost with Unit cost

fireplace fireplace fireplace without fireplace
Design Type 1 130,884 123,632 88,500 87,300
Design Type 2 149,846 141,543 106,041 104,841
Design Type 3 162,328 153,335 120,670 119,470
Design Type 4 172,466 162,911 133,170 131,970
Marginal value Marginal value Marginal cost Marginal cost
with fireplace without fireplace with fireplace without fireplace
Type 2–​Type 1 18,962 17,912 17,541 17,541
Type 3–​Type 2 12,483 11,791 14,629 14,629
Type 4–​Type 3 10,138 9,576 12,500 12,500

development density assumptions. Remember that the original conclusion that Design Type 2 provides
the highest profit per unit was drawn by utilizing in the price calculations the base-​case density value of
9.6 units per acre. As indicated in Table 10.12, within that development density setting, Design Type 2 was
estimated to command a price of $159,926. As indicated in Tables 10.14 and 10.15, however, with the
higher density of 12 units per acre, this design type is expected to command a lower price ($149,846).Thus,
the calculations presented in Table 10.12 were recreated utilizing the optimal density value in the hedonic
price calculations. The results of these calculations, presented in Table 10.16, confirm that Design Type 2
still provides the highest profit per unit within the higher development density setting.
The same series of calculations was carried out for Target Group 2. The results of these calculations,
presented in Tables 10.17 through 10.18 and Figure 10.2, indicate that the optimal development density
setting for Target Group 2 is 8 units/​acre. The calculations presented in Table 10.19 confirm that Design
Type 3 still provides the highest profit per unit within this optimal development density setting.

232
 Example analysis of residential markets

Table 10.17  Example of an estimate of Design Type 3 market price under alternative density settings

Variable Coefficient Attribute [2]‌*ln[3] or [2]*[3]1

[1]‌ [2]‌ [3]‌
Constant 10.917133 1 10.9171
Parking 0.3300 2.5 0.3024
Density –​0.3309 8 –​0.6881
Quality 0.3408 1 0.3408
Living 1 0.0450 0 0.0000
Living 2 0.0830 0 0.0000
Living 3 0.1030 1 0.1030
Living 4 0.1020 0 0.0000
Bedrooms 0.1900 4 0.2634
Bathrooms 0.1020 3 0.1121
Fireplace 0.0300 1 0.0300
School access 0.3604 1 0.3604
Open space 0.3602 1 0.3602
ln(Price) 12.10128
Unit price 180,102

The formula [2]‌*ln[3] applies to non-​dummy variables, which are parking, density, bedrooms and bathrooms. The
1

formula [2]*[3] applies to the remaining variables.

Table 10.18  Identifying optimal development density for Type 3 units

Unit price, P Unit cost, C Units/​acre, D RLV ($/​acre)

(P–​C)*D
284,931 120,670 2 328,521
226,531 120,670 4 423,445
198,089 120,670 6 464,512
180,102 120,670 8 $475,452
167,282 120,670 10 466,123
157,488 120,670 12 441,822
149,657 120,670 14 405,814
143,188 120,670 16 360,288
137,715 120,670 18 306,804
Conclusion: Design Type Unit price Units/​acre
Target Group 2 Type 3 $180,102 8

Determining how many units of each design type to include in the project
Once the optimal design types, densities, and unit prices for each target group are determined, then it has
to be decided how the total acreage that is available for the development of residential uses (i.e., 8.35 acres)
should be allocated to each design type so as to maximize project value, while satisfying the average density
constraint of 9.6 units/​acre or less. As Table 10.20 demonstrates, this can be done by estimating the total
value of all possible acreage allocations to the two design types that satisfy the condition that their sum is
8.35. For example, as indicated in the first line of this table, the analyst can start by assuming that all avail-
able residential acreage is devoted to the development of Design Type 2 units. Developing this acreage at

233
 Analyzing residential real estate markets

Residual land value per acre

Optimal density = 8 units/acre
500,000

450,000

400,000
Residual land value/acre
350,000

300,000

250,000

200,000

150,000

100,000

50,000

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Units/acre

Figure 10.2  Density vs residual land value/​acre for Design Type 3 units

Table 10.19  Design type analysis at optimal density for Target Group 2 (Design Type 3 remains the optimal
design type)

Price with Price without Unit cost with Unit cost without
fireplace fireplace fireplace fireplace
Design Type 1 142,946 138,721 88,500 87,300
Design Type 2 164,065 159,216 106,041 104,841
Design Type 3 180,102 174,779 120,670 119,470
Design Type 4 190,689 185,053 133,170 131,970
Marginal value Marginal value Marginal cost Marginal cost
with fireplace without fireplace with fireplace without fireplace
Type 2–​Type 1 21,120 20,495 17,541 17,541
Type 3–​Type 2 16,036 15,562 14,629 14,629
Type 4–​Type 3 10,587 10,274 12,500 12,500

the optimal density of 12 units/​acre would allow the construction of 100 units, resulting in a total RLV of
$4.38 million. Similar calculations can be carried out by:

1 . Gradually decreasing the number of Type 2 units.

2. Calculating the acreage that would be required for those units given a density of 12 units/​acre.
3. Calculating the number of Type 3 units that can be developed in the remaining acreage at optimal
density.
4. Summing up the RLVs of each section of the development to derive total project RLV.

As Table 10.20 indicates, the acreage allocations to Types 2 and 3, which result in a density of 9.6 units/​
acre or less and maximize total project RLV, are 3.33 and 5.02, respectively.These densities allow the devel-
opment of 40 units from each design type (80 units in total).

234
 Example analysis of residential markets

Table 10.20  Determining the optimal design type mix

Design Type 2 Design Type 3 Constraints

Number Density RLV/​acre Acres Number Desntiy Value/​acre Acres Total RLV Average density Total acres
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌ [6]‌ [7]‌ [8]‌ [9]‌ [10] [11]
100 12 525,658 8.33 0 8 475,452 0.02 4,380,482 11.98 8.35
99 12 525,658 8.25 0 8 475,452 0.10 4,336,678 11.86 8.35
98 12 525,658 8.17 1 8 475,452 0.18 4,352,304 11.86 8.35
97 12 525,658 8.08 2 8 475,452 0.27 4,367,931 11.86 8.35
96 12 525,658 8.00 2 8 475,452 0.35 4,324,126 11.74 8.35

44 12 525,658 3.67 37 8 475,452 4.68 4,126,379 9.70 8.35

43 12 525,658 3.58 38 8 475,452 4.77 4,142,006 9.70 8.35
42 12 525,658 3.50 38 8 475,452 4.85 4,098,201 9.58 8.35
41 12 525,658 3.42 39 8 475,452 4.93 4,113,828 9.58 8.35
40 12 525,658 3.33 40 8 475,452 5.02 4,129,455 9.58 8.35
39 12 525,658 3.25 40 8 475,452 5.10 4,085,650 9.46 8.35
38 12 525,658 3.17 41 8 475,452 5.18 4,101,277 9.46 8.35
37 12 525,658 3.08 42 8 475,452 5.27 4,116,903 9.46 8.35
36 12 525,658 3.00 42 8 475,452 5.35 4,073,099 9.34 8.35

3 12 525,658 0.25 64 8 475,452 8.10 3,935,033 8.02 8.35

2 12 525,658 0.17 65 8 475,452 8.18 3,950,660 8.02 8.35
1 12 525,658 0.08 66 8 475,452 8.27 3,966,287 8.02 8.35
0 12 525,658 0.00 66 8 475,452 8.35 3,922,482 7.90 8.35
Development program
Units Design Type 2 Design Type 3
80 40 40

Note: This development program needs to be tested through the derivation of absorption schedules and detailed
financial feasibility analysis

Estimating project absorption and revenue schedules

Once project design, density, and the number of units to be developed are finalized, the analyst can proceed
to the estimation of project absorption, which, in combination with forecasts of project price at the time of
completion and thereafter, can set the stage for estimating the project’s revenue schedule. Notice that at this
stage the analyst needs to take into account macroeconomic analysis elements, such as expected movements
in market prices and forecasts of market demand and supply. Given the data provided for this exercise, this
whole process involves the following steps:

1. Estimate the nominal market price index, using the forecasting equation and forecast values of the
exogenous variables provided.
2. Determine the schedule of nominal house prices for the two design types to be included in the project.
3. Determine effective market demand for the forecast period by:
a) Estimating the distribution of new households (in terms of levels) in the two target age groups
across the income-​group categories for which percentage distributions have been provided.
b) Estimating minimum qualifying incomes, given the forecast project prices at the time of market
entry and the lender’s underwriting criteria.

235
 Analyzing residential real estate markets

Table 10.21  Estimating the market price index

P(t) = 50 + 0.77 P(t-​1) –​67 USER (t) + 0.00011 INC (t) –​70 SHH (t)
Year Marginal Mortgage Expected User cost Household Stock/​ Real price Real Nominal Nominal
tax rate rate inflation (USER) income (INC) household index change change index
(SHH)
1996 28.0% 8.0% 3.0% NA NA NA 100.0 100.0
1997 28.0% 8.0% 3.0% 2.8% $30,000 0.350 104.0 4.0% 7.0% 107.0
1998 28.0% 8.3% 3.0% 2.9% $30,750 0.342 107.5 3.4% 6.4% 113.8
1999 28.0% 8.8% 3.0% 3.3% $31,519 0.347 109.7 2.1% 5.1% 119.6
2000 28.0% 8.0% 3.0% 2.8% $32,307 0.360 111.0 1.2% 4.2% 124.6
2001 28.0% 7.7% 3.0% 2.5% $33,114 0.380 110.8 –​0.2% 2.8% 128.1
2002 28.0% 8.0% 3.0% 2.8% $33,942 0.350 112.7 1.7% 4.7% 134.1
2003 28.0% 8.8% 3.0% 3.3% $34,719 0.340 114.6 1.7% 4.7% 140.4
2004 28.0% 8.7% 3.0% 3.3% $35,670 0.340 116.2 1.4% 4.4% 146.5

c) Estimating effective market demand for each type of design for 1998, based on minimum quali-
fying incomes for each type of unit and housing type preferences.
d) Using the provided demand growth rates, forecast effective demand for the period 1999–2004.
4. Determine project absorption on the basis of its fair market share adjusted by the project’s competitive
position index.
5 . Estimate project revenue schedules on the basis of the estimated absorption schedule and unit prices
for each design type included in the development.

Estimate the nominal market price index

The market price index, for which a forecasting equation has been provided, reflects expected market price
movements based on an econometric analysis of its historical behavior. Project prices should follow such
movements in percentage terms. Notice that the equation provided refers to real house prices, not nom-
inal. Thus, any estimates derived using this equation need to be adjusted for expected inflation in order to
generate a nominal house price index over the forecast period.
Table 10.21 presents both the real and nominal market price index estimates. The real price index was
calculated by applying the coefficients of the forecasting equation to each year’s forecasts of the exogenous
drivers, that is, income, user cost of capital, and stock/​per household. In each year’s calculation, the con-
stant and the previous year’s price index were also taken into account, as dictated by the equation that was
provided. According to the estimates presented in Table 10.21, real house prices in the project’s market area
are predicted to increase by about 16% from 1996 to 2004. Note that the nominal change was calculated by
using the simplistic formula of adding the inflation rate on the real change and not the more sophisticated
formula that actually compounds the two rates of change (real and inflation).

Determine the schedule of nominal house prices for the two design types to be
included in the project
The hedonic price estimates produced earlier for Design Types 2 and 3 refer to the year of analysis, which is
1996. The project, however, is expected to enter the market in 1998. Therefore, a sensible evaluation of the
project’s potential revenues requires forecasts of project prices for 1998 and thereafter. Such forecasts can be

236
 Example analysis of residential markets

Table 10.22  Generate forecasts of nominal prices for the project

Year Nominal market price index Forecast prices for design type Forecast prices for design type
2 three-​bedroom units 3 three-​bedroom units
1996 100.0 149,846 180,102
1997 107.0 160,261 192,621
1998 113.8 170,564 205,004
1999 119.6 179,192 215,374
2000 124.6 186,637 224,323
2001 128.1 191,915 230,666
2002 134.1 200,963 241,540
2003 140.4 210,312 252,777
2004 146.5 219,534 263,861

Table 10.23  Expected distribution of anticipated household growth across income groups

Percentage distribution Levels

Age Group 1 Age Group 2 Age Group 1 Age Group 2

$50,000–​$59,999 55% 40% 2,475 2,800
$60,000–​$74,999 30% 35% 1,350 2,450
$75,000–​$120,000 15% 20% 675 1,400
Total 100% 95% 4,500 6,650

produced by applying the anticipated percentage changes in the market price index to the 1996 hedonic
price estimates for each design type. These forecasts are presented in Table 10.22 and indicate that in 1998,
the anticipated year of project completion, Design Type 2 units are expected to command a market price
of $170,564, while Design Type 3 units are expected to command a price of $205,004.

Forecast effective housing market demand

Forecasting effective housing market demand, as it pertains to the project, requires the following four steps:

a) Estimate the income distribution of new households in the two target age groups using the percentage
distributions that have been provided.
Estimating the income distribution of new households is a necessary step in order to allow the ana-
lyst to eventually quantify how many of the new households within the different income categories
will be able to afford to buy the specific types of units included in the development. Given the data
provided, this is a simple task and involves multiplication of the predicted increase in households
in each age group with the percentages provided under each income group. The results of these
calculations are provided in Table 10.23.
b) Estimate minimum qualifying incomes, given the forecast project prices at the time of market entry
and the lenders’ underwriting criteria.
The calculation of minimum qualifying incomes for the two design types that are included in the
development is the second step in quantifying how many of the new households will be able to afford
to buy them. These calculations are presented in Table 10.24. Note that the calculations in this table
refer to market prices at the anticipated time of completion, which is 1998, as dictated by the forecasts

237
 Analyzing residential real estate markets

Table 10.24  Estimating minimum qualifying incomes

Three-​bedroom units Four-​bedroom units

House price, year of entry $170,564 $205,004

Annual interest rate, entry year 8.25% 8.25%

Term of loan 360 360
Loan-​to-​value ratio 0.95 0.95
Property tax rate 2.50% 2.50%
Homeowner insurance 0.50% 0.50%
Mortgage underwriting standard 35.00% 35.00%

Calculating minimum qualifying incomes

Mortgage constant (monthly) 0.0075 0.0075
Loan amount $162,036 $194,753
Annual mortgage payments $14,608 $17,557
Annual property tax payments 4,264 5,125
Annual insurance payments 852.8186063 1025.017887
PITI $19,725 $23,708
Qualifying gross annual incomes $56,356 $67,736

Table 10.25  Estimating effective market demand for each type of design

Income groups Qualifying households

Lower-​income High-​income Type 2 units Type 3 units

limit limit
Age Age Age Age
Group 1 Group 2 Group 1 Group 2
$50,000 $59,999 902 1,020 0 0
$60,000 $74,999 1,350 2,450 654 1,186
$75,000 $120,000 675 1,400 675 1,400
Total households qualifying 2,927 4,870 1,329 2,586
Preferences 0.48 0.12 0.10 0.64
Effective demand 1,405 584 133 1,655
Effective market demand (all groups) Type 2: 1,989 Type 3: 1,788

presented in Table 10.22. Since the nature of these calculations has been discussed in detail in the
previous chapter, there is no need for further discussion here. According to these calculations, the
minimum qualifying income for Type 2 units (with three bedrooms) is $56,356, while the minimum
qualifying income for Type 3 units (with four bedrooms) is $67,736.
c) Estimate effective market demand for each type of design for 1998, based on minimum qualifying
incomes and housing type preferences.
Given the estimated minimum qualifying incomes and the distribution of new households across
income groups, we can calculate how many households will be able to afford to buy each of the two
design types. Finally, by taking into account each group’s preferences for housing types, we can esti-
mate overall effective demand for each of the two design types. Bear in mind that these calculations
refer to the anticipated year of project entry, which is 1998. As indicated in Table 10.25, the anticipated
total effective demand is 1,989 for Design Type 2 and 1,788 for Design Type 3.

238
 Example analysis of residential markets

Table 10.26  Effective demand schedules: 1998–​2004

Year Design Type 2 Design Type 3

1998 1,989 1,788
1999 2,103 1,890
2000 2,223 1,998
2001 2,249 2,021
2002 2,275 2,045
2003 2,302 2,069
2004 2,329 2,093

d) Use the provided demand growth rates to forecast effective demand for the period 1999–​2004.
Since not all units included in the project may be absorbed during 1998, when the project will enter
the market, it is prudent to forecast effective market demand and supply for each type for several years
following 1998. According to the data provided for this exercise, effective demand for both types is
expected to be rising rapidly at the rate of 5.75% in 1999 and 2000, but slow down to 1.2% thereafter.
A market demand schedule for each unit type can therefore be developed by applying these growth
rates to the estimated 1998 levels of effective demand. These schedules are presented in Table 10.26.

Estimate project absorption schedule

Given the provided market supply schedules and the number of units available in the project, we can now
proceed to the calculation of the project’s fair market share, which can then be applied to the estimated
market demand schedule to calculate project absorption (see Table 10.27).2 Notice that there is no need to
adjust the project’s fair market share with its competitive position index since the latter has been assumed
to be equal to one. The estimates indicate that 27 units from each type will be absorbed in 1998 and the
remaining 13 in 1999.

Estimate project revenue schedule

Given the number of units from each design type expected to be sold during each year and the forecast
prices for those two years, we can now estimate expected sales revenue. The calculations (presented in
Table 10.27) suggest that, if the assumptions underlying the analysis presented hold, the developer/​investor
will receive about $15 million in sales proceeds from this residential project.

Evaluating residential market studies

In evaluating housing market studies, it is important to have in mind the particular components that make
up a good study, focusing on this property type. To this aim, this section highlights these components in the
form of specific questions that the analyst needs to ask in evaluating a market study for residential real estate.

Macroeconomic analysis of residential markets

• Is the competitive market area carefully delineated so that it encompasses all existing and expected com-
peting residential supply?
• Is expected residential demand analyzed across demographic groups (households stratified, for example,
by age and income)?

239
 Analyzing residential real estate markets

Table 10.27  Estimating project absorption and revenue schedules

Year Effective Competing Market price Unit prices Available Project’s fair Project Sales revenues
demand supply index units in market share absorption
project
1996 100.00 $149,846
1997 106.95 $160,261
1998 1,989 2,900 113.83 $170,564 40 1.4% 27 $4,605,220
1999 2,103 1,980 119.58 $179,192 13 0.7% 13 $2,329,500
2000 2,223 2,287 124.55 $186,637
2001 2,249 2,395 128.08 $191,915
2002 2,275 2,426 134.11 $200,963
Design Type 3: Four-​bedroom units
1996 100.00 $180,102
1997 106.95 $192,621
1998 1,788 2,600 113.83 $205,004 40 1.5% 27 $5,535,097
1999 1,890 1,709 119.58 $215,374 13 0.8% 13 $2,799,867
2000 1,998 1,974 124.55 $224,323
2001 2,021 2,067 128.08 $230,666
2002 2,045 2,475 134.11 $241,540

• Are anticipated changes in market prices/​rents and affordability carefully taken into account in
developing effective demand estimates?
• Do expected competing supply estimates take into account most likely changes in future house prices/​
rents?

Micro, site-​specific analysis of residential development projects

• Are the development’s target groups sensibly defined? Are location and project attributes compat-
ible with the preferences of these target groups? Are their preferences properly taken into account in
product design, in terms of both project amenities and development densities?
• Are project prices estimated through sound techniques? Do such prices reflect potential changes in
market demand and supply conditions?
• Does the project absorption schedule incorporate information on the competitive strength of the
proposed project, as well as future changes in market conditions?

Notes
1 The formula for the calculation is given in Chapter 9. It should be noted that the methodology and formula used
here is a simplistic one and very different than the residual valuation technique, which is based on a similar broader
concept but is a much more meticulous and sophisticated technique as it uses the discounted cash flow model.
2 For a more detailed explanation of the calculations involved in this step, see the discussion on project absorption
estimation in the previous chapter.

240
 PART D

Analyzing the market for retail space

newgenprepdf

11
RETAIL MARKETS AND RETAIL
MARKET STUDIES

Introduction 243
Idiosyncrasies of the market for retail space 244
Idiosyncratic macroeconomic aspects 244
Idiosyncratic microeconomic aspects 244
Retail market fundamentals 245
The demand for retail space 245
Components of retail space demand 245
Purchasing power 247
Spending patterns 249
The supply of retail space 253
Retail market analysis objectives 256
Overview: the three phases of a retail market study 256
Phase I: Analyzing entry 256
Phase II: Location analysis/evaluation 257
Phase III: Project evaluation 257
Chapter summary 259
Questions 260
References and additional readings 261

Introduction
In the last decade, the retail market and the way consumers are shopping has undergone a major transform-
ation due to the rapid growth of e-​commerce, which has been “stealing” sales and market share from the
traditional in-​store retailing. According to data provided by the U.S. Department of Commerce, the share
of total retail sales that is captured by the internet increased from 2.9% in 2006 to 10.2% in the first quarter
of 2019 (preliminary estimate). In some items like books and CDs the penetration of internet sales is very
high. The rapid growth of online sales at the expense of physical store sales is likely to lead to a reduc-
tion in per capita retail space needs (Henderson Research, 2013). However, the magnitude of the negative
effects of online retailing on demand for retail space in the long term is debatable, as recent trends point
to the complementarity of the role of the two sales channels (electronic and physical space), which has led

243
 Analyzing the market for retail space

many companies to seek their seamless integration (referred to as omnichanneling) in terms of all aspects
of the interface between the consumer and the retailer, such as available merchandise, pricing, promotions,
marketing, and delivery (Piotrowitz and Cuthbertson, 2014; Hortacsu and Syverson, 2015).
Within this context, from a sales point of view, there is an increasing realization of the distinction between
physical space and electronic space, and retail analysts that are failing to appropriately take into account
the sales lost to electronic retailing are likely to produce misleading estimates of a retail development’s sales
potential. Furthermore, the developments in logistics and multichannel retailing are blurring the distinc-
tion between manufacturing and retailing in cases in which the retailer demands that the manufacturer
ships an item directly to the customer, or when a manufacturer sets up its own shop or webstore and sells
directly to the customer. All these developments create additional uncertainty in estimating potential sales
for a particular retail project.
Despite the above challenges, market analysis for retail space remains one of the most interesting types
of market studies. A lot of applied research has been conducted in this area, and, in the author’s opinion, the
quality of this research is quite high. Probably, only macroeconomic analysis for office space can be favor-
ably compared to retail market research in terms of quality.
In this chapter, we first outline the idiosyncrasies of the market for retail space. Subsequently we elab-
orate on the fundamentals of retail markets, focusing on the basic determinants of the demand for and
supply of retail space. Finally, we discuss the major objectives and the basic components and analysis
elements of market studies for retail development, as a way of an introduction to the next chapter, which
elaborates on the different stages of retail market analysis.

Idiosyncrasies of the market for retail space

The market for retail space is characterized by a number of idiosyncrasies both at the macroeconomic and
microeconomic level:

Idiosyncratic macroeconomic aspects

• Unlike demand for residential units, demand for retail space is a derived demand stemming from the
demand for retail goods and services. It is driven by population/​household growth, disposable incomes,
and spending patterns. These factors remain the key determinants of the sales potential of a retail devel-
opment despite the rapid growth of e-​commerce, but retail analysts need to sort out as accurately as
possible what part of an area’s purchasing power is allocated to internet purchases and what part to in-​
store purchases for the different product categories.
• The idiosyncrasies of the retail market on the supply side lie in the great variety of retail space typolo-
gies that exist, as well as the dynamic nature of the sector with changing retailing formats and concepts,
as well as the rising competition from electronic retailing and warehousing/​manufacturing selling dir-
ectly to consumers.

Idiosyncratic microeconomic aspects

The idiosyncratic aspects of retail market analysis at the microeconomic level relate to location, tenant mix,
and project design.

• Location in terms of access to consumers plays a very critical role in determining store revenues, and,
therefore, it is a critical determinant of a retail project’s viability.
• Tenant mix in clustered developments is critically linked to shopping externalities stemming from the
co-​location of stores facilitating complementary and comparison shopping. Furthermore, leisure-​related

244
 Retail markets and retail market studies

services and activities, which are increasingly becoming a larger component of tenant mix, are crucial
in enhancing a shopping center’s attractiveness and competitive position.Within this context, the tenant
mix is an important determinant of a development’s sales potential, especially in light of the fierce com-
petition for retail sales from the internet. According to Evans (2014), in today’s e-​commerce environ-
ment, the tenant mix needs to be structured appropriately so as to transform the shopping center to a
destination of “meaningful lifestyle experiences” by including a considerable component of non-​retail
tenants and specifically food and beverages, leisure, entertainment, and services. The structure of retail
lease rates with a percentage component, which is tied to a store’s sales, provides a powerful incentive
for optimizing a shopping center’s tenant mix and maximizing sales potential.
• Project design, in terms of inter-​store relationships, as well as project amenities (including parking), are
thought to play an important role in determining a retail project’s consumer appeal, sales potential, and
profitability.

Retail market fundamentals

Before proceeding into a more detailed discussion of the major phases of retail market analysis, it is important
to review the fundamentals of demand for and supply of retail space within a metropolitan market or a
retail market area. The concepts discussed below provide a very useful framework for understanding the
different analytical components of retail market studies.

The demand for retail space

Components of retail space demand
The demand for retail space is a derived demand stemming from the demand for retail goods and services.
This is evidenced in the positive effect that an area’s retail sales have on retail space rents. Such positive
effect has been empirically verified in several metropolitan retail space markets in the U.S. by Hendershott
et al. (2013). Retail goods and services can be classified into four major categories:

1. Convenience goods, representing standardized goods bought frequently from the store most conveniently
located with respect to the consumer (food, drugs, etc.).
2. Shoppers’ goods, representing less standardized goods that are purchased less frequently and involve some
comparison shopping (furniture, clothing, etc.).
3. Personal services, representing services purchased usually from the store most conveniently located with
respect to the consumer (shoe repair, dry cleaning, etc.).
4. Specialized services, representing services that are bought less frequently and involve some comparison
shopping (insurance, travel, etc.).

The aggregate amount of retail space demanded in a market is in essence the sum of the space demanded
by each product line or tenant category. The retail space demanded by each product line is a function
of expected consumer expenditures (potential sales in dollars) on that product line and the space nor-
mally required per unit of sales, or alternatively, the “normal” amount of sales per square foot of retail
space.1 The key demand factor is the potential sales by product line, which depend on the number of
households in each income group, their disposable income, and the percentage of such income spent on
the specific line of trade considered. These relationships are described by formulas (11.1) and (11.2) in
Figure 11.1.
Obviously, consumer expenditures are at the heart of the retail space demand calculation. Therefore,
demand for additional retail space in a market will depend on metropolitan growth, as it relates to the factors

245
 Analyzing the market for retail space

Retail space demanded, SD

Simply:
SD = Σ E k (NS k ) = Σ E k / NE k ) (11.1) Retail space Potential sales
demanded =
where: by tenant Sales/sq. ft. (Norm)

SD : Retail space demanded across all product lines (tenant categories)

Ek : Expected consumer expenditures on product line k (potential sales)
NS k : Retail space requirement norm per dollar of sales in product line k
NE k : Sales per square foot norm for product line k
Simply:
Potential sales by product line
Consumer expenditure (potential sales) by product line, Ek =
Sum of potential sales by different household groups

Ek = Σ i HH i Y i e ik (11.2) WHERE:
Potential sales by a household group
where:
=
Ek : Aggregate expenditure on product line k Number of households in the group
*
HH i : Number of households in group i Disposable income
Yi : Disposable income per household in group i *
: Percent expenditure on product line k Percent income spent on the product line
ek

What drives consumer expenditure?

Number of consuming units Stage Spending patterns differ

(individuals or households) in life cycle: across pre-nesters, full-
Demographics nesters, and empty-nesters
Metro (age, sex
growth household size, composition)
Consuming unit buying power Income/occupation See Ingene,
(after-tax disposable income) Housing tenure status The Structural
Determinants
Credit conditions of Market Potential,
Consumer psychology/expectations Journal of Retailing,
Spending patterns Relative prices 60, pp. 37-63 (1984).

Figure 11.1  On the demand for retail space

that determine the aggregate level of consumer expenditures and their breakdown into different product
lines. These factors include (see Figure 11.1):

• The number of households

• Household buying power (after-​tax disposable income)
• Spending patterns, which depend on:
• Stage in life cycle
• Demographics (age, sex, household composition, size)
• Income/​occupation
• Housing tenure status

246
 Retail markets and retail market studies

• Consumer psychology and expectations

• Credit conditions
• Relative prices.

The number of an area’s households and their income determine the aggregate dollar amount that is
available for retail spending or what is often referred to as “purchasing power,” while spending patterns
determine how this dollar amount is distributed to the different retail/​product categories. Below we dis-
cuss these factors in more detail.

Purchasing power
The number of an area’s households and their income determine what many analysts refer to in broader
terms as an area’s purchasing power. For example, Carn et  al. (1988a) refer to an area’s total income—​
calculated as the product of the number of households and the mean household income, or as the product
of per capita income and population—​as a measure of purchasing power. 2
The after-​tax disposable income that is available for retail spending represents a more exact measure of
buying power and the demand for retail space. As Table 11.1 indicates, this can be calculated as the personal
income (including interest income and transfer payments, such as social security, public assistance, imputed
rents, and value of services without payments) net of savings, tax payments, non-​tax payments, and other
largely non-​retail expenditures, such as expenditures on housing, health, education, and public transpor-
tation. It should be noted that housing expenditures refer only to the rent or mortgage payment made by
a household, and do not include all other expenditures on housing-​related items, such as home improve-
ment, furniture, and so on. Expenditures on these items are classified by the Bureau of the Census under
the category of “Household Operation.” Notice also that tax payments include income, estate and gift
taxes, personal property taxes, and motor vehicle licenses, while non-​tax payments include passport fees,
fines and penalties, donations, tuition in public schools, and medical fees in public hospitals.
A most common approach in calculating the dollar amount available for retail spending in an area is by
multiplying an estimated total personal income figure for that area with the percentage of personal income
spent on retail purchases, or purchases in a specific line of trade. According to Carn et al. (1988b), this per-
centage can be calculated using data from three alternative sources: the Consumer Expenditure Survey, the
Census of Retail Trade, or a publication entitled Relative Importance of Components in the Consumer Price Indexes.
These sources are discussed in the next section that focuses on spending patterns.
When using expenditure shares to calculate an area’s retail spending on a particular product line, it is
very important to apply them to the same type of income as the one used for their calculation. For example,
the Consumer Expenditure Survey reports two types of income: average income before taxes and average
income after taxes. Therefore, expenditure shares based on data provided by this survey need to be applied
to one of these two income measures for the area under consideration, depending on which one of the two
was used by the analyst to calculate such shares. Furthermore, it is very important to reduce appropriately
the expenditure share used to account for the share spent on online purchases, as the share provided by both
the Consumer Expenditure Survey and the Census of Retail Trade represent all household expenditures on
both online and in-​store purchases. Failure to reduce this share by the online sales component will result
in overestimation of the sales potential of the retail development under consideration.
Depending on the level of disaggregation desired by the analyst, the three alternative approaches/​
formulas listed below can be used for calculating the purchasing power (PP) or sales potential SE within
a project’s trade area. The first and second approach allow the analyst to estimate purchasing power by
line of trade, which is usually desirable when carrying out a retail market study. Furthermore, the first
approach/​formula represents the full application of the disaggregate approach that takes into account the
composition of the area’s household in terms of both age and income. The second formula represents a

247
 Analyzing the market for retail space

Table 11.1  Measuring buying power available for retail spending

After-​tax disposable income available for retail expenditure

Personal income
minus Personal tax and non-​tax payments
equals Disposable income
minus Interest paid
minus Personal transfer payments to foreigners
minus Housing expenditure
minus Expenditure on selected services (health, education,
public transportation)
minus Savings
equals Disposable income net of expenditure on housing,
selected services, and savings

Note: The above disposable income figure proxies more closely the income available for retail spending.

partial application of the disaggregate approach as it accounts for the distribution of the area’s households
only across income groups. Since the segmentation of an area’s households across age and income groups
is non-​uniform, any disaggregation of households across these two dimensions, or either one of the two,
should lead to improved estimates compared to those produced by formula (11.5), which represents the
aggregate approach that ignores completely such segmentation. Note that the formulas below, as well as
formula (11.2), can be applied also using the number of individuals instead of the number of households
and income per capita instead of income per household.3

1. Sum up potential expenditures across each product line, k, for all income groups, i, and all age groups, j,
as in formula (11.3):

S E = ∑ ∑ HH ijYij eijk (1 − oijk ) (11.3)

i j

where
HHij : number of households in income group i and group j
Yij : average or (preferably) median income of households in income group i and group j
eijk : percentage of income spent on line of trade k by households in income group i and age group j
(including both purchases at stores and online)
oijk : percentage of purchases of line of trade k by households in income group i and age group j that
are completed online
2. Sum up potential expenditures across each product line, k, for all income groups, i (so here the age seg-
mentation is ignored):4

PP = ∑ HH iYieik (1 − oik ) (11.4)

ik

3. The percentage of income spent on all retail purchases (er) reduced by the percentage of those purchases
that takes place online (or) is applied to the area’s total income, calculated simply as the product of the
total number of households (HH) times the average or median household income (Y):

PP = HHYe r (1 − or ) (11.5)

248
 Retail markets and retail market studies

In the above formulas, the analyst needs to use a measure of an “average” household income either for
households within a given income group (Yi) or for all households within the project’s trade area (Y).
Carn et al. (1988b) point out that there are measurement issues in either case. In the case of the first two
approaches in which households are disaggregated into income groups, the only measure of “average”
income the analyst can use is the midpoint of the two levels that define the income range within each
group. This midpoint, however, may overestimate or underestimate purchasing power depending on the
distribution of households to more refined income categories within that group. A measurement issue is
also raised when using average income for all households within a market area. In particular, this measure
most likely overstates buying power because income distributions tend to be skewed by a small percentage
of very high-​income households. In such a case, the median household income will provide a more
accurate assessment of the average household’s buying power.5
Many analysts avoid these calculations altogether by purchasing estimates and forecasts of potential
sales from private data vendors. The data are often provided electronically and is incorporated within a
computer program that provides great flexibility to the user in terms of defining the trade area for which
estimates of potential sales are desired (census tracts, zip codes, etc.).
The challenge of a retail market study is not only figuring out an area’s purchasing power at the time
of analysis, but also assessing how that purchasing power may change over the project’s planning horizon,
which can be quite long for shopping centers depending on their size. Obviously, increases in an area’s
income and population will boost purchasing power and sales. Thus, the state and prospects of the metro-
politan economy are very crucial in determining increases or decreases in an area’s purchasing power. For
example, during economic slowdowns, many workers lose their jobs and experience a significant reduc-
tion of their incomes. On the contrary, when the economy is growing fast, incomes increase, as many
unemployed individuals are able to find jobs, and those that are permanently employed receive hefty
bonuses and salary increases. Metropolitan growth analysis can provide very useful insights in terms of the
types of income increases that may be expected in the area and the types of retail goods and services that
may be demanded (see discussion in Chapter 3).

Spending patterns
Spending patterns, that is, the distribution of consumer purchases across different expenditure categories,
represent a very crucial piece of information, because they can help the analyst determine the types and
respective volumes of retail goods that are likely to be demanded in an area given the income levels and
demographic characteristics of its households or population. Figure 11.2 describes the percentage alloca-
tion of disposable personal income to major expenditure categories, as reported in the 2012 U.S. Statistical
Abstract. As indicated in this figure, the expenditure categories with the highest shares are health (19.7%),
housing, utilities and fuels (19%), recreation (9%), and transportation (8.9%). The expenditure categories
with the lowest shares are transfer payments (1.5%), interest payments (1.7%), and education (2.3%).

THE ROLE OF INCOME AND DEMOGRAPHICS  Spending patterns are determined by a number of factors,
the most important of which are income level and demographics.These factors tend to change significantly
through a household’s stage in the life cycle. For example, young couples without children (pre-​nest
stage) tend to have lower incomes and smaller household size. As age increases, income tends to increase
as well. Furthermore, as couples decide to have children, household size increases and the age and sex
composition of the household changes (full-​nest stage). As children grow up and move out of the family
home, household size and composition change again (empty-​nest stage). These life-​cycle changes have
significant effects on a household’s consumption and spending patterns.
According to the Platt Retail Institute (2014), an important dimension of the new frontiers in retailing
is the changing structure of the consumer base. This new consumer structure is characterized by a basic

249
 Analyzing the market for retail space

Food and beverages

Clothing, footwear, and related services

5.8% 7.8%
1.7% 1.5% Housing, utilities, and fuels
3.4%
6.7% Furnishings, household equipment, and
maintenance
19.0% Health
6.0%
Transportation

2.3%
Communication

Recreation
9.0%
Education
4.2%
Food services and accommodations
2.3%
Other goods and services
8.9%
19.7% Interest payments

Transfer payments

Savings

Note: Figures do not add up to 100%

Source: Statistical Abstract of the United States (2012)

Figure 11.2  Percentage allocation of disposable personal income in the U.S. in 2010

dichotomy, consisting of the aging baby-​boomers with high buying power, and a new internet-​addicted
generation, born between the early 1980s and early 2000s, and referred to as generation Y or the Millennial
Generation, with considerably smaller buying power. The other major characteristic of the changing con-
sumer base is the increasing multi-​culturalism and diversity, manifested in an increasing number of diverse
consumer groups seeking specialization, customization, and niche-​interest products and services (Platt
Retail Institute, 2014).
A household’s income not only determines the dollar amount that is available for retail spending, but
also influences the types of goods and services purchased. For example, low-​income households spend
most, if not all, of their income on necessities and seek low-​end discounted goods and services. On the
contrary, wealthy households spend significant amounts on non-​necessities and lifestyle-​related items and
seek high-​end goods and services.
A household’s demographic characteristics also play an important role in determining what types of
goods and at what volumes are purchased. In particular, keeping income constant, a household’s volume
and types of purchases vary depending on its size, age, and other demographic characteristics. For example,
married couples with children spend a significant portion of their budget on child-​related expenditures,
while married couples with no children do not.
In a study of differences in expenditures per capita across 250 MSAs, Ingene (1984) verified the effect
of different life-​cycle stages on consumption patterns. For example, he found that the percentage of
households/​individuals in the full-​nest stage had the greatest positive impact on retail expenditures.6 Such
impact is consistent with theory that predicts large expenditures on food, clothing, and durable goods at
this stage of the life cycle. The percentage of households/​individuals in the pre-​nest stage was also found
to have a positive and statistically significant effect on expenditures, though somewhat smaller than the

250
 Retail markets and retail market studies

Table 11.2  Percentage spent on food away from home across age and income groups (2014–​2015)

Age group Less than $5,000 to $10,000 to $15,000 to $20,000 to $30,000 to $40,000 to $50,000 to $70,000
$5,000 $9,999 $14,999 $19,999 $29,999 $39,999 $49,9991 $69,999 and more
>25 8.3% 8.4% 8.0% 6.3% 6.5% 6.4% 5.5%
25–​34 7.7% 3.8% 4.2% 5.9% 5.8% 5.4% 5.8% 6.1% 5.9%
35–​44 4.4% 4.8% 5.1% 6.5% 4.6% 5.5% 5.0% 5.4% 5.9%
45–​54 4.6% 4.9% 3.0% 4.5% 5.1% 5.4% 5.0% 4.9% 5.4%
55–​64 5.0% 3.4% 3.8% 4.3% 3.9% 4.3% 4.2% 4.6% 5.0%
65 and 4.4% 4.7% 3.9% 4.1% 4.5% 4.4% 4.8% 4.6% 5.1%
older
1
  For the age group of less than 25 years old, this income group is $40,000 or more.
Source: Bureau of Labor Statistics, Consumer Expenditure Survey 2014–​2015.

Table 11.3  Percentage spent on furniture across age and income groups (2014–​2015)

Age group Less than $5,000 to $10,000 to $15,000 to $20,000 to $30,000 to $40,000 to $50,000 to $70,000
$5,000 $9,999 $14,999 $19,999 $29,999 $39,999 $49,9991 $69,999 and more
>25 1.2% 0.9% 0.8% 0.7% 1.0% 1.2% 1.2%
25–​34 0.8% 0.8% 0.8% 1.3% 1.0% 0.9% 0.9% 0.9% 0.9%
35–​44 0.4% 0.3% 0.6% 0.4% 0.7% 0.8% 0.8% 1.0% 0.9%
45–​54 0.6% 0.2% 0.3% 0.3% 0.9% 0.7% 0.6% 0.7% 0.8%
55–​64 0.6% 0.6% 0.4% 0.5% 0.4% 0.4% 0.6% 0.7% 0.9%
65 and 0.7% 0.3% 0.5% 0.7% 0.5% 0.6% 0.6% 0.7% 0.7%
older
1
  For the age group of less than 25 years old, this income group is $40,000 or more.
Source: Bureau of Labor Statistics, Consumer Expenditure Survey 2014–​2015.

full-​nest-​stage effect. This is also consistent with theory that predicts considerable purchases of durables,
furniture, automobiles, and clothing by individuals and households at this life-​cycle stage. Not surprisingly,
the percentage of households/​individuals in the post-​nest stage (referring to two life-​cycle stages, namely
empty-​nest and solitary survivor) was found to have a significant negative effect on expenditures per
household, and, specifically, as they pertain to department, furniture, and variety stores.
Tables 11.2 and 11.3 demonstrate the extent of variations in spending patterns across various age and
income groups for two particular types of expenditures, namely, food away from home and household
furnishings. As Table 11.2 shows, the share of expenditure spent on food away from home varies more
across age groups than across income groups. For example, the largest variation of this share across income
groups is 350 basis points in the age category of 25–​34 (between the percentage spent by the group with
income less than $5,000 and the percentage spent by the group with income $10,000–​$14,999).The largest
variation across age groups, however, is 500 basis points and can be found within the income group of less
than $5,000 (between the <25 years-​old group and the 45 to 54 years-​old group).
It is clear from Table  11.2 that the variation of the share of expenditure spent on food away from
home across age groups narrows drastically as household income increases above $30,000. It appears that
households headed by individuals younger than 25  years old with incomes of up to $9,999 spend the
most on food away from home relative to their total expenditures (8.3–​8.4%), while households that have
income ranging between $10,000 and $14,999 and are headed by individuals 45–​54 years old spend the
least relative to their total expenditures (3%).

251
 Analyzing the market for retail space

It should be emphasized that, although differences in the percentage of expenditures spent on food
away from home between low-​income and high-​income groups may not seem to be very large, the types
of goods and services purchased within this expenditure category by these income groups may be very
different. Furthermore, differences in dollar amounts spent across these income groups are much greater,
since these percentages apply to quite different income levels.
As Table 11.3 shows, the share of income spent on furniture varies also considerably across age and
income groups. For example, the lowest share of expenditures on furniture in 2014–​2015 was 0.2% (by
households headed by individuals 45–​54 years old with income $5,000–​$9,999) while the highest is more
than six times larger (1.3%) and it is spent by households headed by individuals 25–​34  years old with
income ranging between $15,000 and $19,999.
Given the importance of spending patterns in calculating purchasing power for different product lines,
it is worth elaborating on the different sources of such information. The data presented in Tables 11.2 and
11.3 are drawn from the Consumer Expenditure Survey carried out by the Bureau of Labor Statistics (BLS)
and published in annual reports (calendar-​year and mid-​year from July of one year until June of next
year) and biannual reports. The annual reports provide one-​dimensional tabulations of expenditures across
broader lines of trade by various household characteristics (such as age, income, education, occupation,
etc.) and by region. The biannual reports provide expenditure information across more refined household
groups through cross-​tabulations. One shortcoming of these data is that there is about a two-​year lag. For
example, as of June 2019 the latest biannual report was for 2016–​2017.
The most detailed breakdown of spending patterns across different product lines is provided by an
annual BLS publication entitled Relative Importance of Components in the Consumer Price Indexes. One limi-
tation of these data is that households are segmented across a single dimension, that is, either income, or
age, or regional location. The third source of data is the Economic Census, which is carried out every five
years and provides data on establishment sales for detailed SIC codes under the Retail Trade category at
the county and MSA level. These data can also be used to calculate the percentage of expenditure spent
on specific product lines, by dividing sales with aggregate personal income for the county or MSA under
consideration. The big advantage of this dataset is that it refers to in-​store sales, so using these data the
analyst can estimate more accurately the purchasing power or total sales in a particular line of trade and a
particular area, without having to hypothesize what percentage of total consumer expenditures is captured
by e-​commerce. However, no segmentation of sales across different population groups is provided.
In using the retail sales numbers reported by the Economic Census, it is extremely important to keep
in mind that those refer to sales at the point of purchase, and, therefore, may not reflect exclusively retail
expenditures of the residents of the area considered, because of potential exports of retail sales to residents
of neighboring areas. Thus, for example, applying the retail sales figure reported by the census for a small
county (in which a large portion of retail purchases may be made by residents of neighboring counties) to
the reported personal income for that county could lead to a very misleading estimate of the percentage of
income spent on in-​store retail purchases.
Besides the aforementioned household characteristics (size, age, and sex composition), educational level,
occupation, lifestyle preferences, and other psychographic characteristics may also influence spending patterns.
These factors may play an increasing role in the case of households with higher incomes that make consid-
erable purchases on non-​necessity goods and services. Information on an area’s psychographics is typically
collected through consumer surveys.
Finally, another household characteristic that may be an important determinant of the types of purchases
made by a household is housing tenure status, since homeowners tend to spend considerable dollar amounts
on a variety of housing-​related items, while renters do not. As another household characteristic, housing
tenure status seems to also correlate with life-​cycle stages, as young adults and young couples with no chil-
dren tend to be renters, while families with children are more likely to own the house they live in.7

252
 Retail markets and retail market studies

OTHER FACTORS AFFECTING RETAIL SPENDING PATTERNS  Besides income and demographics, retail
spending patterns may be influenced by some additional factors such as consumer psychology and expectations,
credit conditions, and relative prices. For example, a thriving economy shapes optimistic consumer expectations,
encouraging households to save less and spend more. Macroeconomic shocks that may raise fears (hopes)
of potential negative (positive) effects on the economy in the short or long term may also affect consumer
expectations and sentiment. For example, according to data presented by Abel and Bernanke (1995),
and drawn from the University of Michigan Survey of Consumers and the Survey of Current Business
(published by the U.S. Department of Commerce), both consumer confidence and real consumption
registered sharp declines in August 1990, when Iraq invaded Kuwait.8
Consumer spending patterns should be also influenced by the availability of consumer credit. Higher avail-
ability of consumer credit is usually reflected in lower interest rates, which stimulate higher household
borrowing either through home-​equity financing and re-​financing or unsecured loans. This increased
borrowing translates to higher income, which leads to increased spending.
Finally, sizable changes in relative prices may also affect the distribution of consumer expenditures across
the different retail categories. For example, steep increases in the prices of necessity goods will force many
households to devote a considerably higher share of their budget on these items and reduce the share spent
on non-​necessity goods and services.
In sum, retail spending patterns are influenced by income levels, demographics (and psychographics),
housing tenure status, consumer expectations, credit availability, and relative prices. Since many of these
factors may vary considerably both through time and across markets, spending patterns should vary as well
in a similar way.

The supply of retail space

Retail space construction is largely the result of supply decisions by retailers to open a new store and
speculative decisions by developers or investors to pursue a retail development project. The major ini-
tiative, however, for bringing up new retail space in the market rests with the investor and/​or devel-
oper. From the developer’s perspective, retail supply decisions are a function of factor costs (capital,
land, labor, and construction materials), retail property prices/​rents, and expectations regarding future
retail space demand. Retail construction should follow the fundamental law of supply, which suggests
that all else being equal, increasing retail property prices and/​or rental rates should lead to increased
deliveries of retail space (with significant time lag of course due to their large size). Similarly, increases
in labor costs or the cost of financing, while prices and rents remain constant, should lead to reduced
deliveries of retail space.
DiPasquale and Wheaton (1996) present data indicating that retail construction follows closely
movements in the national economy as reflected in movements in retail sales. This finding points to the
strong impact of demand cycles on the nation’s supply of retail space and, in some indirect way, to the sen-
sitivity of the latter to changes in retail space rents/​prices. Hendreshott et al. (2013) confirmed the effect of
retail rents and vacancy rates on retail space supply growth, as well as the slow adjustment process of retail
space construction.9
Retail developments are very heterogeneous and lumpy and tend to cluster in commercial zones or
shopping centers for reasons that will be discussed in the next chapter. Table 11.4 describes the different
typologies of clustered retail developments, as opposed to free-​ standing stores, as classified by the
International Council of Shopping Centers (ICSC). As this table shows, the major typologies of general-​
purpose shopping centers include strip/​convenience centers, neighborhood centers, community centers,
regional malls and super-​regional malls.The major typologies of specialized-​purpose centers include power
centers, lifestyle centers, factory outlet centers and theme/​festival centers (see Table 11.5).

253
 newgenrtpdf
Table 11.4  US shopping center classification: General purpose centers

Type Concept Typical GLA Number of Anchors % Anchor Typical Number Typical Type of Anchors Size of Trade
Range (Sq. Ft.) GLA of Tenants Area
Super-​ Regional Concept similar to that of 800,000+ 3+ 50–​70 NA Full-​line or junior 5–​25 miles
Mall regional malls but offering department store, mass
more variety of goods and merchant, discount
services department store and/​or
fashion apparel store
Regional Mall General merchandise or 400,000–​ 2+ 50–​70 40–​80 stores Full-​line or junior 5–​15 miles
fashion-​oriented offerings. 800,000 department store, mass

Analyzing the market for retail space

Typically enclosed with merchant, discount
inward-​facing stores department store and/​or
connected by a common fashion apparel store
walkway. The outside
perimeter is sourrounded
by parking
254

Community General merchandise or 125,000–​ 2+ 40–​60 15–​40 stores Supermarket, discount store, 3–​6 miles
Center convienence-​oriented 400,000 drugstore, large specialty
stores. Wider range of discount stores (toys, books,
apparel and other soft goods sporting goods, home
compared to neighborhood improvement,etc.)
centers. Usually configured
in a straight line as a strip
Neighborhood Convience-​oriented goods/​ 30,000–​125,000 1+ 30–​50 5–​20 stores Supermarket 3 miles
Center services
Strip/​Convinience Attached row of stores/​outlets <30,000 Not anchored NA NA Convenience store <1 mile
managed as single/​coherent or anchored
entity. On-​site parking by small
usually in front of the stores. convenience
May be configured in a store
straight line or have an L or
U shape
1
Modified from International Council of Shopping Centers (ICSC)
 Retail markets and retail market studies

Table 11.5  U.S. shopping center classification: Specialized-​purpose and limited-​purpose centers

Type Concept Typical Number of % Typical Typical type of Size of

GLA Range anchors Anchor number of anchors trade area
(Sq. Ft.) GLA tenants
Power Includes mostly 250,000–​ 3+ 70–​90 NA Category killers, 5–​10
center category-​dominant 600,000 such as home miles
anchors and a few improvement,
small tenants discount
department store,
warehouse club
and off-​price stores
Lifestyle Includes national chains 150,000–​ 0–​2 0–​50 NA Large-​format 8–​12
of specialty stores 500,000 upscale specialty miles
with dining and outlets
entertainment in
outdoor setting
Factory Includes manufacturers’ 50,000–​ NA NA NA Manufacturers’ and 25–​75
outlet and retailers’ outlet 400,000 retailers’ outlets miles
stores selling brand-​
name goods at
discount
Theme/​ Includes leisure, tourist, 80,000–​ Unspecified NA NA Restaurants and 25–​75
festival retail and service-​ 250,000 entertainment miles
oriented offerings
with entertainment as
a unifying theme
Limited-purpose centers
Airport A consolidated 75,000–​ NA NA NA No anchors, NA
retail collection of retail 300,000 mostly specialty
stores in a commercial retail stores and
airport restaurants

Retail center typologies are differentiated primarily by their size and more specifically by their total
square feet and their gross leasable area (GLA), their anchor and non-​anchor tenant mix, and the size of
the trade area they serve, measured usually as driving time or distance. For example, according to the ICSC
table, strip centers and neighborhood centers serve the smallest trade area, extending up to one mile from
their location, and, as such, they sell primarily necessity goods that are purchased in high frequency. Super-​
regional shopping malls, on the other hand, serve an area extending 5–​25 miles from their location and, as
such, they sell primarily non-​necessity goods that are purchased at low frequencies. Anchor tenants are key
attraction components for most of these clustered developments and differ in nature across the different
center typologies. For example, the typical anchor in a neighborhood center is a supermarket or a drug-
store, while typical anchors in regional shopping malls are full-​line or junior department stores. Factory
outlet centers do not have any anchors, while power centers have several “category killers” such as home
improvement, discount department, etc.
Although the above typologies are the ones most commonly found, retail center formats are often being
challenged, as retailers and developers seek strategies and shopping environment settings that will enable
them to increase market share. The structure of retail leases, which include a percentage of store sales,

255
 Analyzing the market for retail space

provides a significant inducement to developers/​investors for optimizing a center’s design and tenant mix
and venturing into new retail formats. For example, Schmitz and Brett (2001) indicate that shopping center
types are becoming less distinct as tenants typically found in regional malls are venturing into other center
formats and vice versa. Furthermore, new hybrid formats targeting specific market segments are emerging.
Finally, the rising competition from the internet is forcing shopping center developers to rethink tenant
mix in terms of shaping a “destination” offering a diverse experience with increased leisure/​entertainment
and service components (Evans, 2014).

Retail market analysis objectives

Well-​integrated retail market studies deal with the whole range of issues involved in a developer’s deci-
sion to provide retail space and the retailer’s decision to initiate operations at a particular location. As such,
retail market analysis needs to address questions related to market entry, retail site evaluation, formulation
of successful development scenarios, and the evaluation of the feasibility of such scenarios from the market
perspective. The overall goal of a market study for retail development is to assess the need for additional
retail space, help successfully design the retail development concept, and evaluate its prospects. To this aim
its underlying objectives are:

1. To determine the best development scenario in terms of both the most attractive package of pro-
ject attributes and the most productive tenant mix that will maximize the synergistic effects of the
co-​location of different tenants, thereby maximizing the development’s sales and rent-​generating
potential.
2. To assess the market viability of each and every retail tenant included in the planned development
by asserting that an adequate level of market demand (measured by expected sales volume) can be
attracted to the site.
3, To develop the schedule of expected rental revenues for the project.

Overview: the three phases of a retail market study

Given the underlying objectives of retail market studies, we can distinguish three major phases of analysis
(see Figure 11.3):

• Phase I focusing on the assessment of overall retail market strength through the estimation of the area’s
unrealized sales potential.
• Phase II focusing on site and location analysis.
• Phase III focusing on the refinement and evaluation of the viability of the particular retail development
project.

Phase I: Analyzing entry
Phase I, which usually starts where metropolitan growth analysis ends, focuses on questions of entry with
respect to metropolitan location selection and the particular type of center or store that can be developed.
Carried out most often for large developments and retail chains, this analysis involves the assessment of
market strength and the degree/​intensity of competition based on the estimation of the so-​called unrealized
sales potential for a specific metropolitan area and/​or a specific line of retail trade (apparel, food, auto, fur-
niture, etc.). The unrealized sales potential is defined as the potential of an area to capture additional sales
and can be evaluated by comparing actual retail sales to the volume of sales justified by an area’s income
and demographics. Estimates of unrealized sales potential should be very carefully interpreted. Investment

256
 Retail markets and retail market studies

strategies cannot be articulated solely on the basis of these estimates without the analysis of the potential
market share of the specific retail development under consideration.

Phase II: Location analysis/​evaluation

Given a site and a preliminary development scenario (designating the anchor tenant(s), non-​anchor tenants,
size, and mix as specifically as possible), the analysis in Phase II places emphasis on site and location evalu-
ation. More specifically, the analysis during this stage focuses first on the spatial behavior of consumers and
project location requirements, and then on the evaluation of site attributes against the identified location
requirements. The frequency of purchase and the degree of achievable synergies in terms of comparison
and/​or complementary shopping of the goods and services sold by the prospective tenants of the develop-
ment are some of the issues that need to be addressed in the first step of this phase. Proximity and access to
consumer segments (residential-​based, employment-​based, etc.), as well as macro and micro transportation
access, are the most important factors that need to be assessed in the second step of this phase.

Phase III: Project evaluation

Phase III represents an interactive five-​stage analysis process addressing the bottom-​line questions of a
retail market study. Will market demand for retail goods support the amount of square footage devoted to
each and every tenant included in the specific retail development scenario, or alternatively, are potential
trade-​area sales sufficient to cover the retailers’ full operating costs and allow for a reasonable profit? Project
design, tenant mix, capture rate, potential sales, and achievable rents are all issues that Phase III focuses on.
The specific stages involved in this phase of retail market analysis include:

• Refinement of preliminary development scenario

Refinement of the preliminary development scenario (aiming at achieving a rent-​maximizing tenant
mix) based on the insights gained from the estimates of unrealized sales potential by product line and
the analysis of the strengths and weaknesses of the site and location considered.
• Market/​trade area definition
Analysis/​definition of the project’s market or trade area, taking into account the type of development,
its location, the density and demographics of consumers, transport access, travel patterns, topographical,
psychological and socioeconomic barriers, and the location and strength of the competition.
• Competition analysis
The analysis in this stage focuses on competing facilities with respect to tenant mix, inter-​store
relationships, quality of merchandise, consumer market segments, center amenities, design features,
parking facilities, and measures of performance (sales per square foot, rents per square foot, rates of store
failure). Especially important is the identification of project attributes that contribute the most to per-
formance and the refinement of the development concept accordingly.
• Consumer shopping pattern analysis
This stage incudes a thorough analysis of consumer profiles, including buying power (income levels),
demographic characteristics, spending patterns, and shopping preferences. One of the major object-
ives of this stage is to assess as accurately as possible potential in-​store expenditures by line of trade
within the project’s trade area. That is why it is very important at this stage to appropriately account
for the share of consumer expenditures that is spent online as opposed to physical stores. In the case
of larger retail developments with overlapping trade areas, special emphasis is placed on consumer
shopping preferences and the extent to which such preferences relate to development attributes. Linking
preferences to project attributes will shed light on project features that are mostly valued by consumers
and help refine the development concept.

257
 Analyzing the market for retail space

ANALYSIS ANALYSIS ANALYSIS

PHASES OBJECTIVES COMPONENTS

PHASE I Preliminary analysis • Estimation of unrealized sales

of entry in terms of potential (the potential of an area
metro locations and/or to capture additional sales)
lines of retail trade

PRELIMINARY DEVELOPMENT SCENARIO

Size, tenant mix, attributes

PHASE II Location evaluation/ • Analysis of the spatial behavior of

selection of profitable consumers and location
retail locations requirements

• Evaluation of site attributes

against location requirements

PHASE III Refinement/evaluation of Stage 1: Refinement of development

development scenarios: scenario and tenant mix
to maximize rent

Stage 2: Trade area delineation

Trade area demand must Stage 3: Competition analysis

justify a given development
scenario and each and every tenant Stage 4: Shopping pattern analysis
it incorporates (in such a case,
potential trade area sales should be Stage 5: Evaluation
sufficient to at least fully cover the • Capture rate estimation
retailers’ costs). • Estimation of sales potential
The proposed tenant mix must be • Estimation of supportable space
the one that maximizes rent or the • Estimation of achievable
best that can be achieved. anchor and non-anchor rents

Figure 11.3  Overview of the main phases of retail market analysis

258
 Retail markets and retail market studies

• Evaluation of the retail development concept

The information collected in the previous four stages is used to re-​evaluate and further refine the retail
development concept. This more refined development scenario is then thoroughly evaluated through
the following analytical processes:
• Estimation of capture rates (which are especially important in the case of larger developments
with overlapping market areas), taking into account consumer characteristics, competing center
characteristics, and characteristics of shopping trips.
• Estimation of potential sales by tenant type.
• Estimation of supportable Gross Leasable Area (GLA) and Gross Buildable Area (GBA) by tenant
type using sales per square foot norms and efficiency ratios.
• Estimation of achievable anchor and non-​anchor rents through accounting techniques or hedonic
methodologies.

Information gained through each of these analysis stages can help refine the development scenario, espe-
cially with respect to trade areas, tenant mix, and project design features. This is why they are interactive
and should not be considered in isolation. In addition, it should be noted that the emphasis of each of these
steps may vary somewhat across the different types of retail developments, but the overall framework and
methodologies are generally applicable.

Chapter summary
In this chapter, we have discussed the idiosyncrasies of the market for retail space, the demand and supply
fundamentals of retail markets, and the major phases of a retail market study.
Idiosyncrasies of the market for retail space at the macroeconomic level include the determinants of
demand and the rising competition of in-​store retailing with electronic retailing, as well as the high degree
of heterogeneity and the dynamic nature of the supply of retail space. At the microeconomic level, location,
tenant mix, and project design remain the critical components in determining retail project success, even
more so in today’s increasingly competitive environment.
The major drivers of demand for retail space are the number of households residing in the area
under consideration, their incomes, and their spending patterns. A  household’s spending patterns are
primarily determined by its income level and demographic characteristics/​composition. Psychographic
characteristics, housing tenure status, consumer expectations, credit availability, and relative prices
may also affect spending patterns and eventually aggregated demand for space by retailers in different
product lines.
The supply of retail space does obey the fundamental law of supply and, as such, is driven by retail space
rents and prices, developer/​investor expectations, and development costs.The supply of retail space is highly
heterogeneous and manifests mostly in the form of clustered developments, differentiated by their size,
design, tenant/​store mix, and the size of the trade area they serve. The formats of these developments are
often challenged as retailers and developers seek strategies and shopping environment settings to increase
market share in a rapidly changing and increasingly competitive retailing environment.
Given the idiosyncrasies and fundamentals of retail markets, the three major phases of a retail market
study include:

1. Assessment of broader market strength and degree of competition by estimating the unrealized sales
potential either for all retail categories or by line of trade.
2. Retail site and location evaluation and/​or selection with emphasis on site attributes, the shopping
behavior of consumers, and location requirements.

259
 Analyzing the market for retail space

3. Evaluation of the specific development idea through a five-​stage interactive process that focuses
on maximizing project attractiveness and achievable rents and the extent to which market demand
supports the proposed tenant mix.

Questions
1. Discuss the idiosyncrasies of the market for retail space and the extent to which they are applicable to
the retail space market of your city.
2. Provide and explain the formula that can be used to most accurately estimate an area’s aggregate
demand for retail space, as well as demand by line of trade. Also discuss why this formula provides more
accurate estimates than the other two alternative formulas presented in this chapter.
3. Discuss the major determinants of consumer expenditures and spending patterns. How are consumer
expenditures on retail goods/​services related to the demand for retail space in a city?
4. Discuss the most appropriate measure of buying power for assessing an area’s demand for retail space.
5. What are the potential problems in using an “average” measure of household income for the calcula-
tion of aggregate demand for retail space?
6. Describe the various typologies of retail space and the basic attributes that differentiate them.
7. Discuss the three major phases of a retail market study, including the factors considered in each phase
and the five stages of the third phase.
8. Using available data on the number of households (or population), household income (or income per
capita), expenditures on different product lines, and sales per square foot, calculate the required square
feet of retail space for your city for two lines of trade. Present clearly and explain all your calculations
and any assumptions you may need to adopt.
9. Discuss four factors that affect the aggregate demand for renting retail space at the urban area level
(apart from changes in population/​household /​income) and explain how each of these factors should
change in order to cause an increase in the demand for renting retail space in your city. Make sure you
explain clearly how the proposed direction of change of each factor will boost the demand for renting
retail space in your city.

Notes
1 Data on sales per square foot by line of trade or different types of stores may be available from the International
Council of Shopping Centers (ICSC) of other industry associations.
2 When using purchasing power estimates from commercial vendors, the analyst needs to understand very well the
process by which the data are collected and compiled and then carefully evaluate whether the process introduces
biases that would tend to overstate or understate an area’s purchasing power.
3 Carn et al. (1988) suggest that the latter approach should be used for items, such as food and clothing, for which the
consuming unit is the individual.
4 If the percentage of income spent is not available by line of trade, k, then this formula can be applied at a less dis-
aggregate level by simply using the percentage of income spent by households in each income group on all retail
purchases (eir instead of eik).
5 The median of a distribution of values is the value that lies in the middle of the distribution when these values are
arrayed from the smallest to the largest. For example, a median household income of $55,000 indicates that 50% of
the households in the sample or population that was analyzed had income lower than $55,000.
6 The different age/​life-​cycle groups used in the econometric analysis were composite variables, generated through
a procedure called factor analysis and referred to as factors in statistical terminology. This procedure tends to group
together highly correlated variables. This analysis generated (among others) three groups of age/​life-​cycle related
variables out of many demographic characteristics examined in the study. The pre-​nest factor consists of individuals
aged between 25 and 49, and households headed by individuals aged between 25 and 44.The full-​nest factor includes
children of all ages, middle-​aged people, households headed by middle-​aged individuals, and large households.

260
 Retail markets and retail market studies

Finally, the post-​nest factor includes older people and households headed by older individuals and as such it refers to
the life-​cycle stages of empty nest and solitary survivor.
7 DiPasquale and Wheaton (1996) present evidence indicating that homeownership rates do rise with income and age.
8 According to Abel and Bernanke (1995), the survey of consumer sentiment, carried out monthly by the Survey
Research Center at the University of Michigan, is one of the two best known surveys of consumer expectations.The
second one is the one carried out by the Conference Board.
9 Per the results of this study, this was reflected in the strong relationship between each year’s new supply of retail space
and retail space completions in the previous three years.

References and additional readings

Abel, A. and B. Bernanke. 1995. Consumption, Saving, and Investment. Chap.  4 in Macroeconomics. Reading,
MA: Addison-​Wesley Publishing Company.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988a. Market Analysis for Retail Space. Chap. 9 in Real Estate Market
Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988b. Forecasting the Demand for Retail Space. Chap. 10 in Real
Estate Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W.Wheaton. 1996.The Operation of Nonresidential Property Markets. Chap. 11 in Urban Economics
and Real Estate Markets. Englewood Cliffs, NJ: Prentice-​Hall.
Evans, P. (2014). Non-​Retail Tenants Gain in Importance. Retail Property Insights, 21:2, 36–​42.
Hendershott, P. H., M. Jennen, and B. McGregor. 2013. Modeling Space Market Dynamics: An Illustration Using Panel
Data for US Retail. Journal of Real Estate Finance and Economics, 47:4, 659–​687.
Henderson Research. (2013). The Impact of Technology on Real Estate:  Implications for Retail and Logistics. London,
UK: Henderson Global Investors.
Hortacsu, A. and C. Syverson. (2015). The Ongoing Evolution of US Retail: A Format Tug-​of-​War. Journal of Economic
Perspectives, 29:4, 89–​112.
Ingene, C. 1984. Structural Determinants of Market Potential. Journal of Retailing, 60:1, 37–​64.
Piotrowitcz.W. and R. Cuthbertson. 2014. Introduction to the Special Issue Information Technology in Retail: Toward
Omnichannel Retailing. International Journal of Electronic Commerce, 18:4, 5–​15.
Platt Retail Institute. (2014). The Future of Retail:  A Perspective on Emerging Technology and Store Formats. Hinsdale,
IL: Pratt Retail Institute.
Schmitz, A. and D. Brett. 2001. Retail Development. Chap. 5 in Real Estate Market Analysis: A Case Study Approach.
Washington, DC: Urban Land Institute.

261
 12
ANALYZING THE MARKET
FOR RETAIL SPACE

Introduction 263
Phase I: Assessing unrealized sales potential 263
Definition/types of retail markets 264
Analysis techniques 264
Method 1: Estimating unrealized sales potential through accounting techniques 265
Step 1: Define the spatial unit of analysis 265
Step 2: Estimate the potential volume of sales, SE 265
Step 3: Estimate the sales volume supported by existing establishments, S 265
Step 4: Estimate the market’s unrealized sales potential 266
Example of the application of the accounting technique 266
Critique of the accounting technique 267
Method 2: Estimating unrealized sales potential through econometric techniques 268
Step 1: Define the spatial unit of analysis 268
Step 2: Model and predict sales/capita 269
Step 3: Estimate the sales volume supported by existing establishments, S 269
Step 4: Estimate the market’s unrealized sales potential 269
Example of the application of the econometric approach 269
Critique of the econometric approach 270
Strategic implications of the analysis of market potential 270
Phase II: The spatial behavior of consumers 271
Location selection 274
Spatial demand analysis 275
Consumer 275
Macro-access 275
Micro-access 276
Micro-environment 276
Supply analysis 276
Preferred location traits across center typologies 276

262
newgenprepdf

Analyzing the market for retail space

Phase III: Refining and evaluating the retail development scenario 277

Stage 1: Refining the development scenario 277
Stage 2: Retail trade area definition 277
Non-stylized, multiple-criteria methodologies for defining trade areas 280
Stylized methodologies for defining trade areas 281
Stage 3: Competition analysis 283
Identify and collect data on competing facilities 283
Evaluate and rate the planned development against competing developments
using the competitive differentials technique 284
Analyze the links between development traits and performance 284
Revise the development idea 285
Stage 4: Shopping pattern analysis 285
Stage 5: Evaluating the development plan 285
Step 1: Estimating the development’s likely capture rate 287
Step 2: Estimating sales potential by line of trade 292
Step 3: Estimating supportable square footage by line of trade 293
Step 4: Estimating achievable retail space rents 294
Step 5: Assessing the development scenario 298
Chapter summary 298
Questions 299
References and additional readings 301
Appendix 12A: Retail data sources 301
Appendix 12B: Further elaboration on the probabilistic approach used by
DiPasquale and Wheaton 303

Introduction
In this chapter, we discuss in detail the different analytical steps involved in each of the three phases of a
retail market study, introduced in the previous chapter. The discussion starts with Phase I, which focuses
particularly on analytical techniques that can be used to estimate unrealized sales potential and on the
strategic implications of such estimates. The discussion of Phase II focuses on the spatial behavior of con-
sumers and its implications in terms of the distances that consumers may be willing to travel for purchasing
different types of retail goods and services. Spatial demand and supply variables that need to be considered
when evaluating a specific site and location for a retail development are also discussed. The discussion of
Phase III focuses on techniques for defining the project’s trade area, analyzing the competition, identifying
consumer shopping patterns, and evaluating the development’s potential. The discussion of the latter stage
elaborates on the estimation of the project’s expected capture rate, its sales potential by product line, sup-
portable square footage, and achievable retail space rents.

Phase I: Assessing unrealized sales potential

The objective of the first phase of a well-​integrated retail-​market study is the evaluation of market entry
in terms of macro location and/​or product line. Within this context, the analysis focuses on whether
the market or metropolitan area considered is characterized by demand-​supply gaps which a new retail
development can take advantage of. Making such a determination is the foundation of a sound retail-​site
selection when a site is not given. Even if the site is given, this analysis can provide very useful insights

263
 Analyzing the market for retail space

as to whether it is prudent to develop it with a retail use. The basic question addressed at this stage is the
following:

• Is the sales potential of the area of analysis fully met by existing retail facilities?

To this aim, the analysis focuses on the estimation of the unrealized sales potential (referred to also as unmet
demand or market vacuum) as a means for assessing the demand-​supply gap in the market for retail space.
The ultimate objective of this analysis is to evaluate whether a given metropolitan market and/​or line of
retail trade has a positive unrealized sales potential or, in other words, whether potential sales exceed actual
sales. If a development idea/​concept has already been formed, which is usually the case, it would be most
appropriate for the analyst to focus on estimating unrealized market potential for the lines of retail trade
that are most relevant to this preliminary development concept.
In the case that a site is given, estimation of the unrealized sales potential by product line for the metro-
politan area and the site’s trade area can help determine what type of retail development and product lines
may have the greatest sales potential. Thus, the techniques described in the section that follows apply to
these circumstances as well.

Definition/​types of retail markets

The unrealized sales potential (SU) is defined as the difference between an area’s sales potential (SE) and
the actual volume of sales (S). Sales potential is defined as the level of consumer expenditures that can be
expected given the size and characteristics (income, age, household size, household composition, etc.) of
the population residing within the market under consideration, while actual sales refer to the total volume
of sales at the area’s stores (Ingene, 1984). The analysis of the unrealized sales potential can help evaluate
and classify a market in terms of entry possibilities. In particular, the following three categories of markets
can be identified from a strategic point of view:

1. Markets with high potential for entry, in which actual expenditures are considerably lower than the
expected levels (S<SE).
2. Competitive markets, in which actual expenditures are about equal to expected levels (S=SE). Potential
of new entry will depend on the retailer’s competitive strength (market share).
3. Very competitive markets in which actual expenditures clearly exceed expected levels (S>SE).The success
of new entrants would depend on their ability to compete fiercely with existing stores.

Analysis techniques
Embedded in the definition of unrealized sales potential is the concept of expected sales given the
local drivers of demand for retailing goods. Expected sales figures and, therefore, the unrealized sales
potential, can be estimated through two alternative methodologies:  1) accounting techniques, and
2) econometric techniques. Either way, the analysis of an area’s unrealized sales potential involves four
basic steps:

1 . Define the spatial unit of analysis.

2. Estimate the volume of sales justified by the area’s population size, income, and demographics.
3. Estimate the sales volume supported by the existing establishments.
4. Estimate the market’s unrealized sales potential.

Below we discuss how each of these steps is applied in each of the two approaches.

264
 Analyzing the market for retail space

Method 1: Estimating unrealized sales potential through accounting techniques

Step 1: Define the spatial unit of analysis
If a site is not given, the metropolitan area can be defined as the spatial unit of analysis. If a site is given,
then its trade area should be used as the spatial unit of analysis (see the discussion of Phase III on how such
area should be defined). If data from the Census of Retail Trade are used in Step 2 discussed below, then it
is absolutely critical that the spatial unit of analysis be defined in a way that does not allow for exports or imports
of retail sales (see Step 1 of the econometric technique for a more elaborate discussion of this issue).

Step 2: Estimate the potential volume of sales, SE

Estimate the volume of sales justified by the size, income, and demographics of the area’s population and
households for the line of trade under consideration as:

S E = ∑ ∑ HH ijYij eijk (12.1)

i j

where
HHij : number of households in income group i and age group j
Yij : average income of households in income group i and age group j
eijk :percentage of income spent on line of trade k by households in income group i and age group j in
physical stores (as opposed to the internet)

The above formula assumes that the analyst has a breakdown of the area’s households across two dimensions,
that is, income, i, and age, j, which are major determinants of variations in spending patterns. Estimating
sales potential using this level of household segmentation is feasible, since the BLS biannual reports on
consumer expenditures provide cross-​tabulations across these two dimensions.1 Ideally, however, a finer dis-
aggregation of households taking into account additional characteristics, such as education, household size,
etc. is desirable. In theory, such a disaggregation should help derive a more accurate estimate of household
expenditures across the different lines of trade, and the area’s sales potential. In such a case, all the terms
in formula (12.1) will have additional subscripts describing the disaggregation of households across more
than two dimensions.
A note of caution when using BLS expenditure figures, especially by line of trade. These data refers to
total expenditures in both physical stores and online. However, in typical retail market studies that per-
tain to development projects or existing retail clusters, the analyst needs to pin down consumer expenditures
in physical stores only. Therefore, in light of the increasing percentage of sales carried out online these
figures need to be reduced accordingly in order to more accurately reflect in-​store consumer expenditures.

Step 3: Estimate the sales volume supported by existing establishments, S

If survey data are not available on the actual volume of sales by existing establishments for the product line
and area under consideration, then they can be estimated by obtaining two pieces of information:

1 . The total square footage of retail space devoted to the product line under consideration.
2. The average sales per square foot for the type of store and retail development considered. Ideally, the
average sales-​per-​square-​foot figure used in this calculation should refer to the local retail market.
Such information may be obtained by surveying local retail industry experts, as well as store and
center owners. National or regional averages by type of store and shopping center were provided in

265
 Analyzing the market for retail space

the past by Dollars and Cents of Shopping Centers that was published by the Urban Land Institute (ULI).
However, ULI stopped this publication after 2008. Data on sales (but not sales per square foot) are
provided on a monthly basis at the national level by three-​to five-​digit NAICS codes in the Monthly
Retail Trade Report published by the Census Bureau.

Once these two pieces of information are obtained, the analyst can calculate total sales in the product
line under consideration as:

Total Sales (S) = Total Square Feet * Average or Median Sales/​Sq. Ft. (12.2)

In general, the median sales per square foot should be used in (12.2). If, however, the analyst has a pretty good
idea of the strength of the performance of the different shopping centers in the area of analysis, the lower or
upper end of the range can be used accordingly. In any case, the analyst should use the lower end of the range
with extreme caution.According to data that had been published in the 2008 Dollars and Cents volume for many
product lines, this is up to three times lower than the upper end.Thus, its use could result in seriously underesti-
mating the area’s actual sales and signal the presence of unrealized sales potential where there exists none.

Step 4: Estimate the market’s unrealized sales potential

The unrealized sales potential can be calculated using the following formula:

U = SE –​ S
S (12.3)

If the expected sales are greater than actual sales by a decent amount, then that would suggest that there is
indeed unmet demand or unrealized sales potential and that some retail goods and services are purchased
from stores outside the area of analysis. If expected sales are lower than actual sales, it would suggest that
local stores not only capture the full sales potential of the area under consideration, but also sell to residents
outside of the area of analysis.
Notice that if a sales-​per-​square-​foot norm is used for estimating total sales of existing establishments,
then the above approach is equivalent to comparing the supportable (or demanded) square footage with
existing supply. The analyst can make such a comparison by:

1. Calculating the supportable square footage as the ratio of the estimated sales potential over the sales-​
per-​square-​foot norm for the product line and center type under consideration.
2. Comparing the estimated supportable square footage with the total competitive space devoted to the
product line under consideration.

If the supportable square footage exceeds considerably the existing competitive space, the analyst can
infer that there is a demand-​supply gap in the local retail market.

Example of the application of the accounting technique

Table 12.1 demonstrates how the unrealized sales potential for women’s clothing in a hypothetical metro-
politan area X can be estimated using formula (12.3). The data presented in this table are hypothetical
and include the number of households in different income categories and the percentage of household
income spent on women’s clothing by each income group in physical stores. Before applying formula
(12.3), we first need to calculate the average income for each income category as the midpoint of its lower

266
 Analyzing the market for retail space

Table 12.1  Hypothetical example: Estimating unrealized sales potential for women’s clothing

Income Midpoint Number of HHs Percentage of total income Expected or

categories spent on women’s clothing potential sales
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌=[2]*[3]*[4]
<$5,000 $3,000 139,873 4.17% $17,498,107
5,000–​9,999 7,500 219,910 3.66% 60,365,229
10,000–​14,999 12,500 189,446 3.22% 76,252,092
15,000–​24,999 20,000 323,792 3.06% 198,160,991
25,000–​34,999 30,000 302,134 2.95% 267,388,616
35,000–​49,999 42,500 411,452 2.87% 501,868,167
50,000–​74,999 62,500 404,371 2.66% 672,267,247
>75,000 100,000 273,813 2.40% 657,151,704
Total expected sales $2,450,952,153
Minus
Establishment sales (given) $2,000,000,000
Equals
Unrealized sales potential $450,952,153

and upper limit (see discussion in the previous chapter about potential biases of using the midpoint of an
income range). For example, the average income for the income category $5,000–​$9,999 is calculated as
(5,000+9,999)/​2 ≈ 7,500.
A question arises in trying to calculate the midpoint of the last category, which is an open-​ended cat-
egory. In this case, the analyst needs to assume a reasonable upper boundary and calculate the implied mid-
point. In this example an upper boundary of $125,000 was assumed, which implies a midpoint of $100,000.
Also notice that the midpoint of the first income category (<$5,000) reflects the assumption that the lowest
income in that category is $1,000 and not zero.
Once the midpoints of each income category are calculated, then expected sales of women’s clothing
within each income group can be estimated as the product of this midpoint times the number of households,
times the percentage spent on this product line in physical stores. Summing up potential sales across all
income groups provides an estimate of the total volume of potential sales of women’s clothing, SE, in
metropolitan area X, which is approximately $2.451 billion. According to the most recent data, however,
establishment sales of women’s clothing in the metropolitan area under consideration sum up to $2 billion.
Therefore, we can conclude that there is an unrealized sales potential of about $451 million in women’s
clothing in metropolitan area X.

Critique of the accounting technique

One limitation of this approach is that, due to data constraints, it does not allow the analyst to take
into account the effect of multiple population characteristics on retail expenditures when calculating an
area’s sales potential. In particular, a popular source of secondary data on retail expenditure patterns is the
Consumer Expenditure Survey published by the Bureau of Labor Statistics (BLS). This survey provides
only one-​dimensional breakdowns by a number of variables, such as household-​head age, education level,
occupation, household size, etc. The only two-​dimensional breakdowns provided by the BLS survey are
the cross-​tabulations by age and income groups, by income and household size, and by age and house-
hold size, published in the biannual reports. However, this information is still limiting since the effect of
age and income, given an educational level, and an occupation, and a household size, and so on, cannot be
calculated from the data.

267
 Analyzing the market for retail space

As indicated earlier, another issue that arises when using the BLS data is that the various tabulations of
expenditures for different types of goods and services by different household characteristics refer to total
expenditures for both in-​store and online purchases. Since most retail market studies are carried out for
the development of physical store space, such data need to be adjusted to reflect household expenditures
for in-​store purchases only.
The advantage of the accounting technique over the alternative econometric approach is that, depending
on the year of analysis, it may allow the analyst to use the more up-​to-​date data of the BLS Consumer
Expenditure Survey, as opposed to the data from the Census of Retail Trade (required for the application
of the econometric technique), which is available every five years. Furthermore, it is much easier to apply,
and less data-​intensive and time-​consuming. As such, it is often the preferred technique used by retail
analysts for estimating an area’s unrealized sales potential, despite the fact that it may be less accurate than
the econometric approach.

Method 2: Estimating unrealized sales potential through econometric techniques

The basic difference between the econometric technique and the accounting technique is that the effect
of variation in multiple household characteristics on spending patterns is estimated econometrically. Thus,
in trying to estimate an area’s unrealized sales potential through econometric techniques, the key question
is: what are the structural determinants of market potential, or what are the underlying factors determining
demand for retail goods?2
This technique involves the use of cross-​sectional data and, as such, it requires information on con-
sumer expenditures for a large number of markets. Because of the high time and money costs involved in
collecting such data through consumer surveys, this technique is practically applicable only using infor-
mation on actual establishment sales by county and metropolitan area, which is readily available from the
Census of Retail Trade. The big advantage of this methodology is that the sales data do not need to be
adjusted for online purchases since they refer directly to sales in physical stores. Application of this approach
involves the following steps:

Step 1: Define the spatial unit of analysis

If establishment sales are used as proxy for an area’s consumer expenditures when using the econometric
technique, it is very important that the unit of analysis be defined geographically in a way that avoids
problems of exports or imports of sales. Avoiding the problem of leakages when using such data is abso-
lutely critical for the reliability of the results of this technique. Simply, if the area’s establishment sales do not
accurately reflect the retail expenditures of the population of the area under consideration, then estimates
of the relationship between population characteristics (such as income, age, household size, etc.) and sales
will be misleading. This issue does not arise in the case of the accounting techniques, unless the data used
to estimate the percentage of income spent on a product line represents establishment sales.
A city is a net exporter of a good if the retail expenditures of its residents, often referred to as sales by place
of consumption, are lower than the sales of all retail establishments within the city. On the contrary, the city is
a net importer of a good if sales by place of consumption are higher than the sales of all retail establishments
in the city. For instance, consider a city in which expenditures on women’s clothing by place of consump-
tion are $50  million, while sales at all women’s clothing stores within the city sum up to $30  million.
Obviously, the city is net importer of $20 million sales in women’s clothing.
Within this context, the metropolitan area is the appropriate geographic unit of analysis since very little
retail shopping takes place out of its boundaries by its residents, while at the same time the area’s outlets sell
minimal quantities to residents of other metropolitan areas (unless the city is a popular national or inter-
national tourist destination).This definition, therefore, helps ensure that total sales in the area’s stores reflect

268
 Analyzing the market for retail space

the total consumption of its population. As such, it avoids the issue of exports or imports of retail sales to or
from other areas, which is one of the most common and critical errors in retail market studies.

Step 2: Model and predict sales/​capita

An area’s sales/​capita can be modeled as a function of income and other demographic characteristics, as
well as other factors that may contribute to retail expenditure variations across markets (see Ingene, 1984).
The analyst should avoid using supply as an explanatory variable because supply does not generate demand.
To carry out this step, the analyst needs to first collect data on both sales by line of trade and population
characteristics for a sample of market (metropolitan) areas and then estimate a model of retail sales per
capita by line of trade using econometric techniques.3 The linear model that needs to be estimated is
described by (12.4):

SC E = b0 + b1X 1 + b2X 2 +  + bn X n (12.4)

where
SCE : retail sales per capita
X : a set of demographic and other variables that influence the volume of retail sales within a metro-
politan area
Once the econometric analysis is completed, the analyst can use the values of Xs for the particular
metropolitan area under consideration and the estimated parameters of (12.4) to estimate the expected
sales per capita justified by the area’s demographics. Subsequently, total expected sales can be calculated as:

E = SCE * Metropolitan Area Population

S (12.5)

Step 3: Estimate the sales volume supported by existing establishments, S

As described in the accounting technique.

Step 4: Estimate the market’s unrealized sales potential

As described in the accounting technique.

Example of the application of the econometric approach

The application of the econometric approach for the estimation of unrealized sales potential discussed
below is based on Ingene (1984).

S tep 1: The analyst has identified a sample of about 250 metropolitan areas in the United States.
Step 2: The analyst has collected data on sales per capita by line of trade across these markets, as well
as data on the structural determinants of these sales, such as demographics and other relevant
variables. Using these data, the analyst estimated cross-​section regressions of sales per capita (SCE)
for various lines of retail trade—​apparel, department store, drug, furniture, general merchandise,
grocery, hardware, and variety. Take the case of the estimated model for apparel:

SCE = –​9.05 + 39.69 Full-​nest + 1.43 Pre-​nest –​32.18 Low Income

–​45.50 Mid Income + 21.76 Mobility + .... + 51.97 Atmospherics

269
 Analyzing the market for retail space

where
Full-​nest, Pre-​nest : composite demographic variables reflecting the stage of households in their
life cycle
Low Income, Mid Income : the metro area’s average income classification
Mobility : percentage of households with cars
Atmospherics : percentage of stores built in the preceding five years

Step 3: Suppose that for given values of the explanatory variables presented above, the predicted annual
sales per capita for apparel in metropolitan area X are:

SCE = $575

Given that the metropolitan area’s population is 300,000, its expected sales are:

SE = 575 * 300,000 = $172.5 million

With existing sales, S, of $150 million (based on establishment data), it seems that the unrealized
sales potential of this area is:

SU = SE –​ S = 172.5 –​ 150 = $22.5 million

Critique of the econometric approach

The advantage of the econometric approach over the accounting approach in estimating a market’s
retail sales potential is that it allows the analyst to take into account the effect of multiple population
characteristics on retail expenditures in a specific line of trade. Besides age and income, such characteristics
could include household size, occupation, education, mobility, etc. The drawback of this approach (besides
being data-​intensive and time-​consuming) is that the data needed for applying it are only readily available
from the Census of Retail Trade, which is carried out every five years. Thus, depending on the time of
analysis relative to the latest Census of Retail Trade, the analyst may find himself/​herself using out-​of-​date
data on expenditure patterns. This may be a problem if the spending patterns have changed considerably
during the time that has elapsed since the latest census.

Strategic implications of the analysis of market potential

The estimation of the unrealized sales potential can help formulate retail investment and retailer entry strat-
egies because it provides insights into the type of retail market at hand and the strength of the competition
the retailer will likely face. Such insights can help evaluate how competitive both the development and the
retailers have to be for a successful entry in that market. As such, they can provide a broader framework for
the design of the development scenario and the selection of the tenant mix.
Table 12.2, obtained and slightly modified from Ingene (1984), demonstrates more specifically the stra-
tegic implications of unrealized sales potential estimates. For example, a positive unrealized sales potential
(SE > S) suggests that market entry is feasible and viable, especially for strong retailers with a higher-​than-​
average market share. For retailers with share below market average, market entry is more likely to be viable
only in markets where the unrealized sales potential is high. In markets where the expected sales are equal
to or less than actual establishment sales (SE = S, or SE < S), entry is more likely to be viable only for retailers
with a market share considerably higher than average. The implications for retail space developers are the
same. For example, a developer contemplating entering a market in which expected sales are roughly equal

270
 Analyzing the market for retail space

Table 12.2  Strategic implications of the analysis of market potential

Market entry or expansion strategy

Scenarios MSE > MSAV MSE < MSAV
SE > S Enter Enter only if SE –​ S high
SE = S Enter only if MSE –​ MSAV high AVOID
SE < S Enter only if MSE -​ MSAV high AVOID
Notation SE : estimate of expected sales based on their structural determinants
  S : actual establishment sales  
  MSE : retailer’s expected market share
  MSAV : average expected market share

Source: Ingene, C. 1984. Structural Determinants of Market Potential. Journal of Retailing, 60, 37–​63.

to actual sales (SE = S) should do so only if he/she can secure the participation of very strong tenants with
a market share considerably higher than average.

Phase II: The spatial behavior of consumers

When it comes to retail purchases, the spatial behavior of consumers varies considerably depending on the
nature of the retail product. As indicated in Figure 12.1, three broader aspects of retail products are relevant
when trying to understand consumer spatial behavior:

1 . Frequency of purchase
2. Susceptibility to complementary shopping
3. Susceptibility to comparison shopping.

The frequency by which a good or service is purchased is important because consumers are unwilling to
travel long distances very often.4 Thus, for purchases of goods, which are bought at high frequency, such
as food items, consumers tend to shop at the closest and most convenient locations. Stores that are selling
this type of goods, therefore, need to have close-​by access and proximity to consumers.These stores are free
standing or in neighborhood shopping centers, and since they can draw primarily customers that live in
close proximity, they serve smaller market areas. As a result, they have the highest density within the urban
fabric among the different types of retail facilities.
Consumers are in general willing to travel considerably longer distances when it comes to retail products
that are bought at low frequency, such as durables. Thus, stores selling these types of goods serve larger
market areas and, as such, they need to have good broader access. These stores can also be found either as
free standing or in clustered developments, such as regional shopping centers.
In categorizing retail goods in terms of shopping frequency, Carn et al. (1988) refer to those purchased
in low frequency as shopping or high-​order products and to those purchased in high frequency as con-
venience or low-​order products. As Table  12.3 indicates, shopping or high-​order products include personal
shopping goods, such as clothing, jewelry, dry goods, and gifts, purchased several times per year; household
shopping goods such as furniture, appliances, and carpets purchased even more infrequently; and automo-
tive products that are purchased very infrequently. Convenience or low-​order products include convenience
goods, such as groceries, drugs, hardware, and liquors, purchased daily weekly or monthly; personal services,
such as haircuts, dry cleaning, etc., purchased frequently; repair services, such as shoe repair, small appliance
repair, etc., purchased frequently to often; and personal business and financial services, such as banking, real
estate, insurance, etc., purchased frequently to infrequently. Some product or service categories that cannot

271
 Analyzing the market for retail space

THE SPATIAL BEHAVIOR OF CONSUMERS

Susceptibility Susceptibility to comparison shopping

Frequency of to Depends on degree of product differentiation
purchase complemen-
tary shopping and buyer's information needs

Homogeneous Heterogeneous
products products

High Low High Non-

frequency Standardized Standardized
Example: Example: Example: Example: Example:
Food Durables Clothes and New cars, Used cars,
shoes appliances jewelry

Consumers Consumers Usually Lower Lower

shop at travel lower frequency frequency
most con- longer frequency purchases; purchases;
venient distances purchases; consumers consumers
location; less often consumers may travel do travel
unwilling to willing to longer dis- longer dis-
travel long travel longer tances to tances to
distances as to a single clustered clustered
often destination for developments developments
joint purchases for compari- for compari-
son shopping son shopping

Location preferences/patterns

Close by Good Proximity to Sellers prefer Sellers prefer

access, access to sellers of com- to locate close to cluster in
proximity consumers; plementary to stores with central,
to consu- low density products/ similar yet highly
mers; free- of free- clustered differentiated accessible
standing standing patterns of products; locations
stores at or clustered development clusters need
high density stores very accessible
locations

Small Large Large Larger market Very large

market market market areas are re- market areas
areas areas areas quired to en- to ensure
sure adequate adequate
consumer base consumer base

Figure 12.1  Phase II: Site and location analysis

272
 newgenrtpdf
Table 12.3  General categories of retail goods and services1

Category Type of items Frequency of Type of shopping trip Proximity to market

replenishment
A. Shopping or high-​ 1. Personal shopping Clothing, jewelry, dry Often—​several Single-​ or multiple-​purpose Moderate—​15 to 30
order products goods goods, gifts times per year comparison minutes
2. Household Furniture, appliances, Infrequently—​less Single-​purpose comparison Within driving
shopping carpets than yearly shopping range—​over 30
minutes
3. Automotive Automobiles and major Infrequently Single-​purpose comparison Within driving range

Analyzing the market for retail space

products accessories, trailers, boats shopping
B. Convenience or 1. Convenience goods Groceries, drugs, hardware, Frequently—​daily to Multiple-​purpose repetitious Close by—​5 to 15
low-​order products liquors monthly shopping minutes
2.  Personal services Beauty or barber shop, Frequently Multiple-​purpose repetitious Close by
cleaners, health spas shopping
273

3.  Repair services Shoe repair, small appliance Frequently to often Multiple-​purpose repetitious Close by
repair, furniture repair shopping
4. Personal business Branch bank, real estate, Frequently to Single-​or multiple-​purpose Close by to moderate
and financial insurance infrequently repetitious or comparison
services shopping
C. Miscellaneous 1. Eating and drinking Fast foods, dining services Frequently Single-​ or multiple-​purpose Close by
products, mixed with menu foods, repetitious consumption
characteristics cocktails
2. Entertainment and Theaters, bowling arcades Frequently Single-​purpose repetitious activity Close by
recreation
3. Automotive services Gasoline, tires, and batteries, Frequently to often Single-​or multiple-​purpose Close by to moderate
and accessories engine repair and repetitious or comparison
maintenance shopping

Republished partially from Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Real Estate Market Analysis: Techniques and Applications. Englewood Cliffs,
1

NJ: Prentice-​Hall.
 Analyzing the market for retail space

be clearly classified as high-​or low-​order include eating and drinking, entertainment and recreation, and
automotive services and accessories.
The second important aspect of retail products, in terms of influencing spatial consumer behavior, is the
susceptibility to complementary shopping. Complementary products are products that complement each
other in the sense that the purchase of one item typically requires or encourages the purchase of the other.
An example of complementary products is clothing and shoes. Obviously, the co-​location of stores that sell
complementary products allows consumers to save time and reduce travel costs by carrying out all com-
plementary purchases at a single location. For this reason, consumers are in general willing to travel longer
distances to a single destination where they can make such complementary purchases.
Thus, stores selling complementary products benefit from each other’s traffic and, as a result, they do
tend to co-​locate in clustered developments (shopping centers). Since consumers are willing to travel
longer distances to reach them, these developments tend to also serve larger market areas. Good broader
access, therefore, is also an important locational attribute of retail clusters that sell complementary products.
The third aspect of retail products that affects consumer behavior is susceptibility to comparison
shopping. This is directly related to the degree of product differentiation. Obviously, the need for com-
parison shopping arises when the product under consideration is highly heterogeneous, forcing consumers
to inspect/​check several stores/​products before committing to a purchase. That is why retailers selling
products that are susceptible to comparison shopping, such as clothes, shoes, cars, or furniture, tend to
cluster at the same location.
Heterogeneous products can be distinguished as standardized and non-​standardized. For example,
new cars and household appliances are standardized products. Used cars and jewelry, however, are non-​
standardized products. Heterogeneous products (standardized or not) are bought usually at lower frequency
and, thus, consumers are willing to travel longer distances to clustered developments for the opportunity
of comparison shopping. Larger market areas are required to ensure a large consumer base for such
developments.
As Table 12.3 indicates, some retail goods may be susceptible to both comparison shopping and com-
plementary shopping and some not. For example, personal shopping goods are susceptible to comparison
shopping and often require or encourage complementary purchases, allowing for multi-​purpose shopping
trips. Household shopping goods and automotive products rarely require or encourage complementary
purchases but are susceptible to comparison shopping. Most convenience goods do not require comparison
shopping but are often purchased jointly, thus encouraging multi-​purpose shopping trips.
Each of the three aspects discussed above has implications with respect to the level of micro and macro
consumer access and the size of market area required to sustain a certain retail development. In dense urban
areas, which are characterized by high density of demand, trade areas are well defined and mostly non-​
overlapping, especially those for smaller retail developments. Overlapping market areas are more common
in the case of large regional and super-​regional centers serving the whole metropolitan area. Ideally, retailers
would favor being located in a retail development whose market area does not overlap with the market area
of competing retail developments because it would provide them with a spatial monopoly status. In esti-
mating a retail project’s capture rate, it is very important to carefully define the market areas of potentially
competing retail developments and identify the extent to which they are overlapping with the market area
of the development under consideration.

Location selection
The basic focus of retail location analysis is to determine the crucial location attributes that need to be
considered by a retail developer in his/​her search for an appropriate site or when evaluating a given site.
If the type of retail development is known and a site is yet to be found, then the analyst needs to deter-
mine the locational requirements for that type of development and then compare them to the locational

274
 Analyzing the market for retail space

attributes of candidate sites to identify the best match. If a site is given, then the analyst’s major objective
is to determine what type of retail development would be most appropriate for that location given its
characteristics. Whichever is the case, a retail site in search of use, or a retail use in search of a site, the basic
variables that retail-​location analysis needs to focus on can be classified into two broader categories: spatial
demand and supply.
The three different major aspects of retail products discussed in the previous section have important
implications with respect to the location characteristics that matter for a given type of retail development,
or the type of retail development most suitable given certain location attributes. For example, if the type of
retail development and categories of goods that will be sold are known, then site accessibility requirements
and trade area requirements can be determined. If the site is known instead, then its accessibility advantages,
trade area characteristics and the competitive landscape will dictate what type of center and tenant mix
would be mostly in demand at the given location.
Within this context, the spatial demand variables that are mostly relevant in retail site location analysis
have to do with trade area demographics and site accessibility, while relevant supply variables include the
location, characteristics, and performance of competing retail developments. Mapping the spatial distri-
bution of consumers, the transport network, potential competition, vacant land, and underutilized sites
when analyzing these demand and supply location factors (discussed in more detail below) can help make
preliminary site selections more quickly and better evaluate the different alternatives. Thus, Geographic
Information Systems (GIS) combining mapping capabilities with demographic data (including income)
can be very helpful at this and later stages of analysis.

Spatial demand analysis

Assuming that the site is given, the primary purpose of spatial demand analysis is to identify which of
the various demand segments (residential-​based, employment-​based, passers-​by and tourists) will be the
major sources of demand and, therefore, the target groups for the planned development at the particular
location under consideration. Then the growth prospects of these demand segments over the project’s
planning horizon should be carefully evaluated. To this aim the following consumer, accessibility, and
micro-​environment factors need to be examined as they relate to the specific site and location:

Consumer
• Spatial distribution, density, income, and other demographic characteristics of households and
population.
• Expected growth in population/households and changes in their spatial distribution, income, and
demographics (increases in income and population density within a development’s trade area will boost
purchasing power, while decreases will reduce it).5
• Planned residential developments (permits) and relevant public policies, such as growth and zoning
controls that may restrain residential development and population growth within the development’s
trade area.

Macro-​access
• Site linkages to regional access network and traffic counts (special care needs to be taken in identifying
highly congested routes and how they affect accessibility to the site under consideration).
• Expected changes and expansions in transportation network.
• Physical and psychological barriers.

275
 Analyzing the market for retail space

Micro-​access
• Frontage on, or direct access off, major street or interstate freeway in the case of shopping centers.
• Ease of entry from and exit to the street.
• Traffic flows and pedestrian traffic.
• Parking availability.

Micro-​environment
• Proximity to employment centers/​sources of midday shoppers.
• Neighboring uses and positive or negative externality effects.
• Visibility.

It is important that the analyst does not only carefully evaluate the status of the above factors at the time of
analysis, but also very carefully assess how these factors may change in the future and, if so, how anticipated
changes could affect the site’s attractiveness and/​or retail sales potential. For example, new transportation
infrastructure, expected to be completed over the project’s planning horizon, may redirect traffic flows
away from the artery the site is located on, thereby significantly reducing its exposure, visibility, and sales
potential.

Supply analysis
The purpose of supply analysis is to identify the extent of the competition that the planned development
will be facing at its particular location. To this aim, the following factors need to be examined for all retail
developments—existing, potential expansions, under construction, and planned—​that are likely to pene-
trate the planned development’s trade area:

• Spatial distribution, locational, physical, and operational characteristics, including type of center, tenant
mix, and product lines.
• Sales per worker, sales per square foot, rate of business failures.

Survey of all the features of competitive developments that contribute to their sales volume can help the
analyst first assess the advantages and disadvantages of the site and the proposed development and then
identify which features need to be avoided (if possible) and which ones need to be reinforced or added
in order to improve its competitive position. Furthermore, analysis of the competition may help identify
demand gaps in specific product lines. Finally, performance indicators cannot only help evaluate the inten-
sity of the competition and the extent to which it allows additional entries in the area, but also provide the
basis for identifying the development attributes that contribute to higher sales per square foot.6
The intensity of the competition can be evaluated by comparing average sales per worker or per square
foot in competitive developments with the minimum sales required for breaking even. If, for example, average
sales per square foot in competitive developments are close to the minimum required for the retailer to
break even, then that would be an indication of a highly competitive area with little margin perhaps for
additional entries.

Preferred location traits across center typologies

By now it should be clear that the basic types of retail developments are differentiated according to the
frequency of purchase of the products they sell, with neighborhood centers selling primarily goods that

276
 Analyzing the market for retail space

are purchased at the highest frequency, and regional and super-​regional centers selling primarily goods
that are purchased at the lowest frequency. In between, lie the community centers selling primarily goods
purchased at moderate frequency.The analyst needs to carefully examine the locational requirements of the
different types of retail developments in order to determine what type of development better matches the
site’s locational attributes, if a site is given. Alternatively, a site will need to be found that matches the site
and locational requirements of a given retail development type.
The preferred location traits across the various center typologies, as well as the typical tenants they
house, are described by Table 12.4 (republished from Carn et al., 1988). As this table indicates, regional
shopping centers require location at the intersection of two or more regional thoroughfares, as well as close
proximity to major residential and outlying employment areas; community centers require location on a
regional thoroughfare at a major intersection and proximity to two or more residential neighborhoods;
neighborhood centers require location on a major thoroughfare and proximity to at least one residential
neighborhood; and commercial strip centers require location on a major thoroughfare between residential
and employment areas.

Phase III: Refining and evaluating the retail development scenario

Phase III, which is carried out after the site and its location have been thoroughly analyzed, is at the heart
of retail market analysis. It comprises a set of five interactive stages that focus on the question of whether
there is adequate demand to support the different tenants of a retail development scenario. These stages
include:

1 . Refinement of the development scenario

2. Definition of the development’s trade area
3. Competition analysis
4. Shopping pattern analysis
5. Evaluation of the final development concept.

Stage 1: Refining the development scenario

The insights gained from the analyses carried out in Phases I  and II are used to further refine the
development concept in terms of target consumer groups, tenant mix, and center amenities. For
example, site and location analysis may have indicated that the site is well situated to serve high-​
income communities, which would suggest the inclusion of upscale tenants for the different product
lines considered.

Stage 2: Retail trade area definition

A retail project’s market or trade area is defined as the area encompassing all locations from which
consumers can be drawn to the site. Developments appealing to several demand segments, in reality,
may not have a single but multiple trade areas, corresponding to the different product lines included
in the retail program and the frequency by which they are purchased. Ideally, those different demand
segments and trade areas by product line need to be identified and analyzed. However, for a prelim-
inary analysis of the development concept, one trade area may be defined for the whole development,
depending on its type (regional center, community center, etc.) and the analyst’s evaluation of its overall
drawing power.
In many retail market studies, a distinction is made between primary, secondary, and tertiary trade areas.
As indicated in Figure 12.2, the primary market area is defined as the area from which the retailer expects

277
 newgenrtpdf
Table 12.4  Preferred location traits across center typologies

Type of market area Type of location Major locational characteristics Typical types of retail tenants
Regional Central business district At hub and regional thoroughfares; near center of Department stores; variety stores; personal shopping and
commercial and residential areas specialized goods and services; business goods and
services; eating and drinking; entertainment; other
supporting goods and services
Regional shopping At intersection of two or more thoroughfares; near Department stores; two + variety stores; shopping goods;
center or district major residential and outlying areas specialized goods and services; eating and drinking;
entertainment and recreation; other supporting goods
and services
Specialized shopping Parasite to central business district or regional Fashion apparel; factory outlets, eating and drinking;
center district center; at location similar to regional center specialized goods and services

Analyzing the market for retail space

Community Community shopping On a major regional thoroughfare; usually at a Department stores; one or more discount stores; major
center or district major intersection; convenient to two or more drug or variety store; personal shopping goods; fast
neighborhoods; entertainment and recreation; food outlets; automotive services and accessories; other
personal business; and financial services supporting goods and services
Discount or off-​price Same as for community center or specialized Discount store; factory outlets; personal shopping goods;
278

center shopping convenience goods; fast-​food outlets; entertainment and

recreation; other supporting goods and services
Neighborhood Neighborhood On a major thoroughfare close toan intersection; Supermarket, convenience goods, some personal shopping
shopping center or adjacent to at least one neighborhood goods; personal and repair services; personal business,
district financial services; fast foods, entertainment and
recreation; automotive services; other supporting goods
and services
Convenience center or On a major thoroughfare; adjacent to a Convenience food store; convenience goods; automotive
cluster neighborhood services; personal services; repair services
Undifferentiated Arterial street, On a major thoroughfare between residential and Free-​standing units; household shopping goods;
commercial strip employment areas convenience goods; automotive accessories and services,
automotive products; fast-​food outlets; numerous other
 newgenrtpdf
Highway-​oriented On a major highway, usually at outlying locations; Transient facilities; eating and drinking with fast foods;
commercial strip but may be same as arterial street, commercial household shopping goods; automotive services and
strip accessories, automotive products; convenience goods;
entertainment and recreation
Heavy commercial On a major or minor thoroughfare; may be same as Trucking and storage; building materials; major auto repair;
strip highway-​oriented or arterial commercial strip salvage uses; automotive service; limited convenience
goods; eating and drinking with fast foods
Dependent Near, adjacent to, or within a major use or market Gifts, souvenirs, and specialized goods and services; eating
commercial generator and drinking, entertainment and recreation; personal
services
Free-​standing structures On a major thoroughfare; usually in outlying Automotive products; household shopping goods;
and uses location; occupies large individual site; may or distributors with retail outlet; supporting eating and
may not be in a cluster or strip drinking or convenience goods

Analyzing the market for retail space

Source: Republished from Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Real Estate Market Analysis: Techniques and Applications. Englewood Cliffs,
NJ: Prentice-​Hall.
279
 Analyzing the market for retail space

Infrequent visits

Less frequent
but regular visits

Consistently
frequent visits

Primary
60–70% of sales

Secondary
15–20% of sales

Tertiary
5–10% of sales

Figure 12.2  Retail trade area definition

consistent and frequent visitations. Typically, purchases by shoppers drawn from this area make up 60–​70%
of the store’s or center’s total sales (see Figure 12.2). The secondary market area is the area from which
the retailer expects less frequent but regular visits from consumers contributing 15–​20% to the center’s
total sales. Finally, the tertiary market area is the area from which consumers are drawn only infrequently,
accounting for 5–​10% of total sales.
In general, we can distinguish two types of methodologies for defining retail trade areas: 1) non-​stylized
approaches, and 2) stylized approaches. Below we discuss each of these two types of techniques.

Non-​stylized, multiple-​criteria methodologies for defining trade areas

Non-​stylized techniques for defining retail trade areas use multiple criteria and involve the following
major steps:

1. Use of driving times, considering road access and speeds, as well as public transportation routes, to
define the geographic area from which consumers might be attracted to the site.
2. Analysis of travel and consumer shopping patterns (through primary research and secondary data if avail-
able) as well as topographical, socioeconomic, and cultural barriers in order to refine the definition of
the trade area as derived from the previous step.
3. Analysis of the location of competing retail developments and adjustment (reduction) of the trade area to
account for losses to these competing developments.
4. Further adjustment of the boundaries of the trade area so as to coincide with the geographic boundaries of
areas for which secondary population and income data are available, such as counties, census tracts,
zip codes, or blocks. Fanning et al. (1994) suggest that blocks may be used for neighborhood centers,
census tracts or zip codes for community centers, and counties for regional centers.

280
 Analyzing the market for retail space

5. Comparison of the trade area defined as above against the trade areas of comparable retail developments
elsewhere.

Stylized methodologies for defining trade areas

Despite the importance of the factors just discussed in accurately delineating retail trade areas, only well-​
thought-​out retail market studies pay serious attention to all of them. Most market studies follow stylized
methods, which may be too simplistic to rely on. Such techniques include:

1. Concentric ring approach

This approach defines the trade area on the basis of a distance radius around the subject site
depending on the type of development considered. For example, according to Schmitz and Brett
(2001), the primary-​area norm for a neighborhood center is a one-​and-​a-​half-​mile radius, for a
community center is a three-​to five-​mile radius, and for a regional mall is an eight-​to 12-​mile
radius.The pitfall of this approach is that it ignores the uneven accessibility patterns as determined by
the area’s transportation network, topographical and psychological barriers to travel, differentiated
shopping patterns across household groups, and competition. This approach is especially problem-
atic in the case of developments whose trade area overlaps with the trade area of competing retail
developments.
2. Driving time approach
In this case the boundaries of the trade area are defined on the basis of a maximum driving-​time norm
estimated by taking into account the area’s transportation network. For example, the driving time
norms are 15 minutes for a neighborhood center, 30 minutes for a community center, 45 minutes for
a regional center, and 60 minutes for a super-​regional center.This approach suffers from similar pitfalls
to the concentric ring approach except for the fact that it accounts for the idiosyncrasies of the local
transportation system.
3. “Near twin” approach/​customer spotting
“Near twins” or “comparable” developments are selected based on the following criteria:

Radius depends
on the type of
store/retail
center Retail site
1–2 miles if All points on the
3–5 miles if neighborhood curve are within
community center equal driving
center time from the site
Time Center
15 min Neighborhood
30 min Community
45 min Regional
60 min Super-Regional

Figure 12.3 The concentric rings and the modified concentric rings/​driving time approaches

281
 Analyzing the market for retail space

a ) Development type and size

b) Tenant  mix
c) Layout and physical design
d) Surrounding streets
e) Location with respect to downtown and other activity centers
f) Location with respect to competitors
g) Access and distance to sources of demand (residential and employment-​base)
h) Volume of traffic on adjacent streets for stop-​off business.

Once the “near twin” or comparable development is identified, information is collected regarding
the frequency of visits by type of customer (as defined by demographic, income, and spending
characteristics) and by location (block or sector of the city). Information is also collected regarding
the near twin’s performance, measured by dollar sales per customer or per dollars income of the cus-
tomer. This information can then provide the basis for identifying the spatial extent of the site’s area
of influence and for differentiating between primary, secondary, and tertiary trade areas. The required
data for the “near twin” or “comparable” developments can be obtained from franchise information,
and/​or through traffic counts and consumer surveys.This approach is expensive to implement though
and problematic if true “near twin” developments cannot be found.
4. Simple gravity approach
The simple gravity approach explicitly considers the limitations imposed on a retail development’s
draw by competitive developments. According to this approach, the trade area boundary can be
determined using the gravity formula below (Figure 12.4).

d
TAB = (12.6)
S 
1+  2 
 S1 

where,
TAB : trade area boundary of subject store 1 considering competitor 2 (C2)
d : distance between subject store and competitor 2
S1, S2 : “mass” variables, reflecting the subject store’s and the competitor’s attraction (ideally Total
Amenity Indices, or store square footage if such indices are not available)

C2 C3

TAB

C1
C5
C4

C6

Figure 12.4 The simple gravity approach

282
 Analyzing the market for retail space

Since the numerator in this formula is distance and the denominator is a unitless number, the resultant
figure from the calculation of (12.6) is distance. This distance indicates how far the subject store’s
attraction extends given the location of the specific competitor under consideration. The underlying
premise of equation (12.6) is that the greater a center’s distance from its competitors and the greater
its advantage in terms of mass and amenities (S1) relative to the competitor’s mass and amenities (S2),
the greater its attraction and its trade area.7
5. Micro-​analysis/​probabilistic approaches
These approaches allow the analyst to estimate a project’s capture rate from the different residential
zones that are located within its trade area. These capture rates can then provide the basis for more
accurately defining the development’s trade area by including all the residential zones with a minimum
capture rate (for ­example 60% in the case of the primary trade area). Since these approaches can be
used in a more constructive way at a much later stage of the analysis, they are discussed in more detail
in a subsequent section.

Stage 3: Competition analysis

Analysis of competing retail developments has traditionally been the second most important step under-
taken in Phase III once the project’s trade area has been delineated. One of the major objectives of this stage
of analysis is to link sales performance and development attributes and better assess the extent of purchasing
power that may be lost to competing projects. As such, it involves three basic steps:

1 . Identification and collection of data on competing facilities

2. Evaluation and rating of the planned development against competing developments using the com-
petitive differentials technique
3. Analysis of the links between development traits and performance
4. Revision/​refinement of the development idea.

These steps are discussed in more detail below.

Identify and collect data on competing facilities

The first step in competitive analysis is to identify all existing, planned, and under-​construction retail
developments that are likely to capture sales from the project’s trade area. For smaller retail developments,
detailed analysis is usually carried out for competing properties (existing or planned) that are located within
the project’s primary trade area boundary, as well as those located outside of it but close enough to cap-
ture some of that area’s sales. Important aspects that need to be addressed when surveying the competition
include:

a) Location attributes (micro and macro access to consumers, surrounding land uses and

neighborhoods, etc.).
b) Site, structure, and project attributes (including visibility, topography, frontage, utilities, site coverage,
total GLA, GLA by major categories of retail goods and services, parking availability, amenities, signage,
landscaping, and truck service facilities).
c) Tenant characteristics
1. Anchor and non-​anchor tenants (mix, types of stores, size, reputation, etc.)
2. Inter-​store relationships (how traffic in one store helps sales in other stores)
3. Non-​retail uses (i.e., office, entertainment)

283
 Analyzing the market for retail space

4 . Marketing programs and policies

5. Potential for expansion
6. Variety, quality, and price of merchandise
7. Rental rates and other tenant charges
8. Measures of performance (sales per square foot, sales per worker, capture rate, occupancy rate).

It should be noted that it is often not possible to assemble all the pieces of information that are theor-
etically relevant in fully evaluating competitive developments, because either simply the information is
not available or the managers of competitive facilities are not willing to disclose it. Some examples of
difficult-​to-​obtain information include performance measures, such as sales per square foot or sales per
worker, and inter-​store relationships, as measured by the contribution of anchor tenants to sales of sat-
ellite stores.

Evaluate and rate the planned development against competing developments

using the competitive differentials technique
The detailed attributes of the planned project and competing developments are systematically listed and
evaluated using the competitive differentials technique (discussed extensively in Chapter 9), in order to
identify the project’s strengths and weaknesses and assess its competitive position. This analysis can be very
useful for deriving estimates of the project’s capture rate, sales potential, and achievable rental rates, at later
stages of the analysis. Even if more advanced approaches are eventually used for the estimation of the
project’s capture rate and sales potential, estimates derived using the competitive differentials technique can
provide some alternative assessments against which the reasonableness of those estimates can be evaluated.

Analyze the links between development traits and performance

The detailed information collected on competing projects can help identify the attributes that matter the most
from a sales point of view. If consistent information on sales per square foot, tenant mix, project amenities,
and other project characteristics is collected for a sufficient number of competitors, econometric techniques
can be used to analyze and quantify the explicit links between development traits and sales performance. In
particular, the analyst can estimate a regression model with sales per square foot as the dependent variable
and center characteristics as the independent variables (X1, X2,…Xn):

Sales/​sq. ft. = a0 + a1 X1 + a2 X2 + ….. + an Xn (12.7)

Estimation of (12.8) can help identify which retail center attributes have a statistically significant effect on
sales per square foot. Furthermore, it can help evaluate the relative importance of these attributes in deter-
mining project sales (see the discussion in Chapter 7 on the basics of regression analysis). Although estima-
tion of (12.7) provides an objective means for assessing the links between center performance and center
attributes, information on the subjective evaluation of different center attributes by the area’s households
through consumer surveys can also provide very valuable insights on this issue.
If the number of competing facilities is not enough to allow for an econometric evaluation of the rela-
tionship between project attributes and performance, then the analyst may proceed by first sorting out
the most successful of the competing projects in terms of sales per square foot, and then identifying their
advantages over the other projects. Also, the use of the competitive differentials technique can help in this
case to better understand/​quantify qualitative and other differences and how they relate to performance
indicators.

284
 Analyzing the market for retail space

Revise the development idea

The information obtained from the previous step is very important because it can highlight which aspects
and attributes of the preliminary plan are inadequate compared to the competition and point to specific
plan, design, and retail program modifications needed to remedy those shortcomings.

Stage 4: Shopping pattern analysis

This stage represents a considerably more detailed analysis of consumer shopping patterns than the
one carried out during Phase II, as it focuses exclusively on the consumer groups within the planned
development’s trade area and involves primary research.The underlying objective of shopping pattern ana-
lysis is to better understand the shopping choices and patronage preferences of these consumer groups and
how their spending patterns are linked to retail development attributes. Such information will help more
specifically determine what type of retail development and tenant mix will be most in demand at a given
site and what types of consumers are more likely to be attracted to it. To this aim, this stage includes the
following steps:

1. Identification and analysis of shopping choices and spending patterns across different socioeconomic
groups within the trade area through consumer surveys.The information collected in this stage focuses
specifically on:
a) Who are the potential customers (detailed demographics)
b) What each socioeconomic group purchases and where
c) The frequency of purchases and reasons behind specific shopping choices
d) The amount spent on different types of purchases in the different shopping destinations
e) Shopping trip characteristics (travel cost and time).
2. Analysis of the link between consumer choices, consumer characteristics (location, income and demo-
graphics), and the location and other traits of competing projects.

The consumer survey (which has been discussed more extensively in the previous chapter) is the major
step in this stage and one of the most crucial elements of a retail market study. If this survey is carried
out systematically to include samples of all income groups (segmented along reasonable ranges of income
categories) in each residential zone within the development’s trade area, it can allow for the application
of advanced probabilistic modeling for the estimation of the project’s capture rate in the next stage of
analysis.
It is important to emphasize that, before proceeding to the next stage, which involves a more elaborate
assessment and evaluation of the development scenario, it may be prudent to go back to the development
concept/​plan and see if any modification is needed, given the results of this analysis.

Stage 5: Evaluating the development plan

During this stage, the emphasis shifts away from the trade area and to the specific development under con-
sideration. All information collected and analyzed in previous stages is synthesized in this stage in order to
assess and evaluate the viability of the planned retail development in terms of demand and achievable rental
rates. Such evaluation involves the following five steps (also described in Figure 12.5):

1 . Estimation of the development’s likely capture rate

2. Estimation of the sales potential by line of trade

285
 Analyzing the market for retail space

STAGE 5 Evaluate Methodology

the Development
Concept 1. Estimate the development’s likely
capture rate:

• Huff’s probability approach

• Advanced probability modeling
(probability of shopping visits
is estimated as a function
of facility attributes, trip
characteristics, and consumer
characteristics)

2. Estimate sales potential by product

line:

Purchasing Power
× % Spent on Product Line
× Capture Rate

3. Estimate supportable square footage

by line of trade:

Sales Potential / Sales per Sq. Ft. Norm

4. Estimate achievable space rents using:

• The simple accounting approach

• The hedonic approach

5. Assess the development scenario by:

• Comparing supportable space to the

development scenario’s allocated
space
• Evaluating whether potential sales
revenues are sufficient to fully cover
each retailer’s costs

Figure 12.5  Major steps involved in Stage 5 of Phase III

286
 Analyzing the market for retail space

3 . Estimation of supportable square footage by line of trade

4. Estimation of achievable retail space rents
5. Re-​assessment of the development scenario.

Below we elaborate on each of these steps.

Step 1: Estimating the development’s likely capture rate

As is the case of the other property types, one of the major objectives of the microeconomic analysis is the
estimation of the capture rate of the specific project under consideration. In the case of retail developments,
this capture rate needs to be estimated because it is very unlikely that the trade area of a new retail devel-
opment (especially if it is relatively large) will not overlap with the trade area of other retail centers that
already exist within the same metropolitan area.8 Within such a setting, accurate assessment of a retail
development’s sales potential requires a reliable estimate of the project’s capture rate, as reflected in the
percentage of households likely to be attracted to the site (as opposed to competing retail developments).
Depending on the scope and budget constraints of the study, two broader methodologies can be used to
estimate a retail development’s capture rate:

1 . Analysis of similar projects

2. Probabilistic approaches, attempting to quantify patronage probabilities from the different zones that
compose the project’s trade area.

The first methodology represents, in essence, the competitive differentials technique discussed extensively
in Chapter 9. According to this approach, an estimate of the project’s likely market share is derived by
adjusting the share/​performance of competitive retail developments by a factor reflecting the planned
development’s strengths and weaknesses. In assessing the project’s strengths and weaknesses against the
competition, the analyst needs to prudently use the conclusions of the elaborate analysis of consumer
shopping patterns in Stage 4 as they pertain to the attractiveness of the project vis-​à-​vis the competing
developments.
In applying this methodology, the analyst should pay attention to two very important issues. First, since
no two retail projects can be exactly similar in all respects (micro and macro location, site, access, trade area,
structure and development attributes, tenant-​mix, competition, etc.), the analyst needs to very thoroughly
assess the differences of the project under consideration from each “comparable” examined and carefully
take them into consideration when estimating its penetration rate and sales potential. For example, a com-
parable retail development with exactly the same project characteristics will have a higher market penetra-
tion and sales if it faces less fierce competition than the subject does. Careful evaluation of the differences
of “comparables” from the project under consideration is even more critical in the case of large distinct retail
developments for which the comparables used are often located in different metropolitan areas or regions
of the country.
The second issue that needs to be examined when using the experience of comparable projects to
estimate the project’s capture rate is future competition and changes in travel patterns. Since the market
study for a planned retail development is carried out years before the project is completed, the experience
of competitive retail projects can provide clues for the project’s capture rate only at the time of analysis.
Such an estimate, therefore, may not accurately reflect the project’s capture rate at the anticipated time of
its completion, and it should be viewed only as a starting point that needs to be adjusted for future changes
that may influence the project’s relative performance. Such changes include anticipated completions of
new retail developments or expansions of existing centers that are bound to take sales away from the
project’s trade area. In addition, the effect of anticipated changes in the transportation system, which may

287
 Analyzing the market for retail space

Center k=1

Competing Developments, k
Neighborhood j

Figure 12.6  Probabilistic model setting

alter traffic counts on the transportation axis the project is located on, should be carefully evaluated and
incorporated in the forecast of the project’s capture rate. Some additional factors that need to be considered
when attempting to produce long-​term forecasts of the project’s capture rate include potential changes in
consumer preferences and retail formats, which may render the development obsolete and less attractive.
Such effects, however, may be very difficult to quantify.
The probabilistic approaches to capture rate estimation do not present the issues just discussed, because
the differences in center characteristics, future competition, and changes in travel patterns can be gener-
ically and systematically incorporated into the model. Below we discuss in more detail two alternative
approaches that can be used for the estimation of the probability of consumer visitations to a retail center:

1 . The gravity approach

2. Advanced probabilistic modeling.

HUFF’S PROBABILITY MODEL  Huff ’s probability model represents a somewhat more sophisticated version
of the simple gravity model presented by Reilly (1929). In order to apply this model, the trade area can be
thought of as a collection of neighborhoods, j, and competing retail centers (see Figure 12.6). Within such
a setting, Huff ’s formulation postulates that the probability that a consumer located in neighborhood j will
travel and shop at a specific center k can be estimated as:

Sk
T jkλ
Pjk = (12.8)
m S 
∑  T kλ 
k =1 jk

where
Pjk : the probability that a consumer living in neighborhood j will shop at a retail center k
Sk : the size of retail center k
Tjk : the travel time from neighborhood j to retail center k
m : the number of all competing retail centers (including the center under consideration)

288
 Analyzing the market for retail space

λ: a parameter capturing the negative effect of travel time on shopping trips and can vary across the
different lines of retail trade (for example, Huff uncovered λ values of 3.19 for furniture and 2.72 for
clothing)

If, for example, the estimated Pjk is 0.8, it implies that 80% of the households residing in residential zone
j will be shopping at center k. In Huff ’s capture rate formula, a greater λ implies a greater friction or
disamenity from travel and, therefore, a smaller probability that the consumer will travel to the center.
Although the relative simplicity of Huff ’s probability formulation may be appealing, it presents several
shortcomings:

1. Unless there are empirical estimates by line of retail trade for the area within which the project is
located, the value of λ needs to be assumed.
2. The effect of consumer characteristics such as income, age, etc. on patronage probabilities is ignored
completely by the formulation.
3. The effect of numerous shopping center characteristics other than distance and size, such as image,
tenant mix, and appeal, amenities, etc., is also ignored in determining Pjk.

If the total amenity indexes for the project under consideration and the competing centers (estimated as
explained in Chapter 9) are available, they should be entered in formula (12.8) instead of center size (Sk).
Total amenity indices are more appropriate indicators than just project size because they take into account
all center characteristics (including size) that could play an important role in determining sales potential.

ADVANCED PROBABILITY MODELING: THE MIT-​CAMBRIDGE SYSTEMATICS MODEL  A more sophisticated

approach, which involves advanced probability modeling, allows the estimation of a retail project’s capture
rate, taking into account the differential effect of the multiple factors that affect consumer shopping
choices.These factors include a host of center attributes (including travel costs from different locations) and
consumer characteristics. Empirical application of such a modeling approach to the Boston metropolitan
area has shown that taking into account these factors does indeed help more accurately assess a retail
center’s capture rate (Weisbrod et al., 1984; DiPasquale and Wheaton, 1996).9 This approach can be very
useful in retail market analysis because it allows the analyst to forecast a planned or an existing retail center’s
capture rate, taking into account anticipated changes in population characteristics, travel patterns, and the
effect of new competition entering the marketplace.
A model formulation reflecting this approach has been developed and copyrighted by the MIT Center
for Real Estate and Cambridge Systematics.To better understand this formulation, consider a metropolitan
area encompassing several competing shopping centers, and composed of many residential zones that house
a number of different consumer groups with given socioeconomic characteristics. In addition, since it can
be safely assumed that retail sales exports or imports at the metropolitan area level are minimal, consider
that the shopping center choice problem is essentially a zero-​sum game independently of how purchases
are distributed across the different centers. How can the capture rate and sales potential of an existing or a
planned retail development, k, with given characteristics, X, be estimated within such a setting?
The basic analytical concept is to estimate the probability Pijk that a consumer residing in a specific neigh-
borhood j and belonging in consumer group i, which is defined by specific socioeconomic characteristics
Z (primarily income level Y), will choose to visit a specific center with given attributes, X. Estimates of
such probabilities for each neighborhood or residential zone within the project’s trade area can then pro-
vide the basis for calculating total potential sales, E, at center k, using the formula:

Ek = Σi Σj (HHij Yi ei Pijk) (12.9)

289
 Analyzing the market for retail space

where HHij is the number of households (consumers) in group i residing in zone j, and e is the percentage
of the income they normally spent at center k.
The main proposition behind the modeling approach is that consumers facing a number of shopping
alternatives make shopping decisions on the basis of the utility they obtain by visiting each alternative.
Being utility maximizers, consumers choose to shop at the center providing the highest utility. Thus, the
starting point of the analysis involves the examination of the factors that influence the shopping choices
of consumers. Consumer utility, as it relates to shopping, is a function of three sets of factors (Weisbrod
et al., 1984):

1 . The characteristics of the shopping center

2. Travel time and cost
3. The characteristics of the consumer.

Shopping centers may differ in a number of ways, such as size, tenant mix, parking availability, pricing,
design features, and location.The latter determines the cost and time of travel from the different residential
zones to a given center.
The attractiveness of the various shopping center attributes, including travel time and cost, should vary
across households depending on their socioeconomic characteristics. For example, empirical estimates of
shopper utility parameters by income group indicate that in the case of lower-​income households, travel
time has a greater negative effect on shopping-​center choice, while the number of discount and variety
stores within a center has a greater positive effect (DiPasquale and Wheaton, 1996).The influences of these
factors on consumer utility are reflected in (12.10) where utility levels, U, are assumed to depend on center
characteristics, X1 through Xn (including travel time from a given zone or neighborhood j to a specific
center k). Notice that in this formulation the “weight” of each center characteristic, X, is represented by
the expression (ai+biZ) in order to allow variation of the effect of each center characteristic according to
the income and other socioeconomic characteristics of the consumers (Z):

Uijk = (a1+b1Z)X1k + (a2+b2Z)X2k + (a3+b3Z)X3k + ... +(an+bnZ)Xnk (12.10)

where:
a1, ...an : parameters to be estimated
b1, ...bn : parameters to be estimated
Z : income and other socioeconomic characteristics of the consumers
X1k,...Xnk : characteristics of center k (including cost and time of travel from each of the different
residential zones)
(an+bnZ) : marginal utility of center characteristics that vary depending on Z

The parameters in (12.10) measure the effect of the different center attributes (Xs) in attracting consumer
visits from different household groups and residential zones. For example, the term (a1+b1Z) measures the
marginal utility attached to center attribute X1 (and therefore its impact on the probability of center visitations)
by households with characteristics Z.The underlying premise of (12.10) is that patronage percentages should
vary across centers with different characteristics because of the differential utilities enjoyed by the different
types of consumers. Shopping centers offering higher utility to consumers should enjoy higher patronage
rates, while centers viewed less favorably by the consumers should experience lower patronage rates.
According to Dipasquale and Wheaton (1996), with appropriate transformations (explained in
Appendix 12B) the effect that various center attributes and travel costs have on the shopping behavior of
each socioeconomic group can be quantified by estimating equation (12.11).10 This equation postulates
that the logarithm of the ratio of the odds that a consumer will visit center k as opposed to center L

290
 Analyzing the market for retail space

depends on how the attributes of the former compare to the attributes of the latter and the utility
parameters attached to these attributes

ln(Pijk/​PijL) = ai1 (X1k –​X1L) + ai2 (X2k–​X2L)+ ... + ain (Xnk–​XnL) (12.11)

Formula (12.11) represents a linear regression model and can, therefore, be estimated empirically through
OLS, using data on the frequency of consumer visits at competing centers, shopping trip characteristics,
center attributes, and consumer characteristics.11 Information on consumer characteristics and their
shopping choices in the area of analysis requires primary research, and particularly consumer surveys.
Information on travel time and cost from the different residential zones may be available from the census
or local planning departments. Finally, information on the attributes of competing centers may be available
from local brokers or may need to be collected through field surveys.
Information on frequencies of consumer visits, collected through the survey, can be used to calculate
patronage probabilities from each zone for each center. These patronage probabilities, in turn, can pro-
vide the basis for calculating the dependent variable of (12.11) as the log of the ratio of the probability
of patronage, Pijk, of an arbitrarily selected center k over the patronage probability, PijL, of each and every
other competing center L (L=2,...,m) within the market under consideration.12 Notice that if, for example,
data are available for four consumer groups, four different series of ratios of patronage probabilities across
the different centers can be created (one for each consumer group). The independent variables can be
constructed by taking the differences between the attributes of center k and the respective attributes of
each and every other competing center L.
Empirical estimation of the coefficients of (12.11) can help quantify the patronage probabilities for a
center given its characteristics, and its location. These probabilities represent essentially capture rates that
can be used for the calculation of potential consumer expenditures at each center using (12.19). The spe-
cific steps for applying this approach include:

1. Divide the metropolitan area or trade area under consideration into residential zones that are homo-
geneous in terms of their socioeconomic characteristics.
2. Collect data on household income and demographic characteristics in each zone (which can be used
in combination with the survey-​based data on spending patterns collected in step (5) to estimate the
total purchasing power of each zone).
3. Identify and survey the competing retail centers in the area and their attributes, such as square footage,
number of parking spaces, number of discount stores, number of anchor department stores, etc.
4. Estimate, or collect data on, travel time and cost from each residential zone to the zone each com-
peting center is located within. Such data can often be obtained from local planning departments.13
5. Complete interviews with random samples of consumers residing in each zone, in order to establish
information regarding their shopping choices and expenditure patterns.
6. Calculate the dependent variable as the log ratio of patronage frequencies, ln(Pijk/​PijL), using the
information on consumer visitations to the different centers.
7. Estimate the parameters of (12.11) for each income or consumer group.
8. Obtain forecasts of the explanatory variables included in (12.11) that take into account new com-
peting retail developments expected to enter the marketplace, as well as anticipated changes in travel
time and costs due to changes in the area’s transportation system.
9. Use the parameter estimates of (12.11) and the forecast values of the explanatory variables in (12.11)
to estimate first utilities for each center using (12.10) and then patronage probabilities or capture rate
for the center under consideration using (12.11).
10. Apply estimated capture rates to forecast retail expenditures expected to be captured from each zone
upon project completion.

291
 Analyzing the market for retail space

MODEL APPLICATIONS  The model can provide the basis for answering some crucial questions in retail
market research such as:

• Forecasting the capture rate and potential sales of a planned center, taking into account its specific
characteristics and anticipated changes in the distribution of the population by income and other con-
sumer characteristics, as well as expected changes in the transportation network, travel costs, and the
competitive landscape; the estimated sales performance can then be evaluated against the performance
of competing centers.
• Analyzing how the location of alternative sites may affect the sales of a planned retail development, if
the site is not given.
• Predicting the capture rate and potential sales of an existing center, given its characteristics and
anticipated changes in the area’s demographics, travel patterns, and competitive landscape.
• Assessing what type of expansion (such as an additional department store, discount store, etc.) or modi-
fication is likely to have the most beneficial effects on the capture rate and sales of an existing center.
• Assessing which of the area’s existing shopping centers would benefit the most from a major addition,
such as a department store.

Step 2: Estimating sales potential by line of trade

The development’s capture rates from the different zones and consumer groups, Pijk, estimated from Step
1, can provide the basis for estimating its sales potential by product line using (12.10). Simply put, the sales
potential of center k equals the sum of potential sales expected to be captured from all neighborhoods and
all consumer groups within the center’s trade area. As such, its calculation requires two steps:

1. Calculate potential sales for a given line of trade for each consumer group i and each zone j (Sij), as
the product of the number of all households in each group i and zone j (HHij) times their average or
median income (Yi), times the percentage of income spent for in-​store purchases of goods in the par-
ticular line of trade (ei), times the probability that households in group i and location j will visit the
development under consideration (Pijk).

Potential Sales = No. of Households × Average Income × % Spent × Probability of Patronage

(Sij) = (HHij) * (Yi) * (ei) * (Pijk) (12.12)

2. Sum these sales estimates across all consumer groups i and zones j.

Red flag: In applying the above formula, the analyst needs to ensure that ei reflects the share of average household income
spent on the particular line of trade. Otherwise, if the figures used for ei reflect the share of a different type of income, then that
same type of income should be used in the formula above.

One of the most crucial inputs entering the above calculations is the predicted number of households in
each income category and zone. The reasonableness and feasibility of population/​household forecasts used
in estimating future sales potential should be very carefully evaluated, especially in the case of small retail
developments (such as neighborhood and community centers) with small primary trade areas, extending
only a few miles away from the site. Barring the presence of unusually high vacancies in the local housing
sector, strong population/​household growth is not possible without a proportionate increase in the area’s

292
 Analyzing the market for retail space

Table 12.5  Applying Step 2: Estimation of potential furniture sales at Center K

Residential Midpoint of income Total number of Estimated capture Percentage of Potential furniture sales to
zone (j) group i (Yi) households in group i rate from each group income spent on each group i in each zone j
and zone j i (Pijk) furniture by each (2)*(3)*(4)*(5)
group i (ei)
(1) (2) (3) (4) (5) (6)
1 $45,000 1,000 20% 3.2% $288,000
1 55,000 2,000 35% 3.8% 1,463,000
1 70,000 3,000 15% 4.1% 1,291,500
2 45,000 1,500 35% 3.2% 756,000
2 55,000 4,000 50% 3.8% 4,180,000
2 70,000 1,000 25% 4.1% 717,500
Total potential furniture sales at Center K $8,696,000

housing stock. Thus, strong population/​household growth in small trade areas may not be feasible at all
simply because they may be already densely developed with minimal or no vacant land available for new
residential development. Even if there is some vacant land, new residential development may be hampered
by growth moratoria and/​or other restrictive and anti-​g rowth zoning laws and local government pol-
icies. In the case of trade areas encompassing suburban or outlying areas, aggressive population growth
forecasts should be evaluated against the capacity of local governments to provide the required infrastruc-
ture support systems for rapid residential development.
Table 12.5 demonstrates the calculation of sales potential for a hypothetical furniture store at a center
k. For ease of illustration only two residential zones and three income groups have been assumed, but the
analysis could very well include both a greater number of zones (since furniture stores have typically larger
trade areas) and a greater number of income groups.
Whichever approach is used for estimating the project’s capture rate (comparables, gravity approach,
or advanced probabilistic modeling), the final estimate of the development’s total potential sales should
be evaluated against other data in order to test its reasonableness. For example, total predicted sales can be
converted to sales per square foot of the planned development, sales per square foot of an average-​size store,
or sales per capita and then compared to the experience of competing centers or relevant industry norms.
This suggests that the analyst may be better off using a variety of techniques for estimating the project’s
capture rate or potential sales so that each estimate can provide a point of reference for evaluating the rea-
sonableness of the alternative.

Step 3: Estimating supportable square footage by line of trade

The estimates of potential sales by product line (obtained from Step 2) can provide the basis for calculating
the supportable square footage using industry standards or sales per square foot norms:

Supportable Sq. Ft. = Sales Potential /​Sales per Sq. Ft. Norm (12.13)

To demonstrate the application of (12.13) we continue with the example presented in Table 12.5 and
assume that center k represents a community shopping center. We also assume that the prospective tenant’s
store productivity is among the top ten percent in the industry. According to the Dollars and Cents of
Shopping Centers: 1997 (ULI, 1997), the sales per square foot for the top ten percent of furniture stores in
community shopping centers is $276.21.14 The supportable square footage for the product line and tenant
in question can then be calculated as:

293
 Analyzing the market for retail space

Supportable Sq. Ft. for Furniture Store = 8,696,000 /​276.21 = 31,483

The estimate of the supportable square footage is the key in assessing a tenant’s viability at a given location
(we elaborate on this in the discussion of Step 5). In the case of a shopping center with multiple tenant/​
product lines, obviously the analyst needs to estimate supportable square footage for each tenant/​product
line included in the retail program, and then sum up the individual estimates in order to calculate the total
square footage of retail space that should be provided at the center.15

Step 4: Estimating achievable retail space rents

After estimating the supportable square footage by line of trade, the analyst can proceed with the estima-
tion of the rental rates that can be charged to the different tenants. Owners of retail space typically require
from their tenants two types of rent payments: a base rent, representing a specified dollar amount, and a
percentage rent representing a specified percentage that is applied to the tenant’s gross sales once they reach
or exceed a certain level.16 Furthermore, owners of retail space tend to require from anchor tenants con-
siderably lower rents than those they require from non-​anchor tenants, because of the significant contribu-
tion of the former to the development’s overall attractiveness and sales. In estimating achievable rents for a
retail development, therefore, the analyst needs to distinguish between anchor tenant space and non-​anchor
tenant space.
Achievable rents for each of the two types of retail space can be estimated either through accounting
techniques or through hedonic valuation techniques. Whichever methodology is used, the analyst needs
to examine how the estimates compare with actual rental rates charged at comparable centers operating
within the project’s trade area. Thus, the analyst is advised to use some or all of the techniques discussed
below in order to produce alternative estimates, which can provide a basis for cross-​checking their reason-
ableness. Furthermore, the analyst needs to assess the extent of rent discounts that should be offered to the
anchor tenants given their expected contribution in expanding the development’s sales potential.

ACCOUNTING TECHNIQUES FOR ESTIMATING RETAIL SPACE RENTS  We can distinguish two types of
accounting approaches that can be used for the estimation of the rental rates a retail development may
command:

1 . The competitive differentials approach

2. Mathematical approaches focusing on the threshold levels (maximum or minimum) of the rates that
can be charged to non-​anchor and anchor tenants.

The competitive differentials approach can be applied in the same way it is applied in the case of residen-
tial property analysis for estimating the rent a project may command in the local marketplace. In particular,
the analyst needs to collect information on rental rates charged for anchor and non-​anchor tenant space
in competing retail developments, along with information on sales per square foot and all attributes that
may contribute to each comparable’s performance. Subsequently, the analyst can develop total amenity
indices for each comparable and the subject property, calculate relative amenity indices, and use the latter
as a basis for calculating the rental rate it should command relative to the rates charged by the comparable
developments. Since this approach and its application have been discussed extensively in Chapter 9, this
section will focus on the discussion of the second accounting approach.
The mathematical approaches focus on:  1) the estimation of the maximum rent that non-​anchor
tenants can afford to pay given expected sales, and 2)  the minimum rent that anchor tenants could be
asked to pay given their expected contribution in terms of improving the sales of non-​anchor tenants. It

294
 Analyzing the market for retail space

should be emphasized at the outset that these approaches do not produce estimates of market rents, but only
threshold levels of rental rates that can be charged. As such, estimates using this methodology should not
be used without being evaluated against estimates using the competitive differentials or hedonic valuation
techniques in order to assess their reasonableness. This approach can be helpful in forecasting how the
project’s market rent (as estimated at the time of analysis using the competitive differentials or hedonic
valuation techniques) may move during the forecast period.This approach may also provide helpful insights
for developing asking rent strategies for the planned project.

ESTIMATING MAXIMUM NON-​ANCHOR TENANT RENTS  The maximum rent that can be charged to a non-​
anchor tenant can be derived as the residual profit using the simple break-​even condition. According to
this condition, the maximum amount that non-​anchor tenants can pay for occupying retail space at a given
center k can be calculated as the residual profit, that is, the amount remaining after all other components
of the retailer’s costs (that is, the wholesale cost of the merchandise and operating expenses) are deducted
from sales, as in (12.14):

Max. Retail Rent/​Sq. Ft. = [(Sales/​Sq. Ft) × Markup] –​Operating Costs/​Sq. Ft. (12.14)

In the above formula, markup (also referred to as profit ratio) is the percentage difference between retail
price and wholesale cost, while sales per square foot represent either the estimated sales per square foot
(that is, the ratio of the expected sales over the planned square footage) or the sales-​per-​square-​foot norm
of the store in question. Obviously, the data needed for carrying out this calculation include, besides the
expected sales per square foot for the store under consideration, the retailer’s operating costs per square
foot and the typical markup for the line of trade under consideration.This information can be obtained by
surveying retail industry experts.

ESTIMATING THE MINIMUM ANCHOR TENANT RENT  Anchor tenants generate significant externality
effects, as they contribute towards increased sales at other stores co-​locating in the same development, thus
enhancing the center’s total sales.Therefore, the rent anchor tenants pay should be reduced by the extent of
their contribution to total sales. In principle, the minimum rent that an anchor tenant can be asked to pay
should be such that the weighted average rent for the whole development with and without the anchor
tenant are equal. This condition is described by (12.15):

WRA = WRNA (12.15)

where
WRA : weighted rent/​sq. ft. with the anchor tenant
WRNA : weighted rent/​sq. ft. without the anchor tenant

Obviously, the key to applying this approach is figuring out how much lower the development’s sales
would be in the absence of an anchor tenant.The MIT-​Cambridge Systematics model, which allows taking
into account explicitly center characteristics, such as tenant mix, can help estimate sales potential with and
without an anchor tenant.
To understand how this framework can be used specifically for the calculation of the minimum rent
a retail center owner would be inclined to charge an anchor tenant, consider the following hypothet-
ical example of a shopping development with a well-​known anchor tenant likely to occupy 50% of the
development’s total GLA and smaller stores likely to occupy the remaining 50%. For the purposes of this
example let’s make the following assumptions regarding the smaller stores’ performance in the presence
and absence of the anchor tenant:

295
 Analyzing the market for retail space

• In the absence of the anchor tenant, non-​anchor tenant sales per square foot average $78, while their
operating costs per square foot average $30.
• In the presence of the anchor tenant, non-​anchor tenant sales per square foot are expected to increase
by 15%, while their operating costs per square foot are assumed to remain the same.
• The average markup or profit ratio of the smaller stores is 50% under both scenarios.

In order to apply the condition expressed by (12.15), two alternative development scenarios need to be
considered:

Scenario 1: Assumes that a fraction α (in this case α=50%) of the development is allocated to the anchor
tenant and a fraction (1-​α) to non-​anchor tenants.
Scenario 2:  This is a hypothetical scenario that assumes that all of the development’s square footage is
allocated to non-​anchor tenants.

The underlying rationale for considering these two scenarios is to calculate the maximum rental rates that
non-​anchor tenants can pay on average, in the presence (Scenario 1) and in the absence (Scenario 2) of
the anchor tenant (denoted below as RNA and R’NA, respectively). Given the aforementioned assumptions
regarding average sales and operating costs in the presence and the absence of the anchor tenant, (12.15) can
be used for carrying out these calculations as indicated below:

RNA = (78 * 1.15 * 0.50) –​30 = $14.85/​sq.  ft.

R’NA = (78 * 0.50) –​30 = $9/​sq.  ft.

Common sense suggests that RA, the minimum rent the anchor tenant could be asked to pay, should be
such that the developer remains indifferent between the two scenarios (including and not including the
anchor tenant in the development). This implies that (12.15) should hold.

(Weighted) Rent/​Sq. Ft. from Scenario 1 = (Weighted) Rent/​Sq. Ft. from Scenario 2

Given that α percent of the development’s square footage will be occupied by the anchor tenant
and (1-​α) percent of the development will be occupied by the non-​anchor tenants, then (12.15)
translates to:

RA α + RNA (1-​α) = R’NA

 RA α = R’NA –​ RNA (1-​α)

 RA = [R’NA –​ RNA (1-​α)] /​ α (12.16)

Given RNA= $14.85/​sqft, R’NA= $9/​sq. ft., and α=0.50,

RA = [9–​(1–​0.50)*14.85] /​0.50 = $3.15/​sq.  ft.

For further illustrating the usefulness of this analysis in project planning, assume that the anchor tenant is
not willing to pay more than $1.50 per square foot. How would the developer deal with this situation?
Can he modify in any way the development scenario to accommodate the anchor tenant’s requirement?
Looking at (12.16) we can see that, if we consider the anchor tenant’s rent, RA, as known, we can mathem-
atically derive a different fraction of the development’s total square footage, a’, to be allocated to the anchor

296
 Analyzing the market for retail space

tenant so that the indifference principle still holds. In such case, a’ is unknown, while RA = $1.50/​sq. ft.,
RNA= $14.85/​sq. ft., and R’NA= $9/​sq. ft. Solving (12.16) for a’ yields how the anchor tenant’s percentage
space allocation could be altered to accommodate its maximum rent threshold:

a’ = (R’NA –​ RNA) /​(RA –​ RNA)= 0.44 (12.17)

HEDONIC VALUATION TECHNIQUES  As in the case of residential market analysis, econometric hedonic
valuation techniques provide a powerful tool for estimating the market rent a retail development may
command. In applying this technique, the analyst will need to use a sample of recent retail lease transactions
and information on the characteristics of the leases and the space associated with those transactions. Once
such information is collected, then retail rents can be modeled as a function of lease characteristics (Xl),
location characteristics (XL), tenant mix (Xm), and other shopping center traits (Xs) expected to affect tenant
profit and the rent that a tenant may be willing to pay.

R = f(Xl, XL, Xm, Xs) (12.18)

For a more detailed discussion of hedonic valuation techniques, and how they can be used to estimate pro-
ject rent, see Chapter 9 and Chapter 14.

FORECASTING RETAIL SPACE RENTS17  Since the ultimate objective of the analysis in most market studies
is to assess the performance of the market and the project under consideration, the analyst needs to
produce a forecast of the rental rates the project will command at the anticipated time of completion.
Both competitive differentials and hedonic valuation techniques provide the rental rate the property may
command at the time of analysis. The analyst needs to use that estimate as the starting level and forecast
most likely changes in those levels over the forecast period. Whichever technique is used by the analyst to
project those levels forward, the final rental rate forecast needs to be ascertained by the adequacy of the
predicted sales per square foot and the tenant’s ability to pay the predicted rates, taking into account its
operating costs.
The mathematical approach for estimating the maximum rent that can be afforded by non-​anchor
tenants does take into account the predicted sales per square foot and the retailer’s operating costs. As
such, predicted trends in these rates can provide important clues with respect to future movements of the
project’s rents. For example, if the estimated maximum rent afforded by the tenant is predicted to increase it
would suggest that forecast sales per square foot are expected to grow more quickly than the retailer’s other
operating costs, enabling the retailer to pay a higher rent. If that is the case, this analysis would provide some
basis for anticipating rising rents between the time of analysis and the project’s completion. Notice that the
gravity approach, as well as the advanced probabilistic methodologies that can be used for the estimation of
a retail development’s capture rate, can directly account for any anticipated increases in competitive supply.
Thus, sales per square foot estimates generated using these techniques should reflect any potential down-
ward pressures on rental rates from the supply side.
An alternative approach for forecasting retail space rents is to develop an econometric model for the
broader market similar to the one described in the case of the residential market. However, the availability
of retail lease transaction data by type of center is very limited, making any econometric endeavor very
difficult. Furthermore, the applicability of metro-​wide rent index forecasts to a particular retail project
is questionable, due to the great diversity of retail formats and the localized nature of the smaller retail
developments and their trade areas. For this reason, industry professionals are often reluctant to use aggre-
gate metro-​wide indices in the analysis of specific projects.

297
 Analyzing the market for retail space

Step 5: Assessing the development scenario

The final step of Phase III of a retail market study is the assessment of the development scenario on the
basis of the estimated potential sales, supportable square footage, and achievable rents. In particular, the
market viability of each and every tenant in the development scenario can be assessed in two ways:

• By comparing estimates of supportable space to the space allocated to each tenant (which is typically
within an efficient-​store-​size range for the product line and type of center considered). If the support-
able space is greater than the planned square footage or, at least, the lower limit of the efficient-​size
range, then the analyst can conclude that market demand is sufficiently high to support the operations
of the tenant in question.
• By comparing estimates of potential sales revenues to the sales required to cover the retailer’s full oper-
ating costs (wholesale cost of goods, operating expenses, and rent). The minimum required sales to
cover the retailer’s cost can be calculated using (12.5), which represents the break-​even condition. If
potential sales revenues are sufficiently higher than the sales required for breaking even, then the tenant
can be considered viable from the market perspective. This methodology of evaluating the viability of a
retail development is more difficult to apply than the previous one because data on full operating costs
(including wages) for different types of stores may be more difficult to obtain than data on sales-​per-​
square-​foot norms and store efficient sizes.

Continuing the numerical example presented in Step 3, in assessing the viability of a furniture store at the
center in question, the supportable square footage (estimated using (12.15)) is compared to the efficient-​
size range for furniture stores in community shopping centers. This information was available from Dollars
and Cents of Shopping Centers before 2008, but currently it needs to be obtained through industry research
or other reliable industry source. If for the purposes of this example we assume that the efficient-​size range
for furniture stores in community shopping centers is between 12,000 to 48,000 square feet, it can be
inferred that market demand is sufficient to support the operations of a furniture store at the center under
consideration, since the estimated supportable square footage is 31,483.
Once the viability of each and every tenant in the planned development is asserted, the analyst needs
to evaluate whether the estimates of maximum rents (that presumably the tenants of the development will
be able to afford to pay given the estimated sales, operating expenses, and profit ratios) are comparable to
market rents. If they deviate significantly, then the analyst may want to reexamine the data and assumptions
used for the calculation of the maximum achievable rents, and whether the difference is justified by the
center’s considerably weaker or stronger than market-​average performance for the type of center or store
considered.

COMPARING ALTERNATIVE SCENARIOS  Simply identifying a viable retail development scenario is not
enough. Ideally, the developer or investor is interested in maximizing achievable rent/​sq. ft.Thus, alternative
viable development scenarios need to be compared and evaluated with that criterion in mind. Maximizing
sales per square foot for given operating costs and markup or profit ratios provides an equivalent criterion.

Chapter summary
In this chapter, we have discussed the three major phases of retail market studies. Phase I focuses on the
estimation of the unrealized sales potential and can help evaluate decisions of market entry in a specific area
and/​or product line. Such unrealized sales potential, or market vacuum, can be estimated by comparing
potential sales, justified by the income and other demographic characteristics of the area’s population,
to actual sales at the area’s retail establishments. Potential sales can be estimated using either accounting

298
 Analyzing the market for retail space

or econometric techniques. Econometric techniques represent a more sophisticated and comprehensive

approach, in that they allow the analyst to take into account the effect of more factors on spending patterns,
than the accounting techniques do.When applying such techniques, it is important that the area of analysis
is defined in a way that does not allow imports or exports of retail sales, if sales are used to proxy consumer
expenditures.
Phase II focuses on evaluating a given site and location. Such analysis needs to take into account
consumer shopping patterns as they relate to different types of retail goods and services, and particu-
larly as they relate to their frequency of use, as well as susceptibility to comparison and complementary
shopping. These factors are important, as they determine how far consumers are willing to travel for a
given product line and, therefore, the drawing power and locational requirements of the different types of
retail developments housing different types of stores. In evaluating a given site or location, a host of spatial
demand factors as they relate to population/​consumers, macro-​access, micro-​access, and the site’s imme-
diate micro-​environment need to be assessed. Furthermore, supply factors need to be considered as they
relate to the characteristics and spatial distribution of competing retail developments.
Phase III focuses on the evaluation of a specific retail development concept and a specific site, and it
comprises five interactive stages in the sense that conclusions from later stages may provide feedback to
the earlier stages, allowing continuous refinement of the development plan. The first stage of Phase III in
fact involves the refinement of the preliminary project design and retail program based on the insights
gained from the analysis of the unrealized sales potential in Phase I, and the assessment of the strengths and
weaknesses of the site and location under consideration in Phase II.
The second stage of Phase III focuses on the definition of the project’s trade area through either
non-​stylized techniques that use multiple criteria, or stylized methodologies such as the concentric ring
approach, the driving time approach, the “near-​twin” approach, and the simple gravity approach. The third
stage focuses on a detailed analysis of the competing retail developments expected to attract sales from
the project’s trade area. A major objective of the analysis of the competition is to link development per-
formance, as measured by sales per square foot, to development attributes. If data on sufficient number of
comparables are available, hedonic techniques can provide a means for more accurately assessing such links.
The final stage of Phase III focuses on the evaluation of the final development plan. The first step is
the estimation of the project’s expected capture rate. Advanced probabilistic approaches, such as the MIT-​
Cambridge Systematics model, provide the means for more accurately forecasting the project’s capture rate
taking into account expected changes in population characteristics, changes in travel patterns, and future
competition. Project capture estimates provide the basis for the estimation of the project’s sales potential
and supportable square footage by product line. Estimation of achievable rental rates provides the final crit-
ical market input for assessing the project’s viability.

Questions
1 . Discuss in general terms the three phases of a well-​integrated retail market study.
2. Discuss the purpose for estimating unrealized sales potential and list the steps required for carrying out
this task.
3. Describe how an area’s sales potential for a specific product line can be calculated using:  1) the
accounting approach, and 2) the econometric approach.
4. Discuss the shortcomings and advantages of each of the two approaches for estimating unrealized sales
potential.
5. How can unrealized sales potential estimates guide market-​entry decisions?
6. What product type characteristics affect consumer spatial behavior and how?
7. Discuss in specific terms the spatial demand factors that need to be considered when evaluating a
given site for retail development.

299
 Analyzing the market for retail space

8. Discuss the major locational requirements for neighborhood, community, and regional shopping
centers.
9. List the five stages of Phase III and the steps involved in Stage 5.
10. Discuss briefly the different methodologies for defining a retail project’s trade area.
11. List and discuss briefly the four steps involved in the analysis of competitive retail developments.
12. Discuss briefly how a retail project’s capture rate can be estimated using the competitive differentials
technique.
13. Discuss the MIT-​Cambridge Systematics model for estimating a project’s capture rate and potential
sales. In particular elaborate on:
a) The data required for estimating the model
b) The model’s dependent variable
c) The model’s explanatory variables
d) The use of the estimated parameters to forecast the project’s potential sales.
14. What are the advantages of the econometric approach in estimating project capture rate?
15. How can the supportable square footage by line of trade be estimated and how can such estimates be
used to evaluate the viability of a given tenant?
16. Discuss alternative techniques for estimating a retail development’s most likely rental rates.
17. Outline and carefully explain the main steps involved in conducting a market study for a regional mall
at a given site.
a) Start your discussion with the questions such a market study must address.
b) Outline the main methodological steps involved and the expected bottom-​line conclusion of
each step.
c) Expand on the factors that a developer should consider in his/​her attempt to optimize the
development’s sale potential.

Notes
1 The other two alternative sources of data (discussed in the previous chapter) can also be used for calculating the
percentage of income spent on specific product lines, but household disaggregation by these sources is even thinner.
2 The reader is referred to the previous chapter in which such factors are discussed in detail.
3 Sales data for broader lines of retail trade for metropolitan areas and counties are readily available from the Census
of Retail Trade, which is carried out every five years. The last census was carried out in 2012.
4 DiPasquale and Wheaton (1996) argue that frequency of purchase of a given good depends on its rate of use and
storage cost.
5 In evaluating the trade area’s population/​household growth prospects, it may be helpful to try and identify the
path of urban growth by looking at recent major residential and commercial developments, planned or under-​
construction projects, and planned expansions of transportation and infrastructure systems.
6 Added by the editor.
7 For a more detailed presentation of how this technique can be applied, see Clapp (1987).
8 Obviously, if the analyst is certain that the development’s trade area does not overlap with the trade area of any other
retail facility, and that there are no significant leakages of sales to other shopping destinations, no capture rates need
to be estimated, since in theory the subject development should capture all potential purchases within that area.
9 The underlying approach is discrete choice analysis, which has often been used to address urban issues, such as travel
mode choice, as well as household and firm location choice (DiPasquale and Wheaton, 1996).
10 This is the approach used in the application presented by DiPasquale and Wheaton (1996).
11 DiPasquale and Wheaton (1996) point out that if shopping choices of consumers are treated as discrete observations,
that is, 1 if a particular center is chosen and 0 if not, a maximum likelihood estimation technique can be used.
12 As DiPasquale and Wheaton (1996) indicate, the ratio of the patronage probabilities, Pijk /​ PijL, is equivalent to the
ratio of the total number of households in group i actually (or estimated to be) residing in zone j choosing k, VHijk,
over the total number of households in group i actually (or estimated to be) residing in zone j choosing L, VHijL.
Both VHijk and VHijL can be calculated from the raw survey data. For example, if PHijk is the number of surveyed
households in group i and zone j that chose center k, SHij represents all surveyed households in group i and zone j,

300
 Analyzing the market for retail space

and THij represents the total number of households in group i actually, or estimated to be, residing in zone j at the
time of analysis, then VHijk can be estimated as:
VHijk = THij (PHijk /​  SHij)
13 According to DiPasquale and Wheaton (1996), such data are compiled by these agencies for long-​range transpor-
tation planning purposes.
14 This publication has been cancelled since 2008.
15 Fanning, Grissom, and Pearson (1994) suggest that the estimated supportable square footage should be adjusted to
account for a normal vacancy rate in the market. If a normal vacancy rate of 5% is assumed (which is typically the
case), such an adjustment can be carried out by dividing the estimated supportable square footage by 0.95. Although
this point should apply when estimating total supportable square footage at the market level, per the editors’ view
it is debatable whether it should be applied when estimating supportable square footage for purposes of assessing
the viability of a particular development.
16 Additional charges, besides the ground and percentage rent, include typically common area charges (pertaining
to expenses for operating and maintaining the project’s common areas), property taxes, insurance, and utilities
(provided by the center).
17 Added by the editor.
18 This is the approach used in the application presented by DiPasquale and Wheaton (1996).

References and additional readings

Benjamin, J. et  al. 1990. Retail Leasing:  The Determinants of Shopping Center Rents. AREUEA Journal, 18:3,
302–​312.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Market Analysis for Retail Space. Chap. 9 in Real Estate Market
Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
Clapp, J. 1987. Market Capture. Chap. 13 in Handbook for Real Estate Market Analysis. Englewood Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W. Wheaton. 1996. Retail Location and Market Competition. Chap. 6 in Urban Economics and Real
Estate Markets. Englewood Cliffs, NJ: Prentice-​Hall.
Fanning, S., T. Grissom, and T. Pearson. 1994. Existing Shopping Center. Chap.  12 in Market Analysis for Valuation
Appraisals. Chicago, IL: The Appraisal Institute.
Ingene, C. 1984. Structural Determinants of Market Potential. Journal of Retailing, 60:1, 37–​64.
McMahan, J. 1989. Retail Commercial. Chap.  8 in Property Development. New  York:  McGraw-​Hill Publishing
Company.
Reilly, W. 1929. Methods for the Study of Retail Relationships. Austin, TX: Bureau of Business Research, University of
Texas.
Schmitz, A. and D. Brett. 2001. Retail Development. Chap. 5 in Real Estate Market Analysis: A Case Study Approach.
Washington, DC: Urban Land Institute.
Urban Land Institute (ULI). 1997. Dollars and Cents of Shopping Centers: 1997. Washington, DC: Urban Land Institute.
Urban Land Institute (ULI). 2008. Dollars and Cents of Shopping Centers: 2008. Washington, DC: Urban Land Institute.
Weisbrod, G., R. Parcels, and C. Kern. 1984. A Disaggregate Model for Predicting Shopping Area Market Attraction.
Journal of Retailing, 60:1, 65–​83.

Appendix 12A: Retail data sources

The most crucial inputs in retail market analysis involve data on population, households, income, and spending patterns.
Population projections for counties and large municipalities can often be obtained from state, regional, and local
planning agencies. For smaller areas such as combinations of zip codes or census tracts, such projections can be obtained
from private vendors. Every year, the Census Bureau provides population estimates and breakdowns by age, sex, and
race or ethnicity for metropolitan areas and large cities. Detailed household characteristics are only available from
decennial censuses.
Forecasts of population, households, and income, as well as more detailed information on consumer demographics
and household characteristics, including current and forecast retail sales by type of store and product line, can be
obtained from private vendors. Such data are available for metropolitan areas, counties, or even local trade areas defined
by zip codes, census tracts, or longitude and latitude. Below a brief (but certainly not exhaustive) list of major sources
of secondary data useful for retail market analysis is provided. Please see Chapter 18 for additional sources of retail data
and forecasts.

301
 Analyzing the market for retail space

Sources of secondary data for retail market analysis

On demographics, income, and effective buying power
U.S. Department of Commerce, Bureau of the Census, 1990. Census of Population and Housing. Washington,
DC: Government Printing Office. www.census.gov/​prod/​www/​decennial.html.
Population projections at national level by the U.S. Department of Commerce, Bureau of the Census. https://​census.
gov/​programs-​surveys/​popproj.html.
Employment and wages by industry and county from state departments of labor or employment/​labor commissions.
Other private vendors are listed in Chapter 19.

On expenditure patterns
Bureau of Labor Statistics. Consumer Expenditure Survey. Washington, DC: Government Printing Office. www.bls.gov/​
cex.​
The Consumer Expenditure Survey carried out by BLS is often used by retail analysts for identifying differential
spending patterns across income and age groups. The data are collected in independent quarterly Interview and Diary
surveys of approximately 5,000 sample households each. Data on household income and socioeconomic characteristics
are collected in each survey. The Interview survey data include out-​of-​pocket expenditures for housing, apparel and
services, transportation, health care, entertainment, personal care, reading, education, food, alcoholic beverages, tobacco,
cash contributions, and personal insurance and pensions. The Diary survey data include expenditures on frequently
purchased items such as food at home, food away from home, alcoholic beverages, tobacco products and smoking
supplies, personal care, nonprescription drugs, and housekeeping supplies.
The expenditure survey data are published in annual (calendar-​year and mid-​year) and biannual reports.The annual
reports provide tabulations of expenditures across broader lines of trade by income group, by age of household head,
by household composition, by education of household head, by housing tenure, by number of earners, by occupation,
by size of consuming unit, by region, etc. The biannual reports provide expenditure information across more refined
household groups, which are defined through different cross-​tabulations of retail expenditures. In particular, these
reports provide cross-​tabulations of expenditures by household age and income, by household income and size, by
household age and size, as well as expenditure composition for selected metropolitan areas.
Bureau of Labor Statistics. Relative Importance of the Components of the Consumer Price Indexes (RICCPI). Washington,
DC: Government Printing Office.
Provides data on percentage of income spent on very detailed expenditure categories at the national level. Provides
similar data for ten broader expenditure categories for 27 metropolitan areas and the four regions.
www.bls.gov/​cpi/​tables/​relative-​importance/​home.htm.

On establishment sales
Bureau of the Census. Economic Census.
The survey is carried out every five years (every year ending in two and seven) and provides data on establishment
sales for detailed NAICS sectors for the nation, states and metropolitan statistical areas, counties, and places.
www.census.gov/​programs-​surveys/​economic-​census/​data/​tables.html.
The statistics on the Retail Trade sector (NAICS sector 44–​45) provide more information and data on retail sales in
great detail by NAICS sub-​sector codes (up to seven-​digit codes) for the largest metropolitan areas and retail sales by
product line at the national and state level but not the metropolitan-​area level.
www.census.gov/​data/​tables/​2012/​econ/​census/​retail-​trade.html.
Bureau of the Census. Monthly Retail Sales.
Monthly sales at national level for different lines of trade with historical monthly data going back to 1992. This
page also includes the latest Quarterly E-​Commerce Report that provides total retail sales and e-​commerce sales for
the last five quarters.
www.census.gov/​retail/​marts/​www/​timeseries.html.
Bureau of the Census. Annual Retail Trade Survey.
Published every year with a two-​year lag. Provides estimates at the national level of sales, sales taxes, inventories,
purchases, total operating expenses, gross margin, gross margin as a percentage of sales, accounts receivable per capita
sales, U.S.  retail trade sales—​total and e-​commerce, and U.S.  electronic shopping and mail-​order houses (NAICS

302
 Analyzing the market for retail space

4541)—​ total and e-​ commerce sales by merchandise line. www.census.gov/​content/​census/​en/​data/​tables/​2017/​

econ/​arts/​annual-​report.html.
Sales tax receipts for counties, cities, and municipalities may be retrievable from state internal revenue department
websites. This information may be available by industry (NAICS code).

Appendix 12B: Further elaboration on the probabilistic approach used by

DiPasquale and Wheaton
If consumer shopping utility is influenced by some random elements, in addition to the factors described by (12.11),
then patronage probabilities should differ across centers. According to DiPasquale and Wheaton (1996), the patronage
probability, Pijk, that a consumer in group i residing in zone j will visit center k can be expressed as:

U ijk
e
Pijk = m
(12B.1)
∑e
U ijk

k =1

Notice that the denominator of (12B.1) accounts for all centers, m, within the market under consideration, so that the
sum of the patronage probabilities of all centers from all households residing in the market area adds up to 1. Given
(12.B2), the ratio of the probabilities that a consumer will choose center k over another competing center L will be
equal to:

U ijk
Pijk e
= U ijk (12B.2)
PijL e

Taking the logarithm of both sides of (12B.2) and incorporating the consumer utility expression in (12.11) yields the
following:

ln(Pijk/​PijL) = lnPijk –​ lnPijL = Uijk –​  UijL

= [(a1+b1Z)X1k + (a2+b2Z)X2k + (a3+b3Z)X3k + ... +(an+bnZ)Xnk]

– [(a1+b1Z)X1L + (a2+b2Z)X2L + (a3+b3Z)X3L + ... +(an+bnZ)XnL]

= (a1+b1Z)(X1k-​ X1L)+ (a2+b2Z)(X2k-​X2L)+ (a3+b3Z)(X3k-​  X3k)

+ ... +(an+bnZ)(Xnk-​ XnL) (12B.3)

Equation (12B.3) postulates that the logarithm of the ratio of the odds that a consumer will visit center k as opposed to
center L depends on how the attributes of the former compare to the attributes of the latter and the utility parameters
attached to these attributes. An alternative to estimating (12B.3) for all consumer groups is to estimate separately for
each consumer group, i, Equation (12B.4) in which consumer characteristics, Z, such as income, do not enter the right-​
hand side of the equation.This estimation procedure allows the quantification of the different effect that various center
attributes and travel costs have on the shopping behavior of each socioeconomic group.18 That’s why the coefficients of
(12B.4) below carry the subscript i, as they refer to specific consumer groups.

ln(Pijk/​PijL) = ai1 (X1k –​ X1L)+ ai2 (X2k –​X2L)+ ... + ain (Xnk –​ XnL) (12B.4)

303
 13
ANALYZING THE MARKET
FOR RETAIL SPACE
Synthesis and market studies

Introduction 305
Emphasis of retail market studies 305
Regional and super-regional centers 305
Community centers 305
Neighborhood centers 305
Strip centers 306
Discount and warehouse stores 306
Off-price centers 306
Outlet centers 306
Specialty centers 306
Evaluating retail market studies 306
Broader evaluation criteria 306
Common flaws of retail market studies 306
Improper definition of the market/trade area 307
Inaccurate data on spendable income and retail sales 307
Inappropriate assessment of supportable retail activity 307
Naive assumptions regarding future changes 307
Misinterpretation of “comparables” 307
Poor analysis of the risks pertaining to store and tenant failure 307
Use of very broad data sources 308
Retail development: a simple example 308
Project and market data 308
Time of entry 308
Retail GLA 308
Tenant clusters and sources of demand 308
Tenant data 309
Residential-based demand and trade area households 309
Job-based and visitor-based demand 310
Retail rent index 310
Analysis 310

304
newgenprepdf

Synthesis and retail market studies

Questions 312
References and additional readings 314

Introduction
This is the last chapter focusing on retail market analysis. As such, its purpose is to demonstrate how some
of the concepts, issues, and techniques discussed in previous chapters may relate to different types of
retail market studies, what their implications are in evaluating a retail market study, and how they may be
synthesized when analyzing a specific retail development idea.
Within this context, we first outline the major points of emphasis of market studies for different types
of retail developments and then discuss the major pitfalls and evaluation criteria for retail market studies in
general. Subsequently, we present a simple example of evaluating the viability of alternative retail programs
involving a small retail development, as a way of demonstrating the major pieces of information needed
and how they can be synthesized to carry out the task at hand.
It should be emphasized at the outset that the example presented is purely hypothetical and by
no means represents an actual or a complete market study. Some of the calculations presented in this
example, such as the estimation of expected sales volume, supportable square footage, and achievable
rents, constitute some of the most critical estimates that can be found in a retail market study. However,
the most important and major part of a retail market study would involve the detailed discussion, explan-
ation, and documentation of the sources, validity, accuracy, and limitations of the numbers used in these
calculations.

Emphasis of retail market studies

Market studies for retail developments typically follow the five stages of Phase III outlined in the previous
chapter. The emphasis of the analysis, however, varies across the different types of retail developments.
The different points of emphasis placed by market studies across the various types of retail clusters are
summarized below (McMahan, 1989).

Regional and super-​regional centers

• Sufficient demand within the project’s trade area, especially in terms of shopper goods.
• Strength, reputation, and drawing power of the anchor tenants.
• Tenant mix, variety, and quality of non-​anchor tenants.
• Layout and design providing a pleasant environment and a smooth pedestrian circulation pattern, as well
as facilitating complementary and comparison shopping.
• Transportation access and availability of parking.

Community centers
• Identification of market gaps in particular lines of trade and services to maximize capture rate.
• Proximity to large concentration of population, convenient access, and parking.

Neighborhood centers
• Market strength and reputation of the anchor tenant (supermarket or drug store).
• Convenience of micro access.

305
 Analyzing the market for retail space

Strip centers
• Careful selection of location by identifying spatial supply gaps for convenience goods (spatial competi-
tion is intense).
• Convenience of access and parking.

Discount and warehouse stores

• Demand for durables (usually household furnishings and general merchandise).

Off-price centers
• Local demand by moderate-​income households.
• Potential competition from close-​by community and regional centers.
• Competitive analysis focus on community and regional centers.

Outlet centers
• Regional demand by moderate-​and lower-​income households and sometimes tourists.
• Competitive analysis focus on regional and specialty centers attracting tourists.

Specialty centers
• Demand by high-​income households and/or tourists.
• Access to/​from tourist destinations, if targeted specifically for tourists.
• Location within an established shopping area, or close to a regional center, to benefit from its drawing
power, if local consumers are targeted.

Evaluating retail market studies

Broader evaluation criteria
The general criteria for evaluating market studies. These include:

• Clarity of objectives
• Robustness of methods
• Quality and adequacy of data
• Consistency of conclusions with analysis results.

Common flaws of retail market studies

Besides the above general criteria, a study by Cambridge Systematics provides some more specific guidelines
for evaluating retail market studies. In particular, a mimeo authored in 1984 by Weisbrod and Radov,
outlines the “seven deadly sins” most often encountered in retail market studies. Their major conclusion is
that the methods employed in many studies often result in overstatement of a project’s potential to capture
retail sales and, therefore, the project’s potential for generating rental income. Drawing heavily from this
article we elaborate on these seven deadly sins.

306
 Synthesis and retail market studies

Improper definition of the market/​trade area

Market/​trade areas are often overestimated in size due to the use of simplistic assumptions regarding travel
time and distance, made without sufficient consideration of transport and socioeconomic barriers, adverse
neighborhood demographics, and market penetration by competitors. A not necessarily bad shortcoming
of a number of retail market analyses uncovered by the Cambridge Systematics study is the underestima-
tion of the contribution of employment-​based shopping.

Inaccurate data on spendable income and retail sales

Most studies utilize the Survey of Buying Power, which, according to Weisbrod and Randov (1984), provides
estimates of “effective buying income” that is noticeably larger than disposable income available for retail
purchases, and reports retail sales for urban areas that may be considerably lower than those reported by
state sales tax collection records. As a result, many studies tend to considerably overestimate an area’s unreal-
ized sales potential, as they overstate spendable income and understate actual retail sales. It should be noted
that the Survey of Buying Power is no longer published, but the issue of using accurate data on spendable
income and retail sales remains a key issue in terms of the validity of the results of retail market studies

Inappropriate assessment of supportable retail activity

Many studies, even when they appropriately define a center’s trade area, tend to overestimate sales poten-
tial and, therefore, supportable square footage, because they fail to appropriately account for “leakages” of
expenditures to competing stores located in other trade areas. This is typically reflective of poor analysis of
the market penetration achieved by different retailers in the various communities in the trade area. Such a
flaw can be avoided by a more elaborate analysis of the market capture of competitive stores, by broadening
the center’s market area, and by estimating each development’s capture rate.

Naive assumptions regarding future changes

Very often, projections of future sales are made on the basis of naive assumptions regarding the future
that fail to take into account the effect of potential changes in factors that may affect the project’s sales.
Potential harmful developments include changes in the socioeconomic make-​up of the trade area and their
impact on spending patterns and shopping choices, changes in tenant space needs due to new technologies,
expansion plans by competitors, and changes in transportation patterns that may have significant impact on
traffic flows. This finding points to the necessity of testing and evaluating carefully different development
scenarios and sales projections through sensitivity analyses that explore the potential effects of such changes.

Misinterpretation of “comparables”
It is common in many retail market studies to estimate sales potential on the basis of “comparables.” This
practice only leads to inaccurate estimates, as it captures only average performance and fails to evaluate the
contribution, positive or negative, of the project’s idiosyncratic features that may distinguish it from the
average.

Poor analysis of the risks pertaining to store and tenant failure

Retail market studies rarely evaluate risks of failure of a project as a whole or its tenants separately.
Investment decisions and tenant-​mix decisions would certainly benefit from such analyses. Developments

307
 Analyzing the market for retail space

in the field of compiling and processing data on business turnover and growth patterns by community
and street level allow assessment of the risks of failure of a project at a particular location. For example,
in a study of a shopping center in the Boston metropolitan area, Cambridge Systematics found that the
apparel-​store failure rate was 48% greater than the metro average. Apparently, common market indicators
are not sufficient proxies of risk and retail market studies should perhaps start examining additional under-
lying factors more carefully.

Use of very broad data sources

Another critical flaw of many retail market studies is the use of highly aggregated information concealing
important insights with respect to household segmentation across income groups, as well as demand and
sales differences across merchandise lines and locations. This flaw points to the importance of collecting
information on spending patterns by household group and sales information by tenant type as opposed for
the development as a whole.

Retail development: a simple example

(The case is a modified version of a subsection of the Comprehensive Exam given to the Master of Real Estate
Development Program students, at the School of Policy, Planning and Development at the University of Southern
California)
It should be emphasized at the outset that market-​related information provided in this example is strictly
hypothetical. The example involves a cluster of 15,000 square feet of gross buildable retail space (including a
restaurant), which is intended to complement the retail space already planned on a commercial site likely to
house both residential and office-​commercial uses. Preliminary analysis identified three potential retail tenant
groups that could satisfy unmet needs in the site’s trade area. Given space constraints, the “best” of these groups
need to be identified and their total contribution to gross rental revenues evaluated. Finally, a schedule of
rental revenues needs to be prepared in order to be used as an input in the financial feasibility analysis.

Project and market data

Time of entry
Time of entry is year-​end 1997 (beginning of 1998).

Retail GLA
A preliminary scenario calls for the allocation of 15,000 sq. ft. of Gross Buildable Area (GBA) or, equiva-
lently, 13,500 sq. ft. of Gross Leasable Area (GLA).

Tenant clusters and sources of demand

1. Potential tenant clusters
Preliminary research of market gaps by line of retail trade, competing stores, consumer shopping
patterns, retail clustering patterns, and shopping externalities led to the identification of three “groups”
or “clusters” of tenants.
  Note that eliminating a particular tenant from a group will, most likely, have adverse effects on
the other tenants. The question that needs to be addressed is what tenant groups, not what individual
tenants, to incorporate in the retail development scenario.

308
 Synthesis and retail market studies

Table 13.1 Tenant clusters

Tenant group 1 Tenant group 2 Tenant group 3

• Restaurant • Bookstore •  Women’s ready-​to-​wear
• Delicatessen • Stationery • Menswear
•  Ice cream parlor •  Cards and gifts •  Shoe store
•  Computer/​software

Table 13.2 Tenant data (expected 1998)

Tenant groups Efficient GLA Sales/​sq. ft. norm2 Markup3 Operating cost/​sq. ft.2
min–​max sq. ft.1
Tenant Group 1
• Restaurant 3,000–​3,600 $361 0.23 $62.00
• Delicatessen 700–​900 145 0.20 14.50
•  Ice cream parlor 700–​900 155 0.21 18.40
Tenant Group 2
• Bookstore 2,500–​2,800 152 0.15 12.80
• Stationery 1,800–​2,000 150 0.23 22.50
•  Cards and gifts 1,500–​1,800 85 0.25 10.90
•  Computer/​software 1,500–​1,800 480 0.19 76.80
Tenant Group 3
•  Women’s ready-​to-​wear 2,500–​2,800 0.18 13.00
• Menswear 2,300–​2,500 190 0.14 15.20
•  Shoe store 2,800–​3,100 137 0.17 13.00

Notes
1
The proposed GLA for stores included in the development must be within the range provided (including the
lower and upper limit).
2
As always, sales per square foot and operating costs per square foot refer to annual dollars per square foot of GLA.
They are both expressed in nominal terms. The operating cost data are purely hypothetical.
3
Markup ratio figures are hypothetical.

2. Sources of demand
Demand for retail goods at the site comes from three adjoining residential neighborhoods, including
Tract 1, Tract 2, and Tract 3 (described later), as well as area workers and visitors. Information on both
demand sources is provided in the following sections.

Tenant data
Table 13.2 provides information on efficient GLA size, sales-​per-​square-​foot norms, markup, and operating
cost per square foot for the different stores included in each of the three tenant clusters considered. Notice
that the figures provided refer to the anticipated year of project completion, that is, 1998.

Residential-​based demand and trade area households

The relevant trade area (for any of the three alternative groups of stores) consists of three residential census
tracts surrounding the site under consideration. Spending pattern information is available for two distinct

309
 Analyzing the market for retail space

Table 13.3  Expected demographics and income in 1998

Census tract Household Group 1 Household Group 2

Number of households Average income Number of households Average income

Tract 1 371 $57,500 784 $95,000
Tract 2 672 57,500 1,402 95,000
Tract 3 678 57,500 1,530 95,000

Table 13.4  Expected spending patterns in 1998 (as percentage of the income figures presented in Table 13.3)

Tenant groups Household Group 1 Household Group 2

Tenant Group 1
• Restaurant 1.80% 3.35%
• Delicatessen 0.24% 0.44%
•  Ice cream parlor 0.09% 0.20%
Tenant Group 2
• Bookstore 0.45% 0.54%
• Stationery 0.30% 0.40%
•  Cards and gifts 0.40% 0.60%
•  Computer/​software 0.60% 1.50%
Tenant Group 3
•  Women’s ready-​to-​wear 2.28% 3.00%
• Menswear 1.51% 2.30%
•  Shoe store 0.77% 1.50%

income groups. Data on expected household growth and household income across these groups within
each of the three residential tracts are presented below. Note that all figures provided refer to year-​end 1998.

Job-​based and visitor-​based demand

Retail rent index
Nominal retail rents for the development under consideration are expected to increase annually as indicated
in Table 13.8 over the period 1998–​2004. These growth rates represent averages for the mix of tenants
considered and have been derived by evaluating how the maximum achievable rent for each tenant is more
likely to grow given forecasts of expected sales volume, markup ratios, and operating costs.

Analysis
Table 13.9 demonstrates the different estimates that need to be undertaken in order to evaluate which of
the three tenant groups would maximize the rental revenue potential of the development. To this aim the
following calculations/​steps were carried out:

1. Estimation of residential-​based expenditures/​sales for each tenant using the information on demo-
graphics, income, spending patterns, and drawing power, as it pertains to the target household groups
within the three tracts composing the development’s trade area.

310
 Synthesis and retail market studies

Table 13.5  Expected drawing power from residential tracts in 1998 for Group 1

Household Group 1 Tract 1 Tract 2 Tract 3

Tenant Group 1
• Restaurant 0.10 0.11 0.12
• Delicatessen 0.12 0.13 0.14
•  Ice cream parlor 0.25 0.26 0.27
Tenant Group 2
• Bookstore 0.20 0.21 0.22
• Stationery 0.15 0.16 0.17
•  Cards and gifts 0.10 0.11 0.12
•  Computer/​software 0.20 0.21 0.22
Tenant Group 3
•  Women’s ready-​to-​wear 0.07 0.08 0.09
• Menswear 0.07 0.08 0.09
•  Shoe store 0.10 0.11 0.12

Table 13.6  Expected drawing power from residential tracts in 1998 for Group 2

Household Group 2 Tract 1 Tract 2 Tract 3

Tenant Group 1
• Restaurant 0.11 0.12 0.14
• Delicatessen 0.10 0.11 0.12
•  Ice cream parlor 0.20 0.22 0.23
Tenant Group 2
• Bookstore 0.18 0.19 0.20
• Stationery 0.16 0.17 0.18
•  Cards and gifts 0.07 0.08 0.09
•  Computer/​software 0.18 0.19 0.20
Tenant Group 3
•  Women’s ready-​to-​wear 0.05 0.05 0.05
• Menswear 0.05 0.05 0.05
•  Shoe store 0.07 0.08 0.09

Note: The drawing-​power figures reported in Tables 13.5 and 13.6 do take into account the fact that another
restaurant and additional retail space are planned on the site. Planned competition in the broader trade area is also
taken into account.

2. Estimation of maximum achievable rents for each tenant/​store, using the data on sales per square foot,
markup ratios, and operating costs.
3. Choice of final GLA for each tenant/​store based on its range of efficient GLA, the total allowable
square footage in the development, and achievable rents. For example, given that the total square
footage for Cluster 1 is 5,400 and the total allowable square footage 13,500, the space allocated to any
of the other two clusters of tenants is 8,100. This prevents us from choosing the maximum efficient
size for each and every tenant in Cluster 2 because it would add up to a higher square footage. Given

311
 Analyzing the market for retail space

Table 13.7  Expected job-​based and visitor-​based demand in 1998

Tenant group Expected annual sales

Tenant Group 1
• Restaurant $180,000
• Delicatessen $67,100
•  Ice cream parlor $40,075
Tenant Group 2
• Bookstore $100,000
• Stationery $75,900
•  Cards and gifts $55,750
•  Computer/​software $110,500
Tenant Group 3
•  Women’s ready-​to-​wear $189,700
• Menswear $125,000
•  Shoe store $102,340

Table 13.8  Predicted average achievable rent growth

1999 2000 2001 2002 2003 2004

Growth rate (%) 3.0 1.5 2.0 2.2 1.7 3.1

this restriction, the bookstore in Cluster 2 was allocated the minimum efficient square footage because
it has the lowest achievable rent among the four tenants in that cluster.
4. Estimation of total rental revenue for different combinations of clusters based on the chosen GLA for
each tenant and the estimated maximum achievable rents. Notice that such estimates are presented for
only two cluster combinations, although a third combination is also possible.This combination, which
involves Clusters 2 and 3, has not been considered because the required total square footage for such
a tenant mix is 16,200, which is considerably higher than the 13,500 square feet that are available for
this development.
5 . Identification of the tenant mix that maximizes the development’s total rental revenue. According to
the calculations presented in Table 13.9, this tenant mix involves Tenant Groups 1 and 2.
6. Estimate of the rental revenue schedule for the development using the estimated average rent growth
rates for the tenant mix under consideration.

Questions
1 . Discuss the seven major flaws often encountered in retail market studies.
2. Discuss how the focus of market studies for different types of developments, such as regional, commu-
nity, and neighborhood shopping centers, differs.

312
 Synthesis and retail market studies

Table 13.9 Tenant group choice, achievable rents, space allocation

Expected expenditures ($), Supportable Efficient Maximum Chosen

space (GLA)4 GLA gross rent ($/ GLA
Residential based
(sq. ft.)5 sq. ft. GLA)6 (sq. ft.)7
Group 11 Group 22 Visitor/​job
based3
a. Tenant Group 1
• Restaurant $143,708 $1,491,574 $180,000 5,028 3,000–​ 21.03 3,600
3,600
• Delicatessen 22,589 172,228 67,100 1,806 700–​900 14.50 900
•  Ice cream parlor 16,827 155,257 40,075 1,369 700–​900 14.15 900
b. Tenant Group 2
• Bookstore 68,067 367,914 100,000 3,526 2,500–​ 10.00 2,500
2,800
• Stationery 34,665 242,107 75,900 2,351 1,800–​ 12.00 2,000
2,000
•  Cards and gifts 32,100 173,702 55,750 3,077 1,500–​ 10.35 1,800
1,800
• Computer/​ 91,120 1,016,738 110,500 2,538 1,500–​ 14.40 1,800
software 1,800
c. Tenant Group 3
• Women’s 133,082 529,530 189,700 6,556 2,500-​ 10.40 2,800
Ready-​to-​Wear 2,800
• Menswear 88,069 405,973 125,000 3,258 2,300-​ 11.40 2,500
2,500
•  Shoe store 61,792 434,255 102,340 4,368 2,800-​ 10.29 2,800
3,000
Suggested tenant Total space Total rental
groups (GLA) revenue
Tenant Group 1 5,400 $101,493
Tenant Group 2 8,100 $93,550
Total 13,500 $195,043
vs
Tenant Group 1 5,400 $101,493
Tenant Group 3 8,100 $86,432
Total 13,500 $187,925
Schedule of maximum rental revenues ($)
Year 1998 1999 2000 2001 2002 2003 2004
Total rental $195,043 $200,894 $203,907.70 $207,986 $212,562 $216,175 $222,877
revenue

Notes
1,2
  Residential-​based expenditures by product line were estimated by multiplying the number of households in
each residential tract by their average income, the percentage spent on the goods sold by each store, and the
development’s drawing power.
3
Visitor/​job-​based expenditures were taken from Table 13.7.
4
Supportable space by tenant was estimated by dividing total expected expenditures by the sales/​sq. ft. norm.
5
Viable store size.
6
Estimated as ((Sales/​sq. ft.)* Markup) –​Operating Costs/​sq. ft.
7
Depending on the level of rent/​sq. ft., the upper or lower limit of the efficient-​size range was chosen.
313
 Analyzing the market for retail space

References and additional readings

Fanning, S., T. Grissom, and T. Pearson. 1994. Existing Shopping Center. Chap.  12 in Market Analysis for Valuation
Appraisals. Chicago, IL: Appraisal Institute.
McMahan, J. 1989. Retail Commercial. Chap.  8 in Property Development. New  York:  McGraw-​Hill Publishing
Company.
Schmitz, A. and D. Brett. 2001. Retail Development. Chap. 5 in Real Estate Market Analysis: A Case Study Approach.
Washington, DC: Urban Land Institute.
Weisbrod, G. and K. Radov. 1984. Retail Success or Failure –​Any Connection to Market Studies? Cambridge, MA: Cambridge
Systematics, Mimeo.

314
 PART E

Office market analysis

newgenprepdf

14
THE MARKET FOR OFFICE SPACE

Introduction 317
Office market idiosyncrasies and determinants of project success 318
Office space demand and prospects 319
Office-using sectors 320
On net absorption and its determinants 320
The supply of office space 322
Income-related factors 322
Cost-related factors and availability of capital 323
Overview of office market analysis 324
Analysis of office project viability from the market perspective 326
Chapter summary 326
Questions 326
References and additional readings 327

Introduction
This is the first chapter of Part E, which focuses on the analysis of commercial office markets. The pur-
pose of this chapter is to first discuss the idiosyncrasies and economic fundamentals of office markets and
then provide an overview of the major steps involved in analyzing them. Within this context, we first
discuss the determinants of office project success within the context of the idiosyncrasies characterizing
the marketplace for office space and, especially, its cyclical nature. We then elaborate on the demand and
supply fundamentals of such a marketplace. On the demand side, the discussion focuses on the nature of
the office-​using employment sectors and the major determinants of demand for office space. On the supply
side, the discussion focuses on the determinants of office construction investment, and specifically on
income-​related factors, cost-​related factors, and the availability of capital. We conclude the chapter with a
brief overview of the market analysis process, as an introduction to Chapters 15 and 16.These two chapters
elaborate on the macroeconomic and microeconomic aspects of market research, as they pertain to office
development projects.

317
 Office market analysis

Office market idiosyncrasies and determinants of project success

Completing an analysis for an office development project requires a thorough understanding of the idio-
syncratic nature of office markets and the factors that affect the value and profitability of office properties at
the intra-​metropolitan level. Given the idiosyncrasies characterizing the market for office space, the major
determinants of office project success are the following:

• Timing
• Location
• Project design.

Timing is a very important determinant of office project success because the office market is characterized
by wide fluctuations in both (marginal) demand and supply and, as such, it is highly cyclical. The high
cyclicality of the office space market is evident in Figure 14.1, which portrays the time path of the national
vacancy rate over the period 1960–​1999.1 As the figure shows, during the 1970s the national office vacancy
rate increased from 4% to almost 15% and then returned back to 4% by the end of the decade. During
the 1980s, the office vacancy rate climbed to new highs, and by the end of the decade, it was hovering
around 17%.The economic recession of the early 1990s not only pushed it even higher up to 19%, but also
triggered the collapse of property values. As a result, a dramatic decrease in the construction of new office
space helped the national vacancy rate to gradually decrease below 10% by 1999.
Wheaton (1987) showed that the office market is characterized by prolonged endogenous cycles with
long periodicity (6–​10 years) and great amplitude (see Figure 14.2 for the definition of periodicity and
amplitude). Barras’ (1994) analysis of the cyclical movements of the City of London office market points to
an office development/​building cycle of around 12–​14 years. Mueller and Peiser (2015) have documented
the cyclical movements in office occupancy rates in major North American cities in recent decades (1988–​
2012) while Ibanez and Pennington-​Cross (2013) confirmed that office markets adjust at a considerably
slower rate that the other commercial property markets (retail, industrial, and flex). Furthermore, Stevenson
et al. (2014) present evidence of significant concordance of office market cycles across the world’s 20 lar-
gest cities, attributed to commonalities in the underlying structure of the economies of those cities. The
authors cite also the concentration of investments in these markets as another reason for synchronization
in office cycles, as they are all highly sensitive to developments in capital markets. As has been empirically
verified, these cycles are generated by the synergy of the following demand, supply, and rent adjustment
factors underscoring the inefficiency of the commercial office market:

20

15

10

0
1960.1
1963.2
1966.3
1969.4
1973.1
1976.2
1979.3
1982.4
1986.1
1989.2
1992.3
1995.4
1999.1

Figure 14.1 The office vacancy cycle

Source: CBRE Econometric Advisors

318
 The market for office space

Peak Periodicity

Amplitude

Trough

Figure 14.2  Cycle attributes

• Repeated and unexpected office demand shocks triggered by changes in the national and local
economies.
• Sluggish demand, vacancy, and rent adjustments due to informational lags and inefficiencies, high adjust-
ment costs, and long-​term lease structures (Mourouzi-​Sivitanidou, 2002).
• Sluggish supply adjustments, due to informational inefficiencies, investment-​ decision lags, and
inherently-​long construction lags.
• Highly price-​elastic new supply, due in part to myopic expectations on the part of real estate investors
(which reinforce and amplify the effect of exogenous shocks), and the lumpiness of investments in new
office structures.2, 3
• Asymmetric stock adjustments in response to favorable vs. unfavorable exogenous demand shocks, in
the sense that there is a ceiling to rent and price increases in response to positive demand shocks due to
competition and supply responses, but downward rent movements are not constrained.4

Location is the second important factor for office project success because of the role productive amenities
valued by office tenants, and worker (non-​productive) amenities valued by their employees, play in deter-
mining firm costs (labor costs, rent, etc.). Since such amenities may not be uniformly distributed within
a metropolitan area, location selection can help office firms minimize such costs. In addition, because of
linkages among office activities, intra-​metropolitan location selection may also play a role in affecting office
firm costs and revenues.
Project design is thought to be less important than timing and location, but still the choice of the appro-
priate development density can be very crucial in maximizing residual land values.

Office space demand and prospects

Office space is a required production input by firms within several SIC-​code classifications, justifiably
referred to as office-​using sectors. The amount of space per worker required by a firm depends on the
nature of its operations and expected growth. Furthermore, a firm’s realized demand for office space in the
marketplace should be conditioned by prevailing lease rates. Therefore, market demand for office space is
determined primarily by four major factors:

1. Office employment
2. Space-​per-​employee requirements
3. Rental  rates
4. Expectations.

Sivitanidou (2002) has provided indirect evidence of the effect of three of these four factors on office space
demand in a panel study of office rent differentials across the nation’s major office markets. In particular,

319
 Office market analysis

this study has demonstrated that office employment levels, office employment growth rates—​most likely
capturing growth expectations—​and the ratio of FIRE to total office employment (reflecting the higher
space-​per-​employee requirements in this sector) have positive and statistically significant influences on
metro-​wide office rent levels. Such effects are consistent with the expected influence of these factors on the
demand for office space. Tse and Webb (2003) also provide evidence of the strong effect of office employ-
ment on office space absorption.
It should be noted that Sivitanidou (2002) also provides indirect evidence of differences in office space
demand levels across markets attributable to some additional factors affecting tenant search processes.These
additional factors include the diversity of the local office tenant base, and the spatial/​nodal distribution of
the local office space stock. In theory, such local market attributes affect absorption through their effect
on the matching probability between tenants and landlords and on search costs. Given that such attributes
change slowly through time, these factors are more relevant in understanding differences in office demand
levels across markets, as opposed to changes in office demand through time within the same market.
Within this context, accurate assessment and forecasting of office space demand prospects in any market
requires an understanding of how the basic forces that drive its movements through time are likely to
unfold over the project’s planning horizon.

Office-​using sectors
Studies by Torto Wheaton Research suggest that the two major office-​using sectors of the economy
are Finance, Insurance, and Real Estate (FIRE), and Services (including primarily Professional Services,
Business Services, and Government). Estimates show that 100% of FIRE workers and about 36% of Service
workers are office space users and that these two sectors account for about 80% of total office employment,
with the remaining 20% accounted for by the industrial sector (Wheaton, 1987). According to DiPasquale
and Wheaton (1996), the Services subsectors that can be considered as office-​using include Advertising,
Computer and Data Processing, Credit Reporting, Mailing and Reproduction, Legal and Social Services,
Membership Organizations, and Engineering and Management Services.
Carn et al. (1988) present a different analysis that points to a somewhat lower contribution of FIRE and
Services to total office employment. In particular, these authors distinguish between office-​using and non-​
office-​using occupations using 1980 Census data on the occupational composition of the different sectors.
Their analysis concludes that 67% of employees in the FIRE sector and 28% of employees in the Service
sector are office space users.Their analysis suggests also that some non-​negligible shares of employees in the
other one-​digit SIC sectors ranging between 16% and 38% may also be users of office space.
Rabianski and Gibler (2007) caution about relying on FIRE and Service employment as proxies for
office employment because technological changes and changes in work practices are altering the contribu-
tion of these two sectors to total office employment.

On net absorption and its determinants

Absorption of office space at the margin, or net absorption, should depend heavily on the extent of
increases in office employment, as well as the other three major factors driving the demand for office space,
that is, space per employee requirements, office rents, and growth expectations. Each of these factors is
discussed below.

OFFICE EMPLOYMENT GROWTH  According to Bureau of Economic Analysis data, national employment

in key office-​using sectors, such as Finance and Insurance, Real Estate and Rental and Leasing, Professional,
Scientific, and Technical Services, and Management of Companies and Enterprises has been increasing over
the period 2015–2018 at annual rates ranging mostly between 2% and 3.5%.These trends have contributed

320
 The market for office space

6%

5%

4%

3%

2%

1%

0%
2012 2013 2014 2015 2016 2017 2018
Finance and insurance
Real estate and rental and leasing
Professional, scientific, and technical services
Management of companies and enterprises

Figure 14.3  Growth rates in key office-​using sectors in the U.S.

Source: BEA

to rising demand for office space, positive net absorption, and falling vacancy rates, according to data
provided by CBRE (2019) for the 20 largest office markets in the U.S.
Looking ahead, the Bureau of Labor Statistics (BLS) predicts that employment growth rates for the
next period (2018–​2028) for key office-​using sectors will be on average considerably lower than the ones
registered over the period 2015–​2018. In particular, BLS predicts that employment in the overall service
sector will be increasing at an average annual rate of 0.6% over the period 2018–​2028, with employment
in professional and business services rising at an annual rate of 0.8%, and employment in financial services
growing at the considerably lower rate of 0.3%. Some of the reasons that may contribute to the deceler-
ation of employment growth rates include:

• Falling labor force participation rate (according to BLS projections it will fall from 62.9 in 2018 to 61.2
in 2018).
• Reduction of immigration.
• The expanding use of technology by office firms that may contribute to substitution of labor for capital.

SPACE-​PER-​WORKER REQUIREMENTS  Space requirements per office worker vary by industry. For
example, it is often argued that such requirements are higher in the prestigious FIRE sector as opposed
to Services. Sivitanidou (2002) provides indirect evidence that is consistent with the premise that higher
shares of FIRE workers in total office employment are associated with higher office space demand.5 Carn
et  al. (1988) suggest that space-​per-​worker requirements vary also by occupation, with managers and
professionals occupying more square feet, and clerks and secretaries less. Therefore, it is only appropriate
to assume that average square feet requirements across metropolitan markets must vary given differences
in sectoral employment composition and occupational structure. Rabianski and Gibler (2007) advise
strongly against using average office space per worker and point to the need for adjusting such a ratio for
differences across sectors and individual firms within sectors, as well as across occupations, locations, and
market conditions.

321
 Office market analysis

According to data presented by CoStar, occupied office space per employee in the U.S. has been falling
steadily and rather rapidly since 2010 when it reached its latest peak of 197.3. By 2017 this ratio was down
to 181 square feet per employee, representing an almost 20% cumulative drop. Note that according to the
CoStar data, the previous peak of this ratio was registered in 1992 at 202 square feet per employee while
the bottom of that cycle was registered after ten years of an uninterrupted fall to 171.6 square feet per
employee. Looking at the two downturns in square feet per employee (the one that started in 1992 and
the one that started in 2010) they look very similar in terms of both their peaks and rate of decline. This
behavior supports the hypothesis that the consistent decrease in occupied office space per employee that
has been taking place since 2010 represents a temporary cyclical trend rather than a secular trend caused
by structural changes in the way that office-​using companies organize their production activities and use
office space.

OFFICE SPACE RENTS  As the fundamental law of demand postulates, all else being equal, increasing office
rents should exert downward pressure on office space absorption, while decreasing rates are expected to
positively influence office space demand. The potential influence of rental rates on space-​per-​employee
requirements is very important because it has significant analytical implications that cast doubts on the
accuracy of a very commonly used technique for forecasting office space demand (see discussion in the
next chapter).

EXPECTATIONS  We can distinguish two types of expectations that may influence demand for office space
at any given period: rent growth expectations and employment growth expectations.

• Rent growth expectations may have a positive effect on net absorption, since firms may be more motivated
to lease space at the present rather than wait for a later time, in anticipation of rent increases in the
future. Such expectations are very likely to prevail within a tight market environment where rental rates
have already started rising.
• Office employment growth expectations can also have a positive effect on office space absorption, as firms
are likely to lease more space than they currently need in anticipation of their expanding workforce.
Since relatively slow employment growth is expected in the future, growth expectations may not have
a significant effect on net absorption of office space.

The supply of office space

Since office projects take usually about two years to complete from the time of inception and one to two
years for lease-​up, office market studies aiming at guiding investment and development decisions need
to have a three-​to four-​year planning horizon at minimum. Simple counts of projects permitted and
under construction can provide little clues, if any, about the levels of available supply three to four years
ahead of the time of analysis. For this reason, it is very important to examine past trends and evaluate
future prospects in the factors that drive office construction investments at the macro-​analysis level.
Since profitability is the prime motivation of office development, these factors include the determinants
of the income-​earning capacity of an office property, the perceived risk of expected income flows, and
development costs.

Income-​related factors
Factors pertaining to an office building’s income-​earning prospects that may influence construction invest-
ment include the following:

322
 The market for office space

RENTS  Increasing rents are likely to boost development only to the extent they rise (or are expected to
rise) to or above the levels that would justify new construction. In particular, office rental rates must rise to
a level for which the implied asset price would be greater than development costs in order to trigger new
construction.6 The experience of the 1980s has shown the supply of office space to be highly price/​rent
elastic, but the experience of the 1990s points to a more disciplined behavior.

VACANCY RATES  As indicated earlier, office vacancy rates fluctuate considerably through time, thereby
inducing significant fluctuations in a property’s rental income. During periods with sufficiently low vacancy
rates and, therefore, higher property income, investors may be motivated to provide capital for new office
development projects. In this sense, markets with lower vacancy rates may be facing prospects for higher
levels of new office space additions in the future.

EXPECTATIONS  Myopic supplier expectations for improving market conditions due to improving demand
indicators, such as office employment growth and declining vacancy rates, may also contribute to greater
office construction levels. Office space completions are thought to largely be driven by such expectations.

MARKET VOLATILITY AND RISK  Higher employment volatility in office-​using sectors, such as FIRE and
Services, may discourage new construction investments, as it may signal a greater demand uncertainty
and development risk and, as such, raise the level above which rents must rise to justify new construction.
Historical data reveal that FIRE employment is much more volatile than employment in Services.
A study by Sivitanidou and Sivitanides (2000) provides evidence of the effect of all these factors on
office space additions. In particular, their analysis shows that real office rents, vacancy rates, office employ-
ment growth expectations, and FIRE employment growth volatility have a statistically significant effect on
metro-​wide office construction.

Cost-​related factors and availability of capital

COST FACTORS  Keeping constant office-​ property prices, higher development costs, including the cost
of financial capital (interest rates), construction costs (cost of labor and materials) and land costs, reduce
investment profitability and discourage office construction. Sivitanidou and Sivitanides (2000) show that
construction costs and interest rates do indeed have a negative and statistically significant effect on metro-​
wide office completions. Kling and McCue (1987) also provide evidence of the effect of interest rates on
office building investments.

AVAILABILITY OF CAPITAL  The availability of investment capital for new office construction projects
may be influenced by government policies, the public equity market, the private equity market, and
the private lending industry. For example, Clapp (1993) notes that some observers have attributed the
excessive overbuilding of office space in the 1980s to the abundance of capital, as lenders, facilitated
by the reduced regulations put in place in the early 1980s, appeared overly zealous to provide real
estate loans. In the private equity market, the focus of institutional investors on office buildings at the
expense of other property types may have also contributed to the office construction boom of the 1980s.
In the public equity arena, publicly traded real estate management companies (REITS) help channel
investment capital (a share of which may be diverted to new development projects) into real estate
markets by allowing small investors to participate in real estate investments. After the office market crash
of the early 1990s, institutional lenders have adopted stricter lending practices (e.g., increased pre-​leasing
requirements among others), which may have reduced the availability of capital for office construction
during the 1990s.

323
 Office market analysis

Overview of office market analysis

Against this background, the purpose of office market analysis is to help proactive (as opposed to reactive)
office investment decisions and development design by looking at and evaluating the forces that deter-
mine movements in the broader office market, as well as the more specific location and site factors that
may influence project success. As is the case for the other property types, we can distinguish two major
components in office market studies: the macroeconomic component and the microeconomic component
(see Figure 14.4).
Given the high volatility of office markets, the macroeconomic component of office market studies is
especially important. Its three major objectives are to:

• Analyze and evaluate broader office market strength.

• Identify the target market’s position on the rent, vacancy, and construction cycle.
• Address questions of timing.

Analyzing office market strength at the time of analysis and its position on the real estate cycle will help
the analyst better anticipate the market’s potential movements over the project’s expected holding period.
Identifying a market’s position on the cycle requires an assessment of the extent of disequilibrium, or the
degree of mismatching between the demand for and the supply of office space. This is not an easy task
because of the extreme volatility of the office space market. It is, however, a very critical step given that
the fortunes of individual actors/​projects within a metropolitan office market are driven to a great extent
by this idiosyncratic cycle. Within this context, the macroeconomic component of office market analysis
can help address prudently the question of whether it is appropriate to enter a given market at a particular
point in time.
Evaluating office market strength and prospects requires the analysis of office space demand at the
margin (net absorption) and its determinants, existing inventory, completions and their determinants, office
space rents, and supply-​demand imbalances. The interplay of office space demand and supply gives rise to
two summary measures of the state of the market—​rental change and supply-​demand gap, which, in theory,
must be consistent in terms of their implications regarding market strength.
As in the case of the other property types, the analyst can use either accounting or econometric
techniques for analyzing office markets at the macro level. Accounting methodologies can help the analyst
develop rather simplistic estimates of future demand, supply, and the demand-​supply gap.There are, however,
several issues in using these techniques, especially for longer forecast periods. Econometric techniques can
help explore the underlying behavioral relationships and feedback effects among the key variables that drive
the market for office space and then use them as a basis for more accurately forecasting its future movements.
The microeconomic component of office market studies focuses on the competitive strength and most
likely performance of the proposed project. Its focus is the evaluation of office locations, the identification
of the most marketable office development features for a given site, and the assessment of project viability
from the office market perspective. In particular, analysis of office locations focuses on:

• The site’s locational amenities and advantages, including productive amenities, that is, amenities that can
increase office-​firm productivity, such as proximity to other firms, freeway access, access to CBD, etc.,
and worker amenities, that is, amenities that are valued by the employees of office firms, such as the
quality of the public school system, crime rates, access to shopping and entertainment opportunities, etc.
(Sivitanidou, 1994 and 1996).
• Visibility of the site and prestige of the location within the urban fabric.
• Locational constraints in terms of growth regulations, local zoning, and other institutional controls that
may hamper commercial real estate development.

324
 The market for office space

OFFICE MARKET
ANALYSIS

A. Macro analysis of office B. Micro analysis of office

markets projects/marketability analysis

Analyzing market strength/ Assessing office locations

positioning target markets on designing marketable projects/
their office market cycle assessing project viability from
addressing questions of timing the market perspective

Analysis components/ Analysis components:

indicators of market strength:
Analyzing office locations
• Space demand and determinants
• Completions and determinants Analysis elements:
• Space rents • Locational amenities
• Supply-demand imbalances • Firm/production amenities
• summary measures (rental • Worker/residential amenities
change, demand-supply gaps) • Prestige and visibility
• Locational constraints
Analysis methodologies

Accounting techniques: Analyzing office projects/

Analyzing assessing project
demand-supply gaps: “Simplistic” view: feasibility
Net absorption
Supply-demand gap, = Office employment growth
GAP = x Space/worker Analysis elements:
Net absorption • Project rents/lease rates,
+ Replacement demand Informed view: Development densities
- Available supply Net absorption • Project absorption and
= f[Office employment growth, cumulative revenue schedules
office rents, expectations]
Econometric techniques: Analysis techniques:
Analyzing office market dynamics Accounting and hedonic
Forecasting expected movements in: approaches
Net absorption and office
Net absorption employment growth can in turn
New construction be estimated through:
Real rents
Vacancy rates
Aggregate Disaggregate
accounting accounting
approaches approaches
`

Financial feasibility analysis

Figure 14.4  Overview of office market analysis

325
 Office market analysis

Analysis of office project viability from the market perspective

The analysis of office project viability from the market perspective focuses on:

• Estimation of achievable office rents through competitive differential techniques or hedonic meth-
odologies that account for lease characteristics, space characteristics, project amenities, and location
attributes.
• Estimation of the project’s expected absorption and revenue schedules.

In the next two chapters, we elaborate first on the macroeconomic aspects of office market studies
(Chapter 15) and then focus on the microeconomic analysis of office development projects (Chapter 16).

Chapter summary
The purpose of office market analysis is to look at and evaluate those factors that determine office project
success. These factors include timing, location, and project design and are inherently linked to the nature
and dynamics of the forces that drive the demand for and supply of office space.
Office space demand is primarily driven by employment growth in the FIRE and the Services sectors.
Such growth is not expected to reach the highs of the 1980s in the coming years for a number of reasons.
Empirical studies have shown that, for given office employment levels, office space demand responds to
movements in rental rates and office-​using firm expectations for employment growth. This suggests that
estimates of office space demand by simple multiplication of expected office employment growth by a
space-​per-​worker norm may lead to serious inaccuracies, as they ignore the potential effect of movements
in such factors.
A trend of declining square feet per office employee has been observed in the 1990s, in contrast to the
1980s when such a ratio was rising. Although these patterns seem to be connected with declining real
rents in the 1980s and increasing real rents in the 1990s, other more permanent influences may be at work.
Trends in this ratio will be crucial in determining office space demand in the future.
The supply of office space is driven by factors influencing office-​property income and development
costs. As such, new office construction is driven by office rent levels, vacancy rates, expectations regarding
office market strength, market volatility/​r isk, construction costs, the cost and availability of capital, and land
costs. The experience of the 1980s has shown the supply of office space to be highly price elastic, but the
experience of the 1990s points to a more disciplined behavior.
Within this context, the purpose of the macroeconomic analysis of office markets is to evaluate their
strength, assess their position on the rent-​vacancy cycle, and make prudent forecasts with respect to their
future movements. The analyst needs to focus on such crucial market indicators as existing office space
inventory, vacancy rates, lease rates, net absorption, and new construction.
The microeconomic component of office market analysis focuses on the evaluation of project location,
identification of the most attractive office development features, and the assessment of project performance
in terms of lease rates and absorption schedule.

Questions
1 . Discuss which factors are important, and why, for successful office development.
2. Discuss the factors that contribute to the cyclical nature of the office market and assess whether they
apply to the office market of your city.
3. Discuss the recent trends in the square-​feet-​per-​employee ratio during, potential drivers of movements
in such ratio, and prospects in the years ahead.

326
 The market for office space

4 . Discuss four ways by which the demand of office space in your city may increase.
5. Discuss three developments that will cause an increase in office construction activity in your city.
6. Describe how you could go about analyzing (the major steps) the office market of your city.

Notes
1 The vacancy rate portrayed in this figure represents the weighted average of 31 office markets covered by Torto
Wheaton Research.
2 There is evidence that in recent years the majority of office construction is directed towards the suburbs and as a
result the average office investment is becoming less lumpy.
3 This conclusion is based primarily on office supply behavior during the 1980s. There is indirect evidence that, after
the collapse of the real estate market in the early 1990s, investors and capital providers have been approaching real
estate, and specifically office development projects, with greater caution compared to the 1980s (Sivitanidou and
Sivitanides, 2000). If this is in fact the case, office consctruction may be becoming less price elastic. In addition, there
is discussion in the industry of the increasing role public markets may be playing in the real estate market and their
potential contribution in reducing its volatility through their forward-​looking approach and analytical discipline.
4 See Sivitanidou and Sivitanides (2000).
5 Sivitanidou (2002) in an analysis of office rent differentials across the nation’s largest office markets finds that for
given office employment levels, the ratio of FIRE employees to total office employment has a positive and statis-
tically significant effect. One obvious rationalization of such influence is the higher space-​per-​worker requirements
in the FIRE sector and its positive effect on office space demand. See the paper for alternative explanations of this
effect.
6 The reader is reminded that property value or asset price and rental rates are linked through the income capital-
ization formula, which postulates that value is the ratio of the property’s NOI over the market capitalization rate.
Sivitanidou and Sivitanides (1999) have shown that office market capitalization rates are driven by factors that shape
income growth expectations and risk perceptions, as well as interest rates.

References and additional readings

Barras, R. 1994. Property and the Economic Cycle:  Building Cycles Revisited. Journal of Property Research, 11:3,
183–​197.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Office Market Analysis. Chap.  11 in Real Estate Market
Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
CBRE. 2019. Q3 2019 U.S. Office Figures. Los Angeles, CA: CBRE.
Clapp, J. 1993. Empirical Research on Office Market Dynamics. Chap. 2 in Dynamics of Office Markets. Washington,
DC: The Urban Institute Press.
DiPasquale, D. and W. Wheaton. 1996. The Operation of Nonresidential Property Markets. Chap. 11 in Real Estate
Markets and Urban Economics. Englewood Cliffs, NJ: Prentice Hall.
Hendershott, P. and E. Kane. 1992. Causes and Consequences of the 1980s Commercial Construction Boom. Journal of
Applied Corporate Finance, 5:1, 61–​70.
Ibanez, M. R. and A. Pennington-​Cross. 2013. Commercial Property Rent Dynamics in U.S. Metropolitan Areas: An
Examination of Office, Industrial, Flex and Retail Space. Journal of Real Estate Finance and Economics, 46:2, 232–​259.
Kling, J. and T. McCue. 1987. Office Building Investment and the Macroeconomy: Empirical Evidence 1973–​1985.
AREUEA Journal, 15:3, 234–​255.
Mourouzi-​Sivitanidou, R. 2002. Office Rent Processes: The Case of U.S. Metropolitan Markets. Real Estate Economics,
30:2, 317–​344.
Mueller, G. R. and R. Peiser. 2015. International Real Estate Office Market Cycles. Real Estate Finance, 31:4, 147–​156.
Rabianski, J. S. and K. M. Gibler. 2007. Office Market Demand Analysis and Estimation Techniques:  A Literature
Review, Synthesis and Commentary. Journal of Real Estate Literature, 15:1, 37–​56.
Sivitanidou, R. 1994. Urban Spatial Variations in Office-​Commercial Rents:  The Role of Spatial Amenities and
Commercial Zoning. Journal of Urban Economics, 38:1, 23–​49.
Sivitanidou, R. 1996. Do Office-​Commercial Firms Value Access to Service Employment Centers? A Hedonic Value
Analysis within Polycentric Los Angeles. Journal of Urban Economics, 40:2, 23–​49.
Sivitanidou, R. 2002. Office Rent Processes:  The Case of U.S. Metropolitan Markets. Real Estate Economics, 30:2,
317–​344.

327
 Office market analysis

Sivitanidou, R. and P. Sivitanides. 1999. Office Capitalization Rates: Real Estate and Capital Market Influences. Journal
of Real Estate Finance and Economics, 18:3, 297–​322.
Sivitanidou, R. and P. Sivitanides. 2000. Does the Theory of Irreversible Investments Help Explain Movements in
Office-​Commercial Construction? Real Estate Economics, 28:4, 623–​661.
Stevenson, S., A. Akimov, E. Hutson, and A. Krystallogianni. 2014. Concordance in Global Office Market Cycles.
Regional Studies, 48:3, 456–​470.
Tse, R.Y. C. and J. R. Webb. 2003. Models of Office Market Dynamics. Urban Studies, 40:1, 71–​89.
Wheaton, W. 1987. The Cyclic Behavior of the National Office Market. AREUEA Journal, 15:4, 281–​299.

328
 15
OFFICE MARKET ANALYSIS
A macro perspective

Introduction 330
Analyzing office markets: a macro perspective 330
What market variables should be analyzed? 331
Office space demand 331
New office supply 331
Office market demand-supply imbalances 331
Numerical example 332
Assessing office market strength: overview of office market gap estimation 335
Calculating new office space requirements, DS(t) 336
Estimating expected net absorption of office space, AB(t) 336
Simplistic view 336
“Informed” view 336
Simple accounting techniques for forecasting office-using employment and
office space demand 339
Aggregate approaches: using office employment growth 339
Disaggregate approaches: using industry-occupation matrices 339
Example 340
Office supply analysis using accounting approaches 343
Vacant stock 345
Completions 345
Critique 345
Econometric forecasting of office markets 346
Forecasting office space absorption, new construction, vacancies, and rents 346
Alternative rental adjustment models 347
Alternative 1: Traditional model with variable structural vacancy rate 350
Alternative 2: The equilibrium or desired rent model 350
Submarket analysis 351
Types of office nodes 352
Downtown district 352

329
newgenprepdf

Office market analysis

Uptown district 352

Office parks 352
Special nodes 352
Shopping center office nodes 352
How to analyze office submarkets 352
Chapter summary 354
Questions 355
Project #5: Econometric forecasting of office markets 355
The case 355
The questions 356
The report 356
The data 356
References and additional readings 358

Introduction
There is a growing belief that the analysis of the macroeconomic conditions and the health of the broader
office market are the most critical elements of office market studies. This premise became very clearly
understood within the professional real estate community after the collapse of the office market in the
early 1990s.
The macroeconomic analysis of office markets involves assessment of market strength focusing on the
metro-​wide demand-​supply gap and real rent movements. These two indicators are consistent measures of
the weakness or the strength of an office market since when the demand-​supply gap is positive (suggesting
excess demand) real rents should be rising, while when the demand-​supply gap is negative (suggesting
excess supply) real rents should be declining.
Within this context, this chapter elaborates on accounting and econometric techniques that analysts can
use to assess and forecast office space demand, supply, and rental rates. In the case of accounting techniques,
the focus is on aggregate and disaggregate methodologies for assessing office space demand on the margin.
In the case of the econometric approach, a more integrated and sophisticated modeling framework is
presented. The chapter concludes with a discussion of submarket analysis issues.

Analyzing office markets: a macro perspective

Independently of the methodology used (accounting or econometric), macro analysis of office markets
involves the following four major steps:

1 . Analyze current conditions, historical trends, and prospects in office demand, supply, and rents.
2. Generate absorption, and completion forecasts.
3. Estimate and evaluate the demand-​supply gap over the project/​investment planning period.
4. Generate office rent forecasts.

In other property types, such as retail, for example, the first step is to define the project’s market area. In
the case of office, however, because of the wide range of locations that may be suitable for an office-​using
firm’s operations, it is more appropriate to define as the project’s market area the whole metropolitan area.
Such a definition will ensure that competing supply is not underestimated and, therefore, help more accur-
ately capture future market trends. Given the need for employment data that provide the basis for assessing
office space demand, such a definition is very fitting since employment data are regularly updated only for

330
 Office market analysis: macro perspective

cities, counties, and MSAs. Analysts should be very skeptical and cautious before defining the city or county
boundaries as an office project’s market area, since narrow market definition could lead to very misleading
conclusions regarding a project’s prospects and viability.
Ideally, when carrying out an office market study, it is preferable to distinguish between owner-​occupied
and speculative space, since the former is not really part of the competitive market for rental space and, as
such, it does not influence office rents, at least directly.1 However, it is difficult to identify from both the
demand and the supply side what portion of an area’s total office space is owner-​occupied and what is not.
In analyzing historical trends in absorption, completions, vacancy, and rents it is useful to review such
trends by type of office space (class A, B, and C) if data are available. In reviewing such data, it is important
that the analyst understands not only the trends in each segment, but also their interactions, since some
degree of substitutability from one segment to the other does exist. For example, non-​negligible increases
in lease rates for class A space should motivate some firms to seek space in the class B segment.
The reader is also reminded that absorption analysis needs to focus on net absorption as opposed to
gross absorption, since the latter does not account for space vacated (see Chapter 2 for a more elaborate
discussion of alternative absorption concepts) and cannot provide any insights as to how total office space
demand changes from one period to the other.

What market variables should be analyzed?

Based on the discussion in Chapter 14, the variables that need to be examined when analyzing an office
market at the macroeconomic level are summarized below:

Office space demand
Net absorption and its determinants (see Figures 15.1, 15.2, and 15.3).

• Employment growth in the FIRE and the service sectors (see Figure 15.2).
• Real (inflation-​adjusted) office rents (see Figure 15.3).
• Firm expectations regarding office employment growth and rental rates.
• Industry-​structure influences on square-​feet-​per-​employee requirements.

New office supply
Completions and its determinants (see Figures 15.4, 15.5, and 15.6).

• Planned and under construction projects.

• Office rents (see Figure 15.4).
• Investor/​developer expectations regarding office employment growth (see Figure  15.5) and rent
increases.
• Vacancy rates (see Figure 15.6).
• Construction costs and land costs, as well as cost and availability of financial capital.
• Market volatility and risk.

Office market demand-​supply imbalances

Popular measures

• Completions minus net absorption (see Figure 15.7).

331
 Office market analysis

• Vacant stock and nominal vacancy rate movements (see Figure 15.8, and 15.9).
• Overhang rate.

Alternative measures of market strength

• Real office rent movements and deviations of the nominal from the structural vacancy rate (see
Figure 15.10).
• The demand-supply gap.

Numerical example
As a way of an example, the major macro indicators for the Los Angeles office market are presented
and discussed below. The data have been provided by Torto Wheaton Research, a business unit of CB
Richard Ellis.
As Figure 15.1 indicates, office net absorption in the Los Angeles metropolitan area has shown consider-
able variation not only from year to year but also from decade to decade. In particular, over the 20-​year period
covered by the data, it ranged from a negative 860 thousand square feet (in 1991) to a positive 8.4 million square
feet (in 1985). Furthermore, it is clear from the graph that overall absorption of office space in the 1990s was
considerably lower than in the 1980s (2 million square feet on average vs. 6.2 million square feet, respectively).
As Figure 15.2 indicates, such difference is mainly attributable to the considerably lower office employment
growth that took place during the 1990s compared to the 1980s (1% on average vs. 4%, respectively). As
Figure 15.3 indicates, absorption in 1990s was helped to some extent by lower real rents.
Office space completions, in the Los Angeles area, registered a sharp increase in 1982, in response to real
rent increases that took place in the late 1970s and early 1980s. After slowing down in 1983, completions
continued rising until 1985, despite declining real rents since 1982, pointing to the long time lags involved
in the investment and development process.With vacancy rates hovering around or above 16% until the late
1990s (Figure 15.4) and real rents declining through the early 1990s (Figure 15.5), office space completions
continued declining as well (barring a resurgence in 1989) and practically reached near-​zero levels in 1994.
They remained at such low levels until 1998. Office completions seem to have been following movements
in office demand (office employment and absorption) during the 1980s, but not during the 1990s when
the market was overwhelmed by low rents and very high vacancy rates (Figures 15.6 and 15.7).

10,000

8,000
O ffice space absorption
T housands of sq. ft.

6,000

4,000

2,000

–2,000
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

Figure 15.1  Los Angeles MSA: Office net absorption

Source: Torto Wheaton Research

332
 Office market analysis: macro perspective

9,000 60

Office employment growth

Office space absorption
7,000 40

Thousands of sq. ft.

In thousands
5,000 20

3,000 0

1,000 -20

–1,000 -40
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998
Net absorption Office employment growth

Figure 15.2  Los Angeles MSA: Office net absorption and office employment growth

Source: Torto Wheaton Research

10,000 42

8,000 35
Office net absorption
Thousands of sq. ft.

Real office rents

Annaul $/sq. ft.
6,000 28

4,000 21

2,000 14

0 7

-2,000 0
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

Net absorption Real office rents

Figure 15.3  Los Angeles MSA: Office net absorption and real rents

Source: Torto Wheaton Research

14,000 42
Office space completions

12,000 36
Thousands of sq. ft.

10,000 30
Real office rents
Annual $/sq. ft.

8,000 24

6,000 18

4,000 12

2,000 6

0 0
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

Office completions Real office rents

Figure 15.4  Los Angeles MSA: Office completions and real rents

Source: Torto Wheaton Research

333
 Office market analysis

15,000 60

Office employment growth

Office space completions
12,000 40

Thousands of sq. ft.

Thousands
9,000 20

6,000 0

3,000 -20

0 1980 -40

1982

1984

1986

1988

1990

1992

1994

1996

1998
Office completions Office employment growth

Figure 15.5  Los Angeles MSA: Office completions and employment growth

Source: Torto Wheaton Research

14,000 28
Office space completions

12,000 24

Office vacancy rate (%)

Thousands sq. ft.

10,000 20
8,000 16
6,000 12
4,000 8
2,000 4
0 0
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

Office completions Office vacancy rate

Figure 15.6  Los Angeles MSA: Office completions and vacancy rate

Source: Torto Wheaton Research

14,000
Absorption/completions

12,000
T housands of sq. ft.

10,000
8,000
6,000
4,000
2,000
0
–2,000
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

Office net absorption Office completions

Figure 15.7  Los Angeles MSA: Office net absorption vs. completions

Source: Torto Wheaton Research

334
 Office market analysis: macro perspective

40,000
35,000
30,000

T housands of sq. ft.

Vacant stock 25,000
20,000
15,000
10,000
5,000
0
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998
Figure 15.8  Los Angeles MSA: Office vacant stock
Source: Torto Wheaton Research

25

20
Vacancy rate (%)

15

10

0
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

Figure 15.9  Los Angeles MSA: Office vacancy rate

Source: Torto Wheaton Research

Although office construction started declining after 1985, it remained at high levels relative to absorp-
tion. As a result, the vacant stock in the Los Angeles office market kept rising until 1992 when it peaked at
35 million square feet (Figure 15.8).The vacancy rate peaked during the same year at about 21% (Figure 15.9).
It should be noted that the considerable increase in the vacant stock in 1991 was also the result of the collapse
in demand due to the 1991 recession.With vacancy rates hovering around or above 16% from 1983 until the
late 1990s, real rents were declining for most of the 1980s and during the first half of the 1990s (Figure 15.10).
The above analysis demonstrates that a review of historical trends in critical market indicators can pro-
vide very valuable insights regarding the dynamics of the office market. As such, it underscores the need for
examining all critical office market variables in order to obtain a comprehensive view of the market and
better understand its behavior and functioning through time. Such an understanding will provide a strong
foundation for evaluating a market’s future prospects.

Assessing office market strength: overview of office market gap estimation

One of the major objectives of the macroeconomic analysis is to evaluate whether there is a demand deficit
that justifies new development of office space. The office market demand-​supply gap, GAP(t), is estimated
as the difference between new office space requirements, DS(t), and available supply, AS(t).

335
 Office market analysis

24 15

20 10

Office vacancy rate (%)

Real rent change (%)

16 5

12 0

8 –5

4 –10

0 –15
1980

1982

1984

1986

1988

1990

1992

1994

1996

1998
Office vacancy rate Real rent change

Figure 15.10  Los Angeles MSA: Office vacancy rate and real rent change
Source: Torto Wheaton Research

GAP(t) = DS(t) –​ AS(t) (15.1)

Calculating new office space requirements, DS(t)

New office space requirements include two components:

1. The space required to accommodate expected marginal growth in office-​using employment, which is
equivalent to net absorption, AB(t)
2. The space required to replace the depreciating office stock, dS(t-​1)

DS(t) = AB(t) + dS(t–​1) (15.2)

Estimating expected net absorption of office space, AB(t)

Simplistic view
Using a simple accounting framework, net absorption is thought to be proportional to expected employ-
ment growth and a space/​worker norm as described by the formula below.2

Net Absorption = Office Employment Growth * Space/​Worker (15.3)

“Informed” view
Based on a more informed view, net absorption of office space is estimated econometrically as a function
of its structural determinants, that is, office employment growth, rents, and expectations:

Net Absorption = f(Office Employment Growth, Rents, Expectations).

Estimating available supply of office space, AS(t):

At any given point in time, the available supply of office space, AS(t), within a market is a function of
two components:

336
 Office market analysis: macro perspective

1. Excess vacant stock, calculated as the vacant stock of office space from the previous period, V(t-​1)S(t-​1),
minus the required structural vacant stock, V*S(t), where V* is the structural vacancy rate
2. Completions, C(t)

Thus, available supply can be expressed as in (15.4):

AS(t) = V(t–​1)S(t–​1)–​V*S(t)+C(t) (15.4)

Therefore, the demand-​supply gap can be estimated as:3

GAP(t) = (AB(t) + dS(t–1)) –​ (V(t–​1)S(t–​1)–​V*S(t)+C(t)) (15.5)

The numerical example in Table  15.1 aims at demonstrating:  1) how the two measures of market
strength—​demand-​supply gaps and real rent changes—​can be calculated using forecasts of net absorption
and completions; and 2) that expected movements in these two indicators should provide consistent market
outlooks. Note that numbers in bold in this table denote variables that are given, and that stock and com-
pletion figures are in millions of square feet.
In order to demonstrate how the figures in Table 15.1 have been derived, the calculations for Office
Market 1 in year 1996 are presented below:

Stock 1996, S(96)
Using the stock-flow identity and assuming a 1% annual depreciation rate, S(96) was calculated as:
S(96) = S(95) (1–​d) + C(96)
= 20 (1–​0.01) + 0.120
= 19.920

Occupied stock 1996, OC(96)

Starting with the net absorption identity, OS(96) was calculated as:
ABS(96) = OS(96) –​ OS(95)  OS(96) = ABS(96) + OS(95)
= 0.360 + 17.200 = 17.560

Vacant stock 1996, VS(96)

This was calculated simply as the difference between total stock S(96) and occupied stock OS(96):
VS(96) = S(96) –​ OS(96) = 19.920 –​ 17.560 = 2.360

Vacancy rate 1996, V(96)

V(96) = VS(96) /​ S(96) = 2.360/​19.920 = 11.85%

Real rent change (%) 1996, DRR(96)

Assuming a 10% structural office vacancy rate and using the rent adjustment equation introduced in
Chapter 2, real rent change for 1996 is calculated as:
DRR(96) = a [V(96) –​V*] = –​0.80 (11.85% –​10%) = –​1.48%

Real rent index 1996, RRI(96)

Having calculated the real rent change in 1996, the real rent index can be calculated simply by applying the
1996 percentage change to the 1995 level (which is 100) as follows:
RRI(96) = RRI(95) [1 + DRR(96)] = 100 (1–​0.0148) = 98.52

337
 Office market analysis

Table 15.1  Numerical example: On the office market gap

Office Market 11 1995 1996 1997 1998 1999

Stock, S(T)2
20.000 19.920 19.961 20.161 20.460
Completions, C(T) 0.060 0.120 0.240 0.400 0.500
Vacancy rate,V(T)5 14.00% 11.85% 9.92% 7.82% 7.21%
Vacant stock,V(T)S(T)3 2.800 2.360 1.981 1.576 1.475
Occupied stock, OC(T) 17.200 17.560 17.980 18.585 18.985
Net absorption, AB(T)4 0.240 0.360 0.420 0.605 0.400
Structural vacancy rate, V* 10.00% 10.00% 10.00% 10.00% 10.00%
V(T)–​V* 4.00% 1.85% –​0.08% –​2.18% –​2.79%
Real rent change, DRR(T)8 –​3.20% –​1.48% 0.06% 1.75% 2.23%
Real rent index, RRI(T) 100.00 98.52 98.58 100.30 102.54
CPI 100.00 103.30 105.10 106.70 109.60
Nominal rent index, NRI(T) 100.00 101.77 100.30 101.83 105.33
a: Rate of rental adjustment –​0.80
d: Depreciation rate 0.01
Available supply, AS(T)6 0.928 0.604 0.365 0.030
Gap, GAP(T)7 -​0.368 0.015 0.440 0.571

Office Market 21 1995 1996 1997 1998 1999

Stock, S(T)2
14.950 15.100 15.339 15.786 16.428
Completions, C(T) 0.200 0.300 0.390 0.600 0.800
Vacancy rate,V(T) 5
9.03% 7.62% 6.45% 6.44% 7.66%
Vacant stock,V(T)S(T)3 1.350 1.151 0.990 1.016 1.258
Occupied stock, OC(T) 13.600 13.950 14.350 14.770 15.170
Net absorption, AB(T)4 0.300 0.350 0.400 0.420 0.400
Structural vacancy rate, V* 9.00% 9.00% 9.00% 9.00% 9.00%
V(T)–​V* 0.03% –​1.38% –​2.55% –​2.56% –​1.34%
Real rent change, DRR(T)8 –​0.02% 1.10% 2.04% 2.05% 1.07%
Real rent index, RRI(T) 100.00 101.10 103.17 105.28 106.41
CPI 100.00 103.30 105.10 106.70 109.60
Nominal rent index, NRI(T) 100.00 104.44 104.96 106.88 109.30
a: Rate of rental adjustment -​0.80
d: Depreciation rate 0.01
Available supply, AS(T)6 0.291 0.160 0.169 0.338
Gap, GAP(T)7 0.208 0.391 0.405 0.220

Given the office market factors provided (in bold) for Market 1 and Market 2, compute likely movements in office
rents, available supply and office market gaps from 1996 to 1999.
Notes:
1. Stock, completions, vacant stock, occupied stock, net absorption, available supply, and gap are in millions of
square feet
2. S(T) = S(T–​1)(1–​d)  +  C(T)
3. V(T)S(T) = V(T–​1)S(T–​1) + C(T) –​ AB(T) –​ dS(T–​1)
4. AB(T) = OC(T) –​ OC(T–​1)
5. V(T) = (S(T) –​OC(T)) /​ S(T)
6. AS(T) = [V(T–​1)S(T–​1) –​ V*S(T)] + C(T)
7. GAP(T) = AB(T) + dS(T–​1) –​  AS(T)
8. DRR(T) = [R(T) –​ R(T–​1)] /​ R(T–​1) = a [V(T)–​V*]

338
 Office market analysis: macro perspective

Nominal rent index 1996, NRI(96)

The real rent index RRI(96) is converted to nominal by using the following formula:
NRI(96) = RRI(96) CPI/​100 = 98.52 103.30/​100 = 101.78

Available supply 1996, AS(96)

Available supply in 1996 is calculated by adding the vacant stock in 1995 to the new office space that is
completed in 1996 and subtracting the required structural vacant stock in 1996:
AS(96) = [V(95) S(95)] + C(96) –​V*S(96) = (0.14 20) + 0.120 –​(0.10 19.920) = 0.928

DEMAND-​SUPPLY GAP  1996  The demand-​supply gap is simply calculated as the sum of the predicted net
absorption of office space in 1996 and the depreciation stock in 1995 assuming a 1% annual depreciation
rate minus the estimated available supply in 1996:

GAP(96) = ABS(96) + d S(95) –​AS(96) = 0.360 + 0.01 (20) –​0.928 = –​0.368

Obviously, the negative demand-​supply gap in 1996 implies excess supply, which is consistent with the
declining real rent index during that year.

Simple accounting techniques for forecasting office-​using employment and

office space demand
As in the case of residential real estate, the analyst can use two types of accounting approaches for esti-
mating/​forecasting demand for office space:

1 . Aggregate approaches using total office employment growth.

2. Disaggregate approaches using industry-​occupation matrices.

Aggregate approaches: using office employment growth

The aggregate approach is a quick technique used to generate a rough estimate of office employment
growth and office space requirements between year T and T+N using aggregate employment data. More
specifically, it involves the following steps:

Methodological steps
Step 1:  Obtain total employment forecasts.
Step 2: Estimate growth in office-​using employment utilizing historical shares of office employment to total
employment.
Step 3: Estimate space required to accommodate new employment growth by multiplying expected office
employment growth by space per worker requirements. A hypothetical example of the estimation of
future office space requirements using the three steps described above is presented in Table15.2. Note
that the term “white-​collar employment” is equivalent to the term “office-​using employment.”

Disaggregate approaches: using industry-​occupation matrices

Office-​using employment growth and office space requirements, or net space absorption, can be estimated—​
presumably with greater accuracy than the aggregate approach just presented—​ using a disaggregate

339
 Office market analysis

Table 15.2  Estimating office space requirements using the accounting technique1

1998 2001 forecast

Base Optimistic
Total employment 129,416 140,585 143,485
White-​collar, % of total 68.5% 70.5% 70.5%
White-​collar employment 88,650 99,112 101,157
Changes in white-​collar employment, 1998–​2001 10,462 12,507
Office space/​worker (sq. ft.) 225 225
Office space requirements (sq. ft.) 2,354,055 2,814,067

1
All numbers are hypothetical.

approach that focuses on the identification of office-​using occupations in the different sectors of the
economy. In this way, this methodology can help quantify the percentage employment in each economic
sector that uses office space. Assuming then that the occupational mix in each sector is fairly constant
through time, these percentages can be applied to forecasts of employment in the different economic
sectors to generate forecasts of office-​using employment.

Methodological steps
Step 1: Obtain the most recent matrix of employment by sector and by occupation. Using this matrix, carefully
exclude occupations and sectors, which are not office-​using.
Step 2: Estimate the percentage of office-​using workers by sector and by occupation.
Step 3: Apply the percentage of office-​using employment by sector and by occupation to employment growth
forecasts by sector and by occupation.
Step 4: Apply space-​per-​employee estimates to each occupational category to estimate the amount of space to
be demanded due to employment growth. Adjust for non-​speculative space demand.

One problem in applying this methodology is the lack of periodic employment data by occupation and by
metropolitan area. Detailed occupational data by industry are provided only by the decennial censuses. Thus, if
the analysis is performed some years after the end of the most recent decade, these industry-​occupation matrices
may be outdated. Because of the data availability issue, employment forecasts provided by the various vendors
are understandably disaggregated only by sector and not by occupation. In such a case, the analyst can still use
historical census data on occupational structure to estimate the total percentages of office-​using employment by
sector in Step 3. These percentages can then be applied in Step 4 to employment forecasts by sector.
An additional problem of the industry–​occupation approach is that it is more difficult to match the
resulting office employment measure with supply since many “office occupations” in the various SIC
classifications may not be housed in conventional office buildings, but in other non-​office structures that
incorporate some office space. DiPasquale and Wheaton (1996a) point out that an industry approach, using
establishment data collected by the government (and provided in the publication titled County Business
Patterns), allows for a more accurate assessment of the number of employees across the different SIC
classifications that tend to occupy space in separate office buildings.4

Example
Below we present an application of the disaggregate approach using 1980 industry and occupation data for
the state of California. These data have been specifically obtained from Table 224 published in the 1980
340
 Office market analysis: macro perspective

Census of Population and Housing. The presented analytical process draws heavily from the application
presented by Carn et al. (1988) in Chapter 12.

Step 1: Get the most recent matrix of employment by sector and occupation and carefully exclude occupations/​sectors that
are not office-​using.

EMPLOYMENT SECTORS CONSIDERED  All but:

• Hotels and lodging places.

OFFICE-​USING OCCUPATIONS CONSIDERED

• Managerial and professional specialty

Excluded occupations:
• public administration administrators
• education administration administrators
• health diagnosing
• health assessment and treating
• registered nurses
• teachers
• writers, authors, entertainers, and athletes.
• Technical, sales, and administrative support
Excluded occupations:
• sales occupations
• supervisors.

Step 2: Estimate the percentage of office-​using employment by sector and by occupation.

Step 3: Apply the percentage of office-​using employment by sector and by occupation to employment growth forecasts by
sector and by occupation.

The application of Step 4 requires the use of square-​feet-​per-​employee ratios. In this example, the same
square-​feet-​per-​employee ratio is used across all sectors for each occupational category. Assessment of
potential differences in such ratios across sectors may help generate more accurate estimates of office
space requirements. In using square-​feet-​per-​employee ratios for forecasting office space demand, it is also
important to keep in mind that this ratio varies over the rent cycle (as demonstrated graphically earlier). For
this reason, it matters where on this cycle the market is at the time of the analysis and where it is expected
to be over the project’s planning horizon. The variations of occupied square feet per employee over the
rent cycle underscore the importance of econometrically forecasting office space absorption, since contrary
to the simple accounting techniques described here, this technique can scientifically and systematically
incorporate such cyclical effects over a given forecast period.

Step 4: Apply space-​per-​worker estimates to each occupational category to estimate the space required to accommodate forecast
office employment growth. Adjust for non-​speculative space demand.

In the case that econometric forecasting is not feasible, then the analyst needs to adjust the space-​
per-​worker estimates judgmentally based on an understanding of recent market trends and how they

341
 Office market analysis

Table 15.3  Office-​using employment sectors

Total Managerial Office-​ Using [2]‌ as Total Technical, Office-​ Using [5]‌ as
and Professional Managerial and percent Sales, and Technical, Sales percent
Specialty Professional of [1] Administrative and Administrative of [4]
occupations Specialty Support Support
occupations
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌ [6]‌
Agriculture, Forestry, and 21,504 15,938 74.1% 19,252 16,572 86.10%
Fisheries
Mining 9,028 8,798 97.5% 7,780 6,864 88.20%
Construction 65,800 62,112 94.4% 62,417 54,278 87.00%
Manufacturing 453,992 410,362 90.4% 479,626 380,092 79.20%
Transportation, 107,805 93,067 86.3% 304,925 254,641 83.50%
Communication, and
Other Public Utilities
Wholesale and Retail 312,815 277,427 88.7% 1,036,070 259,594 25.10%
Trade
FIRE 197,252 192,390 97.5% 528,103 302,472 57.30%
Business Services 136,021 102,653 75.5% 152,959 124,857 81.60%
Repair Services 15,404 15,138 98.3% 22,474 14,060 62.60%
Other Personal Services 20,177 11,946 59.2% 21,053 11,982 56.90%
Entertainment/​ 77,880 27,867 35.8% 34,317 21,236 61.90%
Recreation Services
Professional and Related 1,073,946 322,155 30.0% 583,142 554,173 95.00%
Services
Public Administration 158,969 120,307 75.7% 199,541 174,690 87.50%
Total 2,671,761 1,660,160 62.14% 3,469,277 2,175,511 62.71%

Note: Occupations not included are service occupations; farming; forestry and fishing; precision, production, craft, and
repair occupations; construction trades; extractive occupations; operators, fabricators, and laborers; transportation and
material moving; handlers, equipment cleaners, helpers, and laborers.

may differ in the future. Given that occupied square feet per employee vary with an area’s industry
and occupational mix, as well as prevailing market rents, it is best for the analyst to use local experi-
ence data as opposed to a national or regional average. Recent square-​feet-​per-​employee ratios, based
on actual building data, can be obtained by state, province, or region and potentially metropolitan area
from the online Office Experience Exchange Report that is available from BOMA (https://​eer.boma.
org/​BOMA/​marketing.aspx), which is published annually. Note that it is very important to gain a full
understanding of what exactly the square-​feet-​per-​employee ratios derived from these data represent
before using them.
When some historical data for the local market are available, the analyst may be better or not using a his-
torical average, depending on the part of the rent cycle represented by the available historical information
and the stage of the cycle the market is expected to be during the forecast period. For example, assume
that the available historical data are relatively short, centered around the trough of the rent cycle, thereby
signifying a period during which occupied space per employee may be the highest. In addition, assume that
market rents during the forecast period are expected to be centered around their peak, thereby signifying
a period during which occupied square feet per employee may be the lowest. In such a case, using the his-
torical average could contribute to significantly overestimating the square feet/​employee to be applied to
forecast employment. For this reason, when using accounting techniques (aggregate or disaggregate) for

342
 Office market analysis: macro perspective

Table 15.4  Office-​space using employment forecast

1993–​1995 1993–​1995 Percent Percent Growth in office-​ Growth in

Forecast growth Forecast Growth office- office-​ using Managerial office-​using total
in Managerial in total Technical, using of using of and Professional Technical, Sales,
and Professional Sales, and [1]‌ [2]‌ Specialty and Administration
Specialty Administration occupations Support
occupations Support
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌ [6]‌
Agriculture, Forestry, 1,000 950 74.10% 86.10% 741 818
and Fisheries
Mining 500 400 97.50% 88.20% 487 353
Construction 3,300 3,300 94.40% 87.00% 3,115 2,870
Manufacturing 22,500 24,000 90.40% 79.20% 20,338 19,019
Transportation, 5,000 15,000 86.30% 83.50% 4,316 12,526
Communication, and
Other Public Utilities
Wholesale and Retail 15,500 50,000 88.70% 25.10% 13,747 12,528
Trade
FIRE 9,900 25,000 97.50% 57.30% 9,656 14,319
Business Services 6,800 8,000 75.50% 81.60% 5,132 6,530
Repair Services 800 1,000 98.30% 62.60% 786 626
Other Personal Services 1,000 1,200 59.20% 56.90% 592 683
Entertainment/​ 3,900 1,750 35.80% 61.90% 1,395 1,083
Recreation Services
Professional and 53,500 29,160 30.00% 95.00% 16,049 27,711
Related Services
Public Administration 8,000 9,800 75.70% 87.50% 6,054 8,579
Total 132,700 170,440 82,409 107,645

Note: [5]‌ = [1]*[3]; [6] = [2]*[4]

forecasting office space absorption, the analyst needs to be very cautious in terms of the square-​feet-​per-​
employee ratio applied to forecast office employment.
Since square-​ feet-​
per-​employee ratios do not distinguish between owner-​ occupied and specula-
tive office space, estimates need to be adjusted to reflect demand for speculative space. For example, not
all office-​using occupations in the manufacturing sector use speculative office space, as a share of these
employees may be housed at company-​owned production plants; similarly, some government employees
may be housed in government-​owned office buildings (see Carn et al., 1988). One way to sort out specula-
tive office space needs is to adjust the estimates using local, regional, or national survey data of speculative/​
non-​speculative space usage by economic sector. If such data are not available, then local brokers may be
able to provide some idea about the ratio of speculative to non-​speculative space in the overall local market,
which can be used instead in order to carry out the required adjustments.

Office supply analysis using accounting approaches

Demand analysis needs to be complemented with supply analysis in order to fully evaluate office market
strength. Carn et al. (1988) suggest segmentation of the existing office inventory by several categories and
subcategories based on the type of space (prestigious, general use, supporting use, mixed-​use), amenity level

343
 newgenrtpdf
Table 15.5  Speculative office-​space requirements

1993–​1995 Forecast Sq. ft./​ 1993–​1995 forecast growth Sq. ft./​ Total space Adjustment factor: Speculative space
growth in Managerial and employee in total Technical, Sales, and mployee demanded speculative space demanded
Professional Specialty Administrative Support
occupations
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌ [6]‌ [7]‌
Agriculture, Forestry, and 741 300 818 180 369,545 100% 369,545
Fisheries
Mining 487 300 353 180 209,701 100% 209,701
Construction 3,115 300 2,870 180 1,451,056 100% 1,451,056

Office market analysis

Manufacturing 20,338 300 19,019 180 9,524,800 50% 4,762,400
Transportation, 4,316 300 12,526 180 3,549,689 100% 3,549,689
Communication, and
344

Other Public Utilities

Wholesale and Retail Trade 13,747 300 12,528 180 6,378,965 100% 6,378,965
FIRE 9,656 300 14,319 180 5,474,177 100% 5,474,177
Business Services 5,132 300 6,530 180 2,714,997 100% 2,714,997
Repair Services 786 300 626 180 348,466 0% 0
Other Personal Services 592 300 683 180 300,551 0% 0
Entertainment/​Recreation 1,395 300 1,083 180 613,577 0% 0
Services
Professional and Related 16,049 300 27,711 180 9,802,623 100% 9,802,623
Services
Public Administration 6,054 300 8,579 180 3,360,619 20% 672,124
Total 82,409 107,645 44,098,765 35,385,276

Note: [5]‌ = [1]*[2] + [3]*[4]; [7] = [5]*[6].

For simplicity, space per worker in [2]‌is assumed to be the same across sectors.
 Office market analysis: macro perspective

(high, normal, low) and location (central city, suburban). Such a segmentation can help the analyst better
understand the local office market, but it would require collection of primary data through an extensive
survey, since secondary data available from vendors do not usually provide such a detailed description of a
market’s existing office inventory. For projects located within large metropolitan areas, such primary data
collection is typically undertaken within the project’s submarket.
A refined segmentation of an area’s inventory is always preferable at the metropolitan level too, and
can be very useful if demand forecasts could be developed econometrically at the same level of segmenta-
tion. This is not feasible, however, because it requires historical time-​series for all the relevant variables by
segment. It is very unlikely that such data will be ever available to the analyst. Furthermore, econometric
modeling at this level of disaggregation could be very difficult and complicated because of substitutability
across segments.
The ultimate objective of the supply analysis is to generate forecasts of the components of avail-
able supply over the project’s planning horizon, which can then be used for the estimation of the office
market gap. As (15.3) indicates, such components include the vacant stock, the structural vacant stock,
and completions.

Vacant stock
Each period’s vacant stock can be estimated as the sum of the previous period’s vacant stock and completions
minus net absorption.Thus, starting from today’s vacant stock the analyst can generate a vacant stock series
for the forecast period if forecasts of net absorption and completions are available. Existing inventory and
vacancy information for the nation’s 50+ largest markets is readily available through secondary data sources.
In assessing a market’s current vacancy rate, sublet space should not be ignored because it takes out from net
absorption and could exert downward pressures on rents, since it is usually available at a discount. As such,
it could seriously affect the viability of new projects.

Completions
The key variable in forecasting the office market gap from the supply side is the new office space expected
to be completed in the market over the project’s planning horizon. Accounting techniques can provide
relatively reliable estimates of the most likely level of office space completions one to two years ahead at
the most, based on a survey of projects under construction and projects permitted, but not started. Projects
that are still in the planning stage can provide some clues for potential construction further ahead, but there
is much greater uncertainty as to whether they will be really completed and when.

Critique
The aggregate and disaggregate approaches can help quantify the potential demand for office space in a
given market. However, this quantification is too simplistic to deal with the intricacies of office market
dynamics and ignores important details about behavioral relationships. By their nature, these techniques
cannot account for the period by period interactions between office space demand, supply, and rents, and
the feedback and lag effects that drive their movements. Even if these estimates were to be relied on, they
are not enough to guide a development and investment decision. There are limitations on the supply side
as well. It may be possible to quantify new office space, most likely to be completed in the next one to two
years, based on projects that are under construction and well into the pre-​development phase. However,
that is hardly enough for guiding informed investment decisions since office development projects may
have a minimum of a three-​to five-​year planning horizon, including leasing-​up time. How does the analyst

345
 Office market analysis

make a decent estimate of the office space that will be available four or five years ahead of the time of
analysis? Moreover, how does she/​he ensure that the office supply estimate is consistent with the demand
forecast, since demand, supply, and rents interact in a dynamic way?
Against this background, the emphasis of the next section will be on how to analyze in an integrated
fashion the dynamic behavioral relationships of office markets, and use them as a basis for more accurately
forecasting all the critical market indicators.

Econometric forecasting of office markets

Econometric models treat demand, supply, and rents in an integrated fashion and provide more sophisticated
tools for forecasting office markets. Such models describe the intertemporal behavior of the basic office
market variables, that is, net space absorption, new construction, office rents, and vacancy rates.The under-
lying objective of econometric modeling is to quantify not only the direct and indirect effect of exogenous
influences, such as office employment growth, on these fundamental office market indicators, but also their
interactions and the adjustment lags that condition these interactions. The typical office model formulation
represents a variant of the stock-​flow model that appropriately incorporates features that are unique to
metropolitan office markets.

Forecasting office space absorption, new construction, vacancies, and rents

The demand-​supply dynamics of the office market are well described by an expanded stock-​flow model,
which is more sophisticated than the one we discussed in the case of the housing market. To better under-
stand the workings of this model, assume that the market is initially at equilibrium. What would happen
if there were an exogenous demand shock caused, for example, by an unexpected huge increase in office
employment?
The response of the office market to a positive demand shock, within the context of the expanded
stock-​flow model, is depicted by the flow-​chart in Figure 15.11. As this figure demonstrates, employment
growth boosts net absorption and aggregate demand for office space. The latter interacts with the existing
inventory of office space to determine a new vacancy rate in the market, which, in turn, triggers office rental
adjustments and shapes new rent levels.The new rent levels trigger feedback effects to both the demand for
and supply of office space, which in turn shape a new vacancy rate and a new wave of adjustments in rents
and so on. These adjustment processes are inherently slow due to time-​consuming search processes, the
long-​term leases that characterize commercial office markets, and the long investment-​decision and con-
struction lags. The speed of demand and rent adjustment processes may be also influenced by idiosyncratic
local market features, such as spatial heterogeneity and tenant base diversity, as they may be associated with
greater informational inefficiencies and contribute to longer search processes on the part of office tenants
(see Sivitanidou, 2002). The sluggish demand adjustments and the long construction lags are two of the
major contributors to the cyclical nature of office markets.
The interactions underlying the stock-​flow model in Figure 15.11 can be described by a set of behav-
ioral relationships and identities that in essence depict the functioning of the office market. The fun-
damental behavioral relationships are captured by the absorption, construction, and rent-​adjustment
equations. The two identities provide the means for calculating how office space absorption and
completions shape the market’s total stock and vacancy rate. The typical model equations are provided in
Box 15.1 (Wheaton, 1987).
It should be noted that all three behavioral equations in Box 15.1 are based on the notion of the
partial adjustment of the office market towards some desired levels of demand, supply, and vacancy. For
example, the absorption equation formulation recognizes that net absorption of office space is only part

346
 Office market analysis: macro perspective

Employment
Employment
Growth
growth

Net
Net
Absorption
absorption

Aggregate
Office
office space
Space
Demand
demand
Vacancy
Vacancy
Rate
rate

Office Space Rent

Office space Rent
Stock Adjustment
stock adjustment

New Office
New Office
Construction Rent Level
construction rent level

Figure 15.11  Office market dynamics: The expanded stock-flow model

of the desired additional space by companies because of institutional restrictions, such as long-​term leases,
that prevent growing office-​using firms from immediately and easily adjusting their office space usage.
It should be also noted that rents need to be introduced with a lag of several periods in the completion
equation, because of the long time that it takes to plan and complete an office building, as well as regulatory
requirements that often prolong the process.
The econometric analysis presented by Wheaton (1987) provides the following important conclusions
regarding the behavior of the national office market:

• Net absorption depends on rents, as well as employment growth and expectations (according to BOMA,
68% of changes in office space needs are attributable to expansion due to internal growth, while the
remaining 32% comes from new office firms).
• Office supply is more responsive to rental rates or lagged vacancy rates than demand, and this tends to
generate great instability in the nation’s commercial office market.
• Office space construction does not only depend on vacancy rates and/​or rents, but also on expectations
regarding office employment growth.

Alternative rental adjustment models

A number of alternative office rent adjustment formulations have been presented in the literature over the
years. The traditional model presented below was inspired by rent adjustment models originally developed
for the rental housing market. Rosen (1984) and Shilling, Sirmans, and Corgel (1987) were the first to
apply this specification to the analysis of rent changes in the office market.

347
 Office market analysis

Box 15.1 Typical office model equations

Behavioral equations
Net Absorption: AB = f[E(t), R(t), Xd(t), OC(t–​1)] (15.7)
Derived from: AB = g[OC*(t)–​OC(t–​1)]
OC*(t) = f[E(t), R(t), Xd(t)]
where
AB : net absorption
g : adjustment coefficient
OC* : desired occupied stock
E : office employment
R : office rent
Xd : other exogenous determinants of desired demand
Premise: Net absorption is the result of aggregate office space demand adjustments towards a desired level, OC*,
determined by rents, office employment, and other exogenous shifters. Such adjustments are sluggish, as only
a fraction g of the difference between prevailing and desired occupied stock is eliminated during each period
through space absorption.
New Supply: C(t)=f[S(t–​1), R(t), Xs(t)] (15.8)
Derived from:C(t)=a[S*(t)–​S(t–​1)]
S*(t)=f[R(t), Xs(t)]
where
C : new construction
a : adjustment coefficient
S* : desired stock
Xs : other exogenous determinants of desired stock
Premise:  Completions are the result of aggregate office space stock adjustments towards a desired level, S*,
determined by office rents and other exogenous supply shifters.

Rental adjustment (traditional formulation):

R(t)–​R(t–​1)/​R(t–​1)=a[V*–​V(t–​n)] (15.9)

Premise: Office rent changes occur in response to deviations of the market vacancy rate, V(t), from the equi-
librium or structural vacancy rate, V*. A  more elaborate discussion of this and alternative rent adjustment
specifications is presented in the next section.
Identities

Stock-flow: S(t)=S(t–​1)(1–​d)  + C(t)

Vacancy rate: V(t) = [V(t–​1) S(t–​1) –​ AB(t) + C(t)] /​ S(t)

348
 Office market analysis: macro perspective

The traditional formulation: The structural vacancy rate model

[R(t)–​R(t–​1))/​R(t–​1) = a[V*–​V(t–​n)] (15.6)

where
R : real office rent
a : speed of rental adjustment (1>a>0)
V,V* : nominal and structural vacancy rates, respectively

The fundamental premise of the structural vacancy model is that office market equilibrium is defined
by a “natural” or structural vacancy rate, V*, and that whenever the market vacancy rate is below or
above this rate, office rents adjust accordingly in order to bring it back to its equilibrium level. Thus,
real office rents must be decreasing when there is excess supply (V>V*) and increasing when there
are supply shortages (V<V*). This behavior is described mathematically below and graphically in
Figure 15.12.

• If V(t)>V* then [R(t)–​R(t–​1))/​R(t–​1)  <0

• If V(t)<V* then [R(t)–​R(t–​1))/​R(t–​1)  >0

Vt
Rt

V*

vt <v* vt >v*

vt is decreasing vt is increasing vt is increasing vt is decreasing

below v*; below v*; above v*; above v*;
rents must be rents must still be rents must be rents must still be
increasing at an increasing, but at a declining at an decliniing, but at a
increasing rate decreasing rate increasing rate decreasing rate

Figure 15.12  Rent-​vacancy adjustments and the structural vacancy rate

349
 Office market analysis

The concept of the structural vacancy rate is analogous to the concept of the structural unemployment
rate. On the demand (tenant) side, V* allows for lower search, information, and relocation costs. On the
supply (landlord) side, it allows landlords facing uncertain demands to take advantage of future (unex-
pected) increases in demand and subsequent increases in rental rates. Thus, V* should depend on expected
economic conditions, as well as marginal inventory costs. Notice that in this formulation, the structural
vacancy rate is assumed to be constant through time. Below we present the estimated parameters of this
model for Dallas (Sivitanides, 1997).

EXAMPLE

Statistical equation for Dallas (1981–​1988)

[R(t)–​R(t–​1))/​R(t–​1) = 0.05 –​ 0.44V(t-​1) (15.10)

          (1.64) (–​
3.19)
R2 = 0.42

The structural vacancy rate, V*, can be estimated as 0.05/​0.44 = 0.11 or 11%.5

ALTERNATIVE 1: TRADITIONAL MODEL WITH VARIABLE STRUCTURAL VACANCY RATE  The traditional
rental adjustment model assumes that the structural vacancy rate is intertemporally constant. An
alternative specification that relaxes this restriction takes the form below, in which the structural
vacancy rate is assumed to vary with indicators of market strength, such as net absorption AB
(Sivitanides, 1997):

[R(t)–​R(t–​1)]/​R(t–​1) = a[V*(AB(t–​m)) –​ V(t–​n)] (15.11)

Premise: The underlying premise of this formulation is that landlords determine the optimal level of vacan-
cies based on their expectations of market strength. Assuming myopic landlords, such expectations are
thought to be determined by current or recent levels or changes in demand and/​or supply indicators. For
example, high absorption may trigger perceptions of a strong market and expectations for rising rents,
thereby shaping a higher structural vacancy rate.6 As absorption fluctuates through time, the optimal office
space inventory must fluctuate as well. An example of an estimate of such a model for Philadelphia is
presented below (Sivitanides, 1997).

EXAMPLE  Philadelphia (1981–​1988)

[R(t)–​R(t–​1)]/​R(t–​1) = 0.01 + 0.09 AB(t) –​ 1.3 V(t–​2) (15.12)

         (0.17) (4.64)    (–​
3.79)
R2=0.64

ALTERNATIVE 2:  THE EQUILIBRIUM OR DESIRED RENT MODEL  A recently formulated alternative by

Wheaton and Torto (1994) is explicitly based on the search theory and the notion of a “desired” rent. As
such, it takes the form below:

R(t)–​R(t–​1) = a [R*(AB(t–​n),V(t-​m)) –​ Rt–​1]; 1 >a> 0 (15.13)

350
 Office market analysis: macro perspective

where:
R* : desired or equilibrium rent
AB : net absorption

Premise: Office markets are characterized by a “desired” or “equilibrium” rent, as opposed to an equilib-
rium or structural vacancy rate advanced by the traditional model. This equilibrium rent is thought to
be determined through a bargaining and negotiation process between tenants and landlords, and assumed
to be shaped by expected lease-​up time. Expected lease-​up time is determined, in turn, by tenant flows and
vacant space. With longer expected lease-​up time, landlords would be willing to accept lower rents. Within
this context, absorption (capturing tenant flows) and the vacancy rate are expected to have a positive and
negative effect, respectively, on the “equilibrium” office rent and, hence, on changes in prevailing market
rents.7 Given market imperfections, rents only gradually adjust towards their “desired” level. An example of
the estimate of this model for Denver is presented below (Wheaton and Torto, 1994).

EXAMPLE  Denver (1979–​1991)

R(t)–​R(t-​1) = 4.64 + 16.07 AB(t–​n) –​ 0.10 V(t–​m) –​ 0.23 R(t–​1) (15.14)

      (5.76) (3.59)      
(–​
8.27)   (–​
4.55)
R2=0.75

Submarket analysis
As we are concluding the discussion of the office market analysis at the macro level and getting ready to
move on to the microeconomic analysis, an important question that comes to mind is the following: Does
the state of metropolitan office markets reflect the state of various submarkets developed around office
nodes, or should these submarkets be analyzed separately? This question refers to the intra-​metropolitan
structure of office markets and is a very intriguing one. In considering this question, two arguments can
be advanced:

• Metropolitan office markets are reasonably integrated because of location substitutability. Given location
substitutability, all nodal locations are subject to similar macro influences.Thus, although levels of micro-​
market indicators may vary across nodes, intertemporal movements in these indicators may parallel each
other. This argument supports the level of aggregate analysis presented so far.
• Metropolitan office markets are segmented. Different submarkets may be responding differently to
macro influences, depending on the type of office activities they house, their degree of heterogeneity,
and/​or prevailing land market conditions. For these reasons, nodal cycles may not necessarily par-
allel each other. This argument points to the need for the submarket-​specific forecasting of market
conditions.

The conventional view is that contemporary metropolitan office markets are neither fully integrated
nor perfectly segmented. Given that office vacancy and rental rate levels vary across office submarkets
within a metropolitan area, submarket analysis is a typical part of office market studies. Two of the major
objectives of such an analysis include: 1) the assessment of the specific rent and vacancy levels that pre-
vail in the submarket and 2) the assessment of the share of metropolitan absorption most likely to be
captured by the submarket. The estimated submarket absorption is then used as a basis for assessing pro-
ject absorption.

351
 Office market analysis

Types of office nodes
Typically, modern metropolitan office markets are composed of (sometimes) stylized location clusters or
nodes that are specialized in terms of the type of office employment they attract. The different types of
nodes typically found in contemporary metropolitan areas include (McMahan, 1989):

Downtown district
Office space concentration at the traditional city center, which typically attracts financial institutions, legal,
and accounting firms, company headquarters, and government agencies. It offers superior accessibility to
dispersed labor or clientele, as well as increased access to other firms.

Uptown district
This is usually a linear district developed along a major arterial axis connecting the CBD with the suburbs.
Space demand here comes mainly but not necessarily from location shifts, usually from downtown.

Office parks
These are suburban developments mostly devoted to specialized service activities, light manufacturing, and
distribution. They are characterized by low-​density and ample parking space. As such, their development
requires more land and this is the main reason why they are located in the suburbs where land is cheaper.

Special nodes
These nodes are usually specialized in specific types of office-​using activities, such as medical offices close
to hospitals; law offices close to a courthouse; R&D nodes close to universities; and firms frequently using
air travel or engaging in businesses associated with airport activities located close to airports.

Shopping center office nodes

Office space concentrations next to or within shopping centers, usually directed towards smaller users, most
of which serve households residing within the shopping center’s trade area.
It should be noted that the presence of some degree of specialization across office submarkets does not
necessarily imply that their reaction to metro-​wide forces will deviate considerably from aggregate metro-
politan market behavior and no formal research has provided thus far compelling evidence to contest this
argument. In any case, however, since submarket conditions may deviate from overall metropolitan market
conditions, at least in terms of levels, analysis at the submarket level is warranted.

How to analyze office submarkets

In theory, the most appropriate approach to perform submarket analysis is to apply to the submarket level
the econometric and other methodologies discussed earlier within the context of the macroeconomic ana-
lysis of office markets. In most cases, however, standard time-​series techniques cannot be applied because
of the limited availability of sufficiently long historical data at the submarket level. Besides the scarcity of
data, an additional difficulty in applying dynamic econometric analysis at the submarket level is the need
for modeling the interactions between the different submarkets that compose the aggregate metropol-
itan market.8,9 Given these difficulties, the only sensible alternative is to use metropolitan-​level absorption

352
 Office market analysis: macro perspective

forecasts and distribute them across the area’s submarkets. This requires the estimation of submarket cap-
ture rates, which can then be applied to the metro-​wide absorption forecast. Submarket-​specific demand
estimates can be used in combination with supply estimates to evaluate submarket strength by calculating,
for example, the office market gap, using the same formulas discussed earlier.
The key element of such an analysis lies in the accurate assessment of a submarket’s share of metro-​wide
employment growth or net absorption. This task can be accomplished by explaining econometrically net
absorption share variations across submarkets or nodes using cross-​sectional data that are more readily avail-
able at the submarket level than time-​series data. The econometric specification needs to account for all
factors that are likely to influence the distribution of metro-​wide absorption across the area’s submarkets.
Such factors, listed in more detail below, include characteristics that affect nodal competitiveness as it
pertains to productive and worker amenities, and the amount of available office space (vacant stock) in each
submarket at the beginning of the period for which submarket absorption is measured.10

Nodal competitiveness

• Specialization/​agglomeration
• Prestige
• Rents
• Accessibility advantages/​disadvantages (to airport, downtown, major freeway junctions), including
congestion
• Transportation infrastructure (mass transit)
• Support facilities (shopping amenities, restaurants, hotels, business support services)
• Overall quality of existing office space
• Available vacant space (see note 3)
• Neighborhood externalities.

Location amenities pertaining to employees

• Crime rates
• Quality of education
• Proximity to amenable residential areas.

In the absence of sufficient data to econometrically estimate submarket absorption rates, nodal capture can
be estimated on the basis of current intra-​metropolitan distribution of employment growth or office space
absorption, the trends underlying such distribution, and the competitive position of the submarket vis-​à-​vis
the other submarkets. This is usually a three-​step process involving:

1. Estimation of current shares of employment (by sector and occupation if possible), or net office absorption
across the area’s submarkets. It would be prudent to compare any submarket employment estimates
by sector and occupation against employment data that can be obtained from the U.S. Department of
Commerce, in order to evaluate their reasonableness.
2. Adjustment of these shares by evaluating recent trends in intermodal shifts and the extent to which such
trends are likely to continue, stabilize, or reverse.
3. Refinement of the estimated share by comparing it to the submarket’s fair market share adjusted by its
competitive position index. A competitive position index (CPI) for the submarket can be developed
using the competitive differentials technique across competing submarkets in the same way it is
applied across competing properties. The difference is that the attributes listed in the table with the
characteristics of each submarket that affect its attractiveness to firms using office space will represent

353
 Office market analysis

submarket averages as opposed to attributes of individual properties. Once derived, the submarket’s
CPI can be used to adjust accordingly (upwards or downwards) its fair market share depending on
whether the submarket is superior or inferior than average (Koepke, 2019).

An issue often encountered when trying to allocate metro-​wide absorption between the downtown and
the suburbs is the historically shrinking absorption share of the former. Looking at these historical trends,
analysts may be tempted to assume that downtown absorption shares will be smaller in the future than they
have been in the past. However, the validity of such an assumption needs to be evaluated in the context of
the fundamental dynamics and prospects of the downtown–​suburban dichotomy in each market, taking
into account prevailing market conditions, as well as the potential effect of differences in rental rates and
the availability of space across these two submarkets.
Once submarket employment or absorption shares are estimated, the same process of demand-​supply
gap analysis used at the metro level can be adopted.Vacant stock information by submarket is usually readily
available from secondary sources. What may be considered as a “normal” or equilibrium vacancy rate in a
given submarket can be determined by talking to local brokers and local real estate market players. If suf-
ficiently long historical vacancy data (over a full cycle) are available for the submarket, then the average
vacancy can be considered as a reasonable approximation of the submarket’s structural vacancy rate.
Absorption share analysis needs to be complemented by a good understanding of vacancy, new construc-
tion, and rental rate trends in the submarket, and if possible, an evaluation of the submarket’s position on the
cycle compared to the metro area. For example, the submarket may have a considerably higher vacancy rate
than the metro area, suggesting that it may be recovering at a slower pace than the broader market.11
Analysis of the character or specialization of an office node, in terms of the sectoral composition of its
current tenant mix, will provide clues with respect to the type of firms that are more likely to be attracted
to that submarket in the future. Examination of growth trends in the industries represented by these firms
will provide further insights regarding the office space demand prospects for the node under consider-
ation. Logically, the node with the greatest excess demand and most restrained supply outlook could be
considered as the most promising in terms of registering the strongest increases in real office rents.

Chapter summary
In this chapter, we reviewed the first phase of office market studies, which focuses on the analysis of broader
market strength and future prospects, with emphasis on the positioning of the market in the real estate cycle.
Assessment of the office space demand-​supply gap and the direction of real rent changes can help make
such an evaluation.We reviewed various methodologies for estimating the office space demand-​supply gap:

• Aggregate accounting techniques, where demand for office space is calculated by applying a square-​feet-​
per-​employee norm to a forecast of aggregate office employment growth; and disaggregate accounting
techniques, where this same approach is applied separately to each of the employment subsectors and
occupations that contribute to office space demand.
• The econometric approach that models and quantifies in an integrated fashion the behavioral relationships
among the key office market variables, that is, office space absorption, vacancy rates, completions, and
rental rates; these relationships are then used to forecast future office market movements on the basis of
an exogenous economic forecast.

As in the case of residential gap analysis, the reliability of the accounting approaches is highly questionable
when used for medium and longer-​term forecasts of market strength, while econometric approaches are
technically and scientifically more robust and reliable in forecasting office market behavior, both in the
short and the long term.

354
 Office market analysis: macro perspective

Questions
1. What are the major steps in carrying out a macroeconomic analysis for a given office market?
2. Describe the basic indicators that need to be examined when analyzing an office market from
the macro perspective. What factors need to be considered when evaluating the prospects of these
indicators?
3. What is the basic formula for calculating the demand-​supply gap in an office market? Explain the role
of the structural vacancy rate in this calculation.
4. How can an analyst estimate econometrically a market’s structural vacancy rate?
5. Describe briefly the different techniques for forecasting office space demand at the macro level.
6. Discuss the steps involved in the econometric forecasting of a metropolitan office market. Describe
briefly the components of a typical office market forecasting model and the rationale behind the
model’s equations.
7. Why may simple employment-​multiplier techniques for calculating future office space demand pro-
vide misleading forecasts?
8. What are the steps in submarket analysis for office space?
9. How would you go about forecasting a submarket’s office absorption share econometrically? Discuss
the required data and econometric formulation.
10. You are interested in developing office space in a metropolitan market that is expected to experience
considerable growth. The following data series for the market in question have been provided to you.

Year 2019 2020 2021 2022 2023 2024

Stock (in million sq. ft.) 90.000 89.950 90.000 90.050 90.200 90.449
Completions (in million sq. ft.) 0.300 0.400 0.500 0.500 0.600 0.700
Nominal vacancy (%) 14.00 12.00 10.88 9.70 8.85 7.90
Expected change in real rents (%) –​1.8 0 1.01 2.07 2.88 3.69

a) Using these data, develop a measure of the need for additional office space in year 2023.
b) Is your answer in (a)  consistent with the expected change in real rents presented in the table
above? Explain your answer.
c) Would you consider having new space ready for occupancy in this market in 2020? Explain your
answer.

Note that the depreciation rate is 1% and the market’s expected structural vacancy rate over the forecast
period is 12%.

Project #5: Econometric forecasting of office markets

The purpose of this project is to better acquaint the students with the workings of office markets and the
applications of econometric models in office market analysis.

The case
The year is 2019. Assume that you are the analyst for a national investor who is contemplating whether to
invest in office property in metropolitan area X. A national real estate forecasting firm has provided you
with the equations that describe the behavior of that area’s office market. In addition, you have purchased
economic forecasts for the metropolitan area in question from a macroeconomic forecasting firm. Your

355
 Office market analysis

objective is to first understand how market X is more likely to behave under the base-​case and alternative
economic scenarios using these behavioral equations, and then assess and evaluate the implications of the
alternative market forecasts for the investor’s plans.

The questions
First, you decide to generate the base-​case forecast and evaluate its investment implications. Of course,
you are well aware that your analysis will need to be complemented with submarket-​specific and project-​
specific analyses at a later point.
Second, you decide to explore the sensitivity of the model to two alternative office employment growth
outlooks of your own choosing, one considerably more optimistic than the base-​case forecast and the other
considerably more pessimistic. You specifically focus your attention on how the different employment
growth scenarios impact the following four endogenous variables of the model:

• Net absorption
• Vacancy rate
• Rent inflation
• Completions.

Do these variables respond in predictable ways to the optimistic and pessimistic economic scenarios? How
quickly? In what ways does the investment outlook of the office market change? What insights do you
gain from this exercise with respect to overall sensitivity of the market and the risks and/​or opportunities
that may lie ahead?
Third, you decide to discard the endogenous completion figures generated by the model and use your
own completion series. Examine the impact of higher and lower completion figures on the three first
indicators listed above. Do these change in predictable ways? How quickly?

The report
Write up your findings in five to seven pages.Try to provide precise explanations and document your ana-
lysis with tables and/​or graphs. Answer all of the questions asked.

The data
The base case economic forecast, the required starting values for the real estate market variables, and the
three forecasting equations for metropolitan market X are provided below.

Absorption equation:
ABSt = –​272.8 + 104.7 SERVt + 184.4 DSERVt + 185.8 FIREt –​ 2750 RRt–​1

where
ABSt : net office absorption in period t
SERVt : service employment in period t
DSERVt : change in Service employment in period t (SERVt –​ SERVt-​1)
FIREt : FIRE employment in period t
RRt-​1 : real office rent in period t–​1

356
 Office market analysis: macro perspective

Table 15.6  Required starting values and base case economic forecast

FIRE employment Services employment Occupied stock Completions Vacancy Real rent
(thousands) (thousands) (thousands) (thousands) rate (%)
2018 156.3 332.9 147,152 3,500 19.0 23.75
2019 157.2 333.9 146,752 4,400 21.1 22.95
2020 157.6 330.5
2021 163.0 347.2
2022 169.3 367.8
2023 173.6 380.6
2024 178.1 393.9

Real rent equation:

RRt = 4.59 –​ 0.10 Vt–1 + 0.908 RRt–​1

where
Vt–​1: vacancy rate in period t-​1

Completions equation:
COMt = –​9357 + 560 RRt-​1 + 146 DRRt-​1 + 0.40 COMt–​1

where
COMt: completions in period t
DRRt-​1: change in real office rent in period t–​1 (RRt–​1 –​  RRt–​2)

Use a 0.5% depreciation rate for creating an inventory series. Note that the different variables enter the
forecasting equations in the same units they are provided in the table. In particular, FIRE employment,
Service employment, and completion figures should be in thousands and the vacancy rate in percent. If you
need to convert real office rents to nominal, assume a 2.5% inflation rate.

Notes
1 One could argue that there is some substitutability between the two markets in the sense that when the
microeconomics of owner occupancy do not favor this tenure mode, some demand will have to necessarily shift to
the rental market.
2 BOMA (1999) reports an average of 293 square feet per office worker.
3 Note that V(t) S(t) = (V(t-​1) S(t-​1)+C(t) –​ AB(t) –​ d S(t-​1)) .
4 These are labeled in the County Business patterns as “separate administrative or auxiliary structures.”
5 By contrasting the theoretical formulation of the traditional model, described below by (F85.1), with the implied
statistical equation that was estimated, described by (F85.2), we can conclude that b0 = aV* and b1 = a. Therefore
V* = b0/​b1.

[R(t)–​R(t-​1))/​R(t-​1) = a[V*–​V(t-​–​n)] = aV* –​ aVt-​n(F85.1)

[R(t)–​R(t–​1))/​R(t–​1) = bo + b1 Vt–​n(F85.2)

6 All else being equal, higher absorption should contribute to a higher structural vacancy rate and, therefore, to
higher rent change. For a given structural vacancy rate level, higher current vacancy should signify a greater amount

357
 Office market analysis

of excess supply (if it is above the structural vacancy rate) or a smaller amount of supply shortages (if it is below the
structural vacancy rate), thereby exerting negative influences on office rent change.
7 For a given level of vacant stock, greater absorption shortens lease-​up time, implying a higher R*, while for a given
level of absorption greater vacancy lengthens lease-​up time, implying a lower R*.
8 Added by the editor.
9 Dynamic intrametropolitan modeling of office markets is still in its infancy.
10 Added by the editor. If the dependent variable is submarket absorption share during period t then vacant stock
should refer to period t–​1 and not period t, since vacancy at (the end of) period t is determined by absorption
during period t and not the reverse. Inclusion of vacant stock at time t would, therefore, introduce simultaneity
bias. However, if a submarket has considerably more available space than other submarkets at the end of period t–​1
it could end up getting a greater share of metro absorption during period t. In that sense there is a rationale for
including in the formulation the vacant stock of the previous period.
11 Added by the editor.

References and additional readings

Building Owners and Managers Association International. 1999. 1999 BOMA Experience Report. New York: Building
Owners and Managers Association International.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Office Market Analysis. Chap.  11 in Real Estate Market
Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W. Wheaton. 1996a. The Operation of Nonresidential Property Markets. Chap. 11 in Real Estate
Markets and Urban Economics. Englewood Cliffs, NJ: Prentice Hall.
DiPasquale, D. and W. Wheaton. 1996b. Econometric Analysis of Metropolitan Office and Industrial Markets. Chap. 12
in Urban Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
Koepke, D. 2019. Improving Market Analysis in Commercial Real Estate Appraisal Assignments. Appraisal Journal,
87:1,  11–​26.
McMahan, J. 1989. Office and Industrial. Chap.  8 in Property Development. New  York:  McGraw-​Hill Publishing
Company.
Rosen, K. 1984. Toward a Model of the Office Building Sector. AREUEA Journal, 12:3, 261–​269.
Shilling, J. D., C. F. Sirmans, and J. B. Corgel. 1987. Price Adjustment Process for Rental Office Space. Journal of Urban
Economics, 22:1, 90–​100.
Sivitanides, P. 1997. The Rent Adjustment Process and the Structural Vacancy Rate in the Commercial Real Estate
Market. Journal of Real Estate Research, 13:2, 195–​209.
Sivitanidou, R. 2002. Office Rent Processes:  The Case of U.S. Metropolitan Markets. Real Estate Economics, 30:2,
317–​344.
Wheaton, W. 1987. The Cyclic Behavior of the National Office Market. American Real Estate and Urban Economics
Association Journal, 15:4, 281–​299.
Wheaton, W. and R. Torto. 1994. Office Rent Indices and their Behavior through Time. Journal of Urban Economics,
35:2, 121–​139.

358
newgenprepdf

16
MICRO ANALYSIS OF OFFICE
MARKETS

Introduction 359
Analyzing office locations 360
Productive or firm amenities 360
Worker/residential amenities 362
Prestige and visibility 362
Location constraints 362
But, do office rents fully reflect location amenity differentials? 362
Analyzing office projects: estimating achievable rents and absorption schedules 364
Collection of data for competing office properties 365
Identify comparable office properties 365
Collect information on the identified comparables 366
Data analysis 368
Using the competitive differentials technique to assess project strength and
achievable rental rates 368
Critique 370
Using hedonic methodologies to estimate achievable office lease rates 370
Variables included in hedonic office rent models 370
Usefulness of hedonic rent analysis 372
Critique 372
Developing office-project absorption and revenue schedules 373
Chapter summary 376
Questions 377
References and additional readings 378

Introduction
Once the macroeconomic analysis of the office market under consideration is completed and a preliminary
development scenario is formed, microeconomic analysis needs to be conducted to help refine the design
of the proposed office project, assess its competitive position, and evaluate its revenue-​generating ability.

359
 Office market analysis

As is the case for other property types, the major components of microeconomic analysis of office markets
are: 1) site and location analysis, and 2) project marketability analysis.
The underlying premise of office location analysis is that location is a major determinant of an office
project’s attractiveness and income-​generating ability and, therefore, of its success. Matching the location
and space needs of potential office tenants to site and project characteristics always enhances an office
project’s competitive strength. Marketability analysis is necessary in order to assess the project’s expected
rental revenue stream and absorption schedule.

Analyzing office locations

In analyzing office locations, it is important to bear in mind that the strength and nature of location
influences exerted by site-​specific factors may vary across the different types of office firms. Although, in
general, office firms are not thought to be highly heterogeneous in terms of their location requirements,
they may be differentiated in terms of:

• General type—​general purpose office firms vs. medical offices or law offices
• Type of operations performed—​front or back office functions
• Type of labor used—​executive vs. clerical etc.
• Firm size.

Below we briefly discuss the factors that are relevant in office location analysis and indicate whether
some types of firms may have stronger or weaker preferences for certain location attributes. These factors
are presented in Box 16.1 and include both location amenities and location constraints. Location amen-
ities include productive amenities, worker/​non-​productive amenities, prestige, and visibility, while location
constraints refer primarily to institutional constraints on commercial development.

Productive or firm amenities

Productive or firm amenities refer to locational attributes the can help a firm increase its productivity
and/​or reduce its costs. Such locational attributes include close-​by access to basic business support ser-
vices, broader access to business services agglomerations, access to white-​collar labor, access to airports and
freeways, and jurisdictional property tax rates.
Close-​by access to basic business support activities, such as banking, restaurants, and shopping, provides
for a more conducive working environment that can reduce the time cost of reaching basic services typic-
ally sought by employees during working hours.
Broader access to business services agglomerations matters because it may reduce the cost of interper-
sonal contacts and, possibly, labor search costs. The more specialized the office-​using activity is, the greater
the need for central control functions and access to business services agglomerations. Such a location
attribute appears to have little relevance for back office functions. Mun and Hutchinson (1995) present
evidence from the Toronto office market that agglomeration economies, measured as the change in prod-
uctivity caused by the increase in the number of firms in an office district, explain well spatial variations in
office rents. Their findings also suggest that agglomeration economies in the office sector are much larger
than those in the manufacturing sector. However, Sivitanidou (1997) also presents evidence pointing to the
importance of access to business service centers in office location decisions but also showing that its import-
ance may have been weakening, potentially due to advances in information and telecommunications tech-
nologies. A more recent study of the location decision-​making process of office firm managers and owners
in Toronto found that agglomeration played only a minor role in such a process (Elgar and Miller, 2010).

360
 Micro analysis of office markets

Box 16.1 What location attributes may matter?

Location amenities
Office firm/productive amenities (all contribute towards lower business costs)

• Close-​by access to basic business support activities

Close-​by access to facilities that can provide basic services typically sought by employees during working
hours, such as banking, restaurants, and shopping can help reduce the time costs of accommodating
such needs.

• Access to business service agglomerations

Traditionally thought to reduce the cost of interpersonal contacts and, possibly, labor search costs, but the
effect of this factor is weakening due to advances in information/​telecommunication technologies.

• Access to white-​collar labor

Thought to contribute towards lower labor costs; central locations may be sought if specialized labor is
dispersed, while suburban locations may be sought if labor skills are clustered.

• Airport and freeway access

Facilitates business trips; airport access becomes increasingly important.

• Property tax rates

Property tax rates matter only when the spatial demand for office sites is inelastic because they are passed
on to office lease rates.

Office worker/​residential amenities (contribute towards lower labor costs)

• Access to amenable residential communities with:

• High-​quality educational system
• Low crime rates
• High levels of public services for given levels of property taxes
• Access to shopping, recreational amenities (entertainment, ocean).

Prestige and visibility

Location constraints (increase business rents, if binding)

• Development constraints: zoning and density controls, growth moratoria.

Access to white-​collar labor can help office firms decrease labor costs, as it may reduce their search costs
and motivate workers living close by to accept lower wages in exchange for the lower commuting costs
they will enjoy. Hou (2016), in a study of firm location choices within the Los Angeles metropolitan area,
provides evidence indicating that firms in office-​related activities favor centers with a larger percentage
of highly educated labor. Central locations, such as downtown, may be sought if labor skills are dispersed
within the urban area or if the firm is using a large and highly heterogeneous labor force. Suburban
locations may be sought if labor skills are clustered.

361
 Office market analysis

Airport and freeway access is sought by office firms since it facilitates business trips. Airport access is
becoming increasingly important, especially for corporate divisions, due to the need for more frequent
business trips. Sivitanidou (1996) presents evidence that strongly supports the proposition that prox-
imity to airports matters in office location decisions. Proximity to freeways and freeway junctions may
be also valued by firms as it provides access to other firms, labor, and potential customers over a larger
geographic area.
Property tax rates matter only when spatial demand for office sites is inelastic. All else equal, higher
property tax rates should discourage development, thereby inducing an upward shift in the office property
supply curve. Such an upward shift will result in significantly higher rents only if office space demand is
inelastic, in which case the property tax rate differentials will most likely be passed on to the office tenant.
In this sense, jurisdictional property tax rate differentials may affect firm production costs through their
potential effect on rental rates.

Worker/​residential amenities
Worker/​residential amenities, such as access to amenable residential areas and communities with high-​
quality education, low crime rates, high quality of public services for a given level of taxes, and convenient
access to shopping and recreational destinations, should matter in office location analysis because they may
indirectly help reduce a firm’s costs. In particular, good access to amenable residential areas can facilitate
recruitment of skilled labor and, thereby, reduce labor search costs. Furthermore, there is considerable
empirical evidence by skill that happier workers are willing to accept lower wages. Such residential amen-
ities become more important the higher the proportion of a firm’s managerial labor.

Prestige and visibility

Prestige and visibility are also important from the demand side, especially for high-​profile firms and com-
pany headquarters. Usually, downtown locations or prestigious suburban nodes are the preferred locations
for these types of firms.

Location constraints
Location constraints refer to institutional constraints imposed by local governments on commercial devel-
opment, such as zoning, density controls, and growth moratoria. If such constraints are binding, that is, if
they restrict the supply of office space to levels below the ones dictated by market demand, they can lead to
supply shortages and, thereby, to higher office space rents. Understanding the degree to which the project’s
node or submarket is constrained can provide useful insights as to whether the site’s locational advantages
are fully capitalized into office rent differentials compared to office buildings located in other competing
nodes. We elaborate on this issue in the next section.

But, do office rents fully reflect location amenity differentials?

The location component of office rents is the result of the interplay between location amenities and
location constraints. An issue that is at the heart of office location analysis is the extent to which office
rents capitalize location-​amenity differentials within contemporary metropolitan office markets (see Box
16.2). This issue is very important because the conventional view is that amenity differentials across office
locations are fully capitalized on office rents. This view is the underlying premise of the competitive
differentials technique often used to derive achievable rents for a new office building at a given location.
However, theoretical analysis presented by Sivitanidou and Wheaton (1992) suggests that location-​amenity

362
 Micro analysis of office markets

differentials may be fully reflected in office rent differentials only if the land market is constrained. In
unconstrained markets, most of the location-​amenity differentials should be reflected in wages.
To understand the argument of rent and wage capitalization, consider for the sake of simplicity a city with
two business centers or office submarkets, which are the same in terms of size, office space rents, and worker
wages. Now let’s introduce an exogenous (not related to rents or wages) productive advantage in Center 1
of, let’s say, $1,000 per worker per year. This simply means that it is cheaper to do business at Center 1 by
that amount. As the city grows, new firms will seek space in Center 1, because of its productive advantages.
As a result, Center 1 will grow bigger in terms of both employment and commercial floor space. As Center
1 grows in size, the average commuting cost of its workers increases too, necessitating higher wages. Center
1 will continue to expand until its wages and rents increase enough to erode its productive advantage. Thus,
at the new equilibrium, the rent and wage differentials between Center 1 and Center 2 should be such that
they compensate for the exogenous production cost advantage of $1,000 per worker, as in (16.1):

a (R1–​R2) + (W1–​W2) = $1,000/​worker; R1 > R2 and W1 > W2 (16.1)

where
a : square feet per worker
R1, R2 : office space rent per square foot in Center 1 and Center 2, respectively
W1,W2 : wage per worker in Center 1 and Center 2, respectively

The question that arises is what percentage of the production cost differential will be reflected in higher
rents and what percentage in higher wages. Should the analyst care? The answer is yes, because if rents
do not fully or nearly fully capitalize production cost differences, the competitive differentials technique
may not be appropriate for estimating the rent of the subject property when comparables from different
submarkets are included in the analysis. Notice, however, that the validity of this technique is questioned
only as far as locational amenities are concerned.1

Box 16.2  But, do office rents reflect locational amenities2 (i.e., production or
tenant cost differentials)?
The location component of office rents is the result of the interplay between location amenities (production cost
advantages) and location constraints (zoning, density controls, development moratoria).The question is whether
office rents fully reflect location amenities under alternative land market conditions.

• Constrained (regulated) markets

Only in constrained land markets may office rents almost fully capitalize the value of location productivity
differentials.

• Unconstrained (non-​regulated) markets

In competitive markets, it is wages, not rents that capitalize the major portion of location productivity
differentials.

Implications for micro-​analysis

The competitive differentials technique may be flawed when applied to competitive office markets for it assumes
that differences in office rents across locations must fully reflect amenity differentials.

363
 Office market analysis

As Equation (16.1) suggests, only if the wage differential between the two employment nodes is zero
(W1 –​W2=0) would rent differentials fully capitalize the $1,000/​worker productivity advantage of Center
1. Theoretical analysis shows that the more constrained (regulated) a market is, the greater the percentage of
the productive advantage that is capitalized on rents as opposed to wages.The reason is that the more Center
1 is constrained to expand due to development controls, the smaller its wage differential from Center
2. On the contrary, if the market is not regulated in a way that restrains office space supply and the growth
of Center 1, office rent differentials between the two centers will be very small and most of the productive
advantage will be capitalized on wages (for a more elaborate presentation of this analysis see Sivitanidou and
Wheaton [1992]).
These findings have not yet been established empirically, but if valid, they do have some important
practical implications regarding the use of the competitive differentials technique for the estimation of
the rent of a planned office project. More specifically, the findings suggest that such applications may
be seriously inaccurate when the area’s land market is not constrained and the sample of competing
properties includes office buildings located in different submarkets/​nodes with considerably different
location attributes. In such a case, the analyst needs to examine carefully whether variations in specific
location attributes translate consistently to proportionate variations in lease rates across properties. This is
a very difficult task given the multitude of attributes that may differ across comparables and underscores
the need for hedonic regression techniques in assessing the true effect, if any, of location-​amenity
differentials on office space rents.
In the case of highly regulated land markets, where office space supply is constrained, the competitive
differentials technique may better measure relative office rent and price differentials across competing
locations, although hedonic techniques would still more accurately capture the effect of the different loca-
tion attributes on rents and prices.

Analyzing office projects: estimating achievable rents and absorption schedules

Once location analysis is completed and the development scenario has been refined to match the space
needs of the tenants most likely to be attracted at a given site, the analyst needs to proceed with the evalu-
ation of the competition. This analysis will help assess the prospects for further enhancing the project’s
competitive strength and derive the bottom-​line figures required for assessing its economic feasibility.
These figures include:

• Office rent schedule

• Absorption schedule
• Revenue schedule.

Assessment of the project’s achievable rental rates requires analysis of the competition and quantification
of how the different office structure and location attributes translate into rental rate differentials within the
local marketplace. Such an assessment will help the analyst not only estimate the most likely achievable rent
for the planned office structure at the specific location, but also evaluate whether modifying certain struc-
ture or project attributes will help maximize achievable rates. Furthermore, the estimation of achievable
rates can help determine the optimal office development density for the specific site considered.
It is important to emphasize that when analyzing office rents the unit of analysis is not the structure
per se, but lease transactions. Thus, for representative samples of lease transactions, the data that need to be
collected include not only the attributes of the space, building, and location associated with the lease
contracts, but also a whole array of lease characteristics that may have an effect on the effective rent paid
by the tenant.

364
 Micro analysis of office markets

Assessment of the project’s absorption schedule, in combination with the estimates of achievable rental
rates, can help estimate the property’s expected revenue schedule and, therefore, assess its feasibility from the
market perspective. Furthermore, evaluation of the project’s absorption schedule under alternative entry
scenarios can help make prudent decisions regarding the most appropriate/​profitable time of market entry,
as well as project phasing for projects involving more than one building.
In estimating the expected revenue for a planned office development, the analyst needs to make sure
that reasonable lease expiration and rollover schedules, as well as rent escalation provisions, are assumed and
properly accounted in the calculation. Special caution needs to be exercised when assessing the absorption
schedule for space expected to be vacated by expiring leases in the future.
Given the above discussion, the estimation of an office project’s most likely rental rate and absorption
schedule involves the following broader steps:

1. Data collection
• Identification of comparable properties.
• Collection of information on lease transactions, lease characteristics, and other attributes associated
with the comparable properties.
2. Data analysis
• Application of simple accounting techniques, and specifically the competitive differentials
technique.
• Econometric techniques and, specifically, hedonic methodologies.

These steps are discussed in detail in the following sections.

Collection of data for competing office properties

Identify comparable office properties
The first step in the collection of data for competitive analysis is the identification of comparable office
properties through brokerage firms, multiple listing services, and/​or windshield surveys. The focus of the
data collection effort is somewhat different depending on whether the competitive differentials or the
econometric technique is used:

• When applying the competitive differentials technique, the analysis typically focuses on compar-
able office properties (existing, under construction, and planned) within the submarket in which
the subject property is located in the case of larger markets. Less detailed surveys of other com-
peting office space clusters in the metropolitan office market are also typical in evaluating the
competition.
• When applying hedonic analysis techniques, it is not necessary to restrict the sample of office properties
only within the submarket of the subject site; actually, it is preferable that the sample includes properties
with significant differences in location and structure attributes. Thus, the sample may be expanded to
include properties that may not be strictly competitive with the project under consideration.

The analyst may be better off by applying both techniques, if time, budget, and data availability allow
it. Although the econometric technique is likely to produce more accurate estimates of the project’s
achievable rental rate, estimates using the competitive differentials technique can provide an alterna-
tive estimate, against which the analyst can evaluate the reasonableness of the result of the former
methodology.

365
 Office market analysis

Collect information on the identified comparables

Once the comparable office properties and their locations have been identified, the analyst needs to
compile information on lease rates, lease characteristics, property, and location attributes from local rental
agents, brokers, and appraisers, and/​or through field surveys. Box 16.3 presents a list (not necessarily
exhaustive) of the data that should be collected. As this Box indicates, data on four sets of factors should
be collected: lease terms and characteristics, space attributes, structure attributes, and location attributes.
Some comments are warranted here regarding the relevance of the specific data collected under these
four categories of factors.

LEASE TERMS AND CHARACTERISTICS

• In terms of lease rates, these are expressed in dollars per square foot per year. In trying to understand
the supply-​demand dynamics of office markets and how they affect market rates, the appropriate rent
measure is the “effective” rent, which is different from the asking rent and the contract or quoted
rent. The effective rent is the actual rent paid by the tenant after taking into account concessions,
such as free rent, excessive tenant improvement allowances, expense caps, and free parking, stipulated
by the lease agreement. In general, the larger the lease the more difficult it is to determine the actual
effective rent paid by the tenant (see Peiser and Schwanke, 1992). Thus, it is important that the ana-
lyst collects information not only on base-​year lease rates, but also the array of lease characteristics
discussed below.
• The date of lease transaction is needed in order to control for the effect of market conditions on lease rates
at the time the lease was signed.
• The base-​year or contract lease rate is the basic information, which, in combination with information
on other lease terms, can help develop a measure of the overall effective rent implied by the lease
agreement.
• The type of lease rate is very important in developing a consistent measure of effective market rent across
lease transactions since it clarifies whether the quoted contract rate is net of taxes (N), net of taxes and
utilities (NN), or net of taxes, utilities, and operating expenses (NNN).
• The length of the lease may influence contract rates through several venues. For example, if expectations
for improving market conditions are prevalent at the time the transaction takes place, then longer leases
may be associated with higher contract rates. However, independently of expectations about future
market conditions, a longer lease reduces the risk or volatility of the landlord’s cash flows. As a result,
the landlord may be willing to accept lower rates for longer-​term commitments by prospective tenants,
especially given the high turnover costs (tenant improvements) typically associated with new tenants
in office buildings. On the other hand, by accepting a long-​term lease, the landlord is forgoing the
ability to take advantage of potential market rent increases above those stipulated by any CPI escalation
clauses included in the contract. Thus, the strength of the market at the time the lease is signed may be
the deterministic factor of the relationship (positive or negative) between the length of the lease and
the contract rate. Data presented by DiPasquale and Wheaton (1996) suggest an average length lease of
5 years with 3-​year and 10-​year leases also having relatively high frequencies.
• Rent escalation clauses provide for a formula for adjusting rents on the basis of widely used economic
indicators, usually the CPI; in soft markets, rent escalation is less steep in that rents are allowed to adjust
every two years or in the middle of the lease term. In the presence of escalation clauses, contract rates
should be lower compared to leases with no escalation clauses.
• Stop clauses represent a ceiling amount for operating expenses (the dollar stop), usually determined on
the basis of estimated expenses during the first year of the lease. Any expenses above that amount are
paid by the tenant. Therefore, the stop amount should be positively associated with the base lease rate,
since the higher the amount the higher the burden on the landlord.

366
 Micro analysis of office markets

Box 16.3 Analyzing office rents/​lease rates: Analysis components

Data sources
• Brokerage firms
• Multiple listing services
• Windshield surveys.

Note: Data for strictly competing properties should be collected if the competitive differentials technique is
applied. There are no such restrictions if the hedonic methodology is used, as this technique can control for a
wide range of property and location attributes.

Required micro data
• Lease rates and lease characteristics
• Date of the lease transaction
• Base year lease rates
• Lease terms
• Type of lease rate: Gross, N, NN, NNN
• Length of the lease
• Escalation clauses
• Stop and other clauses
• Concessions (free rent, etc.)
• Tenant improvements (TIs).

Space attributes
• Amount square feet leased
• Quality and type of space (flexible design or not)
• Location within structure (floor, corner office or not, etc.).

Structure attributes
• Quality of lobbies and elevators
• Structure amenities (such as restaurants, gyms, etc.)
• Parking availability
• Building access (ease of access to the building).
Location attributes (see Box 16.1)

Note: See Brennan, T., R.  Cannaday, and P.  Colwell. (1984). Office Rent in the Chicago CBD.
AREUEA Journal, 12, 243–​260.

• Concessions, including primarily free rent, constitute another important piece of information in
developing an accurate and consistent measure of effective market rent through time. It appears that
when the market is soft, landlords prefer to provide rent concessions and maintain a higher contract rate
rather than directly lower the contract rates. As a result, concessions are more prevalent during periods
of excess supply and high vacancy rates.
• Tenant improvements, representing tenant finish allowances for the interior space (including ceilings,
walls, flooring, and telephone and electrical outlets), and commonly referred to as TIs, are in substance

367
 Office market analysis

an additional form of concessions. Landlords always provide some form of tenant improvement
allowances, but it has been observed that during periods of serious excess supply such allowances are
considerably higher and are presumably provided as an additional incentive to lure tenants. Again,
information on tenant improvements can help further refine the effective rent calculation across
leases and through time.

SPACE ATTRIBUTES  The second set of factors that needs to be examined when analyzing lease rates in
comparable office buildings includes the characteristics of the specific space the lease contract refers to. First
of all, the amount of square footage committed in the lease transaction may be negatively associated with office
rents, as landlords may be willing to offer volume discounts when leasing larger blocks of space. Other space
attributes, such as the floor on which the leased space is located or its position within the structure (corner
offices), may also contribute to variations in lease rates.

STRUCTURE ATTRIBUTES  The attributes of the structure within which the space is located, such as
construction quality, quality of the lobbies and elevators, structure amenities (restaurants, shops, gyms,
etc.), parking availability, ease of access to the building, etc., represent another important set of factors
that need to be taken into account when examining differences in lease rates across comparable
properties.

LOCATION ATTRIBUTES  The final set of data that needs to be collected includes the location attributes
associated with the comparable properties for which lease transaction data are available. These have been
discussed in detail in the previous section.

Data analysis
Once the information has been collected and systematically tabulated, the analyst can apply the competitive
differentials technique or hedonic methodologies in order to evaluate the income-​earning ability of the pro-
ject and assess the possibilities for enhancing its competitive position. In particular, application of these
techniques will help the analyst to:

• Quantify the competitive strength of the project by estimating its competitive position index (CPI).
• Identify the sets of attributes that contribute the most to project value, thereby setting the stage for
refining project design and maximizing effective lease rates.
• Estimate achievable lease rates given project and location attributes.

Using the competitive differentials technique to assess project strength and

achievable rental rates
The premise of the competitive differentials technique, as it applies to the office market, is that differences
in office lease rates across competing office properties reflect differences in lease characteristics, structure/​
project attributes, and location amenities. So the basic difference compared to the application of this tech-
nique that was reviewed in the case of the residential market is the addition of lease characteristics in the set
of attributes examined for both the subject and the comparable projects. Such characteristics may include
the base rent, the escalation clause (as percentage of CPI), the stop clause (as a dollar amount), the duration
of the lease, etc. Also note that location characteristics need to account for location attributes that matter to
office-​using companies. For example, location access attributes may include driving time to the center of
the city, driving time to residential areas suitable for clerical labor, driving time to high-​income residential

368
 Micro analysis of office markets

concentrations (managerial labor), distance from a major transportation artery, etc. Similarly, neighbor-
hood characteristics may include metrics capturing availability and cost of parking, concentration of other
businesses, and business support services, etc.
In order to review the application of this technique in the office market, we will draw from Clapp’s
(1987) example of the competitive spreadsheet used for such an analysis. In reviewing this example, we
will elaborate on how it can be used to estimate project rent and absorption rate. Notice that a base
rent figure for the subject property is included only for illustration purposes. When dealing with a new
project, the base rent typically is not known and this analysis is used as a means for estimating it. Thus,
in the analysis steps described below it is assumed that the subject’s base rent is not known (for a more
detailed discussion of the calculations involved in applying the competitive differentials technique, see
Chapter 9).

1. Construct a competitive spreadsheet where the columns represent competing office properties and the
rows represent important lease, structure, and location attributes that may influence a property’s
lease rate.
2. Assign weights to each attribute based on its contribution to project value or rental rate, using experi-
ence and judgment. Interviewing local brokers and firms occupying competitive office space regarding
the importance of each attribute can help more accurately determine these weights.
3. Convert the values assigned to each of the attributes of the subject property and the competitive prop-
erties to scores/​indices using the formulas presented in Chapter 9.
4 . Estimate total unweighted and/​or weighted amenity indices for each of the comparable properties.
5. Estimate adjusted unweighted (weighted) amenity indices, as a property’s total unweighted (weighted)
amenity index minus the unweighted (weighted) rent score/​index, respectively. For example, the
adjusted weighted total amenity index for comparable 1, AWTAI1, in Clapp’s spreadsheet is calculated
as in (16.2), where WAI1 is the weighted amenity index for comparable 1.

AWTAI1= WAI1 –​Weighted rent score = 5,798 –​ 400 = 5398 (16.2)

6. Estimate the base rent, RS, for the subject property relative to each comparable using the subject’s total
amenity index, TAIS, and each comparable’s adjusted total amenity index, ATAIi, as follows:

TAI S
Rs = Ri (16.3)
ATAI i

where
TAIS : (weighted or unweighted) total amenity index for subject
ATAIi : adjusted (weighted or unweighted) total amenity index for comparable I
Ri : base rent for comparable i

Note that the subject property’s total amenity index (TAIS) does not need to be adjusted since the
rent of a planned project is not known and, therefore, normally a score for this variable will not be
available to the analyst to be added to the total amenity score. In case that the subject is an existing
property, a rent figure will most likely be available. If the analyst wants to simply evaluate whether
this rent figure is consistent with the property’s competitive position, then formula (16.3) provides a
means for making such an evaluation. In such a case, the subject property’s total amenity index may
need to be adjusted too, depending on whether the rent score was taken into account when it was
calculated.

369
 Office market analysis

Applying formula (16.3) using the numbers presented in Clapp’s example for comparable 1’s rent as
basis, the adjusted weighted amenity index for comparable 1, and the subject property’s weighted amenity
index, the subject property’s office rent, RS, can be estimated as:

RS = (4,146/​5,398) $18 = 0.76806 $18 = $13.82

Obviously, using formula (16.3) the analyst can derive as many alternative base-​rent estimates for the
subject property as the comparables for which rent information is available. These rent estimates will, in
all probability, be different. The analyst can use the average of these alternative estimates as the project’s
base rent if a single-​point estimate is sought. In computing this average, the analyst may want to use larger
weights for rent estimates associated with properties considered to better proxy the subject’s performance.

Critique
The results of applications of the competitive differentials technique for the estimation of achievable office
project rents should be viewed with skepticism. First, the methodology assumes that differences in loca-
tion attributes across office properties are fully reflected in their rent differentials. As pointed out earlier,
theoretical analysis suggests that this assumption may not be valid if the land market is not constrained in a
binding way. Second and most importantly, the technique entails serious difficulties in terms of objectively
and accurately assigning weights to the different lease, structure, and location attributes that affect office
lease rates in a way that is consistent both across attributes and across properties. Moreover, the validity of
the assigned weights cannot be confirmed without using econometric techniques. In sum, competitive
differentials techniques should be used with caution in the estimation of an office property’s achievable
rents, but can provide an alternative estimate that can be used to test the reasonableness of estimates derived
through hedonic valuation techniques.

Using hedonic methodologies to estimate achievable office lease rates

Hedonic valuation techniques should in theory provide more accurate estimates of the rent an office pro-
ject will command given its structure and location attributes. Actually, application of hedonic methodolo-
gies using a sample of lease transactions can allow the analyst to estimate the rental rate per square foot that
a particular office space on a particular floor of the building under consideration will command, provided
that the amount of the space leased and the floor on which it is located are among the independent
variables of the estimated regression model. Since this technique has been reviewed rather extensively in
Chapter 9, this discussion will focus only on:

• The set of critical variables that need to be incorporated in the case of hedonic office rent formulations.
• The usefulness of the technique in office project assessment.

Variables included in hedonic office rent models

In estimating hedonic office rent models, non-​linear functional forms should be used for the reasons
discussed in the residential section. Such non-​linear hedonic lease rate models need to account for the
whole range of attributes that may affect lease rates in a given office structure. As discussed in the competi-
tive differentials technique these include:

• Lease terms
• Space and structure attributes
• Location attributes.

370
 Micro analysis of office markets

Table 16.1  Office rent hedonic analysis: The data

ID RENT VAC SF SF1 FLOOR PARK TAX AUTO ACCESS STOP BASE YEAR
1 14.11 3 5,000 35,000 4 5 2.76 32 7.5 0.00 1 84
2 12.70 9 1,000 21,000 1 2 2.76 30 7.5 2.50 0 65
3 11.46 11 3,500 15,000 2 9 2.76 34 7.5 3.25 0 73
4 14.44 2 20,400 49,500 6 21 2.76 28 7.5 0.00 1 68
5 15.01 0 7,800 27,000 3 19 2.76 31 7.6 2.10 0 79
6 12.29 18 3,000 10,500 2 9 2.35 26 8.1 4.50 0 54
7 8.18 4 4,000 13,000 1 15 1.90 23 10.0 0.00 1 64
8 11.65 20 2,500 8,200 1 11 1.90 23 10.0 5.50 0 71
9 12.14 0 1,000 6,000 1 4 1.90 21 10.8 2.00 0 65
10 13.05 8 4,000 11,000 2 20 1.90 25 10.8 3.75 0 68
11 9.44 5 1,000 13,000 2 4 1.90 26 10.5 0.00 1 76
12 8.01 3 2,500 9,000 1 11 1.90 22 10.5 0.00 1 82
13 12.20 13 6,000 22,000 3 8 2.23 28 4.5 3.00 0 80
14 13.40 0 1,000 28,000 4 2 2.23 33 4.6 0.00 1 77
15 11.49 0 4,500 18,000 2 12 2.23 31 4.8 0.00 1 75
16 9.76 9 5,000 15,000 2 16 2.23 30 4.8 2.20 0 63
17 11.20 3 3,600 12,000 1 10 2.40 29 5.0 0.00 1 71
18 11.02 6 2,500 14,000 2 9 2.40 29 5.0 0.00 1 78
19 11.45 9 1,000 8,000 1 4 2.60 32 7.8 3.10 0 68
20 11.42 12 3,000 12,000 2 15 2.60 30 8.0 4.25 0 72
21 10.85 10 1,000 14,000 1 5 2.60 28 8.1 3.60 0 83

Source: From Clapp, J. 1987. Handbook for Real Estate Market Analysis. Englewood Cliffs, NJ: Prentice-​Hall
Variable explanations
Lease terms and characteristics
RENT: Base rental rate per year per square foot of floor space
STOP: Stop clause for common expenses
BASE: Base year escalation clause
FLOOR: Floor in building for tenant
Building/​unit characteristics
VAC: Building vacancy rate
SF: Net rentable square footage occupied by tenant
SF1: Net rentable square footage in building
TAX: Effective tax rate
AGE: Age of the building (87 –​Year built)
PARK: Number of parking spaces assigned to tenant
Location characteristics
AUTO: Average auto time to the building
ACCESS: Average distance to the CBD

Lease Rate = f (Xlease terms, Xstructure traits, Xlocation traits) (16.4)

Table 16.1, reprinted from Clapp (1987), provides an example of what the data used for estimating hedonic
office rent equations may look like. Potentially important variables omitted from Table 16.1 include the
date on which each lease was signed (if the sample includes leases signed in different years), other pro-
ductive amenities besides access, worker amenities, other lease terms, and institutional supply restrictions
(if they vary across the locations with which the leases are associated with). One of the variables included

371
 Office market analysis

in this table is the vacancy rate, but its inclusion in the hedonic rent formulation is questionable from an
econometric point of view.3
The important difference between hedonic office models and the residential ones is the inclusion of
lease terms, as well as some location attributes that may not be as relevant in residential rent analysis, such
as access to other firms, access to labor, airports and, especially, jurisdictional and other institutional supply
constraints on commercial development at the different nodes. As indicated earlier, the latter may be very
relevant in terms of correctly quantifying the capitalization of amenity differentials across office locations
on office lease rates.
If the lease transactions included in the sample are associated with properties located in different office
nodes/​submarkets within the metropolitan area, it is typical to add in the specification dummy variables
indicating the node/​submarket within which each property is located. Archer and Smith (2003), in an
analysis of quoted office rental rates in different office clusters in Houston, found that cluster size had a
positive effect on quoted rental rates.The submarket-​specific dummies will capture such an effect as well as
any additional effects on lease rates due to differences in other cluster characteristics that affect office rents.
So, for example, if available lease-​transaction data involve properties that are dispersed within n submarkets,
then (16.4) should be expanded to include n-​1 submarket dummies.

Usefulness of hedonic rent analysis

Hedonic office-​rent models can help directly or indirectly in developing estimates of the following:

• Achievable lease rates for the planned office space that can provide the basis for developing asking-​rent
strategies and project rental revenue forecasts.
• Marginal values for several structure and project attributes that can provide the basis for selecting the
rent-​maximizing set of features and optimizing office project design.
• Residual office land value estimates (RLV) using (16.5) and (16.6) below, where P denotes office prop-
erty price and C non-​land office development costs.

RLV = (P–​C) FAR (16.5)

Following the simple income capitalization approach to value, the price P of the office property
under consideration can be calculated as:

P = NOI/​CAP RATE (16.6)

NOI is the Net Operating Income of the property (which is a function of lease rates, lease rollover
schedules, and operating expenses) and CAP RATE is the market capitalization rate.
• Optimal office project development density, that is, the density most likely to maximize residual land
value (see Chapter 9 for an elaborate discussion of how the results of hedonic analysis can be used to
calculate optimal development densities).

Critique
Contrary to the competitive differentials technique, the impact (weight) of the various office project
and location attributes, as well as lease characteristics, on office space rents is not assumed in the hedonic
approach, but derived in an objective and reliable way through rigorous statistical analysis of real market
data. The complex, multi-​year, and multi-​dimensional nature of non-​residential lease contracts, in gen-
eral, and office leases, in particular, makes the application of this technique even more necessary. Notice

372
 Micro analysis of office markets

that hedonic valuation techniques allow for an objective estimation of the effect of location and structure
attributes on office rental rates, while controlling for variations in a multitude of lease terms and vice versa.
The major limitation of this technique is that it requires the collection of data on a considerably greater
number of office comps and lease transactions than the one required for the simple competitive differentials
technique. The scarcity of systematic, comprehensive, and consistent data on actual office lease transactions
renders the application of this approach more difficult.

Developing office-​project absorption and revenue schedules

Office project absorption depends on the macro conditions prevailing in the metropolitan area and
submarket within which the property is located and its competitive strength (determined primarily by the
structure and location amenity levels). Within this framework, project absorption analysis brings together
the findings from the macro-​and micro-​analysis stages discussed so far.
Office space absorption for a specific project can be calculated using the same general formula utilized
in the case of residential project analysis and described by (16.7):

Project Capture = (NRAS /​ NRAM) * CPIS * ABM (16.7)

where
NRAS : the subject’s net rentable area (NRA) in square feet that is available for leasing
NRAM : total net rentable office square footage in the market expected to compete for tenants with
the project, including the NRA of the project
CPIS : the subject’s competitive position index calculated as the ratio of the subject’s weighted (unweighted)
total amenity index over the average of the weighted (unweighted) total amenity indices for all prop-
erties considered in the analysis
ABM : anticipated market absorption

In applying formula (16.7) for the estimation of the absorption of a planned office project, the analyst is
cautioned to pay attention to the following points:

• The figures used for all four terms in (16.7) should not refer to the period during which the market
study is carried out, but to the period for which absorption analysis is performed.
• The subject’s net rentable area available for leasing, NRAS, will certainly not be equal to the total NRA
of the planned project during the year of market entry, if some office space has been pre-​leased.
• The estimate of the project’s competitive position index, CPIS, for the anticipated year of completion needs
to account for all competing office projects that are in the planning stage or under construction and are expected
to be completed during the same year as the subject. CPIS should be declining during the subsequent years, unless
no new competing office space is expected to enter the market after the planned project is completed.
• The analyst needs to make sure that NRAM, CPIS, and ABM are consistent in terms of the geographic area
of reference and office building type (see elaborate discussion of this issue in Chapter 9). For example, if
CPIS represents the project’s competitive position relative to competing class A  office properties in
the project’s submarket, then NRAM and ABM must also represent total available square footage and
expected net absorption of class A office space in the project’s submarket, respectively. Using available
square footage and expected net absorption of class A and B office space in the metro area instead could
lead to inaccurate estimates of the project’s absorption.
• Myers and Mitchell (1993) point out that in estimating a project’s capture rate using the fair market
share approach a key decision is what constitutes a competing property, and emphasize the importance
of calculating a segmented project capture rate that is coordinated with segmented demand parameters
(in calculating ABM and of course competing supply).

373
 Office market analysis

Table 16.2  Example of office project absorption and revenue schedule calculation

Year Market Competing Market first-​ Structure Project Project Project Gross rental Cumulative
absorption available year lease rate available competitive absorption lease rate revenue gross revenue
supply index space position index
[1]‌ [2]‌ [3]‌ [4]‌ [5]‌ [6]‌ [7]‌ [8]‌ [9]‌ [10]
1997 1,050,939 1,444,800 138.06 120,000 1 87,287 $31.01 $2,706,770 $2,706,770
1998 1,325,021 1,252,160 154.63 32,713 1 32,713 $34.73 $1,136,122 $3,897,028
1999 1,010,845 1,651,200 173.18 0 1 0 $38.90 $3,974,968
2000 1,016,312 1,135,200 176.65 0 1 0 $39.68 $4,054,468
2001 1,326,922 412,800 181.95 0 1 0 $40.87 $4,135,557

Note: Anticipated time of entry is year-​end 1996; the annual rent escalation is 2% and the length of all leases is
five years.
On inputs and calculations
[1]‌ Year.
[2]‌ Market (net) absorption; estimated through an absorption regression model utilizing as independent variables,
among other factors, office employment growth (provided by an econometric forecasting firm) and real rents
(see Chapter 15 for a discussion of the econometric approach).
[3]‌ Competing available supply of office space; also estimated through the econometric approach (see Chapter 14
for a detailed discussion of the calculation of available office supply).
[4]‌ Office lease rate index was also estimated through the econometric approach.
[5]‌ Project size was determined through design specifications.
[6]‌ The competitive strength of the project is assumed to be average.
[7]‌ Calculated as the minimum of (([5]/​[3])*[6]*[2]) and [5].
[8] The year-​end 1996 hedonic lease rate for the project was calculated through hedonic regression analysis using
a sufficient sample of recent office lease transactions within the market in which the project competes; the lease
rate for subsequent years was calculated by growing the estimated current lease rate for the structure under
consideration at the growth rate dictated by [4]; the numbers reported represent the annual rate per square foot.
[9]‌ Gross rental revenue refers here to the revenue only from new leases signed during the year under
consideration; estimated as [8]*[7].
[10] Cumulative revenue refers here to the revenue from both new and existing leases; estimated from [9]‌taking into
account lease length and CPI escalations stipulated by lease agreements.

Some practitioners derive project absorption rates by adjusting the absorption rates of comparable office
properties according the project’s competitive position relative to each comparable. In particular, such
estimates can be derived by multiplying the project’s relative amenity indices by the absorption rates of
the respective comparables (see discussion in Chapter 9 on how to estimate relative amenity indices). This
methodology is the same as the one used in applications of the competitive differentials technique for the
estimation of the project’s lease rate. As in the case of lease rates, absorption estimates through this meth-
odology are problematic because, by construction, they must refer strictly to the time of analysis and need
to be adjusted accordingly for changes in market conditions expected to take place between the time of the
study and actual project market entry, as well as over the project’s lease-​up period.4 Using formula (16.7) is
a better approach for estimating office project absorption, because it does take into account the demand-​
supply conditions expected to prevail at the time of project entry.5
Table  16.2 presents an example of office project absorption and revenue schedule calculations. The
notes in the table explain clearly the different entries and the calculations and estimation procedures
involved. Notice that in this example, the methodology relates project absorption to total metro absorption.
However, as mentioned earlier, often project absorption is calculated using submarket absorption estimates
as framework. It should be emphasized that the formulas for calculating project absorption and revenue

374
 Micro analysis of office markets

schedules are the same independently of whether “market absorption” and “competing available supply”
refer to the submarket or the metro area. To help the reader better understand the figures presented in this
table we elaborate on the calculations for 1998.

PROJECT ABSORPTION (COLUMN [7]‌)  The rationale underlying this calculation is that the project’s share
of total market absorption (column [2]‌) will be equal to its fair market share adjusted for any competitive
advantages or disadvantages that may enhance or diminish its appeal. Following (16.7), project absorption
for 1998 is therefore calculated as:

Project absorption 1998 = Minimum of [(32,713/​1,252,160)*1*1,325,021]

and 32,713 = Minimum of [34,616 and 32,713] = 32,713

Since expected market absorption exceeds available supply in 1998, the fair market share of the pro-
ject would be greater than the space available in the building and, therefore, project absorption will be
equal to the available space. Now the reader may be wondering how it is possible for net office space
absorption to be greater than the market’s available supply. The reader is reminded that the so-​called
“structural” vacant stock, which is derived based on the assumption of a structural or normal vacancy
rate, is not included in the calculation of an office market’s available supply. So, obviously, what the
numbers for 1998 suggest is that some of that structural vacant stock was absorbed in order to accom-
modate excess demand.
Two points need to be made regarding the project’s competitive position index in Table 16.2. First, it
should be noted that an average competitive position index of 1 (as opposed to greater than 1) is not neces-
sarily unreasonable for a new-​build project. A new-​build project can reasonably be at an average competi-
tive position for a number of reasons. For example, the construction and design quality of the building, as
well as amenities provided by the project, may be inferior to those of other new projects or even existing
buildings. Furthermore, the project’s location may not be as attractive and advantageous as those of other
competing projects.
What may appear unrealistic in Table 16.2 is keeping the competitive position index at 1 throughout the
forecast. This may be unrealistic because new buildings entering the market after the project is completed
may be at a better competitive position than the subject.

HEDONIC LEASE RATE (COLUMN [8]‌)  The rationale underlying the estimate of the project’s lease rate in
1998 is that during that year it will increase (decrease) from its 1997 level at the same rate as the market
lease rate index (MLRI). Within this context, it is calculated below by applying the anticipated percentage
change in the MLRI in 1998 to the project’s 1997 lease rate (the note for column [8]‌in Table 16.2 explains
how the latter has been estimated).

Project lease rate 1998 = Project lease rate 1997 * (1 + % MLRI for 1998)

= 31.01 * (154.63/​138.06) = $34.73

GROSS RENTAL REVENUE (COLUMN [9]‌)  For the purposes of this example, gross rental revenue refers to
the rental income attributed only to new leases, signed during the year of the analysis. In order to keep the
calculations simple it is assumed that all new leases are signed on January 1 of the year under consideration.
Within this context, the gross rental revenue for 1998 will be:

Gross rental revenue 1998 = Square feet leased in 1998 * 1998 lease rate

= 32,713 * 34.73 = 1,136,122

375
 Office market analysis

CUMULATIVE GROSS REVENUE 1998 (COLUMN [10])  For the purposes of this example, cumulative gross
revenue refers to all the rental income earned by the property from both existing and new leases. As such, it
is the sum of the rental income from new leases signed in 1998 and the income from existing leases signed
in 1997. Notice that, according to the assumed lease specifications, the contract rate for the leases signed in
1997 will increase by 2% in 1998. Within this context, the 1998 cumulative gross revenue is calculated as:

Cumulative gross revenue in 1998 = (2,706,770 * 1.02) + 1,136,122 = 3,897,028

The calculation of the project’s revenue schedule is based on the assumption that all tenants in the hypo-
thetical office building under consideration have five-​year leases. What if the tenants that occupied the
building in 1997 had three-​year leases? In that case, we would first put the 87,287 square feet that were
absorbed in 1997 back into column [5]‌(labeled “structure available space”) in 2000. Then we would need
to apply the same formula used to calculate the entries in the project absorption column (column [7])
in order to estimate how much of that space would be absorbed given the anticipated demand supply
conditions for that year. Another approach is to assume a renewal percentage for expiring leases, let’s say
50%, and put back in the market under column [5] only 50% of the 87,287 square feet.

Chapter summary
The microeconomic analysis component of office market studies focuses on the assessment of the strengths
and weaknesses of a given site and location for office development, together with the evaluation of the
competitive position and revenue-​generating ability of the planned office project.
Office location analysis focuses on the site’s advantages and constraints. Locational attributes that are
relevant in office location analysis include all factors that may directly help firms increase their productivity
and reduce their costs. Such productive amenities include access to business services, access to white-​collar
labor, and access to airports and freeways, as well as factors that may appeal to their workers (worker/​
non-​productive amenities), such as low crime rates, a good public education system, high levels of public
services, access to shopping and recreational amenities, etc.
Locational constraints that are relevant in office location analysis include zoning, density, and growth
controls, as well as any other factor that may limit the site’s development potential or the overall supply of
office space in the jurisdiction or county within which the site is located. Analyzing constraints on office
development becomes especially relevant in light of theoretical findings suggesting that office location
advantages are fully or almost fully capitalized on office rents as opposed to office-​worker wages only in
the case of supply-​constrained markets.
Analysis of the income-​generating ability of a planned office development focuses on forecasts
of achievable lease rates, absorption schedule, and revenue schedule. Achievable lease rates for office
developments can be estimated through the competitive differentials technique or hedonic valuation
techniques. The important point to be kept in mind is that, ideally, office rent estimates must be based
on the analysis of actual lease transactions and not asking rents. Proper analysis of lease-​transaction
rates requires collection of data not only on structure and location attributes, but also on a host of
lease characteristics. Hedonic valuation techniques can better measure the effect of these multiple lease
characteristics on office rents, while controlling for differences across properties in structure and location
attributes, and vice versa.
Office project absorption can be estimated using the fair market share concept adjusted by the project’s
competitive position index. Project absorption and lease rate estimates provide the basis for generating the
project’s rental revenue schedule. In deriving this schedule, the analyst usually needs to apply CPI clauses
expected to be included in lease contracts and carefully account for potentially different lease rates for space
expected to be absorbed in different years.

376
 Micro analysis of office markets

Questions
1. Discuss what location attributes may play an important role in the intra-​
metropolitan location
decisions of office firms and why.
2. Discuss what the issue of location-​amenity capitalization in the office market has to do with office
project analysis.
3. List and briefly describe alternative techniques and the steps involved in estimating an office project’s
achievable rent and absorption schedule.
4. Describe the steps in estimating an office project’s achievable rent using the competitive differentials
technique.
5. Discuss in detail the differences between apartment and office hedonic rent formulations.
6. You are considering investing in a brand new office structure expected to be completed in two
years, that is, year-​end 1999, at a site located in one of the Los Angeles area’s most vital-​edge cities.
The market analyst has provided you with the following sets of data, which will help you assess the
property’s income earning potential.
Set A: Forecasts of movements in real base-​year lease rates during the period of 1997–​2004 under a
“base-​case” scenario. These are presented below in the form of inflation-​adjusted year-​end indices.

Year 1997 1998 1999 2000 2001 2002 2003 2004

Index 100 102 104 108 111 112 110 108

Set B: Hedonic results, based on data on 300 office lease transactions that took place in 1997 in ­several
of the greater Los Angeles’ office-​commercial nodes. A  preliminary hedonic analysis of these data
produced the results presented in the table below.

Dependent variable: Natural logarithm of the base-​year lease rate.

Coefficient t-statistic
Constant 3.54 12.20
Net (Dummy; =1 if rent is net; =0 otherwise) –​0.09 –​3.20
TripleNet (Dummy; =1 if rent is triple net; =0 otherwise) –​0.25 –​4.20
Stop (numeric; indicates “stop” dollar amount) 0.02933 5.20
CPI Escalation (numeric; indicates % escalation) –0.44 –​5.67
Ln(Lease Length) (numeric; denotes ln of lease-​term in years) –0.08 –​4.20
Ln(Size) (numeric; denotes ln of sq. ft. involved in transaction) –0.014 –​1.60
R-​Squared 0.74

a) Focusing on Set B’s hedonic valuation results, interpret/​explain the following:

• The R-​squared
• The sign of ln(Lease Length); explain whether this makes sense and why
• The coefficient of ln(Lease Length)
• The t-​statistic associated with ln(Lease Length).
b) Compute the year-​end 2004 base-​year triple-​net lease rate of a lease transaction involving 10,000
square feet of office space and a four-​year lease (e.g., Lease length=4 years), with a stop amount
of $3.75 (e.g., stop=3.75) and a 3% CPI escalation.
c) Explain in detail how and why you would refine the above model so that it yields better estimates.
Again, make sure you provide a clear rational.

377
 Office market analysis

d) Would, in principle, the application of the competitive differentials technique be preferable?

Explain clearly why or why not, based on the specifics of the case.
7. Describe in detail how you would go about estimating the absorption schedule for a planned office
building.
8 . Discuss how you could go about developing a five-​year forecast of the base-​year lease rate for a
planned office project using econometric techniques.

Notes
1 As the reader must know by now, application of the competitive differentials technique involves a host of other
attributes, such as structure and project quality and amenities.
2 See Brennan, T., R. Cannaday, and P. Colwell. 1984. Office Rent in the Chicago CBD. AREUEA Journal, 12, 243–​
260; Sivitanidou, R. and W. Wheaton. 1992. Wage and Rent Capitalization in the Commercial Real Estate Market.
Journal of Urban Economics, 31, 206–​229; Sivitanidou, R. 1996. Do Office Firms Value Access to Service Employment
Centers? A Hedonic Value Analysis within Polycentric Los Angeles. Journal of Urban Economics, 40, 1–​27.
3 The inclusion of the contemporaneous vacancy rate as an independent variable may be introducing a simultaneity
bias, in that building rents are not only affected by vacancy rates, but they also influence vacancy rates. For example,
an office building may have lower vacancy because it offers lower rates.
4 These estimates cannot refer to the time of project completion simply because market studies are carried out well
before project construction starts, let  alone completed. Thus, the only absorption rates that the analyst can use
are those of comparable office buildings that were completed within at least 12–​24 months prior to the time of
the study. Assuming a two-​year period from the time the market study is carried out and the time the project is
completed, and another two years until the building leases up and attains a relatively stabilized occupancy, project
absorption estimates using the competitive differentials approach need to be carried forward for at least three or four
years. This task requires supply-​demand equilibrium analysis of the project’s market over the forecast period either
through econometric or accounting techniques. Equation (16.7) does account for the demand-​supply equilibrium
conditions expected to prevail over the forecast horizon.
5 As explained earlier, project rent estimates through the competitive differentials technique can be carried forward by
applying forecasts of market-​rent growth rates.

References and additional readings

Archer, W. R. and M. T. Smith. 2003. Explaining Location Patterns of Suburban Offices. Real Estate Economics, 31:2,
139–​164.
Brennan, T., R. Cannaday, and P. Colwell. 1984. Office Rent in the Chicago CBD. AREUEA Journal, 12:3, 243–​260.
Carn, N., J. Rabianski, R. Racster, and M. Seldin. 1988. Techniques of Performing Office Market Analysis. Chap. 12 in
Real Estate Market Analysis: Techniques and Applications. Englewood Cliffs, NJ: Prentice-​Hall.
Clapp, J. 1987. Evaluating Competing Properties. Chap.  10 in Handbook for Real Estate Market Analysis. Englewood
Cliffs, NJ: Prentice-​Hall.
DiPasquale, D. and W. Wheaton. 1996. The Operation of Nonresidential Property Markets. Chap. 11 in Real Estate
Markets and Urban Economics. Englewood Cliffs, NJ: Prentice Hall.
Elgar, I. and E. J. Miller. 2010. How Office Firms Conduct their Location Search Process? International Regional Science
Review, 33:1, 60–​85.
Hou,Y. 2016.Traffic Congestion, Polycentricity, and Intraurban Firm Location Choices: A Nested Logit Model for the
Los Angeles Metropolitan Area. Journal of Regional Science, 56:4, 683–​176.
McMahan, J. 1989. Office and Industrial. Chap.  8 in Property Development. New  York:  McGraw-​Hill Publishing
Company.
Mun, S. and B. Hutchinson. 1995. Empirical Analysis of Office Rent and Agglomeration Economies: A Case Study of
Toronto. Journal of Regional Science, 35:3, 437–​455.
Myers, D. and P. S. Mitchell. 1993. Identifying a Well-​Founded Market Analysis. Appraisal Journal, 61:4, 500–​508.
Peiser, R. and D. Schwanke. 1992. Professional Real Estate Development. Washington, DC: Dearborn Financial Publishing
and Urban Land Institute.
Sivitanidou, R. 1996. Do Office Firms Value Access to Office Employment Centers? A Hedonic Value Analysis in the
Los Angeles PMSA. Journal of Urban Economics, 40:2, 125–​149.

378
 Micro analysis of office markets

Sivitanidou, R. 1997. Are Center Access Advantages Weakening? The Case of Office-​Commercial Markets. Journal of
Urban Economics, 42:1, 79–​97.
Sivitanidou, R. and W. Wheaton. 1992. Wage and Rent Capitalization in the Commercial Real Estate Market. Journal
of Urban Economics, 31:2, 206–​229.
Vandell, K. D. and J. S. Lane. 1989. The Economics of Architecture and Urban Design: Some Preliminary Findings.
AREUEA Journal, 17:2, 233–​259.

379
newgenprepdf

17
OFFICE MARKET ANALYSIS
Synthesis and market studies

Introduction 380
Example: multi-tenant office development 381
The project 381
Time of entry 381
Structure typology 381
Analysis of hedonic lease rates, 1996 381
Market information 382
Analysis 382
Evaluating office market studies 384
Macro analysis of office markets 384
Micro, site-, and project-specific analysis of office projects 384
Project #6: Evaluating an office development scenario 388
What to turn in 388
Micro market data 388
Site and structure attributes 388
Office lease rates: hedonic analysis results 388
Macro market conditions 388
Forecast of metropolitan office market fundamentals 388
Space requirements per worker 390
Office market absorption equation 390
References and additional readings 390

Introduction
In this chapter, we first present a hypothetical example that demonstrates how some of the different
macro-​and micro-​analysis methodologies, discussed in the previous chapters, can be synthesized to derive
expected lease rates and revenue schedules for a small multi-​tenant office development. Subsequently, we
conclude the chapter by outlining broader evaluation criteria for office market studies.

380
 Synthesis and office market studies

Example: multi-​tenant office development

The purpose of the example presented below is to demonstrate how elements from the macroeconomic
analysis of the broader office market and the microeconomic analysis of the project under consideration
can be combined to help estimate project rental rate, absorption rate, and revenue schedule. Note that all
the numbers incorporated in this example are hypothetical.

The project
Speculative development of three multi-​tenant office structures, no larger than 45,000 square feet each, is
planned on a large office-​commercial site. Market analysis needs to be conducted to help determine whether
market conditions favor the development of such space on the site and, most importantly, provide sensible
estimates of rental revenue schedules necessary for evaluating the financial feasibility of such a development.

Time of entry
The investor needs to make decisions about the timeline of the project. The earliest time of entry is year-​
end 1997 or beginning of 1998.

Structure typology
Preliminary analysis has shown that the area typically attracts suburban-​type developments with higher-​end
multi-​tenant structures. The latter are no larger than 45,000 square feet and no higher than three stories
of office floor space. Furthermore, such space is of average competitive position (relative to class A multi-​
tenant space in the broader submarket within which the site is located).

Analysis of hedonic lease rates, 1996

A recent analysis of base-​year office lease rates ($/​sq. ft.) in the area within which the site under consid-
eration is located suggests the following relationship between lease rates, lease characteristics, and other
structure and location attributes.

Table 17.1  Hedonic regression results, 20201

Independent variables Coefficient estimate t-​statistic

Constant 2.648 3.46
Stop amount 0.029 2.98
CPI escalation –0.421 –4.36
Lease length 0.012 7.18
Construction quality {1,0} 0.123 4.98
Neighborhood quality {1,0} 0.145 2.75
Retail/​restaurant on site{1,0} 0.083 3.71
N (sample size) 240
R-​squared 0.89

The dependent variable is the natural log of base-​year lease rates. All independent variables are in
1

non-​logarithmic  form.

381
 Office market analysis

Table 17.2  Structure, lease, and other attributes

Variable Value Explanations

Stop amount 3.75 Amount above which tenant pays share of increases in operating
expenses
CPI escalation 0.02 Lease rates escalate by 2% annually
Lease length 3, 5, 10 Measured in years. Typically, one-​third of the leases in the submarket
under question are three-​year leases, another third is five-​year leases,
and the remaining third 10-​year leases. Assume that new leases
signed each year follow this distribution.
Construction quality {1,0} 1 =1 if very good; =0 if otherwise
Neighborhood quality {1,0} 1 =1 if very good; =0 if otherwise
Retail/​restaurant on site {1,0} 1 =1 if yes; =0 if otherwise

The values of the structure and lease characteristics that should be used in association with the
parameters presented in the regression results in Table 17.1 in order to estimate the lease rates for the pro-
ject are provided in Table 17.2.

Market information
The absorption of class A  multi-​tenant space is expected to obey the following relationship that was
derived using regression analysis. The R2 of the estimated statistical equation below was 0.82.

ABS(t) = 262,024 + 52.96 * E(t) –​ 21.35  R(t–​1)

     (3.26)    (3.97)  (–​
4.53)

where:
E(t) : Growth in workers using speculative office space
R(t-​1) : Nominal lease-​rate index, lagged one year (given below)

The information required to assess market demand, that is, ABS(t) in the above equation, is provided
below. In particular, Table 17.3 provides employment growth forecasts by occupation across sectors and
Table 17.4 the percentage of office space usage by occupation and sector.
The information regarding expected competing supply and nominal lease rate growth rates (as described
by a lease rate index) is provided in Table 17.5. Note that the lease rate describes expected movements in
lease rates for all leases independently of duration.

Analysis
In assessing the market viability of this project, the first step is the estimation of expected growth in office-​
using employment sectors. Given the data provided, the disaggregate accounting approach is used to carry
out the necessary calculations. To this aim, the percentages reflecting overall use and use of speculative
office space by sector and type of occupation were applied to respective forecast employment growth fig-
ures (see Table 17.6).
In the next major step, the analysis focuses on the estimation of achievable lease rates for the project. In
particular, using the hedonic regression results, base-​year lease rates for leases of different duration (3, 5, and
10 years) are calculated by applying each time the appropriate figure to the coefficient of “Lease Length”

382
 Synthesis and office market studies

Table 17.3  Expected employment growth by occupation across sectors

2018 2019 2020

Managerial occupations in:
Agriculture 308 387 423
Construction 670 841 918
Manufacturing 5,591 7,020 7,659
Retail/​Wholesale 3,152 3,957 4,319
FIRE 2,018 2,534 2,765
Business Services 1,386 1,740 1,900
Professional Services 10,884 13,670 14,916
Other Services 1,361 1,709 1,867
Public Administration 1,625 2,040 2,227
Technical and Administrative Support occupations in:
Agriculture 262 329 359
Construction 623 782 853
Manufacturing 7,408 9,302 10,152
Retail/​Wholesale 9,519 11,952 13,045
FIRE 4,749 5,963 6,508
Business Services 1,521 1,910 2,085
Professional Services 5,539 6,955 7,591
Other Services 915 1,149 1,255
Public Administration 1,859 2,334 2,547

Table 17.4  Office-​using occupations as percentage of total

Office-​using sectors Managerial Technical and Administrative Percentage of workers using

Support speculative space
Agriculture 80% 86% 100%
Construction 95% 87% 100%
Manufacturing 85% 79% 50%
Retail/​Wholesale 90% 25% 100%
FIRE 98% 58% 100%
Business Services 75% 81% 0%
Professional Services 45% 95% 100%
Other Services 60% 57% 0%
Public Administration 75% 88% 20%

(see 1 in Table 17.7).These base-​year lease rates for the three types of leases are projected forward using the
market price index provided (see 2 in Table 17.7).
The final input needed for the calculation of the project’s revenue schedule is its absorption schedule.
Given the information provided for this exercise, project absorption is calculated in two steps. First, using
the estimated parameters of the absorption equation, the estimated growth in office-​using employment
and the forecasts of the market lease-​rate index, forecasts of market absorption are developed (see 3 in
Table 17.7).1 These forecasts then provide the basis for calculating the project’s absorption using the con-
cept of the fair market share, as explained in Chapter 9 (see 3 in Table 17.7).
Finally, the forecasts of achievable lease rates and project absorption schedule are used to develop project
revenue schedules. Given the assumptions regarding the mix of leases, forecasts of total revenues expected

383
 Office market analysis

Table 17.5  Expected competing supply and lease-​rate index

Year Competing supply (sq.ft.) Nominal lease-​rate index

1996 1,340,000 100.0
1997 1,560,000 105.2
1998 2,341,100 111.3
1999 1,970,700 118.4
2000 1,700,760 126.5
2001 1,850,000 132.6
2002 2,000,100 136.8
2003 1,920,000 140.6
2004 1,723,000 143.0

to be received by each type of lease are first estimated, and then aggregated to derive forecasts of total pro-
ject revenues over the forecast period (see Table 17.8).

Evaluating office market studies

The standard evaluation criteria of market studies in general—​ soundness and consistency of data,
assumptions, methodology, and conclusions—​do apply in evaluating office market studies. In applying these
criteria, however, it is important to pay special attention as to whether the study addresses in an appropriate
and reliable way the crucial issues pertaining to the idiosyncrasies of the office markets at the macro and
micro level. These issues are highlighted below in the form of questions.

Macro analysis of office markets

• Are office market cycles in net absorption, new construction, vacancies, and rents properly analyzed,
well understood, and sensibly interpreted? Is the timing issue, in light of these cycles, carefully
addressed?
• Do estimates for office space requirements account for differences in space needs per worker across eco-
nomic sectors and office-​using occupations?
• Do estimates of space needs per worker account for the effect of rental rates? Are assertions of increasing
demand for office space grounded on sensible and economically defensible expectations of growth in
office employment or ungrounded assumptions of increases in space per employee?

Micro, site-​, and project-​specific analysis of office projects

• Is location analyzed properly both in terms of the advantages it offers to and the constraints it imposes
on the target firms and their workers? Are issues of location prestige addressed?
• Are office lease rates estimated through sound techniques, taking into account both lease characteristics,
project amenity packages, and location benefits or constraints? Do lease rate estimates account for
expected changes in market conditions?
• Do estimates of absorption rates take into account the competitive strength of the project and forecasted
changes in market conditions or are they based on current time-​on-​the-​market estimates?

384
 Synthesis and office market studies

Table 17.6  Developing rental revenue schedules for speculative office development: Estimating

office employment growth
Managerial occupations Technical occupations

Economic sectors 1998 1999 2000 1998 1999 2000

Agriculture 308 387 423 262 329 359

Construction 670 841 918 623 782 853
Manufacturing 5,591 7,020 7,659 7,408 9,302 10,152
Retail/Wholesale Trade 3,152 3,957 4,319 9,519 11,952 13,045
FIRE 2,018 2,534 2,765 4,749 5,963 6,508
Business Services 1,386 1,740 1,900 1,521 1,910 2,085
Professional Services 10,884 13,670 14,916 5,539 6,955 7,591
Other Services 1,361 1,709 1,867 915 1,149 1,255
Public Administration 1,625 2,040 2,227 1,859 2,334 2,547

% Workers
using Specu-
Percentage employment in office-using occupations lative space

Agriculture 80% 80% 80% 86% 86% 86% 100%

Construction 95% 95% 95% 87% 87% 87% 100%
Manufacturing 85% 85% 85% 79% 79% 79% 50%
Retail/Wholesale Trade 90% 90% 90% 25% 25% 25% 100%
FIRE 98% 98% 98% 58% 58% 58% 100%
Business Services 75% 75% 75% 81% 81% 81% 0%
Professional Services 45% 45% 45% 95% 95% 95% 100%
Other Services 60% 60% 60% 57% 57% 57% 0%
Public Administration 75% 75% 75% 88% 88% 88% 20%
Expected growth in number of workers in office-using sectors

Agriculture 246 310 338 225 283 309

Construction 637 799 872 542 680 742
Manufacturing 4,752 5,967 6,510 5,852 7,349 8,020
Retail/Wholesale Trade 2,837 3,561 3,887 2,380 2,988 3,261
FIRE 1,978 2,483 2,710 2,754 3,459 3,775
Business Services 1,040 1,305 1,425 1,232 1,547 1,689
Professional Services 4,898 6,152 6,712 5,262 6,607 7,211
Other Services 817 1,025 1,120 522 655 715
Public Administration 1,219 1,530 1,670 1,636 2,054 2,241

Estimated growth in employment

using speculative space

Economic sectors 1998 1999 2000

Agriculture 471 592 647 = 14,916 × 45%

Construction 1,178 1,479 1614
Manufacturing 5,302 6,657 7,265
Retail/Wholesale Trade 5,216 6,549 7,148 = 7591 × 95%
FIRE 4,732 5,941 6,484
Business Services 0 0 0
Professional Services 10,159 12,758 13,923 = (6,712 × 100%) + (7,211 × 100%)
Other Services 0 0 0
Public Administration 570 716 782

Total 27,628 34,692 37,863

385
 Office market analysis

Table 17.7  Developing rental revenue schedules for speculative office development: Hedonic first-​year office lease
rates and absorption schedule
1. Estimating base-​year lease rates

Independent variables Coefficient Lease rate estimation, 1996

estimate
3-​year 5-​year leases 10-​year leases
leases
Constant 2.648 1 2.65 1 2.65 1 2.65
Stop Clause {1,0} 0.1087 1 0.11 1 0.11 1 0.11
CPI escalation –0.4214 0.02 –0.01 0.02 –0.01 0.02 –0.01
Lease length 0.012 3 0.04 5 0.06 10 0.12
Construction quality {1,0} 0.123 1 0.12 1 0.12 1 0.12
Neighborhood quality {1,0} 0.145 1 0.15 1 0.15 1 0.15
Retail/​restaurant on Site{1,0} 0.0825 1 0.08 1 0.08 1 0.08
ln(Lease rate) 3.13 3.16 3.22
Lease rate 22.98 23.54 25.00
2. Developing first-​year lease rate schedules
Year Lease Lease rate schedules
rate
3-​year 5-​year 10-​year
index
1996 100.00 22.98 23.54 25.00
1997 105.20 24.18 24.77 26.30
1998 111.30 25.58 26.20 27.82
1999 118.42 27.22 27.88 29.60
2000 126.48 29.07 29.78 31.62
2001 132.55 30.46 31.20 33.13
2002 136.79 31.44 32.20 34.19
2003 140.62 32.32 33.10 35.15
2004 143.01 32.87 33.67 35.75
3. Estimating project absorption (applies to each of the three structures)
Year Growth in Lease Market Competing Available Project
office rate index absorption supply space absorption
workforce
1996 100.00
1997 105.20
1998 27,628 111.30 1,722,957 2,341,100 45000 33,118
1999 34,692 118.42 2,096,936 1,970,700 11,882 11,882
2000 37,863 126.48 2,264,720 1,700,760
2001 132.55
2002 136.79 ABS (t) = 262024 + 52.96*Office employment growth -​
2003 140.62 21.35*Rent index (t-​1)
2004 143.01

Note: The project’s competitive strength is assumed to be average.

386
 Synthesis and office market studies

Table 17.8  Developing rental revenue schedules for speculative office development: Developing total revenue
schedules

Three-​year leases (CPI escalation: 2%)

Year First-​year Project Revenue schedules
lease absorption From space From space Total
rate leased leased
in 1998 in 1999
1998 25.58 33,118 847,181 0 847,181
1999 27.22 11,882 864,125 323,386 1,187,511
2000 29.07 881,407 329,854 1,211,261
2001 30.46 1,008,930 336,451 1,345,380
2002 31.44 1,029,108 373,551 1,402,660
2003 32.32 1,049,691 381,022 1,430,713
2004 32.87 1,088,548 388,643 1,477,191
Five-​year leases (CPI escalation: 2%)
Year First-​year Project Revenue schedules
lease absorption From space From space Total
rate leased leased
in 1998 in 1999
1998 26.20 33,118 867,760 0 867,760
1999 27.88 11,882 885,115 331,241 1,216,356
2000 29.78 902,817 337,866 1,240,683
2001 31.20 920,873 344,623 1,265,496
2002 32.20 939,291 351,515 1,290,806
2003 33.10 1,096,355 358,546 1,454,901
2004 33.67 1,118,282 400,023 1,518,306
Ten-​year leases (CPI escalation: 2%)
Year First-​year Project Revenue schedules
lease absorption From space From space Total
rate leased leased
in 1998 in 1999
1998 27.82 33,118 921,419 0 921,419
1999 29.60 11,882 939,847 351,724 1,291,571
2000 31.62 958,644 358,758 1,317,402
2001 33.13 977,817 365,933 1,343,750
2002 34.19 997,373 373,252 1,370,625
2003 35.15 1,017,321 380,717 1,398,037
2004 35.75 1,037,667 388,331 1,425,998
Year 1998 1999 2000 2001 2002 2003 2004
Total Rental 2,636,359 3,695,437 3,769,346 3,954,627 4,064,091 4,283,651 4,421,495
Revenue

Note: For simplicity, it has been assumed that all expiring leases are renewed at the market rate. See discussion in this
chapter in terms of how this issue should be handled.

387
 Office market analysis

Project #6: Evaluating an office development scenario

This project focuses on the development of cumulative gross revenue schedules for a two-​structure office
development. The two-​structure project will be developed in two phases, each involving the construction
of 80,000 sq. ft. The first phase will be completed in year-​end 2001 and the second in year-​end 2002.
Your analysis will be based on two sets of data derived from (i) a detailed macro analysis of the relevant
office submarket and (ii) micro-​analysis of the project and its site. Note that the information provided
allows the use of alternative methodologies.You should be able to identify these alternatives and rationalize
your choices.

What to turn in
A short essay explaining how you have derived the bottom-​line and other supporting figures presented
in your analysis. Number and clearly label the tables presenting these figures and refer to them in the text,
whenever appropriate.

Micro market data
Site and structure attributes
The site is zoned office-​commercial. The maximum allowable density has to be taken as given, as prelim-
inary analysis of the marginal benefits and costs associated with a potential rezoning request suggested
that the latter would not make economic sense. Given the allowable density and the site area, a maximum
of 240,000 square feet of office space can be built on the site. The preliminary development scenario calls
for the construction of two structures in two development phases. Financial and other considerations have
primarily led to such a decision. The site is located within a low crime area, four miles away from the
metropolitan area’s major service employment subcenter, and in close proximity to a smaller node that
houses office-​and retail-​commercial activities, including restaurants. The planned space will be Class A,
of good construction quality. Its competitive position in its specific submarket is expected to be average.

Office lease rates: hedonic analysis results

Leases for space similar to the one contemplated are expected to have a stop clause, as well as CPI escalation
of 2% (meaning that lease rates should increase by 2% a year). Assume four-​year office leases, a $3.75 stop
amount, and, if needed, a 5% vacancy rate.
Table 17.9 presents the results of two alternative hedonic analyses of office leases signed year-​end 1998
in the office submarket under question, as well as other competing submarkets. Attributes with little or no
variation across lease transactions were omitted from the hedonic analysis.You should decide which of the
two sets of hedonic analysis is more sensible to use for forecasting lease rates. Note that the dependent vari-
able is the base-​year lease rate in dollars per square foot of net rentable area. This is expressed in its natural
logarithmic form in the estimated hedonic equations below.

Macro market conditions

Forecast of metropolitan office market fundamentals
Forecasted growth in speculative office-​using employment, expected changes in office market rents, and
competing market supply for the metropolitan area within which the project is located are provided in
Table 17.10.

388
 Synthesis and office market studies

Table 17.9  Hedonic analysis results

Hedonic results I Hedonic results II

Independent Variables Coefficients T-​ratios Independent variables Coefficients T-​ratios

Constant 2.55 3.26 Constant 2.59 3.46
Stop clause {1,0} –​0.15 –​3.98 Stop clause {1,0} –​0.11 –​2.98
CPI escalation –​0.51 –​5.32 CPI escalation –​0.44 –​4.36
Construction quality {1,0} 0.18 3.98 Construction quality {1,0} 0.15 3.92
ln(Center access) –​0.07 –​3.45 ln(Center access) –​0.05 –​3.41
Crime Rate {1,0} 0.19 2.15 Crime rate {1,0} 0.14 2.75
Restaurant access {1,0} 0.24 3.67 Restaurant access {1,0} 0.20 3.71
Vacancy rate 0.05 3.64

R-​squared 0.90 R-​squared 0.88

N (sample size) 188 N (sample size) 188
Variable Definitions
Stop clause {1,0} Dummy takes the value of 1 if there is a stop clause; it otherwise equals 0
CPI escalation Expressed in % inflation adjustment annually. If CPI escalation is 2%, the value of
this variable is 0.02
Construction quality {1,0} Dummy takes the value of 1 if contruction quality is good and the value of 0
otherwise
Center access Distance to the closest of major office clusters, in miles
Crime rate {1,0} Dummy takes the value of 1 if crime rate in the site’s immediate neighborhood is
low and the value of 0 otherwise
Restaurant access {1,0} Dummy takes the value of 1 if the site has good access to restaurants and the value
of 0 otherwise
Vacancy rate Vacancy rate in the structure at the time the lease was signed

Table 17.10  Forecasts of metropolitan market fundamentals

Year Growth in office-​using technical Growth in office-​using Competing market supply

and administrative support occupations managerial occupations (in sq. ft.)
1999
2000
2001
2002 5,000 8,000 1,444,800
2003 5,500 8,000 1,252,160
2004 4,700 4,000 1,651,200
2005 5,500 4,000 1,135,200
2006 5,000 10,000 812,800
2007 4,500 7,500 1,100,000
2008 4,000 8,000 500,000

389
 Office market analysis

Table 17.11  Office space absorption equation

Independent variables Coefficients t-​statistics

Constant 847,643 2.10
Lagged nominal rent index –​2,200 –​7.10
Office employment growth 57.86 7.21
R-​squared 0.97

Space requirements per worker

Recent data indicate that the workers in managerial occupations use, on average, 350 square feet, while
workers in technical and administrative support occupations use, on average, 210 square feet.

Office market absorption equation

Per the results described in Table 17.11, speculative net office space absorption during year t is found to
depend on growth in employment using speculative space during the same year, as well as lagged office
rents, proxied by the market’s nominal office rent index in the previous year (t–​1).

Note
1 Notice that in order to do that we do not multiply the estimated growth in office-​using sectors by a space-
per-​employee  norm.

References and additional readings

Fanning, S.,T. Grissom, and T. Pearson. 1994. Existing Office Building. Chap. 13 in Market Analysis for Valuation Appraisals.
Chicago, IL: Appraisal Institute.
McMahan, J. 1989. Office and Industrial. Chap.  8 in Property Development. New  York:  McGraw-​Hill Publishing
Company.
Schmitz, A. and D. Brett. 2001. Office and Industrial Development. Chap. 4 in Real Estate Market Analysis: A Case Study
Approach. Washington, DC: Urban Land Institute.

390
 PART F

Industrial market analysis

 18
INDUSTRIAL SPACE MARKET

Introduction 394
Industrial market idiosyncrasies 395
Functional heterogeneity 395
Large non-speculative component 395
Industrial cycles parallel business, as opposed to real estate cycles 396
The demand for industrial space 396
Growth in industrial output or employment 396
Shifts in the real after-tax cost of corporate capital 397
Replacement demand 397
Square feet per employee 398
The supply of industrial space 400
Industrial market analysis: an overview 401
Macroeconomic analysis of industrial markets 401
Accounting approaches 401
Econometric approaches 402
Microeconomic site-specific analysis of industrial projects 402
Macroeconomic analysis of industrial markets 402
Estimating the industrial demand-supply gap through accounting techniques 402
Accounting techniques for forecasting industrial space demand 403
Method 1: Demand analysis for general-purpose industrial space 403
Method 2: Best-fit industry analysis 404
Econometric modeling of the industrial market 409
Modeling investment in non-speculative industrial buildings 410
Modeling the speculative industrial market 411
Site-specific analysis of industrial projects 413
Step 1: Evaluation of site attributes 413
Site features 413
Site access and linkages 414

393
newgenprepdf

Industrial market analysis

Traditional manufacturing (production) space 414

Warehouse and distribution space 414
Research and development space 415
Community constraints 415
Step 2: Identification of types of tenants most likely to be attracted to the site 416
Step 3: Analysis of the competition 416
Step 4: Estimation of project absorption schedule and achievable rents or land prices 417
On industrial market studies: evaluation criteria 419
Macro, non-site-specific analysis 419
Site-specific analysis of development projects 420
Chapter summary 420
Questions 421
References and additional readings 422

Introduction
This chapter begins with an overview of the basic features and idiosyncrasies of the industrial market, and
continues with a discussion of its major demand and supply drivers. Subsequently, the focus shifts to analysis
techniques typically used at the macroeconomic and the microeconomic level.
The industrial real estate market has received less attention than the office market due to both
the idiosyncratic nature of the market and the relative scarcity of data. Nevertheless, as in the case of
other property types, analysis at the macro level involves the use of either accounting or econometric
techniques for the assessment of market strength. If the focus of analysis is speculative warehouse, dis-
tribution, and R&D space, accounting techniques typically concentrate on a more aggregate analysis
of industrial space requirements due to the general-​purpose nature of this type of space. Assessment of
demand for land acreage or space needs for manufacturing activities or industrial parks, however, usu-
ally requires a more detailed identification of “best-​fit” industries by examining closely industry-​specific
location requirements and the extent to which the metro area under consideration is appealing to each
of these industries.
Up-​to-​date attempts to develop econometric models for the industrial market have focused on the
aggregate market in an effort to quantify the effect of relevant macroeconomic indicators on the demand
for industrial space, and the interactions between demand, supply, and rents. Given that a significant
portion of the industrial market is non-​speculative, an investment modeling approach is discussed along
with the conventional modeling approach that applies to the speculative portion of the industrial market.
It should be noted, however, that empirical work by DiPasquale and Wheaton (1996) suggests that the
latter approach can be successfully applied to the total industrial market encompassing both rental and
owner-​occupied  space.
Industrial market analysis at the site-​specific level involves the same steps and methodologies discussed
in the case of office projects. These include identification of the most likely tenants to be drawn to the
site; identification and evaluation of competing projects; estimation of the project’s space or land absorp-
tion schedule and achievable lease rates or land prices via competitive differential or hedonic valuation
techniques; and estimation of the project’s revenue schedule. Although these steps apply both to the
general-​purpose industrial space and to the more idiosyncratic manufacturing/​production space, project
demand analysis for the latter focuses on more detailed demand segments, as it does when analyzing market
demand. In particular, project demand analysis for production space focuses on the identification of the
subset of those “best-​fit” industries that were identified in the macroeconomic-​analysis stage that are likely
to exhibit the strongest interest in locating at the site under consideration.

394
 Industrial space market

Industrial market idiosyncrasies

The U.S.  industrial market is a very idiosyncratic market, greatly differentiated from the other non-​
residential real estate markets. Its major idiosyncrasies include:

Functional heterogeneity
The market for industrial space is quite heterogeneous as it includes structures used to perform several
different functions. According to the National Council of Real Estate Investment Fiduciaries (NCREIF),
these include:

• Flex space which allows for a wide variation in space utilization, including office, retail, personal service,
distribution, light industrial, and occasionally heavy industrial use.
• Research and Development space, which includes up to 50% office/dry lab space with the remainder
allowing uses as wet lab, workshop, storage, and other support activities.
• Warehouse (and Distribution) space which is used for inventory storage and distribution and usually
includes up to 15% office space.
• Manufacturing space used for production activities.
• Office Showroom space, which allows for retail display, storage, and distribution, and includes less than
15% office space.

Research and development space constitutes by far the smallest segment of the industrial market. Because
of differences in operations performed, structure characteristics, location, land, and space requirements, as
well as community acceptance, the focus of market studies across the various functional classifications does
and should differ.
The rapid development of e-​commerce in the last two decades has spurred the development of the logis-
tics sector, which has to be acknowledged as a special segment of the industrial space inventory. According
to Ronderos (2010), logistics buildings include refrigerated warehouses, non-​refrigerated warehouses, dis-
tribution or shipping centers, self-​storage and flex buildings with 50% or more of their space being used
for warehouse and storage activities.
The heterogeneity of industrial structures is of course the result of the heterogeneity of their tenants.
Unfortunately, the little information that is available with respect to the industrial classification of the firms
that occupy the different types of industrial space is very aggregate. In particular, DiPasquale and Wheaton
(1996) report that in 1991, 75% of the industrial inventory in 50 metropolitan markets was occupied by
firms with manufacturing and wholesale SIC classifications (53% and 22%, respectively). No further infor-
mation was provided, however, regarding the disaggregation of industrial space tenants at a more detailed
SIC level.1

Large non-​speculative component

Industrial space and, primarily, production space, is largely non-​speculative (Anderson and Guirguis, 2011).
Most production plants are built by their owners (as opposed to speculators) or for specific owners (build
to suite). According to DiPasquale and Wheaton (1996), 48% of the industrial buildings in 50 metropol-
itan areas in 1991 were owner-​occupied, while almost 75% of the total industrial space in these markets
was in single-​tenant buildings. This suggests that many industrial structures are built for particular owners
and demonstrates the large non-​speculative component of this property type. The major implication of
this, then, from a modeling point of view, is that a significant part of industrial space construction is the
result of business investment decisions rather than generic speculative development decisions. Developing

395
 Industrial market analysis

speculative production space is risky due to the special design requirements of production plants. For this
reason, industrial developers often prefer to speculatively prepare and provide industrial land as opposed
to production space.

Industrial cycles parallel business, as opposed to real estate cycles

DiPasquale and Wheaton (1996) present data suggesting that industrial completions follow macroeconomic
and business cycles closely and have a considerably less volatile cyclical pattern than office completions.The
latter is also confirmed indirectly by the Ibanez and Pennington-​Cross (2013) study, which found that flex
and industrial rents return to equilibrium following an exogenous demand shock considerably more quickly
than office rents.Two most likely explanations can be cited for the considerably lower volatility of industrial
completions. The first explanation is the very short lead development time required for industrial buildings
because, typically, they are much simpler and smaller structures than office buildings.The second is the non-​
speculative character of many industrial construction investments, which directly links demand with supply.

The demand for industrial space

The demand for industrial space in a metropolitan area is a derived factor demand and, as such, it should be
driven by current or expected growth in industrial output or employment, changes in the real after-​tax cost
of corporate capital, and replacement demand.

Growth in industrial output or employment

Employment growth may not be a very good predictor of growth in output or demand for industrial pro-
duction space. For example, as Wheaton and Torto (1990) report, manufacturing employment in the largest
52 industrial markets decreased by 1.1% between 1972 and 1989, while the estimated output per worker
increased by 41%, largely as a result of capital intensification and technological change. These statistics
indicate that growth in industrial output may be a better predictor of growth in demand for industrial pro-
duction space and, perhaps, warehouse and distribution space.2 When trying to model the industrial market
at any level other than the national, however, there are data availability problems in using this variable as a
demand driver.
Due to greater data availability, most of the existing empirical studies of the industrial market use
employment, as opposed to output, in trying to explain industrial space demand movements. For example,
Ibanez and Pennington-​Cross (2013) confirmed a positive and statistically significant effect of industrial
employment on industrial rents, obviously capturing demand side effects. In this study, industrial employ-
ment is measured as the sum of Manufacturing employment (NAICS 30000000)  and employment in
Warehouse and Storage (NAICS 43493000). Twist (2002), however, found that total employment predicts
industrial space demand better than other employment subcategories. Anderson and Guirguis (2011)
reached the same conclusion, but eventually developed a forecasting model of industrial space demand that
includes only two output-​related variables. In particular, they found that most of the variations in indus-
trial space demand over the period 1990–​2003 were explained by the Federal Reserve Board’s index of
manufacturing output (IMO) and the PMI index from the Institute of Supply Management (ISMPMI).
The IMO measures the quantity of goods produced, excluding mining and utilities, while the ISMPMI
is a leading indicator of output and is constructed by surveying purchasing managers’ expectations with
respect to new orders, production, employment, deliveries, and inventories. Both variables were found to
be leading indicators of industrial space demand. Anderson and Guirguis (2011), based on the accuracy
of their out-​of-​sample forecasts, argue that these two variables are “effective and unbiased” exogenous
predictors of industrial space demand because they encompass the full supply chain.3

396
 Industrial space market

DiPasquale and Wheaton (1996) report that in the case of warehouse and distribution space, employ-
ment in the wholesale sector is a much better proxy for demand than output. In theory, demand for ware-
house and distribution space should be strongly related to the volume of inventories, but again such an
indicator is not available at the metropolitan level. Anderson and Guirguis (2011) tested freight flows and
other air, rail, and shipping metrics but did not find any strong relationship between these variables and
industrial space demand.

Shifts in the real after-​tax cost of corporate capital

According to Wheaton and Torto (1990), the real after-​tax cost of corporate capital should influence
demand for industrial space because it discourages investment and production. This can be calculated as in
(18.1):

c = (n–​i)/​(1–​τ) (18.1)

where:
c : after-​tax cost of corporate capital
n : long-​term nominal interest rate (AAA bond rate)
i : expected inflation rate (usually calculated as the difference between the short-​term and long-​term
Treasury maturities)
τ : effective corporate tax rate

Data presented by Wheaton and Torto (1990) portray a mainly negative relationship between the after-​tax
cost of capital and wholesale employment. For example, in the late 1960s and most of the 1970s, when the
after-​tax cost of corporate capital was relatively low, wholesale employment was growing rapidly. In the
late 1980s, however, when the after-​tax cost of corporate capital was relatively high, wholesale employment
growth was low.

Replacement demand
In the case of industrial property, replacement demand (as a percentage of total stock) is considerably
greater compared to other property types. As such, it represents a significantly greater source of additional
demand for industrial space. In principle, replacement demand arises out of physical, functional, and eco-
nomic depreciation.

• Physical depreciation refers to the physical deterioration of a building’s condition simply due to aging.
Because of its slow pace, this type of depreciation contributes little to replacement demand.
• Functional depreciation refers to the functional obsolescence of industrial buildings due to technological
and other advancements and evolving manufacturing processes that require horizontally developed
manufacturing structures. In addition, the use of new more efficient vertical storage systems favoring
taller buildings, and a re-​definition of the functions performed at the end of the supply chain to include
value-​added functions (such as labeling, barcoding, simple assembly, etc.), are contributing to the func-
tional obsolescence of many warehouses and distribution centers. Functional depreciation of existing
industrial properties contributes significantly to replacement demand.
• Economic/​location obsolescence refer to obsolescence resulting from changes in manufacturing, ware-
housing, and distribution processes that alter the location-​decision calculus of owners and tenants of
industrial space. In recent years, some industrial properties have become economically obsolete due
to the re-​engineering of supply chain practices (motivated by the challenges of a global and highly

397
 Industrial market analysis

competitive economy), as well as due to advances in transportation systems and shipping technolo-
gies that allow increased inventory velocity. For example, because of the aforementioned trends, many
small warehouses are becoming economically obsolete as companies are consolidating smaller local
warehousing operations into larger regional distribution centers. Another contributor to the locational
obsolescence of industrial structures is their low cost and value, which facilitates their displacement by
other more valuable land uses. As a result, existing industrial uses, especially those located in expensive
and scarce central locations, are often being forced to relocate to new facilities at cheaper locations in
the suburbs.4

Economic and functional depreciation are the most important contributors to replacement demand for
industrial space in most markets. This is supported by data presented by Ronderos (2010). According to
the results of a survey of the logistics building inventory as of 2003, 40% of the building stock, or 4 billion
square feet, was built in the 13 years between 1990 and 2003. According to Ronderos (2010), these data
point to a rapid replacement of older stock by new construction, most likely due to the transformation of
the technology, space utilization, and location requirements of the industry that was taking place during
that period as a result of changing patterns of international trade and “just-​in-​time” inventory practices.

Square feet per employee5

The square feet per employee ratio is a key figure in an industrial market study since it is required to con-
vert industrial employment forecasts into space requirements (demand). As indicated earlier, due to the
high functional heterogeneity, square feet per employee varies considerably across the different types of
industrial space.
The Homes and Communities Agency report that was published in the UK in 2015 has a very good
discussion of the factors that affect industrial employment densities, which apply to the U.S. as well. The
four major driving factors identified by the report include advances in technology, the evolution of new
forms of workspace, changing trading formats, and sector and sub-​sector activity. According to the report,
the effects of these factors, and especially those of advances in technology, on employment densities in
manufacturing and distribution facilities can be summarized as follows:

• Advances in racking techniques and stock management software are reducing labor requirements for
multi-​product order picking, thus contributing to lower employment densities in warehouse and distri-
bution facilities.
• Greater automation, however, creates greater need for skilled employees for the servicing and support
of the machinery, thus contributing to higher employment densities.
• Requirements to improve operating efficiencies lead to the introduction of more “final assembly”
operations in manufacturing plants and distribution centers in particular, thus reducing the amount of
“pure” warehousing space and increasing employment densities.
• Within the context of improving operating efficiencies, more business operations are accommodated
under one roof, resulting in the integration of greater levels of office space within warehouse and dis-
tribution facilities, thereby contributing to higher employment densities.

In the U.S., a 2001 study of employment density ratios in the Southern California region prepared on
behalf of the Southern California Association of Governments found wide variation in employees per acre
and square feet per employee across the different industrial land uses (SCAG, 2001). As shown in Table 18.1,
R&D and Light Industrial uses seem to use a smaller amount of square feet per employee while heavy
manufacturing and warehouses use a greater amount of square feet per employee. In terms of employees
per acre, it seems that warehouse facilities have the lowest employment density while heavy manufacturing

398
 Industrial space market

Table 18.1  Employees/​acre and employees/​square foot by industrial land use (narrow polygon selection)

Median employees/ Average employees/ Square feet/​employee

acre acre
Based on median Based on average
employees/​acre employees/​acre
R&D/​Flex 18.13 20.53 609 569
Light Manufacturing 11.63 17.83 1,089 501
Heavy Manufacturing 17.05 31.14 1,700 931
Warehouse 10.63 11.4 1,274 960

Source: SCAG (2001)

Table 18.2  Employees/​acre and employees/​square foot by industrial land use (broad polygon selection)

Median employees/ Average employees/​ Square feet/​employee

acre acre
Based on median Based on average
employees/​acre employees/​acre
R&D/​Flex 18.13 20.53 527 344
Light Manufacturing 11.63 17.83 924 439
Heavy Manufacturing 17.05 31.14 NA NA
Warehouse 10.63 11.4 1,225 814

Source: SCAG (2001)

has the highest, with the number of employees per acre being considerably higher than the other three
industrial uses presented in Tables 18.1 and 18.2.
The difference between the two tables is that the results in Table 18.1 are based on a more narrow selec-
tion of parcels based on an average employment criterion, referred to in the report as “narrow polygon
selection” while the results of Table 18.2 are based on a broader selection with no restriction in terms of
employment density and thus the estimates are based on the analysis of a larger number of parcels. In both
tables, the square feet per employee numbers based on the median employees/​acre for warehousing have
the smallest deviation from the square feet per employee reported for warehouse and distribution facilities
by Ronderos (2010).
The Ronderos (2010) study that used building-​specific data collected in 2003 reports a considerably
higher amount of square feet per employee for logistics facilities in general, and for warehouse and distri-
bution facilities in particular, than those reported by the SCAG study. This study analyzed data pertaining
to buildings identified by the Energy Information Administration (EIA) as “Warehouse and Storage”
buildings, and included the following subcategories: refrigerated warehouse; non-​refrigerated warehouses;
distribution or shipping centers; self-​storage; and flex buildings (with 50% or more warehouse and storage
activities). According to these data, the average square feet per employee increased from 1,685 in 1999 to
2,241 in 2003. However, there were some significant differences in the square feet per employee across the
different types of logistics buildings. In particular, distribution/​shipping centers and refrigerated warehouses
used the same square feet per employee (1,906 and 1,910, respectively) but non-​refrigerated warehouses
used a considerably higher amount of square feet per employee (2,546). Flex buildings were characterized
by a much lower use of square feet per employee (506).
Average square feet per employee varied significantly across regions, in particular from 1,598 in the
West to 2,727 in the Midwest. The data show also a clearly discernible positive relationship between

399
 Industrial market analysis

Table 18.3  Employment densities for different industrial facilities in the UK

Average square feet per full-​time

equivalent (FTE) employee
R&D Space 431–​646
Light Industrial 506
Industrial and Manufacturing 388
Storage and Distribution
National Distribution Center 1,023
Regional Distribution Center 829
‘Final Mile’ Distribution Center 753

Source: Homes and Communities Agency (2015).

building size and square feet per employee. In particular, with the exception of buildings of 25,000 to
50,000 square feet and 500,000 to 1 million square feet, the average square feet per employee increased
from 1,842 for buildings of 1,000–​5,000 square feet to 3,877 for buildings of 1,000,000 square feet or
larger. With the exception of buildings built before 1946, which are characterized by very high square
feet per employee ratios (in excess of 3,900), there is no clear relationship between age and average square
feet per employee. There is a discernible negative relationship between age and square feet per employee
for buildings that were built after 1980. In particular, square feet per employee increased from 1,592 for
buildings that were built between 1980 and 1989 to 2,546 for buildings that were built between 2000
and 2003.
The Homes and Communities Agency survey in the UK reports generally lower numbers of square feet
per employee across all types of industrial facilities. According to this survey, average square feet per full-​
time equivalent (FTE) ranged from 388 in the case of industrial and manufacturing facilities to 1,023 in the
case of national distribution centers (Homes and Communities Agency, 2015). As indicated in Table 18.3,
the different types of storage and distribution facilities required more square feet per employee compared
to other types of industrial space.
The bottom-​line conclusion from the different surveys in the U.S. and the UK is that the square feet per
employee and employees per acre vary considerably across industries, across types of industrial facilities, and
across metropolitan areas and regions. Furthermore, space requirements per employee are changing rapidly
with new advances in technology and the resultant changes in business practices and the global competi-
tive environment. For this reason, it is very important to collect up-​to-​date primary information about the
square feet per employee used by the particular types of facilities considered and the targeted industries in
the region/​metropolitan area within which the project is located, in order to derive reasonable and reliable
estimates of industrial space requirements.

The supply of industrial space

The supply of non-​speculative industrial space is determined in theory by the factors driving business invest-
ment decisions, which include the cost of capital and expected output. Using an investment model of new
industrial structures, Wheaton and Torto (1990) were able to verify empirically the effect of the cost of
capital on aggregate industrial completions (that is, the sum of speculative and non-​speculative space).They
were also able to verify the effect of wholesale employment, which was used as proxy for output in the
wholesale sector.6 A crude measure of manufacturing output, constructed by the analysts, was statistically
insignificant, but manufacturing employment, used instead of manufacturing output in an alternative spe-
cification, proved to have a clearly statistically significant effect.7

400
 Industrial space market

The supply of speculative industrial space should be driven by the usual factors, that is, industrial-​space
rents, vacancy rates, the cost and availability of real estate development inputs, as well as risk/​uncer-
tainty and expectations regarding future market strength. In an empirical analysis of the Philadelphia
industrial market, DiPasquale and Wheaton (1996) found that industrial completions were driven by
lagged rental rates and absorption. The latter is capturing the effect of investor perceptions regarding
the strength of the market. Thompson and Tsolacos (2000) confirmed also the positive effect of indus-
trial rents and the negative effect of construction costs on industrial property construction activity in
Great Britain.
Ideally, separate supply or completions forecasting equations would be desirable for the different types
of industrial space, as construction investment in each of these types may respond differently to changes in
the cost of capital and market strength indicators, such as absorption and rental rates.

Industrial market analysis: an overview

Macroeconomic analysis of industrial markets
The macroeconomic component of industrial market analysis focuses on the evaluation of the industrial
development potential and space needs of the broader market area. Such needs can be assessed through
accounting or econometric techniques. Given that each approach has its own shortcomings and advantages,
it may be useful for the analyst to gain insights with respect to the area’s industrial demand prospects by
applying both approaches.8
The advantage of accounting techniques in analyzing industrial space demand is that there are no
data limitations in terms of the number and detail of industrial sectors examined. For example, best-​fit
industry analysis, often employed in demand analyses for production space, focuses on the identification
of potential target industries and their space needs at the detailed NAICS level. As such, it may better
assess an area’s production space and land requirements since significant variations in space usage have
been detected at this level of disaggregation (SIR, 1984). The disadvantage of accounting techniques,
however, is that they apply static employee-​per-​acre or employee-​per-​square-​foot ratios that do not
account for the effect of movements in prices/​rents over the forecast period. Such effect should not be
ignored, if possible, because it should cause employment density and/​or output multipliers to fluctuate
through time.
Econometric modeling techniques can account for the potential effect of changes in industrial rents by
including the latter in the right-​hand side of the demand equation. However, because of data limitations,
usually the relationship between employment and industrial space demand can be quantified for only a
few aggregate sectors. As a result, some inaccuracies may be introduced into the space-​demand estimates
because of the failure to account for variations in space-​per-​employee requirements across detailed indus-
trial classifications.

Accounting approaches
Accounting approaches for evaluating industrial market prospects at the metro level differ in their level of
detail and focus depending on the nature of industrial space contemplated. These include:

• Assessment of demand for speculative space (primarily warehouse, distribution, and R&D space)
focusing on the identification of industries using industrial space, typically at a more general NAICS
classification level.
• Assessment of demand for land acreage or space for manufacturing activities or industrial parks, focusing
on the identification of best-​fit industries at the more detailed three-​or four-​digit NAICS level.

401
 Industrial market analysis

Econometric approaches
Given that a significant portion of the industrial market has a non-​speculative character, two econometric
approaches are discussed:

• Non-​speculative industrial market model

• Speculative industrial market model.

Microeconomic site-​specific analysis of industrial projects

Industrial market analysis at the site-​specific level has two major objectives. The first objective is the
assessment of the strengths and weaknesses of a given site in attracting the types of industrial tenants that are
most likely to seek space in the broader market area. The second objective is the evaluation of the project’s
potential in terms of absorption and rental rates or land prices. More specifically, this stage involves the
following steps:

• Evaluation of the site’s attributes and its appeal to the area’s best-​fit industries as identified from the
macroeconomic analysis in the case of manufacturing space, or to more general industrial-​space users,
in the case of warehouse, distribution, and R&D space.
• Identification of the most likely target tenants for the specific project and their growth prospects.
• Analysis of the competition and assessment of the project’s competitive position.
• Estimation of the project’s expected space/​land absorption and achievable industrial space/​land rents.

Macroeconomic analysis of industrial markets

Estimating the industrial demand-​supply gap through accounting techniques
In estimating the industrial market gap through accounting techniques, the contribution of depreciation is
explicitly taken into account.Thus, additional industrial space or land requirements can be estimated as the
sum of demand from new growth and replacement demand, reduced by available supply of land or space
(see hypothetical example in Table 18.4).

RISL = OG + DS –​AS (18.2)

where
RISL : required industrial space or land
OG : demand from growth in employment/​output
DS : space needed to replace depreciated stock
AS : available supply of industrial space or land
Demand from growth in employment/​output refers to industrial space demand stemming from expected
growth in employment and/​or output in economic sectors using industrial space. This portion of indus-
trial space demand can be calculated through accounting techniques that are mostly similar to the ones
discussed in the case of the office market.
Space needed to replace depreciated stock refers to the space that is needed to replace industrial
buildings that are removed from the active marketplace because they are no longer functional. The Center
for Real Estate and Urban Economics at the University of California at Berkeley indicates that industrial
stock obsolescence rates range between 2 and 5% annually depending on the market. In a study of the
industrial space needs in California, these analysts use industrial stock depreciation rates ranging between
1% and 3.5%.

402
 Industrial space market

Table 18.4  Hypothetical example of industrial market gap calculation

2000–​2010
[1]‌ Space required to house new growth 25,600
(Net absorption)
[2]‌ + Replacement demand 82,500
(Depreciated stock)
[3]‌ –​ Available supply 7,300
(Excess vacant stock)
[4]‌ = Additional space needs 100,800
(Industrial market gap)

Source: Center for Real Estate and Urban Economics at University of California at Berkeley.

As in the case of the office-​market-​gap calculation, available industrial supply refers to the sum of the
existing vacant space in excess of some structural or normal vacant stock and the newly completed indus-
trial space. The analyst can use permitted projects and projects under construction to derive estimates of
expected additional supply of industrial space in the near term. In the case of industrial land, the notion of
the structural vacancy rate does not apply. Thus, the concept of available land supply simply refers to the
amount of improved land expected to be available and ready for industrial development over the period
of analysis.

Accounting techniques for forecasting industrial space demand

The fundamental accounting methodology for forecasting industrial space needs from employment or
output growth involves the application of average square-​feet-​per-​employee or square-​feet-​per-​unit-​of-​
output ratios to anticipated employment or output growth in the area under consideration. Given the
scarcity of output data, analysts typically focus on square-​feet-​per-​employee ratios and forecasts of employ-
ment growth. Ideally, it would be preferable to calculate such ratios by type of industry (three-​or four-​
digit NAICS code level) and size of firm. Of course, such analysis would be useful if the analyst can obtain
forecasts of employment growth by the same categorization. Given that forecasts by firm size are not typ-
ically available by vendors, analysts focus on square-​feet-​per-​employee variations across the sectors thought
to be the major contributors to speculative industrial space demand. Before applying square-​feet-​per-​
employee ratios to future employment growth, the analyst needs to evaluate whether such ratios need to
be adjusted, in order to reflect anticipated structural shifts, if any, that may impact space usage per employee
over the forecast period.
Depending on the type of space contemplated, a different methodology may be used for the assessment
of future industrial demand from growth. In particular, in the case of warehouse, distribution, and R&D
space, a more general and aggregate approach is typically used because of the general-​purpose nature of the
space provided. When trying to assess land or space requirements for production activities, a more disag-
gregate and location-​focused approach may be employed, referred to as “best-​fit” industry analysis, because
of the idiosyncratic location and space requirements of the different manufacturing industries. These two
methodologies are discussed in the following sections.

Method 1: Demand analysis for general-​purpose industrial space

Although this technique is often used for general-​purpose industrial space, such as warehouse, distribu-
tion, and R&D, it can also be used in estimating a market’s aggregate demand for manufacturing space,

403
 Industrial market analysis

if for any reason the “best-​fit” industry analysis cannot be used. In particular, this technique involves the
following steps:

1. Determine the economic sectors that are using industrial space (usually manufacturing, wholesale
trade, and distribution/​trucking).
2. Obtain data on employment in these sectors to estimate total employment that uses industrial space,
hereafter referred to as industrial employment.
3. Estimate average square feet per employee as the ratio of the employment estimate from Step 2 over
the occupied industrial stock.
4. Make required adjustments to the square-​feet-​per-​employee ratios estimated from Step 3 to reflect the
impact of anticipated structural shifts over the forecast period.
5. Obtain forecasts of employment growth in the sectors determined in Step 1.
6. Estimate industrial space needs by multiplying the final square-​feet-​per-​employee ratio by the forecast
of total industrial employment.

As indicated earlier, the estimation of square-​feet-​per-​employee ratios by sector, as opposed to an average

ratio for all sectors, is likely to lead to more accurate translation of employment growth forecasts by sector
to industrial space requirements. However, such a task is very difficult to accomplish and is not often
pursued by analysts at the macro level of analysis, because data categorizing a market’s occupied industrial
space by the NAICS code of the different tenants is not readily available. More refined calculations, using
square feet per employee by sector, would require the collection of primary data.

Method 2: Best-​fit industry analysis

This analysis methodology is typically used for the assessment of the demand for land acreage or space
needs for production activities and industrial parks because of considerable variations of location and space
requirements across the different industries. For this reason, best-​fit industry analysis represents a more dis-
aggregate and location-​oriented assessment of an area’s industrial demand prospects. In performing this ana-
lysis, three-​or four-​digit industrial classifications are typically used in order to capture important differences
in production processes and their implications in terms of space and location requirements. As McMahan
(1989) points out, firms classified within the same two-​digit industrial classification may have widely different
production processes and location requirements. Best-​fit industry analysis involves two broader steps:

1 . Identification of the best-​fit industries for the local market at the three-​or four-​digit NAICS level.
2. Evaluation of the area’s industrial development potential given the growth prospects of best-​fit industries.

We elaborate on each of these steps below.

STEP 1:  IDENTIFICATION OF BEST-​FIT INDUSTRIES  An area’s best-​ fit industries are defined as those
industries that have the most promising growth potential and are best positioned to benefit by locating
within the region and metropolitan area under consideration. The underlying premise is that best-​
fit industries stand to benefit by locating in the target metropolitan area because regional and local
attributes render their production and distribution processes cheaper, more efficient, and, hence, more
competitive. Thus, the identification of best-​fit industries involves an analysis of the match between
industry requirements that are largely determined by industry characteristics, and an area’s productive and
non-​productive amenities.
It is important in this analysis to recognize the two-​stage process of industrial location decisions. This
two-​stage process includes first the selection of a state or region and then the selection of the metropolitan

404
 Industrial space market

area. This is an important consideration because it suggests that, even if a metro area has strong locational
advantages that appear to fit very well with the locational requirements of a specific industry, it may not be
considered at all if the relevant state or regional attributes are unfavorable.The analytical implication of this
is that relevant unfavorable regional attributes should perhaps be more heavily weighted when evaluating
the fit of a specific industry. According to Barrett and Blair (1982), important factors for regional selection
include proximity to markets, quality and quantity of available labor, cost of natural resources used in pro-
duction process, and transportation linkages. SIR (1984) indicates that regions or general areas are chosen
on the basis of the most profitable combination of labor and transportation costs.
Industries that are already operating in the metropolitan area under consideration can provide the
starting point for the best-​fit industry analysis. Emphasis is placed on the type of companies being attracted
to the area, the state and prospects of these firms in terms of output growth, and the area’s specific com-
petitive advantages (firm and worker amenities) that appeal to them. Obviously, different industries may
place more or less emphasis on certain locational attributes depending on their characteristics. Within this
context, best-​fit industry analysis involves the examination and evaluation of two sets of data:  industry
characteristics and location/​area attributes.

Industry characteristics
• Nature of manufacturing activity
• Production process
• Composition of production costs
• Organizational structure
• Plant type: main or branch plant
• Spatial extent of operations: local, regional, national
• Utility servicing requirements
• Size of facility
• Product markets.

Location and area attributes

• Agglomeration economies
• Other firm/​productive amenities
• Factor availability and costs: labor, energy, and land costs
• Business environment: taxes—​corporate, inventory, and property—​and regulatory environment
• Infrastructure advantages: quality of interurban transportation network (freeways, airports, and rail
network)
• Regional location and geography, including access to markets
• Worker amenities
• Quality of school system and overall residential environment.

In evaluating the attractiveness of an area’s locational attributes for industrial development, it is useful to
have in mind the company classifications presented by Kinnard and Messner (1971). In particular, these
analysts classify industrial firms into the following five categories on the basis of the emphasis they place
on certain locational factors:

• Market-​oriented (gravitate towards the market for their final product)

• Resource-​oriented (gravitate towards resources used intensively in their production process)
• Transportation-​oriented (gravitate towards a certain type of transportation mode or facility)
• Labor-​oriented (location decision is heavily determined by labor costs and availability)
• Footloose (no particular factor is of primary importance).

405
 Industrial market analysis

Although such classifications may help identify broader industry groups that are most likely to be attracted
in the area, a more elaborate location analysis that examines the numerous factors that may affect industrial
location decisions needs to be carried out at a detailed industry classification level. To this end, a location
evaluation matrix, such as the one discussed in Box 18.1, can be used to evaluate the match between the
area’s location attributes and the location requirements of specific industries.
Such an evaluation is carried out using a rank/​score system and each industry’s fit is assessed by
summing up the scores assigned to the different attributes in an unweighted fashion. An econometric
location study, using hedonic regression techniques, can provide more accurate values for the weights
assigned by the different industries to the different location attributes. However, such an analysis is very
difficult to carry out because of serious data limitations.9 The accumulation of such data would be very
difficult because of the multitude of the three-​digit SIC level industries involved in the analysis and, most
importantly, the lack of readily available information on industrial land prices/​rents paid, or output per
worker produced, by the different industries at different locations. Such an endeavor would require the
collection of primary data.

A closer look at interurban industrial location  McMahan (1989) refers to a three-​stage industrial
location selection process, during which a region/​state, a metropolitan area, and a specific site is selected.
This section elaborates on the various productive and non-​productive location attributes/​amenities that
may be important to industrial firms when trying to choose among different metropolitan areas (not
locations within the same urban area). As indicated earlier, some of these locational factors are also relevant
during the first stage when a state or a region is selected.

Box 18.1 An example of “best-​fit” industry evaluation from an industrial

park study
In a case study of an industrial park near the city of Cheyenne Wyoming presented by Barrett and Blair (1982),
the following factors were used to sort out best-​fit industries:

• Environmental soundness
• Regional market compatibility
• Local market compatibility
• National growth trend
• Regional growth trend
• Labor market fit
• Production process and availability of required inputs
• Compatibility with local transportation system
• Utility requirements
• Value added to transport costs
• Local money market compatibility
• Intra-​industry stability
• Regional competitive position.

Barrett and Blair (1982) first identified 22 types of manufacturing operations that appeared to have potential to
be attracted to the area of study. Subsequently, they evaluated those industries with respect to the above list of
factors using a scoring/​ranking system in which 5 represents the maximum rank and 1 the lowest. The score/​
rankings were then added (without applying any weights) to derive the total score for each industry.

406
 Industrial space market

Productive amenities  Productive amenities, representing amenities that are beneficial to industrial
firms, may relate to any of the following five broader categories of factors:

• Agglomeration economies
• Production factor availability and costs
• Taxes
• Transportation costs
• Business environment.

Agglomeration economies  Mills and Hamilton (1989) refer to agglomeration economies as concentra-
tion advantages stemming from the large scale of an urban area. However, the magnitude and type of such
advantages vary across urban areas depending on their size and the make-​up of their economy. Furthermore,
the importance of agglomeration economies may vary across the different types of industrial firms. It has
been argued, for example, on the basis of the product-​cycle hypothesis, that new firms benefit more than
mature firms from agglomeration economies.The rationale behind this argument is that when firms mature
and move to the mass production stage, their focus shifts to maximizing economies of scale at large plants
using cheap land and labor, while agglomeration benefits become less important (Chapman and Walker,
1991).The implication of this is that an area that does not offer significant agglomeration economies may be
more attractive to large mature companies rather than smaller firms at the early stages of the product cycle.
The major types of agglomeration economies that industrial firms may enjoy within an urban area include
economies of scale in input sharing, labor market economies, and communication economies.
Both input suppliers and producers of final products may benefit by co-​locating through economies of
scale in input sharing: the former through internal scale economies; the latter through lower production costs
due to the cost advantages suppliers enjoy. Although, in general, a tendency of de-​agglomeration has been
observed, tendencies for outsourcing and vertical disintegration are said to increase the links between produc-
tion and service inputs, thus, perhaps, increasing the importance of these benefits.Vertically integrated firms
are typically larger with their own ancillary service functions. Recent trends for vertical disintegration and
subcontracting of certain functions represent efforts to minimize costs and cope with fluctuations in demand.
Concentration of industries may also contribute towards labor market economies in two ways. First, by
generating a large labor pool that helps lower labor search and information costs. Second, by providing an
expanded spectrum of job opportunities to workers, thereby reducing their re-​employment costs. This, in
turn, may allow firms to enjoy additional labor-​cost savings if workers are motivated to accept lower wages.
Not only labor and capital, but also information and ideas are important inputs into industrial production
processes. Co-​location of industries allows access to and exchange of business-​related information and ideas
that may not be easily accessible otherwise, thus creating the so called information/​communication economies.

Factor availability and costs  The availability and costs of factors of production, such as labor, energy, and
land, vary across urban areas and, as such, are quite relevant when evaluating industrial locations. A significant
consideration when assessing the fit of the area’s local labor pool with an industry’s labor requirements is the
importance of managerial vs. blue-​collar labor, measured by their shares as percentage of total workers. Empirical
results reported by Blackley (1985) point to a strong correlation between low-​skill labor and standardized pro-
duction processes. Furthermore, Schmenner et al. (1987) present some evidence suggesting that operations
using higher percentages of low-​skilled workers prefer locations with lower percentages of union membership.
In theory, larger, labor intensive facilities (e.g., textiles) should also be pursuing such locations.10
Within this context, the factors that are relevant when evaluating local labor market fit include:

• Labor wages
• Skill make-​up or occupational mix of local labor pool

407
 Industrial market analysis

• Educational level of population

• Unionization levels.

Energy costs should in theory influence the location decision of energy-​intensive industries. Energy con-
sumption is generally greater in the case of capital-​intensive plants and plants using line-​flow processes
(Schmenner et  al., 1987). Finally, commercial land cost considerations should enter industrial location
decisions when evaluating alternative metropolitan areas and, therefore, should be given due consideration
when evaluating local area fit with an industry’s requirements. Such considerations should matter more to
industries with larger land requirements.

Taxes  Another set of factors that may contribute to production cost variations across metropolitan
markets is tax rates. Although tax rates may be considered as part of the overall business environment
discussed below, they are addressed here separately because there is some special focus in the empirical
literature on this issue. Logically, all else being equal, low-​tax-​rate areas should be favored over high-​tax-​
rate areas. Wasylenko (1980) presents evidence of statistically significant influences of local property tax
differentials in the intra-​metropolitan relocation decisions of manufacturing firms that is consistent with
this hypothesis. Relevant tax rates from an industry point of view include:

• Corporate tax rate
• Business property taxes
• Inventory taxes
• Worker’s compensation.

Transportation costs  Transportation costs are especially important for warehouse and distribution facil-
ities as modern supply-​chain practices place less emphasis on storage and more on distribution. For example,
Just in Time (JIT) distribution techniques, which have been made possible by new information technolo-
gies and increasingly efficient and fast transportation systems, help reduce inventories. The importance of
accessibility to highways and air transport in warehouse location decisions is supported by empirical evi-
dence presented by Bowen Jr. (2008). In particular, Bowen Jr. (2008) found that the number of warehousing
establishments in 2005 and the growth in the number of warehousing establishments over the period 1998–​
2005 were strongly correlated with county-​level measures of accessibility in air and highway networks and,
to a lesser extent, rail networks.Van Den Heuvel et al. (2013) in an analysis of the location of logistics firms
in Netherlands found that logistics employment was growing faster in areas with at least one intermodal
container terminal than in areas without one, underlying the importance of transportation cost savings. Also,
Holl and Mariotti (2018) found that access to transportation infrastructure and proximity to urban areas
(especially Madrid and Barcelona) were important location determinants for logistics firms in Spain.
Transportation costs may also weigh heavily in the calculus of market-​oriented industries that tend to
locate close to the consumers of their final product. A firm’s transportation costs at a given metropolitan
area are influenced by the following factors:

• Regional positioning and access to markets

• Regional geography
• Quality, adequacy, and operational efficiency of transportation infrastructure networks: freeways, airports,
rail systems, ports.

Business environment  The local business environment can also have an impact on an industry’s costs
depending on the nature, intensity, and restrictiveness of local regulations, as well as the extent to which spe-
cial economic and industrial development policies/​incentives are in place. In the past, industrial/​economic

408
 Industrial space market

development policies have placed emphasis on loans for land development and tax incentives, but through
time the emphasis has shifted towards longer-​term policies aiming at improving the area’s amenities and
education infrastructure.

Non-​productive amenities  As indicated earlier, non-​productive amenities refer to worker amenities, that
is, amenities that may appeal to a company’s workers. Although such amenities do not directly affect an
industry’s production costs, there is increasing evidence of their indirect influence on such costs through
their effect on labor costs and availability. Amenities, such as a high-​quality residential environment, a good
local education system, a sufficient and efficient level of public services, low crime rates, entertainment,
cultural and shopping opportunities, as well as any other area amenities that contribute to a better quality
of life, may be especially important for high-​technology firms, which have been shown to be very location
selective (Sivitanidou and Sivitanides, 1995b).

STEP 2: EVALUATION OF THE AREA’S INDUSTRIAL DEVELOPMENT POTENTIAL  Once the best-​fit industries
have been identified, the area’s industrial development potential can be assessed by translating projected
employment growth in these industries to land acreage and space requirements. Based on the application
of this technique by McMahan (1989), the specific steps involved in generating forecasts of industrial land
and space absorption include:

a) Generate or purchase forecasts of employment growth at the three-​or four-​digit NAICS code level. If
forecasts at this level of disaggregation cannot be obtained then forecasts at a more aggregate level for
relevant employment categories can still provide the basis for calculating future employment growth
in target industries.
b) Calculate future growth in target industries using total or sectoral employment growth forecasts and
current target-​industry shares, if forecasts of employment growth in the specific industries analyzed
cannot be obtained (see example in Table 18.3).
c) Collect primary data on employment densities (square feet per employee or employees per acre) in the
target industries identified in the first step within the urban area under consideration. Subsequently, gen-
erate projections of employment densities in target industries by adjusting current ratios to reflect anticipated
industry structural shifts that may have an impact on the amount of land or square footage utilized per
worker. An example of structural shifts that may be impacting employee densities is the wholesale and dis-
tribution sector where the re-​engineering of supply practices, and inventory management techniques (in
response to global competition and rapid technology advances), are changing the scope of warehousing
and distribution operations. See the previous discussion on square feet per employee for more on this.
d) Estimate the required industrial land and space by applying the employment density estimates to the
projected employment growth in target industries. Adjust for vacant land utilized for other-​than-​
building purposes, as well as for land held for expansion.

Econometric modeling of the industrial market

The econometric modeling of the industrial market has not been explored in the empirical literature
as much as in the case of the office market. A major reason for this is that industrial market modeling is
hindered to a significant extent by the lack of sufficiently long time-​series data. The scarcity of disaggre-
gate data is much greater, especially when it comes to distinguishing between the non-​speculative and
speculative components of the industrial market. A first modeling effort presented by Wheaton and Torto
(1990) draws from the macroeconomic investment literature in an attempt to address the largely non-​
speculative character of industrial space. DiPasquale and Wheaton (1996) presented another modeling
approach that draws from the office market literature and describes the functioning of the market for

409
 Industrial market analysis

speculative industrial space. Finally, Thompson and Tsolacos (2000) presented a similar modeling approach
using a simultaneous equation system.

Modeling investment in non-​speculative industrial buildings

Wheaton and Torto (1990) presented a modeling approach that draws from the macroeconomic invest-
ment literature and focuses on modeling investment in non-​speculative industrial buildings.This modeling
framework postulates that at any point in time, t, given the level of desired industrial output, Q, and the
after-​tax cost of corporate capital, c, there is a desired level of industrial stock, K*(Q, c). This desired stock
level implies a level of desired investment in new space, I*, which represents the square footage needed to
be added to the existing stock to achieve K*. Paralleling the notion of desired net absorption of industrial
space, this desired investment, I*, is not realized in a single period but in multiple periods at rate g. A similar
partial adjustment modeling framework is also assumed for I*:

I*t = g (K*t(Q, c) –​Kt–​1) (18.3)

Industrial space completions, I, which represent the delivery of the desired investment, I*, or the desired
additional space, do not occur instantaneously but with some delay (which is reflected in the coefficient
a in the equation below) because of the planning and construction time required to turn an investment
decision into an actual building. In this sense, It can be modeled as a partial adjustment process described
by (18.4) below.11

It –​ It–​1 = a (I*t –​ I t–​1) (18.4)

Consequently, from (18.3) and (18.4) an empirical equation for industrial construction investment can be
derived as:

It = (1–​a) It–​1 + ag (K*t(Q, c) –​Kt–​1) (18.5)

The specification described by (18.5) does not incorporate any rental or vacancy influences because such
concepts are not directly relevant in a non-​speculative market. Note that econometric estimation of (18.5)
can help predict future industrial completions (I) and desired industrial space at the margin (I*), which
can provide valuable insights with respect to industrial land and space development needs. The specific empirical
equation estimated by Wheaton and Torto (1990) is described by (18.6).12

I=
t b0 + b1MEt −n + b2WEt −n + b3ct −n − b4K t −1 + b5I t −1 (18.6)

where
I : annual or semi-​annual industrial completions
ME : manufacturing employment
WE : wholesale employment
K : existing industrial stock
c : cost of capital
n : time lag

Once the coefficients of (18.6) are estimated, they can be used according to Wheaton and Torto (1990) to
estimate the desired industrial space, I*, for time, t, for given values of MEt-​n and WEt-​n using the following
equation:13

410
 Industrial space market

b4  b0 b1 b2 b3 
I t* =  + MEt −n + WEt −n + ct −n − K t −1  (18.7)
(1 − b5 )  b4 b4 b4 b4 

Due to the lack of separate time series for the speculative and non-​speculative components of the indus-
trial market, the dependent variable represents the sum of all industrial space completions (speculative and
non-​speculative) in the 52 industrial markets included in the sample used by these analysts.The conclusions
from the Wheaton and Torto (1990) analysis using the investment approach are the following:

• Demand for industrial capital is influenced by changes in manufacturing employment or output with a
two-​year lag (n in the above equations).
• A similar lag intervenes between changes in the after-​tax cost of capital and changes in the desired
industrial capital.
• Demand for industrial capital appears to respond very rapidly to changes in wholesale employment.
• Industrial space construction seems to begin soon after new industrial capital is demanded, but there is
a phase-​in period as a result of which actual delivery of the new space may take several years.

Modeling the speculative industrial market

The three-​equation system examined in the case of the office market can also be used for modeling the
speculative industrial market. However, the independent variables included in the three industrial market
equations, especially in the demand equation, may be different from those included in the respective office
equations. One difficulty in modeling the demand for speculative industrial space at the metropolitan level
is the lack of historical output series at this geographical level. One way to develop rough output estimates
at this level is to use metropolitan-​specific historical employment series with national productivity figures
that can be obtained from the Survey of Current Business published by the U.S. Department of Commerce.
Great caution is needed when using productivity figures to estimate output because the former varies
considerably across different industries. Thus, an estimate of an output series using total employment and
average productivity for all sectors could be very misleading. Such an approach is bound to produce con-
siderably more accurate estimates if it is applied to each of the detailed NAICS industry classifications.
DiPasquale and Wheaton (1996) have presented an attempt to model together the speculative and non-​
speculative markets for industrial space in Philadelphia using an approach similar to the one discussed for
office space.This modeling approach seems to work, suggesting that the two markets (speculative and non-​
speculative) may be integrated to a significant extent. The link between the two markets may be the rental
rate in the speculative industrial market in the sense that when this rate becomes too low, companies may
be more inclined to rent industrial space rather than owing it.
The industrial market model estimated by DiPasquale and Wheaton (1996) consists of a demand, a
supply, and a rent equation. The underlying modeling framework for all three equations is that movements
of the aggregate industrial market represent adjustments towards a desired or equilibrium level of occupied
industrial stock, OC*, completions, C*, and rental rate, R*. Due to the inefficiencies characterizing the
industrial market, which are similar to those hampering the office market, such adjustments are sluggish.
Thus, only a fraction of the difference between prevailing and desired occupied stock, completions, and
rent is eliminated during each period. The three equations along with the two identities required for gen-
erating the industrial market forecast are presented in Box 18.2.
DiPasquale and Wheaton’s (1996) econometric analysis of the determinants of aggregate industrial
absorption (sum of absorption in all industrial sub-​types) in Philadelphia indicates that it responds rather
rapidly to changes in employment and output. In particular, their findings show that there is a contem-
poraneous relationship between changes in industrial space demand and changes in output per worker,

411
 Industrial market analysis

manufacturing employment, and wholesale employment. Furthermore, it appears that increases in output
per worker trigger much less than proportionate increases in demand for industrial space and that the rental
elasticity of space demand is low (about –​0.08).14

Box 18.2 The DiPasquale-​Wheaton (1996) industrial market model

Net absorption: AB = f[ME(t),WE(t), Q(t), R(t–​1), OC(t–​1)] (18.8)
Derived from:
AB = g [OC*(t)–​OC(t–​1)]
OC*(t) = f[ME(t),WE(t),Q(t), R(t–​1)]

where:
AB : net absorption
g : adjustment coefficient
OC* : long-​run desired occupied stock
ME : manufacturing employment
WE : warehouse employment
Q : industrial output per worker
R : industrial rent

New Supply: C(t) = f[R(t–​n), AB(t–​n), C(t–​1)] (18.9)

Derived from:
C(t)–​C(t–​1) = θ [C*(t)–​C(t–​1)]
C*(t) = f[R(t–​n), AB(t–​n)]

where:
C : industrial space completions
θ : adjustment coefficient
C* : desired new stock
n : time lag

Rental adjustment: R(t)–​R(t–​1) = f[V(t–​n), ABN(t), R(t–​1)] (18.10)

Derived from:
R(t)–​R(t–​1) = a [R*(t)–​R(t-​1)]
R*(t) = f[V(t–​n), ABN(t)]

where:
R : industrial rent
a : adjustment coefficient
ABN : net absorption normalized with stock level
V : industrial vacancy rate

Identities
Stock-flow : S(t) = S(t–​1)(1–​d) + C(t) (18.11)
Vacancy rate : V(t) = [V(t–​1) S(t–​1)–​AB(t)+C(t)] /​S(t) (18.12)

412
 Industrial space market

Thompson and Tsolacos (2000) also developed a forecasting model of the UK industrial market. The
model consists of three structural equations which are estimated simultaneously. The dependent variables
in these three equations are the new industrial space supply (NIBSUP), the industrial rent (RENT), and
the availability of industrial floor space (AVFS). The estimated equations and the independent variables
included in each are described below:

NIBSUPt = a0 + a1RENTt + a2CCt (18.13)

RENTt = b0 + b1RENTt-​1 + b2AVFSt (18.14)

AVFSt = c0 + c1GDPt + c2GDPt-​1 + c3NIBSUPt (18.15)

In the above system, the exogenous variables include from the supply side the construction cost (CCt) and from
the demand side the GDP.The above modeling approach postulates in particular that new industrial space con-
struction is a function of the prevailing industrial rent and construction costs. The industrial rent is modeled
according to the standard rent adjustment equation that includes the rent of the previous period and a measure
of the imbalance between demand and supply, the vacant industrial stock (AVFSt). Finally, the vacant industrial
stock is modeled as a function of GDP (capturing the strength of demand for industrial space) and new indus-
trial space supply. Taking into account potential lag effects between GDP growth and the actual realization of
growth in industrial space demand, the equation includes the GDP of both the current and the previous period.

Site-​specific analysis of industrial projects

As is the case in other property types, the ultimate objective of site-​specific analysis for speculative indus-
trial development is to assess its marketability and income-​earning capacity. To this aim, the following four
broader analytical steps need to be carried out:

1 . Evaluation of site attributes.

2. Identification of specific types of tenants most likely to be attracted to the development given its
attributes.
3. Analysis of the competition and assessment of the project’s competitive position.
4. Estimation of project capture and achievable rents and or land prices.

Step 1: Evaluation of site attributes

In analyzing a site for industrial development, the analyst needs to focus on the three broader categories of
attributes that largely determine the marketability and competitive strength of industrial space.These include:

• Site features
• Site access and linkages
• Community constraints.

Site features
In analyzing site features, the focus is on their suitability and the constraints they may pose with respect to
the industrial development under consideration. According to SIR (1984), the most relevant site features
from an industrial-​development point of view include:

• Parcel size, shape, topography, and geology.

• Availability of utility servicing, capacity, and costs (not only important for traditional manufacturing, but
also for modern high-​tech plants requiring high voltage lines).

413
 Industrial market analysis

• Availability of land for expansion (especially important for land-​

intensive manufacturing
establishments).
• Acquisition and land preparation costs.

Site access and linkages

The underlying premise of site and location analysis is that industrial firms across all categories (production,
warehouse, distribution, and research and development) have idiosyncratic operational requirements and
are largely cost sensitive. Therefore, location attributes that matter are those likely to facilitate production
processes and contribute towards the minimization of production costs. Cost components that may be
more important than others can be uncovered by an analysis of the cost composition of the different types
of operations that use industrial space. The relative importance of the costs of the different inputs along
with other idiosyncratic functional requirements will determine the importance of the different locational
attributes for each industry’s facilities.

Traditional manufacturing (production) space

The location attributes that matter in the case of manufacturing or production operations include:

• Access to other firms (business and producer services) to reduce the cost of interfirm contacts. These
may be less important than in the past but may still be relatively important in the case of young, small,
and non-​vertically integrated companies.
• Good access to labor (managerial and blue collar) and amenable communities to save on labor costs,
which are a major cost component of manufacturing operations.15 Residential amenities, such as
good schools, low crime rates, and access to shopping opportunities, were found to be valued by
users of manufacturing space in the Los Angeles metropolitan area (Sivitanidou and Sivitanides,
1995a).
• Access to thoroughfares, interstate highways/​freeways, and airports to facilitate business trips and reduce
shipment costs.
• At the community level, low tax rates, land availability, and limited regulatory constraints were found to
matter in intra-​metropolitan industrial location decisions.16

Warehouse and distribution space

Recent technological advancements and increasing use of information technologies and modern inventory
management techniques, such as Just in Time (JIT), have reduced safety stock and lead time requirements
from order to distribution. As a result, warehousing functions are shifting away from storage and towards
distribution. Warehouse and distribution functions follow a hierarchical function in the sense that products
are shipped from the manufacturer to larger national or regional warehouses and from there to smaller local
warehouses. In recent years, increasing storage, transportation, and distribution efficiencies have enabled
companies to close smaller local warehousing operations and consolidate them into larger facilities serving
larger areas. Within this context, the following location attributes are important for warehousing and dis-
tribution operations:

• Access to thoroughfares, interstate freeways, airports, rail systems, and ports, because transportation costs
are a major component of the costs associated with warehousing and distribution operations. Access to
multiple nodes and alternative means of transportation adds to managers’ power to negotiate lower rates

414
 Industrial space market

for a given transportation venue (ship vs. train, for example). Access to ports may be especially important
for larger distribution centers.
• Access to labor pools and, especially, to blue-​collar labor, which is an important input into not only
traditional warehouse and distribution functions (see Table 18.4), but also some value-​added functions
(such as labeling, barcoding, packaging, and simple assembly) that are increasingly being performed at
warehouse and distribution facilities.
• Access to amenable communities because it appeals to the preferences of warehouse managers who
often play an important role in the facility’s ultimate location selection.
• Access to high concentrations of consumers (consumption centers). JIT distribution processes favor such
access, as well as distribution processes involving products that are shipped frequently in small orders.
• Access to production centers and manufacturing establishment concentrations.
• Low-​tax-​rate communities (inventory taxes may be of special importance).

A hedonic analysis of warehouse facility rents in the Los Angeles metropolitan area shows that good access
to blue-​collar labor pools, amenable residential communities, freeways, freeway intersections, airports, con-
sumer markets, and manufacturing plant concentrations are valued by warehouse-​and distribution-​facility
operators (Sivitanidou, 1996). The importance of these location attributes to warehouse and distribution
operations is evidenced in the statistically significant rent premiums they were found to command in the
marketplace by this study.

Research and development space

A recent trend that is relevant to the location of research and development operations is the location
decoupling of high-​tech company functions, such as decision, production, research and development, and
sales and marketing, to different metropolitan areas depending on the function performed and those areas’
competitive advantages. This decoupling may have some important linkage implications for research and
development space, such as the need for access to headquarters. The research orientation of these activities
also has important implications with respect to their labor requirements and access to institutions that can
facilitate such functions. As Sivitanidou and Sivitanides (1995b) have demonstrated, the following location
attributes are of importance to research and development firms at the intra-​metropolitan level.

• Access to high-​skill labor and highly amenable residential communities

• Access to local and larger research universities
• Transportation, freeway, and airport access
• Access to headquarters
• Access to high-​tech production centers.

Community constraints
Many communities are not very receptive to the idea of industrial development due to potentially negative
effects on neighboring uses and on the whole community for that matter, in the case of large industrial
plants. For this reason, the analyst needs to identify community constraints and their potential implications
with respect to the industrial development potential of the specific site. The relevant factors that need to
be evaluated include:

• Zoning regulations and development controls

• Environmental regulations
• Local government and community attitudes towards industrial development.

415
 Industrial market analysis

Sivitanidou and Sivitanides (1995a) have demonstrated that jurisdictional zoning and development
controls can constrain the local industrial land market, creating issues of site availability and inflating
rents/prices.

Step 2: Identification of types of tenants most likely to be attracted to the site

The underlying objective of this stage of analysis is to identify the most likely tenants to be attracted to
the site under consideration. In the case of best-​fit industry analysis, this stage involves refinement of the
analysis completed at the macro level by factoring in site and project-​specific characteristics and identifying
the subset of industries most likely to be attracted to the subject site. To this aim, the analyst needs to gain
an understanding of which site and community characteristics may be appealing to certain industries and
constraining or undesirable to others. This stage of analysis therefore requires:

1. Evaluation of the “match” between the location requirements of the best-​fit industries (identified at
the macro-​analysis stage) and site, development, and community characteristics.
2. Refinement of the list of target industries by identifying companies occupying space at comparable
industrial developments and by interviewing local industrial brokers regarding companies looking for
space similar to the one provided by the project.

Obviously, Step 2 applies in broader terms also when trying to identify most likely tenants in the case of
general-​purpose industrial space (warehouse, distribution, and R&D). Although such analysis may not
include a detailed best-​fit industry analysis, it may elaborate on the more specific segments of broader
industrial-​space-​using sectors that are more likely to be attracted to the site.

Step 3: Analysis of the competition

The bottom-​line objective of this step is to evaluate the project’s advantages and disadvantages, relative to
those of competing industrial developments, and come up with an assessment of its competitive position
in the marketplace (such as average, 10% above average, etc.). In most cases, analysts will need to use the
competitive differentials technique to quantify the project’s competitive position, since it is very unlikely
that sufficient data will be available for the application of hedonic regression analysis.
Within this context, analysis of competing industrial developments focuses on three broader set of
characteristics. The first set includes project features such as size, coverage, occupancy, parking availability,
etc. The second set includes structure attributes, such as typology, functionality (including key technological
advantages or disadvantages), mode of tenancy (single tenant, multi-​tenant), adaptability of use (general pur-
pose or limited purpose), size, number of stories, ceiling heights, office buildout, column spacing, building
depth, number of docks and ground level doors, quality of the structure, etc. The third set includes location
advantages and constraints as they particularly relate to the type of operations expected to be performed at
the planned or existing facility.
Some broader submarket characteristics that may be of importance in this analysis include major trans-
portation corridors and nodes, and the overall nature, quality, functionality, and scale of existing industrial
space. For a more thorough coverage of the location attributes to be examined in this stage, the analyst
should refer to the framework discussed in Step 1.
In an example of industrial market analysis, presented later in this section, information on the following
list of attributes was collected for competing industrial park developments (McMahan 1989):

• Total planned acreage

• Sold/​used

416
 Industrial space market

• Vacant
• Total acreage absorbed
• Average annual absorption
• Average price per acre
• Transportation (close-​by access to highway and/​or railroad)
• Availability of utilities (water, electricity, gas)
• Coverage
• Parking (sq. ft.)
• Terms of transaction (sale, build to suit, lease)
• Overall project rating (excellent, very good, etc.).

The above list is certainly not exhaustive.Additional information that can be collected, if applicable, includes
structure attributes, type and identity of tenants/​buyers, contract rents, lease terms, etc. As McMahan (1989)
suggests, the scope of the competitive survey needs to be consistent with the scope of the project. For
example, if the development is directed towards regional or national markets, projects of the same caliber
should be analyzed.

Step 4: Estimation of project absorption schedule and achievable rents or land prices

This stage of analysis focuses on two primary objectives. The first objective is to estimate the share of the
metro-​wide or submarket demand likely to be captured by the subject site. The second objective is to esti-
mate the rent or land price the project will most likely command in the local marketplace, using as basis its
advantages relative to those of competing developments. At this point, the analyst must have from Step 2 of
the macroeconomic analysis the total industrial land or space expected to be absorbed in the overall market
area. Using this estimate as a framework of reference, the project’s absorption schedule can be determined
if an estimate of the project’s market share is available. Within this context, the broader steps involved in
assessing industrial project absorption and achievable land prices or space rents include:

a) Estimate the share of incremental demand likely to be captured by the project under consideration based
on the assessment of its competitive strength. In particular, the project’s fair market share (see Chapter 9
for an explanation of how this share is calculated) can be used as a starting point and then adjusted
accordingly by applying the estimated competitive position index from the previous step. For example,
if the competitive position index suggests that the project is 10% more superior than the average com-
peting project then its capture rate can be calculated as its fair market share times 1.10. Comparison
of the estimated absorption schedule with actual recent experience of similar projects in the market
can help evaluate the reasonableness of the forecast. If the analyst has absorption rate data and project
attributes for several industrial properties, it would be advisable to use econometric techniques to
quantify the effect of the different project attributes on absorption rates and use those estimates as the
basis for predicting the project’s absorption/​capture rate.
b) Estimate achievable space and/​or land rents using the competitive differentials technique or prefer-
ably the hedonic regression technique if the required data are available. In the case of the competitive
differentials technique, the project’s competitive position index can be applied to the average rental rate
or price commanded by comparable developments in its market area. If the analyst has sufficient data,
the development’s rental rate or land price should be estimated via hedonic regression techniques as:

Land Price/​Rent = f (structure/​project attributes, location attributes,

        tenant characteristics, lease characteristics)

417
 Industrial market analysis

Table 18.5 Target industries for a potential industrial park*

NAICS code Target industries Paid employees

2007 2012 2016

3341 Computer and peripheral equipment manufacturing 668 357 527
3342 Communications equipment manufacturing 6,849 4,563 3,629
3344 Semiconductor and other electronic component manufacturing 20,656 7,675 13,234
3345 Navigational, measuring, electromedical, and control instruments 11,658 16,577 15,424
manufacturing
3346 Manufacturing and reproducing magnetic and optical media 486 441 306
3371 Household and institutional furniture and kitchen cabinet 5,818 4,606 4,991
manufacturing
3372 Office furniture (including fixtures) manufacturing 4,076 2,878 3,176
3379 Other furniture related product manufacturing 2,072 1,413 1,278
5111 Newspaper, periodical, book, and directory publishers 16,294 6,369 4,329
5112 Software publishers 9,304 9,465 15,688
Totals 77,881 54,344 62,582
Total manufacturing employment 274,800 223,641 240,025
Percentage in target industries 28.3% 24.3% 26.1%

* Modified by the Editors from McMahan, J. 1989. Property Development. New York: McGraw-​Hill Publishing
Company.
Source: Census Bureau, County Business Patterns.

Ambrose (1990) showed that industrial property characteristics and lease terms yield statistically significant
differences in property-​specific rents, underscoring the importance of using hedonic regression techniques
in accurately estimating achievable rents for specific industrial properties.
Tables 18.5 to 18.7 present an example of industrial market analysis with mostly hypothetical numbers,
modified by the editors from McMahan (1989).Table 18.5 presents the list of the four-​digit NAICS target
industries that have been found to have promising prospects for development in the area of the study and
actual employment data for the years indicated in the table.
Table  18.6 presents the target-​industry absorption trends and projections at the metro-​area level as
derived from total manufacturing employment and the use of an average ratio of employees per acre based
on the 2001 SCAG study. For better understanding of the calculations presented in this table, the formulas
used for the 2020 projections are presented below:

Target industry employment 2020 = Total manufacturing employment * Percentage in target

               industries = 249,771 * 0.26 = 64,940

Land utilized = Target industry employment /​Employees per acre

      = 69,940 /​24.5 = 2,651

otal land absorbed = Land utilized /​(1 –​percentage land held as vacant) =

T
        = 2,651 /​(1 –​0.15) = 3,118

Absorption for period = Total land absorbed in 2020 –​Total land absorbed in 2016
         = 3,118 –​3,005 = 113

418
 Industrial space market

Table 18.6 Target-​industry land absorption, metro area

Actual Projected

2007 2012 2016 2020 2025

Total manufacturingemployment 274,800 223,641 240,025 249,771 262,512
Annualized percentage change –​4.0% 1.4% 1.0% 1.0%
Percentage in target industries 28.3% 24.3% 26.1% 26.0% 26.0%
Target industry employment 77,881 54,344 62,582 64,940 68,253
Employees/​acre* 24.5 24.5 24.5 24.5 24.5
Land utilized (acres) 3,179 2,218 2,554 2,651 2,786
Percentage of land held as vacant** 15.0% 20.0% 15.0% 15.0% 15.0%
Total land absorbed 3,740 2,773 3,005 3,118 3,277
Absorption for period –967 232 113 159

*  Average of Light Manufacturing and Heavy Manufacturing employees per acre per the 2001 SCAG study.
Modified by the Editors from McMahan, J. 1989. Property Development. New York: McGraw-​Hill Publishing
Company.
*  All numbers are completely hypothetical.

Table 18.7  Demand on subject site (in acres)

2016–​2020 2020–​2025 Total

(1) Absorption for period 232 113 345.7
(2) Capture by subject site 0.12 0.15
(3) Subject site absorption 27.9 17.0 44.9
(4) Average annual absorption 7.0 3.4

Modified by the Editors from McMahan, J. 1989. Property Development. New York: McGraw-​Hill Publishing
Company.

Finally, Table 18.7 presents the calculation of the site’s expected absorption by applying a hypothesized
capture rate to the predicted industrial land absorption in the metropolitan area. It is important to note
that any estimate of project market share needs to take into account competing industrial land/​space
supply over the forecast period.

On industrial market studies: evaluation criteria

Some key criteria for evaluating industrial market studies for each level of analysis (macro and micro) are
expressed below in the form of questions.

Macro, non-​site-​specific analysis

Is the industrial development potential within the competitive area properly analyzed?
Are target, or best-​fit, industries properly identified and are their prospects carefully assessed?
If employee density ratios have been used to quantify future demand, have they been carefully evaluated
for their reasonableness based on primary data pertaining specifically to the specific target industries and
urban area under consideration?

419
 Industrial market analysis

Site-​specific analysis of development projects

Are location requirements and constraints, as they pertain to the area’s best-​fit industries at the given site,
carefully identified and evaluated?
Are forecasts of project rents/​prices produced through sound techniques? Do such rent/​price estimates
reflect anticipated changes in industrial demand and supply conditions both locally and at the urban
area level?
Does the project land/​space absorption schedule take into account the competitive strength of the proposed
project, as well as anticipated future changes in industrial market conditions (both demand and supply)?

Chapter summary
In this chapter, we reviewed the major features of industrial markets and commonly used market analysis
techniques at the macro and micro level. Industrial markets are in general very idiosyncratic due to their
functional heterogeneity, their large non-​speculative component, and the importance of the macroeco-
nomic cycles. In addition, demand is driven by expected output and/​or employment, the cost of corporate
capital, and technology-​induced functional and locational obsolescence.
In theory, the supply of non-​speculative space is driven by factors shaping business investment decisions,
while the supply of speculative space is driven by the usual factors that trigger the supply of commercial
real estate. Empirical evidence suggests that the two segments (speculative and non-​speculative) may be
linked to a significant extent. The most likely link between the two segments is the lease rate in the specu-
lative segment, as lower rents may discourage demand and construction of new owner-​occupied space.
At the macroeconomic level of analysis, both accounting and econometric techniques are applicable
to industrial markets. In the case of general-​purpose warehouse, distribution, and R&D space, accounting
techniques often focus on the identification of broader industrial-​ space-​using sectors, their growth
prospects, and their space requirements. If land and space needs for production space and industrial parks
are of interest then accounting techniques focus on the identification of the area’s best-​fit industries, pref-
erably at the three-​or four-​digit NAICS level, and the assessment of their development potential, employ-
ment growth, and space/​land requirements.
Econometric techniques focus on the behavior of the area’s aggregate industrial market and the quan-
tification of three sets of relationships.These relationships include the association between the area’s aggre-
gate employment/​output and industrial space absorption, controlling for the effect of rental rates; the effect
of rents and other market strength indicators, such as absorption, on speculative industrial construction; and
the feedback effects of market tightness on industrial space rents.
Focusing on the analysis of the project’s attributes and competitive strength, micro/​marketability ana-
lysis for industrial space or land development involves:

• Analysis of the appropriateness of a given site for developing industrial space or preparing land for a
subset of the area’s best-​fit industries (identified in the macro analysis). Such an analysis should account
for the full range of location attributes that are relevant for the particular type of industrial development
under consideration.
• Analysis of the competition and assessment of the competitive position of the project within the
marketplace.
• Estimation of the site’s capture rate and absorption schedule, taking into account the project’s competi-
tive position.
• Estimation of achievable industrial space and/​or land rents, preferably through hedonic methodologies,
if sufficient data are available, or competitive differentials techniques otherwise.

420
 Industrial space market

Questions
1. Discuss briefly the main idiosyncrasies of the industrial market and their implications for market
analysis.
2. Discuss the most likely drivers of the demand for industrial space.
3. Describe the various techniques of macroeconomic analysis that can be used to evaluate industrial
development potential within a metropolitan area. What are the advantages of each approach?
4. Describe how the market for speculative industrial space can be modeled. Discuss the main variables
modeled and the behavioral mechanism underlying the model’s equations.
5. Discuss the basic steps of site-​specific analysis for industrial development projects.
6. What are the differences in the locational requirements and access preferences of the different types of
industrial tenants?
7. Discuss the recent trends in the logistics industry and how they may be affecting the location
requirements of warehouse and distribution facilities.

Notes
1 SIC codes have been replaced by NAICS codes.
2 Productivity, measured as output per worker, is thought to vary little across areas for detailed industrial
classifications.
3 The initial industrial space demand model that was estimated by Anderson and Guirquis (2011) utilizing the
Kalman Filter approach (in order to allow the regression parameters to vary with time) was stated as follows:

NAt = β0 + β1 NAt-​1 + β2 NAt-​3 + β3 IMOt-​2 + β4 IMOt-​3 + β5 ISMPMIt-​1 + β6 ISMPMIt-​4

In the above equation, NA represents net industrial space absorption.

4 If, for example, RLVres represents the residual land value for residential uses and RLVind represents the residual land
value of existing industrial uses, the industrial use will be taken over by residential if:

RLVres –​ RLVind ≥ redevelopment costs

5 Added by the Editors.

6 An inventory series could not be used as a proxy for the output in the wholesale sector due to the lack of
metro-​specific  data.
7 An estimated manufacturing output series for the 52 metropolitan areas included in the analysis sample was not
found to contribute statistically to movements in aggregate industrial completions. This might have been the
result of the failure of the estimated series to reflect true movements in industrial output in the metropolitan areas
included in the sample rather than the lack of a relationship between expected or current output and industrial
completions. Please see the discussion on modeling the speculative industrial space for an explanation of why this
may be the case.
8 Added by the editor.
9 The different weights assigned by a specific industry to different locational attributes can be estimated through
cross-​sectional data on industrial rents or output-​per-​worker produced by tenants in the industry under consider-
ation at different locations, and the characteristics of those locations.
10 Shmenner, Huber, and Coor (1987) find weak evidence that plant size (and therefore) labor intensity matters when
companies select states for their manufacturing plants. This finding, however, does not necessarily mean that labor
unionization issues do not matter when evaluating industrial location at the metropolitan level.
11 The reader needs to keep in mind that since we are referring to non-​speculative space, that is, space owned and
occupied by the user, the concepts of absorption, completions, and investment are interchangeable.
12 In order to estimate I*, the underlying coefficient structure of the statistical equation (18.6) has to be determined
first. Using (18.5) as basis, (18.6) can be rewritten as:

I t = a [ g (a0 + a1 MEt − n + a2 WEt − n + a3 ct − n ) − g K t −1 ] + (1 − a ) I t −1

421
 Industrial market analysis

The above implies the following mathematical relationships between the coefficients of the above equation and
equation (8.16).

ag = b4 ; (1 − a ) = b5 ; a = (1 − b5 ); g = b4 / (1 − b5 ); aga0 = b0 ; a0 = b0 / b4 ; a1 = b1 / b4 ;
a2 = b2 / b4 ; a3 = b3 / b4

13 This is derived based on the basis of the coefficient structure described in note 14.
14 The general form of the industrial absorption equation estimated by DiPasquale and Wheaton (1996) is described
by (18.8) in Box 18.2. The corresponding statistical equation is:

ABt = b0 + b1MEt + b2WEt + b3Qt + b4 Rt −1 + gOCt −1

The parameter estimates of this equation can be used to calculate the long-​run square feet per employee ratio
for manufacturing and warehouse space as b1/​g and b2/​g, respectively.
15 Cheng and Stough (2006) and Luker (1998) confirmed the negative effect of labor wages on manufacturing
location.
16 Zoning regulations and growth moratoria were found to create a constrained land-​market environment in the
manufacturing-​space market in the Los Angeles metropolitan area (Sivitanidou and Sivitanides, 1995).

References and additional readings

Ambrose, B. W. 1990. An Analysis of the Factors Affecting Light Industrial Property Valuation. Journal of Real Estate
Research, 5:3, 355–​370.
Anderson, I. R. and H. Guriguis. 2011. Industrial Space Demand. Herndon,VA: NAIOP Research Foundation.
Barrett, G. V. and J. Blair. 1982. How to Conduct and Analyze Real Estate Market and Feasibility Studies. New York: Van
Nostrand Reinhold Company.
Blackley, P. 1985. The Demand for Industrial Sites in a Metropolitan Area: Theory, Empirical Evidence and Policy
Implications. Journal of Urban Economics, 17:2, 247–​261.
Bowen Jr., J. T. 2008. Moving Places: The Geography of Warehousing in the US. Journal of Transport Geography, 16:6,
379–​387.
Chapman, K. and D. Walker. 1991. Industrial Location. Cambridge, MA: Basil Blackwell.
Cheng, S. and R. R. Stough. 2006. Location Decisions of Japanese New Manufacturing Plants in China: A Discrete-​
Choice Analysis. Annals of Regional Science, 40:2, 369–​387.
DiPasquale, D. and W. Wheaton. 1996. Econometric Analysis of Metropolitan Office and Industrial Markets. Chap. 12
in Urban Economics and Real Estate Markets. Englewood Cliffs, NJ: Prentice Hall.
Holl, A. and I. Mariotti. 2018. The Geography of Logistics Firm Location: The Role of Accessibility. Networks and
Spatial Economics, 18:2, 337–​361.
Homes and Communities Agency. 2015. Employment Density Guide, 3rd Edition. London, UK: Homes and Communities
Agency.
Ibanez, M. and A. Pennington-​Cross. 2013. Commercial Property Rent Dynamics in U.S. Metropolitan Areas:  An
Examination of Office, Industrial, Flex and Retail Space. Journal of Real Estate Finance and Economics, 46:2, 232–​259.
Kinnard, W. and S. Messner. 1971. Industrial Real Estate. Washington, DC: Society of Industrial Realtors.
Luker, B. 1998. Foreign Investment in the Nonmetropolitan U.S. South and Midwest: A Case of Mimetic Location
Behavior? International Regional Science Review, 21:2, 163–​184.
McMahan, J. 1989. Office and Industrial. Chap.  8 in Property Development, 201–​211. New  York:  McGraw-​Hill
Publishing Company.
Mills, E. and B. Hamilton. 1989. Urban Economics. Glenview, IL: Scott, Foresman and Company.
Ronderos, L. N. 2010. Logistics Trends and Specific Industries that Will Drive Warehouse and Distribution Growth and Demand
for Space. Herndon,VA: NAIOP Research Foundation.
Shmenner, R., J. Huber, and R. Coor. 1987. Geographic Differences and the Location of New Manufacturing Facilities.
Journal of Urban Economics, 21:1, 83–​104.
SIR Educational Fund. 1984. Industrial Real Estate. Washington, DC: SIR Educational Fund.
Sivitanidou, R. 1996. Warehouse and Distribution Facilities and Community Attributes. Environment and Planning A,
27:7, 1261–​1278.
Sivitanidou, R. and P. Sivitanides. 1995a. Industrial Rent Differentials: The Case of Greater Los Angeles. Environment
and Planning A, 27:7, 11433–​1146.

422
 Industrial space market

Sivitanidou, R. and P. Sivitanides. 1995b.The Intrametropolitan Distribution of R&D Activities: Theory and Empirical
Evidence. Journal of Regional Science, 35:3, 391–​415.
Southern California Association of Governments. 2001. Employment Density Study. Los Angeles, LA: SCAG.
Thompson, R. and S. Tsolacos. 2000. Projections in the Industrial Property Market Using a Simultaneous Equation
System. Journal of Real Estate Research, 19:2, 165–​188.
Torto Wheaton Research. 1999. TWR Industrial Outlook, Spring 1999. Boston, MA: Torto Wheaton Research.
Twist, D. 2002. Determinants of Industrial Real Estate Demand. San Francisco, CA: AMB Property Corporation.
Van Den Heuvel, F. P., P.  W. De Langen, K. H. Van Donselaar, and J. C. Fransoo. 2013. Spatial Concentration and
Location Dynamics in Logistics: The Case of a Dutch Province. Journal of Transport Geography, 28, 39–​48.
Wasylenko, M. 1980. Evidence of Fiscal Differentials and Intrametropolitan Firm Relocation. Land Economics, 56:3,
339–​349.
Wheaton, W. and R. Torto. 1990. An Investment Model of the Demand and Supply of Industrial Space. AREUEA
Journal, 18:4, 530–​546.

423
 PART G

Data sources
newgenprepdf

19
DATA SOURCES FOR REAL ESTATE
MARKET ANALYSIS

Introduction 427
Sources for real estate market data 427
Sources for economic and demographic data 431
References 432

Introduction
Below we provide a partial list of sources of information that can be useful when conducting real estate
market studies. It is emphasized that listed vendors may not provide equal coverage of property types and/​
or geographies listed. Furthermore, the description of information obtainable from each source may not
necessarily reflect the full range of information that may be available. Analysts are strongly advised to con-
tact the listed sources in order to get an exact description of the property types and geography covered,
as well as the type of information that is available. Before using any of the listed sources, analysts are also
advised to carefully evaluate the information collection processes and methodologies utilized by vendors
for compiling data and producing forecasts.
Besides the vendors listed below, additional potential sources of information for real estate market
analysis include studies/​research reports prepared at universities, local Chambers of Commerce, planning
departments, and other government-​agency research divisions. Listings of additional sources of information
may be found in Clapp (1987), and Schmitz and Brett (2001).

Sources for real estate market data

Building Managers and Owners Association (BOMA)

www.boma.org
1101 15th Street NW
Suite 800
Washington, DC 20005
Phone: (202) 326–​6300
Fax: (202) 326–​6377

427
 Data sources

E-​mail: [email protected]
Property type: Office
Publishes among others the Experience Exchange Report annually. This publication provides operating
income and expense data, including contract rent components, for more than 80 metropolitan areas. It
also provides information on square feet per office worker. Data are provided for different building sizes
and submarkets (downtown and suburbs).

CB Richard Ellis
www.cbre.us
400 S. Hope Street,
25th Floor
Los Angeles, CA 90071
Property types: Office, Industrial, Retail, Multi-​Housing
Local market reports providing information on recent economic, vacancy, absorption, rental rate, and con-
struction trends.

Colliers International USA
www2.colliers.com/​en
601 Union Street
Suite 5300
Seattle, WA  98101
1-​206-​223-​0866
Property types: Office, Industrial, Retail, Multi-​Housing
Data provided include local market information on recent economic, vacancy, absorption, rental rate, and
construction trends. Market coverage is more extensive in the case of office and industrial, moderate in
the case of retail, and very limited in the case of multi-​housing.

CoStar Group
www.costar.com
2 Bethesda Metro Center,10th Floor
Bethesda, MD 20814
301-​215-​8300
Property types: Office, Industrial
Data provided include local market information on absorption, vacancy rates, rental rates, existing inven-
tory, new buildings under construction, sale prices, and cap rates.

Cushman & Wakefield

www.cushmanwakefield.com
1290 Avenue of the Americas
New York, NY 10104–​6178
212-​841-​7500
Property types: Office, Industrial, Retail
Data provided include local market information. Residential and more specialized property types may be
also covered at a broader or local level

Federal Housing Finance Agency (FHFA)

www.fhfa.gov
Constitution Center

428
 Data sources for market analysis

400 7th Street, SW

Washington, D.C. 20219
Data provided include house price indices for the 100 largest metropolitan areas in the USA on a monthly
and quarterly basis. Also national statistics for new residential mortgages are also available from this
source.

JLL
www.us.jll.com/​en/​trends-​and-​insights/​research
Jones Lang LaSalle
200 East Randolph Drive
Chicago, IL 60601
312-​782-​5800
Property types: Office, Industrial
Quarterly local market reports for the largest metropolitan areas in the U.S.

Marshall & Swift
www.corelogic.com/​solutions/​marshall-​swift.aspx
40 Pacifica, Suite 900
Irvine, CA 92618
Toll-​Free: (800) 426–​1466
Property types: Office, Industrial, Retail, Residential
Construction cost data for commercial and residential buildings.

M/​PF Research
www.realpage.com
4000 International Parkway
Carrollton, TX 75007-​1913
877-​284-​4938
Property type: Apartments
Data provided include apartment market information for the largest metropolitan areas in the U.S.

National Apartment Association (NAA)

www.naahq.org
4300 Wilson  Blvd
Suite 800
Arlington,VA  22203
703-​518-​6141
Property types: Multi-​Housing
Publishes annually the NAA Income and Expenses Survey, which provides income and expense data for rental
housing for over 90 markets in the U.S.

National Association of Realtors (NAR)

www.realtor.org
430 N. Michigan Ave
Chicago, IL 60611–​4087
800-​874-​6500
Property type: Residential

429
 Data sources

Data provided include information on sales of existing single-​family homes and condo/​co-​ops, housing
permit data for both single and multi-​family units, profiles of home buyers and sellers, housing afford-
ability statistics, and relocation statistics.

National Multifamily Housing Council (NMHC)

www.nmhc.org/​research-​insight​
1775 Eye St., N.W.
Suite 1100
Washington, DC 20006
202-​974-​2300
Publishes quarterly survey of apartment conditions, which provides a snapshot of the state of the apartment
market, sales, equity financing, and debt financing. It also publishes an annual survey of the largest
apartment owners, managers, developers, builders and syndicators in the nation, providing the number
of apartment units owned, managed, started, or syndicated in the current and/​or previous year.

Newmark Knight Frank
www.ngkf.com
125 Park Avenue
New York, NY 10017
212-​372-​2000
Property types: Office, Industrial
Data provided include local market information on office and industrial for the largest metropolitan areas
in the U.S.

REIS
www.reis.com
7 World Trade  Center
250 Greenwich Street
New York, New York 10007
(800) 366–​7347
Property types:  Affordable Housing, Apartment, Flex/​R&D, Land, Office, Retail, Self-​Storage, Seniors
Housing, Student Housing, Warehouse/​Distribution.
Data provided for most of the above property types include comparable group summary statistics for
concessions and lease terms, detailed sales comparables with a cap rate analysis proforma, new con-
struction pipeline, market and submarket historical trends, baseline and downside scenario forecasts, net
absorption, rent, and vacancy trends segmented by property class.

U.S. Census Bureau
www.census.gov/​construction/​bps​
U.S. Census Bureau
4700 Silver Hill Road
Washington DC 20233-​0001
301-​763-​4636
Property Types: Multi-​Housing
Publishes the Building Permits Survey, which provides counts of housing units authorized by building permits
in structures with varying number of units (monthly and annually) at the metropolitan-​area level.

430
 Data sources for market analysis

Sources for economic and demographic data

The following vendors provide primarily economic and/​or demographic data that may be useful for the
analysis of any property type.

Bureau of Economic Analysis (BEA)

www.bea.gov
1441 L Street NW,
Washington DC 20230
202-​606-​9208
Obtainable data include employment histories by general and detailed NAICS classifications as well as
projections for metropolitan areas, regions, and the nation. It also publishes personal income data.

Bureau of Labor Statistics (BLS)

www.bls.gov
U.S. Bureau of Labor Statistics
Postal Square Building
2 Massachusetts Ave., NE
Washington, DC 20212-​0001
202-​691-​5200
Data provided include most current and historical data on employment and wages by NAICS classifications
for counties and MSAs.

Claritas
https://​claritas360.claritas.com/​mybestsegments​
53 Brown Rd.
Ithaca, NY 14850
800-​866-​6511
Data provided include mostly demographic information at detailed geographic level (zip code level).

Moody’s Analytics
www.economy.com
121 North Walnut Street, Suite 500
West Chester, PA 19380-​3166
610-​235-​5299
Data provided include, among others, metro-​specific histories and forecasts of population, income, and
employment.

Federal Reserve Bank of St. Louis –​Economic Research

https://​fred.stlouisfed.org
One Federal Reserve Bank Plaza
St. Louis, MO 63102
Rich source of many historical time-​series data on population, income, employment, construction, and
other drivers of real estate demand and supply at the national and metropolitan-​area level.

The Federal Reserve Board (FRB)

www.federalreserve.gov

431
 Data sources

Contact information is not available.

Obtainable information includes historical and up-​to-​date data on interest rates.

U.S. Census Bureau
www.census.gov
4700 Silver Hill Road
Washington DC 20233-​0001
301-​763-​4636
Data provided include statistics on population, demographics, and housing.

Woods & Poole Economics

www.woodsandpoole.com
4910 Massachusetts Avenue, NW
Suite 208
Washington, DC 20016–​4368
Data provided include, among others, histories and forecasts of population, income, and employment at the
county level and the MSA level.

References
Clapp, J. 1987. Evaluating the Competition with Regression Analysis. Chap.  11 in Handbook for Real Estate Market
Analysis. Englewood Cliffs, NJ: Prentice Hall.
Dasso, J. and A. Ring. Real Estate, Eleventh Edition. Englewood Cliffs, NJ: Prentice Hall.
Schmitz, A. and D. Brett. 2009. Real Estate Market Analysis:  A Case Study Approach. Washington, DC:  Urban Land
Institute.

432
 INDEX

Note: Page numbers in italics refer to figures; page numbers in bold refer to tables.

Abel, A. 253, 261n8 amenity levels, residential markets 103

Abel, J. R. 60 American Community Survey: population data 73;
absolute employment shares 80, 80–​1, 82 commuting time 119
absorption equation 346–​7, 356; industrial American Housing Survey (AHS)
market 411–​12 104–​5
absorption schedules: concepts 24–​6; industrial markets anchor tenants (retail markets) 255; rents 294,
401, 409, 412, 417–​19; office markets 320–​2, 331, 295–​7
346–​7, 351, 352–​4, 365, 373–​6, 383, 386, 390, 390; Anderson, I. R. 396, 397, 421n3
project residential markets 101, 140, 182–​4, 188, Anderson, S. T.  186
207–​14, 235–​9, 240 Annual Retail Trade Survey 302–​3
access issues: industrial markets 414; residential markets Anundsen, A. K. 102, 112
187–​8, 189; retail markets 275–​6 Apergis, N. 109, 110
accounting techniques: aggregate approaches 123–​8, 339, Archer, W.  R.  372
342, 345, 403; disaggregate approaches 128–​35, 136–​8, Arestis, P. 103
210, 339–​43, 345, 382, 403, 404; industrial markets average absorption 25–​6
394, 401, 402–​9; office markets 324, 330, 336, 339–​46, Average Amenity Index (AAI) 196–​7
382; residential markets 118–​37; retail markets 264–​8, average household income 107
267, 294–​5
adaptive expectations 33 Bajari, P. 106, 109, 116n6
adjusted R-​squared 153 Barkley, D. L. 60
affordability, residential markets 106–​9 Barras, R. 318
age factors: residential markets 102, 104, 105, 106, 114, Barrett, G.V. 405, 406
130, 131, 132, 133; retail markets 251, 251–​2 Bartik, T.  J.  60
agglomeration economies: industrial markets 407; basic growth indicators, metropolitan growth
metropolitan growth analysis 59, 66; office analysis 69–​72
markets 361 behavioral/​structural equations 155–​6; residential
airport access: industrial markets 408, 414, 415; office markets 159–​64, 160
markets 361, 362 Bernanke, B. 253, 261n8
Ambrose, B. W.  418 best-​fit industry analysis 404–​9
amenity indices: Average Amenity Index (AAI) 196–​7; beta coefficient 142, 150, 154
formulas 196, 197; Relative Amenity Indices (RAISi) birth rates 104
196, 197–​8, 198, 199, 374; residential markets 191, Blackley, P. 407
192, 194–​8, 195, 198; Total Amenity Index (TAIi) 196, Blair, J. 405, 406
289, 294, 369 Boehm, T.  P.  102

433
 Index

Boston: demand-​induced growth 62–​3; government Cheng, S. 422n15

expenditure 59–​60; retail markets 308 Clapp, J. 190, 191, 192, 199, 300n7, 323, 369, 370,
Bowen, Jr., J. T. 408 371, 427
break-​even point, retail markets 276, 295, 298 Claritas 431
Brett, D. 12, 116n1, 256, 281, 427 Colliers International USA 428
building codes, residential markets 187 community attributes, residential markets 186–​7,
Building Managers and Owners Association (BOMA) 188, 189
342, 347, 427–​8 community constraints, industrial markets 415–​16
building permits 29; data 113; equation 163–​5; community shopping centers, 253, 254, 277, 278,
forecasting apartment rents 156, 160, 165–​75; 281, 305
stock-​flow equation  27–​8 commuting patterns, residential markets 119–​20, 120
Building Permits Survey 179n6 comparative advantage, metropolitan growth analysis 59
Bureau of Economic Analysis (BEA) 430–​1; employment comparison shopping 271, 274
and income data 69–​70; housing depreciation rate competition: industrial markets 416–​17, 419;
127; office employment growth 320 metropolitan growth analysis 83, 84, 85; office markets
Bureau of Labor Statistics (BLS) 431; Consumer 353, 364, 365–​8, 384; residential markets 121, 190–​1,
Expenditure Survey 252, 267–​8, 302; office 210; retail markets 257, 264, 276, 283–​4, 287; see also
employment growth 321; Relative Importance of competitive differential techniques
Components in the Consumer Price Indexes (RICCPI) competitive differentials technique: industrial markets
252, 302; retail markets 265 416, 417–​18; office markets 353, 363–​4, 365–​6,
business cycles 396 368–​70, 378nn1, 5; residential markets 182, 190–​9,
business environment, industrial markets 409 193, 207, 208, 212; retail markets 284, 287, 294,
Business Services, office markets 320, 321, 361 295, 297
business support activities, close-​by access to 360 competitive market area, residential markets 121
Byrne, D. 104 Competitive Position Index (CPI): industrial
markets 417, 418; office markets 353–​4, 368, 373,
California: government expenditure 59–​60; household 375; residential markets 190, 191, 197, 198, 199,
affordability measures 107–​9; immigration of 203, 208–​10
employable population 63; industrial markets 398–​9, complementary shopping 271, 274
399, 402; office markets 340–​3 completion stage, new construction 28–​9
California Association of Realtors (CAR) 107, 108–​9 completions minus net absorption measure 41–​3
Cambridge Systematics 289–​92, 295, 306, 307, 308 completions 28–​9; assessing the extent of disequilibrium
capital availability: office markets 323; residential 40–​3; equilibrium condition 38; office completions
markets 113–​14 30–​2; simple stock-​flow model 38–​9; stock-​flow
capitalization/​cap rates 4, 113, 327n6, 428 model with vacancy 44
capture rate: industrial markets 417, 419; office markets components of market analysis 8–​12
353, 373–​4; retail markets 287–​92, 307 concentric ring approach, retail trade area definition
Carn, N. 107, 127, 130, 138nn2–​4, 247, 249, 260n3, 271, 281, 281
320, 321, 341, 343 concessions, office markets 367–​8
Case, C. 112 Conference Board 261n8
CB Richard Ellis 332, 428 construction: costs 219–​20, 323, 331, 401, 413;
Census Bureau 430, 431; American Housing Survey investment 13, 17n7, 30, 163, 317, 322–​3, 401, 410
104–​5; Annual Retail Trade Survey 302–​3; Building Construction Report Series 29
Permits Survey 179n6; commuting time 119; construction sector, Los Angeles 74–​5, 74, 79, 80, 80,
Construction Report Series 29; Economic Census 302; 82, 82, 84
household definition 102; “Household Operation” consumer expenditures 245–​53, 257, 264–​5, 291
expenditure 247; household size 104; Metropolitan Consumer Expenditure Survey 247, 252, 267–​8, 302
Statistical Areas 69, 122; Monthly Retail Sales 266, 302; consumer information, residential markets 188
population data 70; residential markets 113; residential Consumer Price Index (CPI) 90–​1, 92–​3; 1970–​1998
vacancy rates 127; retail data 266, 301, 302–​3 91; base year change 91, 92; and income 89–​90, 92;
Census of Retail Trade 247, 265, 268, 270, 300n3 residential markets 158; urban consumers (CPI-​U) 90;
censuses 301, 302; employment data 340; residential urban wage earners (CPI-​W) 90
markets 121, 127, 157, 189 consumer psychology, retail markets 253
Center for Real Estate and Urban Economics, consumer surveys: residential markets 204, 228; retail
University of California at Berkeley 402 markets 252, 285
Central Limit Theorem  148 consumer utility, retail markets 290
Chan, P. 116n6 convenience goods 245, 271, 272, 274

434
 Index

Coor, R. 422n10 flow identity 28; see also economic depreciation;

Core Based Statistical Area (CBSA) 9 functional depreciation; physical depreciation
Corgel, J. B. 347 desired demand 20
corporate capital, after-​tax cost of 397 desired (equilibrium) rent: influences 48–​9; office
cost approach to value 17n1 markets 350–​1; vs. prevailing rents 48–​9
cost factors, office markets 323 development controls, industrial markets 416
CoStar Group 322, 428 development cost(s) 4, 33, 113, 218–​19, 259, 322–​3, 326,
County Business Patterns: office markets 340; residential 372; non-​land  224
markets 189 development density, residential markets 218–​20, 227,
credit availability: and demand 23; retail markets 253 229–​32, 230, 231, 232, 232, 233
cross-​section econometric models 140, 141, 146; development potential 139, 376, 401, 404, 409, 416, 420
residential markets 148 development process 332
Cushman & Wakefield  428 development stage, residential markets 186
customer spotting approach, retail trade area diminishing marginal utility, principle of 200
definition 281–​2 DiPasquale, D. 58, 60, 102, 105, 107, 109, 110, 112, 114,
cycles: and stock-​flow model 39, 112; industrial market 122, 164, 179nn4–​5, 217nn6, 13–​14, 219–​20, 253,
396; office market 318–​19 261n7, 290, 300–​1nn4, 9–​13, 18, 303, 320, 340, 367,
394, 395, 396, 397, 410, 411–​12, 422n14
Dallas: apartment market 156–​75; office markets 350 discount stores 306
data sources 427; demographics 302, 430–​2; economic discounted cash-​flow model 17n2
and demographic data 430–​2; income 302; discrete choice analysis 300n9
metropolitan growth analysis 86; office markets 366; disposable household income 107
population 301; real estate market data 427–​30; distribution space, industrial markets 414–​15
residential markets 188–​9; retail markets 301–​3 divorce rates 104
death rates 104 Dollars and Cents of Shopping Centers 266, 293,
demand 20; analysis 102–​3, 275–​6, 394, 403–​4; concepts 298, 301n14
20, 24–​6; endogenous determinants 22; exogenous downtown district office nodes 352, 354
determinants 22–​4; fundamental law of 20, 21, 22; driving time approach, retail trade area definition
industrial markets 396–​400, 403–​9, 411, 413, 417, 419; 281, 281
metropolitan growth analysis 58–​63, 59; office markets Dudensing, R. M. 60
319–​22, 331, 353; replacement 8, 57, 397–​8; residential Duffy, D. 104
markets 100, 101–​12, 123, 128–​31, 133, 134, 136–​7, dummy variables 202
139–​40, 147, 159, 162, 209–​10, 227, 237–​9, 238, 239;
retail markets 244, 245–​53, 275–​6; simple stock-​flow e-​commerce: industrial markets 395; retail
model 36, 37, 38–​40; see also demand curve; demand–​ markets 243–​5
supply imbalances; price elasticity of demand econometric analysis: industrial markets 394, 401, 402,
demand curve 20–​1, 35; shifts 22, 23, 23, 24, 35, 103, 109 406, 409–​13, 417; office markets 324, 330, 336–​9, 341,
demand equation 144; apartment market 159–​61; 345, 346–​51, 352–​3, 355–​7, 365–​6; residential markets
industrial market 401; office market 347–​8; speculative 139–​51, 155–​80, 199–​207, 209, 213; retail markets
industrial market 411 264, 268–​70, 284, 297; see also hedonic valuation
demand–​supply imbalances: advanced assessment techniques
methodologies 44–​9; industrial markets 402–​3, Economic Census 302; retail markets 252
403; market inefficiencies 40; office markets 324, economic depreciation 28; industrial markets 397–​8
330, 331–​2, 335–​51, 354, 362; residential markets economics 20, 49–​50; demand 20–​6; demand–​supply
100; retail markets 263–​4, 266; simple assessment imbalances, assessment of 40–​9; price adjustments
methodologies 40–​3 34–​40; residential markets 101–​12; supply 26–​34
demographics: data sources 302, 430–​2; residential economy, national 59
markets 103–​6, 186, 188, 189, 204, 223, 223; retail education: metropolitan growth analysis 60, 65;
markets 249–​52, 275 residential markets 187
Denver, office markets 351 effective demand 20
Department of Commerce: Consumer Price Index 158; elasticity of supply 30; see also price elasticity of supply
employment data 353; residential markets 157; retail employment: industrial markets 396–​7, 400, 402, 403,
markets 243; Survey of Current Business 404, 409, 411, 412; metropolitan growth analysis
253, 411 69–​72; nodes (residential markets) 121; office markets
Department of Housing and Urban Development 110 31, 31, 320–​1, 322, 332, 333, 334, 339–​43, 346,
depreciation: industrial markets 397–​8, 402; metropolitan 353, 383
growth analysis 57; residential markets 127, 167; stock endogenous: determinants variables 144–​5

435
 Index

energy costs, industrial markets 408 Forestry, Fishing, and Mining sector, Los Angeles 80, 80,
Energy Information Administration (EIA) 399 82, 83, 84, 85
entrepreneurship, metropolitan growth analysis 60 freeway access: industrial markets 408, 414, 415; office
environmental impact studies (EIS), residential markets 361, 362
markets 192 frequency distributions 151n2
environmental regulations: industrial markets 416; functional depreciation 28; industrial markets 397, 398
residential markets 187 functional heterogeneity, industrial markets 395
equilibrium (desired) rent: influences 48–​9; office
markets 350–​1; vs. prevailing rents 48–​9 Gabe, T.  M.  60
Erickcek, D. 60 geographic analysis 9–​10
evaluation of market studies: criteria 14–​15; industrial Geographic Information Systems (GIS), retail
markets 419–​20; retail markets 306–​8 markets 275
Evans, P. 245 Gibler, K. M. 320, 321
ex-​ante demand  20 global financial crisis (2008/​9): construction sector 74;
exogenous: cycles 318; determinants of demand 22–​4, office construction 31; sub-​prime loans 114
159, 163; determinants of new residential construction Goetzmann, W.  122
113–​14; factors 32, 38; variables 38, 85, 144–​5, 156, Gonzalez, A. R. 103
172, 175, 235, 413 government expenditure, metropolitan growth
expectations: adaptive 33; and demand 23, 24; myopic 33, analysis 59–​60
112; office markets 320, 322, 323, 327n6; rational 33; government sector: Los Angeles 78, 78–​9, 80, 80, 82, 82,
residential markets 112, 114, 115, 115; retail markets 84, 85; office markets 320
253; shock 35, 40, 45, 48–​9, 109; and supply 33–​4 Graevenitz, K. 207
Experience Exchange Report 428 gravity approach, retail trade area definition 282, 285,
ex-​post demand  20 288, 297
Green, R. K. 102
factor analysis, retail markets 260n6 Grissom, T. 116n3, 301n15
factor availability and costs, industrial markets 407–​8 gross absorption 24
failure rate, in retail markets 307–​8 gross household income 107
fair market share: industrial markets 417; office markets growth controls, residential markets 187, 216n1
373–​4, 375; residential markets 207–​8 Guirguis, H. 396, 397, 421n3
Fanning, S. 116n3, 281, 301n15
feasibility studies: differences from market analysis 12; Hagen, D. 127
office markets 364–​5; residential markets 101, 140 Hamilton, B. 407
Federal Housing Finance Agency (FHFA) 179n1 Hansen, J. 127
Federal Reserve Bank of St. Louis –​Economic Harding, J. P. 28
Research 431 Harmon, O. R. 106
Federal Reserve Board (FRB) 431; Index of headship rates 104, 105; defined 116n2; formulas 104
Manufacturing Output (IMO) 396; interest rates 112 hedonic valuation techniques 12, 101, 182, 190,
financial data, residential markets example 226, 226 199–​207, 213; industrial markets 406, 415, 416, 417,
financial feasibility analysis: differences from market 418; office markets 364, 365, 366, 368, 370–​3, 375,
analysis 12; office markets 364–​5; residential markets 381–​2, 386, 388, 389; residential markets 182, 190,
101, 140 199–​207, 205–​6, 212, 216, 218–​20, 222, 223–​4, 224,
Financial Services 321 225, 228, 230–​4, 235, 236; retail markets 227, 294,
Fire, Insurance, and Real Estate (FIRE) sector: Los 295, 297
Angeles 75, 75, 79, 80, 80, 82, 82, 83, 84, 85; office hedonic price analysis 223–​5
markets 320, 321, 323, 327n5 Heebøll, C. 102, 112
Fischel, W.  216n1 Hendershott, P. H. 245, 253
FitzGerald, J. 104 heterogeneous products 274
flex industrial space 395 high-​order products see shopping products
floor area ratio (FAR), residential markets 187, Holl, A. 408
216n1, 218–​20 Homes and Communities Agency 398, 400
footloose industries 405 Hou, Y.  361
forecasting: apartment market 155–​67; housing demand households: defined 102; determinants 103–​4; estimation
130; industrial market 409–​12; industrial space 104; qualifying 130–​4; size 104
demand 403–​8; office market 346–​50; office space housing demand see demand
demand 339–​42; residential construction 113; with housing gap, aggregate approach 123–​8, 124, 125
regression models 145 housing mobility patterns 104–​5; resale homes 114, 115

436
 Index

housing rent see rents International Council of Shopping Centers (ICSC) 253,

housing stock 65, 165 255, 260n1
housing supply see supply irreversible investments theory 33
housing types 102, 105, 106, 131 Irwin, E. G. 202
Houston, office markets 372
Huber, J. 422n10 JLL 428–​9
Huff ’s probability model 288–​9 Jones, C. 119, 184, 187
human capital: metropolitan growth analysis 60, 65;
residential markets 187 Katz, B. 66
Hutchinson, B. 361 Kim, J. H. 187
Hwang, M. 23 Kinnard, W.  405
hypothesis testing 154 Kling, J. 323
Krueger, D. 116n6
Ibanez, M. R. 7, 51n2, 318, 396
Ihlanfeldt, K. 30, 113, 114 labor force participation, metropolitan growth
image-​related attributes, residential markets 186 analysis 64
immigration 73; of employable population 63 labor-​oriented industries  405
income approach to value 4 labor supply: industrial markets 415; metropolitan
income capitalization formula, office markets 327n6 growth analysis 63–​5; wage elasticity of 60–​1,
income-​earning capacity, influences on 6, 6–​8 60–​1
income elasticity of housing demand and supply 101–​2 land availability, residential markets 113
income/​wealth: Consumer Price Index 89–​90; data land prices, industrial markets 417–​19
sources 302; and demand 23; metropolitan growth land-​use composition, residential markets 186–​7
analysis 56–​7, 60, 61, 64, 69–​72, 73–​85; office markets lease rates and terms: industrial markets 418; office
322–​3; residential markets 102, 104, 106–​9, 114, 115, markets 366–​8, 372, 382–​3, 384, 386, 388
130, 131, 133, 133, 134, 223, 223, 237, 237–​8, 238; lease transactions, office markets 365
retail markets 249–​52, 251, 307 lease-​up time, office markets 351
Index of Manufacturing Output (IMO) 396 Lehman Brothers 114
indices 88; calculation 89, 89; comparisons 89, 90, 90; linear regression analysis 142–​4, 152–​4, 152; residential
defined 88; see also specific indices markets 146–​50
industrial agglomeration, metropolitan growth linkages, residential markets 187–​8, 189
analysis 59, 66 local fiscal environment, residential markets 187
industrial development potential of an area 409 local government regulations, residential markets 187
industrial markets 394, 420–​1; demand 396–​400, local market conditions, residential markets 188–​9
403–​9, 411, 413, 417, 419; evaluation criteria 419–​20; local public goods, residential markets 187
idiosyncrasies 395–​6; macro analysis 13–​14, 394, 401, location evaluation matrix, industrial markets 406
402–​13, 419–​20; micro analysis 14, 394, 402, 413–​19, location quotient: formula 81; interpretation 81–​2; Los
420; overview 401–​2; supply 400–​1, 411, 412, 413 Angeles 82, 82; metropolitan growth analysis 80–​2;
industrial output 396–​7, 402, 412 origins of idea 81
industrial parks 417, 418 location shifts, metropolitan growth analysis 57
industry effect, metropolitan growth analysis 83, 84, 85 location/​site: industrial markets 404–​6, 413–​17;
industry-​occupation matrices, office markets 339–​40 office markets 319, 324, 353, 360–​4, 366, 368, 373;
industry-​structure trends, metropolitan growth analysis residential markets 103, 184–​8, 185, 192; retail markets
71, 73–​85 244, 257, 272, 274–​5, 276, 277, 278–​9, 283
inflation: and interest rates 116n7; residential logistics space 395
markets 110–​12 long-​run aggregate supply 26–​7, 27
information sources 427; demographics 302, 430–​2; Los Angeles: apartment market 25–​6, 26, 176–​9, 177;
economic and demographic data 430–​2; income completions and net absorption (1994–​2014) 41, 41;
302; metropolitan growth analysis 86; office markets industrial markets 414, 415, 422n16; metropolitan
366; population 301; real estate market data 427–​30; growth analysis 56, 61, 68, 71–​2, 74–​86; office market
residential markets 188–​9; retail markets 301–​3 41, 42, 43, 43, 332–​5, 332–​5, 336, 361–​2; population,
Ingene, C. 250, 269, 270 employment and income (1969–​2017) 70, 71–​2;
innovation, metropolitan growth analysis 59, 60, 66 population index 88–​9, 89; wage elasticity of labor
Institute of Supply Management, PMI Index supply 61
(ISMPMI) 396 low-​order (convenience) products 245, 271, 272,
interest rates 23; and inflation 116n7; office markets 323, 274
327n6; residential markets 109, 112 Luker, B. 422n15

437
 Index

MacKinnon, G. 65 mortgage lender’s affordability criterion 107, 108

macro analysis 8, 9, 11, 67n1; components 10; mortgage rates, residential markets 109–​10, 114, 115
industrial markets 13–​14, 394, 401, 402–​13, 419–​20; mortgage underwriting standards, residential markets 130
metropolitan growth analysis 55; office markets 13, M/​PF Research  429
322, 324, 329–​57, 381, 384, 388–​90; regression models Mueller, G. R. 318
146; residential markets 13, 98, 100, 101–​15, 118–​37, multi-​collinearity, residential markets 217n10
139–​51, 155–​80, 222, 239–​40; retail markets 13, multiple regression analysis 143; beta coefficient 142,
244, 275 150, 154; residential markets 150, 156–​74
macroeconomic forces 7–​8, 8 Mun, S. 361
manufacturing sector, Los Angeles 75–​6, 76, 79, 80, 80, Myers, D. 373
82, 82, 83, 84 myopic expectations 33; residential markets 112
manufacturing space 395, 414
Mariotti, I. 408 National Apartment Association (NAA) 429
market area definition, residential markets 119–​22 National Association of Realtors (NAR) 429;
market-​oriented industries  405 affordability index 107, 108; median time on the
market risk, new construction 33–​4 market 188
market size, and demand 23 National Council of Real Estate Investment Fiduciaries
market strength assessment, industrial market 394; (NCREIF) 395
residential markets 120, 122–​35, 140; retail market national economy 59
256; office market 324–​5, 330, 332, 335–​51 National Multifamily Housing Council
markup, and retail rent estimation 295, 296, 298 (NMHC) 429–​30
marriage rates 104 natural logarithm (ln) 202
Marshall & Swift 429 natural (structural) vacancy rate: analysis 46–​7; industrial
mathematical approaches, retail rent estimation 294–​7 markets 403; influences 46–​7; vs. nominal vacancy
Mayock, T. 30, 113, 114 rate 44–​7; office markets 348, 349–​50, 349, 354, 375;
McCue, T.  323 residential markets 127
McMahan, J. 404, 406, 409, 417, 418 near twin approach, retail trade area definition 281–​2
median household income 107; defined 116n4 Nechayev, G. 7, 23
median time on the market, residential markets 188 neighborhood attributes, residential markets 186–​7, 189
Messner, S. 405 neighborhood centers, retail markets 305
metropolitan growth patterns 6–​7, 8–​11, 55–​6, 66, net absorption 24–​5; office markets 331, 332, 332–​3,
68–​9, 86; broad-​brush framework 71; components 10; 334, 336–​9, 346–​7
defining 56–​7; demand-​induced 58–​63, 59; indices net household formation 103
88–​93; information sources 86; main steps 69–​86; net migration: calculation 73; metropolitan growth
mechanisms of growth 58–​66; in perspective 56; analysis 72–​3; residential markets 104
project 87–​9; and residential markets 121–​2; and retail net operating income (NOI) 4
markets 245, 249, 256; supply-​induced 58, 59, 63–​6 net present value (NPV) analysis, residential
Metropolitan Statistical Areas (MSAs) 9–​10, 69, 122 markets 212–​13
micro analysis 8, 9, 12, 67n1; components 10 new construction 27–​31; determinants 31–​2, 33;
industrial markets 14, 394, 402, 413–​19, 420; office industrial markets 401, 410–​11; office markets 322–​3,
markets 13, 324, 359–​76, 381, 384, 388; regression 332, 333–​4, 335, 345, 346–​7, 354, 375; pipeline 28–​9;
models 146; residential markets 13, 98, 100–​1, residential markets 112–​14, 159, 161, 162, 163–​4,
181–​220, 222, 240; retail markets 13, 244–​5, 276, 165–​7, 168, 169, 170–​1, 172, 173, 192, 209
282, 287 Newmark Knight Frank 430
microeconomic forces 6–​8, 8 nominal house prices 236–​7, 237
micropolitan areas 9 nominal market price index, residential markets 236, 236
migration 73; of employable population 63; residential nominal rents 51n3
markets 114; see also net migration nominal vacancy rate (V) 43; vs. structural vacancy
Miller, D. 116n6 rate 44–​7
Mills, E. 407 non-​land development costs, residential markets example
minimum lot requirements, residential markets 224, 226
187, 216n1 non-​productive amenities, industrial markets 409; office
MIT-​Cambridge Systematics model 289–​92, 295 market 319, 360–​1
Mitchell, P. S. 373 non-​speculative industrial space 395–​6, 400, 402, 409–​11
Monthly Retail Sales 302 non-​standardized products  274
months supply, residential markets 188 normal absorption 25–​6
Moody’s Analytics 431 normal distribution 148

438
 Index

North American Industrial Classification System Pindyck, R. 144

(NAICS) 7, 17n3; industrial markets 401, 404, 409, Platt Retail Institute 249
411, 418; retail markets 266 population: basic population identity 72; data sources
301; indices 88–​9, 89; metropolitan growth analysis
obsolescence: industrial markets 397–​8; metropolitan 59, 69–​73; residential markets 104, 130, 132
growth analysis 57 preferences, residential markets example 226–​7,
occupied stock 24–​5, 42; desired 348, 412; industrial 238
market 411–​12; office market 337–​8 prestige, office markets 361, 362
Office Experience Exchange Report 342 prevailing vs. implicit equilibrium rents 48–​9
Office of Management and Budget (OMB): metro price: nominal house prices 236–​7, 237; nominal
and micro areas 9; Metropolitan Statistical Areas 69; market price index 236, 236; office markets 319, 372;
NAICS 17n3 residential markets 101, 106–​9, 112, 113, 114–​15, 115,
office markets 317, 326, 380; data sources 366; evaluating 139–​40, 147, 182, 190–​214; retail markets 253; see also
studies 384–​90; example 381–​4; idiosyncrasies and price adjustments; price elasticity of demand; price
project success determinants 318–​23; macro analysis elasticity of supply
13, 322, 324, 329–​57, 381, 384, 388–​90; micro analysis price adjustments 34; long-​vs. short-​run changes 35,
13, 324, 359–​76, 381, 384, 388; new construction 35; market price determination 34–​5, 34; residential
and employment growth (1994–​2014) 31, 31; new markets 100; simple stock-​flow model 36–​40
construction and real rents (1994–​2014) 30–​1, 30; price elasticity of demand 20–​2; actual vs. expected price
new construction and vacancy rate (1994–​2014) 31, changes 22; labor demand 64–​5, 64; product demand
32; overview 324–​6, 325; project 385–​90; supply 64–​5, 64
322–​3, 331 price elasticity of supply 27, 30, 35; metropolitan growth
office parks 352 analysis 62, 62; residential markets 101–​2, 113
office showroom space 395 probabilistic approaches: capture rate estimation (retail
office space demand see demand markets) 288, 288–​92, 297, 303; retail trade area
office space requirements per worker 388 definition 282
office space supply see supply product design, residential markets 189; example 223–​6,
office worker/​residential amenities 361, 362 224, 228–​9, 229, 230
office-​using sectors  320 product information, residential markets 188
off-​price centers  306 production space, industrial markets 414
Olaru, D. 189 productive amenities: industrial markets 407; office
omnichaneling 244 markets 360–​1
operating costs, retail markets 295, 296, 298 productivity, metropolitan growth analysis 59
optimal development density 218–​20 Professional Services 320, 321
ordinary least squares (OLS) 144; residential markets 146 profit ratio (markup), and retail rent estimation 295,
O’Sullivan, A. 110 296, 298
outlet centers 306 profitability, office markets 322
output growth, metropolitan growth analysis 56–​7 project absorption schedules, residential markets 182–​4,
207–​14, 235–​9, 240
Panduro, T.  E.  207 project design: office markets 319; residential markets
Panel Study of Income Dynamics 106 223–​34; retail markets 245
patronage probabilities, retail markets 291, 300–​1n12 project evaluation matrix, residential markets 189, 190
Pearson, T. 116n3, 301n15 project features, industrial markets 416
Peiser, R. 318 project price/​rent analysis, residential markets 182,
Pennington-​Cross, A. 7, 51n2, 318, 396 190–​214
pent-​up demand  20 project revenue schedule, residential markets 182,
performance indicators, retail markets 276 207–​10, 208, 212, 235–​9, 240
permanent household income 107 projects: econometric forecasting of office markets
permits: new construction 28–​9; residential markets 157, 355–​7; evaluating an office development scenario
159, 163, 164, 165–​7, 168, 169, 170–​1, 172, 173, 178 385–​90; hedonic price models 216; Los Angeles
persistence rate, residential markets 163 apartment market forecasting 176–​9; metropolitan
personal services 245 growth analysis 87–​8; residential markets and the
Philadelphia: industrial markets 401, 411–​12; office accounting approach 136–​8
markets 350 project-​specific analysis see micro analysis
physical depreciation 28; industrial markets 397 property tax rates, office markets 361, 362
physical growth, metropolitan growth analysis psychographics: residential markets 106; retail
56–​7, 61, 64 markets 252

439
 Index

purchase frequency 271 residential amenities: industrial markets 414; office

purchasing power 247–​9, 248 markets 361, 362
residential markets 97–​101; accounting techniques
qualitative analysis, residential markets 100–​1, 182, 118–​37; basic housing market economics 101–​15; data
183, 184–​90 sources 188–​9; econometrics 139–​51, 155–​80,
quality-​controlled price/​rent indices: defined 116n5; 199–​207, 209, 213; example analysis 222–​40; macro
residential markets 109 analysis 13, 98, 100, 101–​15, 118–​37, 139–​51, 155–​80,
quality factors, residential markets 103, 106, 107 222, 239–​40; micro analysis 13, 98, 100–​1, 181–​220,
quantitative analysis, residential markets 101, 182, 183, 222, 240; overview 98–​101, 99; supply 100, 101–​2,
190–​214 112–​15, 123–​7, 133, 139–​40, 147, 159, 163–​7, 168,
Quigley, J. 23 169, 170–​1, 173, 209, 227, 227
Residis, A. 109, 110
Rabianski, J. S. 121, 216n2, 320, 321 residual land value (RLV): industrial markets 421n4;
Radov, K. 306, 307 inputs 4, 5; office markets 372; redevelopment 28;
rational expectations 33 residential markets 217n13, 218–​19, 229–​31, 234, 234
real market price index, residential markets 227, 228 residual valuation technique 217n13
real rents 51n3; Los Angeles office market (1994–​2014) resource-​oriented industries  405
43, 43; and office construction (1994–​2014) 30–​1, 30 retail markets 243–​4, 259–​60, 263, 298–​9, 305; categories
realized demand 20 of goods and services 271–​3, 272; classification
redevelopment 28 253, 254–​5; common flaws of studies 306–​8; data
regional retail markets 305 sources 301–​3; definition/​types of 264; demand 244,
regression analysis: basics 140–​5; evaluating regression 245–​53, 275–​6; emphasis of studies 305–​6; evaluation
models 145; prediction 145; residential markets 140, of studies 306–​8; example 308–​12; idiosyncrasies
146–​50, 213; see also hedonic valuation techniques; 244–​5; macro analysis 13, 244, 275; micro analysis
linear regression analysis; multiple regression analysis 13, 244–​5, 276, 282, 287; objectives of analysis 256;
Reilly, W.  288 phase I: assessing unrealized sales potential 263–​71;
REIS 430 phase II: spatial behavior of consumers 271–​7; phase
Relative Amenity Indices (RAISi): office markets 374; III: refining and evaluating the retail development
residential markets 196, 197–​8, 198, 199 scenario 277–​98; phases of analysis 256–​9, 258,
relative employment shares, metropolitan growth analysis 263–​98; supply 253–​6, 276
80, 82, 82 retail trade area definition 277–​82, 280, 281
Relative Importance of Components in the Consumer Price retail trade sector, Los Angeles 77, 78, 79, 80, 80,
Indexes (RICCPI) 247, 252, 302 82, 82, 84
relative prices, residential markets 109, 114, 115 revenue schedules, office markets 365, 373–​6, 383–​4,
relocation demand 8 385, 386–​7
rent adjustment equation 46 risk: new construction 33–​4; office markets 323, 327n6;
rent adjustment process 46, 161–​2 residential markets 114
rent controls, residential markets 187 Riverside, metropolitan growth analysis 56
rent escalation clauses, office markets 367 Ronderos, L. N. 395, 398, 399
rent–​vacancy cycle 45, 45–​6 Rosen, K. 127, 179n3, 347
rents 34; industrial markets 401, 411–​12, 413, 417–​19; R-​squared 153; adjusted 153
long-​vs. short-​run changes 35; market rent Rubinfeld, D. 144
determination 34–​5; nominal 51n3; office markets
319, 320, 322, 323, 324, 330, 332, 333, 336, 337, 342, sales comparison approach to value 17n1
346–​51, 354, 362–​73, 375–​6, 385, 386–​7 sales data, retail markets 265–​6, 269, 307
prevailing vs. implicit equilibrium 48–​9; residential sales potential by line of trade 292–​3, 293
markets 101, 106–​7, 109, 113, 139–​40, 147–​8, sales prices 140, 203–​4, 216
157–​9, 161–​5, 166, 167, 169, 170–​1, 172, 173, 174, San Francisco, covariance of housing returns 122
178, 182, 190–​214; retail markets 294–​8, 310–​12, Schlottman, A. M. 102
312, 313; simple stock-​flow model 36–​40; speed of Schmenner, R. 407, 422n10
adjustment calculation 179–​80; see also equilibrium Schmitz, A. 12, 116n1, 256, 281, 427
(desired) rent; real rents; rent adjustment equation; sectoral employment trends, metropolitan growth
rent–​vacancy  cycle analysis 71, 73–​85
replacement demand 8; industrial markets 397–​8; sensitivity analysis, residential markets 127, 128, 167–​74
metropolitan growth analysis 57 Service sector: Los Angeles 76, 76–​7, 79, 80, 80, 82, 82,
resale units, residential markets 112, 114–​15, 115 83, 84, 85; office markets 320, 321, 323
research and development space 395, 414 share effect, metropolitan growth analysis 83, 84, 85

440
 Index

shift-​share analysis, metropolitan growth analysis stock-​flow model, expanded: office markets 346–​7, 347;
82–​5, 84 with vacancy 44–​6, 44
Shiller, R. 112 stock-​flow model, simple 36–​8, 36–​7; assessment 39–​40;
Shilling, J. D. 347 equations 37–​8; residential markets 140, 146, 157, 159,
shoppers’ goods 245 161, 162, 167, 178
shopping center classification 254–​5 stop clauses, office markets 367
shopping center office nodes 352 Stough, R. R. 422n15
shopping frequency 271 strip centers 306
shopping pattern analysis 284–​5 structural/​behavioral models 155–​6; residential markets
shopping products 271, 272 159–​64, 160
short-​run aggregate supply 27, 27 structural (natural) vacancy rate: analysis 46–​7; industrial
simple gravity approach, retail trade area definition 282, markets 403; influences 46–​7; vs. nominal vacancy
285, 288, 297 rate 44–​7; office markets 348, 349–​50, 349, 354, 375;
simple regression models 143 residential markets 127
SIR Educational Fund 405, 414 structure attributes: industrial markets 416; office markets
Sirmans, C. F. 347 366, 367, 368, 373
site/​location: industrial markets 404–​6, 413–​17; submarket analysis, offices 351–​4
office markets 319, 324, 353, 360–​4, 366, 368, 373; sub-​prime loans  113–​14
residential markets 103, 184–​8, 185, 192; retail markets substitutes, and demand 21, 23, 24
244, 257, 272, 274–​5, 276, 277, 278–​9, 283 super-​regional retail markets 305
site-​specific analysis see micro analysis supply 26; concepts 26–​34; elasticity 30; fundamental
Sivitanides, P. 34, 103, 106, 323, 327nn4, 6, 415, 416 law of 26, 29, 29; industrial markets 400–​1, 411, 412,
Sivitanidou, R. 34, 48, 319–​20, 321, 323, 327nn4–​6, 361, 413; long-​run aggregate supply 26–​7, 27; metropolitan
362, 363, 415, 416 growth analysis 58, 59, 63–​6; office markets 322–​3,
Smith, B. 189 331; residential markets 100, 101–​2, 112–​15, 123–​7,
Smith, L. 127, 179n3 133, 139–​40, 147, 159, 163–​7, 168, 169, 170–​1, 173,
Smith, M. T.  372 209, 227, 227; retail markets 253–​6, 276; short-​run
sources of data 427; demographics 302, 430–​2; aggregate supply 27, 27; simple stock-​flow model 36,
economic and demographic data 430–​2; income 37, 38–​40; see also price elasticity of supply; supply
302; metropolitan growth analysis 86; office markets curve; supply–​demand imbalances
366; population 301; real estate market data 427–​30; supply curve 26–​7, 27, 34, 35, 38, 39, 61; shifts 32,
residential markets 188–​9; retail markets 301–​3 33, 362
Southern California Association of Governments supply–​demand imbalances: advanced assessment
(SCAG) 398–​9, 418; immigration 73 methodologies 44–​9; industrial markets 402–​3,
space attributes, office markets 366, 367, 368 403; market inefficiencies 40; office markets 324,
space requirements per office worker 321–​2, 341–​3 330, 331–​2, 335–​51, 354, 362; residential markets
spatial behavior of consumers, retail markets 271–​7, 272 100; retail markets 263–​4, 266; simple assessment
spatial unit of analysis: metropolitan growth analysis 69, methodologies 40–​3
70; residential markets 102; retail markets 265, 268–​9 supportable square footage by line of trade 293–​4
special office nodes 352 Survey of Buying Power 307
specialized services 245 Survey of Current Business 253, 411
specialty centers, retail markets 306
speculative space: industrial markets 395–​6, 401, 402, target markets, residential 189, 223
410, 411–​19; office markets 331, 343, 344 tastes and lifestyle preferences, residential markets 106
speed of rent adjustment 161, 163, 179–​80 taxes: industrial markets 408; office markets 361, 362
spending patterns (retail markets) 249–​53, 250, 251 t-​critical  149–​50
Spiegel, M. 122 t-​distribution  149–​50
square feet per employee ratio, industrial markets technology, metropolitan growth analysis 66
398–​400, 399, 400, 403, 404, 409 tenant improvements (TIs), office markets 368
standard deviation 151n3 tenant mix: industrial markets 395, 416; office markets
standard error: of coefficient (sb) 149, 153–​4; of 354; retail markets 244–​5, 283
regression estimate (s) 149, 153 tenure types, residential markets 102–​3, 105, 106–​7, 109
standardized products 274 Thompson, R. 401, 410, 413
startup stage, new construction 28–​9 time frame, residential market analysis 120, 122
Statistical Abstract 249 time-​series econometric models 140–​1, 141, 142, 143,
Stevenson, S. 318 146; residential markets 148, 157–​9, 158, 176
stock-​flow identity 27–​8; residential markets 127 timing factors, office markets 318–​19, 318–​19

441
 Index

Torto, R. 7, 48, 51n8, 350, 396, 397, 400, 401, 410–​11 U.S. Office of Management and Budget (OMB): metro
Total Amenity Index (TAIi): office markets 369; and micro areas 9; Metropolitan Statistical Areas 69;
residential markets 196; retail markets 289, 294 NAICS 17n3
total household income 107 U.S. Statistical Abstract 249
Transportation and Warehousing, Utilities, and user cost of capital, residential markets 110–​12, 111, 111
Information (TWUI) sector, Los Angeles 79, 79, 80,
82, 82, 83, 84, 85 vacancy rates: industrial markets 413; nominal 43;
transportation costs, industrial markets 408, 415 nominal vs. structural 44–​7; office markets 31, 32,
transportation-​oriented industries  405 40, 41, 42, 43, 43, 318, 318, 323, 332, 334, 335, 335,
Tse, R.Y. C. 320 336, 337, 339, 345, 346–​7, 348, 349–​50, 351, 354,
Tsolacos, S. 401, 410, 413 372, 378n3; residential markets 125–​7, 164, 209; retail
t-​statistic 149–​50, 154; residential markets 164 markets 301n15; simple stock-​flow model 40, 51n7
Twist, D. 396 Van Den Heuvel, F. P. 408
two-​stage least squares, residential markets 217n11 visibility, office markets 361, 362
volatility, office markets 323
uncertainty, residential markets 114
University of California at Berkeley, Center for Real wage elasticity of labor supply 60–​1,
Estate and Urban Economics 402 60–​1
University of Michigan, Survey of Consumers wages: industrial markets 422n15; and office rents
253, 261n8 363–​4
unrealized sales potential, retail markets 256–​7, 263–​71 Wagner, J. 66
uptown district office nodes 352 warehouse space 395, 414–​15
Urban Land Institute (ULI) 13; Dollars and Cents of warehouse stores 306
Shopping Centers 266 Wasylenko, M. 408
U.S. Bureau of Economic Analysis (BEA) 430–​1; wealth/​income: Consumer Price Index 89–​90; data
employment and income data 69–​70; housing sources 302; and demand 23; metropolitan growth
depreciation rate 127; office employment growth 320 analysis 56–​7, 60, 61, 64, 69–​72, 73–​85; office markets
U.S. Bureau of Labor Statistics (BLS) 431; Consumer 322–​3; residential markets 102, 104, 106–​9, 114,
Expenditure Survey 252, 267–​8, 302; office 115, 130, 131, 133, 133, 134, 223, 223, 237,
employment growth 321; Relative Importance of 237–​8, 238; retail markets 249–​52, 251, 307
Components in the Consumer Price Indexes (RICCPI) Webb, J. R. 205, 320
252, 302; retail markets 265 Weisbrod, G. 306, 307
U.S. Census Bureau 430, 431; American Housing West, S. E. 186
Survey 104–​5; Annual Retail Trade Survey 302–​3; Wheaton, W. 7, 23, 48, 51n8, 58, 60, 102, 105, 107, 109,
Building Permits Survey 179n6; commuting time 119; 110, 112, 114, 122, 164, 179nn4–​5, 217nn6, 13–​14,
Construction Report Series 29; Economic Census 302; 219–​20, 253, 261n7, 290, 300–​1nn4, 9–​13, 18, 303,
household definition 102; “Household Operation” 318, 320, 340, 347, 350, 363, 367, 394, 395, 396, 397,
expenditure 247; household size 104; Metropolitan 400, 401, 410–​12, 422n14
Statistical Areas 69, 122; Monthly Retail Sales 302; 266; white-​collar labor, access to 361–​2
population data 70; residential markets 113; residential Wholesale Trade sector, Los Angeles 77, 77, 79, 80,
vacancy rates 127; retail data 266, 301, 302–​3 82, 82, 84
U.S. Department of Commerce: Consumer Price Woods & Poole Economics 432
Index 158; employment data 353; residential markets worker amenities, industrial markets 409
157; retail markets 243; Survey of Current Business
253, 411 Zietz, J. 203
U.S. Department of Housing and Urban zoning: industrial markets 416; residential markets
Development 110 example 222

442