Daniel Beimborn

Cooperative Sourcing
GABLER EDITION WISSENSCHAFT
Daniel Beimborn

Cooperative Sourcing
Simulation Studies
and Empirical Data
on Outsourcing Coalitions
in the Banking Industry

With a foreword by Prof. Dr. Wolfgang König

GABLER EDITION WISSENSCHAFT

Bibliographic information published by the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

Dissertation Universität Frankfurt am Main, 2007

1st Edition 2008

All rights reserved
© Betriebswirtschaftlicher Verlag Dr. Th. Gabler | GWV Fachverlage GmbH, Wiesbaden 2008
Editorial Office: Frauke Schindler / Nicole Schweitzer
Gabler Verlag is part of the specialist publishing group Springer Science+Business Media.
www.gabler.de

No part of this publication may be reproduced, stored in a retrieval system

or transmitted, in any form or by any means, electronic, mechanical, photo-
copying, recording, or otherwise, without the prior written permission of the
copyright holder.

Registered and/or industrial names, trade names, trade descriptions etc. cited in this publica-
tion are part of the law for trade-mark protection and may not be used free in any form or by
any means even if this is not specifically marked.
Cover design: Regine Zimmer, Dipl.-Designerin, Frankfurt/Main
Printed on acid-free paper
Printed in Germany

ISBN 978-3-8350-0946-2
Foreword V

Foreword

“Make or buy?” The question about optimal sourcing is one of the oldest and
most central questions of managerial economics. Consequently, the question
about IT outsourcing has been an important research topic of the Information
Systems discipline for the past two decades. The financial services industry also
has a long tradition in IT outsourcing (ITO) with IT representing, besides people,
the only “production facility” of banks. The cost structure of non-physical prod-
ucts and services in the financial services sector relies heavily on fixed costs and,
thus, holds high potential for the inter-organizational bundling of processes and
achieving cost savings from economies of scale and skill.
In this book, Dr. Beimborn advances the traditional academic view on IT
outsourcing towards a sourcing network perspective. The complexity in this
paradigm, which adds to the complexity of the traditional outsourcing research
perspective and, furthermore, generalizes it to a business process outsourcing
(BPO) perspective, consists in the fact that typically more than two parties will
negotiate and that the roles of insourcer and outsourcer are not necessarily pre-
defined in advance. It is striking that in the otherwise generally mature outsourc-
ing literature the sourcing network issue has been almost completely ignored so
far. The potential impact becomes clear when realizing that both transaction
volumes in similar processes and process competencies are likely to be prevalent
in many firms of an outsourcer’s industry as well, raising the old co-opetition
question of who really is a competitor and who should rather be seen as a part-
ner. The research challenge thus is to explain and guide decision making from
this multilateral perspective and to disclose the market effects resulting from
those decisions.
Daniel Beimborn gives a thorough conceptualization and foundation for co-
operative sourcing research by addressing a triad of research questions: why to
source cooperatively, how to source cooperatively, and what are the resulting
consequences from cooperative sourcing? To answer these challenging ques-
tions, Dr. Beimborn makes use of a multi-theoretical foundation and a compre-
hensive multi-method approach. The core of his work consists in developing a
mathematical agent-based model which is applied to both analytical game-
theoretical analyses and simulation studies. Moreover, the author carried out
extensive empirical research which enabled him both to empirically test his ar-
guments and to feed the simulation studies with real world data.
VI Foreword

The author makes an important theoretical contribution by applying network

effect theory to the cooperative sourcing phenomenon and investigating how
externalities – decision interdependencies between the cooperatively sourcing
firms – affect the sourcing decision and the resulting outcomes of sourcing coali-
tions. He identifies and distinguishes different structural inefficiency sources,
such as individual vs. system lock-in, which can emerge on account of this inter-
dependence. This not only has a theoretical impact but also contributes directly
to the manager’s sourcing decision calculus. As we can see from the dynamic
markets for securities handling or payments processing in Germany, where
changing the provider or coalition is not unusual, decisions of sourcing partners
leaving a coalition can significantly affect the others’ cost situation of others and
therefore must be anticipated and included in the sourcing decision. If a firm
cannot trust in the long-term stability of a cooperative sourcing venture, a start-
up dilemma will emerge.
Another important part of this work shows how coalition stability can be en-
forced. By applying a game-theoretical cost analysis, the author shows how coa-
lition costs have to be allocated to the cooperative sourcing partners in order to
form a stable coalition which ensures a lasting cooperative behavior of all par-
ties. Overall, Dr. Beimborn provides a first theoretical foundation for the analysis
of the existence and efficiency of cooperative sourcing equilibria.
Although this work shows that cooperative sourcing is a complex issue, it
also illustrates that it has tremendous potential benefits. Building on and extend-
ing advancing outsourcing research, the author develops important building
blocks towards a theory of cooperative sourcing, which addresses the complex
mutual feedback between microbehavior and macrobehavior as discussed by
Nobel laureate Thomas Schelling (1978).
The work is impressive on account of its methodologically elaborate struc-
ture and analysis. The sound combination and integration of developing a formal
model and conducting analytical studies, simulations, and empirical research is
very remarkable and rare in the literature, contributing to a new perspective on
the sourcing phenomenon and providing the community with great opportunities
for future research.
The results of Daniel Beimborn, who has also published several articles in
scientific top journals of the Information Systems community, provide a solid
foundation for future cooperative sourcing research and should be read by any-
one who intends to further investigate multilateral sourcing networks.

Prof. Dr. Wolfgang König

Foreword VII

Foreword

This is another impressive success from the Johann Wolfgang Goethe’s Univer-
sity’s E-Finance Lab. Following the tradition of his predecessors, Dr. Beimborn
has put together an intriguing piece of research involving what he calls “coopera-
tive sourcing” which involves looking at the network of relationships that exist in
outsourcing. Using the German financial industry as his base (particularly the
industry’s credit business), Dr. Beimborn explores how these networks of sourc-
ing relationships can effect banks’ competitive positions. His results show that
the German banking industry is still in the early stage of cooperative sourcing
but he further suggests how the market is likely to evolve over time. Fascinating.
Whilst Dr. Beimborn’s research may typically be thought to lie within the
domain of Business Process Outsourcing (BPO), it is much more than just an-
other BPO study. It is truly a tour de force. Employing an impressive array of
methodological methods and theoretical groundings, Dr. Beimborn doesn’t just
use one lens to analyze the cooperative sourcing landscape, he uses four: empiri-
cal, analytical, game theory and simulation. Any one of these would lead to in-
teresting insights but to have all four, makes this far more than a traditional piece
of research. Its originality, its attention to detail, its level of sophistication, its
comprehensiveness, its coherence, and its richness, makes this an excellent ex-
ample of what the IS field needs far more of: applied research which ties together
the best of what academics can offer with sound practical results. Dr. Beimborn
must be congratulated for this truly exceptional piece of research. It should be
required reading for the IS academic and practitioner communities.

Prof. Dr. Dr. h. c. Rudy Hirschheim

Preface IX

Preface

The phenomenon of IT outsourcing has drawn considerable attention among

researchers and practitioners since the early 1990s. About five years ago, the
scientific information systems (IS) community extended its research focus to
business process outsourcing, which is closely related to IT outsourcing but also
includes unique facets that offer new and exciting research opportunities. In
particular, a major challenge is known as “cooperative sourcing”: multiple firms
agree to merge specific parts of their business. Currently, this trend can be ob-
served in the financial services industry, where banks decide on merging their
payments processing, securities administration, or loans processing.
Cooperative sourcing extends existing outsourcing research by introducing a
multilateral perspective, which brings about a substantial level of additional
complexity and, thus, raises important research questions. What is the role of
externalities in these sourcing networks? How are decisions of multiple firms
dependent on each other? How can the stability of sourcing coalitions be en-
forced? What economic inefficiencies can evolve in this complex environment?
With this work, I want to contribute to answering these questions. Over the
past few years, I have had the opportunity to be involved in dozens of case stud-
ies and various industry surveys, which helped me gain great insights into the
financial services industry, its value chains, and its rationales behind outsourcing
and cooperative sourcing. These insights enabled and motivated me to investi-
gate the essential questions of why and how cooperative sourcing should be
established as a governance mode among firms and what the structural effects
would be. It is my hope that this work can both guide future research in this field
and provide decision makers in the banking industry with a more clearly struc-
tured and comprehensive decision framework for evaluating sourcing options.
This thesis would not have been possible without the help of many people,
who I’d like to acknowledge for their support throughout this lengthy project.
First of all, I would like to thank my academic advisor Prof. Dr.
Wolfgang König for guiding me through the dissertation process over the past
few years and offering me a creative and inspiring “home” at the Institute of
Information Systems at Johann Wolfgang Goethe University in Frankfurt am
Main. Likewise, many thanks go to Prof. Dr. Peter Gomber for acting as co-
referee for my dissertation thesis and to Prof. Dr. Rainer Klump and Prof. Dr.
Roland Holten for acting as further members of my dissertation committee.
X Preface

As a former member of PhD Program 492 “E-Commerce: Infrastructures for

the Electronic Market”, chaired by Prof. Dr. Alejandro Buchmann at Technical
University of Darmstadt, I am also indebted to the German Research Foundation
(Deutsche Forschungsgemeinschaft) for their funding. Similarly, I would like to
express my gratitude for the financial support of the E-Finance Lab at Goethe
University in Frankfurt, which enabled me to do extensive empirical research.
My academic progress could not have been achieved without the help of
Prof. Dr. Tim Weitzel, who was always there to guide me since my days as a
diploma candidate. I am very thankful for his great and continuous support while
I was working on my thesis, but also for his general guidance in many other
research projects. I am greatly looking forward to continue our work in various
exciting areas of research and to write many more papers together with him
Special thanks go also to Prof. Dr. Rudy Hirschheim and Prof Dr. Andrew
Schwarz at Louisiana State University in Baton Rouge who gave me the oppor-
tunity to spend time in their department and doing empirical research in the US.
Furthermore, I am very grateful to all my friends and former colleagues in
Frankfurt, who supported me in writing my thesis and preparing my defense, or
who worked with me on other research projects: Dr. Roman Beck, Dipl.-Kfm.
Stefan Blumenberg, Dr. Rainer Fladung, Dipl.-Kfm. Jochen Franke, Dr. Norman
Hoppen, Dipl.-Kfm. Sebastian F. Martin, Dr. Gregor Schrott, and Dr. Heinz-
Theo Wagner. In particular, I want to thank Dipl.-Vw. Christopher Müller for his
ambitious and reliable support as student helper for more than four years – even
at nights and on weekends.
I am especially indebted to my friend Jochen Franke, who breathed his last
in December 2006, passing away far too early. We had great plans for a joint
future in the academic world and I sorely miss him.
Many thanks go to Michael Schmid and Lisa DeCerchio for carefully proof-
reading the manuscript. Furthermore, I would like to thank all those students who
worked with me on this research topic: Ralf Borger, Matthias Frechen, Martin
Heymann, Simone Kraut, Dennis Ungewitter, and Christoph Groß.
Finally, I feel especially indebted to my family. Special thanks go to my
parents who opened the way so that I could have an excellent education and who
encouraged me in my efforts. In particular, I want to thank the most important
person in my life, my wife Heike, who paid a high price for this book. She al-
lowed me to have lots of free space, especially after our daughter Nina Shania
was born in 2006. Many thanks for your invaluable support and patience, without
which I would not have been able to complete this work. Life is wonderful with
you and Nina and I dedicate this book to both of you.

Dr. Daniel Beimborn

Table of Contents XI

Table of Contents

Variables and Symbols ................................................................................... XVII

Abbreviations.................................................................................................XXIII
1 Introduction.................................................................................................... 1
1.1 Cooperative Sourcing in the Banking Industry................................... 3
1.2 Motivation and Research Questions ................................................... 4
1.3 Theoretical Foundation and Methodology.......................................... 8
1.4 Contribution and Main Findings....................................................... 11
1.5 Basic Terms ...................................................................................... 17
1.5.1 Processes, Activities, Business Functions, and the Value Chain.............17
1.5.2 Interorganizational Relationships............................................................18
1.5.3 Outsourcing and Cooperative Sourcing ..................................................19
1.5.3.1 Outsourcing..................................................................................19
1.5.3.2 Cooperative Sourcing...................................................................21
1.5.4 Financial Service Firms ..........................................................................22
1.6 Thesis Structure ................................................................................ 25
2 Theoretical Foundation and Related Research............................................. 27
2.1 Theoretical Foundation for Cooperative Sourcing Research ............ 27
2.1.1 Production Cost Economics ....................................................................29
2.1.1.1 Basics ...........................................................................................29
2.1.1.2 Implications for the Sourcing Decision........................................31
2.1.1.3 Empirical Evidence of PCE in the Banking Industry ...................33
2.1.2 Transaction Cost Economics...................................................................37
2.1.2.1 Basics and Discussion ..................................................................37
2.1.2.2 Implications for the Sourcing Decision........................................42
2.1.3 Agency Theory .......................................................................................46
2.1.3.1 Basics and Discussion ..................................................................46
2.1.3.2 Implications for the Sourcing Decision........................................49
2.1.4 Theory of Incomplete Contracts..............................................................51
2.1.4.1 Basics and Discussion ..................................................................51
2.1.4.2 Implications for the Sourcing Decision........................................53
2.1.5 Porter’s Positioning Framework .............................................................54
2.1.5.1 Basics and Discussion ..................................................................54
2.1.5.2 Implications for the Sourcing Decision........................................55
2.1.6 Resource-Based View and Core Competence View ...............................56
2.1.6.1 Basics and Discussion ..................................................................56
XII Table of Contents

2.1.6.2 Implications for the Sourcing Decision........................................60

2.1.7 Resource Dependency Theory ................................................................63
2.1.7.1 Basics and Discussion ..................................................................63
2.1.7.2 Implications for the Sourcing Decision........................................64
2.1.8 Relationship Theories .............................................................................65
2.1.8.1 Basics and Discussion ..................................................................65
2.1.8.2 Implications for the Sourcing Decision........................................69
2.1.9 Network Effect Theory ...........................................................................71
2.1.9.1 Basics and Discussion ..................................................................71
2.1.9.2 Implications for the Cooperative Sourcing Decision....................72
2.1.10 A Multi-Theoretical Perspective on Cooperative Sourcing............73
2.1.10.1 Theory Synthesis..........................................................................73
2.1.10.2 Business Functions as Unit of Analysis .......................................77
2.1.10.3 Cooperative Sourcing as Unit of Analysis ...................................78
2.2 Outsourcing Research....................................................................... 80
2.2.1 Overview.................................................................................................80
2.2.2 Outsourcing Drivers and Inhibitors.........................................................82
2.2.2.1 Outsourcing Drivers .....................................................................82
2.2.2.2 Outsourcing Inhibitors .................................................................88
2.2.2.3 A Literature Analysis of BPO Drivers and Inhibitors ..................94
2.2.3 Formal Models in the Context of Outsourcing........................................96
2.2.4 Summary.................................................................................................98
3 Cooperative Sourcing in the Banking Industry ............................................ 99
3.1 Current Situation in the German Banking Industry ........................ 100
3.1.1 Structural Deficits in the German Banking Industry .............................100
3.1.1.1 Fragmentation ............................................................................101
3.1.1.2 Overbanking...............................................................................102
3.1.1.3 Vertical Integration ....................................................................104
3.1.2 Current Tendencies ...............................................................................107
3.1.2.1 Consolidation .............................................................................108
3.1.2.2 Deconstruction ...........................................................................111
3.2 Segmentation in the Banking Industry............................................ 112
3.2.1 Generic Value Chain of the Banking Industry ......................................112
3.2.2 Segmentation Models............................................................................115
3.2.2.1 Three Segments Model ..............................................................115
3.2.2.2 Four Segments Model of Hamoir et al. ......................................117
3.2.2.3 Five Segments Model of Dombret and Kern..............................118
3.2.2.4 Conclusion .................................................................................120
3.3 Credit Process as Application Domain ........................................... 120
3.3.1 Overview of the Credit Market .............................................................121
3.3.1.1 Credit Products in the Retail Customer Business .......................121
3.3.1.2 Credit Products in the Corporate Customer Business.................122
3.3.1.3 Development of the German Credit Market ...............................122
Table of Contents XIII

3.3.2 Reference Processes for Process-Based Empirical Research ................124

3.3.2.1 MaRisk as a Foundation for Designing Credit Processes...........124
3.3.2.2 Reference Processes for the Credit Business..............................125
3.3.3 Credit Business Segmentation Model ...................................................131
3.3.3.1 Model .........................................................................................131
3.3.3.2 Banking Supervision Requirements ...........................................134
3.3.3.3 Effect of Bank Size and Sector Membership..............................135
3.4 Cooperative Sourcing in the German Banking Industry ................. 137
3.4.1 General Trends......................................................................................137
3.4.2 Outsourcing of Particular Business Processes.......................................141
3.4.3 Outsourcing of Credit Processes ...........................................................146
3.5 Regulatory Issues............................................................................ 152
3.5.1 General Requirements Related to BPO .................................................152
3.5.1.1 Section 25a of KWG and BaFin Circular 11/2001.....................152
3.5.1.2 Joint Forum – Outsourcing in Financial Services.......................154
3.5.1.3 Basel II .......................................................................................154
3.5.1.4 Particular Legal Domains...........................................................155
3.5.2 Specific Requirements for Credit Process Outsourcing ........................158
3.5.2.1 MaRisk – Minimum Requirements for Risk Management.........158
3.5.2.2 BaFin Memorandum on Credit Factories ...................................158
3.5.2.3 Requirements for Credit Process Outsourcing ...........................159
3.6 Empirical Evidence in the German Credit Business ....................... 160
3.6.1 Demographics .......................................................................................162
3.6.2 Characteristics of the Credit Process.....................................................166
3.6.2.1 Process Performance and Strategic Relevance ...........................167
3.6.2.2 Process Complexity....................................................................177
3.6.2.3 Modularity between Single Business Functions.........................178
3.6.2.4 Similarity of activities between different banks .........................184
3.6.2.5 Process Costs..............................................................................186
3.6.3 BPO Potential of the SME Credit Processes .........................................200
3.6.3.1 Actual State and General Trends in BPO ...................................200
3.6.3.2 Economies of Scale, Skill, and Scope – A PLS Approach .........214
3.7 Summary......................................................................................... 219
4 Developing a Formal Model for Cooperative Sourcing ............................. 221
4.1 Justification of Model Development............................................... 221
4.2 Derivation of the Cooperative Sourcing Model (CSM).................. 223
4.2.1 Actors and Business Functions .............................................................223
4.2.2 Business Neighborhood ........................................................................227
4.2.3 Cooperative Sourcing............................................................................230
4.2.3.1 Transaction Costs Related with Cooperative Sourcing ..............231
4.2.3.1.1 Negotiation Costs ....................................................................232
4.2.3.1.2 Coordination Costs ..................................................................233
XIV Table of Contents

4.2.3.1.3 Adoption Costs ........................................................................234

4.2.3.1.4 Interface Costs.........................................................................236
4.2.3.1.5 Agency Costs...........................................................................237
4.2.3.1.6 Summary .................................................................................238
4.2.3.2 Process Cost Effect ....................................................................238
4.2.3.3 Decision Calculus.......................................................................239
4.2.3.4 Constraints .................................................................................241
4.3 Centralized Model: Global Optimization........................................ 243
4.4 Decentralized Model: Autonomous Actor Decisions ..................... 247
4.5 Extending the Model by Legal and Regulatory Issues.................... 253
4.6 Summary......................................................................................... 255
5 Analytical and Simulative Studies ............................................................. 256
5.1 Game-Theoretical Analysis of Cooperative Sourcing .................... 256
5.1.1 Basic Concepts from Cooperative Game Theory ..................................257
5.1.2 Allocation Mechanisms.........................................................................259
5.1.3 Model....................................................................................................260
5.1.4 Analysis ................................................................................................262
5.1.4.1 Equal Allocation of Gain............................................................263
5.1.4.2 Proportional Allocation of Costs................................................264
5.1.4.3 Shapley Allocation .....................................................................267
5.1.4.4 Threshold for Transaction Costs ................................................268
5.1.5 Experiment............................................................................................270
5.1.6 Conclusion ............................................................................................274
5.2 A Genetic Algorithm for Solving the CSP...................................... 275
5.2.1 Basics....................................................................................................276
5.2.2 GA Design ............................................................................................277
5.2.3 Configuration ........................................................................................281
5.3 Simulation Studies .......................................................................... 282
5.3.1 Agent-based Simulations as Research Approach ..................................282
5.3.2 Simulation Procedure............................................................................285
5.3.3 Parameterization ...................................................................................289
5.3.3.1 Demographics ............................................................................290
5.3.3.2 Business Function Properties and Process Costs........................292
5.3.3.3 Transaction Costs .......................................................................297
5.3.3.4 Simulation and Optimization Control ........................................299
5.3.4 Simulation Results ................................................................................300
5.3.4.1 Results from the Basic Setting ...................................................301
5.3.4.2 Analysis of Process Cost Structure Variations ...........................326
5.3.4.3 The Impact of Transaction Costs................................................340
5.3.4.4 Variation of Additional Decision Calculus Parameters ..............354
5.3.4.5 Impact of Demographic Properties.............................................361
5.3.4.6 Impact of Simulation Control Parameters ..................................369
Table of Contents XV

5.3.4.7 Lock-in to Effects.......................................................................375

6 Conclusion ................................................................................................. 384
6.1 Summary of the Findings................................................................ 384
6.2 Contributions .................................................................................. 397
6.2.1 Implications for Theory ........................................................................397
6.2.2 Managerial Implications .......................................................................400
6.3 Validation of the Research Approach ............................................. 403
6.4 Limitations...................................................................................... 408
6.5 Further Research............................................................................. 412
References ........................................................................................................ 417
Appendix .......................................................................................................... 453
A1 – Performance Tests of the Genetic Algorithm................................ 453
A2 – Parameterization of the Simulation Studies .................................. 457
Variables and Symbols XVII

Variables and Symbols

* Firm i's vector of business functions: a*

ai i ^a i1 ,..., a ik ,..., a i|K| `
aik Indicates, whether actor i runs business function k (binary variable)
Parameter for stepwise changing the empirically derived ratio between fixed
alter_costRatio
and variable costs for conducting sensitivity analyses during the simulations
Parameter for stepwise changing variable unit costs for conducting sensitiv-
alter_cP
ity analyses during the simulation studies
Parameter for stepwise changing fixed costs for conducting sensitivity
alter_KF
analyses during the simulation studies
Parameter for stepwise changing all transaction cost parameters for conduct-
alter_TCcoefficients
ing sensitivity analyses during the simulation studies
alter_TCcoefficients Parameter for stepwise changing setup transaction cost parameters (cN and
_setup cAD) for conducting sensitivity analyses during the simulation studies
Parameter for stepwise changing the variation coefficients of all parameters
alter_vc
for conducting sensitivity analyses during the simulation studies
AvgPC Average process (or production) costs

bn, bnij Business neighborhood in general and in particular for the relationship of
firm i and j
Business neighborhood between i and insourcer of his business function k in
bn ikmt
period t
* *
bnijcust Customer portfolio overlap between i and j with bnij
cust
cos ci , c j 

bnijgeo Geographical business neighborhood between i and j with bnijgeo  ^0;1`

* *
bnijprod Product portfolio overlap between i and j with bnij
prod
cosai , a j 
bne Proportion of agency costs which is explained by business neighborhood
* Vector of customer segments served by j
cj *
with c j retail , sme, large  and retail , sme, large ^0;1`

Adoption costs
C AD , Cikmt
AD
(in general and in particular for firm i joining coalition km in period t)
ckAD Adoption cost basis
Agency costs
C AG , CikAG , Cikmt
AG
(in general and in particular for firm i joining coalition km in period t)
XVIII Variables and Symbols

ckAG Agency cost basis

AG
'C ijkmt Change of i's agency costs if firm j enters coalition km in period t

Coordination costs (in general and for firm i joining coalition km in period t
C C , C ikC , Cikmt
C
in particular)
C
'Cijkmt Change of i's coordination costs if firm j enters coalition km in period t
G G
C ,C ikmt Cost allocation mechanism in an alliance
IF IF IF
C ,C ,C it ikt Interface costs firm i has to bear when taking part in coalition km in period t
IF
c Interface cost basis

Negotiation costs (in general and for firm i joining coalition km in period t
C N , CikN , Cikmt
N
in particular)
ckN Negotiation cost basis
P
c Maximum variable unit costs of all firms and business functions

C P , CikP Process costs in general and of business function k at firm i, in particular

P P
c ,c
i ik Variable unit costs of (business function k) of firm i

C PM , C mPM , C kmt
PM Coalition process costs in general, of coalition m, and of coalition km in
period t, in particular
c mPM , c km
PM
Variable unit costs of coalition m or of coalition km
Variable unit costs of a sub-coalition of coalition m (in the game-theoretical
c mPS
analysis)
CR1, CR3 Market share of the largest and of the three largest coalitions
T
C im Transaction cost parameter in general (in the game-theoretical analysis)
T ,bilateral Individual transaction costs for managing a bilateral coalition
C im (in the game-theoretical analysis)
T ,trilateral Individual transaction costs for managing a trilateral coalition
Cim
(in the game-theoretical analysis)
DC Decision capacity
Exp>NPV @ Expected net present value
> @
Exp NS ikm t W  Expected individual periodical net savings

Exp>'PCS @ leave
j Expected change of j’s process cost savings if leaving the current coalition
Exp>'TC leave
j @ Expected change of j’s transaction costs if leaving the current coalition

Exp>3 alternative
iknt @ Expected individual benefit of i from entering an alternative coalition
Variables and Symbols XIX

Exp>3 enter
ikmt @ Expected individual benefit of i from entering coalition km in period t

Exp>'3 switch
ikt @ Expected individual net benefit of i from switching the current coalition
act
HF Actor-based Herfindahl index
HFvol Process volume-based Herfindahl index
homogenize Parameter for homogenizing the empirically derived ratio between fixed and
_costRatio variable costs for conducting sensitivity analyses during the simulations
homogenize Parameter for homogenizing the random number seed for process volumes
_procVol for conducting sensitivity analyses during the simulation studies
i, j Indices of firm with i, j  I
I, |I| Set and total number of firms

ĵ Index of insourcer

k, l Indices of business functions with k, l  K

K, |K| Set and total number of business functions
F F Fixed costs (of business function k in case of multiple business functions) of
K ,K i ik
firm i
KF Maximum fixed costs of all firms and business functions
FM FM
K m ,K km Fixed costs of coalition m (providing business function k)
Fixed costs of a (hypothetical) sub-coalition of coalition m (in the game-
K mFS
theoretical analysis)
K iG Part of the coalition costs to be borne by member i
LC Legal constraint
km Index of the m-th coalition providing business function k
mktmax Number of coalitions serving business function k in period t

M kmt , M kmt Set of coalition members and size of coalition km in period t

Mm, Mm Set (and number) of firms joining the m-th coalition

max(X) Maximum value of variable X

mean, mean(X) Arithmetic mean (of variable X)
min(X) Minimum value of variable X
mod Modulo function
Set of natural numbers
N Set of players
ND Normal distribution
XX Variables and Symbols

NPV Net present value

NS ikmt i’s individual periodical net savings being member of coalition km in t
stay
NSikmt i’s individual net savings resulting from staying in coalition km

oit Number of outsourcing projects firm i realized up to period t-1

p General abbreviation for probabilities

p CO Crossover probability (genetic algorithm)
leave
p jkm t W  Probability that firm j leaves coalition km in future period t+W
M
p Mutation probability (genetic algorithm)
PC(x) Process costs for process volume x
PCSikt Process cost savings for actor i's business function k in period t
PCSikmt Process cost savings for firm i in period t when joining coalition km
PCSit Total process cost savings
'PCS ijkmt Change of i's process cost savings if j leaves coalition km in period t
leave
'PCS j Change of j’s process cost savings if leaving the current coalition
r(X,Y) Correlation between stochastic variables X and Y
rad Risk-adjusted discount rate
s Number of players in sub-coalition S
S, Sm Sub-coalition (of coalition m)
sd, sd(X) Standard deviation (of variable X)
Maximum amount of core competencies which can be outsourced due to
SCi
strategic reasons
sizei Size of firm i with sizei = [0.0, 1.0]
t, W Period indices
T Total number of simulation periods
TCoSo Sourcing contract duration
TGA Number of generations in the GA
TC Total transaction costs with TC = CAD + CN + CC + CIF + CAG
C IF AG
TCp Total periodical transaction costs with TCp = C + C + C
'TCijkmt Change of i's transaction costs if j leaves coalition km in period t
leave
'TC j Change of j’s transaction costs if leaving the current coalition
Variables and Symbols XXI

Binary variable that indicates whether actor i's process k will be evaluated
uikt
regarding outsourcing options in period t
v Real valued function
v(i) Costs which player i has to bear before entering a coalition
vikmt Indicates, whether firm i is insourcer of alliance km in period t (binary)
VAT Value-added tax rate
vc Variation coefficient: vc = VP
x , xi, xik Process volume (of business function k ) of firm i
M
x kmt Process volume of coalition km in period t
M
x m Process volume of coalition m (in the game-theoretical analysis)
S Process volume of a sub-coalition (i.e. subset of members) of coalition m (in
x m the game-theoretical analysis)
Binary decision variable that indicates whether firm i outsources business
yikmt
function k to alliance km in period t
Binary variable that indicates whether firm i's business function k is oper-
zikmt
ated by alliance km in period t
diff
ziklm 1 Binary variable that indicates whether k and l are operated by different firms

Negotiation/coordination cost parameter (exponent) with 1<D<2 to consider

D
the effect of coalition size on negotiation/coordination costs
E Learning effect parameter (exponent) with E < 0
Fik Complexity of the business function

J Ratio between C
Cikmt N
and Cikmt with 0  J 1

Oik Core competence of firm i’s business function k

P Arithmetic mean
S Imputation of an n-player game
Costs which player i has to bear in the coalition (in the game-theoretical
Si
analysis)
3 Overall global net benefit from cooperative sourcing (net present value)
3ikmt* Individual net benefit of firm i in coalition km in period t
Degree of task interdependence between k and l of firm i with Tikl = [0.0;1.0[
Tikl
defined for l>k
V Standard deviation
Degree of similarity of business function k in firms i and j in period t with
]kijt
]kijt =[0;1] (]kijt = 1 means perfect similarity)
Abbreviations XXIII

Abbreviations

Aareal HM Aareal Hypotheken-Management GmbH (now: Kreditwerk HM)

(http://www.hypotheken-management.de)
ABC Activity-based costing
ABM Agent-based modeling
Abs. Absolute
ACE Agent-based computational economics
ACS Affiliated Computer Services, Inc.
AMEX American Express (http://www.americanexpress.com)
AMOS Analysis of Moment Structures (software package for conducting SEM)
(http://www.spss.com)
ASP Application service providing
Assessm. Assessment (of credit applications)
AT Agency theory
ATM Automated teller machine
AVE Average variance extracted
avg Average
AWD Allgemeiner Wirtschaftsdienst AG (http://www.awd.de)
B2B Business to business
BaFin Federal Financial Supervisory Authority
(Bundesanstalt für Finanzdienstleistungsaufsicht)
BDC Bounded decision capacity
BGB German Civil Code (Bürgerliches Gesetzbuch)
BIS Bank for International Settlements
BPO Business process outsourcing
BDSG German Federal Data Protection Act (Bundesdatenschutzgesetz)
BSpk Thrift institution (Bausparkasse)
BVR Federal Association of the German Credit Cooperatives
(Bundesverband der Deutschen Volksbanken und Raiffeisenbanken)
CAPM Capital Asset Pricing Model
CASE Series of case studies conducted by the E-Finance Lab in 2005
CCV Core competence view
XXIV Abbreviations

cen. centralized (coordination)

CIBC Canadian Imperial Bank of Commerce, Inc. (http://www.cibc.com)
CIR Cost/income ratio
CoBa Commerzbank AG (http://www.commerzbank.de)
CS Cooperative sourcing
CSC Computer Sciences Corporation (http://www.csc.com)
CSM Cooperative sourcing model
CSP Cooperative sourcing problem
CT Cooperation theory
DBB Federal Reserve Bank of Germany (Deutsche Bundesbank)
DEA Data envelopment analysis
dec. decentralized (coordination)
Dec. Decision
DeuBa Deutsche Bank AG (http://www.deutsche-bank.de)
DG Hyp Deutsche Genossenschafts-Hypothekenbank AG (http://www.dghyp.de)
DreBa Dresdner Bank AG (http://www.dresdner-bank.de)
DSGV German Public Savings Banks Association (Deutscher Sparkassenverband)
DtA Deutsche Ausgleichsbank
dwpbank Deutsche WertpapierService Bank AG (http://www.dwpbank.de)
ECB European Central Bank
ECJ European Court of Justice
EDI Electronic Data Interchange
EDS Electronic Data Systems Corporation (http://www.eds.com)
EFL E-Finance Lab (http://www.efinancelab.com)
EI Expert interviews
emp. empirical
EQS Software package for conducting SEM (http://www.mvsoft.com)
ERP Enterprise resource planning
etb European Transaction Bank GmbH (http://www.etb-ag.com)
EUR European currency (Euro)
EURO European currency area
F&A Finance and accounting
ForEx Foreign exchange dealing
FSO (German) Federal Statistical Office
GA Genetic algorithm
Abbreviations XXV

GAD Gesellschaft für automatische Datenverarbeitung eG (http://www.gad.de)

GT Game theory
GTA Gross total assets
GWB Act Against Restraints on Competition
(Gesetz gegen Wettbewerbsbeschränkungen)
HP Hewlett Packard Corporation (http://www.hp.com)
HR Human resources
HVB Hypovereinsbank AG (http://www.hypovereinsbank.de)
IAIS International Association of Insurance Supervisors
IBM International Business Machines Corporation (http://www.ibm.com)
ICT Information and communication technology
IDE Integrated development environment
IOR Interorganizational relationship
IOS Interorganizational (information) system
IOSCO International Organization of Securities Commissions
IPO Initial public offering
IS Information systems
IT Information technology
ITO Information technology outsourcing
ITS International Transactions Services GmbH (http://its-wertpapiere.de)
KfW Kreditanstalt für Wiederaufbau (Reconstruction Loan Corporation)
Kreditwerk Kreditwerk Hypotheken-Management GmbH (formerly: Aareal Hypotheken-
HM Management GmbH) (http://www.hypotheken-management.de)
KSK County-owned public savings bank (Kreissparkasse)
KWG German Banking Act (Kreditwesengesetz)
LISREL Software package for conducting SEM (http://www.ssicentral.com/lisrel)
M Sales unit in the credit business (Markt)
m million
M&A Mergers and acquisitions
MaK Minimum Requirements for the Credit Business of Credit Institutions
(Mindestanforderungen an das Kreditgeschäft der Kreditinstitute)
MaRisk Minimum Requirements for Risk Management
(Mindestanforderungen an das Risikomanagement)
MF Middle / back office in the credit business (Marktfolge)
MFI Monetary financial institution
MIPS Million instructions per second
XXVI Abbreviations

MNE Multi-national enterprise

MU Monetary units
N/A Not available
NET Network effect theory
NPL Non-performing loans
NPV Net present value
NS Net savings
OCSP Optimal Consortia Structure Problem
OpM Operational margin
OSGV Eastern German Public Savings Banks Association
(Ostdeutscher Sparkassenverband)
OTC Over the counter (inter-bank trading of securities, without stock exchange as
intermediary)
PAT Principal agent theory
PCA Principal component analysis
PCE Production cost economics
PCS Process cost savings
PLS Partial Least Squares
PoBa Deutsche Postbank AG (http://www.postbank.de)
p.p. Percentage points
PPF Porter’s positioning framework
Prep. Preparation
Proc. Processing
PSP Processing service provider
R&D Research and development
RBS Royal Bank of Scotland PLC (http://www.rbs.co.uk)
RBV Resource-based view
RDT Resource dependency theory
Rel. Relative
Risk man. Risk management
Risk mon. Risk monitoring
ROE Return on equity
RT Relationship theories
S1, S2 Survey-based studies of the E-Finance Lab in 2004 and 2005
SOA Service-oriented Architecture
sd, st. dev. Standard deviation
Abbreviations XXVII

SDC Sparkasseninternes Datacenter A/S (http://www.sct.dk)

SDV Sparda-Datenverarbeitung eG (IT service provider of the Sparda Group)
(http://www.sparda.de/spardagruppe_sdv_index.html)
SEM Structural equation modeling
Serv. Servicing
sim. simulated
SLA Service level agreement
SLM Service level management
SME Small and medium-size enterprises
SOX Sarbanes-Oxley Act
SSK City-owned public savings bank (Stadtsparkasse)
STP Straight-through processing
TC Transaction costs
TCE Transaction cost economics
TIC Theory of incomplete contracts
TSB (Lloyds) Transaction Services Bank (http://www.lloydstsb.com)
UK United Kingdom
US, USA United States (of America)
VAR Value-added ratio
VAS Value-added to sales index
VAT Value-added tax
vc Variation coefficient (ratio between standard deviation and mean)
VGR National accounts (Volkswirtschaftliche Gesamtrechnung)
VIC Vertical industry connection index
Introduction 1

1 Introduction

IT outsourcing (ITO) has become a tool that is frequently used by firms for re-
ducing their portfolio of activities and achieving, among other things, economies
of scale and skill by using a specialized provider. Because of their high level of
IT reliance and business processes that are mostly fully digitizable, the value
opportunities offered by outsourcing are especially attractive for financial ser-
vices firms. Accordingly, Hirschheim and Dibbern (2002) found the first re-
markable ITO deal in 1963 when insurance company Blue Cross of Pennsylvania
outsourced its data processing to EDS. However, despite many other industries
having utilized advances in information and communication technologies – as a
subsequent step – to restructure their value creation, to outsource, automate, and
integrate business processes, and to form value networks, there is still substan-
tially less disintegration of value chains in the financial services industry. As a
consequence, experts assume that there are significant efficiency potentials in the
industry.
From an academic perspective, outsourcing has developed into quite a ma-
ture research area over the decades, especially in the Information Systems (IS)
discipline, and now also includes business process outsourcing (BPO) (Currie et
al. 2003; Dayasindhu 2004; Holzhäuser et al. 2004; Rouse and Corbitt 2004;
Weitzel et al. 2004; Willcocks et al. 2004). Nevertheless, outsourcing literature is
almost exclusively concerned with 1:1 outsourcing relations and, quite surpris-
ingly, has only very marginally incorporated the possibility of in- and outsourc-
ing networks. Considering that most outsourcing success factors are at least
partially driven by what others do, like for example integration costs, transaction
volume etc., it would seem appropriate to explicitly consider sourcing networks
as a relevant research domain. If one goes a step further from ITO to BPO and
asks what firm is likely most competent to carry out part of a primary business
process for an outsourcing bank, one might certainly expect a different bank to
qualify. Much empirical evidence for this can be found in the German banking
industry, for example (cf. section 3.4).
From a theoretical perspective, this cooperative sourcing a) is a multilateral
instead of a bilateral agreement; b) focuses on core activities being cooperatively
sourced; c) is a close form of horizontal interorganizational cooperation and
therefore inherently represents a form of coopetition as coalition partners usually
are (potential) competitors; and d) in contrast to traditional outsourcing relation-
ships relies on the benefits being allocated by negotiation rather than by market
2 Introduction

mechanism which can have a substantial impact on the stability and efficiency of
the coalitions. Thus, extending the scope of analysis from one insourcer and one
outsourcer from different markets to multiple stakeholders mostly within one
market adds a substantial degree of complexity as coalition building and com-
petitive proximity, to name just a few phenomena, now need to be considered.
This makes cooperative sourcing a theoretically important and relevant research
topic. Therefore, the key questions guiding this work are:
o How can cooperative sourcing networks help individual firms and entire
networks, like industry branches, to improve their value creation?
o What are the effects of cooperative sourcing networks on a bank’s cost and
competitive situation, and how can cooperative sourcing strategies aid in
improving them?
o What are the conditions for stable cooperative sourcing networks?
o What is the effect of cooperative sourcing networks on industry dynamics
and market structure/segmentation?
Using a multi-method (empirical, analytical, game theory, simulation) and
multi-theoretical approach to incorporate a network perspective into outsourcing
research, a formal model is developed and used for game theory and simulation
analyses together with empirical data from the German financial services indus-
try. It is argued that, in contrast to the conventional wisdom on ITO, on the one
hand, BPO is associated with lower (hidden) transaction costs thus making BPO
more likely. But on the other hand, selective outsourcing that is often suggested
for ITO leads to process-oriented diseconomies of scope for BPO which has been
analytically and empirically shown to be a substantial inhibitor to BPO potential.
Using the credit business as our application domain, we reveal why substantial
parts of the German banking industry are still in a comparatively early sourcing
network stage and how the market can be expected to evolve.
Based on the cooperative sourcing model, it is shown which cost allocations
lead to stable coalitions, i.e. ensuring the participation of all potential members.
While mostly neglected so far, cost and benefit allocations in networks are cru-
cial for their stability and efficiency. The analysis reveals that, depending on the
amount of coordination costs, only the process volume-proportional distribution
ensures stable coalitions. This is surprising as the Shapely allocation which takes
power asymmetries of the participants into account is, in particular, unable to
guarantee stable sourcing networks. As a consequence, although the proportional
cost allocation turned out to be the only stable allocation, experiments conducted
alongside the analysis reveal that most of the participants did not agree to a cost
allocation which is close to it, but preferred the Shapley value, which sometimes
even resulted in inefficient outcomes.
Introduction 3

The subsequent simulation analysis shows that if process cost structures are
relatively heterogeneous throughout the industry, benefits from cooperative
sourcing primarily result from economies of skill rather than from economies of
scale. As a consequence, even small coalitions can be beneficial and compensate
for the transaction costs, value-added taxes, and business risks that arise as a
result.
The thesis contributes to the literature in four major areas: First, by extend-
ing the 1:1 outsourcing view by a network perspective that draws from network
effect theory; second, by offering a formal approach to understanding and direct-
ing sourcing networks; third, by providing an analytical game theory approach to
determine conditions for stability of sourcing coalitions; and fourth, by combin-
ing these theoretical and methodical results into a single simulation model that
can be used to determine advantageous sourcing coalitions and to anticipate
market dynamics in order to identify unexploited firm and industry benefits.
The remainder of this chapter is structured into introducing the object of
analysis (section 1.1), motivating and deriving the research questions (1.2), giv-
ing a brief overview of the theoretical and methodological fundament (1.3), and
summarizing the key results (1.4). After defining the main terms and concepts
used in this work (1.5), the chapter concludes with an overview of the structure
of the overall thesis (1.6).

1.1 Cooperative Sourcing in the Banking Industry

The competitiveness of the German banking industry has been frequently dis-
cussed in recent years. The relative underperformance of German banks in terms
of profitability and cost efficiency is usually traced back to high fragmentation
(too low market concentration), overbanking (too many banks and branches), and
high vertical integration (Dombret 2004; Eichelmann et al. 2004; Moormann and
Möbius 2004; Weber 2002). As response, German banks have reshaped their
organizational borders by merging and outsourcing during recent decades (Wal-
ter 2001; IBM 2003). The financial services industry is carrying out a transfor-
mation process which dismantles the borders of the traditional universal bank
and creates new forms of cooperation – even between former competitors (Mar-
lière 2002). The value chain of the banking industry, which was characterized by
a single institution in the past, is becoming disintegrated with the various activi-
ties allocated to different, specialized service providers. Furthermore, the avail-
ability of mature B2B information technologies, which enable straight-through
processing across the boundaries of the value chain partners, enables and drives
these organizational disintegration tendencies (Englert 2000).
4 Introduction

After undergoing a strong consolidation phase – the number of German

banks decreased from 4,500 to less than 2,100 over the last twenty years (DBB
2008) –, the financial institutions started to consolidate parts of their business in
joint subsidiaries or to outsource them to other banks specialized in operating
these activities. This cooperative sourcing of activities has been shown to be a
valid and promising option for breaking up the value chain and for improving
efficiency by exploiting economies of scale and skill (cf. section 3.4).
One of the largest cooperative sourcing deals in Germany took place in
2004, when two of the largest commercial banks outsourced their domestic retail
payments processing to a third one, which now operates one fifth of the overall
domestic payments processing volume in Germany (Buhl et al. 2005). Further-
more, the majority of the remaining 80% is provided by only very few other
“transaction banks” (cf. section 3.4.2). A similar mature market can be found for
securities processing. By contrast, other core activities, such as the credit busi-
ness, show a rather slow and cautious trend with only very few banks outsourc-
ing parts of these activities – although this has been common for many years in
other countries like the USA, the UK, or the Netherlands.

1.2 Motivation and Research Questions

Cooperative sourcing of business activities has been shown to be a major strat-
egy for firms in order to stabilize their competitive position by reducing the size
of their firm without shrinking their product portfolio. The revolutionary trans-
formation in the banking industry and the different developments and patterns
provided the initial motivation to explore the cooperative sourcing phenomenon
in this thesis.
Cooperative sourcing extends the concept of “traditional” outsourcing to-
wards a horizontal and multilateral cooperation, having the following character-
istics:
o Cooperative sourcing focuses on primary business activities. Consequently,
the outsourcer and insourcer of the sourcing relationship are not pre-defined
by their distinct overall business models. Instead, a group of structurally
similar firms either founds a subsidiary or one of the firms decides to be-
come the insourcer.
o From a theoretical perspective, cooperative sourcing is a multilateral instead
of a bilateral agreement. As a consequence, the strategic situation of coop-
erative sourcing is even more intricate than in traditional 1:1 outsourcing re-
lations as there now is the possibility of coalition building. This has a sub-
stantial impact on coalition stability and cost allocation rules.
Introduction 5

o In contrast to traditional outsourcing relationships, the benefits are allocated

by negotiation rather than by market mechanism.
o Aligned interests: Since the insourcer firm not only provides services for
other firms but also for its own needs, it has similar interests in effective and
efficient process management as the outsourcers. Thus, there is a lower in-
centive for moral hazard and resulting service debasement problems.
o Since cooperative sourcing represents a form of horizontal cooperation, it
takes place between (potential) competitors. Thus, it inherently is a form of
coopetition.
Whereas “traditional” outsourcing has been well researched over the last 15
years, particularly in the information systems domain, cooperative sourcing of-
fers a range of additional important and exciting research opportunities. While
cooperation from a general perspective is tackled by research on alliances or
“hybrid” organizational structures, placed between hierarchical and market
mechanisms of coordination (Williamson 1991), the particular concept of coop-
erative sourcing is covered by this research work in oder to contribute both a
theoretical extension to outsourcing research and a managerial contribution for
successful management of the transformation process in the banking industry.
Why do firms decide for (or against) cooperatively sourcing a certain busi-
ness function? What are the determinants, drivers, and inhibitors which influence
cooperative sourcing behavior compared with insourcing or traditional outsourc-
ing to a third party vendor? When multiple firms decide to cooperatively source
a business function, how should the benefits be distributed? Since the partners
are usually not perfectly similar and bring different resources, capabilities, and
process volumes into this emerging cooperative sourcing coalition, it has to be
determined how the gains should be distributed to guarantee a stable partnership.
Cooperative sourcing activities lead to significant changes in the relevant
market structure. For example, we can observe strong consolidation tendencies in
transaction banking which might lead to very few powerful players (or maybe
only one) in particular market segments (e.g. payments or securities processing).
Which determinants will lead to which market effects? Which market structures
can be anticipated? How efficient are they from an economic perspective, and
how should decision makers react to upcoming cooperative sourcing dynamics in
their markets?
To provide initial answers in approaching this complex research object, the
following three research questions are chosen to outline the scope of this work:
6 Introduction

theoretical 1 2 3
What are the drivers and Which cost allocation enables Which segmentation effects
implications
inhibitors of cooperative stability of cooperative can be expected? How stable
sourcing? What is their sourcing coalitions? and how desirable are they?
differential impact?

Managerial implications:
How can managers in the banking
industry be supported in making
managerial decisions about cooperative sourcing
implications opportunities?
IS economic
focus focus
Figure 1: Research questions and managerial implications
These questions can be divided into more detailed questions which guide the
reader through this thesis:
1) Drivers and inhibitors:
o What are the relevant theories and associated influential factors which
help to explain cooperative sourcing?
o How can the relevant theories be aggregated to provide an integrated
framework of drivers and inhibitors for cooperative sourcing?
o What is the role of competitiveness between partners? What is the role of
externalities?
2) Cooperation stability:
o What are the conditions for cooperation stability?
o Which allocation mechanisms lead to cooperation stability?
3) Market effects:
o What are the effects of the identified determinants on cooperative sourc-
ing behavior and on the resultant savings?
o How do market dynamics (outsourcing, switching of sourcing coalitions,
and backsourcing) evolve over time and how are they affected?
o What are the resulting segmentation effects, given certain process charac-
teristics?
o How do externalities and bounded rationality affect the efficiency of co-
operative sourcing decisions?
The design of this research is further motivated by three issues highlighted
by the scientific community:
o In outsourcing research, the development of formal and quantitative decision
support models which capture a huge range of relevant influencing factors is
still lacking (Buhl et al. 2005; Weill and Ross 2005). In this work, a model
is developed which focuses on capturing factors relevant to cutting and
Introduction 7

modularizing financial processes into services and determining optimal co-

operative sourcing strategies for these. While many papers focus on empiri-
cal research and investigate singular or disjoint causal relationships between
IS (information systems) characteristics and outsourcing decisions, this
model tries to capture different characteristics simultaneously and to handle
the resulting complexity by applying simulation methods.
o In the outsourcing literature, industry-specific research is increasingly de-
manded and it is expected to deliver more valuable and concrete research re-
sults (Hirschheim 2003, Dibbern et al. 2004, 87). Consequently, this re-
search work strongly focuses on a particular application domain and applies
cooperative sourcing in the (German) banking industry as its empirical
foundation throughout the whole work.
o The IS community already provides a comprehensive range of literature
explaining the degree of IT outsourcing. During the last two decades, many
authors have identified determinants of IT outsourcing decisions (e.g. Hu et
al. 1997; Lacity et al. 1994; Lacity and Willcocks 1995; Lacity et al. 1995;
Loh and Venkatraman 1992a; Loh and Venkatraman 1992b; Saunders et al.
1997) and success factors of outsourcing deals (e.g. Lacity and Hirschheim
1993; Ang and Straub 1998; Feeny and Willcocks 1998; Hu et al. 1997). As
a result, researchers increasingly call for the focus of outsourcing research to
be extended towards the outsourcing of business activities (i.e. BPO) instead
of only considering the IT layer of the firm (Dibbern et al. 2004, Hirschheim
2003, Rouse and Corbitt 2004).
Therefore, the main objective of this research is to provide a generic formal
model for the cooperative sourcing of business functions as well as empirical
data and simulation studies for the banking industry as a particular and relevant
application domain.

Outsourcing in Information Systems Research

Especially in terms of the scope and theoretical background, this thesis addresses
a core objective of information systems (IS) research, i.e. to answer questions
about the organizational effect of the employment of information systems (e.g.
Gurbaxani and Whang 1991; Clemons et al. 1993; Clemons and Reddi 1994).
Since IS are the essential enabler of efficient interorganizational cooperation and
outsourcing of activities (Malone and Rockart 1991; Snow et al. 1992; Hammer
and Champy 1993; Mertens et al. 1998), they lead to a significant organizational
transformation of the affected industry and its member firms. Transaction costs,
which determine the optimal governance mode (hierarchy vs. cooperation vs.
8 Introduction

market) (Williamson 1985, 1991), are strongly determined by information sys-

tems (Brynjolfsson et al. 1994; Clemons et al. 1993; Barney 1999; Afuah 2003).
A further reason why the research domain of cooperative sourcing – at least
in the banking industry – should be addressed by IS research stems from the
production side: IS are the critical resource in the banking business (Ang and
Straub 1998; Roy and Aubert 2002; Schott 1997, 131). Banks’ process cost
structure – i.e. the ability to achieve economies of scale on the one hand and
interdependencies between business functions (economies of scope) on the other
hand – are primarily affected by IS characteristics and IT costs (Adams et al.
2002; Holzhäuser et al. 2004; Hughes 1999; Köhler and Walter 2002; Lamberti
and Pöhler 2004; Loetto et al. 2003, 34; Schmiedel et al. 2002). Consequently,
several authors define business process outsourcing as the outsourcing of IT-
intensive business processes (Dayasindhu 2004; Gartner Group 2002; Rouse and
Corbitt 2004) because the advantageousness of outsourcing is often driven by IT-
related cost advantages. In addition, Sperber and Günther (2004) state that coop-
erative sourcing in the banking industry is an “IT-driven business” (see also Nitz
et al. 2004). Determining the optimal sourcing provider in a cooperative sourcing
coalition is strongly based on selecting the partner with superior information
systems, which covers “a mixture of assets and capabilities formed around the
productive use of information technology” (Wade and Hulland 2004, 132).
Research on outsourcing has almost exclusively taken place in the IS re-
search domain. Dibbern et al. (2004) carried out an extensive literature review
which showed that almost all articles related to outsourcing (including all major
business journals!) only tackled the IT outsourcing phenomenon. As already
mentioned above, it is consequently claimed that BPO has to be more thoroughly
explored (Dibbern et al. 2004, 88; Hirschheim 2003) and that outsourcing issues
should be considered more generically to answer the needs of practitioners who
are not willing to accept unnecessary and maybe artificial borders between IT
outsourcing and BPO (for a differentiation of the terms see section 1.5.3.1).
Recently, BPO has been increasingly addressed by IS journals and conferences
(e.g. Currie et al. 2003; Dayasindhu 2004; Feeny et al. 2003; Holzhäuser et al.
2005; Mani et al. 2006; Rouse and Corbitt 2004; Weitzel et al. 2004; Willcocks
et al. 2004).

1.3 Theoretical Foundation and Methodology

Academic research is based on three interdependent basic elements, called the
“triad network of justification” (Laudan 1984, 63) or “diversity of research”
(Robey 1996; Benbasat and Weber 1996) – being aims, theories, and methods.
Introduction 9

Aim refers to the research problem and the research questions addressed by the
research. The aim of this work has been defined in the previous section.
Theories represent the conceptual foundation which is applied to address the
research question (Landry and Banville 1992). They provide orientation in cap-
turing a complex real phenomenon, although a single theory is not able to pro-
vide an all-embracing explanation of the real world (Marlière 2002). Hence, this
work adopts several theoretical perspectives to tackle the research questions.
They can be classified into economic theories (production cost economics, trans-
action cost economics, agency theory, incomplete contract theory, and network
effect theory), organizational theories (resource dependency theory and relation-
ship theories), and strategic theories (Porter’s market-based view, resource-based
view, and core competence view.) These are briefly discussed and linked to the
cooperative sourcing phenomenon in section 2.1.
From a theoretical standpoint, the multilateral character of cooperative
sourcing, compared with “traditional” outsourcing, requires the incorporation of
decision interdependencies. Since efficient and effective cooperative sourcing
strategies rely on the agreement of multiple partners, these interdependencies are
covered by incorporating cooperative game theory and network effect theory in
addition to the economic and organizational theories usually applied in outsourc-
ing research. This extends the classical view of the firm (and on outsourcing)
which focuses on the trade-off between production cost economies and transac-
tion costs (Coase 1937, Williamson 1985, 1991) or on strategic issues (Prahalad
and Hamel 1990; Cheon et al. 1995; Pfeffer and Salancik 1978) for determining
the optimal size and shape of the firm.
As the third basic element of the “triad”, methods represent the techniques
which are used to answer the research questions. Depending on the research
questions, an appropriate methodology for conducting research has to be chosen.
Basically, knowledge can be created by empirical and non-empirical ap-
proaches (Alavi et al. 1989; Dibbern et al. 2004). Empirical approaches are
based on evaluating any sort of data resulting from surveys, case studies, action
research, experiments, etc. Following Dibbern et al. (2004) and Orlikowski/Ba-
roudi (1991), there are three different empirical approaches in IS research: posi-
tivist, descriptive, and interpretive.
Research “can be classified as positivist if there is evidence to formal propo-
sitions, quantifiable measures of variables, hypothesis testing, and the drawing of
interferences from a representative sample to a stated population” (Klein and
Myers 1999). By contrast, descriptive research primarily investigates relation-
ships between various empirical constructs by means of a rather explorative and
straight-forward analysis (Dibbern et al. 2004). As a third empirical approach,
interpretive studies neglect the existence of “objective” and “factual” events and
10 Introduction

situations (Orlikowski and Baroudi 1991). They use methods which are “aimed
at producing an understanding of the context of the information system, and the
process whereby the information system influences and is influenced by the
context” (Walsham 1993, 4-5).
Non-empirical research can be sub-divided into conceptual and mathemati-
cal methods. While conceptual research typically designs classification frame-
works or develops decision support models for developing guidelines and other
forms of decision support, mathematical approaches try to formalize and analyze
problem structures and therefore often develop highly abstract mathematical
models (Dibbern et al. 2004).
Another approach, which is rather difficult to classify within the given cate-
gorization, are simulations. These have to be placed between empirical and non-
empirical research approaches because, on the one hand, simulation studies in-
clude the observation and analysis of data, but on the other hand, the data is
artificially created, based on formal models (non-empirical approaches) (cf.
section 5.3.1).
The main approach of this work is to develop a formal model of cooperative
sourcing, which is based on a thoroughly developed theoretical foundation, and
to use it for both analytical (mathematical) analyses as well as for simulation
studies. The model consists of decision functions for a system of autonomous
firms which independently decide on cooperatively sourcing business functions
and thereby mutually affect their decision-relevant environment. Based on this
agent-based modeling approach, the appropriate mathematical method stems
from game theory (cf. section 5.1 for analytically determining the cost allocation
conditions of coalition stability) while the simulation part of this work is carried
out by applying the agent-based computational economics (ACE) paradigm in-
troduced later on in section 5.3.1. Although a purely analytical approach would
be the best choice (Axelrod 2000), this is not possible due to the complexity of
the interdependent system of cooperative sourcing determinants and the aim of
analyzing not only the static effects but also the dynamics of the system when it
moves towards a stationary state. A discussion on applying simulations is given
in section 5.3.1.
Moreover, this work includes empirical studies for gathering evidence and
for parameterizing the simulation model. Based on two large cross-sectional
studies in the German banking industry, actual trends in cooperative sourcing in
a particular business segment (SME credit business) as well as the corresponding
drivers and inhibitors are analyzed (descriptive approach, cf. section 3.6), and a
theory-based causal model regarding the impact of production cost economic
determinants on the BPO potential is tested (positivist approach, cf. section
3.6.3.2). Thus, this thesis applies a complementary multi-method approach to
Introduction 11

provide valid answers to the proposed research questions. The various methods
applied are described in detail in each of the corresponding sections.
The empirical perspective focuses on the German SME credit business al-
though we can find many similar trends in other countries and business seg-
ments, which often occurred much earlier (Marlière 2002). The reason for this
choice is that it is part of the research activities of the E-Finance Lab1 which
offered great opportunities to gain access to resources, data, and case study part-
ners which otherwise would not have been achievable to that extent.

1.4 Contribution and Main Findings

By answering the research questions, this work offers several contributions to
theoretical work as well as to practitioners’ decision behavior regarding coopera-
tive sourcing.
In this thesis, the concept of cooperative sourcing is theoretically developed
and empirically analyzed. Therefore, the research streams dealing with outsourc-
ing and B2B cooperation are brought together. As an extension to the existing
theoretical foundation of outsourcing research, the concept of externalities is
integrated by incorporating network effect theory. The dependence of sourcing
decisions based on other entities’ activities adds a further component of com-
plexity in the form of behavioral uncertainty to the already complex outsourcing
decision. In accordance with newer arguments from the network effect theory
(Weitzel et al. 2000), it is shown that externalities – in a positive occurrence
(generating additional scale economies) as well as in a negative occurrence (in-
creasing coordination efforts and agency costs) – strongly depend on who enters
the coalition (and not only on how many entities), incorporating various charac-
teristics such as relative cost efficiency, different process volumes, or even the
competitive relation between the firms.
The main conceptual contribution lies in the development of a formal agent-
based cooperative sourcing model which allows compound investigations of the
effect of sourcing drivers and inhibitors on cooperative sourcing activities and
resulting market effects. Thus, an integrated investigation of otherwise disjointly
investigated constructs and causal relationships is possible. This theory-based
model helps to explain and anticipate structural effects resulting from a system’s
cooperative sourcing dynamics by integrating microbehavior (decisions of the
individuals) and macrobehavior (resulting market effects) (Schelling 1978). The

1
The E-Finance Lab is a private-academic partnership between J. W. Goethe University in Frank-
furt/Main, Technical University in Darmstadt, and several large banks, software vendors, out-
sourcing providers, and consulting companies. It is located in Frankfurt am Main, Germany
(www.efinancelab.com).
12 Introduction

model also allows us to conduct game-theoretical equilibrium analyses to deter-

mine the inherent stability of cooperative sourcing coalitions.
From a practitioner’s perspective, the application of simulations to the com-
plex and interdependent field of cooperative sourcing helps managers to deter-
mine stable cooperative sourcing clusters and evaluate a repositioning of their
firm on a more sound decision basis. Determining and utilizing stable clusters
can increase both behavioral certainty and entry barriers (Emmelhainz 1987;
Large 1987).
The following paragraphs summarize the main findings related to each of
the research questions.
Research Question 1: What are the drivers and inhibitors of cooperative
sourcing?
Answering the first research question primarily consists of extracting the out-
sourcing determinants identified by previous research on IT outsourcing and
evaluating their relevance for cooperative sourcing by deduction and empirical
research. Since cooperative sourcing is a special form of BPO, the drivers and
inhibitors of the latter can also be adapted to cooperative sourcing.
The main reasons for BPO are cost reduction and capital reduction (i.e. cost
variabilization), accompanied by strategic issues (core competence focus). Ac-
cess to superior skills is considered ambivalently – usually it is also a major
argument for outsourcing decisions, but as cooperative sourcing usually covers
parts of the core business, because no other industry has superior skills, this
argument often has merely a secondary impact – as supported by various empiri-
cal studies (cf. sections 2.2.2 and 3.6.3). Nevertheless, the empirical results on
process cost differences (section 3.6.2.1) and the simulation studies indicate that
this argument tends to be underrated by practitioners because there are often
significant process performance differences between the firms.
As inhibitors of outsourcing, expected and hidden transaction costs are
commonly mentioned. Incentive conflicts between insourcer and outsourcer or
insufficient capabilities on the part of the provider to take over additional process
volumes can lead to cost escalation (contractual renegotiations, monitoring,
claim management, etc.), service debasement, and a loss of quality. A major
strategic issue is the loss of the outsourcer’s own skills and becoming too de-
pendent on the provider’s capabilities.
BPO, in particular, is supposed to lead to lower transaction costs because
taking entire business functions out of the firm does not lead to the interorganiza-
tional severance of the tight relationship between IT and business as often asso-
ciated with ITO. By contrast, selective outsourcing of parts of a business process
leads to vertical diseconomies of scope which were found to be a substantial
Introduction 13

inhibitor of BPO. However, particularly in the banking industry, there are still
“cultural” problems with thinking in terms of modular activities instead of mono-
lithic business segments. This hinders the selective outsourcing of singular busi-
ness functions. Moreover, agreeing on a common process design (i.e. process
standardization) is a major problem.
In cooperative sourcing, there is a basic congruence of interests between the
insourcer and the outsourcers because the insourcer firm usually operates its own
process volume on the same platform as well and therefore has the incentive to
provide high process performance (in terms of costs, time, and quality). This
reduces the risk of opportunistic behavior.
Research Question 2: How to ensure stable cooperative sourcing coalitions?
When deciding on establishing a cooperative sourcing coalition, the parties have
to agree on a cost or benefit allocation mechanism which fulfills their criteria for
participation and thus leads to a stable coalition.
Section 5.1 defines the conditions for an existing set of allocation vectors
which lead to stable coalition clusters, based on a simple formal cooperative
sourcing model. Furthermore, different cost allocation mechanisms are tested for
their inherent stability, such as an equal allocation of gain, a process volume-
proportional allocation of costs, and the Shapley allocation. It will be demon-
strated that only the proportional distribution inherently ensures stable coalitions.
Although the other schemes do not always lead to unstable coalitions, determin-
ing them ex ante when founding a coalition results in the problem that, with new
members joining the coalition in later periods, the allocation scheme would have
to be completely renegotiated between all members. Thus, this analysis contrib-
utes to the sourcing literature by providing a sound theoretical foundation for
cooperative sourcing and the analysis of the existence and efficiency of sourcing
equilibria and also offers some intriguing and maybe surprising findings for
practitioners.
Research Question 3: What is the impact of cooperative sourcing on overall
market structure and economic efficiency?
Based on the theoretically derived cooperative sourcing drivers and inhibitors,
the third research question focuses on the market effects resulting from coopera-
tive sourcing activities in a system of independent firms. Based on a mathemati-
cal, agent-based model of a system of firms, which autonomously decide on
sourcing their activities, simulation studies of system dynamics and the resulting
market structures are conducted by incorporating empirical data from the SME
credit business of the German Top 1,000 banks.
14 Introduction

Transaction costs are frequently argued to be one of the primary inhibitors

of outsourcing. Correspondingly, in the basic simulation scenario with low trans-
action costs, cooperative sourcing activities lead to significant cost savings.
When increasing the different types of transaction costs (interface costs,
agency/control costs, negotiation/coordination costs, and adoption costs), busi-
ness functions with low process volumes are most strongly affected by interface
costs, while agency costs inhibit outsourcing of processes with high complexity,
in particular. Overall, the net savings react most sensitively to changes of nego-
tiation costs and coordination costs.
If we take into account that firms do avoid outsourcing strategic business
functions, we see that individual banks do less cooperative sourcing. To simulate
this, a strategic constraint is introduced, which ensures that strategically relevant
core competencies are less outsourced. As a consequence, the autonomously
acting firms in the simulations significantly reduce their cooperative sourcing
activities.
But what is the degree of inefficiency induced by transaction costs and stra-
tegic constraints? The simulations are accompanied by an optimization routine
which determines the “optimal” cooperative sourcing configuration for the whole
system and thus determines the degree of inefficiency resulting from individual
decision behavior (i.e. decentralized vs. centralized coordination). For example,
while the autonomous(ly acting) firms in the simulations simply reduce their
cooperative sourcing activities in the case of a more restrictive strategic con-
straint, the optimal solution would be a system consisting of more and smaller
coalitions where more banks act as insourcers and thus keep their strategically
important business functions inhouse without abandoning cooperative sourcing.
Hence, individual optimization behavior increasingly fails when more decision
determinants have to be considered.
As well as analyzing the market structure and related monetary effects that
eventually result, the analysis focuses on the individual behavior of the firms
over time. Coalition switching and backsourcing activities are tracked to measure
behavioral uncertainty. More of these activities result from the situation that the
banks have to test more sourcing configurations unless they find their local opti-
mum2. Due to the existence of externalities, more activities in the system lead to
more changes in the decider’s environment which in turn hinders the search for
the optimal coalition.
Although the majority of monetary savings are achieved in the first periods,
the individual switching behavior often takes a long time before the overall sys-

2
In our definition, the local optimum describes the firm’s eventually chosen sourcing strategy
(insourcing vs. cooperative sourcing and choice of coalition).
Introduction 15

tem reaches a stationary state. Some firms search for the optimal coalition for a
long time although the benefits are only marginal and furthermore usually induce
negative externalities for the firms left in the previous coalition. Since the
evaluation of a coalition membership is influenced by this behavioral uncer-
tainty, higher dynamics lead to more frequent sub-optimal decisions which either
lead to a lock-in to sub-optimal local optima or to a longer and repeated search
process for a more beneficial coalition.
As a conclusion, the divergence in the overall net savings from centralized
vs. decentralized coordination represents a global inefficiency dilemma. In gen-
eral, the lack of cooperative sourcing activities (start-up problem) can be ex-
plained by negative individual monetary results or by behavioral uncertainty
which again can be traced back to the existence of externalities. Both reasons
represent two fundamentally different arguments. While the first is an ineffi-
ciency dilemma “only” from the overall system’s perspective, the second directly
represents an inefficiency dilemma also for the individual decision maker (Weit-
zel 2004, Weitzel et al. 2006).
The relevance of these problems strongly depends on the type of the busi-
ness function. When the potential savings are low (or when the perception of
possible savings is low, as found in the empirical studies) or when multiple deci-
sion criteria have to be considered (e.g. cost savings and strategic constraints) the
efficiency gap widens. The problem of behavioral uncertainty usually arises in
the second step after the firm has already entered a first coalition. Due to the
system dynamics, the stability of another, more beneficial, coalition cannot be
guaranteed. Since substantial cost savings have already been exploited with the
first coalition entered, the impact of switching costs and externalities will be
much stronger in this second step and will lead to a lock-in of the actor (individ-
ual lock-in). Furthermore, high dynamics may lead to a globally inefficient coali-
tion structure (e.g. many small coalitions) which can no longer be resolved by
individual decisions since the single actors are not even able to identify superior
coalitions (system lock-in).
These dilemmas are reduced by learning effects and increasing process stan-
dardization resulting from cooperative sourcing activities or other measures
because transaction costs for switching the coalition are reduced. As a counter-
argument, the simulations also show that process standardization that only results
from cooperative sourcing activities might be insufficient to overcome a sub-
optimal system structure of too small coalitions or clusters. When standardiza-
tion takes place only in sourcing coalitions, an industry-wide homogenization
will fail to appear and a potential migration to or merger with larger clusters may
even become more aggravated.
16 Introduction

Table 1 summarizes the identification of the three different inefficiency

sources in cooperative sourcing. In particular, the differentiation of individual
and system-wide lock-in represents a major theoretical contribution of this work.
Inefficiency
Description Reasons
source
Firm is reluctant to cooperatively No sufficient cost savings (economies of
Start-up
source business functions or does scale and skill)
problem
not find a coalition to join Too high transaction costs, agency costs, or
diseconomies of scope
Uncertainty about the partners’ behavior and
thus the savings (externalities)
Coalition members discard “application” (too
Individual Firm is reluctant to switch to high additional transaction costs, too low
lock-in superior coalition additional economies of scale)
Too high competitive degree in cooperation
(strategic risks which drive transaction costs)
Strategic constraint (only start-up problem)
Single firm does not find a superior Whole system runs into inefficient coalition
coalition to switch to, although structure, due to decentralized decision
System-wide another market structure would be making and due to “too fast” cooperative
lock-in more beneficial from a global sourcing activities (in the simulations, over-
perspective (i.e. higher aggregate reaction sometimes leads to a more frag-
benefits) mented system configuration)
Table 1: Sources of global and/or local cooperative sourcing inefficiencies

Apart from these structural inefficiency sources, the dynamics of switching

coalitions represent a further behavioral inefficiency source. The higher the
dynamics in the system, the higher is the behavioral uncertainty for the individu-
ally deciding firm. As a consequence, sub-optimal decisions can occur which
either lead to a lock-in to sub-optimal local optima or to a longer and repeated
search process (inducing higher cumulated transaction costs) for the “optimal”
coalition.

Managerial Implications
Apart from the contributions to research and theory development, the results of
this work yield several implications for practitioners who are responsible for
cooperative sourcing decisions and activities. The cooperative sourcing model
and the simulation routines provide a valuable tool for identifying promising
cooperative sourcing coalitions.
The major implications of the combination of empirical data and simulation
studies are:
Introduction 17

o In BPO of primary processes, economies of scale tend to be overestimated

and economies of skill are underestimated. As long as players with superior
capabilities take on the role of insourcers, significant cost savings can be re-
alized. Nevertheless, the primary motivation of these dominant players to
become the insourcer of a cooperative sourcing coalition is only economies
of scale. However, due to high process volumes, economies of scale will of-
ten be exploited quite fast (depending on the cost structure).
The simulation results suggest that the future of the German credit busi-
ness will consist of many small sourcing network clusters. The empirical
status quo – different regional cooperative sourcing activities in the savings
banks sector vs. a large player in the cooperative sector that has not yet been
able to attract a sufficient number of cooperatives – supports this forecast.
o Based on the simulations, several decision-relevant factors have been identi-
fied as critical with regard to the system’s cooperative sourcing activities
and subsequent dynamics. Potential members have to discuss how to control
them in terms of reducing variance and influencing them in the desired di-
rection. Examples are task interdependencies which lead to diseconomies of
scope, technology investments of the insourcer which help to stabilize the
coalition, or cluster-wide vs. industry-wide process standardization which, in
the first case, hinders or, in the second case, facilitates coalition changes.
The major implication of this research is to focus on the particular aspect of
externalities as part of the decision calculus. When firms cannot assume a coali-
tion to be stable in the long term, they might decide not to outsource their busi-
ness functions (leading to the start-up problem becoming an individual ineffi-
ciency dilemma for them). By contrast, there can also be too many activities. If a
firm changes the coalition, it might achieve higher benefits but will also induce
negative externalities for other firms. This can in turn affect the firm’s own situa-
tion. Thus, the “take-away” is that a firm’s cooperative sourcing behavior poten-
tially induces a feedback loop, too.

1.5 Basic Terms

This section defines and classifies the basic terms and concepts used in this re-
search work.

1.5.1 Processes, Activities, Business Functions, and the Value

Chain
Porter’s value chain concept provides a process-based picture of the firm (Porter
1985). It separates business processes of the original value chain (primary proc-
18 Introduction

esses or core/customer processes) from secondary processes (supporting proc-

esses) (Crux and Schwilling 1996; Davenport and Short 1990; Porter 1985).
Primary processes create value by transforming input factors into market-
oriented output. Secondary processes represent cross-sectional business functions
which support the value creation (Spiegel 2002), e.g. IT services, HR administra-
tion, finance & accounting (F&A), etc. Some authors add a third category of
management processes or controlling processes which represent all of the firm’s
non-operational activities (Crux and Schwilling 1996; Davenport and Short
1990; Dernbach 1996). Section 2.3.1.1 shows a generic value chain for the bank-
ing industry (p. 114).
A business process represents an operational workflow which is processed
within a firm or across multiple firms. In contrast to the generic value chain
concept, business processes are geared towards achieving a concrete and defined
business outcome (Davenport and Short 1990) and do not try to incorporate a
firm’s complete value creation within one construct. The creation of each prod-
uct requires its own business process. Within a bank or banking network, there
are business processes for credit products, securities services, payment and treas-
ury services etc.
Business processes can be hierarchically subdivided into subprocesses,
process steps and single activities although a clear distinction is hardly possible.
“Value activities are the physically and technologically distinct activities a firm
performs. These are the building blocks by which a firm creates a product valu-
able to its buyers. [… E]very value activity employs purchased inputs, human
resources, and some form of technology to perform its function. Each value
activity also uses and creates information [… and] also creates financial assets.”
(Porter 1985, 38)
In the following, the generic term business function – describing an encapsu-
lable subprocess, process step, or activity with well-specified input and output –
is used. Firms have to examine sourcing strategies for particular business func-
tions that can be done at different granularity levels: a bank can outsource the
complete payment process, only the payment transactions subprocess, or only the
process step of scanning remittance slips.

1.5.2 Interorganizational Relationships

The term interorganizational relationship (IOR) embraces “hybrid organiza-
tional forms, which contain elements of both markets and hierarchies” (Zhang
and Liu 2005, 54) and can be defined as relatively enduring transactions, flows,
and linkages that occur among two or more organizations (Oliver 1990). A sub-
set is described by the term alliance, “commonly defined as any voluntary initi-
Introduction 19

ated cooperative agreement between firms that involves exchange, sharing, or

co-development” (Gulati and H. 1998, 781).
The literature discusses different forms of IOR, spanning a continuum from
discrete transaction (pure market mechanism) over long-term relationships, stra-
tegic alliances, joint ventures, and network organizations (Webster 1992) to
vertical integration (i.e. merging two firms or business units) (Fontenot and Wil-
son 1997, 5). While the pure form of a truly discrete transactional exchange is
independent from all other exchanges – being quite rare in the real world –, most
B2B exchanges contain more or less relational activity. For close relationships,
Porter uses the term coalition, which describes “long-term agreements among
firms that go beyond normal market transactions but fall short of outright merg-
ers (Porter 1985). The section on relationship theories (2.1.8) provides a discus-
sion on the determinants of IOR creation and IOR success. Classifications of
different cooperation forms and cooperation motives can be found, for example,
in (Buse 1997; Cheon et al. 1995; Englert 2000; Porter and Fuller 1986). Out-
sourcing or cooperative sourcing as the research object of this work can be de-
scribed as establishing an IOR where one firm provides services to another on a
permanent or at least regular basis (cf. next section).
Cooperation on a horizontal layer, i.e. interconnecting similar process steps
of the value chain (like R&D, production, or logistics) often takes place between
competitors, shaping the term of coopetition (Brandenburger and Nalebuff
1996). Chen (1997) found about 50% of all strategic cooperations between the
largest 2,000 US firms to be between competitors. Some authors use a wide
definition of the term coopetition, combining cooperation and conflict: “Coop-
eration arises with respect to establishing arrangements that create value that
would not have existed otherwise. Conflict arises with respect to dividing up the
resulting surplus value” (Elitzur and Wensley 1997, 54). Consequently, Afuah
(2000) argues “that a firm should view its suppliers, customers, rivals, and poten-
tial new entrants as competitors”. By contrast, our work uses a more narrow
definition of coopetition, i.e. the partnership of competitors on a horizontal layer
(Hippel 1989; Nueno and Oosterveld 1988).

1.5.3 Outsourcing and Cooperative Sourcing

1.5.3.1 Outsourcing
In its generic form, the term outsourcing is defined very broadly as procuring
particular goods or services from outside the firm (Finken 1997, 2; Gilley and
Rasheed 2000, 764; Nagengast 1997, 47; Petzel 2003). According to this defini-
tion, outsourcing would cover every procurement activity. Most authors restrict
20 Introduction

this definition by defining outsourcing as shifting a business function from inside

the firm to outside the firm (Gilley and Rasheed 2000, 764; Lei and Hitt 1995,
836). Therefore, it can be seen as synonymous to “contracting-out” (Bartell
1998) or as a special form of make-or-buy decisions subject to a prior “make”
state. From an organization theory perspective, “outsourcing or insourcing deci-
sions are basically equivalent to the question of optimal vertical integration”
(Anderson and Parker 2002, 315). Outsourcing “represents a significant shift in
the mode of governance – from the traditional locus of control and coordination
within the hierarchy (combined with relatively standardized market transactions
with vendors) towards newer modes that could be characterized as hybrids or
partnerships” (Loh and Venkatraman 1992b, 237). Thus, outsourcing describes a
particular form of IOR (cf. section 1.5.2).
Some authors emphasize that outsourcing does not mean founding a spin-
off, but that the process must be given to a partner who already operates on the
market (Lacity et al. 1996; Meyer and Schumacher 2003; Riedl 2003; Schott
1997, 37). Furthermore, the term outsourcing is often inconsistently used not
only for the process of changing the governance mode (from make to buy) but
also for describing the result (“using outside resources”). Hence, Dibbern distin-
guishes the dynamic and static view of outsourcing (Dibbern 2004).
In our work, outsourcing describes the change in a firm’s sourcing mode of
a particular business function from in-house to outside the firm – either to a third
party or to a subsidiary. By contrast, insourcing describes the same process from
the sourcing provider’s point of view, whereas backsourcing describes the proc-
ess of reintegrating a formerly outsourced task back into the firm.
The term outsourcer will be used for the firm that outsources the business
function while insourcer represents the sourcing provider.
IT outsourcing (ITO) as a particular form of outsourcing is the permanent or
temporary delegation of IT operations to an external service provider (Heinzl
1993; Loh and Venkatraman 1992a, 9; Schott 1997, 37). A company can out-
source all of its IT functions (total outsourcing) or just a subset (selective sourc-
ing) (Aubert et al. 1996a). If only minor IT functions are outsourced (e.g. desk-
top support, network management, IT help desk services) or human resources are
brought in for a particular IT project, some authors call this outtasking (Allen
and Chandrashekar 2000; Freedman 2002, 6; Kooymans 2000). Some writers
also distinguish between outsourcing and application service providing (ASP),
based on the degree of specificity of the IT object that is contracted out. ASP
describes deploying, managing and hosting standardized software applications
such as standard ERP and desktop software by means of a centrally-located ser-
vice in a rental agreement (Currie and Seltsikas 2001). By contrast, in an out-
sourcing deal, the sourcing provider has to bear more specific investments.
Introduction 21

Orthogonally to ITO, Business Process Outsourcing (BPO) is defined as se-

lective outsourcing of complete business functions, including the required re-
sources, such as IT & HR (Dayasindhu 2004, 3478; Friend et al. 2002; Greaver
1999; Halvey and Melby 1996; McCarthy 2003; Meyer and Schumacher 2003).
Authors often restrict the definition of BPO to the outsourcing of IT-intensive
business processes because otherwise the outsourcing of facility management,
physical security etc. would also be part of their research (Dayasindhu 2004;
Gartner Group 2004; McCarthy 2003; Pfannenstein and Ray 2004; Rouse and
Corbitt 2004), e.g. “BPO […] is defined as outsourcing all (or most) of a reengi-
neered process that has a large IT component” (Pfannenstein and Ray 2004, 73).
The provider takes over the complete business function and is free to choose the
implementation; the outsourcer receives only the process result (Braun 2004).

1.5.3.2 Cooperative Sourcing

The investigation object of this research is cooperative business process sourcing
or – in short – cooperative sourcing. While outsourcing in a classical sense fo-
cuses on a bilateral IOR between an outsourcer and an insourcer firm, reality
shows more complex sourcing constellations. Sometimes, multiple firms jointly
decide to merge parts of their business activities, e.g. R&D in the automotive
industry or securities processing in the financial services industry.
In this work, cooperative sourcing (CS) describes the process and the result
of cooperatively merging (primary) business functions of different firms at a
horizontal level. This can be realized either by founding a joint subsidiary or by
bundling the process volumes so that they can be operated by one of the coopera-
tion partners (i.e. the insourcer). In (Lammers 2005), this alternative is labeled as
“share”, in contrast to outsourcing to a provider from other industries as common
in IT outsourcing. Since CS represents a close variant of horizontal cooperation,
inter-firm coordination is not based on market mechanisms (Buse 1997; Dowling
et al. 1996). Furthermore, it inherently represents a form of coopetition, since
cooperation usually takes place between (potential) competitors (Bloch 1987;
Englert 2000; Li 1995; Porter and Fuller 1986; cf. section 1.5.2).
Similar terms exist in the literature. CoSourcing, as used by consulting com-
panies, simply describes the known concept of outsourcing, but with a greater
focus on developing a long-term collaborative partnership between insourcer and
outsourcer instead of establishing a pure transaction-based contractual mecha-
nism (Maasjost 1995). This partnership involves sharing the risks that exist in the
outsourcing relationship (Jäger-Goy 1998; Willcocks and Lacity 1998). Wib-
belsman and Meiero (1994) use the term to describe temporary assistance from a
third party to reform the internal IS department. In (Gallivan and Oh 1999;
Sharma and Yetton 1996), the term co-sourcing is used in a slightly different
22 Introduction

way and describes several client companies – often firms in the same industry –
that have the same need which can be met more efficiently by forming an alli-
ance for obtaining services from a single sourcing provider (bundling bargaining
power). Moreover, the term cooperative sourcing is already used in the German
logistics domain, where it is a synonym of consortium purchasing (Essig 1998).
With the increasing availability of innovative internet technologies and a
widespread adoption of the service-oriented paradigm (Beimborn and Weitzel
2003; Koch and Rill 2005; McGovern et al. 2006), the concepts defined above
will not be sufficiently distinguishable because the establishment of service pro-
vision will become more and more dynamic and flexible. For example, Currie
notes that many SMEs “are now shifting their strategies to BPO as the emer-
gence of web services will enhance integration of software applications across
business processes. Some of these firms may develop partnerships or alliances
and call themselves business service providers” (Currie et al. 2003). Conse-
quently, the concepts of outsourcing, cooperative sourcing and insourcing are
unified in this citation.

1.5.4 Financial Service Firms

Banks and their business functions are the object of analysis of this work. Since
no general international definition of this term exists, the working definition is
usually derived from the legal definition of the respective regulating country.
Therefore, this section classifies the different types of financial service firms that
exist in Germany.
In the following, the terms bank, credit institution, and credit institute syn-
onymously describe an institution that follows the European Central Bank’s
definition of a monetary financial institution (MFI) but excluding money market
funds. All banks following this definition accept public deposits (or equivalent
substitutes, e.g. by emitting securities) and grant credits (also by purchasing
securities) for own accounts (DBB 2008, 110). This largely corresponds to the
definition of the German Banking Act (KWG), which defines a credit institution
by its business, which includes deposit business, lending business, discount busi-
ness, principal broking services, safe custody business, investment management,
investment fund business, guarantee business, giro business, underwriting busi-
ness, and e-money business (KWG, section 1 (1)). German banks are subject to
control by the Federal Financial Supervisory Authority (BaFin3).
A particular characteristic of the German banking market is the three-sectors
model, which subdivides the banking market into (private) commercial banks,
(private) cooperative banks (or credit cooperatives), and public savings banks,
3
Bundesanstalt für Finanzdienstleistungsaufsicht – BaFin.
Introduction 23

usually owned by cities, counties, or states. The cooperative sector and the public
savings sector are tightly organized within associations (BVR and DSGV4). Both
associations include large banks which provide special services for the smaller
and regionally operating savings banks and cooperatives (12 state banks or giro
centers in the public sector, DZ Bank and WGZ Bank in the cooperative sector).
Credit institutes can be divided into universal banks and specialized banks.
Universal banks offer all essential banking services as described by section 1 (1)
of KWG whereas specialized banks cover only some of these services (e.g.
mortgage banks, building and loan associations) (Sauter 2002). In terms of regu-
lation, all universal banks are treated equally; in particular, there is no legal im-
pediment to expanding a bank’s operations to other German regions. However,
“within the savings sector and within the cooperative sector a self-imposed prin-
ciple of regional demarcation holds, implying very little competition within these
sectors” (Lang and Welzel 1998, 70). Furthermore, the differences in legal forms
have consequences for the collection of equity capital, which in turn influences
the maximum credit volume permitted by law.
Some authors use the term universal bank in a process dimension rather than
in a product dimension (e.g. Polster 2001). In their understanding, a universal
bank provides the full range of business functions that are necessary to offer and
provide a particular banking product. In this work, we will refer to this concept
by the term fully integrated bank.
The German banking market predominantly consists of universal banks.
Historically, German banks have followed the business model of an omnipotent
and omnipresent service provider (Itschert and ul-Haq 2003, 127). This model is
linked to the principle of a house bank, which is a characteristic of the German
market. Customers often have only one bank, which provides all the services
they require (Kopper 1998, 49). German banks therefore try to offer a large
product portfolio. Moreover, in the 1990s, German banks were able to extend
their business to become all-round financial service providers. The main idea
underlying this strategy was that banks would be competitive in the future only if
they were able to offer a wide range of financial products to their customers, who
do not want to have multiple FSPs. Therefore, they also started offering non-
banking products such as insurances, building society savings, real-estate busi-
ness, etc. (Jasny 2001, 25; Schulte-Noelle 1998, 325). Figure 2 gives a quantita-
tive overview of the German banking market and shows the predominant posi-
tion of the universal banks.

4
Bundesverband der Deutschen Volksbanken und Raiffeisenbanken and Deutscher Sparkassen-
und Giroverband
24 Introduction

Advantages of a universal bank include economies of scope through cross-

selling, shared sales resources, more and improved long-term customer relation-
ships, increased flexibility, and market risk compensation through product diver-
sification (Büschgen 1994, 39-43; Jasny 2001, 25; Schulte-Noelle 1998, 325).
However, a universal banking system must also solve interest conflicts. Since
universal banks are able to offer different product alternatives to their customers,
they may be tempted not to offer the optimal product package but the one which
maximizes their own profit. One example of competing businesses is the compe-
tition between deposit business and principal broking services.
German Banking Sector
(12/06) (2156)

Universal Banks Specialized banks

(2090) (66)

Commercial banks Cooperative Mortgage banks Banks with special

Public banks (469)
(360) banks (1261) (48) functions (18)

- Big banks (5) - Credit cooperatives - Public savings banks (457) - Real-estate credit e.g.
- Regional and other (1259) - State banks /giro centers (12) associations (22) - Deutsche Ausgleichsbank
commercial banks (217) -Deutsche Zentral – - Home loan banks (26) - Kreditanstalt für Wiederaufbau
- Branches of Genossenschaftsbank
foreign banks (138) - DZ Bank AG

Figure 2: Classification of German banks (Data from the German Central

Bank as of 12/2006, classification scheme from (Sauter 2002, 325))
Related to banks or credit institutes are the financial services institutions,
which offer financial services but are not considered banks. These services in-
clude investment broking, contract broking, portfolio management, trading on
own account, foreign currency dealing, money transmission services, and credit
card business (KWG, section 1 (1a)). Financial service institutions are also sub-
ject to control by the BaFin but have to follow less strict rules. Examples are
sales-oriented FSPs, asset and fund managers, and credit card companies.
As a third group, financial enterprises are defined as companies which are
no institutions and whose business consists of stock acquisitions, closure of leas-
ing agreements, trade in financial instruments for their own account, investment
advice, other financial consulting, or the money-broking business (arranging
loans between credit institutions) (KWG, section 1 (3)). Typical examples are
private equity firms, leasing firms, and consulting companies specialized on
mergers & acquisitions (M&A). Financial firms are subject to only very few
regulatory issues (only sections 10a and 13b of KWG).
Finally, ancillary banking services enterprises “are enterprises which are
neither institutions nor financial enterprises and whose main activity comprises
Introduction 25

administering real estate, operating computer centers or performing other activi-

ties which are ancillary activities relative to the main activity of one or more
institutions” (KWG, section 1(3c)).
It has to be noted that the term bank will sometimes be substituted with the
more generic term firm during this work as long as the remarks and findings can
be generalized for other industries.

1.6 Thesis Structure

The structure of this thesis is as follows (cf. Figure 3 on p. 26 for visualization):
II – Theory: Chapter 0 starts with the theoretical foundation of cooperative
sourcing (section 2.1). Several economic and organizational theories are re-
viewed and applied to outsourcing in general and to cooperative sourcing in
particular. As a conclusion, a multi-perspective foundation for cooperative sourc-
ing research is derived. The second part of the chapter (section 2.2) reviews
previous research works on outsourcing related to this thesis by focusing on the
drivers and inhibitors of outsourcing and on the development and application of
formal models in the outsourcing context.
III – Banking industry as application domain and empirical data: After
theories and outsourcing research have been reviewed at a generic level, the third
chapter introduces the application domain of this work. After giving an overview
of the structural problems of the banking industry and related industrialization
tendencies (section 3.1), segmentation models which help to structurally analyze
segmentation activities in this field (3.2) are compared. Since the credit business
has been chosen as the concrete application domain of the empirical research, the
chapter zooms into this particular business segment with a brief overview of
products and processes (3.3). After presenting secondary data regarding coopera-
tive sourcing activities in the banking industry in general and in the credit busi-
ness in particular (3.4) and discussing the regulatory issues related to outsourcing
activities (section 3.5, again for the banking industry in general and for the credit
business in particular) section 3.6 aggregates evidential results from several own
surveys and case studies on the potential of outsourcing and cooperative sourcing
in the German credit business.
IV – ACE model: Chapter 4 presents the conceptual core of this research
work. Based on the theoretical foundation and on the related research, a formal,
agent-based model of cooperative sourcing is developed, which makes it possible
to analyze the effects of cooperative sourcing behavior of multiple organizational
entities. From the basic model (section 4.2), a centralized variant (4.3) and a
decentralized variant (4.4) of the model are derived. While the first is intended to
26 Introduction

determine the globally optimal cooperative sourcing configuration for an entire

system of firms, the second provides the foundation for simulating the individual
behavior of the autonomous firms.
V – Analysis: The fifth chapter represents the application of the cooperative
sourcing model. First, the model is applied in order to determine the conditions
for the stability of cooperative sourcing clusters by using game theory (section
5.1). Second, to solve the cooperative sourcing problem which results from the
centralized variant of the cooperative sourcing model, a heuristic optimization
routine based on the concept of genetic algorithms is developed (5.2). As the
main part, section 5.3 provides the application of the decentralized model to
agent-based simulation studies. The simulation studies analyze effects resulting
from cooperative sourcing and explore the role of process cost structures, trans-
action costs, and strategic constraints. Based on these results, the sources of
inefficient decisions and market structures are identified.
VI – Implications and summary: Finally, chapter 6 provides a summary of
the results (section 6.1) and derives the implications for research and practitio-
ners (6.2). Validation steps (6.3), a discussion of the limitations (6.4), and an
overview of promising future research (6.5) conclude this work.
I: Motivation

II: Theoretical foundation III: Empirical foundation

Theories and related research Cooperative sourcing in the German banking industry
validate model structure feed model

IV: Cooperative sourcing model

(Centralized and decentralized variants)

V: Analysis - Application of the cooperative sourcing model

5.1: Stability of 5.3: Individual cooperative

5.2: Solving the cooperative
cooperative sourcing sourcing behavior,
sourcing problem
clusters structural effects

VI: Conclusion, implications

Figure 3: Structure of the thesis

Theoretical Foundation and Related Research 27

2 Theoretical Foundation and

Related Research
“The purpose of theory is to increase scientific understanding through a systematized structure
capable both of explaining and predicting phenomena.”
(Hunt 1991, 149)

This chapter provides the foundation for our research on cooperative sourcing. In
the first part (section 2.1), the theoretical foundation is developed from different
economic and organizational theories, while the second part (2.2) reviews the
application of these theories to the outsourcing phenomenon in earlier related
research on identifying the determinants of outsourcing decisions. Based on the
results of this chapter and the empirical data in chapter 3, the formal cooperative
sourcing model is developed in chapter 4.

2.1 Theoretical Foundation for Cooperative

Sourcing Research
Theory is intended to fulfill the objectives of explanation and prediction in order
to understand the relationships among particular objects of analysis (Dubin
1969). A theory is represented by a “system of constructs and variables in which
the constructs are related to each other by propositions and variables are related
to each other by hypotheses” (Bacharach 1989, 498), with the underlying as-
sumptions limiting the theory’s application to the real world (Dubin 1976). Re-
searchers do not set up theories to describe objective reality; instead, theories are
used as “social constructions of reality people use to help ascribe meaning to
existence” (Lacity and Willcocks 1995, p. 218, based on Allison 1971; see also
Astley 1985; Chua 1986).
This section will briefly discuss the main theories which have been identi-
fied as foundation for research on outsourcing and cooperative sourcing. Led by
the research questions of this work, the selection is restricted to theories which
contribute to the question of why (or why not) to engage in cooperative sourcing.
The following list gives a short overview of these theories, using an adapted
classification scheme from (Cheon et al. 1995; Dibbern 2004; Lee et al. 2000).
While Cheon et al. (1995) consider only the economic and the strategic perspec-
28 Theoretical Foundation and Related Research

tive for outsourcing research, the later articles extend this classification by the
social/organizational category.
o Economic theories:
o Production cost economics (PCE) (section 2.1.1)
o Transaction cost economics (TCE) (section 2.1.2)
o Agency theory (AT) (section 2.1.3)
o Theory of incomplete contracts (TIC) (section 2.1.4)
o Network effect theory (NET) (section 2.1.9)
o Strategic theories:
o Porter’s positioning framework (PPF) (section 2.1.5)
o Resource-based view (RBV) (section 2.1.6)
o Core competence view (CCV) (section 2.1.6)
o Organizational and social theories:
o Resource dependency theory (RDT) (section 2.1.7)
o Relationship theories (section 2.1.8)
Economic theories describe the impact of cost or other efficiency criteria to co-
ordinate economic agents (Cheon et al. 1995, 211) while strategic theories sup-
port the understanding of “how firms develop and implement strategies to
achieve a chosen performance goal” (Dibbern et al. 2004, 17), e.g. determining
and developing the strategic resources and competencies of the firm and gaining
a sustainable competitive advantage by positioning their business. Thus, they can
be applied to connect outsourcing decisions to the firm’s overall business strat-
egy (Lammers 2005, 14). Organizational/social theories have a meta-organiza-
tional focus and concentrate on dependencies and relationships between agents
(Dibbern et al. 2004, 17). In contrast to some of the articles cited above, the ana-
lysis will draw only on the resource dependency theory and on relationship theo-
ries, while ignoring other social theories that have been applied to outsourcing,
such as the power theory, in particular.
In the following, the theories are briefly introduced in general and then dis-
cussed with regard to their impact on sourcing decisions and consequences. The
theoretical foundation concludes with a synthesis of all discussed theories (sec-
tion 2.1.10), including a reflection on the main research objects – business func-
tions and cooperative sourcing – in light of the different theories.
Theoretical Foundation and Related Research 29

2.1.1 Production Cost Economics

2.1.1.1 Basics
In neoclassical production economics, a firm is viewed as a “production func-
tion” which is driven by profit maximization (Williamson 1981). The exogenous
variables of this function in its basic form are labor and capital while the result is
described by the output volume of products or services.
If an activity-based view is adopted, the firm is seen as a collection of re-
lated production functions (instead of a single one) (Porter 1985). This allows a
more detailed specification of the actual production function(s) and the respec-
tive input factors and outputs.
If the production factors are weighted with prices, the corresponding produc-
tion cost function can be derived. The production cost function can be based
either on input factors (common in economics and econometrics) or – if the
minimal cost combination has already been determined – on output volume (as in
product cost accounting).
Production costs include any direct and indirect costs for producing a service
or product as well as the delivery and service costs. Differences in production
costs of different firms result from a different scale and scope as well as from
different production resources and skills (Baumol et al. 1982; Chalos and Sung
1998; Poppo and Zenger 1998).
Economies of scale arise from the ability to perform activities more effi-
ciently at larger volumes (Porter 1985, 71). They result both from fixed-cost
degression and from a declining ascent of the cost function, i.e. decreasing aver-
age unit costs when expanding the output (Baumol et al. 1982; Murray and
White 1983). For an output-based production cost function, this can be formally
described as declining average production costs (Baumol et al. 1982):
dAvgPC PC  x 
0 with AvgPC
dx x
Equation 1: Economies of scale ([Avg]PC = [average] production costs,
x = output)
On the other hand, there may be diseconomies of scale, which could arise
from increasing complexity, decreasing motivation of HR, etc. (Porter 1985, 71).
Economies of skill can be distinguished into ex ante and ex post. Ex ante
economies of skill determine the form and the position of the cost function (Au-
guste et al. 2002, 55). If a firm’s cost function leads to lower costs at a certain
output level compared to other firms, this implies comparatively higher skill
economies, representing a certain core competence of the firm (Langlois 1995;
30 Theoretical Foundation and Related Research

Prahalad and Hamel 1990). Comparatively higher economies of skill can result
from technology leadership or superior resource bundles (cf. section 2.1.6 on
RBV). Dynamic or ex post economies of skill are represented by a down-shift of
the cost function over time (Lamberti and Pöhler 2004, 21; Lammers 2005, 31).
This can happen both continuously by realizing learning effects5 (Porter 1985,
72+73) and step-wise by technology changes.
As a consequence, economies of scale and skill are interrelated. On the one
hand, economies of skill are, among others, dependent on scale. If the process
volume increases, unit cost reductions cannot only be realized in the short term
by fixed-cost degression or by reaching lower marginal cost regions of the cost
function but the increase also leads to stronger learning effects in the long term
(Ewert and Wagenhofer 2003, 167; Simon 1992) and therefore to a down-shift of
the cost function (Porter 1985, 74). On the other hand, economies of scale are
(co-)determined by economies of skill. For example, if learning effects or newly
acquired capabilities lead to a change of the shape of the production cost func-
tion towards (relatively) higher fixed costs, economies of skill will lead to even
greater economies of scale.
Assuming a constant output volume x over time (t)6, the relationship can be
expressed as follows:
wAvgPC w 2 AvgPC PC
 0 and  0 with AvgPC and PC PC ( x, t )
wt wtwx x
Equation 2: Dynamic economies of skill (learning effects)
Economies of scope refer to the advantages resulting from the shared utiliza-
tion of common resources when it is less costly to combine two or more product
lines in one firm rather than handling them separately (Panzar and Willig 1977;
Panzar and Willig 1981). If an activity-oriented view of the firm is chosen,
economies of scope can also describe task interdependencies or linkages between
different business functions (Knolmayer 1993; Porter 1985), which are either
connected within a business process (vertical or process-oriented economies of
scope) or which belong to different business processes (horizontal or product-
oriented economies of scope). Economies of scope may result from re-using
resources in different activities (e.g. different organizational units having access
to centralized client data or employee knowledge applied in different process
steps), from the need to closely coordinate activities (e.g. time critical processes)

5
The learning effect in production economics was first discovered by Wright (1936). Argote
(1999) provides a comprehensive review of empirical studies on learning effects.
6
Otherwise, the production costs would be a function of cumulated process volumes over time
instead of the number of periods (t).
Theoretical Foundation and Related Research 31

(Porter 1985), or from the easier and more effective optimization of business
processes which are composed of those business functions7.

process subprocess subprocess subprocess

1 1a A 1b A 1c

process subprocess subprocess subprocess

2 2a 2b 2c

Figure 4: Vertical (A) and horizontal (B) economies of scope between a

firm’s business functions
Examples for horizontal economies of scope in banking exist in sales, for
example (same resources for selling different products/services and increasing
customer benefits by cross-selling), or when jointly using an indivisible special-
ized asset (Pfeiffer et al. 1999) or utilizing expertise, customer relationships, and
technology of a product segment in another domain (Börner 1998).

2.1.1.2 Implications for the Sourcing Decision

As Ang and Straub point out “Firms provide goods and services to markets
where they have cost advantages and rely on the marketplace for goods and ser-
vices in which they have comparative cost disadvantages” (Ang and Straub 1998,
537). From a production cost economics (PCE) perspective, firms treat outsourc-
ing (or make-or-buy considerations in general) as a decision that compares pro-
duction costs of internal operations with the price offered on the market (Ford
and Farmer 1986). Firms therefore outsource activities in order to achieve the
production cost advantages of vendors (Ang and Straub 1998; Loh and
Venkatraman 1992a; Slaughter and Ang 1996). This theory has received more
empirical support in explaining outsourcing decisions than any other theory (e.g.
Ang and Cummings 1997; Ang and Straub 1998; Loh and Venkatraman 1992a;
Loh and Venkatraman 1992b; Slaughter and Ang 1996; Walker and Weber
1984).

7
The “fundamental principle of effective and efficient value chain modification” states that the
opportunities for improvement will increase disproportionally to the extension of the improve-
ment attempt of superior system layers. Furthermore, the relative optimization costs will decrease
(Pfeiffer et al. 1999).
32 Theoretical Foundation and Related Research

Economies of scale are often considered to be one of the main reasons for
outsourcing. The service provider is expected to provide services at lower costs
by bundling similar processes of several firms, thereby reducing average costs
per unit (Cachon and Harker 2002; Gurbaxani and Whang 1991; Matiaske and
Mellewigt 2002; Schott 1997). By contrast, in the field of IT outsourcing, it has
often been argued that economies of scale could be reached sufficiently within a
single (large) firm (Bloch and Spang 2003; Dibbern et al. 2003; Earl 1996; Lac-
ity et al. 1996). The economies of scale argument alone therefore does not hold
because large firms still outsource some of their activities (Chabrow 2002;
Levina and Ross 2003; McDougall 2002).
A necessary precondition for realizing economies of scale in BPO is stan-
dardizing the business process or accepting a reference process (i.e. the standard)
which is provided by the sourcing provider (Rouse and Corbitt 2004; Wüllenwe-
ber et al. 2008).
Economies of skill, as another argument for outsourcing, result from core
competencies and superior learning effects of the provider (Langlois 1995; Pra-
halad and Hamel 1990). Economies of skill can be realized by the service pro-
vider because (from the provider’s perspective) the insourced business process
represents a primary process, i.e. his core business (Dibbern and Heinzl 2001).
The vendor firm’s strategy to be a cost leader will only be possible if it has lower
average costs (cf. section 3.5 (Porter’s positioning framework)). On the other
hand, a firm will tend to outsource a business function if it has a comparatively
high production cost function (Loh and Venkatraman 1992a). Due to increased
process volumes, the insourcer can realize even more learning effects. From the
outsourcer’s perspective, negative learning effects do also occur, the so-called
“organizational forgetting” (Conway 1959, Epple and Devadas 1991). Anderson
and Parker (2002) explicitly link outsourcing and the learning curve literature
into a formal outsourcing model.
Economies of scope, as defined above, can be inhibiting factors for selec-
tively outsourcing particular business functions since they might get lost and
result in severe interfacial costs (Bruch 1998). Various studies show the loss of
economies of scope to be a major problem of outsourcing (Bettis et al. 1992; Earl
1996; Kotabe 1992; Loh and Venkatraman 1992a). Links between different ac-
tivities often remain unrecognized unless a business function is outsourced (Lei
and Hitt 1995, 840). One of the main risk sources in IT sourcing is not ex ante
discovered technological indivisibility and interdependencies between informa-
tion systems and processes (Aubert et al. 2000; Earl 1996; Lee and Kim 2003).
Interdependencies increase information-processing costs (Emery and Trist 1965)
and can lead to a decline in organizational performance (Gulati and Singh 1998).
Theoretical Foundation and Related Research 33

On the other hand, selective outsourcing of business functions (BPO) is seen

as less problematic than IT outsourcing because the tight interfacial area between
the IT and the business domain is not separated interorganizationally (Orton and
Weick 1990; Rouse and Corbitt 2004). In the case of IT-intensive businesses, in
particular, several authors argue that business functions can be sufficiently
modularized to selectively outsource them (Beimborn et al. 2005b; Gurbaxani
and Whang 1991) and that the German banking industry follows this modulariza-
tion path to reduce task interdependencies (Hertel 2004; Petzel 2003). Further,
Anderson and Parker (2002) showed that higher product modularity in physical
products leads to higher fractions of outsourcing.
Economies of scope can also have positive effects on outsourcing. For ex-
ample, Loh and Venkatraman argue that an IT outsourcing provider can realize
substantial economies of scope by the variety of conducted IT projects (Loh and
Venkatraman 1992a); similar arguments for BPO can be found in (Rouse and
Corbitt 2004).
From a PCE perspective, the primary functional relationship regarding out-
sourcing can be described as
Outsourcing = f((additional) economies of scale (+), (relative) economies of
skill (+), economies of scope (-))
The same arguments obviously hold true for cooperative sourcing. If there
are opportunities for achieving economies of scale from bundling process vol-
umes, this will drive cooperative sourcing to the coalition member which offers
cost efficient processing (economies of skill). Economies of scope can get lost
from the perspective of the outsourcing firms in the coalition.

2.1.1.3 Empirical Evidence of PCE in the Banking Industry

Compared with many other industries, banks do not generate value by
physical product development but by information processing and risk assump-
tion. While the value chain of a non-bank can often be modeled quite easy and
sequentially, the “steps of production” in a bank are much more difficult to
model. Consequently, a comprehensive production theory of the banking firm
has not yet emerged and there is still disagreement about what constitutes input
and output of the banking business (see below) (Polster 2001).
Production cost economics have been applied to econometric research works
in the banking industry for many years. The basic step of these econometric
works is to determine the design of the production function. The choice of how
to define input and output has not reached a consensus in the literature. “Thus,
when assessing efficiency issues, results differ not only because of the different
techniques used to estimate efficiency but also because of our beliefs about what
34 Theoretical Foundation and Related Research

banking firms produce” (Tortosa-Ausina 2002, 199). Particularly, customer de-

posits are variably used either as input factor (funding) or as output (deposit
management as product/service) (Braeutigam and Daughety 1983; Tortosa-
Ausina 2002, 201).
The basic assumption of the analyses of economies of scale is that a signifi-
cant proportion of the input into bank production functions is quasi-fixed in the
short run. A bank’s operational costs are predominantly formed by IT and per-
sonnel (Lamberti and Pöhler 2004, 8). IT costs, in particular, do not increase in
proportion to the bank's size (Hughes 1999, 2) and – since most banking proc-
esses are IT-intensive – they have a significant influence on economies of scale
in banking activities. The bank’s physical capital (Hunter and Timme 1995, 167)
and customer deposits (if the respective econometric model uses them as an input
factor) are further quasi-fixed input factors – because both the bank and its cus-
tomers incur setup costs when establishing new accounts (Flannery 1982; Klein
et al. 1978).
Lacity and Willcocks discuss the different costs related to IT in order to ex-
plain where scale advantages of an insourcer would occur (Lacity and Willcocks
1995, 234). While data center operating costs are often cited as achieving scale
economies, they show that these are achieved at 150 MIPS which is approxi-
mately equal to the size of one large IBM mainframe. This was often realized
within single firms which were investigated in Lacity’s and Willcocks’ case
study series. As another factor, hardware purchasing costs may be slightly lower
for larger customers due to volume discounts which are often offered by the
vendors. Similar discounts can be realized when purchasing software (either via
volume discounts or site licenses). Furthermore, as long as it runs on a single
machine, software provides unrestricted scalability (Englert 2000, 119-20).
Englert argues that, since hardware prices show constant decreases, software
(purchases or in-house development) becomes the primary cost factor in IT.
Consequently, major savings can also be realized in R&D costs. Higher scales
justify more complex and sophisticated developments in information systems
which allow for higher automation. In an own case study, credit back-office
experts of a large German credit cooperative argued that, although human inter-
action is necessary in credit application processing, the larger scale of credit
factories allows them to implement sophisticated workflow management systems
which lead to significant reductions in HR activity. Larger scales often might
justify a change of technology leading to higher fixed costs but lower variable
costs.
Another economic reason for economies of scale in the banking industry is
the ability for better risk diversification if capital stock increases. As “the num-
ber of a bank’s loans and deposits increases with its size, the bank’s exposure to
Theoretical Foundation and Related Research 35

credit risk and to liquidity risk can be reduced by better diversification of assets
and liquid liabilities” (Hughes 1999)8. By contrast, larger banks can use this
diversification advantage to engage in larger risks, which would consequently
lead to stable risk costs (Hughes 1999; Hughes et al. 2001). Furthermore, this
argument only holds true for outsourcing credit refinancing but not for (the much
more common) outsourcing of credit processing and other repetitive tasks.
Economies of scale in the banking industry are mostly surveyed within
econometric banking efficiency studies. The focus of these investigations is on
“estimating an efficient frontier and measuring the average differences between
observed banks and [the efficient] banks on the frontier” (Berger and Humphrey
1997, 896). Primary estimation techniques are DEA (data envelopment analysis)
and comparable frontier approaches (Berger and Mester 1997, 905).
The level of economies of scale within most of these surveys only ranges
from 3% to 10% (Berger and Humphrey 1991; Hunter and Timme 1991; Hunter
et al. 1990), i.e. the bank’s product mix could be produced at minimum average
cost by increasing output by 3-10%.
Berger and Mester (1997) found substantially greater (potential) economies
of scale. According to their study, more than 90% of American commercial
banks operate below an efficient scale; they would have to be two to three times
larger in order to maximize the cost scale efficiency for their product portfolio
and the given input prices (Berger and Mester 1997, 926). Moreover, the results
differ for different classes of bank size. Larger banks show slightly larger unex-
ploited economies of scale, which is explained by the hypothesis “that larger
banks ($10 to $25 billion GTA) choose product mixes that are more conducive to
large scale” (Berger and Mester 1997, 926). Similar results were found by
Hunter and Timme (1995). By contrast, the same analysis of data from 10 years
before did not find any unexploited economies of scale (Berger and Mester
1997). Although the reasons are difficult to identify, the authors partly explained
this phenomenon with IT-related arguments such as improvements in technology
and information processing.
The differences between the studies that found only weak potential econo-
mies of scale in the banking industry (such as Berger and Humphrey 1991;
Hunter and Timme 1991; Hunter et al. 1990) and the ones that found substantial
economies of scale (such as Berger and Mester 1997; Hughes et al. 2001; Hunter
and Timme 1995) can generally be explained by the scope of the data used. In-
vestigations which neglected banks’ capital and risk structures often did not find
economies of scale, whereas taking them into consideration often lead to the

8
If the risk portfolio management captures different credit products of the bank and balances risks
among them, this argument will also account for economies of scope.
36 Theoretical Foundation and Related Research

opposite result (DeYoung et al. 2001; Hughes et al. 2001, 2170; Tortosa-Ausina
2002, 200).
Studies on (product-oriented) economies of scope are quite uncommon for
the banking industry. At firm level, there is no empirical support for their exis-
tence (Amel et al. 2004). Some studies showed that universal banks which offer
a wide range of products were more cost-efficient than product specialists (Ber-
ger et al. 1996; Lang and Welzel 1998; Parsian et al. 1996; Vander Vennet
2002). “Nevertheless, there is a major difficulty in measuring economies of
scope at banking level, as the benchmark should consist of single product firms”
(Lammers 2005, 35) which usually do not exist in a pure form.
One of the main shortcomings of the econometric analyses introduced is the
firm level approach. Of course, there is usually no more detailed data available;
however, estimating economies of scale at firm level instead of doing investiga-
tions at the level of particular activities (and on the corresponding cost functions
from the banks’ accounting systems) obviously contains certain drawbacks. The
multi-product characteristics of banks, which include several business processes
with different degrees of automation etc., as well as different output definitions
throughout the banking industry are some of these drawbacks (Berg et al. 1992;
Berger and Humphrey 1997; Favero and Papi 1995; Tortosa-Ausina 2002). Only
an analysis of each of the bank’s activities enables management to evaluate the
production cost effect (economies of scale, skill, and scope) of sourcing deci-
sions (Bátiz-Lazo and Wood 1999; Canals 1994).
There are a few works analyzing production costs at process level. For ex-
ample, Schmiedel et al. (2002) analyzed the existence of economies of scale and
skill (efficiency improvements) in IT-intensive banking processes such as securi-
ties depository and settlement of equities. In both processes, they found econo-
mies of scale which would result in a cost increase of about 18% (securities
depository) or 70% (equities settlement) if the process volumes were doubled.
Learning effects and improvements in IT (dynamic economies of skill) reduce
costs by about 4.5% per year. Other surveys found significant economies of scale
in payment processing (Adams et al. 2002) and securities processing (Malkamäki
1999). The cost structure of these processes is dominated by fixed costs because
they are highly automated (IT) and the remaining HR is only necessary for moni-
toring the process and for exception handling (Schrauth 2004, 59). In the German
securities processing market, which is already quite consolidated (cf. section
3.4.2), there would be only a few further economies of scale. Bongartz found that
increasing the processing volumes of the large transaction banks listed in Table
13 (p. 143) by 50% would lead to average cost reductions of less than 10%
(Bongartz 2004, 51). Nevertheless, this does not disprove the hypothesis that the
Theoretical Foundation and Related Research 37

existing transaction banks have already exploited substantial economies of scale

from cooperative sourcing.
Apart from economies of scale, Malkamäki also found significant economies
of scope at activity level (between the activities of trade processing and listing of
additional companies). Since there are specialized service providers (“single
product firms”) on the market for offering these services to banks, a comparison
was possible which showed that separate production leads to about 50% higher
costs compared with joint production. Further, initial evidence for vertical
economies of scope in the credit business has been found by Beimborn et al.
(2005a).

2.1.2 Transaction Cost Economics

2.1.2.1 Basics and Discussion
Transaction cost economics (TCE) basically goes back to the fundamental works
of Coase (Coase 1937), Klein, Crawford (Klein et al. 1978), and Williamson
(Williamson 1975; Williamson 1979) and represents a major strand in new insti-
tutional economics. As an extension to the neoclassical paradigm, new institu-
tional economics focuses on the legal and social mechanisms and rules that un-
derlie economic activity. It covers theories such as TCE, agency theory (cf. sec-
tion 2.1.3), and property rights theory.
In contrast to production cost economics, TCE focuses on costs which occur
with transferring property rights, i.e. a transaction (Commons 1934, 652). The
term transaction covers all exchanges of goods and services between separable
business functions (Williamson 1985, 552). The object of analysis of TCE is the
(loss of) friction which accompanies the transaction. Distinguishing them from
production costs, Arrow defines transaction costs as the operating expenses of
the economic system (Arrow 1969).
Transaction costs are usually classified as ex ante or ex post transaction
costs. While the first include all “costs of drafting, negotiating, and safeguarding
an agreement” (Williamson 1985, 20), the latter “include the maladaptation costs
incurred when transactions drift out of alignment […], the haggling costs in-
curred if bilateral efforts are made to correct ex post misalignments, the setup
and running costs associated with the governance structure […] and the bonding
costs of effecting secure commitments” (Williamson 1985, 20). Classifications
by different authors and the relationship between them are given in Figure 5.
38 Theoretical Foundation and Related Research

(Picot 1982) (Kreikebaum 1998) (Albach 1998)

Initiating costs Search costs
(search, decision) Initiating costs
ex ante (search, decision, negotiation)
Information and
transaction
Agreement costs agreement costs
costs
(negotiation, processing (decision, negotiation,
and agreement processing) Agreement costs
the agreement) (agreement processing)

Coordination costs Exchange costs

(process changes, interfaces) (interface costs) Adaptation costs
(process changes, interfaces,
ex post Adaptation costs ex post changes of contract)
Ex post change
transaction ex post adaptation
costs
costs of the contract

Control and Hedging costs

Control costs
enforcement costs (control system)

Termination costs

Figure 5: Mapping of transaction cost categories

Although some classification divergences exist within the ex ante stage, all
authors clearly separate between ex ante and ex post transaction costs. Before the
transaction is processed, costs arise for evaluating the make-or-buy decision, for
searching for an appropriate partner, for negotiating, and for processing the
agreement. Afterwards, coordination costs arise for establishing a successful
exchange, possibly by developing interfaces in the case of more complex trans-
actions (e.g. entering an outsourcing relationship). The term “adaptation costs” is
used with different meanings. It may describe adaptation of process changes
(necessary in BPO, for example) but also later changes to the contract if one
party is discontent with the transaction ex post. Only Albach takes the reversal of
the transaction into account, which again generates costs. Interestingly, all au-
thors incorporate control costs in their classifications but none cites costs for
relationship management, which proved to be critical within complex transac-
tions such as large projects or outsourcing relationships (Goles 2001; Lee et al.
2003; Van der Meer-Kooistra and Vosselman 2000).
The objective of TCE is determining the optimal configuration of contracts
by minimizing total transaction costs. The choice between hierarchy or market
(make-or-buy), and thus the comparison of internal and external transaction costs
(Afuah 2003, 37), has been theoretically supported by TCE for several decades.
The application domains include determination of the optimal degree of vertical
integration (Klein et al. 1978; Williamson 1971; Williamson 1975; Williamson
1985), analysis of employment relationships (Williamson et al. 1975), internal
Theoretical Foundation and Related Research 39

design of organizations (Williamson 1981), optimal design of IT infrastructures

(Ciborra 1987), configuration of accounting systems (Albach 1988), selection of
financing instruments (Williamson 1988), outsourcing of IT and business proc-
esses (cf. next section), and analysis of the success of forming innovation-
oriented firms (Goshal and Moran 1996; Picot et al. 1989).
Before TCE emerged, other economic theories had already tried to deal with
the contracting problem. In the classical contracting system, a contract was sim-
ply related to a particular point in time, with output and reward occurring at the
same moment. The neo-classical approach refined the last assumption but still
focused on unique transactions, while the more recent research on TCE also
considers relational contracts in long-term business relationships, and contracts
which will be reconfigured over time (Williamson 1985, 29).
TCE argues that transaction costs are primarily influenced by three different
dimensions of the particular transaction – asset specificity, uncertainty, and fre-
quency (Williamson 1985). The degree of asset specificity is determined by the
extent to which the particular asset is developed or adapted to the particular
transaction, i.e. how difficult it is to re-use it within an alternative transaction. It
can be operationalized as quasi-rent – the difference between the value of a re-
source (as transaction object) in its current use and in its next best possible use
(Burr 2003; Picot and Dietl 1990, 179). Ex ante and ex post specificity are dif-
ferent. If an ex post specificity but no ex ante specificity is related to a particular
transaction, there is competition at the beginning but the receiver of the transac-
tion develops transaction-specific competencies and idiosyncratic knowledge.
This process is called fundamental transformation (Caballero and Hammour
1996).
Furthermore, specificity typically refers to three different types of assets:
human assets, which represent the learning progress and knowledge needed for
operating a transaction or a process; physical assets, which consist of the appara-
tus required for completing the transaction; and site specificity, which describes
the need for a party to be physically located near to the other party in order to
participate in the transaction (Aubert et al. 1996a; Williamson 1981). More re-
cent sources add “dedicated asset specificity”, i.e. the investment required to
fulfill the special requests of a particular customer (Sjurts and Stieglitz 2004, 6-
7), and brand name capital (Williamson 1985; Williamson 1989).
Uncertainty describes the impact of bounded rationality on transaction risks.
Bounded rationality, as one of the behavioral assumptions of TCE (see below),
refers to the deficit in the decision maker’s “processing power”, i.e. there ability
to evaluate the (potential) consequences of all possible decision alternatives
(Williamson 1985). Since transaction partners are unable to foresee all the con-
sequences of their actions, specific risks occur. Unavoidably, uncertainty leads to
40 Theoretical Foundation and Related Research

long-term contracts being incomplete, which may require renegotiation and fre-
quent adjustments over time (Bahli and Rivard 2004; Pilling et al. 1994; Pisano
1990; Williamson 1985). Coase argues that the existence of uncertainty is a nec-
essary reason for the emergence of firms (Coase 1937).
In the context of long-term contracts, uncertainty can be classified into the
categories of technical uncertainty (not completely predictable development of
used resources or technology), task uncertainty (fuzzy and complex description
of tasks), and behavioral uncertainty (customers do not define their requirements
precisely enough and therefore frequently request changes) (Burr 2003, 114).
Finally, frequency describes the rate of a transaction’s recurrences. Transac-
tions that are processed frequently are more likely to be integrated within a firm
because specific investments become more justifiable if a transaction is carried
out more frequently (Burr 2003, 114; Coase 1937, 8; Lammers 2005, 25; Picot
1991, 347). However, there are also arguments that internal control costs rise
with increasing frequency, leading to external provision being efficient (Coase
1937, 390-403).
In addition to the basic transaction attributes of specificity, uncertainty, and
frequency, some authors add the criterion of nondisclosure needs of a transaction
object (Alchian and Woodward 1988). Furthermore, plasticity, which describes
the degree of difficulty in describing the transaction object and thus the extent of
discretionary scope, is named as an important driving force of transaction costs,
particularly in the context of IT, where huge opportunities for hidden action exist
(Monteverde 1995). Additionally, Monteverde (1995) supplemented the “extent
of unstructured non-technical dialogue” or “unstructured interaction degree”
which is necessary between the different parties involved. The problem here is
the difficulty of separating uncertainty (behavioral uncertainty when not com-
municating well enough) and human capital specificity, caused by specific in-
vestment in establishing company-wide shared communication codes. Further-
more, Picot (1995) adds “strategic relevance of the transaction” as a characteris-
tic which should be considered. Strategic relevant transactions are those which
allow a firm to differentiate its business from that of its competitors. Thus, Picot
incorporates concepts from Porter (section 2.1.5) and from the RBV (section
2.1.6).
Finally, TCE is based on two basic assumptions about the deciders’ decision
behavior: bounded rationality, explained above, and opportunism, “which refers
to the claim that humans act not only out of self-interest, but also with guile”
(Bahli and Rivard 2004, 177).
TCE compares different governance structures with regard to efficiency and
performance (Williamson 1985): market, hierarchy, and hybrids. While markets
are completely based on price mechanisms, hierarchy shows a coordination
Theoretical Foundation and Related Research 41

mechanism based on direction, incentive, and assignment of responsibilities

(Moore 1992). Hybrid governance structures use coordination mechanisms from
markets and hierarchies, including subsidiaries and joint ventures, for example.
Based on the constructs introduced above, TCE formulates the following re-
lationships: transaction costs increase with increasing specificity, frequency,
uncertainty, and plasticity; unstructured non-technical dialogue; non-disclosure
needs and strategic relevance. Based on the application domain, particular con-
clusions are possible, e.g. the make-or-buy decision shifts to “make”, with in-
creasing transaction costs.
TCE has been adversely criticized in different ways. The next paragraphs
will outline the most important deficiencies of TCE as discussed in the literature.
First, many authors criticize the focus of TCE. TCE places too little empha-
sis on production and too much emphasis on exchange. In order to be useful, any
organizational theory must place elementary objects such as production and
distribution costs at the center of its focus (Burr 2003). When applying PCE to
the question of make-or-buy of highly specific input factors, the answer would
be the same as from TCE: “make” – because a provider could not achieve any
scale effects if production is applied specifically to the customer’s needs. Never-
theless, in evaluating a particular governance design (e.g. a sourcing alternative),
deciders will have to incorporate insights from both PCE and TCE: the govern-
ance structure with the lowest sum of (expected) production costs and transaction
costs has to be chosen. Even Williamson claimed that TCE was complementary
to PCE and that production costs and transaction costs should be evaluated si-
multaneously for different governance modes (Williamson 1985; Williamson
1989). TCE researchers often make a basic assumption about production costs.
They argue that external production costs are lower than internal production
costs, since an external supplier can realize economies of skill and scale to an
extent that an internal organization unit cannot (Afuah 2003; Chiles and
McMackin 1996; Hill 1990; Monteverde 1995; Williamson 1985). “Thus, the
default governance structure for firms is market contracts, unless it can be shown
that external transaction costs are high” (Afuah 2003, 37). However, as shown in
the debate on economies of scale in the PCE section (2.1.1), this assumption does
not hold true in general.
Moreover, apart from ignoring the production costs, TCE does not consider
the benefits side (i.e. transaction benefits) (Zajac and Olson 1993). This weakens
the theory’s applicability to the domain of analyzing cooperation, joint-ventures,
etc. TCE focuses on minimizing transaction costs of a singular firm and not on
interorganizational arrangements, which aim at common achievement of advan-
tages and value creation of more than one partner. Important benefit factors in
this field are exchange of information and knowledge as well as adjustment of
42 Theoretical Foundation and Related Research

partners’ interests. Incorporating and evaluating these factors could lead to com-
pletely different decisions than when focusing only on transaction costs (Burr
2003).
Some authors focus on Williamson’s concepts of market and hierarchy and
argue that there is conceptually no difference between them and that the notion
of authority has weaknesses. In a firm, the worst possible sanction is to fire
someone. However, the difference between firing a worker and “firing“ a sub-
contractor is not of a structural nature (Moore 1992). And on the other hand:
“Why should integration reduce ex post conflicts of interest and thereby amelio-
rate the hold-up problem[9]?” (Slater and Spencer 2000). Instead, Hoetker (2002)
argues that opportunism within hierarchies can be constrained much more effi-
ciently and effectively by the management; a greater legal burden on employees
compared with contractors (Masten 1988), and judicial forbearance, which re-
stricts access by employees to the courts to resolve internal disputes (Williamson
1991).
Another element of TCE which is often criticized is the concept of uncer-
tainty. Neither Coase nor Williamson precisely define this concept: Coase does
not specify risk and uncertainty at all while Williamson conflates both (Slater
and Spencer 2000). Williamson’s “approach posits a fully determined future, in
which there exists a known list of possibilities. Bounded rationality limits are
foreseeable such that efficient governance can always be implemented” (Burr
2003). Williamson (at least in earlier works) equates complexity (which leads to
bounded rationality) with uncertainty (Slater and Spencer 2000, 74-75).
Finally, one of the most relevant points of critique is the almost non-existent
operationalization and measurability of transaction costs. However, this point is
usually dealt with by establishing a relative cost analysis. Transaction costs are
simply compared for different organizational forms, e.g. for market vs. hierar-
chy. Relative dominance criteria replace the evaluation criterion of efficiency
(Moore 1992). Here, the costs of integration (hierarchy) are often given rather
scant attention (Geyer and Venen 2001, 33).

2.1.2.2 Implications for the Sourcing Decision

In the outsourcing domain, TCE has provided a major reference point for
identifying the main cost and risk factors and for operationalizing them (Wang
2002). TCE is widely used to determine the boundary of the firm because “trans-
action cost economics specifies the conditions under which firms should manage

9
The hold-up problem describes a situation where two parties are able to work efficiently by
cooperating, but refrain from doing so due to concerns that they may give the other party in-
creased bargaining power (Klein et al. 1978).
Theoretical Foundation and Related Research 43

a particular economic exchange within their organizational boundary as well as

conditions under which it should be outsourced” (Barney 1999, 137). The ge-
neric results from TCE can be directly applied to the outsourcing domain, i.e.
low specificity of the resources underlying the particular business function, low
uncertainty and low complexity (i.e. high transparency about the dimensions of
execution, such as quality, accuracy, etc.), as well as low frequency of use, lead
to increasing advantageousness of outsourcing (Aubert and Patry 1998; Aubert et
al. 1996b; Dibbern and Heinzl 2001; Sjurts and Stieglitz 2004, 9; Williamson
1985). Therefore, the functional relationship between transaction costs and out-
sourcing of a business function deriving from the TCE perspective can be de-
scribed by
Outsourcing = f(transaction costs (-))
with transaction costs = f(asset specificity (+), uncertainty (+), frequency(+))
Langlois argues that transaction costs are only a short-term phenomenon
within sourcing partnerships. With increasing duration they will approximate
towards zero because the relationship will become more transparent and “inti-
mate” while the transaction becomes a matter of routine (Langlois 1992). By
contrast, many other authors argue that transaction costs for coordination and
renegotiation are a significant cost factor during the whole relationship (e.g.
Barthélemy 2001).
However, the application of TCE to the outsourcing phenomenon could only
seldom be confirmed empirically (see (Dibbern et al. 2003) for a list of empirical
works). One major survey on IT outsourcing based on TCE was (Ang and Straub
1998). It showed that managers consider transaction costs when evaluating out-
sourcing options and that transaction costs have a negative impact on outsourc-
ing decisions. They also recognized an instrumentalization of the transaction
costs argument by managers. If intending not to outsource, their estimate of the
transaction costs is higher than the estimate of those who are willing to out-
source. The problem of not being able to calculate transaction costs ex ante leads
to the situation that managers who favor internal IT services rely much more on
transaction costs than those who outsource IT. Other studies showed that transac-
tion costs are often not adequately considered and are under-estimated ex ante
(“hidden costs”) (Barthélemy 2001; Barthélemy and Geyer 2000; Wang 2002).
Most empirical surveys show that high specificity of the asset (i.e. the object
to be outsourced) reduces outsourcing activities because, on the one hand, speci-
ficity is often related with core competence and is therefore a strategic resource
and, on the other hand, it is often assumed that the potential insourcer is not able
to carry out the task at lower costs (at defined service and quality levels) because
realizing scale effects would not be possible (Ang and Straub 1998). In particu-
44 Theoretical Foundation and Related Research

lar, the role of human asset specificity was validated in (Aubert et al. 1996b) by
investigating why there are less software development outsourcing activities than
outsourcing of IT operations.
A further argument regarding the impact of specificity is the threat of ex-
post specificity, which can be created by the provider, leading to increasing costs
for terminating the outsourcing relationship (Monteverde and Teece 1982). By
contrast, studies also showed that, in special cases, specificity can increase out-
sourcing potential. Due to the specificity, both partners have to make specific
investments which reduce opportunism (Ang and Cummings 1997).
Within the IT outsourcing domain, an asset is specific10 if it contains a pat-
ented, self-developed technology or a proprietary innovation which is developed
by a third party but cannot be rebuilt and resold readily (Oh and Gallivan 2004,
6) or if it is cooperatively developed by the client and the vendor (Schott 1997,
3). Hence, the diffusion of standard software reduces the specificity of IT in
many areas (Englert 2000, 119/120; Schott 1997, 3), while hardware as such is
rather unspecific (Aubert et al. 1996b). The more an IT provider firm gets in-
volved in the provided business functions of the client firm (e.g. when develop-
ing, configuring, and operating business applications), the more it is able to raise
ex post specificity (Schott 1997, 14). In the IS domain, the main specificity driv-
ers are embedded business rules and human knowledge carriers. Based on Wil-
liamson, most IS functions (which regularly show low to medium factor specific-
ity) should be sourced via bilateral cooperation to legally independent firms,
based on relational contracts (Bahli and Rivard 2004; Pilling et al. 1994; Pisano
1990).
It is sometimes argued that modularization of processes and products will
lead to more outsourcing of services. Modularization decreases complexity and
(to a certain degree) specificity, making selective sourcing (Lacity et al. 1996)
more favorable and reducing the risk of opportunism by making it easier to
switch vendors (Hoetker 2002).
The second determinant for outsourcing from a TCE point of view is uncer-
tainty. If there is any uncertainty, contracts will be incomplete (cf. section 2.1.4)
and likely require renegotiation and frequent adjustments when unexpected con-
tingencies occur (Dibbern et al. 2003; Knolmayer 1993). As a consequence,
uncertainty decreases the advantageousness of outsourcing. In order to reduce
risks which evolve from the cooperation, the following risk reducing instruments
are suggested by TCE: (1) both should make specific investments to increase the
symmetry of the relationship, thereby reducing opportunistic behavior (hold-up

10
It should be noted that quantifying the specificity of IS (i.e. determining value and alternative
value of IS operations) is hardly possible. Determining the value of IS is still one of the most im-
portant tasks of IS research.
Theoretical Foundation and Related Research 45

problem) (Williamson 1985); (2) the sourcing provider is bound to the out-
sourcer by implementing “dedicated assets” (Krahnen 1991, 101) in order to
create contract-based specificity (Geyer and Venen 2001, 18; Sjurts and Stieglitz
2004).
Outsourcing IT services or business process operations requires the formula-
tion of detailed service level agreements and the implementation of dedicated
control mechanisms (Lancellotti et al. 2003; Marlière 2004b). This helps to re-
duce uncertainty and thus leads to lower transaction costs (Englert 2000,
119/120).
Empirical studies show that the implementation of interorganizational sys-
tems (IOS11) will lead to reduced coordination costs (Venkatraman 1991; Zaheer
and Venkatraman 1994), making cooperation more favorable. “This integration
could involve automation of exchange procedures and documents and the sharing
of applications and databases” (Grover et al. 2003, 222), which is especially
relevant in the IT-intensive banking industry. Based on this proposition,
Clemons developed and confirmed the “move to the middle” hypothesis, which
describes the impact of IT making cooperation relatively more favorable com-
pared to hierarchy and market (bilateral IT investments lead to closer and long-
term relationships) (Clemons and Reddi 1994; Clemons et al. 1993).
It is argued that low transaction frequency increases the opportunity of out-
sourcing, e.g. empirically validated in (Aubert and Patry 1998). However, par-
ticularly for IT outsourcing, frequency is not a relevant decision variable because
IT services are procured on a continuous basis (except in application develop-
ment) (Aubert et al. 1996a), and is therefore only considered in very few empiri-
cal studies on outsourcing (Aubert et al. 2003).
In a survey of the IT outsourcing activities of 50 firms, Barthélemy tried to
quantify transaction costs (Barthélemy 2001). He split them into search and
contracting costs, transition costs, and costs for “managing the effort” (i.e. costs
for controlling the insourcer and managing the relationship). While the respon-
dents were not able to evaluate transition costs, they could determine search and
contracting costs to be 3% of the IT outsourcing volume on average. Of course,
there are scale effects: a firm with a $500 million contract spent around 0.4% on
searching and contracting, whereas a firm with a $2.5 million deal spent around
4%. The same could be observed in the control and coordination costs. While
firms with small deals (under $10 million) spent 14% on average, medium deals
(between $10 million and $100 million) required costs of about 6%, and large
deals (over $100 million) only had control and coordination costs of 1% of the

11
IOS are defined as the technological means by which integration of multiple firms’ information
systems are carried out (Cash and Konsynski 1985; Johnston and Lawrence 1988).
46 Theoretical Foundation and Related Research

total ITO volume. Barthélemy also showed that outsourcing experience signifi-
cantly reduces transaction costs.
Finally, the following points from Lacity and Willcocks show why TCE, in
particular, provides a useful theoretical framework for research in outsourcing
(Lacity and Willcocks 1995, 204):
o TCE specifically addresses sourcing decisions, i.e. the make-or-buy deci-
sion.
o TCE captures the widely-held perception that organizational members make
sourcing decisions based on an economic rationale (cf. section 2.2.2.1 on
outsourcing advantages).
o Practitioners often use a terminology consistent with TCE to explain why
outsourcing is predicted to reduce IT costs (e.g. “commodity service”)
o TCE has enjoyed an abundance of empirical and theoretical academic atten-
tion, suggesting that other researchers find this theory to be a useful interpre-
tation of organizational reality.

2.1.3 Agency Theory

2.1.3.1 Basics and Discussion
In addition to the TCE, the second branch of institutional economics with impli-
cations for outsourcing decisions is agency theory (AT), which is based on the
works of Alchian (1950), Ross (1973), and Jensen/Meckling (1976). AT is not a
general theory of the firm and “not a general solution to the problem of organiza-
tional design” (Levinthal 1988, 155) but focuses specifically on contracting be-
tween transaction partners, when the performance of an agent cannot be observed
deterministically by the principal.
In short, agency theory deals with the diverging interests between principals
and agents (Eisenhardt 1989). Similar to the hierarchy understanding of TCE, a
firm is seen as an agency relationship built on a set of contracts between owners
(“principals”) and selfish, utility-maximizing managers (agents). AT introduces
the problem of private (asymmetric) information, i.e. the agent is able to diverge
from contractually determined employment (“moral hazard”) (Jensen 1983). “As
a consequence, when decision-making authority is delegated to agents, it cannot
be guaranteed that the decisions will be aligned with the interest of the principal”
(Gurbaxani and Whang 1991, 60). Divergent interests of the principal and the
agent, together with information asymmetry about the agent’s behavior (“hidden
action”), lead to the agency problem (Jensen and Meckling 1976, 309).
Theoretical Foundation and Related Research 47

The principal agent theory (PAT) (Ross 1973; Spence and Zeckhauser
1971), as the normative strand of agency theory12, tries to determine efficient
contracts which solve this problem, i.e. encourage the agent to behave in the
principal’s interests. The choice between input-based or behavior-based contracts
(e.g. hierarchy, vertical integration) and outcome-based contracts (e.g. market,
vertical disintegration) depends on the related level of agency costs, which repre-
sents the result of discrepancies between the objectives of the principal and those
of his agent (Gurbaxani and Whang 1991, 61). Thus, agency costs are the sum of
the principal’s monitoring costs, bonding costs, and the residual loss – depending
on the uncertainty, task complexity, and measurability of the agent’s perform-
ance (Jensen and Meckling 1976; Levinthal 1988).
The principal has the option either to implement an information system to
reduce uncertainty, making the agent’s effort observable, or to contract on the
outcome to align the agent’s behavior. If uncertainty increases, this might be the
more expensive option. Basic models of PAT assume deterministic measurability
of the outcome (in appropriate time) although this is often not the case. Outcome
measurability problems lead to outcome-based contracts being less favorable.
Agency costs represent the difference between first and second-best opti-
mum. The first-best optimum represents the principal’s gain in the case of per-
fect information while the second-best optimum represents the result after taking
asymmetric information into account. The efficiency loss as difference between
first and second-best optimum represents the upper limit for spending in an in-
formation system (Levinthal 1988).
Optimal incentive-compatible contracts are determined by using formal
mathematical models (Ewert and Wagenhofer 2003; Milgrom and Holmström
1991; Rumelt 1995). PAT is highly abstract and mathematical, and therefore less
accessible to those who study organizations (Eisenhardt 1989). It aims to be a
general theory of principal-agent relationships, being applicable to employer-
employee, buyer-supplier, client-lawyer, and other relationships (Harris and
Raviv 1979).

12
Agency theory surfaced in two different branches (Williamson 1985, 27-28). The second strand,
positivist agency theory (Alchian 1950; Alchian and Demsetz 1972), or the nexus-of-contract
view of the firm, is “concentrated on modeling the effects of additional aspects of the contracting
environment and the technology of monitoring and bonding on the form of the contracts and or-
ganizations that survive. Capital intensity, degree of specialization of assets, information costs,
capital markets, and internal and external labor markets are examples of factors in the contracting
environment that interact with the costs of various monitoring and bonding practices to determine
the contractual forms” (Jensen 1983, 334-5). The common approach of this research strand is to
empirically identify situations in which principal and management interests diverge and to dem-
onstrate which mechanisms (a certain contract or implementation of an information system) will
solve this problem (Eisenhardt 1989).
48 Theoretical Foundation and Related Research

Due to their analytical approach, the models developed are usually very ab-
stract and simple (mostly only one principal, one agent, and one period). Some
models display an agent’s performance over multiple periods (Holmström 1979;
Rogerson 1982) or implement incentive systems which are based on the output
from multiple agents’ work (Rumelt 1995). With longitudinal measures being
available, the variance of uncertain but observable output decreases and gives
more insight into the agent’s effort. If the relationship can be repeated over an
infinite number of periods, it could be shown, in terms of game theory, that the
first-best solution is achievable (Selten 1975). By incorporating concepts of
bounded rationality (see TCE, section 2.1.2.1, and Radner 1979; Simon 1955),
this phenomenon can also be transferred to a finite number of periods (Kreps et
al. 1982; Neymann 1985)13.
While moral hazard is the primary and basic focus of the majority of PAT
models, there are also approaches which deal with adverse selection and hidden
information in the context of budgeting or goal setting. In the first case, the prin-
cipal has to adjust the contract in a way that only the most competent (potential)
agents are attracted, e.g. by making the size of rewards and penalties for the
agent dependent on the success or failure of the outcome (Levinthal 1988). Goal-
setting problems can be avoided by having sufficient information about previous
results. The agent may be tempted to give inaccurate figures in order to be given
easily achievable future goals. The principal must beware of this (Levinthal
1988; Osband and Reichelstein 1985).
The most important propositions of agency theory, empirically validated by
multiple studies, can be summarized as follows (Eisenhardt 1989):
o If the contract between the principal and agent is outcome-based, the agent
is more likely to behave in the principal’s interests (reducing agency costs).
o If the principal has information to verify the agent’s behavior, the agent is
more likely to behave in the principal’s interest (reducing agency costs).
o Implementation of information systems leads to behavior-based contracts
being more favorable than output-based contracts.
o Outcome uncertainty leads to behavior-based contracts (and implementation
of an information system) being more favorable than outcome-based con-
tracts.
o Outcome measurability leads to outcome-based contracts being more favor-
able than behavior-based contracts.

13
One criticism of all of these models of repeated relationships is the assumption of independence
among periods. Intertemporal interdependence is introduced for risk-sharing and incentive pur-
poses but is not inherent in the structure of the models. For example, an agent’s activity may pro-
duce long-term effects. (Lambert 1983)
Theoretical Foundation and Related Research 49

Apart from its extreme simplification and formalization, which prevents the
PAT being applicable to most prescriptive issues, agency theory as a whole is
criticized as being too narrow and trivial, having few testable implications
(Hirsch and Friedman 1986; Perrow 1986). In conclusion, authors call for more
application-oriented and context-embedded models (Levinthal 1988, 154), which
should incorporate a broader range of contract alternatives instead of the pure
dichotomy between behavior-based and outcome-based contracts (Eisenhardt
1989).

2.1.3.2 Implications for the Sourcing Decision

Applying agency theory to the context of outsourcing requires the transformation
of the principal/agent relationship to an outsourcer/sourcing provider relationship
(Gellrich et al. 2005). From an AT perspective, the principal has to choose be-
tween an inter-firm and an intra-firm agency relationship. AT insights regarding
the outsourcing relationship can be distinguished by the following perspectives:
Ex ante (or: pre-outsourcing): The selection of an unsuitable vendor has to
be avoided. Information about the vendor’s capabilities is only partially available
to the potential outsourcer (hidden characteristics), provoking adverse selection
(Akerlof 1970).
Ex post: After establishing the sourcing partnership, problems of hidden ac-
tion or moral hazard and hold-up have to be prevented (Schott 1997). As pointed
out earlier, the outsourcer has to choose between implementing an information
system (monitoring) and outcome-based contracting (including positive and
negative benefits) to insure incentive compatibility.
Consequently, outsourcing a business function usually raises agency costs
because, in most cases, it is easier to control efforts within the firm than to moni-
tor autonomous business partners (Williamson 1990, 134), due to more easily
implementable controlling and incentive instruments (Alchian and Demsetz
1972). By contrast, sometimes measurement problems and opportunistic behav-
ior exist within a firm, which actually leads to outsourcing (Wang 2002).
The problem of hidden characteristics is generally insignificant because
many specific investments are necessary in most cases and the market of sourc-
ing providers in a wide range of fields is quite clearly set out (Schott 1997, 187).
Investments in outsourcing relationships increase specificity and therefore in-
clude signal quality. If a service provider did not have the necessary clout and
capability, it would be detrimental for him to undertake a too great business risk.
Another way the sourcing provider can (help to) reduce the outsourcer’s
risks resulting from hidden characteristics and hidden action is by taking over
equity of the outsourcer (Cunningham and Fröschl 1995). Or, often related to
outsourcing deals in the banking industry, the sourcing provider places deposits
50 Theoretical Foundation and Related Research

in the customer bank (Schott 1997). Other bonding activities such as implement-
ing one’s own control and logging systems or using standardized resources (to
facilitate an exit option for the outsourcer) are also common in order to raise the
outsourcer’s trust in the partnership.
As already discussed in the section on TCE, the implementation of interor-
ganizational information systems (IOS) leads to a reduction of coordination costs
(p. 45). From an AT perspective, these IOS represent (at least partially) the im-
plementation of monitoring capabilities which can reduce the risk of opportunis-
tic behavior.
Furthermore, the problem of hidden action might be significantly less prob-
lematic in the context of cooperative sourcing. Since the insourcing firm pro-
vides the same services to the outsourcers as well as to itself, an inherent congru-
ence of interests does already exist. Moreover, since the outsourcer in a coopera-
tive sourcing scenario knows a lot about the outsourced business function, the
level of uncertainty will be lower.
The formal models based on PAT show that the less risk-averse entity has to
bear the risk. For example, in the (hypothetical) case of a risk-neutral agent and a
risk-averse principal, PAT models suggest an incentive system which is similar
to selling the business area to the manager and letting the principal earn just a
fixed amount; which is simply another way of describing outsourcing (Shavell
1979). On the other hand, it can be shown that legal company types with access
to capital markets – in contrast to individual managers or share holders – are able
to act on a risk-neutral rationale because they can hedge any existent risk aver-
sion by capital market transactions14 (Ewert and Wagenhofer 2003, section 5.3).
Consequently, adopting formal PAT models to the outsourcing field leads to
limiting models to this special case of a two-sided risk-neutral principal/agent
relationship15.
In conclusion, the agency theory view of outsourcing hypothesizes the fol-
lowing relationships regarding the outsourcing decision (adapted taken from
Cheon et al. 1995):

14
If the following conditions are given (DeAngelo 1981):
spanning: the set of existing financial instruments, available at the capital market contains all pos-
sible cash flow structures which can be created by the firm’s own activities.
competitivity: static valuation system for market values (in a certain environment comparable to a
stable interest rate for maximizing net present value).
15
By contrast, it can be argued that all decisions within a firm are made by individuals who are risk-
averse (and their individual benefits are usually also directly related to the results of their deci-
sions). Nevertheless, this effect is ignored in this work since firms are considered as monolithic
entities in this research, regardless of internal governance structures.
Theoretical Foundation and Related Research 51

Outsourcing = f(agency costs (-))

with agency costs16 = f(information asymmetry (+), length (-))
with information asymmetry = f(uncertainty (+), measurability (-))
Outsourcing is less favorable if agency costs can be assumed to be high.
“Agency costs [...] increase in outsourcing relationships with high uncertainty,
high risk aversion, […] low outcome measurability and greater length of rela-
tionship” (Cheon et al. 1995, 215). As discussed above, the level of risk aversion
might be negligible, therefore, Cheon’s function was reduced by this term.
There is some discussion about adopting PAT as a theoretical lens for inves-
tigating why firms are outsourcing particular business functions. Some authors
such as Lammers (2005) argue that this theory “is useful for defining how the
contractual relationship with the insourcer should be drafted but not very helpful
for determining what activities shall be sourced internally and what externally”
(p. 39). By contrast, other authors have adopted this theory for investigating the
research questions of why and what to outsource (Loh 1994; Nelson et al. 1996;
Poppo and Zenger 1998) and are surprised that it does not receive much attention
(Logan 2000; Poppo and Zenger 1998).
As empirical evidence, studies by (Loh 1994; Loh and Venkatraman 1995;
Nelson et al. 1996; Poppo and Zenger 1998) show most of the propositions de-
rived from AT to be supported in different fields of IT outsourcing. Risk of op-
portunism and difficulties in contractual regulation, goal alignment, or vendor
monitoring lead to outsourcing being less favorable. The negative relation be-
tween measurement difficulties with the vendor’s performance and outsourcing
has shown to be weak (Poppo and Zenger 1998).

2.1.4 Theory of Incomplete Contracts

2.1.4.1 Basics and Discussion
One idea that is essential to modern theories of the firm (including TCE) is the
idea that contracts are imperfect or incomplete – meaning that contracting parties
are not able to specify or observe all contingencies that may affect the outcome
of their relationship (Milgrom and Roberts 1992) – so that the first-best effort
levels cannot necessarily be obtained (Feenstra and Hanson 2003). Complete
contracts are considered infeasible or at least associated with high transaction
costs (Williamson 1975).

16
Being more precise, one must argue that length decreases agency costs per period but not overall
agency costs.
52 Theoretical Foundation and Related Research

The theory of incomplete contracts (TIC) (Grossman and Hart 1986; Hart
1988; Hart and Moore 1990) is based on the same behavioral assumptions as
TCE (bounded rationality, opportunistic behavior) (Gietzmann 1996) and applies
the property rights concept in organizational contexts by using formal models to
identify situations that prevent the formulation of complete contracts as a conse-
quence of real-world complexity combined with bounded rationality (Gebauer
1996).
TIC argues that certain variables (e.g. efforts of the agent/insourcer) might
be observable by the parties but not verifiable by a third party, such as a court or
an intermediary (Bakos and Brynjolfsson 1993a). Therefore, contracts based on
the outcome of such variables are inappropriate.
Hart and Moore (1990) show that, if complete contracting is not possible,
optimal investment levels generally cannot be reached; thus only a "second-best"
outcome can be achieved. The parties involved have to divide the resulting out-
come of the relationship by ex post bargaining. This leads to the hold-up problem
because the agent is reluctant to make relation-specific investments in dedicated
assets when he has to fear ex post exploitation by the principal, who knows about
the specificity, and thus the decreased bargaining power of the agent. In this
case, the principal, in particular, has to implement mechanisms which build up
trustworthiness and reputation.
The completeness of a contract is assumed to be influenceable by the con-
tract parties. But, increasing a contract’s degree of completeness leads to higher
transaction costs. Uncertainty of a transaction directly refers to the inherent in-
completeness of the contract and high frequency increases the involved risk,
usually due to lower efforts and specific investments with unique transactions
(Aubert et al. 2003). Thus, contract incompleteness can be seen as an mediator in
the positive impact of uncertainty and frequency on transaction costs (cf. TCE,
section 2.1.2).
To safeguard against the threat of hold-up, deciders adopt relational con-
tracts. Relational contracts specify only general terms and objectives of a rela-
tionship as well as mechanisms for dispute resolution (Milgrom and Roberts
1992). They do not try to explicitly deal with of all future contingencies but are,
nevertheless, long-term arrangements (Furubotn and Richter 1998). Often they
include informal and unwritten agreements (Baker et al. 2002). The objective is
“to provide a framework for resolving unforeseen disputes and a social process,
based on norms of trust, mutuality, and solidarity” (Poppo and Lacity 2002,
256). Relational and formal contracts are interrelated in a substitutive as well as
in a complementary sense. The joint use of both mechanisms provides more
efficient outcomes (Poppo and Zenger 2002) but contract completeness can also
Theoretical Foundation and Related Research 53

damage the exchange performance by undermining relational governance

(Macaulay 1963)17.
TIC has been used to treat issues in buyer-supplier relationships, such as de-
termining the optimal number of suppliers, the optimal type of coordination
between buyers and suppliers, and investigating the impact of information sys-
tems on organizational form, on coordination of suppliers, and on the supplier
structure itself (Bakos and Brynjolfsson 1993a; Bakos and Brynjolfsson 1993b;
Clemons et al. 1993).

2.1.4.2 Implications for the Sourcing Decision

TIC emphasizes that relationship-specific investment is distorted due to the
hold-up problem arising from the inability to fully reward investments within the
context of incomplete contracts (Spencer 2005). This will affect outsourcing
contracts, as long as specific investments in the relationship are necessary. Usu-
ally, business process outsourcers are forced to adopt a reference process offered
by the vendor, enabling economies of scale (cf. section 2.1.1.2). This adoption
leads to at least some homogenization of process design and standardization of
data formats within the industry. Consequently, the hold-up problem triggered by
incomplete contracts may be less problematic in outsourcing and in cooperative
sourcing relationships, in particular. Nevertheless, interfaces and certain steps of
customization have to be provided by the insourcer.
The design of the outsourcing contract follows a trade-off between the mar-
ginal benefits and the marginal costs of contractual completeness (Aubert et al.
2003). For (industry-wide) standardized, transparent, and highly automatable
processes (e.g. payments or securities processing in the banking industry), this
implies a higher degree of completeness than for more complex tasks (e.g. SME
loans processing) (Wüllenweber et al. 2008). TIC proposes the degree of out-
sourcing to be – with contract completeness as mediator – related to uncertainty,
measurability, and complexity as well as to the standardization degree (Nam et
al. 1996).
The impact of relational governance, as the second main factor for outsourc-
ing, is two-fold. First, together with contract completeness, relational governance
leads to outsourcing being more favorable (Poppo and Zenger 2002) but it also
substitutes contract completeness as argued above (Macaulay 1963; Corts and
Singh 2001)18.

17
The relationship element represents the analysis object of relationship theories discussed in
section 2.1.8.
18
cf. also section 2.1.8 on relationship theories.
54 Theoretical Foundation and Related Research

Outsourcing = f(contract completeness (+),relational governance (+))

with contract completeness = f(uncertainty (-), measurability (+), complexity (-),
standardization degree (+), relational governance (-))
TIC has received hardly any attention in empirical outsourcing research.
Lacity and Hirschheim (1993a) argue that it is a myth that an outsourcing rela-
tionship can be primarily set up as a strategic partnership. This leads to specify-
ing contracts which are too incomplete, include high risks and provoked many
outsourcing failures in the early years.
Nam et al. (1996) based their study on TIC and TCE and found a significant
relationship between uncertainty and the degree of outsourcing/ownership.
Poppo and Zenger (2002) showed that managers were less satisfied with the cost
performance of outsourced IT services when they could not easily measure its
performance. They also found a complementary relationship between relational
governance and contract completeness (or what they call “complexity”). Aubert
et al. (2003) investigated factors influencing the completeness of (IT) outsourc-
ing contracts and found that higher uncertainty, less measurability, and less per-
manency in the relationship led to less complete contracts. Loh (1994) analyzed
the different cost categories related to outsourcing and showed that “contract
resolution” (i.e. contract completeness) had the highest impact on total transac-
tion costs, with a negative impact on the outsourcing degree. In another study, he
and Venkatraman found the risk of opportunism to be related to “the extensive
and complex range of details and contingencies for IT-based contracting”, lead-
ing to fewer ITO activities (Loh and Venkatraman 1995).
The following sections will outline the strategic perspectives of outsourcing.

2.1.5 Porter’s Positioning Framework

2.1.5.1 Basics and Discussion
During the 1980s, concepts emerged which tried to explain market success by
firm-external factors (market focus). Porter’s concept of the “five forces” has
been one of the basic works in this field. In his comprehensive theory of the firm,
he argues that the attractiveness of industries is determined by the threat of po-
tential new entrants, threat from substitutes, bargaining power of suppliers and
customers, and the competitive environment (number and power of competitors)
(Porter 1985). Driven by this market-oriented focus, Porter further deducts a
complementary firm focus. A firm’s success depends on choosing the appropri-
ate competitive strategy within the given industry structure (Porter 1985). Porter
proposes analyzing firm-specific value chains to identify the firm’s strengths and
Theoretical Foundation and Related Research 55

weaknesses and its particular activities (cf. section 1.5.1) compared with those of
its competitors. He develops two different basic positioning strategies (Porter’s
positioning framework, PPF), which are cost leadership and differentiation, and
he argues that only a sustainable, dominant position leads to competitive advan-
tage; otherwise it is only a temporary skimming of market opportunities.
Cost leadership can be defined as making a product or offering a service of
identical quality at lower costs compared with competitors. Cost leadership de-
pends on production costs and thus it is driven by economies of skill and scale as
well as by economies of scope (or “linkages” inside the firm) and, also by “verti-
cal linkages” between one’s own and one’s partners’ activities (Porter 1985,
70ff) (cf. section 2.1.1.1 on PCE). In order to determine the sources of cost ad-
vantage, not only the product itself but each part of the value chain of the firm
must be analyzed (Lammers 2005) to identify all relevant cost drivers. In addi-
tion, “absolute and relative costs will change over time independent of its [i.e.
the firm’s] strategy” (Porter 1985, 95). Therefore, drivers of cost dynamics must
also be taken into account (e.g. industry growth, differential scale sensitivity,
different learning rates, aging, etc.) (Porter 1985, 95-97).
As the second strategy, a “competitive advantage through differentiation can
be achieved if a company is able to serve specific needs of customers that are not
provided by other companies within the industry and for which the customer is
willing to pay a premium on the market price” (Lammers 2005, 17). Again,
competitive advantage can be achieved in every activity of the value chain.
In conclusion, the positioning framework makes it possible to structurally
analyze both the competitive environment and the internal strengths and weak-
nesses, to select one of two basic strategy types for each product and each market
the firm is operating on. Nevertheless, authors argue that the PPF is solely mar-
ket-oriented (i.e. outside-in perspective: the strategy that is enforced by the mar-
ket is adopted by the firm) and call for a complementary internal view (Barth
2003, 113) as provided by the resource-based view (section 2.1.6).

2.1.5.2 Implications for the Sourcing Decision

Porter proposes that the object of analysis – especially with regard to sourc-
ing decisions – should be the different activities of the value chain. Each activity
generates costs and values, thus it is possible to be a cost leader for particular
activities but not for the overall product. Given sufficient separability of the
different activities, cost improvement in inferior activities can be reached by
selectively outsourcing them (Lammers 2005). Thus, the PPF adopts the PCE
arguments (cf. section 2.1.1.2).
56 Theoretical Foundation and Related Research

On the other hand, from an insourcer’s view of generating further revenue, it

is necessary to differentiate one’s own business from that of one’s competitors in
terms of cost leadership or to provide higher quality at equal cost.
Several authors hypothesize that the type of strategy chosen has a moderat-
ing impact on other factors driving the outsourcing decision (e.g. (Grover et al.
1994)). For example, if a firm follows a cost leadership strategy, efficiency will
be critical. This leads to cost-related questions (cf. PCE and TCE) being more
relevant for the outsourcing decision, compared with a differentiator. Due to this
moderating effect, this section does not present a functional relationship as in the
sections before, but will include the moderating relationship in the literature
synthesis given in section 2.1.10.1 and especially also in Figure 6.
The PPF has not received much attention as a foundation for empirical re-
search on outsourcing phenomena. Porter’s theory of the firm is more norma-
tively oriented towards supporting strategic management than towards explaining
markets and organizational phenomena.

2.1.6 Resource-Based View and Core Competence View

2.1.6.1 Basics and Discussion
According to Porter’s positioning framework (section 2.1.5), sustainable advan-
tage can mainly be realized by market-oriented competitive strategies. The firm’s
market position is safeguarded by installing competitive barriers. The existence
of differences in the firms’ success is explained by the value chain concept,
which represents the interface to the resource-based view (RBV). With regard to
Porter's perspective, the RBV mainly criticizes (1) the assumption of basic simi-
larity between competing firms, and (2) a too static view, which over-emphasizes
established industries and is especially problematic, for example, when investi-
gating industrialization tendencies in the banking business. This major transfor-
mation of a whole industry requires a more dynamic and modular view. The
basic message of the RBV is to utilize one’s own capabilities and resources to
develop a competitive advantage within a field in which the firm’s borders are
increasingly blurred (Barney et al. 2001, 625-626; Hoopes et al. 2003, 889-891;
Wernerfelt 1984, 172-175).
Building on the basic works of Penrose (Penrose 1959), the RBV handles
the question of why firms within the same industry differ in success and market
position. In order to explain sustainable competitive advantages and the resulting
inter-firm heterogeneity, Penrose started “to look into the firm itself” (Penrose
1959, 44). She conceptualized the firm as a collection of resources, each of them
being a bundle of potential productive services, bound together in an administra-
Theoretical Foundation and Related Research 57

tive framework (Penrose 1959). The RBV introduces the concept of individual
and innovative assembling of firm resources and describes how the resources and
capabilities of a firm may contribute to its competitive success (Barney 1999;
Foss and Eriksen 1995, 54; Peteraf 1993). Resources are often defined as the
superset of assets (“anything tangible or intangible the firm can use” (Wade and
Hulland 2004, 109)) and capabilities (or skills, processes) (Sanchez et al. 1996;
Wade and Hulland 2004), including everything that is used or applied to generate
a product or a service. Barney (1991) groups firm resources into three categories:
physical resources, human resources, and organizational resources. He defines
the capability of a firm as its capacity to perform an activity as a result of orga-
nizing and coordinating the productive services of a group of resources. More-
over, if a firm possesses a certain capability which is superior to those of its
rivals, this capability becomes the firm’s competence (Tsang 2000). Wernerfelt
(1984) argues that resources and products are just two sides (input/output) of the
same coin.
The goal of assembling resources is to create a situation that strengthens the
firm’s position and impedes competitors from catching up. Resources in the
RBV’s sense are, therefore, only relevant if they have the following properties
(Barney 1991; Beimborn et al. 2005b; Grant 1996b):
o Valuable – Every business consists of resources that enable it to carry out
the necessary activities in order to produce goods or deliver services
throughout the value chain. While some of these resources might perform
adequately (or poorly), others must be superior if the business is to outper-
form its competitors (Day 1994). These “valuable” resources make a contri-
bution to the generation of "perceived customer benefits” (Prahalad and
Hamel 1990) and thus lead to competitive advantage (Foss et al. 1995; Por-
ter 1991). In a less strict sense, valuable resources cover everything which is
necessary to conduct the intended business.
o Rare – In order to be rare, a resource must be immobile and it should not be
(or only at prohibitive costs) reproducible.
o Imperfectly imitable – The non-imitability of resources is considered by
many authors to be one of the key factors of competitive advantage (Barney
1991; Grant 1991; Peteraf 1993; Prahalad and Hamel 1990; Rasche 1994;
Reed and DeFillippi 1990). For example, the reason for Penrose’s claim that
the firm’s accumulated pool of knowledge can be a source of sustained
competitive advantage is believed to be grounded in the nature of knowl-
edge, which is more “experience-based” than “objective” (Knudsen 1995),
thus making it difficult for other firms to imitate. Although Porter argues
that “barriers to imitation are never insurmountable” (Porter 1985, 20), the
height of those barriers is a determining factor in how contestable or how
58 Theoretical Foundation and Related Research

sustainable the firm’s advantage will be (Reed and DeFillippi 1990). Reed
and DeFillippi consider tacitness, complexity and specificity of a firm’s
skills and resources as the generating factors for causal ambiguity which on
its part can raise barriers to imitation. In his discussion about factors which
influence the rents from resources, Grant considers limited replicability and
transferability of a resource position (due to geographic immobility, imper-
fect information, specific resources, and immobile skills) as being generators
of sustained competitive advantage (Grant 1991).
o Non-substitutable – A firm’s resources cannot be sources for sustained com-
petitive advantage if there are strategically equivalent valuable resources
that are themselves neither rare nor non-imitable so they can be imple-
mented by other firms (Barney 1991). Only capabilities which are difficult
to substitute either by similar or by different capabilities can be sources of
sustained competitive advantage and should be treated in the same way as
non-imitable resources.
o Interconnected – Apart from the “traditional” list of attributes required for a
resource, some works have introduced the property of interconnectedness of
a resource or a capability. Interconnectedness describes how “deep” a re-
source or capability is anchored into existing business processes (Beimborn
et al. 2005b; Grant 1996a). The more interconnections a capability is in-
volved in, the more complicated it will be to extract it out from an existing
context or even make simple changes to the capability itself.
One can argue that competitive advantage is realized by a valuable and rare
resource but that it is only sustainable if the resource is also imperfectly imitable,
and non-substitutable (Barney 1991, Teng et al. 1995).
The concept of a firm’s resources or capabilities can be extended to repre-
sent shared capabilities, also called industry capabilities. These are defined as
non-proprietary capabilities that are shared among a group of firms and may
yield rents, even in the absence of explicit coordination (Foss and Eriksen 1995).
The RBV “has led to a much improved understanding of firms’ diversifica-
tion strategies (Montgomery and Wernerfelt 1988) and of the underlying condi-
tions for sustained competitive advantage (Barney 1991; Peteraf 1993)” (Foss
and Eriksen 1995, 43). Empirical studies of firm performance using the RBV
have found differences not only between firms in the same industry (Hansen and
Wernerfelt 1989) but also within more granular industry segments. They imply
the effects of individual, firm-specific resources on performance being signifi-
cant (Mahoney and Pandian 1992; Wade and Hulland 2004). Further studies are
showing that the impact of firm-specific attributes on firm performance is more
important than industry factors. One interesting finding is that the differences in
the “long-term rates of return of firms within industries is five to eight times as
Theoretical Foundation and Related Research 59

large as the variance in return across industries” (Rumelt 1981, 5; recited from
Venkatraman and Camillus 1984, 519).
The RBV is very closely linked to the core competence view (CCV). Core
competencies differentiate a firm from its competitors and form the base of sus-
tainable competitive advantage (Prahalad and Hamel 1990). Core competencies
are what make an organization ‘unique in its competitiveness’ (Quinn and Hil-
mer 1994; Stewart et al. 2002, 3). Characteristically, they provide potential ac-
cess to a wide variety of markets (or are applicable to multiple purposes), make a
significant contribution to perceived customer benefits, and are difficult to imi-
tate (Krüger and Homp 1997; Lammers 2005; Prahalad and Hamel 1990).
Researchers have tried to explain firm behavior and competitiveness in
terms of concepts such as capabilities (Collis 1996; Hoopes et al. 2003; Tsang
2000), competencies (Duhan et al. 2001; Sanchez et al. 1996), and knowledge.
Other authors additionally use the terms skills (Grant 1991), strategic assets
(Amit and Schoemaker 1993), assets (Ross et al. 1996), and stocks (Capron and
Hulland 1999) more or less synonymously. These perspectives are either based
on or closely related to the logic of the RBV. For example, if there are huge
overlaps between resources, firm capabilities, and competencies, a firm’s capa-
bility is defined as the capacity to “perform an activity as a result of organizing
and coordinating the productive services of a group of resources” (Tsang 2000,
216). If a firm can perform an activity better than its competitors, this capability
represents a competence (Prahalad and Hamel 1990). In RBV terms, this capabil-
ity is a valuable resource that can be used to gain competitive supremacy
(Barney 1991)19.
Since both RBV and CCV stress the need for firms to capitalize on their
unique assets and to develop management strategies to exploit the advantages
from strategically positioned resources, the terminology for both lines of thought
will be used in the remainder of this work, as also done in (Stewart et al. 2002,
3), for example.
RBV researchers acknowledge that a firm’s resources include its ability to
implement and exploit valuable IT functions (Barney 1991; Stewart et al. 2002,
2). The RBV, for example, has been used to explain the relationship between IT
and sustained competitive advantage (Clemons 1991; Clemons and Row 1991)
and the strategic role of the firm’s IT resources (Brynjolfsson and Hitt 1996;
Clemons 1991; Clemons and Row 1991; Sethi and King 1994; Wade and Hul-
land 2004). Other authors define a firm’s “information resource” as the actual
business-relevant information but also include the information systems that fa-

19
There is much discussion about the similarities and differences between these concepts. For more
information, see (Amit and Schoemaker 1993; Dierickx and Cool 1989; Stewart et al. 2002).
60 Theoretical Foundation and Related Research

cilitate information access and acquisition (King and Grover 1991; Strassmann
1989; Teng et al. 1995).
Just like any other theory, the RBV is faced with a number of criticisms.
One major point of criticism concerns the RBV’s core result. Porter and other
authors claim that it involves circular reasoning: “successful firms are successful
because they have unique resources. They should cultivate these resources to be
successful” ((Porter 1991, 108), see also (Mosakowski and McKelvey 1997;
Tsang 2000)). When faced with the complexities of a real firm, it is often diffi-
cult to identify which of a firm’s resources are critical for its success. Often it is
a reverse argumentation: “once a firm is recognized as successful, the resources
behind the success are labeled as valuable” (Foss et al. 1995, 8). Moreover, the
difficulty in assessing the value of resources might be due to the fact that it is
impossible to measure them separately: “it may be a system of resources that
matters, not the individual resources taken separately” (Foss et al. 1995, 8).
Another item of discussion is the endogenous construct of the RBV which
focuses on the sustainability of competitive advantage without conceptualizing
competitive advantage itself (Barney 2001; Priem and Butler 2001). One solution
might be to use Porter’s framework to argue that competitive advantage is
achieved either by cost leadership or by differentiation (cf. section 2.1.5). This
would lead to the suggestion of integrating Porter’s strategic management
framework and the RBV (Mahoney and Pandian 1992), which include comple-
mentary aspects: “it is not possible to determine a competitive advantage by just
analyzing the internally developed resources” (Gewald and Lammers 2005, 3).
Finally, a minor point of criticism is the previously mentioned terminologi-
cal ambiguity when using concepts such as resources, capabilities, competencies,
etc. “This terminological ambiguity stems from the fact that the RBV is far from
a coherent perspective” (Tsang 2000, 216-217). A detailed critical discussion of
the RBV is given in (Foss 1998).

2.1.6.2 Implications for the Sourcing Decision

Most of the literature on RBV and CCV is of an explanatory nature, pointing out
that firm-specific resources enable competitive advantage and, implicitly, are at
the origin of heterogeneity among firms on the market. Several sources (Dibbern
and Heinzl 2001; Roy and Aubert 2002) give an explanation of how outsourcing
of resources can affect a firm’s ability to achieve and sustain competitive advan-
tage.
As described above, the main assumption of the RBV and CCV is that firms
are equipped with different rare and valuable resources, which enable them to
develop core competencies and a sustainable competitive advantage (Williams
2001). Therefore, business functions that do not belong to a firm’s core compe-
Theoretical Foundation and Related Research 61

tence or not to a relevant resource in the RBV’s sense, can be outsourced. Out-
sourcing is thus only preferable if the resource is neither scarce nor imitable nor
non-substitutable. Otherwise the firm’s competitive position would be weakened
(Ang 1994; Stewart et al. 2002). Furthermore, to achieve efficiency, firms should
refocus on their core competencies (Barney 1991; McFarlan and Nolan 1995).
For resources that are not within this focus, outsourcing not only can be, but even
should be considered (Knaese 1996, 56-59; Quinn et al. 1990). Outsourcing
depends both on the resources’ attributes (value, rareness, imperfect imitability,
non-substitutability, interconnectedness) and on the relation of the resources to
the activity that is to be outsourced. The more distinctly these attributes are de-
veloped, the less favorable it is to outsource the activity. Thus, the outsourcing
decision can be formulated as the following relationship (Cheon et al. 1995;
Moore 1992):
Outsourcing of an activity = f(attributes of related resources)
Moreover, one can argue that cooperative sourcing in contrast to traditional
outsourcing (to third parties from different industries) will take place if the rele-
vant resources imitable, substitutable, and rather less interconnected, but also
valuable and rare. Subsequently, cooperative sourcing can lead to a joint industry
capability for the participating firms which, based on the cooperation itself, can
create unique characteristics which help to gain a sustainable competitive advan-
tage (Foss and Eriksen 1995; cf. the previous section). For example, higher proc-
ess volumes can lead to the development of a superior transaction system being
advantageous, which in turn will lead to the generation of unique competencies
resulting in cost leadership not only from economies of scale but also from supe-
rior capabilities (or “economies of skill” in PCE terms).
In the IT outsourcing context, Lacity et al. (1996) evaluate (IT) capabilities
regarding their contribution to business processing (critical or only useful) and to
business positioning (commodity or differentiator) and suggest capabilities
which are “useful commodities” be outsourced. In the case of critical commodi-
ties, outsourcing could also be an option after evaluating further criteria.
The strategically oriented RBV is often applied as complementary to the ef-
ficiency-based TCE (Astley 1985; Tsang 2000). Empirical works on B2B coop-
eration and outsourcing try to incorporate both as theoretical foundations. They
show that the explanatory power of organizational behavior will significantly
increase by applying this theoretical double lens. TCE focuses on the costs while
RBV recognizes the strategic value of a transaction (Tsang 2000).
Nevertheless, the application of the RBV to empirical outsourcing research
appeared rather late (Teng et al. 1995) and surprisingly seldom (Dibbern et al.
2004; Stewart et al. 2002) while efficiency reasons (cost orientation) were much
62 Theoretical Foundation and Related Research

more frequently discussed. In IT outsourcing research, Stewart argued that

“whereas case study evidence suggests that managers think value is created in
the IT function through strategic focus on core competencies (Lacity et al. 1994),
there has been no quantitative effort to study the relationships between views
about IT as a strategic resource and control of ITO” (Stewart et al. 2002, 4).
RBV links the prudent management of strategic resources to increased firm
performance. Grover et al. (1996) found strong relationships between the degree
of ITO and particular aspects of firm performance, more precisely “the ability to
focus on core competencies, the ability to utilize human and technological re-
sources of the provider, and the ability to gain access to leading-edge IT and to
avoid the risk of technological obsolescence”. Similar results were found by
Goles (2003) and Dibbern and Heinzl (2002). Advantages of vertical disintegra-
tion were also shown in (Bettis et al. 1992; D'Aveni and Ravenscraft 1994;
Gilley and Rasheed 2000; Lei and Hitt 1995).
The impact of these findings is slightly weakened by turning around the first
critical remark on the RBV discussed above (section 2.1.6.1): Determining “can-
didates” to be outsourced by using the suggestions of the RBV implies outsourc-
ing non-performing functions, which obviously should lead to an increase in
overall performance (when assuming effective and efficient outsourcing man-
agement). Furthermore, many other empirical studies did not find positive rela-
tionships between IT outsourcing and overall firm (financial) performance (Loh
and Venkatraman 1992b; Poppo and Zenger 1998; Teng et al. 1995).
When applying the RBV to BPO in the banking industry, it can be argued
that, for example, many of the highly automated business functions, such as
payments processing or securities processing, do not rely on relevant resources
(neither heterogeneous nor non-imitable) for many banks (Schrauth 2004, 59)
although they constitute a major part of the banking business.
While Brandenberger (1995) describes the core competence of a bank as the
ability to mediate between financial capital and information, Marlière (2002)
more concretely distinguishes the following six core competencies in the banking
industry: payments processing; access to and pooling of financial resources (e.g.
capital market); temporal and spacial transformation of business resources; risk
management; pricing (supporting decentralized decision making); and overcom-
ing incentive incompatibilities and information asymmetries between market
participants (supporting the design of “optimal” contracts (e.g. in the swap busi-
ness)).
Deregulation has enabled foreign players and non- and near-banks to enter
the markets, while the increasing capability and availability of information tech-
nology has made banking services easier to copy, with the result that these core
competencies have increasingly lost their “core character” in recent years (Mar-
Theoretical Foundation and Related Research 63

lière 2002). Consequently, it is now normal in the German banking industry to

outsource some of these functions, such as payments and security processing (cf.
section 3.4).
This example reveals the close relationship between PCE and RBV. The in-
sourcer can strengthen its capabilities by increasing volume (economies of scale
drive economies of skill, cf. section 2.1.1.1) and making specific investments,
leading to reduced imitability, which is more favorable.

2.1.7 Resource Dependency Theory

2.1.7.1 Basics and Discussion
While the RBV focuses on company strengths, resulting from superior resources,
the resource dependency theory (RDT) recognizes that a firm is dependent (to
varying degrees) on resources that have to be acquired from the external envi-
ronment to ensure its survival (Aldrich 1976; Pfeffer and Salancik 1978; Thomp-
son 1967). This dependence usually relies on the control of necessary resources
by others (Kotter 1979). “Thus, RDT stresses the organizational necessity of
adapting to environmental uncertainty, coping with problematic interdependence
and actively managing or controlling resource flows” (Cheon et al. 1995, 212).
The RDT provides three dimensions of organizational task environments
(Cheon et al. 1995; Pfeffer and Salancik 1978):
o Concentration: the extent to which power and authority between the firm
and the environment is dispersed
o Munificence: the availability or scarcity of critical resources
o Interconnectedness: the number and pattern of linkages between the firm
and the environment
Each dimension differs depending on “to the nature and the distribution of
resources in environments with different values on each dimension implying
differences in appropriate structures and activities” (Aldrich 1976, 54). In the
context of these dimensions, firms adopt strategies to secure access to critical
resources, to stabilize relations with the environment, and to ensure survival
(Pfeffer and Salancik 1978; Zeithaml and Zeithaml 1984). Consequently, the
objective of firms to ensure their resource supply but not to lose autonomy is the
reason for cooperation between firms (Pfeffer and Salancik 1978, 257; Ulrich
and Barney 1984; Van Gils 1984). Cooperation has to be designed in a way so
that resource dependency (and the loss of control involved) becomes minimal
(Pfeffer 1982).
64 Theoretical Foundation and Related Research

Cooperation strategies that help provide access to scarce and valuable re-
sources include controlling or acquiring other firms, cooperating with other firms
(e.g. joint ventures, outsourcing) or aligning with and linking to more powerful
organizations (Ulrich and Barney 1984, 472).
Pfeffer and Salancik describe three resource dimensions as critical in deter-
mining the external dependence of a firm on another (Pfeffer and Salancik 1978):
o Importance: extent to which the organization requires the resource for con-
tinued operation and survival
o Discretion: extent to which the interest group has control over the resource
allocation and use
o Alternatives: extent to which there are alternatives
The RDT only provides a complementary perspective for investigating or-
ganizational behavior. It neither takes into account any efficiency arguments as
TCE does (Sydow 1992, 198) nor does it incorporate a comparative internal view
of firm-internal resources as RBV does (Dibbern 2004, 35).

2.1.7.2 Implications for the Sourcing Decision

Since the RDT emphasizes the dependence of organizations on external re-
sources, it provides a useful additional perspective for investigating a firm’s
outsourcing decision (Cheon et al. 1995). Through outsourcing, a firm can obtain
scarce human resources, expertise, and technological resources from the external
environment in order to enhance its long-term survival (Teng et al. 1995). Thus,
outsourcing from the RDT perspective can be described as a strategic step to
increase the dependence of one firm on another in order to obtain critical re-
sources (Jayatilaka et al. 2003).
Based on the description of the RDT given above, the following functional
relationship can be derived (Cheon et al. 1995):
Outsourcing = f(resource gaps (+))
œ outsourcing = f(importance (+), discretion (+), alternatives (+), strategy)
with importance, discretion, alternatives =
f(concentration (+), munificence (+), interconnectedness(+))
Of course, as argued above, the RDT perspective on the outsourcing phe-
nomenon can be applied only in a complementary way because it neglects effi-
ciency issues and a comparative internal view (Dibbern 2004, 35).
Grover et al. (1994) and Teng et al. (1995) explicitly considered RDT as part
of their empirical outsourcing research. Combined with an internal view (based
on RBV), they could, for example, validate the hypothesis that perceived dis-
crepancies (between the desired and actual level) in IS quality (information qual-
Theoretical Foundation and Related Research 65

ity and IS support quality) lead to significantly more IT outsourcing. However,

when using a monetary measure (IS cost-effectiveness), both studies showed
only insignificant results. Levina et al. (2003) conducted a single case study
partially based on RDT where they investigated the ITO relationship between a
sourcing vendor and a large telecom company. They argued that an outsourcer
cannot realize particular capabilities in the same way as the vendor is able to do.
For example, the vendor firm can pay higher salaries to IT experts and offer
them more attractive career plans; further, it exploits stronger experience-based
learning due to having multiple clients.
When applying the RDT to the banking industry, the sourcing behavior of
new market entrants is particularly noteworthy. For example, the American
banks GMAC-RFC and GE Money Bank entered the German retail banking
market in 2002 and 2004. While they focused strongly on sales as their core
competence, they had to acquire knowledge about the provision of banking
products and services in the strongly regulated German banking market. For this
reason, they outsourced their back-office processes to transaction banks and
credit factories (cf. section 3.4).
In conclusion, the RDT offers a valuable complementary perspective on out-
sourcing activities by explaining a firm’s need to acquire resources from other
firms in order to survive. Nevertheless, the RDT has not received much attention
as a foundation for outsourcing research. The next section will put a stronger
emphasis on the cooperation aspect of the outsourcing phenomenon.

2.1.8 Relationship Theories

2.1.8.1 Basics and Discussion
Relationship theories (RT) focus on interactions “between parties that are geared
towards the joint accomplishment of the individual parties’ objectives” (Dibbern
et al. 2004, 20). Relationship theories address several forms of interorganiza-
tional relationship (IOR, cf. section 1.5.2) such as alliances, partnerships, and
supplier-buyer relationships. There is a form of exchange at the root of each
IOR. Participating parties find a mutual agreement so that the resulting outcomes
of the exchange are better than could be attained in any other way (Dibbern et al.
2004; Klepper 1995).
Research that is based on relationship theories asks why and under what
conditions organizations decide to establish linkages among themselves (Oliver
1990). It looks for the environmental and interorganizational factors that increase
the likelihood of establishing an IOR. Furthermore, researchers are investigating
66 Theoretical Foundation and Related Research

the life-cycle, i.e. how IORs emerge, grow, and dissolve over time (e.g. Ring and
Van de Ven 1994)
In order to conceptualize and to empirically investigate IORs, several au-
thors have suggested a set of determinants which are individually or collectively
responsible for the formation of IORs (e.g. Goles and Chin 2005; Oliver 1990).
Concepts from the various theories, including those discussed above (e.g. TCE,
RBV, and RDT), have been adopted:
o Necessity: Firms are required to form IORs to meet legal or regulatory re-
quirements (Porter and Fuller 1986).
o Asymmetry: Potential to exercise power or control over another organization
or its resources (Benson 1975; Pfeffer and Salancik 1978), not only for sta-
bilizing access to scarce resources but also to increase market power and en-
try barriers (Kogut 1988). Thus, a potentially weaker organization will
rather join an IOR when its competitive position is less likely to be eroded
(Oliver 1990). Hence, asymmetry represents both a driving force and an in-
hibitor to IORs, depending on the perspective.
o Reciprocity: Usually, relationship literature assumes that relationship forma-
tion is based on reciprocity. It is seen as the main reason and overall objec-
tive for IOR (Berg and Friedman 1981; Porter and Fuller 1986; Wöhe 1990,
199-200; Zhang and Liu 2005). “IORs occur for the purpose of pursuing
common or mutually beneficial goals or interests” (Oliver 1990, 244), e.g.
obtaining synergies in resources, technology, or information sharing. Recip-
rocity motives will be stronger if the bargaining position of the two partners
is balanced, i.e. not negatively influenced by one-sided specific investments
or the non-substitutability of a partner (Pfeffer and Salancik 1978; Porter
and Fuller 1986).
o Efficiency: This argument, adopted from economic theories, considers IORs
which are prompted by the attempt to improve efficiency (McConnell and
Nantell 1985).
o Stability: IORs serve as a strategy to forestall or absorb environmental un-
certainty (Kogut 1988; Pfeffer and Salancik 1978, 154). For example, risks
in entering new markets can be shared (Oliver 1990). Osborn and Hage-
doorn (1997) found that IORs “are more common in areas in which firms
face daunting technological challenges”.
Many authors classify types of cooperation as either efficiency-oriented
(synergies) or strategy-oriented (e.g. bundling market power) (Afuah 2003, 36;
Picot et al. 1999, 163ff). Relationship theories are a layer which is orthogonal to
the theories discussed earlier. While those mainly focus on theorizing the reasons
for reshaping the firm’s borders to be either economically reasoned (PCE, TCE,
Theoretical Foundation and Related Research 67

AT) or strategically reasoned (PPF, RBV, CCV, RDT), relationship theories

build on the firm’s need or desire for inter-firm cooperation and investigates
appropriate governance designs and mechanisms.
Getting straight to the heart of relationship theories, Parkhe states that the
“essence of voluntary inter-firm cooperation lies in ‘coordination effected
through mutual forbearance’ (Buckley and Casson 1988, 32)” (Parkhe 1993,
227). “Forbearance[20] becomes possible only when there is reciprocal behavior
(Axelrod 1984) and mutual trust (Thorelli 1986), which in turn only come about
given an absence of opportunism (Williamson 1985)” (Parkhe 1993, 227). If the
IOR fulfills these characteristics over time, trust is established within the alli-
ance, leading to a partnership (Marcolin and Ross 2005) and a substitution of
coordination costs and agency costs over time (Tsang 2000) (cf. sections 2.1.2.1
and 2.1.4.1).
The relationship literature suggests that stable, well-performing partnerships
must develop several characteristics such as aligning the partners’ objectives and
managing partnership controls and conflict (Marcolin 2002; Marcolin and Ross
2005). Thus, relationship research provides lists of factors which influence each
other and – in combination – are supposed to lead to successful IORs, such as
communication, commitment, coordination, cooperation, trust, etc. (Dwyer et al.
1987; Fontenot and Wilson 1997; Goles and Chin 2005; Macneil 1974)21. The
items are classified into action variables (e.g. communication, coordination,
conflict resolution) and context variables (e.g. trust, commitment, consensus),
which are influenced by the former, in order to build a relationship exchange
theory (Goles and Chin 2002). See, for example, the relationship management
models in (Goles and Chin 2005; Henderson 1990; Kern 1997; Mohr and Spek-
man 1994; Willcocks and Kern 1998).
From those models, the following constructs are relevant to later argumenta-
tion:
o Congruence: “Congruent sense making among parties increases the likeli-
hood of concluding formal negotiations to a cooperative IOR” (Ring and
Van de Ven 1994, 101).
o Interdependencies: these can be “analyzed in terms of common objects that
are involved in some way in both [parties’] actions” (Malone and Crowston
1990, 362). Interdependencies can either be of a prerequisite type (i.e.

20
Forbearance is defined as forgoing certain behavior that is not in the best interest of both parties
(Marcolin and Ross 2005, 33). (Footnote is not part of the citation.)
21
A comprehensive list of fundamental works in the literature of relationships and the concepts and
dimensions used can be found in (Kern 1997, table 2+3). Goles and Chin (2005) give a more ac-
tual overview about the different items used in the analysis of IORs.
68 Theoretical Foundation and Related Research

common production object: one task has to be conducted after another), of a

shared resource type (common resource object: resource required by multi-
ple activities), or of a simultaneity type (synchronization required) (Malone
and Crowston 1990).
o Coordination: defined in this context as managing the interdependencies
described above (Malone and Crowston 1994; Van de Ven et al. 1976). Syn-
chronizing business activities in organizations is an ongoing challenge in
dynamic environments (Goles and Chin 2005). Successful relationships are
characterized by coordinated actions (Mohr and Spekman 1994; Narus and
Anderson 1987; Pfeffer and Salancik 1978).
o Cooperation: describes the patterns of complementary activities which are
undertaken by related parties to achieve mutual benefits (Anderson and Na-
rus 1990; Fritsch 2004, Kern and Willcocks 2002). “Each participant has its
own objectives, but for the most part they are compatible, and the parties
cooperate to help each other achieve their respective objectives (Fontenot
and Wilson 1997)” (Goles and Chin 2005, 57). Nevertheless, it is an incom-
pletely specified exchange relationship, leaving room for opportunistic be-
havior (Gurbaxani and Whang 1991).
Apart from the conceptual and empirical research strand, which applies and
advances relationship theories by conceptualizing and testing relationship fac-
tors, there is another – strongly formal – branch of research on cooperation, fo-
cusing on the latent instability of cooperation. Axelrod’s cooperation theory
(Axelrod 1984; 1987; Axelrod and Dion 1988; Axelrod and Hamilton 1981) is
based on game theory and analyzes the common tension “between what is good
for the individual actor in the short run, and what is good for the group in the
long run” (Axelrod 2000, 131), following the question under what conditions
cooperation can emerge and be sustained among parties displaying opportunistic
behavior. Applying the prisoner’s dilemma in multi-period computer simula-
tions, Axelrod shows that only the tit-for-tat strategy leads to reliable cooperative
behavior of the involved parties and, thus, to realizing cooperation rents. Coop-
eration rents represent the return of joint usage of individual resources which are
higher than the total of individual returns from separately utilizing them (by
market coordination) (Alchian and Demsetz 1972). Their potential level depends
on achievable productivity advantages from cooperative labor division while the
real level is determined by the activation of accessible cooperation chances
(Wieland 2000).
Theoretical Foundation and Related Research 69

2.1.8.2 Implications for the Sourcing Decision

The literature which focuses on B2B relationships leads to the idea that relational
components of an IS sourcing decision can be regarded as distinct concepts sepa-
rate from other theories (Marcolin and Ross 2005). Grover and Malhotra intro-
duced the construct “relationalism” to extend their TCE-based view on IT out-
sourcing and to incorporate the “intrinsic motivations of trust and a long-term
perspective” (Grover et al. 2003, 220). Relationalism is shown to be inversely
related to the transaction costs involved in monitoring and controlling the sourc-
ing relationship. Thus, investments in relationship management with all of its
facets is evaluated as an important determinant for successful outsourcing (Goles
and Chin 2005; Grover et al. 2003; Kern 1997; Marcolin and Ross 2005; Poppo
and Zenger 2002) and to control coordination/transaction costs in the relationship
(Artz and Brush 2000; Buvik and Gronhaug 2000; Zhang and Liu 2005, 54).
As noted above, IORs are especially common in areas of major technologi-
cal challenges (Osborn and Hagedoorn 1997). From a banking perspective, these
dynamics can be supplemented by other environmental factors such as major
regulatory changes. Since in times of globalization, banks are confronted with
huge investment needs to meet the regulatory requirements, they seek for coop-
eration to cope with this task.
Marcolin and Ross (2005) introduced the concept of equifinality to IT out-
sourcing research, which means that different sourcing modes (insourcing, joint
venture, outsourcing) combined with different degrees of partnership intensity,
may lead to the same result, explaining heterogeneity of results in outsourcing
research. Therefore, although relationship theories are more often applied to
questions regarding the design of an effective sourcing relationship, they also
allow inferences to be drawn about whether to outsource a certain business func-
tion. From a relationship perspective, the outsourcing function can simply be
described as follows (Henderson 1990; Marcolin and Ross 2005):
Outsourcing = f((anticipated) outsourcing success (+))
with Outsourcing success= f(forbearance (+))
with Forbearance = f(reciprocity (+), trust(+), contract duration (+))
Outsourcing will only occur if deciders are confident of establishing a suc-
cessful relationship. Relationship theories imply a successful relationship based
on forbearance which in turn is driven by reciprocity and trust. As already dis-
cussed in the section on incomplete contract theory (TIC, section 2.1.4), formal
relationships can be substituted over time (contract duration) by a certain level of
trust, leading to increasing stability and decreasing coordination costs and
agency costs (Coleman 1990; Ring and Van de Ven 1994, Martinsons 1993).
Thus, initial investments in the relationship can be leveraged (Kavan et al. 1999).
70 Theoretical Foundation and Related Research

Empirical outsourcing research which is based on relationship theories can

be frequently found in recent literature. Starting with Klepper (1995) and Kern
(1997), who investigated the process of relationship management (Klepper 1995)
and extracted relationship theories to develop a framework for the “gestalt of an
IT outsourcing relationship” (Kern 1997) based on case studies, survey-based
works soon followed, such as (Goles and Chin 2002, 2005; Lee and Kim 1999),
validating the importance of relationship constructs for explaining outsourcing
success. Lee et al. (2004) showed the significant positive influence of contract
duration on outsourcing success.
As already mentioned in the sections on TCE and agency theory, another as-
pect, which is especially relevant to investigating partnerships in the banking
industry, is the question how the implementation of interorganizational systems
(IOS) (cf. p. 45) affects the characteristics of an interorganizational relationship.
Several empirical works investigated this question in B2B relationships and
found that the implementation of IOS not only decreases coordination costs (as
part of the transaction costs) as well as agency costs and business risks, but also
strengthens the closeness of the partnership (Grover et al. 2003; Stump and
Sriram 1997). This shows the tight interrelation between the different theories’
arguments: high perceived transaction costs and risks lead to higher use of IT
(decrease of transaction costs/agency costs), which in turn leads to closer rela-
tionships. The impact of IT is dual, being “both a control mechanism as well as a
relationship building mechanism” (Grover et al. 2003, 223).
Grover et al. give some reasons for this impact of IT on relationalism: first,
bilateral investments in IOS increase reciprocity and would, in case of missing
cooperation, lead to suboptimal benefits from the IT investments. Second, auto-
mation of interorganizational business processes enables the managers to spend
more time on cooperative activities than on operational tasks (Grover et al.
2003).
As already introduced in chapter 1, cooperative sourcing represents a form
of coopetition between firms. Hence, theoretical findings for coopetition research
can be adopted: coopetitive alliances show to be shorter-term, smaller, and imply
a more intensive evaluation and monitoring of strategic risks (higher threat of
opportunistic behavior) than non-coopetitive partnerships (Nueno and Oosterveld
1988; Hippel 1989; Polster 2001).
Coopetition is especially useful in cases where theory recommends out-
sourcing but no market exists for providing the activity (Polster 2001). Accord-
ingly, it is quite established in parts of the banking business. The classical exam-
ple is syndicate operation, where multiple banks provide joint funds (or other
financial instruments) which they cannot stem alone. Other well-known exam-
ples are the provision of cash to retail customers by forming ATM alliances (e.g.
Theoretical Foundation and Related Research 71

CashGroup of DeuBa, DreBa, PoBa, CoBa, etc), shared self service centers (in
particular between public savings banks and credit cooperatives), or the coopera-
tive sourcing alliances in the transactions business22.

2.1.9 Network Effect Theory

2.1.9.1 Basics and Discussion
While the previous theoretical perspectives have already been applied to out-
sourcing research, our work on cooperative sourcing will apply the network
effect theory (NET) as a concept which investigates the interrelationship of co-
operating agents’ benefits. Positive network effects have been defined as "the
change in the benefit, or surplus, that an agent derives from a good when the
number of other agents consuming the same kind of good changes" (Liebowitz
and Margolis 1995). The externality property implies coordination problems that
are said to be endemic in high-tech and software industries in particular (Westarp
2003). Representing economies of scale on the demand side, network effects lead
to decisions of agents, which could otherwise be autonomous, being interde-
pendent. The discrepancy between private (individual) and collective gains re-
sults in coordination problems23, possibly leading to Pareto-inferior results
(Weitzel et al. 2006). With incomplete information about other actors’ prefer-
ences, excess inertia can occur, as no actor is willing to bear the disproportionate
risk of being the first adopter of a good and then becoming stranded in a small
network if all others eventually decide in favor of an alternative (Besen and Far-
rell 1994; Katz and Shapiro 1985). This start-up problem can prevent the adop-
tion of goods, even if it is preferred by everyone.
Since the diffusion process is path-dependent, the ultimate outcome is un-
predictable (Arthur 1983; 1989). Analogously, Besen and Farrell (1994) show
that “tippiness” is a typical characteristic in networks, meaning that multiple
incompatible technologies rarely co-exist and the switch to a single, leading
technology suddenly occurs.
Some serious limitations of the explanatory power of many contributions
have become apparent, especially for the dynamic ICT markets (Liebowitz and

22
Cf. section 3.4.2 on the market for payments and securities processing. Some of the transaction
banks represent coopetitive relationships within and even between the three sectors of the banking
industry.
23
In the literature on standardization research (i.e. why and how a network of autonomous agents
decides to adopt a common technology/standard), this is called the “standardization problem”
(Weitzel et al. 2006; Wiese 1990, 1)
72 Theoretical Foundation and Related Research

Margolis 1999; Weitzel et al. 2000). The “scrupulous and implicit simplifica-
tions” (Wiese 1990, 101) of most network effect models make them essentially
useless for deriving concrete managerial implications. For example, the prevail-
ing propensity to monopolization (networks tipping into monopoly) cannot cope
with the co-existence of different IT products, despite strong network effects
(e.g. server market, EDI networks) (Westarp 2003). One important reason for
these problems is the (mostly implicit) assumption of continuously increasing
homogeneous network effects (Liebowitz and Margolis 1995); another is the
structure and properties of connections between the agents (e.g., information
flow linkages, social and historical ties, competition, group membership, etc.)
(Coleman et al. 1957; Goldenberg and Efroni 2001; Westarp 2003).

2.1.9.2 Implications for the Cooperative Sourcing Decision

Since cooperative sourcing is quite a new theoretical domain, NET has hardly
been used so far in this context. Nevertheless, this theory can offer specific com-
plementary perspectives on the particular situation of cooperative sourcing. Even
though a cooperative sourcing relationship uses economies of scale on the pro-
duction side, the concept of network effects (defined as economies of scale on
the demand side) can be interpreted and applied with regard to the producer’s
“demand” to join a certain coalition. From this perspective, multiple parties join-
ing a coalition becomes equivalent to multiple agents adopting a common tech-
nology or good; each decision influences the other partners’ or potential part-
ners’ benefits from the similar decision. Of course, as discussed above, benefits
from joining a coalition may not increase with every new member and with
every extension of the sourcing coalition. New members and a certain size of the
coalition may also lead to decreasing benefits, resulting from increasing coordi-
nation costs or cooperation risks. Based on the NET, the cooperative sourcing
decision can be described as follows:
Cooperative sourcing decision = f(cooperative sourcing decision of poten-
tial coalition members (+),
number and attributes of existing coalition members (+))
Thus, cooperative sourcing decisions underlie a particular form of external-
ities which leads to the decisions of autonomous actors being interdependent.
This in turn leads to a complex mutual feedback between the microstructure (i.e.
local individual decision behavior) and the macrostructure (overall system be-
havior) which has been considered by e.g. Hayek (1948) or Schelling (1978).
One of the rare discussions of the relationship between NET and outsourcing
is (Smith and Kumar 2004). Without explicitly referring to NET, they incorpo-
Theoretical Foundation and Related Research 73

rated the critical mass argument as one of several determinants for ASP adoption
into their conceptual work.
Since the sourcing decision of other firms is generally not known ex ante (or
can be changed in future), the decider has to make assumptions about the others’
behavior, considering that their behavior is based on assumptions about their
own behavior as well. Chapter 4.4 will present a model which applies mecha-
nisms of making individual assumptions from analytical approaches which are
based on NET.
Selecting the optimal sourcing coalition is a major problem because change
costs (transaction costs for switching to another coalition or backsourcing) for
BPO are quite large. This can lead to a lock-in into the chosen state although it
might become sub-optimal.

2.1.10 A Multi-Theoretical Perspective on Cooperative

Sourcing
In this section, the different facets provided by the theories discussed above are
put together to achieve a sound foundation for conducting cooperative sourcing
research. One of the strengths of organizational research is its plurality of theo-
ries that yields a more realistic view of organizations (Hirsch et al. 1987). Any
“theory presents a partial view of the world that, although it is valid, also ignores
a good bit of the complexity of organizations. Additional perspectives can help
to capture the greater complexity” (Eisenhardt 1989, 71).
In the following section, we will follow a multiple perspectives approach
which uses theories as different lenses on cooperative sourcing to give comple-
mentary explanations of empirical phenomena (Barney 1999; Dibbern 2004) and
to derive a formal model of cooperative sourcing decisions.

2.1.10.1 Theory Synthesis

The research question of this thesis focuses on the circumstances which lead to
the cooperative sourcing of business functions of different firms. Some of the
theories discussed above can be applied directly to this research question while
others only allow for complementary inferences about whether a firm should
cooperatively source a particular business function.
Cooperation between firms is argued by many authors to be either economi-
cally reasoned (and therefore explained by PCE, TCE, PAT, TIC, and NET) or
strategy oriented (explained by PPF, RBV, CCV, and RDT). In reality, firms
generally will tend to weigh up all arguments before deciding about a sourcing
relationship The combination of perspectives is therefore essential and applied
by many empirical works, e.g. on IT outsourcing. Nickerson et al. address this in
74 Theoretical Foundation and Related Research

the positioning-economizing perspective which argues that “for each alternative

market position alternative strategies are identified by the transaction-cost and
production-cost minimizing resource profile and by the governance mode for
each asset in the profile. The optimal strategy of a firm is the profile that offers
the greatest resulting profit. Hence, decisions regarding market position, re-
sources, and governance are made jointly” (Nickerson et al. 2001, 253).
The theories introduced above enable a company to address resources, mar-
ket position, and governance mode to determine the ”optimal” activity profile.
The firm’s strengths and weaknesses have to be determined by identifying those
resources of the firm (RBV/CCV) which represent a sustainable competitive
advantage. Because the RBV/CCV provides only an internal view, the strengths
have to be compared with the market in order to determine their strategic impor-
tance (PPF).
If the firm has chosen a cost leadership strategy, the value of its own re-
sources must also be reflected by an efficient production cost structure (PCE:
economies of scale, scope, and skill). The non-production costs associated with
different governance structures can be approximated using TCE, complemented
by TIC and AT.
For cooperative sourcing, the future cost and benefit development of partici-
pating in a coalition has to be estimated. The externalities induced by existing
and future partners’ behavior are captured by NET. Comparing the sum of pro-
duction costs and transaction/agency costs for different alternatives leads to an
optimal sourcing configuration from an economic perspective.
Table 2 summarizes the synthesis of theories and captures the relationship of
the different theories regarding cooperative sourcing. The first column classifies
whether the theory has an efficiency focus or a strategy focus while the second
states whether it has a firm-internal focus (focusing on the firm’s own production
costs and resources) or whether it instead focuses on the market or on inter-firm
relationships. In the third column, theories are classified according to the primary
outsourcing research question which they help to answer. This classification has
been adopted from (Dibbern et al. 2004) (cf. Figure 7 on p. 81): Why describes
the research aim that analyzes reasons for and against outsourcing. Results from
these theories and corresponding research are the main foundation for our work.
What is related to why and asks which part of a firm’s set of activities should be
outsourced. Third, the how question analyzes how a sourcing relationship can be
designed in order to be successful (e.g. type of governance mode, contractual
design, etc.). As argued above, although this research question is not the objec-
tive of our research, it nevertheless allows backward inferences to the why ques-
tion because those theories also give implications about the likelihood that an
effective partnership can be established in a certain context. If this likelihood is
Theoretical Foundation and Related Research 75

too small, the outsourcing decision itself has to be reconsidered. The last column
summarizes the decision determinants discussed in the previous sections.
Efficiency or Internal or
Research
strategy- external Decision determinant
question
oriented focus

Production cost economics Production cost function

Efficiency Internal Why (economies of scale, scope, and
(PCE, section 2.1.1) skill)
Transaction cost econom-
Efficiency External Why Transaction costs
ics (TCE, section 2.1.2)
Network effect theory
Efficiency External Why Network effects
(NET, section 2.1.9)
Agency theory
Efficiency External How Agency costs
(AT, section 2.1.3)
Ability to manage relationship
Theory of incomplete efficiently & effectively, either
contracts by sufficient contract com-
Efficiency External How
pleteness or by successful
(TIC, section 2.1.4) implementation of relational
contracts

Relationship theories Ability to realize congruent

n/a External How behavior (cf. relational con-
(RT, section 2.1.8) tracts in TIC)
Resource-based view /
core competence view Strategy Internal What Resource properties
(RBV/CCV, section 2.1.6)
Resource dependence
theory Strategy External What Resource gaps
(RDT, section 2.1.7)

Porter’s positioning Impact of outsourcing on

framework competitive position, depend-
Strategy External What
ing on the chosen business
(PPF, section 2.1.5) strategy
Table 2: Theory synthesis (based on (Beimborn 2005; De Looff 1995;
Jayatilaka et al. 2003; Lammers 2005; Poppo and Zenger 1998))
Two theories primarily have a firm-internal focus – PCE and RBV/CCV.
The first evaluates a firm’s own production capabilities from an economic per-
spective, while RBV represents the complementary strategic view. Both theories
have their corresponding partners coming with the “external view”: TCE (com-
plemented by AT and TIC) covers the remaining costs not to be incorporated by
76 Theoretical Foundation and Related Research

PCE, i.e. the costs related to coordination and exchange, while RDT takes into
account the outside-in view of the strategic perspective by not investigating the
resources of the firm but the resource gaps which have to be filled (from outside)
to stabilize the competitive position. Similar derivations come from Porter’s
framework. The NET incorporates the decision behavior of potential and existing
partners into the decision function.
The TIC and the relationship theories basically examine the design of an ef-
ficient and effective partnership mode. If the likelihood of failure is expected to
be too high (e.g. if the degree of competition between the potential partners is too
high and cannot be managed adequately), this will also be crucial for the initial
outsourcing decision.
In conclusion, the following nomological model captures the relationships
derived in the previous theory sections. For PPF and RT, a simplified representa-
tion has been chosen to reduce visual complexity.

Internal focus External focus

information asymmetry, uncertainty, measurability, specificity,

length, uncertainty, complexity, uncertainty,
measurability standardization degree frequency
Efficiency focus

contract completeness
agency costs transaction costs
relational governance
AT TIC TCE
PCE NET

economies of behavior and properties

scale, scope, skill of (potential) partners
(Cooperative)
Outsourcing
Decision
resource properties Forbearance*
Strategy focus

(value, rareness,
imperfect imitability,
substitutability, RT
interconnectedness)
R/CBV resource gaps
(importance, discretion, strategy type** reciprocity, trust,
alternatives, strategy) contract duration
RDT PPF
*= Simplified representation: **= Strategy type moderates the impact
concentration,
forbearance impacts anticipated of the other factors (cf. section 3.5.2).
munificence,
outsourcing success, which in turn This representation has been
interconnectedness
affects the outsourcing decision. chosen for reasons of simplification.

Figure 6: Nomological model of theory synthesis for (cooperative)

outsourcing (partially based on (Cheon et al. 1995))
All of these theories provide factors which encourage or inhibit sourcing de-
cisions. In the second part of this chapter, the reasons for and against outsourcing
Theoretical Foundation and Related Research 77

will be discussed from an empirical perspective. The end of section 2.2.2 pro-
vides the mapping of these items to the corresponding theoretical arguments
(Table 3 and Table 5 on page 85 and 92, respectively).
Finally, the question remains whether a combination of these theories can be
applied to the units of analysis, which are the business functions as sourcing
objects and cooperative sourcing (of these business functions) as the correspond-
ing action alternative.

2.1.10.2 Business Functions as Unit of Analysis

The objects of cooperative sourcing in this research are business functions or
activities which are encapsulable process steps that result in a measurable output
(Lammers 2004).
From the view of production cost economics, a business function is de-
scribed by a production function and a corresponding production cost function.
While the first describes the transformation of any input quantity into output, the
latter determines the resulting production costs. Thus, PCE can be applied to
determine the production costs that are expected for a specific output generated
by an activity. Production cost functions (resp. process cost functions) in this
work are defined for each business function, i.e. economies of scope are not
taken into account. This problem will be solved by integrating interface costs as
a separate cost factor.
Transaction cost economics focuses on transactions rather than on internal
business functions, but when applying TCE to the outsourcing phenomenon,
these prove to be two sides of the same coin. “Therefore, it may be concluded
that the term ‘transaction’ would need to be substituted by ‘tasks’ or ‘business
functions’ in Williamson’s definition. No transactions are organized in alterna-
tive governance modes, but certain tasks or business functions” (Dibbern 2004,
47). Although Williamson argues that a transaction is a technologically separable
interface where one stage of activity terminates and another begins (Williamson
1985, 1), he also discusses bundling transactions and unbundling firms into
transactions (Williamson 1985, section 4). In his treatment of a firm’s efficient
boundaries, he applies his framework to a process of activities (Williamson
1985, 96-98) and “conceives of the vertical chain as a set of technologically
separable stages of production – activities – in which each stage may be involved
in multiple transactions with other stages” (Nickerson et al. 2001, 253).
Agency theory and the theory of incomplete contracts do not explicitly focus
on business functions but they of course assume a certain task as the object to be
fulfilled by one of the contract partners (agent, insourcer). Therefore, they can be
applied to BPO.
78 Theoretical Foundation and Related Research

Network effect theory does not focus on business functions but on network
effect goods. Nevertheless, the participation in a cooperative sourcing alliance
deals with a very similar type of externality if the service provision of the coop-
erative sourcing instance is seen as the “good” in demand. For our requirements,
this view provides a valuable complementary perspective. Although NET has not
yet been adopted to the domain of outsourcing research, similar steps are done
both in organizational acceptance models (which substitute a technology to be
adopted by an outsourcing decision) (Gewald and Lammers 2005) and in the
theory of innovation diffusion, which originally also focused on diffusion of
technology or other goods and has already very early been adopted to outsourc-
ing research (outsourcing as an organizational innovation) (Loh and Venkatra-
man 1991; cf. also Hu et al. 1997).
Porter in his positioning framework focuses directly on business functions as
we understand them. Based on the value chain concept, Porter presumes that
activities are more or less encapsulable and can be unbundled to reconfigure the
system of (existing and potential) activities in order to strengthen a firm’s com-
petitive advantage (Porter 1996).
The resource-based view, core competence view, and resource dependency
theory all have employed a concept of resources, capabilities, or competencies
although they are often criticized for not defining and demarcating these terms
precisely. Business functions, in turn, deploy resources (or assets, capabilities,
competencies), otherwise the latter would offer no value to the firm (Ray et al.
2004). Thus, the units of analysis of these theories are resources but they have to
be deployed in activities in order to result in a competitive advantage (Lammers
2005). For simplification, we will conceptually encapsulate resources in the
corresponding business functions, as is also done in (Lammers 2004; Beimborn
et al. 2005b). This ignores the fact that resources might be employed by different
business activities (leading to economies of scope), which is a limitation of our
approach that will be discussed later on.
Finally, relationship theories focus on the development of the relationship
itself, rather masking the object of the relationship. As argued above, in the
work on hand relationship theories will provide only complementary comments
on whether to outsource a business function or not.
In summary, it can be argued that the applied theories sufficiently employ
consistent and complementary views on business functions as unit of analysis
and are well suited to form the foundation for this research work.

2.1.10.3 Cooperative Sourcing as Unit of Analysis

Cooperative sourcing, as the decision alternative examined in this work, can also
be appropriately dealt with by the different theoretical perspectives. While most
Theoretical Foundation and Related Research 79

of the chosen theoretical lenses have a primary focus on the relationship between
several business entities, two of them (PCE, RBV/CCV) follow more of a firm-
internal perspective. Nevertheless, PCE in its classical form already discusses the
bundling of the production of similar output (economies of scale). Cooperative
sourcing of business functions from different firms is a natural extension of this
concept.
The RBV/CCV as well as RDT give suggestions about outsourcing particular
business functions but, in their basic form, do not explicitly consider cooperative
sourcing as a separate governance mode. However, in addition, Foss and Eriksen
(1995) extended the capability concept to “industry capabilities” which represent
non-proprietary capabilities that are shared among a group of firms. Although
this concept does not actually require the explicit coordination of firms, it can be
mapped onto cooperative sourcing if opportunities which result from the bundled
processing volumes are exploited to generate superior capabilities (e.g. higher
degree of automation).
TCE has often been applied to outsourcing research and it will be similarly
helpful with cooperative sourcing, which has no propositions that are structurally
different. Similar conclusions can be derived for AT and TIC. We can expect less
threat of opportunistic behavior from an agency-theoretical perspective, since
cooperative sourcing usually implies incentive-congruent behavior. Moreover,
cooperative sourcing of business functions will lead to a homogenization of
those processes and, consequently, to less inherent contract incompleteness (cf.
section 2.1.4.2). However, business risks might be evaluated differently if the
cooperative sourcing relationship is between competitors.
NET, if applied to the sourcing of business functions, directly addresses the
issue of cooperative sourcing because it subsumes the use of the same “good” by
different entities.
Strategic management theories such as PPF do not explicitly address out-
sourcing questions. Outsourcing as well as the special form of cooperative sourc-
ing remain one of various options to strengthen or to stabilize the competitive
position of the firm.
Finally, RT directly address the topic of forming an interorganizational rela-
tionship, be it an alliance, a joint venture, or even a coopetive partnership. They
help determine and design the optimal governance mode.
Cooperative sourcing can be described as horizontal and vertical coopera-
tion. Since similar processes are merged, the relationship is basically horizontal.
However, because one bank now provides services which are used within an-
other bank’s value chain, the relationship can also be described as vertical. Since
we did not adopt a process perspective in this work (i.e. we did not explain how
the business functions are structurally interrelated) and therefore do not have to
80 Theoretical Foundation and Related Research

determine whether two business functions are sequentially organized within a

particular business process, this dual property remains unproblematic and uncon-
sidered. Interconntectedness of the several business functions will be captured,
but not the process layer itself. This activity-based focus can be imagined e.g. as
a service-oriented architecture, where one business function requires services of
another, without hard-coding the process flow ex ante.
Furthermore, cooperative sourcing coalitions can be classified into non-
competitive (or “pre-competitive”24) and competitive (Nueno and Oosterveld
1988). The coopetition character of coalitions has been discussed within the
section on relationship theories and will be considered by the model develop-
ment in chapter 4.

Cooperation as well as outsourcing has been empirically investigated many

times. Most of the works adopted a multi-theoretical perspective because a single
theory cannot sufficiently explain interorganizational dynamics and relations.
This section aimed at providing a sound theoretical base for a combination of
both – cooperation and outsourcing – by integrating the theoretical views com-
monly used by those research streams and adding another one, namely the net-
work effect theory. The model developed in chapter 4 tries to incorporate the
essential propositions to provide a proper analysis tool for cooperative sourcing.
The next section provides a view complementary to this theoretical section
and discusses related research on outsourcing determinants from the application
domain perspective while referring to the relevant theoretical arguments.

2.2 Outsourcing Research

This section gives a brief overview on outsourcing research. After a very brief
classification of the different research strands (section 2.2.1), the literature analy-
sis focuses on drivers and inhibitors of outsourcing in general and of BPO, in
particular (section 2.2.2). Further, since the main approach of this work is to
develop and to apply a formal model on cooperative sourcing behavior, section
2.2.3 summarizes the research works which also adopt analytical or simulative
approaches for investigating the outsourcing phenomenon.

2.2.1 Overview
There has been research on outsourcing in the scientific literature for more than
15 years. Since the early 1990s, IT outsourcing has become a major research
24
In unregulated markets, there is always a potential threat of competition between cooperating
firms. Knowledge transfer can always lead to the opportunity of entering the partner’s market.
Theoretical Foundation and Related Research 81

topic esp. in the IS discipline. Although the work in hand focuses on BPO which
shows fundamental differences to IT outsourcing (Dibbern et al. 2004, 9; Will-
cocks et al. 1996), many results can be transferred and represent the closest area
of related research. In fact, recent works on BPO stem from the IS discipline,
adopting and adapting the concepts and models developed there (e.g. Beimborn
et al. 2005a; Currie et al. 2003; Franke and Gewald 2005; Ganesh and Moitra
2004; Mani et al. 2006; Rouse and Corbitt 2004; Willcocks et al. 2004).
In order to get an overview of the outsourcing research of the last decade,
the interested reader is referred to the literature survey of Dibbern et al. (2004).
Based on Simon’s (1960) Decision Making Model, the authors developed a sche-
me which allows the classification of outsourcing research relating to the re-
search question and the stage of the outsourcing process, respectively (Figure 7).
Simon 1960 Application of
Decision Outsourcing Outsourcing
Making Model Stages Stages

-Determinants
Phase 1: Decision Process

Intelligence Why -Advantages/disadvantages

Outsourcing alternatives:
-Degree of ownership
Design What
-Degree of outsourcing

Guidelines, procedures and

stakeholders of decision
Choice Which
initiation, evaluation, and
making

-Vendor selection
-Relationship building
Implementation
Phase 2:

Implementation How
-Relationship management

Outcome -Experiences/learning
-Types of success
-Determinants of success

Figure 7: Stage model of outsourcing (Dibbern et al. 2004, 15)

In the first phase – the decision phase – outsourcing research can be distin-
guished into research foci on why to outsource, what to outsource, and which
choice to make (Dibbern et al. 2004, 15-17). Why to outsource asks for the rea-
sons for and against outsourcing and represents the earliest focus of ITO research
(e.g. Loh and Venkatraman 1991, 1992a+b). More specifically, the second ques-
tion asks for what part of the business should be outsourced. It focuses on the
range of outsourcing (selective vs. total outsourcing from an ITO perspective or
the optimal degree of vertical integration from a BPO perspective). Little of the
82 Theoretical Foundation and Related Research

literature is dedicated to analyzing the process of outsourcing decision-making in

the firm i.e. which choice to make.
After having made the outsourcing decision, the implementation phase fol-
lows opening research opportunities in the fields of how the outsourcing deal is
implemented and what the outcomes are. How focuses on the topics of vendor
selection, contract design, and relationship management. Finally, outcome asks
for the results of outsourcing and thus closes the loop. Have goals such as cost
cutting or performance increase been met by the vendor? What undesirable (and
unforeseen) impacts did additionally occur?
With this classification in mind, the work in hand aims at tackling the ques-
tions of why and what to outsource as well as the outcomes of (cooperative) BPO
of primary business processes in the banking industry. In fact, the questions
regarding why to outsource and outcomes of outsourcing turn out to be two sides
of the same coin (Aubert et al. 1998; Dibbern et al. 2004, 79; King and Malhotra
2000). Many of the ex ante drivers and inhibitors of an outsourcing decision will
be (potential) ex post outcomes of the outsourcing decision (Dibbern et al. 2004,
79). Thus, research models relevant to both questions often use similar con-
structs, first addressing “outsourcing potential” and then “outsourcing success”
(e.g. Aubert et al. 1998; King and Malhotra 2000). The models on outsourcing
success often incorporate additional variables from the remaining implementa-
tion research topic (how), such as contract completeness, partnership quality, etc.
(e.g., Grover and Teng 1996; Lee and Kim 199; Saunders et al. 1997).
Following this integrated perspective, as adopted by earlier papers25, the next
section will discuss BPO drivers and inhibitors as possible outcomes and, thus,
not distinguish between an ex ante (why) and ex post (outcome) perspective.

2.2.2 Outsourcing Drivers and Inhibitors

In the first section of this chapter, the question of why (or why not) to outsource
was answered from the perspective of different theories. Based on this analysis,
the following sub-sections will recapitulate the different outsourcing drivers and
inhibitors within a classification scheme to provide a sound overview about the
why strand of outsourcing research.

2.2.2.1 Outsourcing Drivers

Many research works have discussed the relevance of different reasons and in-
hibitors for outsourcing, often focusing on IT outsourcing. However, most of

25
E.g., Hirschheim and Lacity (2000), Loh (1994), and Reponen (1993) combined both research
questions for why to outsource and outcome of outsourcing.
Theoretical Foundation and Related Research 83

these factors are not unique to IT outsourcing and thus “may be realized in out-
sourcing other forms of resources” as well (Teng et al. 1995, 78).
Impacts of outsourcing – positive as well as negative – can be classified in
terms of two dimensions. One is time while the second refers to the basic objec-
tive (economic/efficiency vs. strategic reasons).
temporal perspective
short-term mid-term long-term
fixed cost
variabilization

cost reduction from (dynamic) skill

efficiency/ scale and (ex ante) economies
economic skill economies (learning effects)

business performance
objective

improvement

quality improvement

core competence focus

Access to superior
strategic skills and technology
strategic flexibility
technical and
operational flexibility

Figure 8: Classification of outsourcing reasons

Along the first dimension, impacts of outsourcing can be classified into
short term, mid term and long term (King and Malhotra 2000). Primary short-
term positive impacts of outsourcing are cost savings, task efficiency gains, and
service level enhancements. Regarding the mid-term focus, firms follow the
goals of overall firm performance enhancement, risk mitigation, and the utilizing
of access to specialized external resources (Loh and Venkatraman 1992a),
whereas the main long-term impact is to focus the business on its core compe-
tencies and to increase strategic flexibility (Prahalad and Hamel 1990).
The most cited reason for outsourcing is reduction of costs, esp. operational
costs (Barthélemy and Geyer 2000; Friend et al. 2002; Kakabadse and Kaka-
badse 2002; Lacity and Hirschheim 1993b; McFarlan and Nolan 1995). Opera-
tional costs can be reduced by provider-side economies of skill (the sourcing
provider has superior process knowledge and can achieve improved learning
effects (Loh and Venkatraman 1992a)) and economies of scale (merging process
outputs of multiple outsourcers (Loh and Venkatraman 1992a)).
A frequent argument in favor of outsourcing related to economies of scale is
reduction or “variabilization” of fixed costs (Alexander and Young 1996; Lacity
and Hirschheim 1993a; Lacity et al. 1996), including amortization of real capital
84 Theoretical Foundation and Related Research

(Accenture 2002; Barthélemy and Geyer 2000). Outsourcing a business function

and adopting a transaction-based pricing model leads to costs that dynamically
follow process volume oscillations and, thus, decrease in times of less activity.
This strategy often enables the outsourcer to reach a lower break-even point
(Bettis et al. 1992; Gilley and Rasheed 2000). Nevertheless, fixed cost variabili-
zation from the insourcer’s point of view is only possible if process volumes of
multiple firms can be aggregated and economies of scale can be achieved.
Another efficiency argument of outsourcing is the intended improvement of
internal control routines. Having outsourced particular processes, performance
can be measured and controlled more easily (Alexander and Young 1996; Kaka-
badse and Kakabadse 2002; McFarlan and Nolan 1995). Sometimes, firm leaders
just use the intimidation of outsourcing or the benchmarking with outsourcing
providers as an effective instrument to regain control over internal departments
(Lacity et al. 1996) and to encourage them to work more efficiently.
D’Aveni/Ravenscraft (1994), Dess et al. (1995), and Gilley/Rasheed (2000)
showed that even not outsourcing IT functions after evaluating this option, led to
IT cost reductions of up to 45%.
The main strategic objective of outsourcing is to improve the organizational
performance by concentrating the firm’s own management staff on core compe-
tencies, allocating scarce management resources more effectively (Bloch and
Spang 2003, 12), thus raising their strategic flexibility potential (the firm is more
agile to react to business threats and opportunities (Klein et al. 1978; Williamson
1975)). Various academic works reveal a significant positive relation between
focusing on core competencies and a firm’s performance (Dess et al. 1990; Ko-
tabe and Murray 1990; Quinn 1992), due to reduced effort on commoditized
processes and improving “managerial attention and resource allocation to those
tasks that the firm does best” (Gilley and Rasheed 2000, 766).
Flexibility crops up in another area, too: outsourcing often leads to the re-
designing of processes and to the defining of both organizational and technical
interfaces between in- and outsourced activities. Those well-defined and stan-
dardized interfaces enable faster changes of technology as well as of sourcing
providers (Harrigan 1985; Quinn 1992), leading to reduced threat of lock-in (i.e.
irreversibility of the outsourcing decision and provider choice) and, furthermore,
to increased competition on the provider market. Increased competition on the
provider market enforces innovation and the delivery of high quality services at
low costs. Technology changes per se, of course, are eased by outsourcing, too,
because if a technology is not used in-house, changes provide no internal costs,
problems, and operational risks. Thus, increasing flexibility often comes along
with quality improvements (Dess et al. 1995; McFarlan and Nolan 1995; Quinn
and Hilmer 1994) and operational risk reduction (Accenture 2002; Quinn 1992).
Theoretical Foundation and Related Research 85

Table 3 summarizes the reasons for outsourcing from the scientific litera-
ture’s perspective, including the theoretical base from section 2.1 and (positivist)
empirical support26 from outsourcing research.

Construct Outsourcing driver Theoretical Other

Empirical support26
Focus (why to outsource) foundation references27
(Ang and Cummings 1997; Apte et
Economies of al. 1997; Clark et al. 1995; (Palvia 1995;
Cost reduction scale & skill Lancellotti et al. 2003; Loh and Teng et al.
(PCE) Venkatraman 1992a; McLellan et 1995)
al. 1995; Smith et al. 1998)

Capital reduction / fixed Economies of

(Apte et al. 1997; Loh 1994) (Huber 1993)
Economic/ cost variabilization scale (PCE)
Efficiency Economies of
Business performance skill, improve- (Loh and Venkatraman 1992a;
improvement ment of control 1995; Teng et al. 1995)
(PCE, PAT)

Increased service quality, Resource gaps (Apte et al. 1997; Clark et al. 1995; (Cheon et al.
Grover et al. 1994; Lancellotti et al. 1995; Cross
reducing processing time (RDT) 2003; Teng et al. 1995) 1995)

Resource gaps
Access to superior skills (RDT), Envi- (Apte et al. 1997; Loh 1994;
(Cross 1995;
and resources, reduction ronmental McLellan et al. 1995; Nelson et al.
Huber 1993)
of technological risks (techn.) uncer- 1996)
tainty
Strategic
Increased technological Resource gaps (Apte et al. 1997; McLellan et al.
and process flexibility (RDT) 1995; Slaughter and Ang 1996)
Core competence focus / Resource
(Smith et al.
increased business properties (Slaughter and Ang 1996)
1998)
flexibility (RBV/CCV)
Table 3: Desired outcomes of outsourcing and theoretical foundation
Other drivers for outsourcing, not necessarily being perceived as advantages,
are the institutional pressure from major players or federal regulators (Ang and
Cummings 1997) and the outsourcing behavior of other market players (“band-
wagon effect”, Lacity and Hirschheim 1993b; Loh and Venkatraman 1992b;
Rouse and Corbitt 2004). The latter describes outsourcing decisions influenced
by competitors’ decisions (imitative behavior) and was empirically shown in by
Loh and Venkatraman (1992b). The theoretical foundation is provided by the

26
All of the listed works have successfully tested for a positive relationship between the factor or
desired outcome and outsourcing potential or degree of outsourcing.
27
Conceptual works (e.g. classifications, decision support models, construct development) and case
studies
86 Theoretical Foundation and Related Research

innovation diffusion theory (Parker 1994; Teece 1980a), while institutional pres-
sure is covered by the institutional theory (DiMaggio 1988; Jepperson 1991;
Powell 1991).
Finally, it should be mentioned that the different drivers are interrelated and
often stand in causal relationships. As one example, access to superior skills is
often used as an argument for predicting cost advantages from outsourcing
(economies of skill).
Two different approaches can be used in order to determine the relative im-
portance of the different outsourcing arguments: The first is directly asking out-
sourcing decision makers, which is the method used in many studies by experts
and consulting firms. Although the relative importance of outsourcing reasons is
always based on the type of outsourcing object (e.g. IT vs. secondary vs. primary
business process) and on the type of firm (type of industry, large vs. small firm),
the primary reason for outsourcing is usually cost saving (Ang and Straub 1998;
Dibbern et al. 2003; Lacity and Willcocks 1998). Although numerous researchers
contrarily argue that cost savings are not the main benefit of outsourcing (Beau-
mont and Costa 2002; Saunders et al. 1997, Gewald and Dibbern 2005), they are
still the critical factor when it comes ex ante to evaluating an outsourcing deci-
sion and ex post to measuring and determining outsourcing success (Ang and
Straub 1998; Lacity and Willcocks 1998; Lee 2001; Lee and Kim 1999). Decid-
ers usually have to calculate a positive business case before suggesting outsourc-
ing a certain activity. Consequently, a study on outsourcing in the European
banking industry (Pujals 2005) found cost savings to be the dominating outsourc-
ing reason (cf. Table 4). After costs follows focusing on core competencies and
access to superior resources. Table 4 shows the results of three exemplary de-
scriptive studies on IT outsourcing (Landis et al. 2005; Schott 1997) or outsourc-
ing in general in the banking industry (Pujals 2005).
A second approach to determining the importance of outsourcing arguments
measures the effect of outsourcing decisions (and their reasons) on the value of
the outsourcing firms. It is argued that, although this is more objective, this ap-
proach is weaker because stock price volatility is subject to many other influ-
ences and activities of the overall firm that may neutralize the effect of a singular
outsourcing decision. Peak et al. (2002) showed that capital markets react posi-
tively to outsourcing initiatives that are aimed at getting access to superior skills
and resources, while they are indifferent to those whose motivation was improv-
ing service quality or focusing on core competencies. Oh and Gallivan (2004)
showed that there were no market reactions when the published rationale for
outsourcing was cost advantage.
Theoretical Foundation and Related Research 87

Outsourcing reason (Schott 1997) (Landis et al. 2005) (Pujals 2005)

Cost savings (overall) #3 (87%) #1 (70%) #1 (89%)
- Scale economies in particular #5 (70%) #4 (29%)
Access to superior skills and inno-
#1 (92%) #5 (22%) #2 (60%)
vative technology
Core competence focus #3 (35%) #3 (58%)
Performance improvement due to
#4 (79%) #2 (57%) #5 (29%)
market-based relationship
Improvement of control #2 (91%)
Increase of flexibility #3 (35%) #7 (16%)
Mitigation of technological risk #5 (22%)
123 German firms 25 very large global
from different corporations, differ- 82 European banks,
Demographics
industries and all ent industries survey of the ECB
sizes (PhD thesis) (consulting study)
Table 4: Reasons for outsourcing: results from exemplary studies (multiple
answers possible). Values give ranks and frequency of positive
answers.
Business Process Outsourcing
The particular topic of business process outsourcing, esp. in the banking indus-
try, still has not stimulated much attention in academic literature. Typically,
papers on BPO cite similar advantages and disadvantages from outsourcing as
the ITO literature (cf. (Pujals 2005) in Table 4 and section 2.2.2.3). In the bank-
ing industry this can be even more justified because outsourcing of business
processes in the banking industry usually includes outsourcing of IT because it
represents the “production infrastructure”.
The existence of the advantages discussed always depends on the particular
activity to be outsourced and its position in the value chain (IBM 2003). Banks
“who outsource their middle office activities such as treasury, risk management
or products do so mainly to better focus on their core competencies and to de-
liver value-added services to their clients. [Study] Participants outsource the
back-office and support activities mostly to achieve cost savings” (IBM 2003,
46). Further, there is a structural difference between outsourcing primary vs.
secondary processes. For the first (e.g. payment transactions or credit processing
in banks), getting access to superior resources and skills shows not to be an out-
sourcing reason because banks consider themselves competent in their primary
processes (Gewald and Dibbern 2005). Consequently, it can be argued that –
88 Theoretical Foundation and Related Research

from research on outsourcing reasons – secondary process outsourcing is more

related to IT outsourcing than outsourcing core processes.

2.2.2.2 Outsourcing Inhibitors

Negative impacts of outsourcing can be classified in the same manner as the
outsourcing reasons above (Figure 9). In the short term and mid term, the main
outsourcing inhibitors are start-up and periodically reoccurring transaction costs
(TC). As discussed in section 2.1.2, TC includes all costs arising from a transac-
tion, i.e. costs for evaluation, negotiation, contracting, transition, coordination,
and – possibly – roll back (Williamson 1979). The largest factor – often cited by
practitioners, in Germany, in particular, – is costs for staff transfer and reduction
(cf. section 3.5.1.3).
temporal perspective
short-term mid-term long-term

periodical transaction
transaction
costs for coordination,
costs for backsourcing or
monitoring, re-negotiation
efficiency/ start-up transaction costs changing provider
(incl. risk of hidden costs)
for evaluation,
economic negotiation, migration under-performance
(incl. risk of hidden costs)
(loss of quality)
objective

scope diseconomies

loss of own skills

loss of operational
& business uncertainty
flexibility
strategic
loss of strategic
Confidentiality problems flexibility, lock-in
dependent on provider

Figure 9: Classification of outsourcing inhibitors

Large parts of transaction costs could be determined ex ante, but often are
not. Transaction costs “are traditionally excluded by outsourcing consultants on
the basis that good performers should always be benchmarking internal perform-
ance against competition, and that the costs of going to the market are once off”
(Rouse and Corbitt 2004, 6). This obviously ignores the fact that parts of the
transaction costs occur periodically (mid term). Hidden transaction costs often
include unexpected migration and management costs, change requests, and costs
for lawsuits caused by insufficiently defined SLAs and performance measures
(Aubert et al. 1999, 2000, 2002). In severe cases, Lacity and Hirschheim talk
Theoretical Foundation and Related Research 89

about “cost escalation” (Lacity and Hirschheim 1993, 1995). The outsourcer
loses control over the sourcing relationship with related costs and services.
Finally, in the long run, costs for backsourcing or changing the provider
might occur, which are higher the longer the outsourcing relationship spans
(Peng and Wenhua 2004).
Besides periodical transaction costs, mid-term impacts include a loss of
economies of scope (cf. section 2.1.1) and of alignment between internal and
outsourced organizational units, resulting in interface costs, friction loss, and
coordination problems (Bettis et al. 1992; Kotabe and Murray 1990; Picot et al.
2001). In fact, these can also be integrated into the transaction cost construct.
Generally, it can be shown that outsourcing due to diseconomies of scope often
leads to an overall loss of a firm’s performance (Accenture 2002; Bahli and Ri-
vard 2003).
While it can be argued that outsourcing has a positive impact on flexibility
(cf. section 2.2.2.1), the opposite can be found as well (Beimborn et al. 2006a).
Standardization of services by the provider inevitably leads to more efficiency at
the cost of flexibility (Levina and Ross 2003). Not all the requirements and
change requests of a single mandator can be considered. Modular and flexible IT
systems may mitigate this because they allow more individual customization.
Further, in the mid term a loss of quality caused by agency problems be-
tween the insourcer and the outsourcer can occur (Gurbaxani and Whang 1991).
Since SLAs cannot be sufficiently defined (incomplete contracts, cf. section
2.1.4) and controlled for every particular task, the relationship is vulnerable to
moral hazard28 (cf. section 2.1.3.2 and Lacity et al. 1996). The agency problem
gets worse when the outsourcing firm has a weak management, which actually is
often a reason for outsourcing. If the firm’s own management is incompetent, the
task of designing SLAs and control measures, which should be done by know-
how carriers in the relevant domain, will be carried out even more inadequately
(Accenture 2002; Earl 1996; Petzel 2003). The opportunistic behavior of the
provider can furthermore lead to confidentiality problems. In certain circum-
stances, the provider might be tempted to use accessible, but sensitive data (in IT
outsourcing as well as in BPO) for their own purposes (Earl 1996; König and
Beimborn 2004).
Loss of quality can also be caused by bad alignment. If there are marked cul-
tural differences between the outsourcer and the insourcer, there can be a lack in
shared knowledge regarding the mandator’s business and related problems (Lac-
ity et al. 1996). Thus, problem-solving processes and improvement tasks are

28
Of course, moral hazard is not only a problem of interorganizational relationships.
90 Theoretical Foundation and Related Research

difficult to handle in this relationship. Besides a loss of quality, this can result in
increased operational risks that cannot be dealt with as effectively as in-house.
In the long term, outsourcing-related risks turn out to be the loss of in-house
competencies (Bahli and Rivard 2003; Bettis et al. 1992; Khosrowpour et al.
1995; Martinsons 1993; Quinn and Hilmer 1994) and, as a consequence, depend-
ency on the service provider’s capabilities (Alexander and Young 1996; Aubert
and Patry 1998; Lacity et al. 1996). This problem of lock-in, caused by high
backsourcing costs and lost competencies, becomes even more damaging if the
provider turns out to be less capable (e.g. inexperienced staff, outdated technol-
ogy skills (Earl 1996)) of providing services than assumed ex ante (problem of
adverse selection) (Aubert and Patry 1998; Dess et al. 1995).
By outsourcing its competencies, a firm trims its own strategic flexibility.
Even a commoditized activity can become strategically important in the future
and needed to exist within the firm. If outsourcing is not reversible, substantial
problems can occur (Lacity et al. 1996). Therefore, an outsourcing strategy
should always be built on previous business process modularization and optimi-
zation activities to alleviate future backsourcing and to sustain strategic flexibil-
ity.
Beside economic and strategic outsourcing disadvantages, there are further
outsourcing inhibitors such as legal and regulatory constraints (cf. section 3.5) as
well as social problems like low organizational performance from demoralized
staff (Orton and Weick 1990; Rouse and Corbitt 2004).
Table 5, which is based on (Aubert et al. 2005), summarizes this section on
outsourcing inhibitors by mapping the different factors to the theoretical back-
ground in section 2.1. Again, the reader should be aware that the different inhibi-
tors are strongly interrelated and often affect one another. As in Table 3, the right
hand columns give references to conceptual and empirical work which validate
the negative impact of the discussed arguments on outsourcing. Surprisingly,
(positivist) empirical analyses of outsourcing inhibitors are much rarer, com-
pared to those of the outsourcing reasons referred to in Table 3.
Theoretical Foundation and Related Research 91

Construct Outsourcing Empirical

Theoretical foundation Other references29
focus inhibitor support
Expected (Ang and
Need to search for partner, (Cheon et al. 1995a;
“start-up” Straub 1998;
persuading, negotiating, migrat- Hancox and Langlois 1992; William-
transaction son 1985)
ing (TCE) Hackney 1999)
costs
Lack of expertise of the client
with the activity (RBV, PCE),
(Hirschheim (Earl 1996; Lacity et al.
Hidden migra- lack of expertise of the client
and Lacity 1995; Lancellotti et al.
tion costs with outsourcing (TCE), inter- 2000) 2003; Sappington 1991)
dependencies, diseconomies of
scope (PCE)
Contractual
amendments Uncertainty, e.g. technical (Alchian and Demsetz
(expected and discontinuity (TCE), complexity 1972; Barzel 1982; Earl
hidden trans- (TIC) 1996; Lacity et al. 1995)
action costs)
Measurement problems (PAT),
Unexpected
lack of expertise of the client or (Aubert et al.
disputes and (Alchian and Demsetz
of the supplier with outsourcing 1999; Lacity
litigation 1972; Barzel 1982; Earl
Economic/ (hidden trans- contracts (RBV), poor cultural and Hirschheim 1996; Lacity et al. 1995)
fit and relational governance 1993b)
Efficiency action costs)
(relationship theories)
Lack of expertise of the client
with contract management (Earl 1996; Lacity et al.
(TCE, TIC, PAT), measurement 1995; Sappington 1991)
Cost escala-
problems (PAT), lack of exper- (Lacity and Hirschheim
tion, loss of 1993b; Lacity et al. 1995)
tise of the supplier with the
control (Alchian and Demsetz
activity (PCE, RBV), uncertainty
(TCE), no critical mass reached 1972; Barzel 1982)
(NET)
Lack of expertise of the supplier
with the activity (PCE, RBV), (Clark et al. (Alchian and Demsetz
task complexity, measurement 1995; Currie 1972; Aubert et al. 1997;
problems + moral hazard and and Willcocks Aubert and Patry 1998;
adverse selection (PAT), incom- 1998; Hirsch- Bahli and Rivard 2003;
Loss of quality plete contracts (TIC), interde- heim and Barzel 1982; Cheon et al.
Lacity 2000; 1995; Earl 1996; Jurison
pendence of activities (PCE), Loh 1994; Loh 1995; Lacity and Hirsch-
cultural barriers and misalign- and Venkatra- heim 1993b; Langlois and
ment (relationship theories) man 1992a) Robertson 1992)

29
Conceptual works (e.g. classifications, decision support models, construct development) and case
studies
92 Theoretical Foundation and Related Research

Provider cannot fulfill all indi-

Loss of
vidual demands (PCE)  (Beimborn et
operational al. 2006a)
(Levina and Ross 2003)
flexibility o outsourcer cannot differenti-
ate (PPF)
(Schott 1997; Lancellotti
Confidentiality Moral hazard (PAT), coopetition
et al. 2003; Landis et al.
problems (relationship theories) 2005)
(Burr 2003; Earl 1996;
Grover et al. 1994;
Strategic (Currie and Hancox and Hackney
Uncertainty about future strate- Willcocks 1999; Lacity et al. 1995;
Loss of own
gic importance of outsourced 1998; Nam et Langlois and Robertson
skills al. 1996; Teng 1992; Nelson et al. 1996;
activity (CCV, RBV)
et al. 1995) Prahalad and Hamel
1990; Quinn and Hilmer
1994)

Loss of skills (RBV), switching (switching

Lock-in, loss (Earl 1996; Langlois and
costs: Apte et
costs (TCE), small number of Robertson 1992; Lacity et
of strategic al. 1997;
al. 1995; Nam et al. 1996;
suppliers (PPF), behavioral Aubert et al.
flexibility Williamson 1985)
uncertainty (NET) 1996b)
Table 5: Undesirable outcomes and theoretical foundation
(based on (Aubert et al. 2005))
When asking for the relative importance of the different arguments against
outsourcing, many research papers name the problem of hidden transaction costs
as the main problem (Aubert and Patry 1996a, 1998; Barthélemy and Geyer
2000; Dibbern et al. 2003; Earl 1996; Lacity and Hirschheim 1993a). Disecono-
mies of scope and coordination problems in particular are expected to lead to
severe problems (Aubert et al. 2000; Earl 1996; Lee and Kim 2003; BIS 2003).
In contrast, Venkatraman and Loh (1993) found the irreversibility of the out-
sourcing decision (lock-in) to be the largest risk.
Table 6 shows the empirical relevance of outsourcing inhibitors based on
three examples of descriptive studies, introduced earlier in Table 4.
Theoretical Foundation and Related Research 93

(Schott (Landis et (Pujals

Outsourcing inhibitor
1997) al. 2005) 2005)
Loss of control (cost escalation) #5 (55%) #1 (35%) #1 (71%)
Loss of own skills #2 (74%) #3 (30%) #3 (31%)
30
Operational risk #4 (58%) #2 (40%)
Transaction costs #6 (30%) #3 (30%) #5 (25%)
Loss of flexibility, #3 (69%) #7 (17%) #4 (30%)
dependency on provider (lock-in)
Cultural/social problems #1 (82%) #6 (22%) #7 (19%)
Loss of quality #8 (15%) #1 (35%) #6 (20%)
Confidentiality problems #7 (17%) #5 (26%)
Table 6: Reasons against outsourcing: results from exemplary studies
(multiple answers possible). Values give ranks and frequency of
positive answers.
Business Process Outsourcing
In general, BPO should have fewer disadvantages than ITO because the tight
interfacial area between IT and business process is not divided interorganization-
ally (Walker and Weber 1984). BPO instead of ITO should lead to less “tightly
coupled” outsourcing and so to fewer problems (Orton and Weick 1990). “BPO
should produce easier to manage, and more successful outsourcing” than IT
outsourcing (Rouse and Corbitt 2004, 2).
In contrast, in BPO, process standardization especially has to be taken into
account (Lancellotti et al. 2003). As argued above, the outsourcer is required to
accept a sourcing provider’s process design, to enable economies of scale (Petzel
2003). In the banking industry, which is the application domain of this work, this
problem might be of less consequence than in other industries, due to a quite
homogeneous product portfolio. But although the banking industry in particular
delivers quite homogenous products, the actual workflow of the different banks’
processes is very heterogeneous. Sometimes standard processes of a potential
sourcing provider are a welcome solution to the updating of internal processes
(Herrmann 2004), but on the other hand, banks often see advantages in running
processes in their own way and are not willing or even able to accept the differ-
ent workflow design (Rouse and Corbitt 2004).
Primary business processes are often outsourced to competitors or jointly
sourced with them (cf. section 3.4). Thus, the problem of ensuring confidentiality

30
Operational problems during the transfer period
94 Theoretical Foundation and Related Research

might be especially parlous because the insourcing firm is able to get an insight
into data and contextual and strategic concerns of the outsourcer. Be this as it
may, this scenario will probably lead to fewer cultural problems between in-
sourcer and outsourcer because both stem from the same industry (Lancellotti et
al. 2003). Employees of both firms have the same understanding of their business
and of the problems that have to be jointly solved. Furthermore, if the processes
are cooperatively, sourced, problems of opportunistic behavior which result in a
loss of quality will be less relevant since the insourcer firm will provide the ser-
vices to itself as well.
The next section will provide a small meta-study of BPO advantages and
disadvantages.

2.2.2.3 A Literature Analysis of BPO Drivers and Inhibitors

There is still a lack of scientific literature regarding BPO (Rouse and Corbitt
2004). In September 2005, we researched the database of three major journal
archives (EBSCO, Science@Direct by Elsevier, and JSTOR) to look for state-
ments regarding advantages and disadvantages of BPO. The databases (title,
keywords, and abstracts of articles) were searched for key terms. In total, only 38
usable documents were found for the term “BPO” or “(Business) Process Out-
sourcing”. Together with “finance” or “bank”, eleven and seven articles were
found, respectively.
Most of the documents found either have been published in the proceedings
of academic conferences or represent studies from large outsourcing providers,
consulting firms, or analysts. Almost all of them are either conceptual or repre-
sent descriptive empirical studies. The following analysis incorporates all of the
works, but distinguishes between scientific literature (peer-reviewed) (13 arti-
cles), scientific and technical literature (non peer-reviewed) (16), and consult-
ing/analyst studies (7). Not surprisingly, outsourcing disadvantages are only
seldom discussed in the last group since it is not in the interest of outsourcing
providers and consultants to emphasize those. Table 7 lists all articles used.
First, papers dealing with BPO drivers or inhibitors, in favor of which they
argue in conceptual terms or of which they empirically approve, were analyzed.
If more than one item was discussed in the paper, then rankings (for drivers as
well as for inhibitors) based on the frequency of the term in the text and on the
order of their first appearance, or based on the empirical results were compiled
for the paper. The average ranking leads to the result tables below, the order is
based solely on the rankings and does not consider how many papers referred to
the outsourcing reason. Nevertheless, this number often follows the same order.
Of course, this is a weak approach, but nevertheless, it gives a good indication
Theoretical Foundation and Related Research 95

for the relevance of the different items. Similar word count approaches have
been used in information retrieval (Drori 2003; Yang et al. 2005) and qualitative
research in social and medical disciplines (Malterud 2001, 487). Empirical stud-
ies have shown the relevance of a term being related to the frequency with which
it appears in a document (Salton and McGill 1983)31.
Journal (Feeny et al. 2005; Fröschl 1999; Lancellotti et al. 2003; Nag 2004;
A: Scientific articles Willcocks et al. 2004)
literature (peer- (Beimborn et al. 2005a; Dayasindhu 2004; Franke and Gewald 2005;
reviewed) (13) Conference
Ganesh and Moitra 2004; Gewald and Hinz 2004; Kshetri and William-
proceedings son 2004; Peng and Wenhua 2004; Rouse and Corbitt 2004)
Book (Halvey and Murphy Melby 2000)
B: Scientific (Alt and Zerndt 2005a+b; Dillmann and Sioulvegas 2003; Grebe et al.
literature and Journal 2003; Herrmann 2004; Kiely 1997; Klaemmt 2003; Margulius 2003;
technical litera- articles Namasivayam 2004; Rebouillon and Matheis 2004; Riedl 2003; Rusch
ture (not peer- 2003; Schneider Traylor 2003)
reviewed) (16) Working
(Feeny et al. 2003; Katre 2005)
papers
(Disher et al. 2004; Friend et al. 2002; Knowledgestorm 2004; Linder et
C: Consulting/analyst studies (7) al. 2002; N.N. 2002b; Snowdon 2004; Tornbohm 2005)
Table 7: Literature used in the analysis
The analysis shows that in all groups the most important argument again
turns out to be cost savings, followed by the less important argument of core
competence focus.
Group A Group B Group C
Number Avg. Number Avg. Number Avg.
Rank Reason Rank Reason Rank Reason
of papers ranking of papers ranking of papers ranking

1 Cost savings 13 1.2 1 Cost savings 15 1.5 1 Cost savings 7 1.5

Core Core Risk

2 5 1.6 2 6 1.7 2 1 2.0
competencies competencies mitigation

Performance Performance / Performance /

5 6 2.7 3 4 2.0 3 5 2.6
/ skills skills skills

3 Quality 7 2.8 4 Quality 6 2.3 Core

4 5 2.6
competencies
4 Flexibility 6 2.9 5 Flexibility 6 2.8
5 Flexibility 5 3.0
Risk Risk
6 1 5.0 6 2 3.5
mitigation mitigation 6 Quality 2 4.0

Table 8: Literature analysis of BPO drivers

In addition to cost savings, outsourcers usually seek to increase the perform-
ance and quality of the process by utilizing specialized resources of the service
provider. Another reason for BPO is to increase the firm’s flexibility. Usually it
is not the core objective but an expected positive side effect. Generally, there is
quite consistent accordance between the different groups of literature. The high

31
Lebart et al. (1997) give an introduction to the different methods of statistical text analysis.
96 Theoretical Foundation and Related Research

ranking of risk mitigation in the consultants’ literature is based on only one

study; if the items were reordered on the basis of the number of papers in which
they were cited, it would drop to the last place as it was the case in the other
groups.
The next table shows the results for outsourcing inhibitors, following the
same approach.
Group A Group B Group C
Number Avg. Number Avg. Number Avg.
Rank Reason Rank Reason Rank Reason
of papers ranking of papers ranking of papers ranking

Dependence Dependence Loss of

1 3 1.0 1 5 1.5 1 2 1.5
on insourcer on insourcer flexibility

Loss of Loss of Dependence

2 4 2.0 2 4 1.7 2 2 2.0
flexibility flexibility on insourcer

3 Loss of skills 2 2.5 2 Loss of skills 3 2.0 2 Loss of skills 1 2.0

Hidden Hidden Hidden

3 2 2.5 2 1 2.0 4 0 n/a
costs costs costs

Table 9: Literature analysis of BPO inhibitors

Although there are not many results relating to BPO risks that can be ex-
tracted from the literature, the main risks of becoming dependent on the in-
sourcer and losing strategic flexibility are clear. Surprisingly, hidden transaction
costs do not play a major role in the discussion of outsourcing inhibitors. This
can be explained by the argument that cutting whole business functions out of a
firm might be easier than unpicking the business from the IT layer.
A second possible explanation is that transaction costs might be underrated
by firms, as it was done in the beginning of the IT outsourcing era, too. Accord-
ingly, Corbitt and Rouse (2003) found that cost savings from BPO are quite low
when taking transaction costs into account (Rouse and Corbitt 2004). In another
study carried out by them, 20% of the 240 participants even stated that BPO
leads to increased instead of reduced total costs.

2.2.3 Formal Models in the Context of Outsourcing

This section gives a very brief overview of the few formal contributions to out-
sourcing research that use mathematical models for analytical investigations.
These can basically be split into two distinct groups according to their underlying
research questions: why to outsource and how to outsource (in the sense of estab-
lishing an efficient and effective relationship mode in terms of contractual de-
sign, managerial control etc.), again following the classification of Dibbern et al.
(2004). Naturally, all of these models employ certain restrictive assumptions,
ignoring some of the real world’s complexities. They just focus on particular
aspects often ignored by empirical and conceptual research (Dibbern et al. 2004),
Theoretical Foundation and Related Research 97

such as uncertainty (Lammers 2005), task interdependencies (Knolmayer 1993),

information asymmetries (Sridhar and Balachandran 1997), contract design (Ak-
sin et al. 2004), risk attitudes, and incentives (Chalos and Sung 1998).
In order to answer the research question regarding why to outsource, several
decision models are proposed. Lammers and Tsang propose decision models
based on PCE, TCE, and RBV, to evaluate the advantageousness of in-house
production vs. outsourcing vs. cooperative sourcing (Lammers 2004) or own
production vs. cooperative sourcing (Tsang 2000) of a single activity. Further,
Lammers developed a decision calculus which captures uncertainty about future
process volumes by using a real options approach (Lammers 2005). Knolmayer
developed a quadratic assignment model to support decisions about the sourcing
location of multiple interrelated activities (Knolmayer 1993; 1994), taking task
interdependencies and the strategic value of business functions into account.
The works investigating the question of how to outsource usually assume
that a firm can increase the performance of the activity under consideration by
outsourcing it (Dibbern et al. 2004). Primarily based on agency theory but also
on the theory of incomplete contracts and cooperative game theory, these works
try to determine efficient and effective contractual designs, relationship modes,
and incentive mechanisms, given different settings of information asymmetry
and levels of risk aversion (e.g. (Chalos and Sung 1998; Demski and Sappington
1993; Feenstra and Hanson 2003; Gietzmann 1996; Lewis and Sappington 1991;
Van Mieghem 1999; Wang et al. 1997; Whang 1992)). Some of them assume
monitoring efforts or agency costs to be higher in outsourcing than in hierarchi-
cal relationships (Sridhar and Balachandran 1997), others, in contrast, assume
that firms outsource in order to improve managerial incentives (Chalos and Sung
1998). As well as avoiding moral hazard during a sourcing relationship, there are
also suggestions for the pre-contract phase in order to avoid adverse selection
and to reduce costs by implementing an appropriate selection/bidding mecha-
nism when multiple potential vendors are present (Chaudhury et al. 1995; Klotz
and Chatterjee 1995; Sarkar and Gosh 1997).
Further, a group of papers investigates analytically how the advent of IT
(and therefore an increase of information symmetry and an reduction of monitor-
ing costs) affects the degree of outsourcing (or sometimes, viewed more gener-
ally, as the buyer/supplier relationship), the relationship mode, and the contrac-
tual design, e.g. (Bakos and Brynjolfsson 1993a; Bakos and Brynjolfsson 1993b;
Banker et al. 2000; Clemons and Reddi 1994; Clemons et al. 1993).
Outside outsourcing research, authors have developed formal cooperative
sourcing models that describe demand and cost functions of two players who
have to decide about the level of investment and about merging their manufactur-
ing into a single firm or (in some models) outsourcing it to a third party (An-
98 Theoretical Foundation and Related Research

upindi and Bassok 1999; Lee and Whang 2002; Lippman and McCardle 1997;
Plambeck and Taylor 2001; Spulber 1993; Van Mieghem 1999). The models
differ in their assumptions about exogenous vs. endogenous and deterministic vs.
stochastic parameters, available strategies to choose from, degree of competition
between the potential partners, and information available to the parties. The
models are used to develop mechanisms for allocating the cooperation gain
which lead to stable partnerships (cf. the game-theoretical analysis in section
5.1).

2.2.4 Summary
This chapter gave an overview of previous research regarding the question why
or why not to outsource. Based on the theoretical derivations in section 2.1, em-
pirical research works on IT outsourcing have been briefly reviewed showing
that the main reasons for outsourcing are cost savings and variabilization, core
competence focus and access to superior technologies (section 2.2.2.1) while
major inhibitors prove to be hidden transaction costs and the risk of lock-in into a
sub-optimal sourcing location (2.2.2.2). The short analysis of the sparse literature
on BPO (2.2.2.3) reveals that the results from ITO research can mostly be trans-
ferred to this related domain. Only the impact of hidden transaction costs seems
to be less important in this new field, explained by two possible reasons. First,
BPO might be less difficult in terms of cutting whole business functions out of a
firm because the close relation between IT and business does not have to be cut
interorganizationally. Second, maybe even these recent studies of BPO show the
same lack of awareness of the risk of hidden costs as in the early years of IT
outsourcing.
Preparatory to the development of the formal cooperative sourcing model in
chapter 4, the last section (2.2.3) summarized the few existing formal works on
outsourcing.
Cooperative Sourcing in the Banking Industry 99

3 Cooperative Sourcing in the

Banking Industry
“An attractive option for many financial institutions is to create joint ventures aimed at sharing ex-
ternal sourcing, operations, and platforms for systems and delivery. Although such initiatives are rela-
tively new in financial services, they have proved to be critical differentiators for top retailers.”
(Riera et al. 2003)

The aim of this chapter is to provide insights into the chosen application domain
and to clarify the motivation behind selecting cooperative sourcing behavior in
the banking industry.
There is lot of discussion that German banks are being confronted with mul-
tiple problems which endanger their competitiveness (e.g. Dombret and Kern
2003; Koetter et al. 2004). Driven by these structural issues, which will be high-
lighted in the following sections, banks are starting to follow the principles of the
common hype term industrialization, incorporating strategies such as process
automation, vertical disintegration, horizontal integration, standardization, and
modularization of their business. Their aim is to achieve international competi-
tiveness through cost efficiency, increased flexibility and differentiation (Eng-
stler and Vocke 2004; Licci 2003; Linn 2005).
Starting with a discussion and empirical evidence relevant to the general
problems facing German banks, this chapter will gradually zoom in on the cho-
sen research object of cooperative sourcing in the banking industry. Since the
credit business is the empirical application domain of this thesis, the chapter will
focus particularly on this particular business domain.
The chapter is structured as follows: section 3.1 gives an actual overview of
the current situation in the German banking industry as well as of the ongoing
“industrialization” activities. Based on this, section 3.2 analyzes normative litera-
ture regarding a possible future configuration of the banking industry (segmenta-
tion models). Section 3.3 focuses on the particular banking business of granting
and processing loans. Credit products are classified, followed by a process view
that applies the general segmentation models to the credit business. Section 3.4
summarizes the current outsourcing activities in the banking industry with a
particular focus on credit process outsourcing while section 3.5 completes the
picture with a brief discussion of the legal and regulatory issues governing out-
sourcing and cooperative sourcing in the financial industry.
100 Cooperative Sourcing in the Banking Industry

Subsequently, section 3.6 presents our own empirical evidence of BPO and
particularly of cooperative sourcing in the credit business. Based on empirical
studies with German banks, the status quo and potential of credit process out-
sourcing is analyzed. BPO drivers and inhibitors, discussed in the theory chapter
(esp. section 2.2.2) are compared with empirical data and the various process
characteristics which are relevant to the outsourcing potential are explored.

3.1 Current Situation in the German Banking

Industry
The current competitive situation of the German banking industry is frequently
discussed. High fragmentation (section 3.1.1.1), overbanking (3.1.1.2), and
strong vertical integration (3.1.1.3) are thought to be the causes of the underper-
formance of German banks (section 3.1.1). Section 3.1.2 describes two generic
strategies to overcome those deficits – consolidation and deconstruction – and
classifies the concept of cooperative sourcing as a combination of both strategies.

3.1.1 Structural Deficits in the German Banking Industry

Compared with the banks of other European countries, German banks show a
very poor overall cost structure and profitability. Figure 10 compares the average
return on equity (ROE)32 and the average cost/income ratio (CIR)33 of different
European countries in 2006. German banks (at national aggregate level) showed
one of the worst positions of all European countries with a CIR of 65.2% and a
resulting ROE of 10.2%. At international level, a ROE of 15% is frequently
argued to be necessary for covering the costs of capital (Moormann and Möbius
2004); on overall European average, the ROE was 16.6% in 2006.

32
ROE represents the return on a bank’s equity, i.e. profit related to equity (including reserves).
33
CIR is a measure for evaluating a bank’s cost efficiency. It results from dividing operating ex-
penses by operative income (= interest surplus, commission surplus, and trade surplus).
 Cooperative Sourcing in the Banking Industry 101

45

unweighted mean

weighted mean
Estonia
40
Return on Equity (ROE)

35
Hungary
Slovakia
30 Latvia

25 Belgium
Luxemburg
unweighted mean Spain Sweden France Romania
20
UK Ireland Slovenia Poland
Greece Italy Bulgaria
weighted mean Portugal
Malta Austria
15
Czechia Lithuania Finland
Cyprus Denmark Netherlands
Germany
10
25 30 35 40 45 50 55 60 65 70 75
Cost/Income Ratio (CIR)
Figure 10: CIR and ROE in 2006, compared at national level (the weighted
mean takes the number of banks in each country into account,
data source: ECB 2007)
Owing to this low profitability, German banks show a very low market capi-
talization. Therefore, as international markets become increasingly liberalized,
German banks are more likely to become targets for the acquisition strategies of
large international players, constituting a threat to the autonomy and strength of
the German financial industry.
Of course, Figure 10 would show different results if the data represented
single banks instead of aggregate national data. Especially the large players in
the German banking market show significantly better figures. Moreover, the
overall ROE of German banks more than doubled from 2004 to 2006 which
indicates a significant improvement of the situation in the German banking in-
dustry.
Researchers and experts have identified three different reasons for this
alarming situation in Germany: strong market fragmentation, overbanking and a
high level of vertical integration. These are discussed in the following sections.

3.1.1.1 Fragmentation
Germany has the most fragmented banking market in Europe. Figure 11 shows
the market share of the five largest banks in the different European countries.
102 Cooperative Sourcing in the Banking Industry

100
90
80
70
60
50
40
30
20
10
0
Latvia

Slovenia

Slovakia
Ireland
Italy

Spain
Austria

France

Greece

Malta
Luxemburg

Poland

Czechia

Lithuania
Finland

Estonia
Portugal

Belgium
United Kingdom
Germany

Hungary

Denmark
Cyprus

Netherlands
Figure 11: Aggregate market share of the five largest banks in different
countries in 2006 (data from (BDB 2006))
Over the last years, the five largest German banks (by total assets) continu-
ously showed a combined market share of only about 22% while the (un-
weighted) European average is around 59% (BDB 2006). The profitability of the
German Top 5 banks has significantly increased in recent years (to 15% in
2006), the market shares remained fairly stable in all countries.
Two alleviating factors have to be taken into account when making compari-
sons between countries. First, there is quite a strong negative correlation between
a country’s size and its market concentration (r=-.58 between population size and
cum. market shares). Second, the banking structure in Germany shows close
cooperation between the public savings banks and credit cooperatives, which
means that individual banks within these sectors sometimes are only partially
seen as separate firms. An appropriate concentration measure would have to
consider the structure of those associations.
Nevertheless, other large countries, such as France and the UK, show that
high market concentration is not restricted to small countries. Moreover, some of
the European countries have association structures that are quite similar to the
German associations of public savings banks and cooperatives.

3.1.1.2 Overbanking
The high fragmentation of the German banking market (section 3.1.1.1) is
strongly related to overbanking. Compared with other European countries, Ger-
many has a lot more banks, bank branches and bank employees (Moormann and
Möbius 2004, 26-28; Weber 2002, 458).
Cooperative Sourcing in the Banking Industry 103

Figure 12 shows the relationship between CIR and different measures of

banking density (banks, bank branches, and bank employees per 100,000 inhabi-
tants) for different European countries in 2003.
75% 75%
efficiency problems overbanked efficiency problems overbanked
Germany Germany
Benelux Benelux
70% 70%
Switzerland Switzerland
France France
Cost/Income-Ratio

Cost/Income-Ratio
mean mean
65% 65%
Italy Italy

60% Scandinavia 60% Scandinavia

Spain/Portugal
Spain/Portugal

55% 55%

Great Britain Great Britain

mean

mean
efficient overbanked but efficient efficient overbanked but efficient
50% 50%
0,00 1,00 2,00 3,00 4,00 5,00 6,00 10 20 30 40 50 60 70 80 90 100
banks per 100K capita bank branches per 100K capita

75%
efficiency problems overbanked
Germany
Benelux
70%
Switzerland
France
Cost/Income-Ratio

mean
65%
Italy

Scandinavia
60%
Spain/Portugal

55%
Great Britain
mean

efficient overbanked but efficient

50%
400 600 800 1000 1200 1400
bank employees per 100K capita

Figure 12: Relationship between CIR and number of banks, bank employees,
and number of branches in 2003 (data from FBE 2003)
Germany is the only country that had a CIR below average and a banking
density above average for every density measure. It follows that – irrespective of
the measure in view – Germany has an overbanked, inefficient banking market.
Linear regressions confirm the assumption that a high density of branches and a
high number of employees lead to costs that are too high and thus increase the
CIR (and decrease the ROE)34.
Although the degree of overbanking has decreased in recent years by closing
branches, reducing human resources, and merging banks (Weber 2002, 456), this
analysis shows that the goal has not yet been reached. A particular problem,

34
R2=.22 (effect of number of banks on CIR), R2=.53 (effect of number of bank branches on CIR)
and R2=.32 (effect of number of bank employees on CIR). The following outliers have been re-
moved before conducting the regression analyses: Spain/Portugal for measuring the effect of bank
branch density and Switzerland for measuring the effect of the number of bank employees.
104 Cooperative Sourcing in the Banking Industry

especially for mergers & acquisitions, are the three sectors of commercial banks,
public savings banks, and credit cooperatives. Mergers between banks of differ-
ent sectors are still seen as quite unrealistic (although some acquisition tenden-
cies have already happened).

3.1.1.3 Vertical Integration

Measurement of Vertical Integration

Vertical integration describes the ratio between in-house business functions and
all business functions needed to make a product or to carry out a service (Adel-
man 1955; Picot 1992, 104). The most common approach to measuring the verti-
cal integration of a firm is the VAS (value added to sales index) (Martin 1986),
based on the works of Adelman (Adelman 1955; Gort 1962; Nelson 1963). It is
defined as 100% less purchases per sales, as represented by Equation 3.
sales  purchases
u 100%
sales
Equation 3: Value added to sales index (VAS)
The index was adapted to a vertical integration index for the banking indus-
try by Gellrich et al. (2005). Figure 13 shows the formal representation and the
different components.

VA VA  ( NIAT  IT )
VI AVI
sales sales  ( NIAT  IT )

Vertical Integration Index Adjusted Vertical Integration Index

Sales IIn  FIn  CIn  TIn  OIn VA IE  LE  LLP  IT  NIAT

IIn = Interest income LE= Labor expenses

FIn = Fee income IE = Interest expenses
CIn = Commission income LLP = Loan loss provisions
TIn = Trade income IT = Income taxes
OIn = Other income NIAT = Net income after tax
VA = Value added
Figure 13: Vertical integration index and adjusted vertical integration index
Gellrich et al. defined the value added as the sum of loan loss provision, in-
terest and labor expenses, income taxes, and net income, whereas the vertical
integration index itself is described by the ratio of value added to sales. Sales
Cooperative Sourcing in the Banking Industry 105

include commission income, fee income, interest income, trade income, and
others. The effects of changes in profitability and taxation should be eliminated,
resulting in the adjusted vertical integration index (Gellrich et al. 2005; Tucker
and Wilder 1977).
A very similar method of calculating the degree of vertical integration is the
measurement of the value added ratio (VAR), which is calculated by the German
Federal Statistical Office (FSO). As Figure 14 shows, first, the bank’s “revenue”
– the gross output value – is calculated (difference between interest income and
interest costs plus further income) (Glöckeler 1975, 20). Revenue less further
costs forms the gross value added. The degree of vertical integration is repre-
sented by the ratio of gross value added to gross output value (Weisser 2004).
Gross value added
Value added ratio
Gross output value

Gross value added Gross output value  LE  LLP

Gross output value IIn  IE  FIn  CIn  TIn  OIn
IIn = Interest income LE= Labor expenses
FIn = Fee income IE = Interest expenses
CIn = Commission income LLP = Loan loss provisions
TIn = Trade income
OIn = Other income
Figure 14: Value added ratio (VAR) (Weisser 2004)
Both forms of measurement have weaknesses. Given a vertical supply chain,
the measurement correlates with the firm’s position in the chain. The nearer the
firm’s business is to the primary level of value creation, the higher is the degree
of vertical integration. Therefore, the measurement shows a firm’s position
within the value chain rather than its coverage of the chain (Bauer 1997, 32-33).
Furthermore, the measurements are influenced by factors that do not relate to the
degree of vertical integration (Weisser 2004, 50), such as the deployment of
expensive resources (e.g. technology), increasing prices on the sales side, and the
firm’s profit. The higher the profit, the higher will be the measured degree of
vertical integration.
A general problem of the measurements is their comparability to other
branches. Because there are structural differences between the profit and loss
accounts of banks and of other industries, the resulting values are not really
comparable. The “revenue” of a bank is quite difficult to compare with the value
creation of other industries.
Another important method of measuring the vertical integration is the verti-
cal industry connection index (VIC) (Maddigan 1981; Maddigan and Zaima
106 Cooperative Sourcing in the Banking Industry

1985) which is based on input-output-matrices of the Leontieff Model (Leontieff

1951). The VIC is much more sophisticated and more precise because it incorpo-
rates input and output data of the different products and production factors in-
volved in a firm’s business. Because the data for those input-output-matrices is,
unfortunately, not publicly available, an empirical estimation of the VIC of the
banking industry is not possible. Although there are some problems with the
application of VAS and VAR, these are normally used for conducting empirical
analyses.

Vertical Integration in the German Banking Industry

German banks are typically characterized by a very high degree of vertical inte-
gration. Alongside high market fragmentation (section 3.1.1.1) and high banking
density (section 3.1.1.2), this represents a third reason for their high CIR and low
profitability.
The comparison of different industries in Figure 15 (left) reveals the high
degree of vertical integration in the banking industry. By contrast, other indus-
tries have radically reduced their degree of integration over recent decades and
optimized their supplier network.
100%
100%
90% 20%
90%
80%
50% 80% 45%
70% 70% 60%
60% 78% 80%
89% 60%
50% 50% 98%
40% 80% 40%
30% 30%
50% 55%
20% 20% 40%
10% 22% 20% 10%
11%
0% 0%
Investment
Cayenne

processing

processing

Loans
Electronics

Banks
Automobile

Porsche

Payment
Industry

Security
goods

Buy Buy
Make Make

Figure 15: Degree of vertical integration in different industries (left) and in

different banking businesses (right) in Germany (data from: (Bösch
1999; Dombret 2004; Eichelmann et al. 2004; Gellrich et al. 2005))
Many reports estimate the degree of vertical integration in the German bank-
ing industry to be around 80% (e.g. Platzer and Riess 2004; Sauter 2002). Unfor-
tunately, many of those works do not empirically validate their results, but only
cite each other. An actual VAS calculation approving this value (83.7%) was
done by Gellrich et al. (2005). However, there are also differing results, for ex-
ample in (Kassner 2004), where a vertical integration degree of only 67.6% is
calculated. The national accounts of the FSO show a significant decrease in ver-
Cooperative Sourcing in the Banking Industry 107

tical integration. In 1996, the vertical integration of the German banking industry
was 69% and decreased to 51% in 2002. This tellingly shows the effect of the
different outsourcing activities in the industry (Weisser 2004, 49).
Why is there a relationship between the degree of vertical integration and
profitability? When discussing the competitiveness of German banks, two
prominent success factors are commonly discussed in the literature: focused
business models and effective cost management (Licci 2003). Both factors relate
directly to disaggregating the banking value chain.
Specialized providers, who could insource particular areas of the banking
value chain, would be able (due to economies of scale and economies of skill, cf.
section 2.1.1) to generate cost reduction (Alms 2003; Benna et al. 2003, 91;
Bösch 1999, 32; Hackethal 2003, 33). Furthermore, a focus on core competen-
cies promotes flexibility potential (cf. section 2.1.6 on RBV and CCV). A the-
ory-based discussion of reasons for and against disaggregating the banking value
chain by outsourcing was given in the previous chapter.
Gellrich et al. were able to partially prove the correlation between the degree
of vertical integration and profitability (ROE). They showed that banks with
either a low or high degree of vertical integration were more likely to work prof-
itably. By contrast, banks that neither had a clear integration strategy nor fol-
lowed a disaggregation strategy were “stuck in the middle” (Gellrich et al. 2005,
12).
The diagram on the right in Figure 15 shows the degree of vertical integra-
tion for different key banking products. While in the credit business almost eve-
rything (98% of the value chain) is provided within the bank, transactional proc-
esses (payments and securities processing) have lower levels of integration, indi-
cating more outsourcing activities in these business segments (cf. section 3.4).

3.1.2 Current Tendencies

The section above discussed three main reasons for the low international com-
petitiveness of the German banking industry. This section will discuss how Ger-
man banks are reacting to those problems. A visualization of the argumentation
path is given by Figure 16 (König and Beimborn 2008).
108 Cooperative Sourcing in the Banking Industry

Problem Reason Countermeasure

Consolidation
Low profitability Mergers &
Fragmentation
(ROE) acquisitions (M&A)

Deconstruction
Limited Outsourcing
Overbanking
competitiveness Cooperative Sourcing

Low cost efficiency

Vertical integration Divestments
(CIR)

Horizontal integration
(Universal Banking)

Figure 16: Structural problems of the German banking industry

(König and Beimborn 2008)
There are two basic strategies to react to the problems under discussion. The
first is to increase the firm size (“consolidation”) by mergers & acquisitions
(M&A) or cooperative sourcing, whereas the other is to decrease firm size (“de-
construction”) by outsourcing/cooperative sourcing and divestments (Walter
2001, 39). A retail banking survey, conducted by IBM, showed a high take-up
for both of these strategies. As of 2003, already 70% (30%) of participating
banks had considered (undertaken) M&A, while 57% (31%) had considered
(undertaken) to take part in a joint venture (o cooperative sourcing). 37% (28%)
had considered (realized) divestments (IBM 2003, 19).

3.1.2.1 Consolidation
Consolidation leads to an increase in market concentration. Concentration can be
operationalized as an absolute or relative measure. The statistical term “(relative)
concentration” focuses on disparities and describes an unequal distribution of the
sum of attributes to the different attribute carriers (Börner 1998). Absolute con-
centration, by contrast, involves the sum of attributes being distributed to a low
number of attribute carriers. Following the first definition, the German banking
industry has always been quite concentrated because it has always shown a very
heterogeneous structure in terms of institute size. If the definition of absolute
concentration were applied, we would have to discuss what would constitute a
low number of attribute carriers (Börner 1998). In order to avoid this, we will
only talk about increasing (absolute) concentration as the macro effect of con-
solidation activities.
Cooperative Sourcing in the Banking Industry 109

There are both strategy and efficiency reasons for seeking to consolidate dif-
ferent firms by M&A. While the main strategic reason is to increase market
power, efficiency reasons primarily focus on economies of scale, scope and skill.
In the past, these reasons led to strong and continuous M&A activities in most
countries of the European Union (Börner 1998, 36-38). Most countries are at a
far more advanced state than Germany. Nevertheless, Figure 17 shows that be-
tween 12/1985 and 12/2007 the number of reporting banks in Germany de-
creased strongly from 4,659 to 2,015 with a break between 1989 and 1990 due to
the German reunification.
5000
4500 Number of reporting credit institutions
4000
3500
3000

2500
2000
1500
1000

500
0
85

87

89

91

93

95

97

99

01

03

05

07
19

19

19

19

19

19

19

19

20

20

20

20

Year
Figure 17: Number of reporting banks in Germany (data source: (DBB 2005))
This almost linear consolidation trend was mainly due to the merging of
small credit cooperatives. Half of the banks participating in the IBM retail bank-
ing survey cited above (IBM 2003) believe that the industry will carry out further
substantial consolidation steps (54%) and that achieving economies of scale will
be a key success factor (46%) in the future.
In recent years, larger banks have also focused on mergers, although really
big M&A deals have not yet occurred in Germany, except for one case in 1998
(Bayerische Vereinsbank + Bayerische Hypotheken- und Wechselbank = Hy-
poVereinsbank). In the past, most consolidation processes in which large banks
were involved consisted of large banks acquiring small ones. This trend has
changed now so that banks of similar size merge, too. Since the number of big
consolidation candidates in Germany is very low, it is assumed that cross-border
mergers will increasingly occur (Börner 1998, 32-34; Walter 2001, 39). A first
example is the acquisition of HypoVereinsbank by UniCredit (Italy) in 2005,
which, however, has not yet been integrated. By contrast, a major argument
110 Cooperative Sourcing in the Banking Industry

against cross-border mergers are the estimated smaller efficiency potentials. The
synergy benefits of national mergers of similarly large banks are estimated to be
three times higher than for those of cross-border mergers because there is more
of an overlap of business segments in domestic mergers than in cross-border
deals (Hamoir et al. 2002). Further difficulties are posed by different regulatory
settings, which lead to different business process designs and cultural barriers
(different corporate philosophies). Harmonization efforts of the European Union
are an important step in changing this. Some authors believe that a domino effect
will occur once a big merger deal has been realized. The number of suitable
partners will then decrease, thus rapidly forcing banks to react (Hamoir et al.
2002). A similar phenomenon occurred in the airline industry where the airlines
did not merge but rapidly formed quite tight alliances, leaving some late movers
behind.
The political dimension cannot be ignored. “The domestic banks in
EURO[35] were – and are – protected as domestic flagships. The fundamental
belief that financial institutions should not be controlled by foreigners has (so
far) prevented almost any type of cross-border merger” (Boot 1999, 2). More-
over, it is not only governments who want to strengthen the power of “their”
banks, but also the managers themselves. Consolidation trends are, therefore, not
only driven by efficiency and strategy reasons but also by personal incentives.
To put all this in an international context, the development of the US bank-
ing market will be briefly outlined in the following paragraphs. In the 1990s, the
USA had a much more fragmented industry than any other developed country –
with about 10,000 more financial institutes than the remainder of the G-1036
combined (Berger et al. 1999), which, moreover, showed to be a very dynamic
market. For example, between 1985 and 1990, 200 banks failed while 200 new
institutes were formed (Berger et al. 1999). Nevertheless, the industry is not
over-branched. Even in 1997, there were only 36 branches per 100,000 inhabi-
tants, which – even compared with the European figures from 2003 (Figure 12) –
is a very low value and was the lowest of all G-10 countries in 1997 (Berger et
al. 1999).
The beginning of the 1980s saw the start of a trend towards consolidation
that accelerated further at the end of the decade. Megamergers (i.e. mergers be-
tween banks with an assets total over $1 billion each) became very common and

35
European currency area
36
G-10 = “Group of Ten”: Belgium, Canada, France, Germany, Italy, Japan, the Netherlands, Swe-
den, Switzerland, the United Kingdom and the United States.
Cooperative Sourcing in the Banking Industry 111

between 1992 and 2007 the number of credit institutions dropped from 27,210 to
16,82637 (Figure 18).
30,000
Credit Unions
25,000 Savings Institutions
number of institutions

Commercial Banks
20,000

15,000

10,000

5,000

0
1992 1994 1996 1998 2000 2002 2004 2006
year
Figure 18: Consolidation of the US banking market37
The consolidation trend is mainly driven by the motive of improving the in-
stitute’s market position – on the one hand by increasing market power and being
able to set prices, and on the other by increasing cost efficiency and diversifying
business risks. The former explains the national mergers of very large players
while the latter explains the common trend that large banks acquire smaller
banks once they have reached a certain size (Berger et al. 1999; Vander Vennet
1999). Some side-effects have been the suspension of inter-state merger restric-
tions and other deregulation steps, governmental activities in financial crises
(e.g. between 1984 and 1991 the US government provided financial assistance to
allow healthy banks to purchase over 1,000 insolvent US banks), and the techno-
logical progress which increases potential economies of scale in all areas of the
banking business (electronic sales and service delivery channels, new financial
engineering tools, improved transaction systems) (Berger et al. 1999).

3.1.2.2 Deconstruction
In addition to the tendency to consolidate, there is an emerging trend to decon-
struct German universal banks. Deconstruction or disintegration describes the
logical splitting of the value chain and its subsequent reorganization (Walter
2001, 39).

37
Data sources: Federal Deposit Insurance Corporation (http://www2.fdic.gov/sdi/sob/ (as of 07 Feb
2008)) and Credit Union National Association (http://advice.cuna.org/econ/cu_stats.html (as of 15
Feb 2008).
112 Cooperative Sourcing in the Banking Industry

Deconstruction can involve both a vertical disintegration as well as a hori-

zontal disintegration (from universal banks to product specialists). Since out-
sourcing reduces a firm’s involvement in successive stages of production, it may
also be viewed as vertical disintegration (Gilley and Rasheed 2000, 764). By
contrast, divestments from whole product segments represent the strategy of
horizontal disintegration.
Strategies to deconstruct the banking value chain have only been possible
since the introduction of information technologies that make interorganizational
information systems (IOS) possible. One of the first outcomes of deconstruction
tendencies has been the emergence of internet-oriented direct banks. Other out-
comes are the outsourcing of business processes to transaction banks and credit
factories (cf. section 3.4.2 and 3.4.3).
The combination of both strategies – deconstructing the monolithic univer-
sal bank and consolidating particular business units of different banks – de-
scribes the concept of cooperative sourcing, which is the research object of this
work.
Consolidation and deconstruction are the main organizational strategies to
solve the problem of the decreasing competitiveness which banks are currently
faced with. Both concepts have been successfully applied by other industries in
previous decades. While those have developed their value chain network over a
long time and are still optimizing it, banks are forced to reshape their business
much faster, due to rapidly changing environments and high competitive pres-
sure. As a result, they often copy industrial concepts, not knowing whether this
approach is valid for their completely different business and process characteris-
tics. This research work tries to shed some light on the implicit basic hypothesis
underlying all banking industrialization tendencies.

3.2 Segmentation in the Banking Industry

Based on the strategy options discussed above, this section will conceptualize the
possible outcomes of deconstruction and consolidation activities in the banking
industry. While the first subsection (3.2.1) develops a generic banking value
chain for the typical German universal bank, the second subsection (3.2.2) sum-
marizes and discusses the current state of the literature regarding the layout of
the banking industry of the future.

3.2.1 Generic Value Chain of the Banking Industry

In contrast to physical industries, banks do not create value by producing and
refining material goods. Their value creation can be described by risk takeover
Cooperative Sourcing in the Banking Industry 113

and information processing (Polster 2001, 15). Banks have a “production proc-
ess” consisting of three steps (Dombret and Kern 2003): first, the bank develops
the product and prepares its technical readiness for executing transactions or for
performing services. It has to be determined whether the product can be managed
effectively by the bank’s risk management. The second step consists of cus-
tomer-initiated sales and provision of a product. This step generally includes all
activities ranging from branding and marketing through sales and cross-selling to
customer management (i.e. the customer interface). The third step of the value
chain represents the fulfillment and comprises administration and transactions
which actually provide the service.
The essential characteristic of a typical banking product is that it is produced
to order, i.e. it will not be produced before the customer initiates the provision
(Dombret and Kern 2003, 29). The three steps of the banking value chain, par-
tially, run in parallel. For example, in many cases, sales cannot be completed
without being supported by the administration and transactions infrastructure as
well as by risk management.
The following generic value chain model of the typical German fully inte-
grated and universal bank (Figure 19) distinguishes between primary and secon-
dary activities, following Porter’s value chain (Porter 1985) and details the three
value chain steps.
114 Cooperative Sourcing in the Banking Industry

Primary activities

Product Customer interface Fulfillment

development Branding/marketing Sales Client management Product admin. transactions
Customer care &
Marketing Current accounts
Customer data analysis information Payments processing
intelligence administration
(after sales)
Design product/ Savings and time
Marketing Activity
financial Receive orders deposits Clearing & settlement
intelligence management
engineering administration
Refine product/ Manage, amend &
Acquisition/product Loans
financial Branding update customer Securities trading
offering administration
engineering data
Complaint Credit cards
Legal affairs Targeting Advisory Securities custody
management administration
Financial Product/pricing Loans processing &
Advertising Brokerage
supervision configuration servicing
Managing marketing Other Credit card processing
Pricing Contract closure
campaigns investments & servicing
Business process Insurance
Sales processing
implementation administration
Manage product portfolio Sales monitoring Foreign trade services Foreign exchange
dealing (ForEx)
Channel-management/multi-channel management Safe deposit&custody
Customer relationship management Other services

Secondary activities
Enterprise planning and management
Strategic management Property, firm infrastructure
Financial management ICT management
Risk management Stakeholder relationship management
Internal auditing & compliance Procurement
Legal & tax management Insurance management
Financial supervision management Expedition
Human Resources

Collaboration management
Strategic collaboration Planning collaboration Operational collaboration

Figure 19: Detailed generic banking value chain, based on (Dombret and Kern
2003; Lammers et al. 2004; Petry and Rohn 2005; Porter 1985)
Primary activities are part of the core banking business while secondary ac-
tivities are supporting business functions with an internal focus, including cross-
sectional functions for managing the bank (enterprise management) and its inter-
firm partnerships (collaboration management) (Spiegel 2002) (cf. section 1.5.1).
Enterprise management has two functions which are specific to the banking
business and are closely related to all steps of the value chain: financial man-
agement and risk management. Risk management involves integrated controlling
of all relevant risks for all primary activities, such as market risks, credit risks,
and operational risks (BIS 2004). Financial management mainly represents the
treasury function (i.e. refinancing management, liquidity management, asset and
liability management). The treasury is very closely related to almost all primary
activities. For example, during the development and marketing of a new loan
product, it has to determine whether (and on what conditions) incoming credit
exposures can be refinanced and the refinancing has to be arranged. A second
Cooperative Sourcing in the Banking Industry 115

main part of the financial management apart from the treasury is proprietary
trading of securities.
As already shown in section 3.1.1.3, banks still integrate the most important
parts of the banking value chain within their own boundaries, i.e. only minor
parts of the primary activities have been outsourced38. The next section will de-
scribe normative segmentation models which can be found in the literature and
which describe how fully integrated and universal banks may be transformed
into a banking value network in the future.

3.2.2 Segmentation Models

Many experts believe that the model of the typical fully integrated and universal
German bank will not persist into the future (e.g. Jasny 2001; Marighetti et al.
2001; Salmony 2002; Walter 2001). They assume that banks will focus on their
core competencies while outsourcing the remainder. The following basic models
describe how the banks are supposed to disintegrate their business to form cost-
efficient and more flexible business value networks which consist of independent
but interlinked banks with different business models (i.e. segments39). These
models will be mapped to the generic value chain in the following sections.

3.2.2.1 Three Segments Model

The three segments model, as the most common segmentation model, assumes a
segmentation of the market into sales banks, portfolio/product banks, and pro-
duction/transaction banks (Hamoir et al. 2002; Salmony 2002; Steffens 2002).
The core competence of a sales bank would be marketing and sales activities. It
manages the sales of different banking products and services which are offered
by the other segments and provides the customer interface (traditional and elec-
tronic channels) to retail and corporate clients (Flesch 2000; Jasny 2001, 20-23).
The portfolio bank (or: product bank) provides the function of risk transfor-
mation and manages market risks and credit risks (Flesch 2000). It receives debit
items (savings deposits, bonds, etc.) and sells credits via the sales banks. Portfo-
lio banks specialize in the development of new products and services, and in
portfolio management (Ketterer and Ohmayer 2003).
The primary task of the transaction bank, also called production bank or
processing service provider (PSP), will be to fulfill the tasks of the banking
value chain that follow the sales process and that are repetitive and can be organ-
38
Section 3.4 describes the present state of BPO of core activities in the German banking industry.
39
The term segments (i.e. sets of banking specialists with different business models) should not be
confused with sectors, which describe the current classification of the German universal banking
market into commercial banks, public savings banks, and credit cooperatives.
116 Cooperative Sourcing in the Banking Industry

ized as an industrial production process (N.N. 2003). A transaction bank is the

central provider for processing services such as payments processing, securities
processing and trading, clearing and settlement, custody, credit processing, and
other back-office processes. In the following, these processes will be described
by the term transaction banking40. Transaction banks are often founded by out-
sourcing internal processing units (Ketterer and Ohmayer 2003, cf. section 3.4).
Processing services are mainly repetitive services with large-scale volumes.
These can often be standardized and bundled across different banks (Krichel and
Schwind 2003, 768). Furthermore, because many activities in the processing and
transactions domain do not require the legal form of a bank, the transaction bank
can be substituted by non-bank PSPs. One major reason for this would be a re-
duction in personnel costs since non-banks are not covered by the tariffs which
have to be paid in the banking industry (Bongartz 2004, 50).
Figure 20 shows a mapping of the three segments model to the generic value
chain, introduced in section 3.2.1.

Primary activities

Customer interface Fulfillment

Product
development Branding/ Client Product
Sales Transactions
marketing management administration

Transaction
Portfolio bank Sales bank Portfolio bank
bank

Figure 20: Three segments model, based on (Salmony 2002)

The different segments will be very different in size. Salmony forecasts a
large number of sales banks and a moderate number of portfolio banks but only
very few transaction banks in each particular product domain (Salmony 2002). It
should also be noted that, depending on the particular product, each segment can
be served by a non-bank. For example, product development today is also carried
out by other financial firms, and sales is carried out by independent financial
consultants (e.g. MLP, AWD) or by large commercial retailers (loans, credit
cards, insurances), airlines (credit cards), etc., which often start their own banks
(so-called “non-bank banks”) (Ang et al. 1997).
While the business models of the sales bank and of the transaction bank are
described very well, a point of critique is the diffuse mapping within the area of

40
It should be noted that this term – especially in the German literature – is also used for describing
a particular set of customer services, such as cash management, depot services, credit lines, secu-
rities services etc. These retail services can of course be supported or completely executed by a
transaction bank; nevertheless the term describes a subset of the retail banking business (Lamberti
and Pöhler 2004, 6).
Cooperative Sourcing in the Banking Industry 117

product administration. Although refinancing as the major task of the portfolio

bank is part of product administration, many other parts have not been dealt with
and which segment would cover them is not explained. Today, for example,
some administrative tasks in the lending business are already provided by trans-
action banks (credit factories). The model is too generic (i.e. it does not distin-
guish between a customer and a product perspective) to provide clear boundaries
between the different segments.

3.2.2.2 Four Segments Model of Hamoir et al.

Hamoir et al. expect that four different types of banks will be established in
Europe in the mid term: regional retail distributors, pan-European product spe-
cialists, European and global wholesale banks, and pan-European service provid-
ers (Hamoir et al. 2002, 122).
Figure 21 shows the mapping of the four segments model from Hamoir et al.
to the banking value chain. The representation of the value chain had to be ex-
tended by a customer type dimension.
Primary activities

Customer interface Fulfillment

Product
deve- Client Product Tran-
lopment Marketing Sales manage- administrati- sacti-
ment on ons

Mass
Retail banking

Affluent Pan-
Euro- Regional retail distributors Pan- Pan-
Private pean European Euro-
product product pean
specia- specialists service
Small lists
banking

provi-
SME

ders
Midsize
European and global
Multinational wholesale banks
corporations

Figure 21: Four segments model, based on (Hamoir et al. 2002).

Regional retail distributors function as sales banks for retail and corporate
customers on their respective national markets. Hamoir et al. argue that this kind
118 Cooperative Sourcing in the Banking Industry

of sales bank will not operate at an international level because cross-border

mergers for this bank type will not imply any economies of scale or scope. The
product banks of the four segments model (pan-European product specialists)
operate at a European level and provide particular products or product groups
such as accounts, credits, or brokerage. They offer their products primarily via
regional retail distributors and partially via global wholesale banks. The sales
banks can offer these products under their own label or by using the product
specialist’s brand.
International wholesale banks focus on the business of mid-size and large
corporate customers as well as on institutional investors. They offer the whole
range of corporate banking and investment banking products (loans, IPO, securi-
tization etc.) and develop individual solutions for their customers’ needs.
The pan-European service providers correspond to the transaction banks of
the three segments model. They take on the processing of payments, securities,
custody, etc. The authors of the four segments model believe that only very few
service providers will exist in Europe in the future, each large enough to enable
all possible economies of scale (Hamoir et al. 2002, 124).
The introduction of the customer dimension can be considered an advantage
compared with the three segments model. Nevertheless it is still too generic to be
applied to a particular banking business. In the credit business, for example, the
processes of credit management and refinancing would be combined within the
business model of the product specialist, although more differentiated business
models are thinkable and can be observed in reality.

3.2.2.3 Five Segments Model of Dombret and Kern

Dombret and Kern describe a model for the retail banking domain, which
shows five different business models. In contrast to the previous models, these
are not disjoint. The different models are called product developers, distributors,
administrators, client specialists, and engineers (Dombret and Kern 2003) and
are mapped to the value chain in Figure 22.
The main role of the product developer is to develop new products. Since
the common products in retail banking (such as check account, savings deposits,
or time deposits) have a low level of complexity and are mostly identical be-
tween different banks, product developers will focus on designing complex
products within the investment domain or for tax optimization. Apart from the
financial engineering (determining the product characteristics) they will particu-
larly focus on marketing aspects as part of the product development and on
aligning their products to particular target customer groups.
Cooperative Sourcing in the Banking Industry 119

Primary activities

Client interaction Fulfillment

Product
development Client Product
Marketing Sales Transactions
management administration

Product
developers

Distributors

Administrators

Client specialists

Engineers Engineers
Figure 22: Five segments model, based on (Dombret and Kern 2003)
The distributors correspond to the sales banks of the three segments model
or the regional retail distributors in the four segments model. Distributors con-
centrate on sales of banking products for all or specific customer groups within
retail banking. Administrators combine the business model of a product bank and
of a transaction bank. They offer their integrated product and services package to
the distributors.
By contrast, client specialists are a combination of product developers and
distributors. This business model will be followed particularly by small and
highly specialized banks. Sales banks that follow a niche strategy will tend to
this model because they possess comprehensive knowledge about their custom-
ers and their particular needs and preferences. This can be directly taken into
account for the product development. As final group, engineers are a combina-
tion of product developers and administrators. Since they both develop products
and provide the processing services, they can consider product characteristics
and optimize processes during the product development in order to achieve cost
advantages (Dombret and Kern 2003, 95)
Dombret and Kern’s model only describes the retail business. This restricts
its area of application and does not cover business models which include corpo-
rate customers. Further, there is less detail than in the four segments model.
There is no distinction between product administration and transactions, which
means that mapping business models which cover only one of the two are not
120 Cooperative Sourcing in the Banking Industry

considered in this model. In order to apply the model to the credit business, there
must be a further distinction between refinancing and credit management.

3.2.2.4 Conclusion
Although containing some weaknesses, all segmentation models presented are
appropriate for showing what a future disaggregated banking industry could look
like. Since they try to cover the complete value chain of the banking industry,
they can only give very generic statements and would have to be adapted for
application to a particular business domain (e.g. to the credit business). In this
context, Spiegel mentions the lack of operational relevance and a too limited
granularity of the value chain analysis (Spiegel 2002, 58). Holzhäuser et al. sup-
port this view and argue that the advantages resulting from disintegration should
be investigated at business process level rather than at firm level or generic value
chain level, in order to achieve better and more detailed results (Holzhäuser et al.
2005, 109).
In the following section, we will restrict our view to one particular business
domain – credit processing – and merge the reviewed segmentation models into a
single credit business segmentation model to discuss possible outcomes of the
segmentation trends in this particular area and in order to get a conceptual base
for empirical and simulative research in later chapters.

3.3 Credit Process as Application Domain

In order to have a consistent application base for this research, the credit business
has been chosen as the particular application domain throughout this work. In
this section, it will be briefly introduced with its different products and processes
in order to provide a better understanding of the reasonings in the subsequent
chapters. The credit business was chosen for several reasons:
o dynamics in the credit process market, high awareness of possible BPO
strategies, and antithetical assessments of their potential in the banking in-
dustry
o business process with balanced IT utilization and human interaction
o unrealized process standardization potential
o access to empirical studies available
After giving an overview of the German credit market (3.3.1), reference
processes are developed for three major credit products (3.3.2). Based on these
reference processes and on the segmentation models discussed in the previous
sections, section 3.3.3 develops a segmentation model particular to the credit
Cooperative Sourcing in the Banking Industry 121

business, which allows for a discussion of different sourcing configurations of

the credit process.

3.3.1 Overview of the Credit Market

The market for credits can generally be divided into providing credits for retail
customers and credits for corporate customers, public bodies, and other organiza-
tions. The latter will be handled as corporate credits for simplification reasons.
The following section briefly describes the different credit products.

3.3.1.1 Credit Products in the Retail Customer Business

The main credit products in the retail customer business are open accounts, con-
sumer credits, and private building loans (usually mortgage loans). Further prod-
ucts are aval credits, three-ways-financing (mixture of credit and leasing in the
car sales business), revolving credits, and building society savings credits.
Open accounts or credit lines are short-term41 credits which are mobilized
on a running account (check account) and can be used by the customer as part of
regular payment transactions and without explicit agreements (Sauter 2002, 299-
300). Compared with other credit products, open accounts have the highest inter-
est rates (10% to 18% p.a.) and are – despite their small volume – very attractive
to the offering banks.
Consumer credits or installment loans are highly standardized mid-term or
long-term credits (generally 2-5 years) which have a fixed duration, fixed credit
amount, fixed interest rate, and fixed monthly redemption rates (Sauter 2002,
301-303). They are commonly used for financing private acquisitions (cars,
furnishing, etc.). Compared with the credit line they are more favorable from the
customer’s point of view (interest rates between 7% and 11% p.a.). However,
they are also attractive for the bank, due to higher credit amounts and low proc-
essing efforts, but they also contain a comparatively high level of default risk.
Private building loans and mortgage loans include all credits for building,
buying, or renovating private homes. Similar to consumer credits, building loans
have a fixed duration, fixed credit amount, fixed interest rate, and fixed monthly
redemption rates. Building loans run from 4 to 30 years. In general, the interest
rate will be renegotiated after 10 years (prolongation). Compared with other
credits in the retail customer business, building loans have the lowest interest
rates (5%-8% p.a.). Despite the collaterals and the huge credit amounts, they are
only moderately attractive to banks because the interest margins are very low

41
The German Central Bank classifies credit durations as short-term (up to one year), mid-term (1
to 4 years), and long-term (more than 4 years).
122 Cooperative Sourcing in the Banking Industry

(usually lower than 1%) due to competitive pressure and the very high process-
ing efforts for building loans. The efficient management of building loans is
therefore crucial for ensuring their profitability (Holtmann and Kleinheyer 2002).

3.3.1.2 Credit Products in the Corporate Customer Business

The corporate customer business shows a greater variety of products than the
retail. Despite a very complex product spectrum, three major product groups can
be distinguished: credit lines, revolving credits, and investment loans (Sauter
2002, 473-480). Further financing products are discount credits, factoring, leas-
ing, aval credits, acceptance credits, and capital market-based instruments such
as corporate bonds, conversion bonds, mezzanine capital, and private placements
(Platzer and Riess 2004, 154-156).
Credit lines have the same characteristics as open accounts in the retail cus-
tomer business and represent short-term credits which are commonly used to
ensure a firm’s liquidity. Revolving credits are a mixture of credit lines and in-
vestment credits. The corporate customer gets a special account with a defined
credit line which can be called on demand. In contrast to a normal credit line, the
redemption occurs by means of fixed rates. After the credit has been paid back,
the revolving credit can be called again. Revolving credits are used for satisfying
short-term liquidity demand, but also for small investments.
Investment loans are used to finance the firm’s mid-term and long-term in-
vestments. Similar to the consumer credits or building loans in the retail busi-
ness, they are entered on separate loan accounts, have a fixed credit amount as
well as fixed and periodical repayment rates.
In the corporate credit business, for assuring a credit, all liabilities of the
debtor are always taken into account. Based on this overall picture, the initial
credit decision and ongoing risk monitoring are carried out. The interest rates can
vary a lot between different company sizes, solvency classes, etc.

3.3.1.3 Development of the German Credit Market

In December 2007, the total volume of the German credit market amounted to
€2.27 trillion42. As shown in Figure 23, corporate credits accounted for 55%,
while the remaining 45% were retail customer credits. The latter can be subdi-
vided into €791.6 billion for private building loans, €129.3 billion for consumer
credits, and €17.2 billion for open credits. During the last five years, the overall
credit volume remained rather constant (+2.9%).

42
i.e. million x million
Cooperative Sourcing in the Banking Industry 123

buildung loans
791.6
corporate loans loans to retail
1259.7 cusomers
1015,2
consumer
credit
129.3
others credit
77.1 lines
17.2
Figure 23: Volume of the German credit market (in € billion) (DBB 2008)
Figure 24 shows the market shares of the different sectors for the markets of
building loans and consumer credits. The market for building loans is not only
served by banks but also by insurance carriers and thrift institutions. The frag-
mentation of the banking industry has also had consequences for the credit busi-
ness: even large players do usually not manage more than 250,000 loans, hinder-
ing the realization of substantial economies of scale (Focke et al. 2004, 11).
market shares - private building loans market shares - consumer credits

life 5% others 5% others

insurance 6% foreign banks
carriers

thrift public savings 16%

inst.
9% banks
28%
32% public savings
credit cooperatives
banks
mortgage
13% 15%
banks

15% 40%
credit
20% private banks
cooperatives private banks

note: norisbank has been taken into the sector of private banks instead of credit cooperatives.

Figure 24: Market shares in the loans market (as of 2003) (Focke et al. 2004)
There are several external influences on the loans market which are forcing
banks to react and to develop strategies for optimizing and redesigning the loan
business. Apart from major changes in the regulatory requirements, new com-
petitors such as car banks, banks set up by large retailers, and direct brokers are
taking over market shares in the consumer credit business. Moreover, due to poor
economic conditions and increasing transparency on the banking markets, cus-
124 Cooperative Sourcing in the Banking Industry

tomer loyalty is decreasing (Hertel 2004): retail customers have become very
cost-sensitive while companies are trying to substitute traditional loans by other
financing instruments (leasing, mezzanine capital, etc.). As a result, the interest
margin has continuously decreased in recent years (Koetter et al. 2004).
To counteract these trends, those costs which can be influenced by banks,
are considered sales costs and processing/administration costs (in contrast to risk
costs and equity costs which cannot be reduced by organizational actions). Con-
sequently, banks focus on automating, integrating (straight-through processing –
STP), and modularizing the credit processes as well as on the standardization of
internal modular activities in order to develop an industrialized setting for more
individualized products (Hertel 2004). These re-engineering approaches are often
accompanied by a necessary change of the underlying information systems; this
huge investment leads banks to evaluate outsourcing and insourcing strategies in
the credit business. A modularization of the credit process can lead to demand
for cost-efficient providers of single process steps and helps banks to differenti-
ate their businesses by developing a competitive advantage within one of these
process steps.
Due to their characteristics and volumes, processing and administration of
consumer credits and private building loans especially, but also corporate credits
to SMEs, are candidates for BPO (for our own empirical research see section
3.6). Therefore, the further focus will be restricted to these credit products.

3.3.2 Reference Processes for Process-Based Empirical

Research
In this section, reference business processes for private consumer credits and
building loans as well as corporate (SME) investment loans will be developed, to
be used for conducting process-oriented empirical analyses. The first section will
discuss the regulatory requirements regarding the design of credit processes
which create the initial framework for the generic process design. The second
section presents the various reference processes.

3.3.2.1 MaRisk as a Foundation for Designing Credit Processes

The “Minimum requirements for risk management” (Mindestanforderungen an
das Risikomanagement – MaRisk) (BaFin 2006) address the minimum require-
ments for governance, risk strategy and risk management, employee qualifica-
tion, and business process design of the banking business, formulated by the
German Federal Financial Supervisory Authority (BaFin). After a general intro-
duction, the MaRisk requirements include particular rules for each part of the
Cooperative Sourcing in the Banking Industry 125

banking business. BTO1 comprises the requirements for the credit business
(formerly MaK (BaFin 2002)). The core of BTO1 is formed by the requirement
to functionally separate the credit process into two parts: credit sales (customer
interface, “Markt”) and credit processing and servicing (back office,
“Marktfolge”). Basically, a positive vote from each domain is necessary for
granting a credit43. As will be discussed later on, this has proved to be a signifi-
cant facility for outsourcing in the credit business.
Apart from the general separation of the credit process into sales and proc-
essing, the MaRisk requirements contain a number of regulations concerning the
workflows which have to operate within both parts. Therefore, the MaRisk re-
quirements provide a generic structure of the credit process into the steps of
sales, loan processing, monitoring of loan processing, intensified loan manage-
ment, treatment of problem loans, and risk provision (BaFin 2006; Zanthier and
Gärtner 2003). Table 10 summarizes the MaRisk requirements for the organiza-
tional design of the credit business.
Intensified Treatment of
Loan Risk man- Monitoring of loan
Granting loans loan man- problem
processing agement processing
agement loans
Verify the Monitor that Special Winding up Value Mechanisms for moni-
debtor’s borrow- debtor meets monitoring or restructur- adjustments toring of loan process-
ing capacity and contractual of credit ing of credit Write-off of ing must be imple-
credit-worthiness agreements exposures exposures uncollectible mented (ensuring
Verify and assess Monitor the with high In both cases accounts compliance with
collaterals before loan purpose contingency the bank Forming
organizational guide-
granting risk must develop lines)
Annual re- loan loss
Assess the guar- evaluation of Defining a plan. provisions In particular, monitor-
antee’s sound- counterparty decision The bank ing that loan agreement
ness risk rules for must super- is in line with defined
further vise the decision-making
First (and possi- Periodical re-
treatment restructuring. hierarchy prior to
ble second) vote evaluation of
collaterals Realization granting
of collaterals (may be conducted via
principle of dual
control)
Table 10: Activities of the credit process as required by the MaRisk
(BaFin 2006, section BTO1)
3.3.2.2 Reference Processes for the Credit Business
This section defines reference processes for different products of the credit busi-
ness which underlie the empirical analyses in later sections of this work. The

43
In the standardized retail customer business, a bank’s board can abandon the second vote.
126 Cooperative Sourcing in the Banking Industry

development of all credit processes follows a five-step procedure. Based on

reviewing the literature on generic credit processes, the MaRisk requirements
were incorporated. The result was validated in a discussion with experts from the
banking industry and from consulting firms which operate in the credit business.
Afterwards, this generic process was differentiated into the particular processes
for consumer credits, private building loans, and SME loans. Finally, the result-
ing reference processes were validated and refined by multiple expert interviews.

Reference Process for Consumer Credits

Figure 25 shows the reference process for consumer credits. The process consists
of the subprocesses of product development, sales/preparation, assessment &
decision and processing of the contract and related documents, as well as back-
office activities of administration/servicing, risk monitoring, and workout (if the
loan fails). Certain tasks require interaction with the customer which is done by
the customer interface (either the sales unit or a dedicated service unit). During
the credit decision step, the refinancing of the loan has to be arranged with the
bank’s treasury.
Particular product variants and their justification for specific customer seg-
ments will be developed in the subprocess product development, which is where
the development and refinement of scoring models for automated credit deci-
sions also takes place.
Sales/preparation can be subdivided into the marketing, acquisition, consul-
tation and final offer of the loan including the preparation of the necessary data.
In the middle office, the data is analyzed (proving creditworthiness by a scoring
model) and a final decision (assessment & decision) is made, possibly including
adjustments of the conditions.
A customer deciding to accept the bank’s credit offer will sign the contract
in the next step (processing), the credit account will be created and the credit
amount will be paid off. The treasury of the bank will be notified about the credit
acquisition44 and the credit documents will be archived.
The subprocess administration/servicing includes all administrative activi-
ties that follow the initial granting of a credit, such as credit monitoring (repay-
ments), archiving, reporting, and closure. The risk monitoring observes whether
the customer’s financial situation has deteriorated. Workout handles credits
which do not follow the “usual” process of a credit exposure (dunning, intensi-
fied loan management, recovery, write-down).

44
Since consumer credits always have a fixed credit amount and a fixed duration (in contrast to
open credits), refinancing occurs for every single credit.
Cooperative Sourcing in the Banking Industry 127

Credit risk and portfolio management

develop risk strategy in accordance with MaRisk
develop and manage risk classification algorithms
monitor risk at overall portfolio level (all loans)
set credit risk limits

Sales / preparation Risk

Product Assessment Adminis-
monitoring
develop- & Processing tration / Workout
&
ment Marketing Acquisition
Consulting & decision servicing
offer management
market research select customer address determine check credit collect administer loan monitor process
group potential financing needs standing and customer’s accounts personal risk dunning
financial
(targeting) customers pre-check signature attributes and
engineering: collect customer credit- monitor the business try to
define product data
branding arrange worthiness set up loan repayments situation of the restructure
attributes
meeting (including data account debtor loan
development marketing from Schufa) preterm
and redemption flat-rate value
on site outpayment conduct
advancement of sales support promotions check credit- adjustments of provision on
scoring models closure
worthiness report loan data loan portfolio problem loan
keep legal ( or: by using to treasury process legal
requirements automated cancel loan
reporting
(e.g. BGB) scoring model
customer archiving of
proactively credit files process notify Schufa
keep regulatory automated
uses accounting
requirements credit decision
electronic process
(Basel I+II, (based on provide
channel enforcement
MaRisk, KWG) resulting score) account
p
design pricing acquisition statements
depreciate
model step will be determine risk-
losses
skipped ) adjusted
develop concept conditions
and process (based on closure
documentation resulting score)
according to
MaRisk
test stage
Service / customer interface
accept and process customer requests regarding
account status, collaterals, prolongation, etc.

Refinancing /
treasury
manage refinancing

Figure 25: Reference process for consumer credits

Credit risk and portfolio management is usually provided as a cross-
sectional business function covering all credit processes. Organizationally, it is
placed at the business segment level (retail business and corporate business) or at
the top level of the bank as a whole. Therefore, risk and portfolio management is
identical for all processes described in this section. Their main interaction within
the operational credit process happens with credit analysis & decision and credit
monitoring.
Since consumer credits are granted based on standardized decision models, a
dedicated credit middle office for the analysis & decision activities is not neces-
sary. Instead, these steps are usually provided by the sales department. More-
over, no collaterals have to be dealt with and administered in the consumer credit
process because consumer credits are primarily collateralized by receiving salary
and income payments.
128 Cooperative Sourcing in the Banking Industry

Reference Process for Private Building Loans

Figure 26 shows the reference process for private building loans. The macro-
structure is similar to the reference process above.
Credit risk and portfolio management
develop risk strategy in accordance with MaRisk
develop and manage risk classification algorithms
monitor risk at overall portfolio level (all loans)
set credit risk limits

Sales / preparation Risk

Product Assessment Adminis-
monitoring
develop- & Processing tration/ Workout
Consulting & &
ment Marketing Acquisition decision servicing
offer management
market research select customer address determine verify claim produce and administer loan monitor process
group potential financing needs documents authorize loan accounts personal risk dunning
financial
(targeting) customers contract and attributes and
Engineering: collect customer
check credit- collateral administer the business try to
define product data
branding arrange worthiness contracts collaterals situation of the restructure
attributes
meeting (process rating) (changes, debtor loan
check credit
development marketing set up loan clearing,
standing and
and on site check and account increases) monitor conduct
pre-check
advancement of sales support promotions evaluate creditworthiness provision on
creditworthiness monitor
scoring models collaterals notification (possible re- problem loan
(including data
repayments evaluation)
keep legal ( or from Schufa)
requirements second vote register prolongate loan cancel loan
(e.g. BGB) determine and final mortgage and monitor
customer
contract decision land charge in collaterals
proactively preterm notify Schufa
keep regulatory structure land register (possible re-
uses redemption
requirements (redemption evaluation)
electronic encash and
(Basel I+II, structure,
channel report loan data closure realize
MaRisk, KWG) interest rate,
p to treasury collaterals
collaterals, process legal
design pricing acquisition
duration) reporting
model step will be outpayment flat-rate value process
skipped ) first vote (often in several adjustments of enforcement
develop concept process
tranches) loan portfolio
and process aggregate files accounting
documentation depreciate
and forward to
according to archiving of provide losses
deciders
MaRisk credit files statements
closure
test stage
Service / customer interface
Accept and process customer requests regarding status
of claim processing, outpayment, etc.
accept and process customer requests regarding
account status, collaterals, prolongation, etc.

Refinancing /
treasury
manage refinancing

Figure 26: Reference process for private building loans

While the product development is quite similar to the same subprocess for
consumer credits, the sales and granting of credits organizationally falls into two
parts: sales/preparation is provided by the sales unit while assessment/decision
and processing take place in a dedicated middle office. Within sales, the steps of
marketing and acquisition are again similar to the consumer credit process. The
actual consultation takes place in the consulting & offer step which consists of
determining the financing needs and the contract structure. If the customer ac-
cepts the contract proposal, the customer data is collected (personal data, finan-
Cooperative Sourcing in the Banking Industry 129

cial data, and data regarding the financed object). Based on this collection, an
advisor adequately supported by information systems which provide a first rating
can give a first vote within the same meeting. If the first vote is positive, the
proposal and the collected information will be transferred to the middle-office,
along with the contract which may already have been signed by the customer.
The analysis & decision step examines the customer’s credit worthiness and
the object to be financed in more detail. Based on this analysis, the middle-office
passes a second vote. If it is positive, the credit can be granted to the customer.
In processing, the contract is finalized and the credit amount is paid out (some-
times in multiple tranches). Although the MaRisk requirements do not insist on
more than one vote in the private building loan segment, the reference process
must take a first and second vote. The advisor in the front-office is able to get a
reasonably clear picture of the customer’s credit worthiness and of the financed
object, but due to competence advantages in the middle-office, administrative
tasks such as a structured evaluation of the collaterals can be provided there
much more efficiently.
The subprocesses of refinancing and administration/back office are structur-
ally equivalent to the homonymous consumer credit subprocesses. Since building
loans are collateralized by real estate, the administration must additionally man-
age the collaterals. Moreover, financing of buildings regularly leads to extremely
long financing durations (20-30 years) which lead to a prolongation of credits
(contract renewal after expiring interest binding).

Reference Process for SME Credits

The reference process for SME credits, shown in Figure 27, is quite similar to the
private building loans process. In fact, many banks do at least handle credit re-
quests from small corporate customers such as investment loans from retail cus-
tomers.
Once again, sales and proposal preparation includes all consultation meet-
ings with the customer and his or her preparation of all relevant data. In the
credit assessment and decision subprocesses, the SME is rated with reference to
risk classification and credit conditions. Afterwards, the second vote is passed,
which leads to the final decision. In the next step (processing), all administrative
operations are carried out, including an initial data archiving, the authorization of
the contract and the outpayment. As in the building loans process, administration
covers all following activities (data collection, repayments monitoring, prolonga-
tion, closure, etc.). All periodical data required for observing the SME’s credit-
worthiness and risk classification over time (periodic repetition of the rating) is
collected in this step. Controlled by the bank’s overall credit risk management,
risk management monitors the risks related to the several exposures. It analyzes
130 Cooperative Sourcing in the Banking Industry

the data that is periodically provided by the corporate customers. A proactive

risk management not only analyzes the bank’s risk portfolio but also feeds rele-
vant data back to the sales subprocess, where credits may be sold only if they are
in line with the current risk situation and the risk strategy of the bank as a whole
(e.g. applications from particular high-risk SME segments are accepted only up
to a certain volume). If the credit exposure is endangered, the credit documents
are transferred to the workout subprocess, which will deal more intensively with
the credit and try to realize a successful reverse transaction but in negative cases
will also exploit the collaterals.
Credit risk and portfolio management
develop risk strategy in accordance with MaRisk
develop and manage risk classification algorithms
monitor risk at overall portfolio level (all loans)
set credit risk limits

Sales / preparation Risk

Product Assessment Adminis-
monitoring
develop- & Processing tration / Workout
&
ment Consulting & decision servicing
Marketing Acquisition
offer management

market research select customer address determine verify claim produce and administer loan monitor process
group potential financing needs documents authorize loan accounts personal risk dunning
financial
(targeting) customers contract and attributes and
engineering: collect customer
check credit- collateral administer the business try to
define product data
branding arrange worthiness contracts collaterals situation of the restructure
attributes
meeting (process rating) (changes, debtor loan
check credit
development marketing set up loan clearing,
standing and
and on site check and account increases) monitor conduct
pre-check
advancement of sales support promotions evaluate creditworthiness provision on
creditworthiness
rating models monitor (periodic rating)
(including data collaterals notification problem loan
repayments
keep legal ( or from Schufa)
requirements second vote report loan data prolongate loan monitor cancel loan
product choice and final to treasury collaterals
(e.g. BGB) customer
proactively decision preterm (possible re- notify Schufa
keep regulatory determine
uses outpayment redemption evaluation)
requirements contract
electronic structure (possibly in encash and
(Basel I+II,
channel (redemption several closure realize
MaRisk, KWG)
p structure, tranches) collaterals
design pricing acquisition interest rate, process legal flat-rate value
model step will be collaterals, archiving of reporting adjustments of process
skipped ) duration) credit files loan portfolio enforcement
develop concept process
and process first vote accounting
documentation depreciate
according to aggregate files provide losses
MaRisk and forward to statements
deciders closure
test stage
Service /customer interface
accept and process customer requests regarding status
of claim processing, outpayment, etc.
accept and process customer requests regarding
account status, collaterals, prolongation, etc.
Refinancing/treasury
manage refinancing and equity
of the bank (Basel II)

Figure 27: Reference process for SME credits

Cooperative Sourcing in the Banking Industry 131

3.3.3 Credit Business Segmentation Model

The segmentation models presented in section 3.2.2 show a number of shortcom-
ings. In order to overcome some of these, this section presents a segmentation
model particular to the credit business domain. The macrostructure of the refer-
ence credit processes developed in section 3.3.2.2 will serve as basis for the
development.

3.3.3.1 Model
The credit business segmentation model defines ten different business models,
which can be combined in five different ways (A to E) (Figure 28, next page).
Scenario A shows the lowest complexity and essentially consists only of the
fully integrated bank. It represents the most common business model, today.
Only within product development or workout, might the bank acquire external
providers. Developers can design new credit product variants and advanced scor-
ing models (e.g. consulting companies). The workout specialist supports banks in
the workout step of collecting or enforcing delinquent or omitted credits. Today,
these steps are often provided by lawyers or collection firms45.
Scenario B consists of two different business models: the processing out-
sourcer and the processing service provider (PSP or “credit factory”). The first
outsources the credit processing to the PSP. The outsourcer does the marketing
and the sales as well as the refinancing (credit is part of the outsourcer’s balance
sheet), while the PSP takes on any administrative or processing steps. There is
close collaboration in the analysis & decision. The processing outsourcer, bear-
ing the actual credit risk, will usually also carry out the decision steps of granting
a new credit as well as prolongating an existing credit. Alternatively, the PSP
may make the decision, based on the outsourcer’s first vote and on predeter-
mined decision rules (guidelines and scoring model). This scenario has become
quite common in recent years (e.g. Aareal Bank, Hypo Real Estate, GMAC-RFC,
et al. & Kreditwerk HM; or Lloyds TSB & EDS).
Scenario C also consists of two business models: the branding & sales spe-
cialist and the credit product bank (white label). The first develops the products
(closely together with the product bank which will provide the product) and does
the marketing and sales. The credit is also branded by the branding & sales spe-
cialist. The credit product bank provides all back-office processes and the refi-
nancing. It has the role of a “grey eminence” (Dombret and Kern 2003) because
it holds the credit and will usually take the second vote and the final decision

45
Since this is the same for all scenarios (A – E), we will ignore the developer and the workout
specialist in the following scenarios.
132 Cooperative Sourcing in the Banking Industry

because it bears the credit risk. Thus far, this combination is not known to be
existent in practice.

Sales / preparation Assessment

Product Refinancing/ Admin. Risk
& Processing Workout
development Consulting treasury / servicing monitoring
Marketing Acquisition decision
& offer

Fully integrated bank

A alt. alt.

Developer Workout
specialist

Processing outsourcer Processing outsourcer

PSP PSP

B alt. alt.

Workout
Developer specialist

Branding & sales specialist

Credit product bank (white label)

C alt. alt.
Workout
Developer specialist

Distributor
Credit product bank
Credit product bank (branded)
(branded)

alt. alt.
D PSP PSP

alt. alt.

Workout
Developer specialist

Branding & sales specialist

PSP PSP

E alt.
Portfolio bank
alt.

Workout
Developer specialist

Figure 28: Credit business segmentation model

Scenario D consists basically of two business models as well, the credit
product bank (branded) and the distributor. The credit product bank again pro-
vides the processing activities as well as the financing. However, compared with
the credit product bank (white label) from scenario C, this product bank develops
its own products, branding, and marketing. The distributor follows the role of a
“pure” intermediary which sells the product bank’s products and is paid on
commission basis, for example. A major advantage of this scenario is the sim-
Cooperative Sourcing in the Banking Industry 133

plicity of the segmentation because the distributor just sells the credit to the cus-
tomer and passes on any other tasks to the credit product bank. This enables the
implementation of simple and cost-efficient process interfaces between the part-
ners. In the simplest case, the sales staff may just complete a paper or web-based
form and send the necessary documents to the credit product bank. Since the
distributor does not interact with the customer during the credit contract, there is
no need for further system integration. On the other hand, there is an incentive
problem which must be overcome. Not all relevant information about the credit-
worthiness of a customer can be documented by structured data. In our own case
studies, interview partners from credit sales departments perpetually emphasized
that “feelings and instincts” proved to be crucial for a successful evaluation. If
the distributor only receives a commission for a successful contract closure and
does not share the credit risk (e.g. by a compensation mechanism), there will be
no incentive to reconsider “soft” information.
Scenario D may involve variations, such as incorporating a PSP that pro-
vides the back-office functions. This combination can be found in reality: for
example, several banks (distributors) sell easyCredit consumer credits from
Norisbank (credit product bank (branded)), while eC-Factory (PSP, subsidiary of
Norisbank) does the processing and the administration. Another example is the
sales of KfW development loans by credit cooperatives (distributors) which are
processed by VR Kreditwerk (PSP).
A further variant (not displayed) would be bundling credits and either issu-
ing them as asset-backed securities on the capital market – as is already common
practice – or selling them to other banks or financial investors (e.g. Lonestar or
Fortress), i.e. outsourcing of refinancing (not displayed in Figure 28).
Scenario E is a combination of C and D and represents the “classical” three
segments model. Refinancing is provided by a portfolio bank which issues the
loans. The remaining parts (processing and administration) are done by a PSP.
This scenario will primarily result from traditional fully integrated banks out-
sourcing everything except sales in order to become pure sales banks (branding
& sales specialists). Initiatives of public savings banks to establish joint credit
factories and to partly transfer the refinancing to the state banks lead to this type
of credit process configuration.
As mentioned in the discussion of the different segmentation models (sec-
tion 3.2.2), it is easy to imagine that PSPs, credit product banks of different
types, and portfolio banks will, in the main, specialize on credit products. A PSP
which provides securities or payments processing in addition to the credit proc-
essing seems implausible because it cannot achieve substantial economies of
scope. However, within the credit business, most PSPs will not limit themselves
to one credit type (e.g. mortgages) but try to offer their services to a broader
134 Cooperative Sourcing in the Banking Industry

range of credit types (e.g. all types of retail credits and SME loans) to stabilize
their market position. Existing credit factories commonly broaden their portfolio
of credit processing services over time, mostly starting with mortgage loans.
By contrast, sales banks, such as processing outsourcers, branding & sales
specialists, and distributors, will certainly not only provide credit sales but fol-
low a holistic strategy and offer other products like insurances, brokerage, etc.

3.3.3.2 Banking Supervision Requirements

As explained in section 1.5.4, the German Banking Act (KWG) distinguishes
between different forms of financial firms, each with different supervision re-
quirements. Table 11 shows the mapping of the different business models intro-
duced in the credit process segmentation model to these different legal types.
The X mark indicates the supervision minimum requirement. The level of re-
quirements decreases from the left to the right46.
Financial
Credit institu- Financial Ancillary banking
services
tion (bank) enterprise services enterprise
institution Others
(KWG, sec- (KWG, (KWG, section
(KWG, section
tion 1(1)) section 1(3)) 1(3c))
1(1a))
Fully integrated bank X
Processing outsourcer X
Processing service
X
provider (PSP)
Branding & sales
(X) (X)
specialist
Credit product bank
X
(white-label)
Credit product bank
X
(branded)
Portfolio bank X
Distributor (X) (X)
Developer X
Workout specialist X
Table 11: Minimum requirements for the different business models
(based on (Ade and Moormann 2004))

46
Example: A fully integrated bank always has to be a credit institution according to the KWG
definition – other forms are not possible. A developer firm is not required to be a financial firm,
but of course it may be a bank, financial service institution etc.
Cooperative Sourcing in the Banking Industry 135

Players who do the refinancing must be credit institutions because refinanc-

ing represents the core of the credit business (KWG, section 1 (1)) (Ade and
Moormann 2004, 163).
Processing service providers do not conduct banking business (as defined by
KWG, section 1 (1)), do not provide financial services (as defined by KWG,
section 1 (1a)) and do not carry out the activities of a financial enterprise (as
defined by KWG, section 1 (3)). As a result, PSPs are not required to be credit
institutions, financial services institutions, or financial enterprises. However,
since credit processing is an ancillary activity as defined by KWG, section 1
(3c), PSPs must be supervised as ancillary banking services enterprises (Ade and
Moormann 2004, 163-164).
Developers and workout specialists do not match any of the KWG defini-
tions. They can be classified as “other companies”. The branding & sales special-
ist and the distributor cannot be classified according to one single definition.
Since credits do not belong to the group of financial instruments as defined by
the KWG, both business models are “other companies” as long as they only
arrange loans. As this business model is rather unrealistic as discussed above and
both business models would normally also cover the sales of financial assets etc.,
both business models will have to be classified as financial service providers.
The pure distributor business model – as long as it is restricted to credit sales –
can virtually be adopted by almost every company, especially by near-banks
(insurance companies, finance brokers etc.), but also by non-banks (mail-order
firms, estate agents, etc.).

3.3.3.3 Effect of Bank Size and Sector Membership

From an economic perspective, the size of a bank essentially influences its future
positioning in a segmented credit process and its adoption of one (or more) of the
described business models. Bösch assumes that due to higher unit costs, smaller
banks, in particular, will become dependent of processing providers (Bösch
1999, 24). Large banks have high process volumes in all parts of the retail credit
business and therefore can realize most of the cost effects in-house. Up to now,
all credit factories in Germany administer a smaller number of credit contracts
than the big private banks. As a result, large banks will possibly keep their credit
processing in-house (scenario A). Some of the big banks actually evaluated the
business case of outsourcing the back office (scenario B) but found that no sub-
stantial cost savings were realizable47. Due to VAT and high costs for personnel
transfer and reduction, in-house processing has shown to be cost-efficient. Ow-
ing to their good credit rating, big banks can refinance themselves on good con-

47
Our own expert interviews. Cf. section 3.6, data source “EI”.
136 Cooperative Sourcing in the Banking Industry

ditions so that outsourcing the refinancing never becomes advantageous. Finally,

the role of a pure branding & sales specialist or of a distributor would be absurd
for a big bank.
For small banks, the arguments can be reversed. Smaller banks often lack a
sufficient number of credits to boost significant cost advantages from in-house
process optimization. Smaller process volumes lead to a relatively higher volume
variability which cannot be absorbed. Therefore, BPO will become an increas-
ingly desirable option for optimizing their credit business. Apart from scenario
B, scenario D, in particular, would be very interesting. Outsourcing the credit
processing to a PSP (scenario E) still generates costs for coordination and im-
plementation of interfaces. If a small bank adopted the role of a distributor, the
interface problems would be significantly reduced. Furthermore, the problem of
VAT can be avoided (cf. section 3.5.1.3). Today, PSPs are often still not in a
position which allows them to adopt the process volume of small banks. Each
new client firm requires some system adaptation, even if it accepts the PSP’s
reference process completely. If a small bank has only a small number of credits
to be administered, the transaction costs for the PSP would be too high.
The different sectors vary in their internal market structure. The credit coop-
erative and public savings bank sectors include a significantly higher proportion
of small banks. Therefore, the distributor business model might be more favored
in these sectors, at least for credits with comparatively high default risks such as
consumer credits, than in the private bank sector as discussed in section 3.3.3.3.
In smaller banks, outsourcing is often circumvented for “emotional” reasons.
Public savings banks and credit cooperatives see all parts of the credit business
as their core competence. Therefore, BPO is a very sensitive topic in the credit
business. In addition, for smaller banks, the problem of personnel transfer or
reduction is much more complicated than in big banks because they are so
strongly embedded in their various regions.
In Germany, the structure of a future banking value network is significantly
influenced by the three-sectors structure. Presently, it is noticeable that credit
factories are often formed in and by one particular sector, but the borders are
fading. For example, in 2006, the only credit factory in the credit cooperatives
sector (VR Kreditwerk) purchased the largest privately owned credit factory
(Aareal HM). Section 3.4.3 gives some overview of the current situation of the
credit BPO market. Public savings banks and credit cooperatives in particular
will outsource processing and maybe refinancing activities to banks of the same
type. Likewise, private banks experience difficulties with outsourcing their proc-
esses to PSPs from one of the other sectors.
Cooperative Sourcing in the Banking Industry 137

3.4 Cooperative Sourcing in the German Banking

Industry
3.4.1 General Trends
Although outsourcing of peripheral elements of a firm (e.g. security, facility
management, cleaning, and canteen management) has been common for decades,
outsourcing really took off with the advent of the first big deals in IT outsourc-
ing. Despite some activities in the 70s and 80s, its popularity was dramatically
increased by the Kodak deal, when Kodak outsourced its complete IT business to
IBM in 1989. This event triggered a huge bandwagon effect of subsequent deals.
Even the finance industry now has its own outsourcing tradition, although most
of the large deals have been done within the last four years. In 2002, outsourcing
contracts with a value of almost $33 billion were signed in the banking industry
(Gellrich 2004), mainly accounted for by six international mega ITO deals,
which are listed in Table 12 (next page).
A pan-European survey conducted by IBM showed that 22% of the partici-
pating banks had outsourced their IT business (IBM 2003). A recent survey of
the E-Finance Lab48 with the 1,000 largest banks in Germany showed that 84%
of them have outsourced major parts of their IT services49. Most cooperatives
and savings banks use data processing centers and core applications from the IT
units of their respective associations. For example, FinanzIT and Sparkassen
Informatik50, as the IT providers of the German Savings Banks and Giro Asso-
ciation (DSGV), serve most of the savings banks, while Fiducia, VR Netze,
GAD, and SDV provide IT services to most of the credit cooperatives in Ger-
many. With commercial banks, the picture is more heterogeneous. The largest
German bank (Deutsche Bank, cf. Table 12) outsourced its IT unit to IBM in
2002/3 – constituting the first mega ITO deal in Germany – while Commerz-
bank51, at this time, evaluated any IT outsourcing options as inefficient: the CIO
stated that outsourcing would only lead to operational cost savings of 4%, which
would be greatly exceeded by forthcoming taxes and coordination costs (Froh-
müller 2005) (as of 2005). Nevertheless, in Oct 2007 he redecided and out-

48
Cf. footnote 1 on p. 11.
49
See section 3.6 for more detailed results and information about the study.
50
FinanzIT and Sparkassen Informatik are currently discussing about merging their businesses
(state: Feb 2008).
51
Fourth largest German bank in total assets by end of 2004.
138 Cooperative Sourcing in the Banking Industry

sourced complete desktop services and particular infrastructure services to Hew-

lett Packard52.
Contract
Com- Sourcing Geographi- Duration
value Deal mechanics
pany provider cal scope (years)
(€ bill.)
Outsourcing of most of the technology infra-
structure including data centers, help desks,
JP
distributed computing and voice networks
Morgan 5.0 IBM Worldwide 7
Transfer of 4,000 employees to IBM
Chase53
Create a virtual pool of computing resources
Leverage supplier’s intellectual property
Outsourcing of voice and data network
Transfer of 1,000 FTEs to EDS
Bank of Establishment of a one-stop shop for voice and
4.5 EDS Worldwide 10
America data services, re-design and implementation of
solutions to optimize Bank of America’s optical
network, provision of help desk support
Outsourcing of the IT technology infrastructure
Ameri-
Transfer of 2,000 employees to IBM
can 4.0 IBM Worldwide 7
Granting AMEX access to IBM’s computing
Express
resources
Outsourcing of data centers and smaller server
Deutsche sites
2.5 IBM EMEA54 10
Bank Transfer of 900 employees to IBM
Establishment of a new data center
Outsourcing of IT/I including desktop, mission
critical systems, software, midrange servers,
and networking gear
CIBC 1.5 HP Canada 7
HP also provide technology related procure-
ment, asset management, and IT vendor man-
agement services
Outsourcing of technology services and appli-
ABN-
1.3 EDS Worldwide 5 cation development in the wholesale client
Amro
strategic business unit
Table 12: “Mega ITO deals” in the international banking industry (Klein
2004)
On the vendor side, the IT outsourcing market is dominated by very few and
very large international IT insourcers. In Germany, the biggest four providers
serve 80% of the ITO market (not restricted to the banking industry) (Schaaf
2004): T-Systems, Siemens Business Services, IBM, and EDS. Worldwide, the

52
Source: http://www.cio.de/financeit/aktuelles/843570/ (as of 20 Feb 2008).
53
The JPMorganChase-IBM deal was canceled and rolled back in 2004 (JPMorganChase 2004).
54
Europe, the Middle East, and Africa
Cooperative Sourcing in the Banking Industry 139

largest players are Accenture, CSC, EDS, IBM, ACS, and HP – called the “Big
Six” by TPI55.
From IT outsourcing it is a small step to outsourcing of secondary processes
such as HR management, procurement and secondary F&A processes (e.g. in-
voicing, claiming, etc.). Since these internal administrative processes usually do
not represent specific business competencies of the outsourcing firm and since
information systems have become more process-oriented, leading to activities
being increasingly transferred to the information systems (and therefore mostly
to the IT sourcing providers), the IT sourcing vendors have started to offer the
processing of whole business functions. For example, in 2004 Accenture in-
sourced the invoice management, procurement, and parts of the HR administra-
tion of Deutsche Bank (Müller 2005).
In the USA, large banks have shown a higher adoption rate of outsourcing
secondary processes than comparably large firms of other industries. In 2003,
36% of the large banks had outsourced secondary processes such as HR and
F&A, while overall it had been only 20.6% ((Scholl 2003), cited in (Dayasindhu
2004, 3479)).
After the success of secondary process outsourcing, the trend has developed
further to outsourcing parts of the banking value chain, i.e. outsourcing of pri-
mary processes. The IBM survey (IBM 2003) showed that operational parts of
the banking business are outsourced by many banks across Europe. The main
operational functions to have been outsourced so far are custody (36%), trad-
ing/execution (23%), settlement (23%), and securities processing (22%) (as of
2003).
The German banking industry is focusing more on outsourcing primary
processes than on outsourcing secondary processes. Due to competitive pressure
(cf. section 3.1.1), major changes of the regulatory requirements (cf. section 3.5),
and high IT intensity in banking processes, close partnerships between banks and
their IT providers have led to the re-engineering of core banking processes.
Banks which are about to undergo this major change in a particular business
often seek to amortize their investments faster by insourcing process volumes of
other banks. For example, the small private bank HSBC Trinkaus & Burkhardt
(TuB), together with its IT provider T-Systems, founded a subsidiary (Interna-
tional Transactions Services – ITS) and developed a completely new securities
processing system which allowed them to insource other banks’ securities proc-
essing volumes and to become one of the largest securities processing providers
in Germany56.
55
TPI offers the quarterly TPI outsourcing index, reflecting the current state of the global outsourc-
ing market. http://www.tpi.net/knowledgecenter/tpiindex/ (as of 02 Feb 2008).
56
Source: http://www.sds.at/files/downloads/HSBC_T-Systems.pdf (as of 20 Feb 2008).
140 Cooperative Sourcing in the Banking Industry

In other cases, banks have decided to source their processes cooperatively to

get the critical mass for realizing particular projects and reducing the investment
risks or just to achieve cost savings from economies of scale. This strategy is
often used by the large commercial banks which – after attempts to completely
merge have failed – cooperatively source major parts of their business. One ex-
ample is the merger of the mortgage business of Deutsche Bank, Dresdner Bank,
and Commerzbank (Krabichler and Krauß 2003, 28), leading to EuroHypo, or
the consolidation of domestic payments processing of Deutsche Bank, Dresdner
Bank, and Deutsche Postbank.
The association of public savings banks in Germany decided to follow a co-
operative sourcing strategy, which offers joint product development and transac-
tion banking activities at the state bank level. The goal is to have only one insti-
tution for executing each task (Krabichler and Krauß 2003, 29).
As shown in the previous sections, the banking industry will necessarily be-
come more segmented, with banks disintegrating their value chain and focusing
on a particular business (Hoppenstedt 2000; Lacity et al. 1996, 13; Petzel 2003;
Rampl 2003). Additional reasons, such as volatile transaction volumes, the need
for expensive technological advancement, and increasing regulatory require-
ments (Basel II, MaRisk, SOX, etc.) caused many banks to outsource processing
activities and focusing on sales (Middendorf and Göttlicher 2003, 4).
Although outsourcing of primary processes is in a premature phase in Ger-
many, 60% of German banking executives believe it can be a (highly) effective
instrument (Herrmann 2004). A questionnaire-based survey by Fraunhofer IAO
shows that BPO was the second most important strategic focus of the German
banking industry in 2005 (Engstler and Vocke 2004)57.
As the examples of outsourcing of primary processes show, banks must al-
ways consider whether they should either become an insourcer for a particular
task and increase process volume or whether they should instead outsource it to
another bank (or become partners in a joint subsidiary) (Aubert et al. 1996a).
Therefore, outsourcing of primary processes usually follows our definition of
cooperative sourcing; banks themselves are the “natural” insourcer for primary
banking processes. Nevertheless, it should be mentioned that IT outsourcing
providers – which are very experienced in providing IT services to banks – have
the opportunity to develop expertise in particular banking processes and will play
a role in the emerging sourcing landscape of the banking industry, despite being
market-external entities (Focke et al. 2004, 5; Marlière 2004a). For example,
“EDS is the fifth largest mortgage processor in the world” (Fairchild 2003).

57
68.3% of the participating banks marked it as an important strategic field of activity. Process
optimization was regarded as most important (72.4%), while increasing the cross-selling ratio was
considered the third most important strategic field of activity (49.0%).
Cooperative Sourcing in the Banking Industry 141

3.4.2 Outsourcing of Particular Business Processes

This section gives a brief overview of BPO tendencies in selected domains of the
banking business. The most common business processes which are outsourced
by German banks are payments and securities processing. Funds and the credit
business as well as outsourcing of typical secondary processes such as HR and
F&A are still in a rather immature state.
Figure 29 from an A.T. Kearney survey on transaction banking shows the
diffusion of outsourcing primary banking processes in Germany over time.
(Please note that the ordinate is not to scale.)
diffusion rate of
cooperative sourcing (selected cases)
negotiations: merger:
transfer:
etb-Dreba, etb-FMS WPS-bws
Dreba-dwp
50-100% formations: to dwp securities
TxB, Setis, Plusbank, LB BW processing
merger: & administration
TxB-
formation: FMS Plusbank merger:
acquisition: DS + SSG
HVB paym. to DSGF
formations: etb,
20-50% acquisition: by SZB several payments
bws, WPS negotiations
DB & Dreba paym.
processing
by Postbank
formation:
TAI funds depot
formation: ebase, several services
formation: FSB, (Coba) negotations
formation:
5-20% (HVB/MEAG)
Fondsdepotbank
negotiations: e.g.
(dit) mortgage loans
formation: etb/Dreba/FMS HM: acquisition services
formations: HVB Payments small HRE-Retail
etb, ZVS & Services acquisition: mortgage trans- loans
consumer credits
loans for several actions
formations: insurances and brokers services
formations: VR Kreditwerk,
0-5% Stater, Depfa, HM Prompter DZ: acquisition VR Bank: account services
Norisbank White Labeling
funds
SSK & KSK Köln administration
1999 2001 2003 2004 2005 time

Figure 29: BPO diffusion path of different banking business functions (Source:
A.T. Kearney transaction banking survey (Focke et al. 2004, 4))
The authors of this survey argue that 10% of the German banks’ administra-
tive costs are related to transaction banking and that outsourcing has resulted in
up to 50% savings in the administrative costs. Such an opportunity explains the
substantial progress of the securities and payments businesses, in particular. But,
although the picture shows increasing diffusion rates for all parts of the banking
business shown, it cannot be assumed that there is simply a time lag between the
already consolidated market segments and the more recent additions to the BPO
142 Cooperative Sourcing in the Banking Industry

market, e.g. the credit business. These activities vary strongly in terms of degree
of automation, interaction needs, and strategic impact.
Securities processing is the most widely developed BPO market segment in
the banking industry58. Transaction banks provide more than 60% of the domes-
tic retail securities processing volume and offer both processing services and
depot administration services. Within the last ten years, a number of players have
emerged and have strengthened in a subsequent consolidation phase. Transaction
banks primarily offer services in the business of standardized listed securities
and tend not to support institutional trading.
Table 13 gives an overview of the actual market structure in this market
segment. dwpbank, the merger result of WPS Bank and bws Bank, is the market
leader for both processing and depot administration, followed by ITS and
Xchanging Transaction Bank (cf. Table 13). WPS, bws, and etb started their
transaction business in 1999. WPS was founded by several state banks while bws
was created by the large banks from the cooperatives sector. Both banks had
already served a few banks from the commercial sector, but the merging of both
banks in 2003 and the forming of the dwpbank (Deutsche WertpapierService
Bank AG) resulted in one of the largest examples of inter-sectoral cooperative
sourcing. Moreover, in 2007 dwpbank purchased txb, which still is operating on
its own but contributes to the market dominating position of dwpbank.
Also founded in 1999 (by Deutsche Bank), etb started to offer payments and
securities processing. While the insourcing of the payments processing of
Dresdner Bank and HVB failed after year-long negotiations (Fehr and Mussler
2003), the securities business was successfully established and also offered to
other sectors: in September 2002, NetBank and all Sparda banks outsourced their
securities processing to etb (Fehr 2002). To increase the attractiveness of etb
services to third parties, who feared the dominant role of Deutsche Bank, etb was
transferred to Xchanging, a large British processing provider, in 2004 and re-
named “Xchanging Transaction Bank” in 2006.

58
It has to be noted that, although many analysts talk about the diffusion of outsourcing, Figure 29
as well as Table 13 and Table 14 describe not only the outsourced volumes but also include the
processing volumes of the insourcer. For this reason, the term “cooperative sourcing” is used in
Figure 29.
Cooperative Sourcing in the Banking Industry 143

Share of
Share of
Used IT system process-
Insourcer Major clients Founded in securities
or vendor ing
accounts
volume
Mandators from WPS: 150 2003 by merg-
dwpbank (Deut- public savings banks, ing the transac-
sche Wertpa- NordLB, WestLB, SaarLB, tion banks of
pierService- Bremer LB, Deutsche
several coopera- WP2 (provided
Bank) Postbank. Mandators from 36.3% 21.6%
bws: DZ Bank, tive central by FinanzIT)
(merger of WPS
GZB Bank, SGZ Bank, banks and of
Bank and bws
WGZ Bank,Dt. Verkehrs- several state
Bank)
bank. New: Dresdner Bank banks
2004 by Bay-
TxB-Plus Bank
ernLB, Landes-
Purchased by 200 public savings banks
bank Hessen- WIS Plus 8.5% 3.3%
dwpbank in and smaller private banks
Thüringen, HSH
2007
Nordbank
HSBC Trinkaus &
International Burkardt (TuB), Sparkas- 2005 by HSBC GEOS, pro-
Transaction sen Broker, DAB Bank, TuB and T- vided by T- 2.7% 15.0%
Services (ITS) Fimatex, FondsService- Systems Systems
Bank
1999 by
Xchanging
Deutsche Bank, Sparda Deutsche Bank,
Transaction Euroengine2 +
banks, Sal Oppenheim transferred to 13.6% 13.7%
Bank (formerly FORSS
NetBank, Citibank Xchanging in
etb)
2004
ACTIS
Financial Mar- 2000 by Pro- PABA/Q
HypoVereinsbank,
kets Service bank and Hy- provided by 3.8% 10.1%
Vereins- u. Westbank
Bank (FMSB) poVereinsbank ACTIS.BSP
Services
60.7%
64.9%
(100% =
Total share of the processing (100% =
183m
volume in Germany ca. 20m
transac-
accounts)
tions)
Table 13: Major securities processing insourcers in Germany (state of market
shares: 06/2006) (data from public company information sources)
Figure 30 visualizes the consolidation path of the securities processing mar-
ket in Germany.
144 Cooperative Sourcing in the Banking Industry

Institutions 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Bayern LB

Sourc.
Bayern LB

Coop.
LB Transaktionsb.

Merger
LB HT LB HT TxB Trans-
HSH Nordbank Plus B. Plus aktionsbank

Sourc.
HSH Nordbank

Coop.
Bank

Dwp purchases TxB

Norisbank Norisbank

Dresdner Bank dwp

Dresdner Bank

Sourc.
Coop.
bank
(with
West LB, LB RP, LB SH WPS

Sourc.
Coop.
WPS TxB)
Deutsche Postbank Deutsche Postbank
Deutsche Wertpapier
DZ Bank, Südwestdt. GZ, BWS Service Bank (dwpbank)

Sourc.
Coop.
Westdt. GZ, GZ Stuttgart Bank für Wertpapierservice
und –Systeme (BWS)
Citibank Citibank

Ci
tib
an
k
Deutsche Bank European Transaction Bank (etb)

Sourc.
Coop.
etb Xchanging TB (xtb)
Sparda banks Sparda banks

Probank
Sourc.
Coop.

DAB Financial Markets Service Bank (FMSB)

20 f H
HVB HVB

07 VB
o
:p
DA

ar
B
ITS

ts
HSBC Trinkaus Trinkaus

Figure 30: Consolidation path of the German securities services market

(selected institutions, structure based on (Eichelmann 2004))
A possible extension of the securities processing market might be the han-
dling of equity funds. Today, the market of funds handling is very concentrated:
Deka Bank, Union Investment, and DWS cover two-thirds of the German funds
market but do not offer their services to other banks. Since the large processing
volumes have already been consolidated, there is not much growth potential for
pure funds transaction banks such as ebase, FondsDepotBank, FondsService-
Bank, and Frankfurter Fondsbank. Therefore, analysts expect a converging trend
of securities processing and an increase in the market potential of funds process-
ing (Focke et al. 2004, 14).
Another part of the banking business where the “breakthrough”59 of the BPO
market has already appeared is the market for payments transactions – document
processing as well as domestic clearing (Focke et al. 2004, 10; Marlière 2004a).
Many providers have specialized their business to particular processing steps but
there are also some players who offer the full spectrum of services. One of the
primary examples is the transfer of payments processing from Deutsche Bank,
Dresdner Bank (in 2004), and HVB (in 2006) to Deutsche Postbank BCB, which
now handles around 7.2 billion payment transactions per year60. Further, TAI
AG, a subsidiary of DZ Bank as the main competitor in the payments business,

59
Defined by consulting firms as a market share of more than 30% of the total domestic processing
volume being cooperatively sourced (Focke et al. 2004)
60
Data from the provider website
Cooperative Sourcing in the Banking Industry 145

recently (11/06) merged with Interpay Nederland B.V. to form a new corporation
named Equens which now represents the first pan-European payments process-
ing provider. Table 14 gives an overview of the payments processing market in
Germany.
Used IT system Share of do-
Insourcer Major clients Founded in for payments mestic process-
processing ing volume
Deutsche Bank,
(Insourcing since SAP Payment
Deutsche Postbank Dresdner Bank, 20%
2004) Engine
HVB
Equens (former DZ Bank, Citibank, 2003 by DZ Bank
GPayS (Mosaic
Transaktionsinstitut, 1,100 cooperatives 2006 merged with 16%
Geva)
TAI) and others Interpay Nederland
Deutsche Service- 120 public savings
2006 by merger of
gesellschaft für banks (15 from SSG,
SSG Köln and DS N/A approx. 13%
Finanzdienstleister 31 from DS, >50 61
Dresden
(DSGF) from SZB), WestLB
Bankgesellschaft
1998 by Bankge- EBS 2000 (Beta
Bankenservice Berlin, multiple N/A
sellschaft Berlin Systems)
savings banks
Bankservice- Ca. 25 public savings
2000 by Fraspa
gesellschaft Rhein- banks, LRP, 2 credit N/A under 1%
and Naspa
Main (bsg) cooperatives
over 50%
Total share of the
(100% = 15.9
German process-
billion domestic
ing volume
clearing items)
Table 14: Major payment processing insourcers in Germany (State of market
shares: 12/2005) (data from public company information sources,
Wernthaler 2004, DBB 2006)
Analysts expect an ongoing consolidation trend in the German payments
processing market, ultimately resulting in three or four national providers (Ei-
chelmann et al. 2004). Figure 31 gives an overview of previous consolidation
activities in this market segment.

61
Source: http://www5.rsgv.de/static/0F020048_.pdf and http://www.dsgf.de/ (as of 02 Jan 2006).
146 Cooperative Sourcing in the Banking Industry

Institutions 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

DZ Bank DZ Bank - Business Unit TAI

Sourc.
Coop.
TAI equens
Citibank Citibank

SSK Köln

Sourc.
SSK

Coop.

Coop. Sourcing
Merger
Sparkassen-Service-Gesellschaft (SSG)
KSK Köln KSK DSGF
DS
Ostsächsische Spk DS Dresdner Sparkassenservice GF
Bayern LB Bayern LB & several savings banks Servicezentrum Bayern

Sourc.
Coop.
Servicezentrum
Bayern
HypoVereinsbank HypoVereinsbank HVB P&S

H
VB
Dresdner Bank DreBa - Business Unit Global TxB ZVS

Sourc.
Coop.
Postbank
Postbank Postbank PoBa BCB
BCB
Deutsche Bank European Transaction Bank (etb) DB Payments
Deutsche Bank
Merger

BB-Bankenservice BB
BankenService der Landesbank Berlin
LBB Betriebsservice LBB

Fraspa Fraspa
Coop.
Sourc.

Bankservicegesellschaft Rhein-Main
Naspa Naspa

Figure 31: Consolidation path of the German payments processing market

(selected institutions, structure based on (Eichelmann 2004))
After giving an overview of two of the most important cooperative sourcing
segments in the German banking industry, the next section will focus on sourc-
ing activities in the credit business.

3.4.3 Outsourcing of Credit Processes

As already indicated in Figure 29 above, outsourcing or cooperative sourcing of
credit processes is still not very common on the German banking landscape. The
private mortgage loans business is the precursor in this domain. Although experts
believe that there are unit cost differences of about 300% between different
banks (Focke et al. 2004, 11), outsourcing is still not a major trend and there are
only very few credit factories – as transaction banks are usually called in this
domain62 – active in the German market.
The consolidation of the German loans industry is very far behind that of
other nations. In the USA, a large loans servicing industry has emerged over the
last 25 years. Apart from credit factories (“primary servicers”), there are special-
ists for the handling of problem loans. Moreover, master servicers control the
complete credit processing and administration across primary servicers and spe-
cial servicers, provide backup capacities, and report to refinancers and investors
62
Credit factories are defined as transaction banks or processors which focus on the credit business.
Their services portfolio usually includes parts of the loan granting (check of creditworthiness,
check of collaterals, documentation, outpayment), loan processing and administration, permanent
monitoring, and handling of non-performing loans (Ade and Moormann 2004, 158).
Cooperative Sourcing in the Banking Industry 147

(Pieske 2005). The servicing market is dominated by large banks such as Citi-
bank and Wells Fargo, as well as by specialized providers (e.g. Owen, Midland
Loan Services, GMAC). The leading servicer in the private home loans business,
Countrywide, manages a total loans volume of $915 billion; 65% of all Ameri-
can mortgage loans are managed by servicers (Pieske 2005). Another example is
the Netherlands, where 40% of all mortgage loans are managed by service pro-
viders.
There are many reasons for the lack of activity in the German market. One
reason is that there is no inter-bank coordination to create common standards in
the credit business; another is the lack of evidence of relative cost superiority
(Focke et al. 2004, 11). Some banks evaluated the benefits of outsourcing their
loans processing, but did not get attractive offers by the existing credit factories.
A further “historic” reason is that many German banks consider mortgaging, in
particular, to be their core competence. Outsourcing parts of this business is not a
plausible approach for German banks, which have had a monolithic firm struc-
ture up to now. At present, the majority of the German credit factories’ custom-
ers are insurance companies and new market members such as foreign retail
banks which strictly focus on sales (e.g. GE Money Bank and GMAC-RFC)
(Focke et al. 2004, 12).
Compared with the processing of payments or securities, where transaction
banks already cover large segments of the market volume, there are three main
differences to the processing of loans, which makes it difficult to draw analogies.
First, in payments and securities processing, there is a very high degree of auto-
mation. The cost structure mainly consists of fixed costs, which enables strong
economies of scale from bundling transaction volumes. Second, since payments
and securities transfers predominantly occur between different banks, there is
much more standardization of processes and formats in these areas (Bongartz
2004) which in turn facilitates BPO. The development of the US payments proc-
essing market during recent decades impressively showed the impact of process
and data format standardization (a short review can be found in (Bongartz
2004)). Therefore, bundling processes from different banks seems to be com-
paratively easy. The third – related – reason is integration needs. The credit busi-
ness consists of making and communicating more or less complex decisions,
which is not the case in processing payments or securities. In the credit business,
real-time integration between the outsourcer and the service provider must be
realized and an extensive service level management (SLM) might have to be
implemented (Focke et al. 2004, 6; Krichel and Schwind 2003, 768-769). On the
other hand, credit factories already claim to provide this kind of flexible integra-
tion as well as modular services which can selectively be embedded within the
client’s credit process. The following figure shows different possible configura-
148 Cooperative Sourcing in the Banking Industry

tions of labor division between an outsourcer and a credit factory along the
mortgage loan granting process as they are offered by one of today’s major credit
factories today.
example: mortgage loans processing

outsourcer credit factory

collect customer data

check credit standing

enter data

check creditworthiness

evaluate property

granting decision

produce loan contract

check collaterals

outpayment

cover booking

Figure 32: Different examples for labor division between outsourcer and credit
factory along the mortgage granting process, as offered by a
German credit factory (Hertel 2004; Aareal 2005)
Apart from outsourcing the processing or administration, an international
comparison also shows other differences which might be responsible for a lack
of BPO in other parts of the credit process, such as sales or refinancing. The
latter is typically done by German banks themselves. By contrast, in Anglo-
Saxon markets it is usually done on the capital market while the loan is managed
by a credit factory (Focke et al. 2004, 12). Similarly, the sales of mortgages are
traditionally done by the German bank itself while only 20% are mediated by
brokers. In the Netherlands, brokers mediate around 60% of mortgages (Focke et
al. 2004, 12). Banks typically do not want to transfer the responsibility for direct
customer contact in this field, in contrast to the securities business or payments
processing. As argued in section 3.3.3.1 (credit business segmentation model),
outsourcing of sales implies agency conflicts, and banks have had bad experi-
ences with mortgage intermediation which significantly increased the number of
bad loans63.

63
Result of one of our own case studies, which is partially documented in (Wagner et al. 2006,
Beimborn et al. 2007a), cf. section 3.6.
Cooperative Sourcing in the Banking Industry 149

Consequently, there is no likelihood of a breakthrough in BPO of credit

processes for quite some time (Focke et al. 2004). Nevertheless, a basic hypothe-
sis is that competitive pressure will increasingly force banks to cooperate in the
core domains of their business.
Based on observations of the Anglo-Saxon markets, A.T. Kearney assumed
that if a processing provider can get two major players on his platform and proc-
ess at least 400,000-500,000 loans, the critical efficiency threshold would be
reached and a diffusion trend would be started (Focke et al. 2004, 12). By con-
trast, the market shares of credit factories in Germany together with their muted
expectations of future market growth (see below) indicate that this assumption
might not hold. Table 15 provides an overview of all credit factories operating in
the German market in 2006, including product portfolio, corporate information,
and clients. The last row but one gives the numbers of credit contracts managed
by the providers.
West- Nord-
Proceed Kredit- BHW VR Credit-
Stater deu. deu.
Portfolio werk Kredit- Kredit- plus
Germany Immo.- Retail-
Services HM center werk Bank
Bank Service
Consumer Consumer
loans, Consumer Mort- Mort- loans,
Mortages, Con-
Supported mortgages, loans, gages, gages, mort-
Mortgages corporate sumer
products corporate mortgages, corporate corporate gages,
loans loans
credit lines corporate loans loans corporate
and loans loans
02/99 by
07/00 by 1995 by
1998 as Aareal. In 07/06 by
DG Hyp 1960 by WestLB,
Founded in 01/97 by GFKL 01/06 HaSpa
N/A and BSpk Crédit LB RP
… Stater N.V. Portfolio transferred and Spk
Schwäbis Agricole and LB
Services to VR Bremen
ch-Hall BW
Kreditwerk
Size (FTE) 2,600 500 1000
450 N/A 250 N/A 370
ca. (10/07) (12/05) (07/06)
Number of approx.
1.5 million ca.
adminis- N/A N/A 215,000
8 million
N/A N/A N/A
tered loans (12/05)
e.g. ABN
9 clients, e. g. Allg.
Amro, NordLB
e.g. Aareal, Hyp Rhein- 12 credit Several
Münchner BSpk. Several
AXA, boden, BHW, coopera- savings
Clients Hypobank, N/A
GMAC- DEVK, KfW, tives
Mainz, savings
banks in
Hypobank Bremer banks
RFC, Hypo all PSD banks (2005) S-H
Essen, DBV, LB
Real Estate (12/07)
Argenta
Table 15: Service offers for different credit products by German credit
factories (state: 12/06, unless stated elsewise) (data from Hertel
2004; Krawietz et al. 2003; Lehmann 2005, provider websites).
150 Cooperative Sourcing in the Banking Industry

All of the service providers focus on the mortgage loans business (BaFin
2003, 2; Krawietz et al. 2003, 12); some of them additionally focus on consumer
credits or corporate investment loans. Proceed Portfolio Services GmbH is a
special case because it primarily specializes on the handling of non-performing
loans (NPL), i. e. liquidation of bad loans, evaluation, re-bundling, and resale of
NPL portfolios (Krawietz et al. 2003, 12).
Unisys conducted a survey of credit factories in the German market and in-
terviewed all players (Lehmann 2005). No common IT platform has yet been
established. Some providers developed their own systems while five of them use
SAP products (CML, CRM, FI/CO), extended to their particular needs.
The results of the study also show that hardly any of the actors expect strong
increases in the number of mandators within the next five years. Although most
of them had only a few clients (average: 9.3, range: 2–14), they did not believe
that this number will be more than doubled by 2010 (Lehmann 2005). The parent
companies and small retail banks (often new market entrants) currently represent
the overwhelming part of the credit factories’ clients. In order to reduce average
costs by economies of scale and to improve their market position, the partici-
pants in the study plan to extend their business to other European countries by
2007 (Lehmann 2005).
Furthermore, possible strategies for reducing average costs would not only
increase economies of scale, but also realize economies of scope. Almost all
credit factories intend to extend their original product portfolio from only serving
mortgage loans to also processing and administering consumer credits and parts
of the corporate loans business (e.g. corporate building loans or investment loans
to SMEs) (Lehmann 2005). Today, consumer loans are processed by universal
banks themselves or by retail product specialists who carry out processing, refi-
nancing, and pricing but not sales (i.e. credit product banks in the sense of the
credit business segmentation model (cf. section 3.3.3.1)) (Holzhäuser et al.
2005). These large providers (e.g. Citibank, GE Moneybank, Norisbank,
Santander Consumer CC-Bank) dominate this rather small market segment,
which, due to its highly standardized and automated business, is suited for bun-
dling and realizing economies of scale. Nevertheless, because the providers have
already realized the critical mass in-house, they do feel pressured into opening
their processing infrastructure to third parties.
Another possible market for generating economies of scope would be incor-
porating the processing of corporate loans. Compared with retail banking loans,
corporate loans are more individual and less standardized between different
banks. Presently, only three credit factories offer services to the corporate loans
business. For example, Aareal HM started offering services to the corporate
building loans segment (Krawietz et al. 2003, 13), which was recently extended
Cooperative Sourcing in the Banking Industry 151

to other kinds of corporate loans. More than 60% of the participating banks of a
FORSA survey64 agreed that there is huge automation and standardization poten-
tial in the corporate loans business (Mummert 2005, 6). In our own study, we
showed that 33.7% of the participating institutes believe that outsourcing the
processing of SME loans would be an efficient strategy (cf. section 3.6.3.1).
In August 2004, the German Federal Government tried to start an initiative
for a “national” corporate (SME) loans factory, together with the publicly owned
KfW (Kreditanstalt für Wiederaufbau / Reconstruction Loan Corporation). The
idea was to standardize small corporate loans – which were evaluated as too
expensive to administer – in order to stimulate the SME sector by cheaper in-
vestment opportunities. The majority of both the large commercial banks and the
public savings banks disagreed with this idea (Rettig 2004), with only Dresdner
Bank signaling a willingness to join the project (N.N. 2004). Managers from DZ
Bank (the largest bank in the cooperatives sector), which was also interested,
stated that it would be very difficult to standardize SME loans (N.N. 2005b).
As in the payments and securities processing market, some of the providers
introduced were formed by cooperatively sourcing the relevant business units of
different banks. For example, in 2002, Deutsche Bank, Dresdner Bank, and
Commerzbank founded the EuroHypo by bundling their mortgage business (resp.
their mortgage bank subsidiaries) (Holzhäuser et al. 2005). This was followed by
EuroHypo outsourcing the processing parts of its business by founding Prompter.
Today, Prompter has been reintegrated and EuroHypo is completely owned by
Commerzbank (N.N. 2005c)65.
In the public savings bank sector, there are several regional activities driven
by the largest players (i.e. G866) to establish credit factories in four different
regions in Germany. The earliest was Norddeutsche RetailService AG which was
founded by Hamburger Sparkasse and Bremer Sparkasse in 2006. Prior to this,
Hamburger Sparkasse, one of the world’s largest savings banks, established an
internal centralized credit center with highly standardized and automated proc-
esses, which administered 260,000 loans of all types (including credit lines,
private mortgage loans and corporate investment loans) (Rösemeier 2005). An-
other credit factory is currently being established by Stadtsparkasse Köln and

64
38 public savings banks, 34 credit cooperatives and 28 private banks + thrift institutions (n=100)
were asked about the automation and standardization potential within their institution. In all sec-
tors, around 65% of the participants agreed that there is potential in the corporate loans business
(Mummert 2005, 6).
65
At present, there is no information about the future of EuroHypo as a stand-alone institute.
66
Eight largest public savings banks in Germany: Hamburger Stadtsparkasse, Sparkasse Bremen,
Stadtsparkasse Hannover, Stadtsparkasse + Kreissparkasse Köln, Nassauische Sparkasse, Frank-
furter Sparkasse, Münchner Stadtsparkasse.
152 Cooperative Sourcing in the Banking Industry

Kreissparkasse Köln and a third one by Nassauische Sparkasse and Frankfurter

Sparkasse.
In the first step towards consolidation in the credit processing market,
Aareal HM was acquired by VR Kreditwerk in 2006 (N.N. 2005a) and now op-
erates under the name “Kreditwerk Hypotheken-Management”.

3.5 Regulatory Issues

Legal and regulatory issues are of great significance for outsourcing potential in
banking processes. Compared with other industries, banks are much more regu-
lated and controlled by legislatory and supervisory authorities. The most relevant
legal controls for outsourcing in the German banking industry are sections 6 and
25a KWG67 as well as BaFin68 Circular 11/2001 (Ketterer and Ohmayer 2003, 9).
The following sections will first discuss general legal conditions for BPO in
banking and subsequently focus on particular issues regarding the outsourcing of
parts of the credit business.

3.5.1 General Requirements Related to BPO

3.5.1.1 Section 25a of KWG and BaFin Circular 11/2001
The legal foundation of outsourcing is given in section 25a (2) of KWG, which
explicitly governs the outsourcing of major parts of the banking business. The
application of this paragraph is “limited to outsourcing solutions relating to
banking business or financial services requiring a license pursuant to section 1
(1) sentence 2 or (1a) sentence 2 [of KWG]” (BaFin 2001a)69.
BaFin Circular 11/2001 represents a flexible and liberal embodiment of sec-
tion 25a (2). This extension of the regulatory framework enables banks to make
use of the cost saving potential of outsourcing in order to ensure their competi-
tiveness (BaFin 2001b). Basically, section 25a (2) allows the outsourcing of all
business activities provided that the following aspects are ensured, independently
of whether the sourcing provider is affiliated or from outside the group (BaFin
2001a, IV.12):

67
German Banking Act (Kreditwesengesetz)
68
Federal Financial Supervisory Authority (Bundesanstalt für Finanzdienstleistungsaufsicht),
formerly Bundesaufsichtsamt für das Kreditwesen (Federal Banking Supervisory Office),
www.bafin.de.
69
If other business functions are affected by outsourcing, section 25a (1) KWG must be taken into
account (Lehnsdorf and Schneider 2002). This is particularly relevant for business activities listed
in section 1 (3) of KWG (leasing, factoring, financial advising, etc.) (Frank 2004), cf. section
1.5.4.
Cooperative Sourcing in the Banking Industry 153

o The banking business is conducted in an orderly manner.

o The managers are able to manage and monitor the business (and thus take
responsibility for the outsourced business function).
o The supervisory authorities have auditing rights and access to oversee the
business.
In any case, the bank’s core management functions must remain the execu-
tives’ responsibility. Therefore, these activities (corporate planning, organiza-
tion, management and control) generally cannot be outsourced (BaFin 2001a,
IV.13). Furthermore, the total of the outsourced operational areas must not ex-
ceed the areas remaining in-house in terms of size and importance (BaFin 2001a,
IV.17).
Outsourcing of decision-making activities will only be possible if the man-
agement of the outsourcing firm retains control of all business risks by imple-
menting appropriate organizational governance structures and control proce-
dures. This can only happen if decision-making can be based completely on
evaluation and decision criteria defined ex ante, which can be incorporated into
the outsourcing contract and which conform to the existing internal decision
rules (section 25 (2) KWG).
In order to fulfill all regulatory requirements stemming from the BaFin Cir-
cular, the following principles must be ensured in addition to KWG, section 25a
(BaFin 2001a, section V): (1) qualitative and quantitative service level require-
ments must be defined and measurable (defining service level agreements), (2)
responsibilities and interfaces must be explicitly determined and documented, (3)
internal and external auditing units must be granted access to all relevant areas
within the insourcer firm, especially if the activities are outsourced to another
country, (4) in the case of service debasement, the outsourcer must prepare alter-
native (backup) solutions, and (5) sufficiently flexible cancellation rights must be
negotiated.
In the USA, the Sarbanes-Oxley Act (SOX) (USA 2002) – the mandatory
guideline for business reporting – contains similar requirements relating to the
management and control of business processes derived from KWG, section 25a
(1). For German firms listed on a US exchange, section 404 of SOX requires
internal documentation and control of all business processes as well as the im-
plementation of control systems. Section 302 (a) of SOX defines the responsibili-
ties of executives involved in business reporting. Through outsourcing, these
duties will be extended to the sourcing provider’s processes, provided that the
outsourced activities are relevant for their internal controls (Lamberti 2005, 520).
The main goal of SOX is to improve business reporting. BPO usually affects
services which can have an impact on the outsourcer’s business reports. Thus, if
154 Cooperative Sourcing in the Banking Industry

processes are not clearly defined, documented and monitored, bias in reporting
becomes possible (Mensik 2004).

3.5.1.2 Joint Forum – Outsourcing in Financial Services

The Basel Committee on Banking Supervision, which formed the Joint Forum
together with the International Organization of Securities Commissions (IOSCO)
and the International Association of Insurance Supervisors (IAIS) in 1996, sees
its main goal in analyzing trans-sectoral problems from the banking, insurance,
and securities business. For example, the committee published general principles
which support firms and national supervisory authorities to minimize risks in-
volved in outsourcing. The published paper “Outsourcing in Financial Services”
(BIS 2005) provides principles and guidelines, which should be considered by all
financial firms when outsourcing their business functions. The paper suggests the
definition of comprehensive outsourcing policies, for evaluating outsourced
activities and for risk management programs, which allow for a permanent moni-
toring of outsourced (and especially of the more complex) business processes.
This is balanced by the supervisory authorities being urged to monitor the sys-
temic risk from increasing consolidation of particular parts of the whole national
banking industry (BIS 2005, 14-19).

3.5.1.3 Basel II
In 2004, the Basel Committee on Banking Supervision published the final version
of the new Basel equity standards (Basel II, BIS 2004). The main goal of Basel II
is to strengthen the stability of the international financial system, to be achieved
by a better consideration of the economic situation of debtors and by a more risk-
appropriate determination of banks’ capital requirements. With the introduction
of Basel II, the solvency of credit users has become directly relevant for deter-
mining the equity needs of the lending bank. Moreover, apart from credit risks
and market risks, operational risks will now have to be included in the assess-
ment of the bank’s capital requirements.
Operational risk is defined as “the risk of loss resulting from inadequate or
failed internal processes, people and systems or from external events” (BIS 2003,
8). Examples are process risks, legal risks, technical risks, but not strategic or
reputation risks (BIS 2003, Rebouillon and Matheis 2004, 347). For outsourcing,
this implies that the sourcing provider must operate its processes exactly as de-
fined by the outsourcing bank (Dittrich and Braun 2004, 61). In addition to the
extended risk coverage by equity, Basel II contains a multitude of statutory re-
porting requirements which should ensure a transparent representation of risk
management (of credit risks, market risks and operational risks).
Cooperative Sourcing in the Banking Industry 155

3.5.1.4 Particular Legal Domains

Labor Legislation
From a labor law perspective, section 613a of BGB70 has a high importance for
outsourcing. This section defines the rights and responsibilities associated with a
transfer of ownership. When a business (or part of a business) is handed over to
another owner, the latter must take on all the rights and duties associated with the
employment contracts of the transferred employees (BGB, section 613a). There-
fore, for any intended outsourcing deal, the question of whether it represents a
transfer of ownership must be clarified71 (Mahr 2004). In case of a transfer of
ownership, the new owner must continue to fulfill individual agreements (em-
ployment contracts including supplementary grants, vacation entitlements, re-
tirement provisions) and collective arrangements (works committee, works
council agreements, labor contracts, etc.). Thus, the negative influence on the
advantageousness of an outsourcing agreement can be significant (Simon 2004).
In actual practice, the potential insourcer firm will usually already have the
necessary resources (HR and IT) and the partners will therefore try to avoid a
transfer of ownership and its consequences. This can be brought about by break-
ing up the identity of the affected business unit, by integrating it into a com-
pletely new organizational structure or by temporarily closing it down (Mahr
2004).

Contract Law
The comprehensive supervisory guidelines concerning outsourcing contracts and
the contractual requirements for allowed outsourcing activities are given in Cir-
cular 11/2001 (BaFin 2001a, cf. section 3.5.1.1). The outsourcing partners can
either design an all-embracing agreement, which governs all aspects of the busi-
ness to be outsourced, or they can agree on a general framework agreement
which is supplemented by detailed and modular service level agreements (SLAs)
(Wullenkord et al. 2005, 139-141). The particular service level descriptions and
quality requirements are attached to the contract (Schrey 2004, 349-350). SLAs
are legally binding agreements and lead to sanctions and penalties in the case of
non-fulfillment. If designed properly, SLAs are an appropriate instrument to
ensure the quality of outsourced services (Cullen and Willcocks 2003). Finally,
the framework agreement is supplemented by price and volume schedules.

70
German Civil Code (Bürgerliches Gesetzbuch)
71
A transfer of ownership exists when a business unit retains its original identity after being trans-
ferred to the insourcer firm (Clever 2004, 227). In contrast, a succession in function exists when
no resources, personnel or customer bases are adopted. For more information, see (Mahr 2004).
156 Cooperative Sourcing in the Banking Industry

Merger Control and Antitrust Law

Large outsourcing deals sometimes represent a major consolidation of an indus-
try’s activities and are therefore subject to antitrust provisions (GWB, section 37
(1.2)72). Basically, agreements which restrain competition are forbidden (GWB,
section 1). Whether a merger, acquisition, or outsourcing deal is relevant to anti-
trust regulations depends on the defined thresholds which have to be reached
(Schrey 2004)73 and which take into account the past revenue of the participating
firms (including subsidiaries) and of the business unit to be outsourced. Since
intra-company sales are often difficult to quantify in cases of outsourcing, the
contract volume is often taken instead (Schrey 2004).

Tax Law (Value-Added Tax74)

An outsourcing project consists of two phases which must be viewed separately
from a tax law perspective (Söbbing 2002): (1) the outsourcing process and (2)
the continuous taxation of subsequent externally procured services.
In the first phase, material and immaterial assets (and sometimes employees)
are transferred from the outsourcer to the insourcer firm. This gives rise to fiscal
effects for the insourcer, in particular. The asset transfer can result in uncovering
hidden reserves and to the capitalization of immaterial assets which then lead to
increased value-added tax (VAT) as well as to a singular increase of the corpo-
rate income taxation base. Therefore, the participating firms will try to avoid any
disclosure (Söbbing 2002, 337-338).
The form of outsourcing is essential for the periodical taxation during the
second phase. In contrast to non-banks, where the procurement of external ser-
vices is unproblematic75, VAT is a major problem for banks involved in out-
sourcing. Banking products and services are usually not charged with VAT
(UStG76, section 4 (8)). Consequently, deduction of input VAT is not possible
(UStG, section 15 (2.1)), resulting in the fact that VAT on externally procured
services causes costs for the outsourcer. Thus, when outsourcing an internal

72
Act against Restraints on Competition (Gesetz gegen Wettbewerbsbeschränkungen)
73
The European Merger Control Regulation (FKVO) becomes relevant if worldwide consolidated
revenue exceeds €5 billion p.a. and if at least two participating organizations (the insourcer and
the outsourced business unit of the outsourcer) each achieve more than €250 million of joint
revenue (FKVO, section 1 (2)). The German merger control regulation has lower threshold values
(GWB, section 37).
74
Apart from the VAT problem, there are a number of other tax related questions regarding corpo-
ration taxes, trade taxes, property transfer taxes, etc. which are not discussed in this work.
75
If the insourcer and outsourcer firms are based in the same country.
76
Turnover Tax Act (Umsatzsteuergesetz)
Cooperative Sourcing in the Banking Industry 157

business unit, services which cannot justifiably be labeled as banking services

might then be priced, in Germany, with 19% VAT.
On this account, the service providers are keen to offer their services portfo-
lio in a way that is exempt from VAT. For example, the ECJ77 gave a ruling on
the VAT exemption of electronic data processing services by SDC within the
payments processing of several Danish public savings banks78. Similarly, the
securities processing offered by CSC to different client banks is not subject to
VAT79. In the UK, the Supreme Court gave a ruling on the tax exemption of a
retail credit process outsourcing deal between Lloyds TSB Bank and EDS80. The
main reason for the outcome of the latter ruling was that EDS has the main re-
sponsibility for granting credits and will induce any legal and financial changes
(Menner 2004). By contrast, the German Federal Ministry of Finance does not
agree with the jurisprudence of the ECJ and refuses to grant tax exemption for
services from transaction banks and data processing centers (Menner 2004). The
European Commission is currently conducting a consultation process on the
modernization of VAT liabilities for financial services (EC 2006), which ad-
dresses this problem and could lead to a harmonization of VAT handling in BPO
in the European Union.
Another way to avoid the VAT problem is the creation of an affiliation
structure. In this case, the sourcing partners found a new firm which provides the
outsourced services (Jorczyk 2004).

Privacy
Outsourcing of parts of the banking business is usually connected with granting
the insourcer access to sensitive data, such as customer information and bank
account data. The BDSG81 allows the processing and storing of customer-
specific data only if the individual has explicitly agreed to it (BDSG, section 4
(1)). If business processes are outsourced, all parties must ensure the confidenti-
ality, integrity and privacy of the data. The security requirements must be part of
the outsourcing contract and the outsourcer firm must continuously monitor the
provided services regarding security and privacy issues (BDSG, section 4 (1)). In
addition to these security and privacy issues, section 25 (2) of KWG and Circular

77
European Court of Justice (ECJ), Brussels
78
ECJ, case: RS. C-2/95 = ECJ ruling 1997, I-3017 (1997-06-05).
79
ECJ, case: RS. C-235/00 (2001-12-13).
80
Supreme Court of Judicature, London, Case no. C3 2002 109, 2003-04-19. http://www.hmcourts-
service.gov.uk/judgmentsfiles/j1707/cce_v_electronic_data_systems.htm (as of 19 Jul 2006).
81
German Federal Data Protection Act (Bundesdatenschutzgesetz)
158 Cooperative Sourcing in the Banking Industry

11/2001 (BaFin 2001a) require a contractually ensured compliance with rules

governing banking and business secrets.

3.5.2 Specific Requirements for Credit Process Outsourcing

3.5.2.1 MaRisk – Minimum Requirements for Risk Management
If a credit institution offers credit products, its business must comply both with
the Minimum requirements for risk management (MaRisk) (BaFin 2006) in gen-
eral and with the specific requirements for the credit business (BaFin 2006, sec-
tion BTO1). Section BTO182 of the MaRisk defines the minimum requirements
for both the structural and process organization of the credit business (cf. section
3.3.2.1 for a process-oriented presentation). Adequate organization as well as
effective monitoring of the credit risks must be ensured (Grill and Perczynski
2004, 349). If parts of the credit business are outsourced, the partners must en-
sure that the MaRisk are fulfilled (Bausch et al. 2004, 49).
Compared to Basel II, the MaRisk contain more detailed and specific re-
quirements for particular bank-internal processes (cf. section 3.3.2.1). Outsourc-
ing is mentioned in the general part (BaFin 2006, AT 9), but the paragraph only
refers to section 25a (2) of KWG and Circular 11/2001 (BaFin 2001a) (Anger-
müller et al. 2005).

3.5.2.2 BaFin Memorandum on Credit Factories

In December 2003, the BaFin issued a memorandum which addresses the in-
creasing tendency to outsource parts of the credit process and concretized the
terms of Circular 11/2001 (BaFin 2001a) with regard to credit factories (BaFin
2003). The memo states that outsourcing of credit processing and servicing to a
credit factory is basically possible within the meaning of KWG, section 25a (2).
However, it involves significant risks, for which the outsourcing bank’s execu-
tives are responsible (BaFin 2003).
When outsourcing parts of the credit process to a credit factory, it is impor-
tant to know who decides on the granting of the credit. If the bank itself grants
the credit and outsources only the processing and servicing, this is permitted by
section 25a (1+2) of KWG. If the credit granting decision is to be outsourced as
well (in terms of the service provider taking over the role of a proxy), then pre-
cise, objective decision criteria must be formulated, which must not allow for
any decision alternatives at all (BaFin 2003, section I). Thus, the decision made

82
formerly MaK – Minimum requirements for the credit business (BaFin 2002), now integrated into
MaRisk.
Cooperative Sourcing in the Banking Industry 159

by the credit factory has the same legal value as a decision of the bank itself
would have. This is almost impossible to realize outside the standardized credit
business. For large retail loans and corporate loans, the bank itself is usually
responsible for deciding the granting of a credit. However, the credit factory can
make the necessary preparations for the decision (ratings, collection of data, etc.)
(BaFin 2003, section II).

3.5.2.3 Requirements for Credit Process Outsourcing

This section summarizes the concrete regulatory terms for the example of out-
sourcing parts of the SME credit process (cf. section 3.3.2.2).
In sales, consulting and intermediation activities constitute risks that are
relevant from a supervisory perspective (acquisition of customer data implies
operational risks). Thus, sales is an activity covered by KWG, section 25a (2).
By contrast, mere agency activities of a customer adviser do not fall into the
application domain of Basel II or section 25a (2) of KWG (Ade and Moormann
2004, 166). Outsourcing is therefore possible because no fundamental corporate
decisions are required in this process step.
After the credit proposal has been prepared, the MaRisk require two inde-
pendent granting decisions in the sales unit and in the back office (first and sec-
ond vote) (BaFin 2006, section BTO1). If the decision can be made solely based
on a rating system (rating based on objective measures (e.g. business measures)
and on qualitative assessments by the sales staff), an automated credit granting
decision might be possible at least for standardized smaller credit products in the
SME business. In this case, it is possible to outsource the decision task (BaFin
2003, section III c).
Tasks before and after the credit decision, such as checking creditworthi-
ness, evaluating collaterals, documentation and outpayment, are not subject to
section 25a (2) of KWG and may therefore be outsourced to a credit factory (Ade
and Moormann 2004, 166-169).
The risk management process involves the management of the overall risk of
the bank’s credit portfolio. According to Basel II, the bank must implement an
internal credit risk controlling unit, which is responsible for the internal rating
system of the overall bank (BIS 2004, 85). The executive board is responsible for
the correct implementation of the credit risk strategy and cannot delegate it.
Credit risk management can therefore not be outsourced (BaFin 2005). By con-
trast, the operational risk monitoring of single exposures can be outsourced as
long as it can completely be carried out by applying ex ante defined risk classifi-
cation criteria (BaFin 2005; BaFin 2003, 2-3; Szivek 2004, 57-58).
160 Cooperative Sourcing in the Banking Industry

Investigative and consulting services in the workout subprocess are not sub-
ject to section 25a (2) of KWG if they represent only supporting and advisory
tasks (Theewen 2004, 109-110). In this case, outsourcing is unproblematic.
To summarize, it is possible, in principle, to outsource major parts of the
credit business and this was significantly facilitated by the explications of the
MaRisk. Nevertheless, the regulations require that banks make significant in-
vestments in the handling of risks. Efforts to transfer ownership and the VAT
problem further reduce the economic advantageousness of credit process out-
sourcing.

3.6 Empirical Evidence in the German Credit

Business
This section presents our own empirical research on BPO in the banking in-
dustry. Based on empirical studies of German banks, the status quo and potential
of credit process outsourcing is analyzed. BPO drivers and inhibitors, discussed
in the theory chapter (esp. section 2.2.2), are reflected against empirical data.
The aim of this section is twofold. First, it empirically analyzes the rele-
vance of BPO in general and cooperative sourcing in particular for the credit
business of the German banking industry. Second, the data will be used to feed
the parameterization of a model on cooperative sourcing developed in chapter 4
with as much realistic and complete data as possible for the subsequent simula-
tion studies in chapter 5. Different empirical projects of the Institute of Informa-
tion Systems and Cluster 1 of the E-Finance Lab at Goethe University in Frank-
furt/Main conducted questionnaire-based surveys and case studies. These data
sources are used in the following:
o Source S1: E-Finance Lab (EFL) Survey 2004 (“Credit Process Manage-
ment”):
S1 focused on process efficiency, optimization potentials, and business
process outsourcing opportunities in the SME loans business of the German
banking industry. Based on a reference credit process (which has been re-
duced in complexity compared to the reference process introduced in section
3.3.2.2 for practical reasons, cf. Figure 33) the questions and scales were de-
veloped and refined in several pre-tests and interviews with experts. Finally,
a questionnaire consisting of 156 largely closed questions was sent to the
Chief Credit Officers of Germany’s largest 519 banks (according to total as-
sets), prior identified and individually contacted by phone. A follow-up by
resending the questionnaire as well as a second contact by phone was con-
ducted. 129 analyzable questionnaires were returned, resulting in a response
Cooperative Sourcing in the Banking Industry 161

rate of 24.9%. The resulting sample can be seen as reasonably representative

in terms of sector sizes (commercial banks, credit cooperatives, public sav-
ings banks) and bank size distributions (total assets, number of employees).
The full survey results have been published in (Wahrenburg et al. 2005).

Sales / Risk
Assessment Processing /
preparation monitoring and Workout
and decision servicing
of credit claim management

x Acquisition of clients x Verify claim documents x Authorize contract x Management of x Management of

x Product choice x Internal rating documents loan portfolio bad loans
x Collect claim data x Meet §18-requirements x Payment x Loan monitoring x Dunning process
x First vote x Second vote x Ongoing transactions x Loan reminder x Encashment
x Loan claim decision x Ongoing data collection x Risk controlling

Figure 33: Reference credit process of S1 and S2

o Source S2: EFL Survey 2005 (“Alignment and Flexibility in Financial Proc-
esses”):
S2 was designed as a follow-up survey of S1, but focused more specifically
on particular criteria relevant for the performance of the SME loans process.
Based on a theoretical model on process performance drivers, incorporating
business competence and flexibility, IT usage, IT flexibility, and IT business
alignment, 170 indicators were developed which allowed for measuring
those constructs and testing the hypothesized model. Again, the question-
naire was sent to the executives responsible for the SME credit process, but
in this case to the largest 1,020 German banks (according to total assets).
Similarly, a follow-up by resending the questionnaire and by phone contact
was conducted. S2 resulted in 136 analyzable questionnaires returned (re-
sponse rate of 13.3%). The sample is not representative in terms of sector
sizes, though: it included significantly fewer public savings banks and sig-
nificantly more credit cooperatives than the basic population. But, it is rep-
resentative regarding bank size. The results of S2 have been published in
(Gomber et al. 2006) and partially in (König and Beimborn 2008).
o Source CASE: EFL Case Studies 2005: As a further follow-up of S1 and as
preparation for S2, a series of case studies was conducted with six German
banks, covering similar topics as S2. In five cases, the unit of analysis was
the SME loans process, while the sixth bank did not offer SME loans. In this
case, the analysis of the private building loans segment was conducted. In-
formation was collected by multiple interviews with the Chief Credit Offi-
cers and sometimes sales managers, IT managers, and controllers (in total 21
people were interviewed). Additional data was gathered from the business
reports. The size of the participating banks ranged from €900 million to
162 Cooperative Sourcing in the Banking Industry

€130 billion in total assets. The results have been documented in internal
case study reports by the E-Finance Lab and have partially been published in
various conference papers (Franke et al. 2005a; Wagner et al. 2006, Beim-
born et al. 2007a).
o Source EI: Expert Interviews 2004: In 2004, the E-Finance Lab conducted a
research project which developed a reference capability map for the German
banking business (Beimborn et al. 2005b) and analyzed the outsourcing po-
tential of parts of the private building loans process from both a legal and an
economic perspective. During this project, eight experts from the banking
industry and from consulting firms (serving the banking industry) were in-
terviewed83. Some of the findings of this project have already been presented
in section 3.2 of this work.
o In some sections, our own empirical research is complemented by empirical
results published by third parties.
In the following, the empirically relevant data derived from the studies is
aggregated from the different sources and presented in sub-sections on basic
demographics (section 3.6.1), process characteristics such as process perform-
ance, task interdependencies, process costs etc. (3.6.2), and BPO potential
(3.6.3). In chapter 5, the parameterization of the simulation model will refer to
these results.

3.6.1 Demographics
The first study (S1) in 2004 addressed the largest 519 German banks which can
be grouped into public savings banks (including state banks) (352 = 67.8%),
credit cooperatives (122 = 23.5%), and private commercial banks (45 = 8.7%).
The sample is representative as regards the size of these groups.

83
Information about the interview participants and the results of the interviews are internally ar-
chived by the E-Finance Lab.
Cooperative Sourcing in the Banking Industry 163

C ommercial banks C ommercial banks

8.7% 10.1%
C redit C redit
cooperatives
cooperatives

23.5%
30.2%
59.7%
67.8%

Public Public
n=129
n=519 savings banks
savings banks
Figure 34: Distribution of bank sectors in the population (left) and in the
sample (right) in S1 (2004)
The follow-up study (S2) in 2005 incorporated the largest 1,020 German
banks (476 public savings banks, 465 credit cooperatives, and 79 commercial
banks). In this enlarged population the relative number of credit cooperatives is
almost doubled, while the proportion of public savings banks has decreased
(Figure 35, left). Unfortunately, in 2005 many of the savings banks decided not
to take part in the survey (Figure 35, right), leading to the dataset being non-
representative as regards the proportion of public savings banks and credit coop-
eratives. Therefore, all of the documented results were tested on structural differ-
ences between the three groups. Unless otherwise noted, the different bank sec-
tors did not give significantly different answers in the survey.
C ommercial banks C ommercial banks

7.7% 7.4%
C redit C redit
cooperatives cooperatives

34.6%
46.7%
45.6%
58.0%

Public Public
n=1,020 n=129
savings banks savings banks

Figure 35: Distribution regarding bank sectors in the population (left) and in
the sample (right) in S2 (2005)
The next table presents the distribution of the studies’ populations and sam-
ples in terms of firm size (measured by total assets and number of employees).
While in S1 the demographic data was added from a third party database, in S2
both measures were asked in the questionnaire. About 70% of the respondents
did not state the number of employees of their bank.
164 Cooperative Sourcing in the Banking Industry

S1 (2004) S2 (2005)
84 Total assets Number of Total assets Number of
Measures
(mill. €) employees (mill. €) employees
.25 quartile ~1,200 ~360 ~480
Not analyzed due
Median ~2,000 ~520 ~850 to the large
.75 quartile ~5,300 ~1,020 ~2,000 number of miss-
ing values
Average 28,349 2,448 12,000
SD 100,437 8,661 52,097
Table 16: Descriptive statistics of bank size distributions of both samples
As can be seen from the quartiles and the comparison of median and average
value, the distributions are very right skewed, containing very few very large
banks but many institutions in the lower field. Figure 36 provides a visualization
of the total assets distribution (please note the logarithmic scale of the abscissa).
Obviously, the distribution of S2 is positioned more left because the studies
differed in targeted population size but focused on the largest (519 resp. 1,020)
banks in both cases.
40%
S1
S2
35%

30%

25%
frequency

20%

15%

10%

5%

0%
100 1,000 10,000 100,000 1,000,000
total assets (in million EUR)

Figure 36: Size distribution of banks in both samples (S1 and S2),
based on total assets
The next figure shows the credit volumes (of all credit types) of both sam-
ples as well as the ratio between credit volume and total assets. Again, in S1 the
data was gathered from secondary sources, whereas it was directly achieved by
the survey in S2 (explaining the lower n due to missing values).

84
To ensure anonymity of the participating banks, no precise values were given for the quartiles.
 Cooperative Sourcing in the Banking Industry 165

40% 35%
35% S1 30% S1
30% S2 25% S2

frequency
frequency

25% n(S1)=129 20% n(S1)=129

20% n(S2)=117
n(S2)=91
15%
15%
10% 10%
5% 5%
0% 0%
100 1,000 10,000 100,000 1,000,000 0 0.2 0.4 0.6 0.8 1
credit volume in million EUR (total) credit volume (total) / total assets

Figure 37: Credit volumes of both samples (left) and ratio between credit
volume and total assets (right)
In S2, we asked for both the SME credit volume and the number of SME
loans in stock. The results are shown in Figure 38. Based on both measures, the
average SME loan size can be determined as €470,000, admittedly with a large
spread (standard deviation = €945,000), expressing the huge outliers on the right.
On average, SME loans amount up to 50% of the total credit volume in German
banks. Again, there is a rather high standard deviation of 26 p.p. (Figure 39).
25% 35%

P=9,560 30%
20% P=581
V=57,809
25% V=1,167
n=98
n=116
frequency

frequency

15%
20%

10% 15%

10%
5%
5%

0% 0%
10 100 1.000 10.000 100.000 10 100 1.000 10.000
number of SME loans in portfolio SME credit volume in million EUR

Figure 38: Number of SME credits (l.) and SME credit volume (r.) in S2
25% P = 50%
V = 26%
n = 130
20%
frequency

15%

10%

5%

0%
0-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 91-100
ratio in % ( SME credit volume / total credit volume )

Figure 39: SME credit volume as part of the total credit volume (S2)
 166 Cooperative Sourcing in the Banking Industry

Finally, the question to be answered is how bank size and credit volume are
interrelated. In the S1 dataset we tested a linear regression between total credit
volume and total assets. As can be seen in Figure 40, there is a very strong linear
relationship85. In S2, where the respondents were explicitly asked for the SME
credit volume, a similar significant relationship could be found for this particular
type of credit86 Besides the credit volume, the number of SME loans in stock is
strongly correlated with total assets, too (Pearson correlation = .391, p<.01).
1,000,000

100,000
total assets (=y)

10,000

1,000

n = 129
100
100 1,000 10,000 100,000 1,000,000
total credit volume (=x)
Figure 40: Relationship between bank size (total assets) and total credit volume
(S1)
3.6.2 Characteristics of the Credit Process
In the following, some of our empirical results, which are relevant to BPO, are
highlighted. First, a short overview about the perceived performance and strate-
gic value of the credit business is given. Afterwards, particular characteristics of
the loans process are empirically investigated in order to fill the parameters of
the cooperative sourcing model (chapter 4) such as process costs, task interde-
pendencies, and complexity of business functions, and similarity between proc-
esses of different banks.

85
Pearson correlation = .836, Spearman correlation = .893, Pearson correlation on logarithmized
data =.968 (The latter were chosen in order to diminish the bias from exponential distributions.)
86
Pearson correlation = .404, Spearman correlation = .836, Pearson correlation on logarithmized
data = .672. (The latter were chosen in order to diminish the bias from exponential distributions.)
Cooperative Sourcing in the Banking Industry 167

3.6.2.1 Process Performance and Strategic Relevance

Process performance is usually measured in terms of costs, time, and quality
(Droge et al. 2004). By contrast, surveys primarily have to use qualitative and
“perceived” (by the respondent) performance measures since asking for quantita-
tive data usually leads to lots of missing values in the data set. Bank managers
often either do not know the “true” values or are not willing to communicate
them (Gomber et al. 2006; Wahrenburg et al. 2005).
As a first global performance measure (in accordance with Chan et al. 1997;
Gopal et al. 1993, for example), both studies – S1 and S2 – asked for the general
satisfaction of the respondent (executive manager responsible for the overall
process) with the SME loans process. As can be seen in Figure 41, the majority
of the credit process executives is more or less satisfied with their process, with
an significant increase in satisfaction in the more recent study which included
more and smaller banks.
60% 56.5% 60%
50% P=.23
50% P=.82 41.9%
n=100
40% n=130 40%
30.2%
30% 30%
21.4% 20.9%
20% 20%
10.7% 10.7%
10% 10%
2.3% 3.9% 3.1%
0% 0%
2 - content 1 - rather 0- -1 - rather -2 - 2 - content 1 - rather 0- -1 - rather -2 -
content indifferent discontent discontent content indifferent discontent discontent
Figure 41: Satisfaction with the overall process (left: S2, 2005, right: S1, 2004)
In S2, the managers were also asked about their satisfaction with the five
sub processes of the SME credit process. The highest satisfaction can be found
for the step assessment/decision while workout received the worst evaluation;
nevertheless there are no strong differences.
frequency
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales/preparation P=3.12 n=134

Assessment/decision P=2.75 n=134

Processing/servicing P=3.07 n=134

Risk management P=3.15 n=134

Workout P=3.18 n=132

1 - totally content 2 - content 3 - rather content 4 - indifferent

5 - rather discontent 6 - discontent 7 - totally discontent don't know

Figure 42: Satisfaction with process steps (data source: S2)

168 Cooperative Sourcing in the Banking Industry

For measuring the process time, we asked for the average number of days
needed from submitting the credit proposal and all necessary documents by the
customer to the final commitment or refusal by the bank. In order to get more
precise data the question was asked twice – for a standardized loan and a rather
specific and complex financing proposal. The result is an average duration of 8.1
working days for the first and 14.3 days for the latter (Figure 43). The process
time correlates significantly with the managers’ satisfaction with the process87.
Pstandardized loan= 8.06 Pcomplex loan= 14.25 n=133
40%
standardized loan
35%
complex loan
p ro p o rtio n o f b an ks

30%
25%
20%

15%

10%
5%

0%
1-3 >3-6 >6-9 >9-12 >12-15 >15-18 >18
avg. process time in man-days
Figure 43: Process times for loan proposals (data source: S2)
Other studies on credit processes detailed this analysis to single tasks of the
credit process to determine inefficiencies more precisely. For example, a survey
of the Eastern German public savings banks association (OSGV) showed that
times for the processing of private building loans of eastern German public sav-
ings banks varied about 441% (Figure 44) (Holtmann and Kleinheyer 2002). In
the administration process step deviations of up to 124% were found.

Processing new building loans Administration

4,525 304

Execution times +124% 136

+441% 3,689
in minutes
136
836

Figure 44: Differences in actual processing times of Eastern German public

savings banks (Holtmann and Kleinheyer 2002, 478)
87
Pearson correlation with satisfaction (Figure 41, right): .228 (standard loans), significant on .01-
level and .218 (complex loan), significant on .05-level.
Cooperative Sourcing in the Banking Industry 169

An own case study on the retail building loan processing of a German com-
mercial bank showed that huge deviations not only occur between different
banks but also between different loans processing units within the same bank
(Table 17). Experts from other large commercial banks (source EI) confirmed
these results and stated that variations between 50% and 100% between different
internal service centers are quite common.
Deviation betw.
Average times per single Processing Processing Processing Processing
lowest and
private building loan in hours unit 1 unit 2 unit 3 unit 4
highest value
Taking over credit agreement 0:43 0:32 0:28 0:35 53%
Taking over collaterals 0:36 0:29 0:22 0:25 64%
Processing of collaterals 0:22 0:25 0:17 0:18 47%
Processing of credit 0:23 0:36 0:32 0:28 57%
Discharging credit 0:16 0:18 0:14 0:17 29%
Releasing collaterals 0:34 0:38 0:28 0:30 36%
Table 17: Processing times of different service centers of a German
commercial bank (data source: own analysis in 200388)
Apart from process times, process quality is an important item to determine
process performance. In S2, this item was measured by the proportion of loans
being processed without problems. As can be seen in Figure 45, on average three
quarters of the loans are processed without winding up for any reason.
P = 74.9% n= 129
35%

30%

25%
responses [%]

20%

15%

10%

5%

0%
<30% 30%-<60% 60%-<80% 80%<-90% 90%-<95% 95%-100%
proportion of loans being processed without any problems [%]
Figure 45: Proportion of granted loans being processed without any problems
(data source: S2)

88
In 2004, the processes were reengineered and standardized, leading to significant reductions in
execution times and variation ranges.
170 Cooperative Sourcing in the Banking Industry

Understandably, this result correlates with the processing time, particularly

for processing complex financing proposals89. Evidently, these not only take
longer but cause more problems (and costs) during processing. On average over
all banks, 74.9% of the granted loans are processed without problems.
As another indicator for process quality, respondents were asked for the
proportion of SME loans with an at least “good” credit rating and for the propor-
tion of loans which fail. The results are displayed in Figure 46. 80% of the re-
sponding banks have less than 5% of loans failing (on average 1.83%). Further,
there is a moderately positive correlation between the fraction of failed loans and
firm size90. In contrast, the proportion of non-good loans increases up to 55% for
80% of the respondents, while the median is at 31.5% (mean = 37.25%).
80%
not "good" loans not "good" loans
failed loans n = 110
maximum proportion in credit volume

P = 37.25%
V = 18.96 p.p.
60%

40%

failed loans
n = 88
20% P = 1.83%
V = 1.895 p.p.

0%
0% 20% 40% 60% 80% 100%
proportion of banks

Figure 46: Maximum percentage of failed loans (lower border) and of loans
without a “good” credit rating (upper border) (data source: S2)
While the proportion of failed loans is a “standardized” measure, the number
of loans not having at least a “good” rating depends on the scheme of rating
classes of the particular bank, partly explaining the high values. Therefore, these
answers are not comparable in all respects.
Apart from process time and quality, process costs are the third factor to
evaluate the performance of a business process. When asked for the process costs
associated with a single SME loan application, the following picture is revealed
(Figure 47).

89
Pearson correlation: -.343 compared to -.276 for standardized loans, both significant on .01-level
90
Pearson correlation with total assets = .250, p<.05 (n=84).
Cooperative Sourcing in the Banking Industry 171

P=1,357.50 n=38
35%

30%

25%
responses [%]

20%

15%

10%

5%

0%
<300 300 - <500 500 - <700 700 - 1,000 - 2,000 - >= 3,000
<1,000 <2,000 <3,000
process costs [EUR]
Figure 47: Total process costs for a single SME loan application
(data source: S2) (König and Beimborn 2008, 190)
Irrespective of the fact that not even a third of the respondents (n = 38) could
(or wanted to) give an answer, the wide span of answers is noticeable. While the
average costs are €1,357.50, a quarter of the respondents stated costs of more
than €2,000. These huge differences between different banks have been found in
other studies as well (e.g. Hölzer 2004). Process costs are analyzed in detail in a
separate section (3.6.2.5).
The high correlation between process time and process costs is coherent
(Pearson correlation: .526, p<.01). Figure 48 visually shows the relationship for
standardized loans91. The regression function (also depicted in the diagram) im-
plies that by reducing process time by about one day, process costs for standard
loans could be reduced by about €131. Nevertheless, the large deviation of the
data points allows only for a very cautious interpretation.

91
Two extreme outliers have been removed.
172 Cooperative Sourcing in the Banking Industry

4000

3500
process costs / loan [EUR] (=y)

3000

2500

2000

1500

1000

500 y = 131.24x + 226.71

R2 = 0.2768
0
0 2 4 6 8 10 12 14 16

process tim e [days] (=x)

Figure 48: Process times and process costs (data source: S2, n=37)
The survey did not find any relationship between process costs and the size
of the bank or the bank sector. Section 3.6.2.5 will take a closer look on the
process costs and their allocation to the different process steps.
By and large, we found a rather positive picture of the performance of the
SME credit process in German banks. The majority of study participants were
quite content. Nevertheless, the high spans when asking for processing times,
problems, and costs (and here even more the high number of missing answers)
indicate that there is still high optimization potential in many banks.
The ultimate goal of process design and optimization is to generate a com-
petitive advantage in the particular market. The former measures only focused on
process performance from an internal perspective but do not ensure a competi-
tive advantage (or “external performance”) per se. Therefore, S2 in particular
asked for information about further indicators which took the bank’s perform-
ance on the market into account.
Almost a third of the participating banks shared the opinion that their par-
ticular credit process design represents a competitive advantage for their firm,
while almost the same number was of the contrary opinion (Figure 49).
Cooperative Sourcing in the Banking Industry 173

0.7% 4.4%

8.1%

19.1% 27.2% 1 - totally agree

2 - rather agree
3 - indifferent
4 - rather disagree
5 - totally disagree
40.4% don't know

P = 2.99
n= 136

Figure 49: “The design of our SME credit process represents a sustainable
competitive advantage to our business.” (S2)
A measure to estimate the realized competitive advantage is the bank’s mar-
ket share in the relevant market (Bergeron et al. 2004; Chang and King 2005)
(Figure 50).
80%
P= 37.5% n = 114
70%
savings banks
60% credit cooperatives

50% commercial banks

frequency

40%

30%

20%

10%

0%
0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89
market share [%]
Figure 50: Stated market share of the SME credit business
in the relevant market [in %] (data source: S2)
On average, the banks stated their own market share in the SME credit busi-
ness in their relevant market to be 37.5%. Differentiated by bank sectors, com-
mercial banks report much smaller market shares than cooperatives and savings
banks because the SME credit business is often a very local business whereas the
large commercial banks often have a less distinctive branch infrastructure.
Moreover, for a savings bank or a credit cooperative the relevant market usually
consists only of the SMEs in the bank’s local surrounding.
174 Cooperative Sourcing in the Banking Industry

When incorporating a dynamic perspective, 36.1% of the participants stated

that they had increased their market share within the last three years, while 21%
mentioned a decreasing trend (no figure). Credit cooperatives in particular lost
market share, while savings banks claimed the strongest increases. Further, over
half of the banks (55.9%) plan to extend the SME credit business in the future.
Another objective indicator of the competitive position of a bank’s credit
business is the interest margin, i.e. the difference between the customer interest
rate and the refinancing conditions (maturity matching inter-bank interest rate).
This measure is not a profitability measure because it does not consider opera-
tional costs and risk costs, but it does measure market-reflected performance for
a particular part of the banking business, whereas available profitability meas-
ures are usually firm-oriented, such as ROE92 or OpM93 rather than process-
oriented.
To reduce the difficulty of comparing different risk structures between the
banks, the survey did not ask for the average interest margin but for the interest
margin of SME loans with a good rating (Figure 51). The majority of the partici-
pating banks achieve interest margins of .5 – 2.5 percentage points (average =
1.68). Similar results were found in S1. The interest margin seems not to be
related to bank size, bank sector, and, what is most surprising, to the banks’ own
or their competitors’ market share or to the number of competitors.
Only one third (36.3%) of the study participants were content with their cur-
rent interest margin while a few more (37.8%) stated the opposite (no figure).
One year before (S1), the proportion of discontent respondents was significantly
higher (52.4%). Banks which have increased their market shares during the last
three years and which see a competitive advantage resulting from their SME
credit process design (Figure 49) are especially content with the interest margin
(Pearson correlation: .361 and .382, p<.01).

92
Return on Equity
93
Operational Margin = 1 – (administrative costs + risk provisioning)/ operating income
Cooperative Sourcing in the Banking Industry 175

P = 1.68% n= 94
40%

35%

30%
frequency [%]

25%

20%

15%

10%

5%

0%
0-.5 >.5-1.0 >1.0-1.5 >1.5-2.0 >2.0-2.5 >2.5-3.0 >3.0-3.5 >3.5-4.0
avg. interest margin [%]

Figure 51: Average interest margin for SME loans with a good rating
(data source: S2) (König and Beimborn 2008, 192)
The final question tackled in this section targets the strategic impact of the
SME credit business. What is the strategic value of this particular business seg-
ment? While the “output” side of this question is at least partly answered by the
indicators measuring competitive advantage (see above), we want to focus more
on the prerequisites of strategic value as argued by the core competence view (cf.
section 2.1.6): strategic value can only be provided by those capabilities of a firm
which represent core competencies. Figure 52 shows that most banks evaluate
the initial stages of the credit process (sales/preparation and assess-
ment/decision) as a core competence of their bank. Following the segmentation
models in chapter 3.2, this indicates that many banks would like to make their
primary focus reducing their business to a sales bank (cf. section 3.6.3). Never-
theless, the evaluations for the three remaining process parts are quite high, too.
frequency
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales/preparation P=4.50 n=128

Assessment/decision P=4.52 n=127

Processing/servicing P=3.69 n=127

Risk management P=4.12 n=128

Workout
P=3.49 n=124

5 -totally agree 4 - rather agree 3 - indifferent 2 - rather disagree 1 - totally agree

Figure 52: “Process step ... represents a core competence of our bank.”
(Data source: S1) (Wahrenburg et al. 2005)
176 Cooperative Sourcing in the Banking Industry

The answers show quite a high correlation among themselves (Table 18).
The results in Figure 52 and the correlations are used for parameterization in the
simulation studies (section 5.3.3.2).
Sales/ Assessment/ Processing/ Risk
preparation decision servicing monitoring
Assessment/
.415, p<.01
decision
Processing/ no significant
.305, p<.01
servicing correlation
Risk no significant
.428, p<.01 .388, p<.01
monitoring correlation
no significant
Workout .342, p<.01 .477, p<.01 .424, p<.01
correlation
Table 18: Correlation between perceived core competence of different process
steps (data source: S1)
The high correlation between mid and back-office functions but the lack of
correlation between sales and the back-office functions, shows that there is a
competence focus on one of these areas, usually the front office. Nevertheless,
the efficient management of the compound bundle of back-office activities can
also provide a competitive advantage.
To get a complementary perspective, the banks were also asked (for the
overall process) whether a BPO provider would be more competent in designing
and optimizing the SME credit process than the outsourcing bank. 37.2% agree
with this statement while only 21.7% of the respondents refute it (Figure 53).
3.9%
3.9% 2.3%

1 - totally agree
17.8% 2 - rather agree
34.9%
3 - indifferent
4 - rather disagree
5 - totally disagree
don't know

37.2%
P = 2.85
n= 129

Figure 53: “A sourcing provider would be more competent in designing and

optimizing the SME credit process than our own bank.”
(Data source: S1) (Wahrenburg et al. 2005)
Further indicators of strategic value and core competence are incorporated in
the section on process outsourcing potential in the SME credit business (section
Cooperative Sourcing in the Banking Industry 177

3.6.3). From the RBV’s perspective, business functions can be outsourced if they
do not take valuable resources with them. Thus, analyses of the BPO potential
can help to establish the strategic relevance of the business process under inves-
tigation.
It can be concluded that there are some banks which achieve a competitive
advantage from the design of their SME credit process and which exploit this to
achieve cost leadership or a differentiation (cf. section 2.1.5). These banks rather
tend to have a higher market share in the relevant market, which has also in-
creased over the last three years and they are more satisfied with their profitabil-
ity. Nevertheless, the market for SME loans seems to offer potential for both
increasing internal performance by raising efficiency and consequentially in-
creasing the competitive advantage of the business by focusing more strongly on
core competencies.

3.6.2.2 Process Complexity

As shown in the literature review, process complexity has a major impact on
transaction costs and agency costs (cf. section 2.2.2.2). Therefore, in S2 we
asked for the degree of complexity of the five process steps based on a 5-Likert
scale.
frequency
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales/preparation P=2.59 n=136

Assessment/decision P=2.46 n=136

Processing/servicing P=2.57 n=136

Risk management P=2.37 n=136

Workout P=2.17 n=126

1 - very high 2 - high 3 - medium 4 - low 5 - very low don't know

Figure 54: Complexity of SME credit process steps (data source: S2)
As shown in Figure 54, the different parts of the credit process are, on aver-
age, estimated to be quite similar regarding complexity (medium to high). There
are no significant differences. In the same study, we tried to devise more indica-
tors to get deeper insights into the characteristics of the workflow. For example,
are the tasks highly repetitive, capable of being automated, or does every credit
application demand attention to its individual characteristics? The following
diagram presents an interesting picture.
178 Cooperative Sourcing in the Banking Industry

percentage of respondents
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

only routine tasks tasks always new

infrequent changes frequent changes

low variabilty high variability

high repetitiveness low repetitiveness

Figure 55: Characteristics of the SME credit process (data source: S2)
Overall, 40% of the responding managers evaluate the credit process tasks
as being routine, while 26.5% believe novel tasks appear with every application.
Further, changes to the process (legal issues or optimization activities) happen
quite often in almost 30% of the banks, while a few more do not change their
process often. The SME credit process is more variable and less repetitive. The
question now is, whether more complex tasks are less suited for outsourcing.
Respondents of S1 were asked whether only “simple” standardized credit proc-
esses like consumer loans or other credit products in retail banking are suitable
for outsourcing. The overwhelming majority agreed (85.3%) and stated the proc-
essing of complex products like corporate loans has to remain in-house. Never-
theless, there is a huge range from retail loans to large corporate investment
loans. Corporate loans for small firms and business customers, in particular, are
located somewhere in the middle in terms of complexity as shown in Figure 55.

3.6.2.3 Modularity between Single Business Functions

As discussed in the section on production cost economics (section 2.1.1), selec-
tive outsourcing of single business functions can cause diseconomies of scope. If
a particular business function is resected from its business process, process costs
may increase due to a loss of synergy effects (Bahli and Rivard 2003; Bruch
1998; Van der Vegt et al. 1998) and internal alignment (Gomber et al. 2006),
arising interface costs etc. In section 2.1.1.1 we distinguished between vertical
economies of scope (within one business process) and horizontal economies of
scope (between multiple business processes serving different (e.g. loan) prod-
ucts. In the following, we primarily focus on vertical (dis)economies of scope
because the surveys S1 and S2 only focused on one particular business process.
Only the expert interviews (EI) took horizontal economies of scope into account.
In S1, several indicators which cover (potential) task interdependencies of
the overall process level were surveyed. First, participants were asked to evaluate
the statement that only a combined outsourcing of the several process steps
Cooperative Sourcing in the Banking Industry 179

would be possible (Figure 56). The results are quite heterogeneous with 22.7%
agreeing to and 25.2% disagreeing. Surprisingly, there is a huge number of re-
spondents who “don’t know” or are indifferent.
7.8% 3.9%

8.6%
18.8%
1 - totally agree
2 - rather agree
3 - indifferent
26.6%
4 - rather disagree
5 - totally disagree
34.4%
don't know

P = 3.19
n= 128

Figure 56: “Only combined outsourcing of the several credit process steps
would be possible.” (Data source: S1) (Wahrenburg et al. 2005)
In the next step, the survey asked whether selective outsourcing of process
parts to specialized processing providers would be not only possible, but also
efficient (Figure 57).
1.6%
5.4% 3.1%

1 - totally agree
27.9% 36.4% 2 - rather agree
3 - indifferent
4 - rather disagree
25.6% 5 - totally disagree
don't know
P = 2.96
n= 129

Figure 57: “Selective outsourcing of process parts to specialized servicing

providers would be efficient.” (S1) (Wahrenburg et al. 2005)
Interestingly, the majority of respondents (39.5%) agreed to the idea that se-
lective outsourcing offers some efficiency potential. Nevertheless, when turning
the statement around and reminding the managers explicitly of the economies of
scope, the proportion of banks who still think selective sourcing would be effi-
cient decreases to 29.4% (Figure 58). Almost half of the responding banks
(47.3%) agree that the different parts of the process parts are so closely intercon-
180 Cooperative Sourcing in the Banking Industry

nected that selective sourcing cannot be efficient. Of course, the answers in

Figure 57 and Figure 58 are highly correlated (Pearson correlation: -.537, p<.01).
5.4%

7.8%

24.0% 1 - totally agree

2 - rather agree

39.5% 3 - indifferent
4 - rather disagree
23.3%
5 - totally disagree
P= 2.80
n= 129

Figure 58: “The parts of the credit process are so tightly interconnected that
selective outsourcing to specialized providers cannot be efficient.”
(Data source: S1) (Wahrenburg et al. 2005)
The interviewees of EI argued that, even if the task interdependence between
the activities is low, communication between the partners has to be much more
formalized. This is seen as a significant cost driver.
In S2 the analysis on task interdependencies was done in more detail. The
participants were asked about the loss of synergy if one of two process steps of
the credit process was outsourced.
frequency
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales <-> Decision

P=4.42 n=133

Sales <-> Risk managem. P=3.52 n=132

Sales <-> Workout P=2.74 n=130

Decision <-> Processing P=3.46 n=133

Processing <-> Risk managem. P=3.05 n=133

Risk managem. <-> Workout P=2.94 n=131

5 - very high 4 - rather high 3 - medium 2 - rather low 1 - very low

Figure 59: Level of synergy loss if one of two business functions is outsourced
(data source: S2) (König and Beimborn 2008, 197)
Diseconomies of scope would be most likely to occur if sales and credit de-
cision were interorganizationally separated; 87.2% of the participants rated them
as high or rather high. This result is comprehensible because most opportunities
to achieve a successful credit business are located at this interface. The more
“soft” information about the credit applicant and application is available in the
Cooperative Sourcing in the Banking Industry 181

decision step, the more effective can the decision be. A separation into sales and
middle office units that is not only organizational (as legally required) but inter-
organizational can create serious problems. Further, there is a substantial agency
problem if sales is outsourced. In the past, banks have generated additional busi-
ness from external agents who did not have strong interest in extensively re-
searching the application and the applicant’s economic situation. The increased
volume of applications overwhelmed the deciders’ capacity and led to a signifi-
cant increase of loans with an (ex post) bad rating. One of the banks participating
in the case studies series (CASE) still has considerable problems with its risk
structure, which occurred when this sales strategy was used in the past.
Further interfaces, which would be strongly affected by selective outsourc-
ing, are between sales and risk management (53.8% rated diseconomies of scope
as (rather) high) and between assessment/decision and processing/servicing
(51.9%). The second result is very interesting because it represents a typical
break point between an outsourcing bank and a credit factory as already realized
by some German banks.
Between risk management, processing/servicing, and workout the loss of
synergy would be much lower for many banks (35.3% and 30.5%). The interface
between sales and workout offers high synergies for 26.9% of the participants.
This can be explained by the banks’ desire not to annoy the credit taking SME
during a dunning process and not to completely destroy the customer relationship
for all future. The relationship between the customer consultant and the customer
should not be needlessly stressed by interaction from workout units (CASE).
The same results as in S2 were found during the expert interviews (EI)
which were not focused on SME loans, but on building loans in the retail busi-
ness. All interviewees stated similar scope effects as above. Furthermore, EI also
incorporated the activities of product development and refinancing. The inter-
viewees found scope effects to be low between the operational process and prod-
uct development and refinancing, representing further potential break points for
efficient selective outsourcing.
The investigation of the SME credit process suggested two main reasons for
economies of scope. First, the joint usage of IT systems and the involvement of
employees in several different process steps often promises advantages (com-
pound resources) as has already been said of horizontal economies of scope.
Second, process redesign and optimization activities can be executed more effi-
ciently and effectively if the whole process is governed by the same entity (Pfeif-
fer et al. 1999, cf. section 2.1.1.1).
64.8% of the responding managers agreed that the joint usage of resources
by several process parts includes competitive advantages (Figure 60). Surpris-
ingly, a quite high number was not able to give any answer (8.6%).
182 Cooperative Sourcing in the Banking Industry

8.6%
3.9%
7.8%
5.5%

1 - totally agree
2 - rather agree
3 - indifferent
19.5% 4 - rather disagree
5 - totally disagree
57.0% don't know

P = 2.30
n= 128

Figure 60: “The joint usage of resources (IT, HR) by several process steps
includes important competitive advantages.” (Data source: S1)
The survey took a closer look at two distinct areas. When asked whether
processing and servicing of a credit (application) are usually provided by the
same person or whether they are split between two people (who are sometimes
organized in a middle office for processing and a back office for servic-
ing/administration), 79.7% of the participating banks answered that both process
steps are provided by the same person, arguing for significant task interdepend-
ence between these business functions (no figure).
A further indicator of the degree of task interdependence between different
activities is the design of the underlying application landscape. Are the different
process steps supported by the same application or do different information sys-
tems exist?
1.6%
7.0%

17.2% 1 - totally agree

34.4% 2 - rather agree
3 - indifferent
4.7%
4 - rather disagree
5 - totally disagree
don't know
35.2%
P = 3.74
n= 128

Figure 61: “The whole credit process is supported by a single IT application.”94

(Data source: S1) (Wahrenburg et al. 2005)

94
To be able to also represent the predominant use of particular application system, a five-level
scale was used instead of a binary one.
Cooperative Sourcing in the Banking Industry 183

In the majority of the participating banks (69.6%), the credit process is not
supported by a single “core application” (Figure 61). Thus, the question of inte-
gration of the several systems becomes a critical issue. Repeated data entries
represent an avoidable cost and error source. Nevertheless, more than half of the
banks (58.0%) still have media discontinuities which lead to the necessity of re-
entering data which is already electronically available (Figure 62).
4.7%

20.3%
23.4%
1 - totally agree
2 - rather agree
3 - indifferent
14.1% 4 - rather disagree
37.5%
5 - totally disagree
P = 3.45
n= 128

Figure 62: “During the processing no data has to be manually entered which
has already been collected and entered in the sales step.” (Data
source: S1) (Wahrenburg et al. 2005)
One year later and with a population containing a higher proportion of
smaller banks, the result shifted slightly towards more integration. In S2, 37.6%
said they had no or almost no media discontinuities (compared to 28.1% in S1),
while only 38.4% (compared to 57.8%) stated the opposite (no figure). The
analysis further showed that credit cooperatives seem to have more strongly
integrated systems than the public savings banks95.
The efficiency potential for reducing media discontinuities is crucial for the
German banking industry; 87.6% agreed that their reduction would contribute
significantly to the optimization of the credit process.
Figure 63 shows the proportion of banks with media discontinuities between
the different process steps. They exist between sales and the subsequent prepara-
tion of the decision in more than half of the participating banks. The same situa-
tion occurs between servicing and risk monitoring and workout. Often, the latter
process steps are not only supported by different applications but also conducted
by different organizational units, while the middle steps of the figure usually take
place in the credit office.

95
Means: cooperatives: 2.9, savings banks: 3.4, Kruskal-Wallis test: p<.037.
 184 Cooperative Sourcing in the Banking Industry

n=115
70%

57.4%
60%
53.0% 53.0%
percentage of banks

50%

38.3%
40% 36.5%

30%
21.7%
20%

10%

0%

Preparation Assessment Risk

Sales Processing Servicing Workout
of decision and decision monitoring

Figure 63: Existing media discontinuities between the several process steps of
the credit process (data source: S1) (Wahrenburg et al. 2005)
To summarize, the credit process is organized quite heterogeneously in the
different banks. In many banks there is a dedicated employee who is responsible
for a particular loan (application), but who has to operate multiple systems. The
study showed that the assignment of different tasks to either the sales unit or the
back office is realized very differently across the participating banks. In the ma-
jority of the banks, data cannot be completely processed straight through without
re-entering any data. From an operational perspective, the interdependence of
sales and back-office tasks, as well as between back-office tasks and risk moni-
toring or workout seems to be rather limited, while within the credit office (deci-
sion, processing, administration, servicing) it is rather high.

3.6.2.4 Similarity of activities between different banks

A basic condition for realizing economies of scale from outsourcing is to stan-
dardize the merged business functions (Cachon and Harker 2002; Matiaske and
Mellewigt 2002; Schott 1997; cf. sections 2.1.1.2 and 2.2.2.1). If a bank is un-
able or not willing to adopt the standard process provided by the service pro-
vider, cost efficient sourcing will not be possible.
It is quite impossible to measure the similarity or even the standardization
potential of business functions in different banks because an explicit inter-firm
comparison and in-depth analyses would be necessary. To get at least a vague
approximation, in S1 the participating bank managers were asked about the in-
dustry-wide standardization potential of the several process steps (Figure 64).
Cooperative Sourcing in the Banking Industry 185

frequency
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales/preparation P=2.51 n=125

Assessment/decision P=2.34 n=125

Processing/servicing P=3.87 n=125

Risk monitoring P=3.02 n=124

Workout P=3.33 n=121

5 - totally agree 4 - rather agree 3 - indifferent 2 - rather disagree 1 - totally disagree

Figure 64: “Standardizing business function … across the industry is possible.”

(Data source: S1) (Wahrenburg et al. 2005)
Many respondents see significant standardization potential for the latter
process steps, first of all processing/servicing (71.2% agree to some extent),
followed by risk monitoring (41.9%) and workout (49.6%)96. Large banks par-
ticularly tend to see these process steps not as being unique to themselves.
The answers are partly biased by the respondent’s perception of their own
core competencies. If a manager stated that a business function could not be
standardized (“1” in Figure 64), in almost all cases they also stated this business
function was a core competence of their own bank (“5” in Figure 52). Hence,
standardization might be possible but is not wanted.
Process standardization would primarily be reflected on the IS layer (data
formats, implemented workflows, used applications, interfaces), forcing firms to
accept the standards of the interorganizational system (Van der Vegt et al. 1998;
Wybo and Goodhue 1995). Therefore, the participants were asked to answer two
IS-related questions which complement the findings on standardization potential.
First, they were asked whether information systems currently used in the SME
credit process were customized in such a way that they could not be replaced by
standard software (Figure 65). Only 25.8% of the respondents said yes while
48.5% said no.

96
The workout indicator correlates to bank size: Pearson correlation between total assets and stan-
dardization potential of workout: -.245, p<.01.
186 Cooperative Sourcing in the Banking Industry

3.9% 5.5%

13.3%
1 - totally agree
20.3%
2 - rather agree
3 - indifferent
4 - rather disagree
35.2% 5 - totally disagree
21.9%
don't know

P = 3.32
n= 128

Figure 65: “The information systems in the SME credit process are customized
in such a way that they cannot be replaced by standard software.”
(Data source: S1) (Wahrenburg et al. 2005)
Second, when asked if the credit process already uses industry-wide stan-
dardized data formats (which would significantly facilitate cooperative sourc-
ing), about one third (31.2%) confirmed this (Figure 66). On the other hand,
47.7% denied it.
7.0% 3.9%

13.3%
27.3% 1 - totally agree
2 - rather agree
3 - indifferent
4 - rather disagree
34.4% 5 - totally disagree
14.1% don't know

P = 3.28
n= 128

Figure 66: “The credit process primarily uses industry-wide standardized data
formats.” (Data source: S1) (Wahrenburg et al. 2005)
Summarizing, there is some standardization potential for parts of the back-
office processes and for the information systems being used. The current degree
of standardization of the data format further supports this verdict.

3.6.2.5 Process Costs

To investigate the economic effects of cooperative sourcing, information about
process cost structures and process volumes has to be gathered. Unfortunately,
many banks still have not employed an activity-based costing system (ABC) or
Cooperative Sourcing in the Banking Industry 187

anything similar. S1 showed that about 30% of the Top 500 banks in Germany
use ABC (Wahrenburg et al. 2005, 30). 37.5% of the participating banks have (in
2004) already calculated the process costs for the total SME credit process, while
one fifth have also carried out a more detailed calculation for each of the process
steps. The larger the bank, the more likely it is to have already determined the
total process costs (Spearman97 correlation: .303, p<.01) or process costs for each
process step (Spearman correlation: .199, p<.05).
Although we did not request the same information in S2, it can be assumed
that – because another 500 smaller banks were additionally surveyed – it would
have shown a significantly lower proportion of banks which had already carried
out a monetary analysis of their credit process. In fact, only 28.7% of the S2
respondents were able (or willing) to confirm their overall process costs (n=38).
Initial results have already been presented in section 3.6.2.1. The average process
costs98 for sales/preparation plus assessment/decision were found to be
€1,357.50, but with big deviations (st. dev. = €1,377). The quartiles have been
given in the last row of Table 20 (p. 189). The process costs are not negatively
correlated to firm size or number of loans in stock, as one would have assumed,
due to potential economies of scale. Actually, the rank correlation even shows
moderate positive correlation99. This indicates that either economies of scale are
not being exploited by large players or – less likely – large players are “too
large”, i.e. already experiencing diseconomies of scale (cf. section 2.1.1.3).
The survey participants were asked to allocate the total process costs to the
five parts of the credit process. Table 19 gives the statistical results while Figure
67 summarizes them by depicting the frequency of the “most expensive” process
steps in the largest 1,000 banks.
Table 19 shows the distributions to be quite symmetrical: median and aver-
age values match quite well and the quartiles are quite symmetrical, too. Further,
a Kolmogorov-Smirnov-Test on normal distribution found that the data is quite
well normally distributed for all process steps100. A visual check of QQ-plots
validated these results.

97
A rank correlation coefficient was used to avoid bias resulting from the extremely skewed distri-
bution of total assets.
98
As usual in ABC, in our process analysis, process costs have been defined as total costs over the
loan life time. All direct and indirect costs for personnel, IT, material, calculatory write offs und
rents are allocated to the different process steps based on a singular compensation key (e.g. han-
dling time in each process step). Cf. (Joos-Sachse 2002, Nadig 2000) for the activity-based cost-
ing approach.
99
Spearman correlation with total assets = .335, p<.05, with SME credit volume = .485, p<.01.
100
The following significance levels have been estimated: assessment/decision, risk monitor-
ing/management, and workout: .001, sales/preparation: .023, processing/servicing: .038.
188 Cooperative Sourcing in the Banking Industry

min .25 .75 max

business function median avg. st. dev.
value quartile quartile value
Sales/preparation 10% 25% 30% 40% 70% 32.2% 11.7%
Assessment/decision 5% 10% 15% 20% 45% 16.2% 9.9%
Processing/servicing 2% 20% 30% 35% 60% 28.8% 12.6%
Risk monitoring 3% 5% 10% 15% 40% 12.0% 7.4%
Workout 0% 5% 10% 15% 30% 10.9% 6.1%
Table 19: Allocation of process costs to single process steps
(data source: S2, n=115)
On average, most of the total process costs come from sales/preparation
(32.2%), followed by processing/servicing (28.8%). Assessment/decision
(16.2%), risk monitoring/management (12.0%), and workout (10.9%) represent
significantly lower cost factors in the SME credit process. Most of the respond-
ing managers assign highest costs to sales/preparation or processing/servicing
(Figure 67).
70
n=115
number of respondents

60

50

40
30

20

10

0
ng

ut
g
n

on

in
io

ko
ri
si

ic

ito
at

or
ci

rv
ar

on
de

W
se
ep

m
t/

g/
pr

en

sk
in
s/

Ri
le

es
ss
Sa

oc
se

Pr
As

process step (multiple answers possible)

Figure 67: Most expensive process step (data source: S2)

The next table combines absolute total process costs with relative cost as-
signments on process steps. The values describe the statistics of the absolute
costs of executing a single loan in the relevant process step.
Cooperative Sourcing in the Banking Industry 189

min .25 .75 max st.

business function median avg. skew
value quartile quartile value dev.
Sales/preparation 108 348 580 1,000 4,500 812 802 2.92
Assessment/decision 56 125 296 621 3,000 545 667 2.18
Processing/servicing 6 174 580 1,084 3,000 788 754 1.33
Risk monitoring 16 94 194 560 2,395 482 617 1.89
Workout 0 61 225 617 2,250 396 467 2.31
Total 282 1,083 2,000 4,303 15,000 3,023 2,958 2,19
Table 20: Process costs [in €] of single process steps per loan
(data source: S2, n=38)
The following figure, stemming from data of a large German savings bank,
complements our picture of process costs by showing the relationships between
margin, process costs, and credit risk costs for different credit products101.
100%
14% 11%
3%
3% 6%
5% 39%
80%
44%

20%
21%
risk costs
fraction of revenue

3%
60% 7% unit costs residuum
2%
5% unit costs back office
unit costs Sales
8% 21%
40% profit margin (DB III)

57% 60%

20% 41%
39%

0%
consumer private corporate corporate
credit building loan credit 2 credit 3 credit product
small SME medium-size
SME
Figure 68: Allocation of credit revenue for different products for a large
German public savings bank (data source: (Rösemeier 2005))

101
The work on hand is not explicitly concerned about credit risk costs because basically they play
no major role in the outsourcing decision, as long as refinancing is not outsourced. In this case,
risk costs would only be affected if the sourcing provider provided less effective risk monitoring,
or if the bank’s overall risk of the credit portfolio would increase due to lower diversification op-
portunities. Risk costs primarily are of a calculatory nature (Hölzer 2004, 236; Platzer and Riess
2004, 162) since credit risks have to be covered by equity.
190 Cooperative Sourcing in the Banking Industry

The relationships between different products differ greatly. SME loans show
a lower profitability than retail products but significantly higher credit risk
costs102. Even more interestingly, the ratio between process costs in the sales
department and the back office is 3-4 : 1 for retail products and loans to medium-
size SMEs, while it is only 1.6 : 1 for small SMEs. However, relatively low sales
costs are compensated for by higher risk costs in that market segment. Of course,
this is only an exemplary snapshot of one particular bank.
In the expert interviews on retail building loan credit business (EI) we tried
to conduct a more detailed analysis to get a sounder understanding of the cost
drivers. Again, the interview partners were asked to assign the respective total
process costs to the different process steps. In this case, the analysis was based
on a more fine-grained credit process (based on the reference process in section
3.3.2.2). Although only estimations of comparative cost levels were requested,
the interview partners had huge problems in assigning percentage values to the
different process steps. Therefore, after the second interview the approach was
changed. Based on the initial data and on further input from a consulting com-
pany, an initial estimation was developed which was discussed with the inter-
view partners. After all interviews had been conducted, the values were adapted
following the discussion results. Table 21 presents the final results.

Risk management & credit portfolio management

(level: whole bank or business division)

Sales/preparation Middle office Back office

Refinancing
/ treasury
Consulting Assess.
Admin./ Risk
Marketing Acquisition & & Processing Workout
decision
servicing monitoring
offer

proportion of total
process costs
33% 30% 3% 30% 4%

proportion of total
process costs 3% 8% 22% 10% 20% 3% 20% 6% 4% 4%

proportion of
fixed costs
40% 90% 90% 90% 90% 100% 90% 90% 70% 100%

proportion of
variable costs
60% 10% 10% 10% 10% 10% 10% 30%

Table 21: Cost allocation in private mortgage loan processing

33% of the total process costs have been assigned to sales/customer inter-
face, 30% to processing, and 30% to administration (including servicing, risk
monitoring, and workout). Within sales, the major part of the costs is created by

102
Consumer credits usually have only a low amount of risk which is covered by the debtor’s while
building loans are collateralized by a mortgage on the financed object. The value of the building
usually exceeds the credit volume by 20-25%.
Cooperative Sourcing in the Banking Industry 191

consulting and offer. Retail building loans need a lot of consultation and frequent
meetings with the customer until the necessary documents are completed for the
credit application. Within processing, two third of the costs are assigned to con-
tract closure & outpayment. The main cost drivers in this process step are han-
dling collaterals (esp. land charges) and the outpayment in several tranches (be-
fore each partial outpayment the customer’s and the object’s situation have to be
reviewed again). In administration/servicing, the most cost-intensive task is
prolongating the loan because this basically represents granting a new loan (in-
cluding new creditworthiness evaluation and negotiation of updated credit condi-
tions).
After the relative process costs of each business function have been deter-
mined, the follow-up question is how these costs can be divided into fixed and
variable costs. Process costs resulting from ABC usually follow a long-term full-
cost consideration, making no explicit differentiation between fixed and variable
components (in the long term everything is variable). Requesting such estimates
in a questionnaire would lead to an unacceptable high effort needed to fill it out;
therefore we did not try to gather such detailed data in the surveys.
Even when experts were asked (EI) to divide relative process costs into
fixed and variable parts, they only understood this question in a short-term per-
spective. They stated a 90-to-10 relationship between fixed and variable costs for
most of the process steps (except sales: 40/60 and workout: 70/30) (Table 21
above) but considered personnel spending to be fixed. In a long-term considera-
tion, e.g. relevant for outsourcing decisions, major parts of HR costs nevertheless
are variable because the cost allocation base of HR is work time and major parts
of it can be explicitly assigned to a single credit or credit proposal in many of the
process steps103. Therefore, when considering outsourcing, the results above
cannot provide an accurate enough estimation of fixed-cost degression from
outsourcing.
Lamberti and Pöhler (2004) argue that a bank’s operational costs are pre-
dominantly formed by IT and personnel. IT costs, particularly, do not increase in
proportion to bank size (Hughes 1999, 2) and – because many banking processes
are IT intensive – have a significant influence on economies of scale. Thus, a
potential proxy for long-term fixed and variable costs is dividing process costs

103
A sourcing provider can also realize economies of scale from short-term fixed costs by pooling
process volumes because usually there will be volume oscillations over time which are not per-
fectly correlated between different clients. Nevertheless, the insourcer has to keep additional “ca-
pacities” in reserve.
192 Cooperative Sourcing in the Banking Industry

by the main input factors employed, HR and IT, with HR assumed to be variable
and IT costs to be fixed in the long term104.
We followed two different paths to get an estimation of the relationship be-
tween input of IT and HR in the five process steps of the SME credit process (S1
+ S2). As a first and qualitative approach, in S2 the managers responsible for the
credit process had to evaluate the degree of IT usage in each process step on a 7-
Likert scale from “no IT” to “only IT”. In all process steps, intensive IT usage
could be found, strongest in processing/servicing. In contrast, a few banks stated
that they do not use any IT in sales, decision, and workout (Figure 69).
frequency
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales/preparation P=3.60 n=129

Assessment/decision P=3.81 n=131

Processing/servicing P=2.55 n=130

Risk monitoring P=3.02 n=127

Workout P=3.74 n=122

1 - only IT 2 3 4 5 6 7 - no IT

Figure 69: How intensive is IT usage in the single process steps?

(Data source: S2)
These usage indicators are not correlated with bank size. Hence, larger
banks seem not to employ a higher degree of automation in the SME credit proc-
ess than smaller banks105.
Additionally, managers were asked what degree of IT usage they would pre-
fer in order to increase process efficiency. Figure 70 compares the results to the
answers regarding current IT usage (Figure 69) and shows that in all process
steps at least 50% of the respondents would prefer higher IT usage.

104
Of course, this approach is rudimentary and does not take into account other cost factors as
equipment and, particularly, costs for physical resources. Nevertheless, for determining a rela-
tionship of fixed and variable costs the classification of physical resources is quite difficult be-
cause if HR is variable in the long term, office space will be, too. Further, minor parts of HR are
largely independent of the credit volume (e.g. administration of credit processing units etc.)
while, on the other side, parts of IT depend on processing volume: at least, when insourcing larger
processing volumes from other banks, IT capabilities usually have to be extended.
105
There might be biased perceptions of the intensity of IT usage between large and small banks
which relativize this conclusion.
Cooperative Sourcing in the Banking Industry 193

100%

80%
n n
n
frequency

60% n n
40%

20%
= = = = =
0%

ut
n

ng
n

ri n
io

io

ko
ci
at

is

ito

or
vi
ar

ec

on
er

W
ep

more IT
t/d

/s

m
pr

en

ng

k
s/

is
si
m

no change
le

R
es
ss
Sa

oc
se

less IT
Pr
As

Figure 70: Difference between desired and current degree of IT usage

(data source: S2) (König and Beimborn 2008, 196)
The second path followed a quantitative approach. In S2, several quantita-
tive measures were requested which help to develop a proxy for the level of HR
and IT costs as well as the relationship between them. First, the number of em-
ployee equivalents in the organizational units of “Markt” (M = sales unit) and
“Marktfolge” (MF = middle/back office) that are involved in the SME credit
process were requested. On average, in M 17.8 people are employed and 18.2 in
MF. Figure 71 shows the distributions.
40%
35% PM=17.8 PMF=18.2 n=132
30%
frequency [%]

25%
M
20%
MF
15%
10%
5%
0%
0 to 5 5 to 9 10 to 19 20 to 49 50 to 99 > 100
number of employees
Figure 71: Number of employees involved in the SME credit process in sales
unit (M) and middle/back office (MF) (data source: S2)
Further, the survey asked for the annual IT budget dedicated to the overall
SME credit process. Only 32 managers were able to quantify this measure. Of
course, due to the very skewed distribution of bank size, the IT budgets vary in
the same way. The average budget was stated to be about €675,120, while the
median was €250,000.
194 Cooperative Sourcing in the Banking Industry

9
n=32
8 P=675.12
frequency of answers

7 median=250
6
5
4
3
2
1
0
0-100 100-200 200-300 300-500 500-1,000 1,000-3,000 3,000-5,000
IT budget for the SME credit process [in €1,000]
Figure 72: IT budget106 for SME credit process (data source: S2)
In the next step, based on the data of these 32 banks, IT budget and HR costs
(employees multiplied by an average labor cost factor of €55,000 per year107)
were assigned to the five steps of the credit process following the keys listed in
Table 22.
Of course, this allocation approach has some shortcomings. One major one –
apart from the statistical problems of multi-step average computation – is that the
IT budget usually cannot be additively allocated to the different process steps.
For example, if we assume that the whole credit process is supported by a single
core application (cf. Figure 61), then the IT budget will barely decrease if one of
the process steps is outsourced. This again represents vertical economies of
scope from shared resources. Furthermore, product-based (i.e. horizontal)
economies of scope may occur because the same application might be used for
multiple related credit products. Consequently, to provide a more robust ap-
proach, we chose two different allocation mechanisms for each input factor
which results in four (resp. three, see below) different combinations when esti-
mating the ratio between IT and HR costs.

106
Please note that the ranges on the abscissa differ in size.
107
As reported by the Federal Statistical Office (FSO 2004), the average labor costs in the banking
industry have been €56,693 in 2004. In one EI, average labor costs for the credit department of a
medium-sized public savings bank were reported to be around €50,884.
Cooperative Sourcing in the Banking Industry 195

Cost Risk
Sales/ Assessment/ Processing/
allocation monitoring/ Workout
preparation decision servicing
key management
IT I IT cost allocation based on individual relative process costs (Table 19)
IT II IT cost allocation based on individual relative degree of IT usage (Figure 69)
Sum of M staff and MF staff (Figure 71) allocated based on relative individual
HR I
process costs (Table 19)
83% of M 17% of M
staff staff
HR II108
14% of MF 57% of MF 10% of MF 19% of MF
staff staff staff staff
Table 22: Cost allocation keys
While the combination of IT I and HR I makes no sense (same allocation
base), the three remaining combinations of IT and HR cost allocation schemes
are used to determine the relationship between IT and HR costs (Table 23) for all
banks which have provided the necessary data in the questionnaire.
Cost allocation
HR I HR II
scheme
IT I B
IT II A C
Table 23: Cost allocation schemes
Table 24 provides the resulting distributions of IT and HR costs based on the
different cost allocation keys for all five process steps of the SME credit process.
Due to the large variation in the number of loans (last row) total costs vary
strongly. Because there are very few very large banks, average values and stan-
dard deviation have been computed without extreme values109.

108
Fixed distribution based on average/median values of relative process costs (Table 19) and based
on the following assumptions: a) the decision step (including preparation of necessary documents)
is equally split to M and MF, b) risk monitoring/management is only partially operated (assumed
to be 50%) by MF staff, because these activities are usually executed by other centralized organ-
izational units. A similar assumption can be made for workout.
109
Extreme values are defined here as exceeding or falling short of the quartiles by more than three
times the inter-quartile distance.
196 Cooperative Sourcing in the Banking Industry

Cost Trimmed statistics Quartiles

Process
alloca- [€/year] [€/year]
step
tion key mean sd .00 .25 .50 .75 1.00
IT I 84,725 68,756 9,520 33,125 88,750 150,000 800,000
Sales/ IT II 57,912 49,761 0 20,750 46,320 93,957 1,249,812
prepa- HR I 502,388 488,030 88,000 187,688 396,000 731,500 3,091,000
ration
HR II 535,207 371,042 91,300 262,488 456,500 924,413 7,121,400
Assess- IT I 46,862 45,369 2,500 11,875 38,750 98,125 1,000,000
ment IT II 78,397 97,388 0 10,423 40,302 109,301 777,726
/ deci- HR I 288,017 298,541 27,500 85,250 192,500 365,750 3,091,000
sion HR II 249,957 193440 60,500 118,938 187,275 371,525 2,421,100
Process- IT I 89,509 107,433 3,600 25,625 62,500 156,250 1,500,000
ing/ IT II 87,770 104,656 5,174 29,851 53,036 95,095 1,666,000
servic- HR I 495,639 445,159 15,400 188,375 332,750 748,000 3,091,000
ing HR II 497,420 407,985 94,050 242,963 376,200 650,513 3,918,750
IT I 52,503 85,636 1,190 8,750 21,500 98,125 1,250,000
Risk
IT II 61,752 61,617 5,174 29,851 53,036 95,095 1,666,000
moni-
HR I 195,672 192,671 24,750 76,313 151,250 251,625 4,636,500
toring
HR II 87,267 71,576 16,500 42,625 66,000 114,125 687,500
IT I 47,082 70,841 0 7,125 22,500 98,125 504,000
Work- IT II 53,587 59,759 3,105 18,825 47,473 91,817 1,000,000
out HR I 164,890 141,455 0 69,850 154,000 247,500 1,980,000
HR II 165,807 135,995 31,350 80,988 125,400 216,838 1,306,250
Number of loans
3,520 6,120 80 800 1,400 3,810 30,000
in stock
Table 24: IT and HR cost distributions based on the different cost allocation
keys (mean and sd based on trimmed data (no extreme values have
been incorporated))
Extreme values have been removed; the results are presented as box plots
(Figure 73) and data plots in relation to firm size (Figure 74, example for cost
allocation scheme A). Table 25 gives the statistical results.
Cooperative Sourcing in the Banking Industry 197

3.00

2.00

ratio
Value
IT/HR


1.00








0.00
Sales/
Sales Assessm./ Processing
Handling / Servicing Workout
preparationPreparation
decision/ Decision Risk monitoring
Process part
3.00 3.00

A C

2.00 2.00
IT/HR ratio

Valueratio
Value

IT/HR




1.00 1.00

0.00 0.00
Sales/ Sales/
Sales Assessm./ Processing
Handling / Servicing Workout
Sales Assessm./ Processing
Handling / Servicing Workout preparationPreparation
preparationPreparation
decision/ Decision Risk monitoring decision/ Decision Risk monitoring
Process part Process part

Figure 73: Distribution of IT/HR cost ratio, presented as box plots without
extreme values (data source: S2, n=32)
The box plots in Figure 73 give an overview of the value ranges for the dif-
ferent process parts resulting from the applied cost allocation schemes A, B, and
C. Table 25 lists the corresponding values. Interestingly, risk monitoring shows
the highest average and median values of all process parts. This indicates that
risk monitoring is comparably strongly determined by IT costs. Other process
parts that show relatively high IT costs are processing/servicing and workout.
Nevertheless, all the analyses show that most ratios are firmly below 1.0. This
means that IT costs represent only the minor part of the process cost structure of
the SME credit business. The Pearson correlation values between the IT/HR cost
ratio and the number of loans in stock all show a negative orientation (cf. Table
25, outer right column): the larger the total credit engagement, the smaller is the
198 Cooperative Sourcing in the Banking Industry

relative portion of IT costs. If IT costs were assumed to be fixed in the long term
and HR effort is variable in the long term, this indicated an under-proportional
cost trend (i.e. economies of scale) at least for the range investigated.
Cost Correlation
Correlations
alloca- with number
Process step Quartiles between cost
tion
allocation schemes of loans in
scheme stock
110
(Pearson)
avg sd .00 .25 .50 .75 1.00 A-B B-C A-C
A .15 .15 .00 .04 .07 .21 .54 -.15
Sales/ .63
B .17 .14 .02 .05 .12 .24 .55 .92 -.31
preparation .69
C .12 .12 .00 .03 .08 .17 .52 -.25
A .35 .41 .00 .05 .18 .45 1.36 -.32
Assessment/ .36
B .22 .18 .01 .06 .16 .33 .60 .82 -.18
decision .70
C .24 .24 .00 .07 .13 .40 .96 -.27
A .24 .24 .03 .05 .11 .40 .87 -.24
Processing/ .43
B .19 .20 .02 .06 .14 .18 .68 .76 -.25
servicing .89
C .21 .19 .03 .06 .14 .32 .68 -.28
A .36 .36 .05 .20 .28 .51 1.06 -.34
Risk .13
B .74 .96 .04 .16 .23 1.21 3.90 .45 -.18
monitoring .78
C .84 .65 .20 .35 .61 1.02 2.27 -.32
A .31 .31 .00 .16 .30 .40 .90 -.19
.42
Workout B .28 .33 .00 .07 .12 .32 1.20 .60 -.24
.68
C .31 .23 .00 .17 .23 .41 .96 -.15
Table 25: Statistical measures of distribution of IT/HR cost ratio
without extreme values (data source: S2, n=32).
The following figure visualizes this relationship between cost ratio and
number of loans for cost allocation scheme A. Schemes B and C show structur-
ally equivalent results.

110
Correlations are not significant due to small sample size.
Cooperative Sourcing in the Banking Industry 199

Sales/preparation
1.40
Assessment/decision
Processing/servicing
1.20
Risk monitoring
fixed/variable cost ratio

1.00 Workout
Linear (Sales/preparation)
0.80
Linear (Assessment/decision)

0.60 Linear (Processing/servicing)

Linear (Risk monitoring)
0.40 Linear (Workout)

0.20

0.00
0 5 10 15 20 25 30
num ber of SME loans in stock [in Thsd.]
Figure 74: Distribution of IT/HR cost ratio in relation to number of SME loans
in stock without extreme values for cost allocation scheme A
(data source: S2, n=32)111.
In this diagram, the trend line111 declines most strongly for assess-
ment/decision, representing the greatest economies of scale. Second is risk moni-
toring. For B and C (not displayed), risk monitoring shows the most strongly
decreasing trend line. The final table gives the slope values of the regression
functions of all the cost allocation schemes used. Due to strongly divergent
scales on the abscissa and the ordinate, the slope values are given for a 1,000
loans scale112.
Cost allocation scheme
Slope values
A B C
Sales/preparation -.0015 -.0046 -.0033
Assessment/decision -.0202 -.0027 -.0077
Processing/servicing -.0053 -.0061 -.0066
Risk monitoring -.0086 -.0210 -.0237
Workout -.0035 -.0088 -.0019
Table 26: Slope values of linear regression between number of loans and cost
ratio (Grey cells show the highest value in their column.)
The results suggest that risk monitoring and assessment/decision show rela-
tively high automation potential. Both processes are related to the rating which
111
Trend line computed by linear regression using least squares method
112
For example, a slope value of -.02 represents cost ratio decreasing by .02 if loans stock increases
by 1,000 units. Although a logarithmic regression would be more fitting in some cases, we de-
cided to solely use linear regression to enable comparability. Thus, the R2 values are very low, al-
lowing only a weak representation of data by trend lines.
200 Cooperative Sourcing in the Banking Industry

has to be executed once at the beginning but also has to be repeated periodically
during the contract period to uncover possible changes in the bank’s risk portfo-
lio.
Although this cost analysis has been done on a small empirical base, the re-
sults can be used for simulation studies which are able to compensate shortcom-
ings in the empirical investigation by numerical sensitivity analyses. The differ-
ent cost allocation schemes A, B, and C have been applied to get a more robust
quantitative insight into the cost structures of the process being investigated and
can be used in the simulations as different data seeds for varying parameter set-
tings within the sensitivity analyses (cf. section 5.3.3).
In the following section, the results on process costs will be complemented
by direct questions regarding the estimated cost savings potential (and further
effects) from BPO.

3.6.3 BPO Potential of the SME Credit Processes

3.6.3.1 Actual State and General Trends in BPO
Section 3.1 gave an overview about the actual defragmentation and sourcing
tendencies in the German banking market. The typical, fully integrated German
universal bank will more and more split up into specialized institutions which
only cover parts of the banking product and services range and only parts of the
value chain (Focke et al. 2004, Krawietz et al. 2003, Marlière 2004a).
When generally asked for the segment the bank would concentrate on (based
on the 3-segments model introduced in section 3.2.2.1), an overwhelming num-
ber of respondents of S1 claimed that they aim at the sales bank business model
(89.3%, cf. Figure 75). In the credit business this would imply outsourcing of
assessment/decision, processing/servicing, risk monitoring, and workout, or – in
a more practical understanding of the term “sales bank” – at least process-
ing/servicing and workout. Only a few banks (9.8%) chose more than one of the
three segments to specialize on in the future.
Cooperative Sourcing in the Banking Industry 201

100%
frequency of responses

n=123
80% 89.3%

60%

40%
10.0%
20%
16.5%
0%
Sales bank Portfolio bank Transaction bank
Figure 75: “In which segment will your bank primarily specialize itself in
future?” (Data source: S1, multiple answers possible)
The remainder of this section will analyze the status quo and the potential
for BPO in the German credit business. BPO projects have been realized on the
German banking landscape for many years, esp. in transaction banking like pay-
ments and securities processing (cf. section 3.4.2). In the less automated and
automatable credit process, these tendencies are still significantly rare. Accord-
ingly, 91.4% of the respondents of S1 evaluate BPO of credit processes to be in
its infancy. Nevertheless, if asked for their personal attitude towards outsourcing,
only 35.3% gave a positive answer.
Regarding the status quo of credit process outsourcing, the survey partici-
pants of S1 were asked for every process step of the reference process whether
they run it in-house (make) or not (buy). If they run it in-house, they have been
further asked whether they offer it to other banks as a sourcing provider (offer).
percentage of respondents
0% 20% 40% 60% 80% 100%

Sales/preparation n=125

Assessm./decision n=125

Processing/servicing n=125

Risk monitoring n=124

Workout n=121

offer partially offer make partially make & buy buy

Figure 76: Current sourcing strategy for the different process steps
(data source: S1) (Wahrenburg et al. 2005)
Corresponding to the results in section 3.4.3, Figure 76 shows that the ma-
jority of the German banks currently follow the make strategy for all of its SME
credit process parts. Assessment/decision and risk monitoring/management espe-
202 Cooperative Sourcing in the Banking Industry

cially are done in-house (offer + partial offer + make: 98.4% resp. 99.2%). In
contrast, outsourcing is most common for workout (6.5%) and process-
ing/servicing (4.7%) in the German Top 500 banks. In total, 14.2% of the par-
ticipating banks have (partially or totally) outsourced parts of the SME credit
process (10.4 of savings banks, 15.3% of cooperatives, and 30.8% of commercial
banks). Only 4.7% have completely outsourced at least one of the specified proc-
ess steps and 2.3% chose “buy” or “partially make and buy” for more than one
process step.
6.2% of the responding banks insource processes from other banks. Two of
them offer the whole SME credit process (all process steps) to other banks.
When a bank had outsourced parts of its credit process, the manager was
further asked about the success of the sourcing project113.
0.0%

12.5%
1 - successful
29.2%
12.5% 2 - rather successful
3 - indifferent
4 - rather unsuccessful
5 - unsuccessful
45.8% P = 2.55
n= 24

Figure 77: “In total, I evaluate our outsourcing as… .” (Data source: S1)
None of the responding 24 banks is totally content with the result of credit
process outsourcing, not even a third are fairly content (Figure 77). Moreover,
25% of the respondents evaluate their outsourcing as (rather) unsuccessful. This
perfectly matches with the studies of Caldwell/McGee (1997) and Corbett (2002)
who both found a “relationship failure” in 25% of investigated IT outsourcing
relationships, which finally broke down.
The number of firms which have at least evaluated an outsourcing strategy
for their credit business is only slightly higher (33 banks = 27.5%, no figure)
than the number of banks which already did outsourcing. Again, the commercial
banks appear as “fast movers”, 46.1% of them having already evaluated out-
sourcing, while only 30.7% of the credit cooperatives and 22.0% of the savings
banks had done the same (data source: S1, state: 2004). The small number of
banks which have evaluated outsourcing of their credit process but have not
113
In Figure 76, not all respondents gave an answer which explains the slightly higher number of
outsourcers in Figure 77.
Cooperative Sourcing in the Banking Industry 203

realized it, indicates that the market for credit process outsourcing is still in the
early stages of development. To get a better perspective on possible future dy-
namics, the participants of S1 were asked about the optimal sourcing configura-
tion, compared to the actual one (Figure 76). Figure 78 gives the corresponding
picture.
percentage of respondents
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Sales/preparation n=125

Assessm./decision n=125

Processing/servicing n=125

Risk monitoring n=124

W orkout n=121

offer partially offer make partially make & buy buy

Figure 78: Optimal sourcing strategy for the different process steps
(data source: S1) (Wahrenburg et al. 2005)
Processing/servicing and workout will be the process steps most affected by
future reorganizing activities. About 60% of the respondents would change the
sourcing strategy for those process steps. Almost 20% of the banks which today
“make” their credit process consider complete outsourcing to be optimal for
processing/servicing (19.1%) or workout (18.5%). Another 14.6% or 17.6%
evaluates at least partial outsourcing to be optimal. On the other hand, 23.6%
(14.8%) stated that they optimally should offer at least parts of the process-
ing/servicing (workout) to other banks.
For the risk monitoring/management activity the picture changes: only
24.6% would change the current make-strategy and the ratio between buy and
offer is reversed: for these process steps, more banks would concentrate on offer-
ing services instead of outsourcing them.
Strategy changes of banks that already follow a buy- or offer-strategy do not
play any role in this analysis. Two of the responding banks which currently offer
workout services intend to completely outsource this process step, two others
want at least to cancel their service offering and to go back to a pure make-
strategy.
After conducting a single activity analysis, Figure 79 shows the resulting
overall optimal credit process sourcing configuration. The analysis only consid-
ers participating banks which gave answers for all process steps (n=105). The
figure aggregates the results of 76% of these banks; all others chose other
unique, partially “exotic” combinations.
204 Cooperative Sourcing in the Banking Industry

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout
20.0%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout 16.2%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout 8.6%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout
7.6%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout
6.7%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout 3.8%

Risk
Sales/
preparation
Assessment/
decision
Processing/
servicing
management/ Workout 3.8%
monitoring

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout 2.9%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout 2.9%

Risk
Sales/ Assessment/ Processing/
preparation decision servicing
management/
monitoring
Workout
2.9%

Make Pure
Legend: Buy
& offer make

Figure 79: Optimal sourcing configurations (data source: S1, n=105)

(Wahrenburg et al. 2005)
The figure gives evidence of the arrival of industrial process understanding
into the banking industry, which leads to modularizing business activities and
determining the optimal sourcing mode for each. Despite this, 20% of the par-
Cooperative Sourcing in the Banking Industry 205

ticipating bank managers still believe that at least for the SME credit business,
the concept of the fully integrated bank is optimal. Other configurations being
favored are outsourcing of both processing/servicing and workout (16.2%) as
well as the complementary opposite business model of offering both activities as
services to other banks (8.6%). Further 7.6% and 6.7% can only envisage out-
sourcing of either processing/servicing or workout. Only very few banks (3.8%)
believe the outsourcing of the overall back office (processing/servicing, risk
monitoring, workout) to be optimal. A similar number of banks intends to be-
come a service provider for exactly this portfolio of business functions.
For our further research we define the following business models (for analy-
ses in the simulation studies in chapter 5, in particular). We distinguish between
the traditional fully integrated bank without service provision to other banks and
an “innovative” fully integrated bank with service provision which may deliver
any of the business functions to third parties. Further, there are selective out-
sourcers (outsourcing of only one business function of the back office), major
outsourcers (outsourcing of two business functions of the back office), and sales
banks (outsourcing of the whole back office). Pure sales banks (outsourcing of
everything but sales) and PSPs are classified also, although the respondents did
not prefer these options. The reason is that these business models are not real
banking business models because the first is only the intermediation of loans
while the latter represents, in large part, activities which do not necessarily have
to be provided by banks (cf. section 3.3.3). Thus, bank managers will obviously
not reshape the business model of their institute to one of these business models
but would rather arrange for a subsidiary to take over those tasks (cf. section
3.4.3).
206 Cooperative Sourcing in the Banking Industry

Figure 80: Business models judged to be optimal by S1 respondents

The follow-up survey S2 in 2005 focused more precisely on cooperative
sourcing. The participants were asked about both outsourcing parts of the SME
credit process to a joint credit factory (shared subsidiary) or directly to another
bank.
22.1% of the respondents basically evaluate outsourcing of credit process
parts to a joint credit factory as reasonable; the majority of 58.8% rejects this
assessment.
Cooperative Sourcing in the Banking Industry 207

0.7% 2.2%

19.9% 1 - totally agree

33.1% 2 - rather agree
3 - indifferent
4 - rather disagree
18.4% 5 - totally disagree
25.7% don't know
P = 3.68
n= 136

Figure 81: “It is reasonable to consolidate parts of the SME credit process with
those of other banks into a joint credit factory.” (Data source: S2)
The respondents were even more negative about the joint credit factory be-
ing substituted by another bank. Only 7.4% of the banks stated that it is generally
reasonable to outsource parts of the credit process to another bank; however,
there is a slightly more positive attitude among larger banks114.
1.5% 5.9%

8.8% 1 - totally agree

2 - rather agree

47.1% 3 - indifferent
36.8% 4 - rather disagree
5 - totally disagree

P= 4.22
n= 136

Figure 82: “It is reasonable to outsource parts of the SME credit process to
another bank.” (Data source: S2)
This more cautious attitude can be explained by the increased strategic risks
(see below) and maybe because cost savings potential is thought to be lower,
because credit factories in Germany do not usually operate as banks and thus can
avoid paying the high cost of the banking tariff. By contrast, large US banks, for
example, insource the back-office tasks of other banks quite often (e.g. Citibank,
Wells Fargo) (Pieske 2005).
Compared to this distinctly unwelcoming overall attitude, an analysis of the
level of single activities uncovers a differentiated picture. The negative attitude
114
Pearson correlation with total assets: -.189, p<.05
208 Cooperative Sourcing in the Banking Industry

indeed remains for sales/preparation and assessment/decision, but strongly

changes for the remaining process steps. 32.6% and 46.1% can imagine out-
sourcing processing/servicing or workout to another bank (Figure 83).
percentage of responses
0% 20% 40% 60% 80% 100%

P=4.81 n=133
Sales/preparation

Assessm./decision P=4.70 n=132

Processing/servicing P=3.10 n=135

Risk monitoring P=3.81 n=133

Workout P=2.80 n=128

1 - totally agree 2 - rather agree 3 - indifferent 4 - rather disagree 5 - totally disagree

Figure 83: “I could imagine outsourcing process step ... to another bank.”
(Data source: S2) (König and Beimborn 2008, 203)
Outsourcing a business function to another bank implies that it might be co-
operation between competitors (“coopetition”, cf. section 2.1.8.1). Consequently,
when the survey participants were asked what strategic risk would arise from
outsourcing the different credit process parts to another bank, a similar picture
emerged (Figure 84). While strategic risks from outsourcing sales/preparation or
assessment/decision to another bank are evaluated as too high by almost all of
the survey participants, for the latter parts of the credit process this does not
necessarily apply. For processing/servicing and workout, strategic risk is evalu-
ated as being rather low by 34.8% and 47.3% of the respondents.
percentage of responses
0% 20% 40% 60% 80% 100%

Sales/preparation P=1.33 n=136

Assessm./decision P=1.43 n=136
Processing/servicing P=2.99 n=132
Risk monitoring P=2.23 n=134
W orkout P=3.18 n=129

1 - totally agree 2 - rather agree 3 - indifferent 4 - rather disagree 5 - totally agree

Figure 84: “The strategic risk from outsourcing process step ... to another bank
would be too large.” (Source: S2) (König and Beimborn 2008, 203)
In the expert interviews (EI), none of the interviewees saw increased risks
from coopetition. The access to customer data would be a serious threat, but
Cooperative Sourcing in the Banking Industry 209

since the banking industry is very sensitive to confidentiality problems, none of

the partners would “dare to exploit this”. Each violation would inevitably lead to
the firm’s market exit (EI). Nevertheless, the restrictive attitudes towards out-
sourcing in S2 lead to the assumption that in this case even more intricate moni-
toring systems and governance mechanisms would have to be established. An-
other coopetition risk debated in EI was the threat of dominance of a large co-
opetitor which could exploit its bargaining power. In the interview partners’
opinion, this threat could be minimized by appropriate contractual structures.
In S1, the participants were generally asked about the relevance of different
risk factors regarding BPO in the credit business. The most essential problem
seen by the responding managers is becoming dependent of the insourcer
(59.7%) and losing control of the process design and execution (59.7%) (Figure
85).
percentage of responses
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

becoming dependent on
P=2.31 n=129
insourcer

loss of control P=2.36 n=129

appropriate SLA not

P=2.90 n=127
possible

opportunistic behavior P=3.00 n=129

security problems P=3.26 n=129

1 - very high 2 - high 3 - medium 4 - low 5 - very low

Figure 85: How do you evaluate different risks resulting from outsourcing parts
of your SME credit process? (S1) (Wahrenburg et al. 2005)
Another important factor is the inability to sufficiently specify a service
level agreement (SLA) in order to align the insourcer to the outsourcer’s own
objectives. Closely related but seen as less problematic, is the risk of the in-
sourcer behaving opportunistically (24.8%), i.e. exploiting contract incomplete-
ness and the lack of or an incomplete control system. Matching the argumenta-
tion above, the least worrying item appears to be security problems (21%) arising
from the exchange of sensitive information between different companies (integ-
rity problems either during communication or in the insourcer’s systems because
the latter might follow less restrictive security regulations (Earl 1996; Accenture
2002; Petzel 2003)).
210 Cooperative Sourcing in the Banking Industry

In contrast to outsourcing risks, Figure 86 asks for the benefits of an out-

sourcing decision. Outsourcing advantages are classified into two different
groups  economic and strategic (cf. section 2.2.2.1). To find out about the for-
mer, questions were asked about cost reduction and cost variabilization (i.e. fixed
cost and capital reduction), Moreover, questions about strategic benefits have
been included from the CCV (core competence focus) and from an RDT perspec-
tive (access to external superior resources).
percentage of responses
0% 20% 40% 60% 80% 100%

Variabilisation of
P=1.95
costs

Reduction of costs P=1.96

Focussing on core
P=1.98
competencies

Getting access to
specialized P=2.33
resources

1 - totally agree 2 - rather agree 3 - indifferent 4 - rather disagree 5 - totally disagree

Figure 86: “… would be a major advantage of outsourcing credit processes.”

(Data source: S1) (Wahrenburg et al. 2005)
About 80% of the responding banks evaluate both economic arguments as
well as the core competence focus as being important. The cost reduction argu-
ment can also be validated by quantitative data from S2. Banks with high process
costs would be more likely to outsource their back-office functions to other
banks115.
Access to superior resources is less relevant. This matches results from Ge-
wald and Dibbern (2005) who also investigated the driving and inhibiting factors
for the BPO of banking processes. Since banking processes are the core domain
of the outsourcer, no superior resources are assumed to exist on the insourcer’s
side.
Generally asked for the anticipated economic effect of cooperative sourcing,
22.8% estimated operational cost savings to be significant (Figure 87, left) while

115
Spearman correlation: -.190 (not significant due to low number of samples: n =35). For measuring
the correlation, the items regarding outsourcing potential of the three back-office business func-
tions were aggregated by a principal component analysis (PCA) to achieve single measures.
Cooperative Sourcing in the Banking Industry 211

19.3% believe that these savings will not be wasted by occurring transaction
costs (Figure 87, right).
2.9%
7.4% 4.4%

19.9%

19.1% 1 - totally agree

16.2% 25.0% 1 - totally agree
2 - rather agree 2 - rather agree
3 - indifferent 4.4%
3 - indifferent
4 - rather disagree 4 - rather disagree
5 - totally disagree 14.7% 5 - totally disagree
26.5% 27.9% don't know don't know
31.6%
P = 3.37 P = 2.87
n= 136 n= 136

Figure 87: Left: "Cooperative sourcing would lead to significant operational

cost savings." Right: "The occurring transaction costs incurred for
migration and controlling would exceed the operational cost
savings." (Data source: S2)
Interestingly, both questions regarding the cost development show a very
high number of bank managers who either did not know the answer or who were
indifferent. This might be an indicator that many German banks still have not
evaluated outsourcing strategies and, furthermore, fear transaction costs which
are often underestimated ex ante (“hidden costs”, cf. section 2.2.2.2).
Asked for the general minimum savings of operational process costs when
outsourcing a business function (S1), a broad range of answers anticipated sav-
ings of between 10% and 80% while the average is 30.8% (standard deviation =
12.7 percentage points).
45
n=100
40
number of respondents

P=30.8% V=12.7p.p.
35
30
25
20
15
10
5
0
0-10% 11-20% 21-30% 31-40% 41-50% 51-60% 61-70% 71-80%
Figure 88: What are the minimum operational cost savings which have to be
met to make BPO a favorable strategy? (Data source: S1)
(Wahrenburg et al. 2005)
As already argued in the theoretical section, one major reason for cost reduc-
tion from outsourcing lies in realizing additional economies of scale by bundling
212 Cooperative Sourcing in the Banking Industry

similar processes of different firms. Above, we showed that the majority of the
responding bank managers believe that parts of the credit process can be suffi-
ciently standardized, which is a necessary precondition (cf. section 3.6.2.4).
Based on this, the respondents of S1 were requested to estimate whether there are
economies of scale which an insourcer could realize by serving multiple clients.
6.3%
1.6% 4.7% 1.6%

14.8%
8.6% 16.4%
1 - totally agree 12.5% 1 - totally agree
2 - rather agree 2 - rather agree
12.5% 3 - indifferent 3 - indifferent
10.2%
4 - rather disagree 4 - rather disagree
5 - totally disagree 5 - totally disagree
56.3% don't know 54.7% don't know
P = 2.19 P = 2.87
n= 128 n= 136

Figure 89: “A sourcing provider could achieve economies of scale by reducing

HR (left) or IT costs (right).” (Data source: S1)
As Figure 89 shows, most of the survey participants believe economies of
scale to be possible by reducing HR and/or IT. The answers are highly correlated
(Spearman correlation: .592, p<.01) and independent of bank size and bank type.
At a first glance, this seems to contradict the argumentation in the process
cost analysis (section 3.6.2.5), where HR costs have been used as a proxy for
variable costs. In the case studies (CASE) we found a possible explanation: The
managers of the back office in one bank argued that a credit factory can exploit
more automation potential by IT investment. Based on their own investigations,
they estimated that a credit factory can reduce operational HR costs by 20% by
implementing a workflow management system which takes over many adminis-
trative tasks and allows for less qualified personnel in several activities of the
credit process (as e.g. archiving documents or authorizing payments). Efficient
dynamic staffing is seen as a critical factor in efficient credit process design
(CASE).
Thus, doing this leads to a cost function with higher IT costs and less HR
costs, leading to greater economies of scale. Further factors are fixed HR costs
(e.g. administrative functions for process control and management) and the short-
term focus of the respondents (pooling of capacities, cf. section 3.6.2.5).
Some authors of the outsourcing literature opine that economies of scale – at
least in large firms – are often already exploited within the firm (Bloch and
Spang 2003; Dibbern et al. 2003; Lacity et al. 1996; Schott 1997). As a comple-
mentary indicator for Figure 89, this statement was placed in the survey but
Cooperative Sourcing in the Banking Industry 213

rejected by the majority of the participants (Figure 90). As expected, the answers
were related to firm size116.
2.3% 0.8%

9.3%

23.2% 1 - totally agree

17.1% 2 - rather agree
3 - indifferent
4 - rather disagree
5 - totally disagree
don't know

47.3% P = 3.85
n= 129

Figure 90: “Scale economies are already exhausted within the firm. A sourcing
provider cannot realize further significant scale economies.”
(Data source: S1) (Wahrenburg et al. 2005)
A quantitative investigation of realizable cost savings from BPO in the
German credit business is not possible as only very few outsourcing deals have
been realized so far. All participants in EI were unable to quantify cost savings
from outsourcing process parts to a credit factory, due to both a lack of knowl-
edge about in-house processing costs and lack of BPO experience. They just
argued that the optimization of in-house processes would significantly decrease
possible further savings from outsourcing, but process modularization would
increase them. The interview partners argued that, based on their observations,
process cost savings through outsourcing today would not exceed 10–15% (after
considering VAT), based on the fact that a credit process had not yet been opti-
mized in-house. These savings would be too low to justify outsourcing (cf.
Figure 88, where survey participants on average wanted 30% cost savings to
make outsourcing a favorable option).
A reason for the current lack of cost advantages offered by credit factories is
the lack of efforts being made to standardize. Credit factories have still not been
able to completely standardize their client’s processes to one reference credit
process. In EI, it was stated by multiple sources that credit factories usually only
reach a standardization degree of 40–60%. One reason is that some credit facto-
ries are “outsourced problems” of banks which just separate their credit depart-
ments organizationally without establishing a new and industrialized process
(EI+CASE). Thus, the organizational and process structure of credit factories is
often similar to the former credit departments. Opening this structure to third
116
Spearman correlation with total assets: -.335, p<.01 and credit volume: -.317, p<.01.
214 Cooperative Sourcing in the Banking Industry

parties just leads to parallelization of tasks without strategically focusing on

achieving all possible scale effects.
Another argument regarding the BPO potential is about profitability. EI
showed that it is still unclear whether SME loans are appropriate for BPO at all.
Some experts argue that since SME loans are simply not very profitable, it would
be better to outsource them completely (sales bank model) or at least they should
become highly standardized (e.g. as private consumer loans) in order to increase
profitability. Standardization in turn would increase the BPO potential. Contrast-
ingly, other interviewees put forth that serving corporate customers will always
require a high amount of personal treatment and individual (human) credit as-
sessment, leading to less BPO potential.
In summary, it can be said that today outsourcing of credit processes still is
not perceived as meeting the cost saving requirements, leading to stagnation in
the credit processing market. Nevertheless, when credit factories have estab-
lished industrialized (modular, standardized and highly automated) credit ser-
vices which are accepted and used by all of their clients, there is the potential to
outsource major parts of the credit process as the estimations of the credit proc-
ess managers both in S1 and S2 and in EI argue.

3.6.3.2 Economies of Scale, Skill, and Scope – A PLS Approach117

In S1, we not only generated descriptive results but also conducted a positivist
analysis of drivers and inhibitors on BPO. We focused on a PCE perspective and
argued that economies of scale and skill are drivers for outsourcing, while
economies of scope have the opposite impact (task interdependencies, related-
ness, cf. section 2.1.1 and 2.2.2.2). For this analysis, we applied the Partial Least
Squares (PLS) method (Chin 1998; Wold 1985) by using the software package
SmartPLS, version 1.1 (Hansmann and Ringle 2004). Like other structural equa-
tion modeling approaches, it allows for the testing of hypotheses based on latent
variables118.
Figure 91 depicts the basic research model for analyzing the impact of
economies of scale, scope, and skill on the BPO potential of the credit process.

117
This analysis was published in the proceedings of the 11th Americas Conference on Information
Systems in Omaha (NE), USA (Beimborn, Franke, and Weitzel 2005a).
118
In contrast to covariance-based approaches as e.g. LISREL, AMOS, or EQS, commonly used,
e.g., in marketing science, sociology, or psychology, PLS has minimal requirements for meas-
urement scales, residual distribution, and sample size (Chin 1998).
Cooperative Sourcing in the Banking Industry 215

Economies
of scope

-
Hypotheses to be tested:
H1
Perceived economies of scope reduce
H1 (-)
the perception of BPO potential.

Economies BPO
Additionally achievable economies of
+
of scale H2 potential H2 (+) scale promote the perception of BPO
potential.
Perceived economies of skill of
+ H3 (+) service providers promote the percep-
H3 tion of BPO potential.

Economies
of skill

Figure 91: Research model and hypotheses

From the literature discussed in sections 2.1.1 and 2.2.2, we would expect a
negative impact of economies of scope and a positive impact of economies of
scale and skill (which can additionally be achieved by means of a sourcing pro-
vider) on BPO potential.
Each of the latent constructs displayed in Figure 91 is amplified by several
indicators (items in the questionnaire of S1), which have already been descrip-
tively analyzed in the previous sections. Table 27 lists the indicators used and
refers to the descriptive results above. All indicators are used in reflective
mode119. For some indicators the scales had to be reversed to consistently get
high values representing high levels of the assigned construct.

119
Reflective mode represents indicators reflecting the variance of the construct. All indicators of the
construct are supposed to move in the same direction if the construct score shifts.
216 Cooperative Sourcing in the Banking Industry

Scales Descriptive
Construct Indicator Description (reference) Loadings
(* = scale reversed) results
BPO would imply scale
1 – totally disagree* avg = 3.81
SCALE1 effects from HR reduc- .742
5 – totally agree* sd = .894
Economies tions. (Figure 89)
of scale There are no further scale
1 – totally agree avg = 3.85
SCALE2 effects realizable by BPO. .896
5 – totally disagree sd = .924
(Figure 90)
Credit processing/
1 – totally agree* avg = 2.31
SKILL1 servicing is our core .759
5 – totally disagree* sd = .999
competence. (Figure 52)
Economies Risk monitoring is our
1 – totally agree* avg = 1.88
of skill SKILL2 core competence. .845
5 – totally disagree* sd = .891
(Figure 52)
Workout is our core com- 1 – totally agree* avg = 2.51
SKILL3 .711
petence. (Figure 52) 5 – totally disagree* sd = 1.108
Competitive advantage
1 – totally disagree* avg = 3.70
SCOPE1 from shared resources .945
5 – totally agree* sd = .768
(Figure 60)
Economies
Selective outsourcing is
of scope
inefficient due to tight 1 – totally disagree* avg = 3.20
SCOPE2 .424
interconnectedness. 5 – totally agree* sd = 1.063
(Figure 58)
Optimal sourcing strategy 1 – make (incl. offer)
avg = 2.02
BPOPO1 of credit process- 3 – partially make/buy .807
sd = 1.543
ing/servicing (Figure 78) 5 – buy
Optimal sourcing strategy 1 – make (incl. offer)
BPO avg = 1.25
BPOPO2 of risk monitoring 3 – partially make/buy .697
potential sd = .866
(Figure 78) 5 –buy
1 – make (incl. offer)
Optimal sourcing strategy avg = 2.18
BPOPO3 3 – partially make/buy .773
of workout (Figure 78) sd = 1.613
5 – buy
Table 27: Indicators used in the PLS analysis
For economies of skill, those indicators have been included which show the
highest heterogeneity of answers and which focus on the business functions
which primarily have optimization potential mainly from a cost perspective (lat-
ter activities of the credit process back-office activities). Nevertheless, the indi-
cators do not explicitly distinguish between core competence resulting from cost
advantages and those resulting from time or quality advantages and thus also
incorporate an RBV/CCV-perspective. However, we can argue that differences
in process quality, particularly in back offices, can usually be transformed to cost
advantages since quality is measured in error rates (affecting process costs) and
effectiveness of the rating (influencing risk costs) (Wahrenburg et al. 2005).
Furthermore, the time dimension shows to be correlated with process costs in our
sample (cf. Figure 48 on p. 172).
Cooperative Sourcing in the Banking Industry 217

Due to quite few BPO activities being found in the investigated process do-
main, we decided not to implement “BPO” itself as the affected construct and
used “BPO potential” instead. We applied the indicators that asked for the opti-
mal sourcing strategy (Figure 78 on p. 203). The answers for “make&offer”,
“partially make&offer” and “make” were aggregated to “make” (=1).
Statistical tests of causal models that use latent variables (i.e. constructs) are
conducted in two steps. First, the measurement model (i.e. the relationships be-
tween a construct and its indicators) has to be tested in order to validate that the
construct is well represented by its indicators. Second, the structural model (i.e.
relationships between constructs) is analyzed to test the proposed hypotheses.

Test of the Measurement Model

First, it has to be ensured that any indicator loads sufficiently well on its related
construct. For reflective indicators, Chin (1998) claims factor loadings to be
larger than .707120, which is fulfilled by all but one (SCOPE2) of the indicators
used (cf. Table 27 above). Second, the composite reliability tests each construct
for its internal consistency. The required minimum threshold differs between .6
(Bagozzi and Yi 1988) and .7 (Nunnally 1978). Except for economies of scope,
all constructs fulfill the more rigorous threshold (Table 28).
Economies Economies Economies BPO
of scale of scope of skill potential
.805 .669 .816 .804
Table 28: Composite reliability
Third, a latent variable should share a higher fraction of variance with its
own indicators than with indicators assigned to other constructs (Hulland 1999).
Here, the average variance extracted (AVE) as a measure for discriminant valid-
ity which should be higher than .5 (Diamantopoulos and Winklhofer 2001) is
used. The diagonal of Table 29 shows that all constructs fulfill this requirement,
too. The remaining cells represent the correlations between the latent scores,
which are sufficiently lower than the AVE square roots.

120
A threshold of .707 ensures that at least 50% of the indicator’s variance can be explained by the
(latent) construct (Götz and Liehr-Gobbers 2004).
218 Cooperative Sourcing in the Banking Industry

AVE/correlations Economies Economies Economies BPO poten-

of Scale of Scope of Skill tial
Economies of Scale .676
Economies of Scope .002 .537
Economies of Skill .061 .041 .598
BPO potential .176 -.232 .290 .578
Table 29: Average variance extracted (AVE) (diagonal)
and correlations between constructs
It can be summarized that most of the criteria for appropriate measurement
models are fulfilled.

Test of the Structural Model

The results of testing the structural model (Figure 91) are presented by Figure 92.

Economies
of scope
-.242*
(t = 1.316)

Economies .159 BPO

of scale (t = 1.218) potential

r2 = .168

.290**
(t = 1.901)
Economies
of skill

Figure 92: PLS Results (significance levels: *” .9, **” .95121)

The path coefficients represent the causal relationships between the exoge-
nous constructs and the BPO potential. While it can be strongly confirmed that
economies of skill operate as drivers for BPO potential, and economies of scope
act as inhibitors, economies of scale show a positive, but insignificant relation-
ship. r2=.168 represents 16.8% of the variance of the BPO potential construct
being explained by the modeled factors. This rather low figure is unproblematic

121
t-values were generated by using the Bootstrapping algorithm with 500 samples.
Cooperative Sourcing in the Banking Industry 219

here because the model did not attempt to cover all relevant reasons for and
against BPO, but adopted a PCE perspective. Integrating more theoretical per-
spectives would lead to overlaps between the constructs which have to be han-
dled.
The insignificant impact of economies of scale on the perceived BPO poten-
tial in this model is due to the domination of economies of skill. As the huge
differences of process costs show (section 3.6.2.5), there are significant differ-
ences in processing capabilities between the banks.
This, partly, contradicts the results of (Gewald and Dibbern 2005) where ac-
cess to superior capabilities was the only hypothesized outsourcing driver which
was empirically shown to have no significant impact on perceived outsourcing
potential at an overall credit process level. Naturally, bank managers will seldom
agree to the statement that sourcing providers would offer superior process per-
formance compared to their own core business. But, if conducting the analysis at
a more granular level (single process steps), differences can be found, and thus
the perceived core competence in the process steps of the back office, which are
to some extent not seen as core business (although the overall process is), shows
the variance which explains outsourcing potential. As long as core competence
can be mainly expressed as cost efficiency, what is the case in back-office proc-
esses, the argumentation from PCE and RBV/CCV does overlap: core competen-
cies can be transformed to economies of skill.

3.7 Summary
For several years, the German banking industry has shown relative underperfor-
mance in an international context which, it is agued, is caused by high fragmen-
tation of the market, overbanking, and high vertical integration. Analysts rec-
ommend a drastic structural change to be achieved by transforming the tradi-
tional German universal banking system, consisting of specialized players pro-
viding only a subset of banking products and parts of the value chain. This
change can be achieved by both deconstruction and consolidation strategies; a
unification of both strategies is met by the cooperative sourcing concept. Based
on a generic banking value chain, different segmentation models have been in-
troduced and briefly discussed. It has been shown that a more specific business
domain focus is necessary to provide effective analyses and conceptual proposi-
tions. Therefore, the credit business, as a major business domain of most banks,
was chosen and reference processes for three main credit products have been
developed. Based on the common structure of these reference processes, a seg-
mentation model dedicated to the credit business has been developed and dis-
cussed.
220 Cooperative Sourcing in the Banking Industry

In order to investigate the actual state of transformation, section 3.4 took a

close look at BPO activities in the German banking market. While the securities
and payments processing domain has already taken big steps towards reaching a
future banking value network, especially by cooperatively sourcing processing
and administration activities between multiple banks, the credit business is still
unchanged. Although some credit factories have been established in Germany,
their services have hardly been made use compared with other countries. The
path towards a banking value network as envisaged by the segmentation models
is only sparsely followed in this field. The section on legal and regulatory issues
presented some of the reasons which partially explain this situation, e.g., the
VAT problem or high expenses for the transfer of ownership.
The last and largest part of this chapter shed light into the BPO opportunities
of one particular business (SME loans) by conducting our own empirical re-
search. The empirical research showed heterogeneity between the participating
banks in terms of process performance and process costs. Although there is dis-
like of BPO of parts of the core business and against cooperative sourcing in
general, the analysis also showed that managers do not reject the strategy of
outsourcing of back-office parts of the SME credit process out of hand, and they
do see benefits from cooperative sourcing. Economic outsourcing advantages
seem primarily to be based in economies of skill rather than in economies of
scale. This is comprehensible when the strong heterogeneity in process perform-
ance (costs, time, and quality) is taken into account. Banks with superior capa-
bilities in credit processing, e.g. represented by lower process costs, are more
likely to see themselves as potential insourcers of future cooperative sourcing
coalitions in the banking industry.
The following chapter will develop a formal model of cooperative sourcing
which will allow a more dynamic representation of this snapshot of the current
and anticipated future state of cooperative sourcing, and to uncover how coali-
tion forming processes develop over time, by applying a simulation approach.
The model will incorporate the different (cooperative) outsourcing determinants
derived from the literature and allow not only the examination of the effect of the
single drivers and inhibitors but also the impact of their interplay on the exis-
tence and efficiency of cooperative sourcing equilibria.
Developing a Formal Model for Cooperative Sourcing 221

4 Developing a Formal Model for

Cooperative Sourcing
“That kind of analysis explores the relation between the behavior characteristics of the individuals
who comprise some social aggregate, and the characteristics of the aggregate.”
(Schelling 1978, p. 13)

Based on the literature research in the preceding parts, this chapter introduces a
formal model of cooperative sourcing (referred to as cooperative sourcing model
– CSM) which allows for both analytical and simulative studies in the remainder.
First, based on the previous summary of related literature on mathematical out-
sourcing models (section 2.2.3), the motivation behind choosing this formal
approach is explained (section 4.1). Based on the theoretical foundation (chapter
0), the model is successively developed – from different cost functions (process
costs and transaction costs) to the cooperative sourcing decision calculus. After
providing the basic structure in section 4.2, we distinguish a centralized perspec-
tive (4.3) and a decentralized perspective (4.4), completing the model by deci-
sion calculi either of the central planner (binary non-linear optimization problem)
or the autonomously deciding agents (maximizing individual benefits from co-
operative sourcing with uncertainty about the partners’ behavior). Finally, sec-
tion 4.5 extends the model by considering legal and regulatory constraints which
are specific to the banking environment.

4.1 Justification of Model Development

Section 2.2.3 gave an overview of analytical works in the outsourcing research
strand. The model developed and applied in this chapter basically unifies most of
the singular arguments made there and differs in two terms from most of the
formal approaches cited above:
1. It models an n-agent scenario where each agent has both the chance to be-
come an insourcer or an outsourcer.
2. It incorporates a significantly larger level of formal power trying to cover
essential parts of real-world relationships and being able to incorporate em-
pirical data for parameterization and to analyze the influence and interplay
of determinants such as transaction costs, task interdependencies, firm sizes,
business function similarity, and competitiveness between firms.
222 Developing a Formal Model for Cooperative Sourcing

Obviously, raising the formal power corresponds to an increase of mathe-

matical complexity. Thus, the model is capable of capturing real-world phenom-
ena as shown in the subsequent analysis and of applying empirical data within
the model. Parkhe criticizes that research often deals with complexity of organ-
izational behavior “by following Ohm’s Law (‘path of least resistance’), that is,
by simply ignoring it. However, this solution to the problem, acceptable in the
well-established paradigm in economics (cf. Bettis 1991), is hardly suitable for
management scholars, inasmuch as these complexities are among the primary
phenomena demanding concerted attention” (Parkhe 1993, 239). Researchers are
often impelled by this high complexity. As a consequence, they rather study
isolated variables instead of the system of interrelationships between them
(Mintzberg 1977; Parkhe 1993).
Furthermore, Parkhe argues that research attempts to validate existing theo-
ries rather than to create new theory. “The result, according to (Lindblom 1987),
is that theorists often write trivial theories because their process of theory con-
struction is hemmed in by methodological strictures that favor validation rather
than usefulness” (Parkhe 1993, 244).
As already stated in the introductory chapter, this work will try to provide a
circumspective first step in tackling these criticisms by an exploratory approach
built on a mathematical model which significantly differs from the other mathe-
matical approaches above in terms of formal power. This is of course a risky
attempt; therefore the very end of chapter 6 will provide several steps of valida-
tion to substantiate the results.
Nevertheless, the model still remains very abstract, compared to organiza-
tional reality. In the discussion of the usability of game-theoretical models,
Aumann and Maschler note that the “analysis of such a highly simplified abstrac-
tion can very seldom lead to any specific recommendations in a specific situa-
tion. But it can lead to insights of general nature. These insights can then be used
by policy makers in reaching specific decisions or in formulating general poli-
cies” (Aumann and Maschler 1995, 1). Deductions from the model analysis are
presented in the following chapters.
The following model tries to capture the sourcing decision determinants
identified in the previous chapters122. To satisfy the requirements of analytical
modeling as well as the method of agent-based simulations, we will conduct a
bottom-up formula-based approach for deriving the model, as it was done e.g. in
(Beck et al. 2003; Hoppen et al. 2003; Weitzel 2004).

122
The model has been discussed and endorsed by the Doctoral Consortium of IRMA International
Conference 2005 in San Diego (Beimborn 2005; Best Doctoral Submission Award) and on the
39th Hawaii International Conference on System Sciences 2006 (Beimborn 2006).
Developing a Formal Model for Cooperative Sourcing 223

4.2 Derivation of the Cooperative Sourcing Model

(CSM)
4.2.1 Actors and Business Functions
Consider an economy that runs during periods t = 0, 1,.., T. In an initial period
t=0, the economy is populated by a finite number of actors which represent
members of a particular industry or market sector. Because the financial industry
is the application domain of this work, one could, e.g., imagine the actors to
represent the banks of a domestic finance industry or simply the German associa-
tion of public savings banks. The banks are identified by a unique index i. Each
bank is handled as an autonomously acting and monolithic entity; decisions are
not disturbed or thwarted by its own management staff. Further, it is assumed
that all banks have unrestricted access to perfect capital markets; consequently,
they can obtain risk neutral decision behavior by hedging risks according to their
risk preferences (DeAngelo 1981; Ewert and Wagenhofer 2003) (cf. section
2.1.3.2, footnote 14).
Furthermore, we assume a finite set of capabilities to be existent in the mod-
eled market sector, which are represented by more or less modular business func-
tions. The model handles those as the core decision object of cooperative sourc-
ing and therefore follows a activity-based view as already discussed in the theory
chapter (2.1) and as also adopted by other empirical works (Lacity and Willcocks
2003) or formal models (Knolmayer 1993; Lammers 2004)123.
Each bank i utilizes a subset of these business functions to fulfill its busi-
*
ness. Vector ai – consisting of binary variables aik – indicates if business func-
tion (or activity) k is part of bank i’s business or not. It is assumed that the bank’s
*
business model does not change over time; consequently, ai does not either.
Divestments or development/acquisition of new business functions is not within
the scope of the model.

123
Although some of these works adopt this view for a particular application domain (IT activities),
the structure of these models is so generic that they could easily be adopted for modeling business
functions in general.
224 Developing a Formal Model for Cooperative Sourcing

i
Figure 93: Banks and business functions (example)
i, j actor (bank) indices
i, j  I resp.
I, |I| set and total number of actors
i, j = 1... |I|
k, l business function indices
set and total number of business functions k, l  K resp.
K, |K|
available in the economy k, l = 1... |K|
indicates (binary variable), whether actor i
a ik  ^0;1`
runs business function k
*
ai actor i's vector of business functions
*
ai ^a ,..., a
i1 ik ,..., ai K `
Table 30: Basic indices, sets, and indicators
Cooperatively sourcing these business functions is the core decision object
of the scenario. Each bank can decide about re-locating the execution of the
activity by deciding either to operate it in-house on its own or it can be coopera-
tively sourced within an alliance, together with other banks. As argued in chapter
1, cooperative sourcing (and this model) focuses on primary banking activities,
assuming that, because (potential) sourcing providers have to be core-competent
regarding the provision of the sourcing object (i.e., the business function), (po-
tential) sourcing providers for banking processes are primarily banks themselves.
Consequently, the model will not cover third-party sourcing providers and their
decision behavior; it is sufficient to model banks similarly as outsourcers and as
potential insourcers. Nevertheless, the model is developed in a way that it can
easily be extended in order to capture those industry-external players.
Developing a Formal Model for Cooperative Sourcing 225

In order to make effective decisions, the particular characteristics of busi-

ness functions, identified in the literature and in empirical studies, have to be
taken into account. The essential aspect is given by the structure of process costs
of a particular business function. Different alternatives are possible here, but,
because there is quite little research undertaken in the banking domain (cf. sec-
tion 2.1.1.3), the model adopts a linear cost function for every business function
in each bank. It is a common approach to utilize functional structures as simply
as possible for reducing the complexity, as long as there is no significant empiri-
cal evidence that it does not describe reality sufficiently. For example, (Barua et
al. 1991; Gal-Or 1983; Quan et al. 2003; Thatcher and Oliver 2001) also use
linear, output-based cost functions in their analytical models on determining a
firm’s optimal IT investment.
The assumption of equal process cost structures in different banks is based
on (Lamberti and Pöhler 2004, 17). The fixed costs also include efforts for man-
aging and controlling the operations of the activity. Any business function has a
measurable output xik which determines the resulting level of process costs. All
of these parameters are assumed to be constant over time as usual in this kind of
formal modeling (e.g., Anderson and Parker 2002).
fixed costs and variable unit
K ikF , cikP costs of business function k in
firm i
output of business function k
xik in firm i
process costs of business
CikP with C ikP K ikF  cikP ˜ xik
function k at firm i
Table 31: Process costs
In the literature review, it was shown that the strategic argument for focus-
ing management to the firm’s core competencies has always been an important
driving force towards outsourcing. Core competence in a particular business
function is at least partially represented by process cost advantages compared to
the competitors. As argued in the empirical section 3.6.3.2, this holds true espe-
cially for repetitive activities such as back-office functions or transactional proc-
esses (e.g. payments, securities), which are the primary objects of cooperative
sourcing decisions in reality. Nevertheless, cost advantages do not sufficiently
reflect core competencies and strategic value. Competitive advantage might also
result from having an especially distinctively valuable customer base or partner
network, higher process quality, or other factors which argue for the need to
implement an additional parameter. Therefore, the model additionally contains a
competence measure Oik for every business function, which actually only repre-
226 Developing a Formal Model for Cooperative Sourcing

sents “residual” core competence and, thus, the strategic value, not reflected by
process cost advantages. This measure is parameterized by a value between 0
and 1.
As all business functions are embedded in one or more business processes of
the firm, the interdependencies between them have to be captured by the model.
Business processes demand different resources and capabilities and request out-
put of different business functions. Actually, a business process is a sequential
interconnection of several business functions (Bruch 1998, 68f; Meyer and
Schumacher 2003) because the performance of one piece of work depends on the
completion of other defined pieces of work (Van der Vegt et al. 1998; Wybo and
Goodhue 1995). Outsourcing one or more business functions will increase the
risk of breaking down the business process and potentially reduce business per-
formance, owing to inflexibilities and poor responsiveness to the market (Bahli
and Rivard 2003; Lacity et al. 1996). Moreover, activities may utilize shared
resources or indivisible skills (Hitt et al. 1993; Lei and Hitt 1995) or there may
be such a great demand for interaction between some activities that selective
outsourcing is not a valid option. The stronger the interdependence, the higher is
the need for coordination, joint problem solving, and mutual adjustment, which
may impede cost control (Earl 1996). Several authors define activities as “sys-
temic” if they are closely interrelated with other activities of the firm (Langlois
and Robertson 1992; Roy and Aubert 2002).
Interdependencies will not only appear in an operational sense (although
these are the most obvious) but also on a strategic layer: the alignment (Hender-
son and Venkatraman 1992) and “extent of unstructured technical dialogue”
(Monteverde 1995) between different business functions are an essential part of
the overall task interdependence because they encourage organizational learning
(Dibbern et al. 2003; Lei and Hitt 1995) and allow for a comprehensive and more
integral consideration of configuration and optimization issues at the overall
business level.
To represent this relationship the model introduces a measure for the busi-
ness function interdependence. Tikl describes how difficult it is to divide two
functions k and l organizationally from each other. If the interdependence meas-
ure is near to one, there is an almost complete dovetail between both (1 would in
fact mean that it would be one single business function), while, if the value is 0,
there will be either no interaction between them within any business process or
the interfaces between them are well-defined and automated in such an efficient
way, that it is not relevant to interacting if it takes place intra- or interorganiza-
tionally. This measure corresponds to the task interdependence measure of
Knolmayer’s outsourcing decision model (Knolmayer 1993, 78).
Developing a Formal Model for Cooperative Sourcing 227

Figure 94: Degree of task interdependencies

(not all relationships are labeled)
Again, it should be noted that the model adopts an activity-based perspec-
tive, not a process or product perspective. If activities are tightly interrelated
within several business processes or exploited for the production of different
products (leading to product-based economies of scope, cf. section 2.1.1.1) this is
only implicitly captured by the concept of task interdependencies. Because the
model defines a process cost function for each activity, economies of scope can-
not be expressed by the production (cost) function itself. Instead, they become
apparent in the form of interfacial costs (based on the task interdependence
measure) if one of two business functions is being outsourced (i.e. economies of
scope getting lost), as realized in other models as well (e.g. Knolmayer 1993).
(Residual) core competence of
Oik
business function k at firm i
Interdependence between business
Tikl function k and l in firm i
Tikl = [0.0;1.0[ defined for l>k

Table 32: Core competence and task interdependence

4.2.2 Business Neighborhood
As discussed above, cooperative sourcing alliances can be characterized as co-
opetive (Brandenburger and Nalebuff 1996, cf. section 2.1.8).
The business neighborhood, describing the degree of competition or rivalry
between two firms, is incorporated into the model as a determinant for strategic
risk resulting from coopetition. Outsourcing involves different kinds of risk for
228 Developing a Formal Model for Cooperative Sourcing

the firm, which might be even higher in coopetive relationships. For example,
“revealing confidential information and bringing an outsourcing partner into a
company’s core business processes inherently reveals part of a company’s strat-
egy. A breach in confidential strategic and competitive market information can
come from differences between the two companies’ corporate culture, ethics, and
governance and can threaten the outsourcer’s competitive advantage and
heighten strategic/market risk” (Beasley et al. 2004, 26) (for the threat of confi-
dentiality problems see also (Lacity and Willcocks 1995, 238)). Further, com-
petitors might be strengthened by an unplanned transfer of technology and know-
how (Nueno and Oosterveld 1988, 17) or the insourcing competitor might exploit
an increasing asymmetric dependence that the outsourcer faces (Afuah 2000,
388; Oliver 1990; Ring and Van de Ven 1994) (cf. section on agency theory
(2.1.3)).
Therefore, the model includes a parameter describing business neighborhood
bnij with bnij = [0.0; 1.0], which is higher, the higher the competitive degree
between actors i and j is.

Figure 95: Business neighborhood

The business neighborhood is determined by three different factors: similar-
ity of customer portfolios bnijcust , similarity of product portfolios bnijprod (loans,
securities and funds administration, payments processing, deposit and giro busi-
ness etc., cf. section 1.5.4) and overlap of geographical domains bnijgeo (based on
Porter’s “competitive scope dimensions” and “industry segmentation parame-
ters” (Porter 1985, 233+238)). For example, two public savings banks have the
same product portfolio and the same customer segments but usually are serving
disjointed geographical markets. Therefore, their degree of competitive
neighborhood is bnij = 0.0. On the other hand, Deutsche Bank and Commerzbank
also have quite similar (customer and product) portfolios and additionally the
same national (or, even more, international) focus, therefore their bnij is near to
one.
Developing a Formal Model for Cooperative Sourcing 229

In order to determine these values consistently throughout the whole mod-

eled market segment, the neighborhood values cannot be determined independ-
ently of each other because they describe relations between (and not attributes
of) the firms. To determine bnij, firstly, the participating actors are positioned in
a virtual product and customer space. For the product portfolio we assume that
similar business functions result in similar products. Therefore, the more similar
*
the business function vectors a i of two particular banks (cf. Table 30), the more
similar is the product portfolio. For determining the degree of similarity bnijprod ,
* *
the cosine value of the angle between both vectors a i and a j is taken124. Be-
cause an angle between two vectors consisting of binary values is equal to or
smaller than 90°, the resulting cosine value is between 0 and 1125. For determin-
ing likeness in the customer segments the same method is used; instead of the
cust
vector of business functions, we define a customer segment vector bnij con-
sisting of three binary values for representing retail banking, sme (small and
medium enterprises) banking and large investors banking, reflecting a segmenta-
tion of the banking business common in reality (cf. section 3.2.2.1). To describe
a geographical overlap, bnijgeo is simply set as a binary parameter to 0 or 1.
For determining the overall business neighborhood bnij from these values,
the three parameters have to be unified. Due to the fact that different dimensions
of business neighborhood are captured here, the arithmetic average cannot be
applied. Instead we will apply the average operator from fuzzy set theory126,
which takes the minimum of the three values (Zadeh 1965). Adopting this con-
cept, it becomes clear that the degree of competition is based on the “flimsiest”
influencing factor: e.g., if there is no overlap in product portfolio, customer port-
folio similarity and geographical overlap can be as high as possible without hav-
ing any competition. The resulting computation of business neighborhood is
described by Equation 4.

* * * *
* * a ˜b a1 ˜ b1  a 2 ˜ b2  ...  a n ˜ bn
124
 
cos a , b * * (= scalar product of any vector a and b )
a˜b a12  a 22  ...  a n2 ˜ b12  b22  ...  bn2
125
This measure was applied in the context of social network analysis (Alstyne and Brynjolfsson
1997) and in information retrieval (Salton 1971), for example. For a more detailed discussion of
similarity measures, see (Jones and Furnas 1987).
126
Fuzzy set theory, which comes from Zadeh, modifies traditional set theory by defining an ele-
ment’s membership within a set using a membership function instead of a binary value. Thus, an
element belongs to a certain set with a particular degree of between 0 and 1 (Zadeh 1965).
230 Developing a Formal Model for Cooperative Sourcing

bnij 
min bnijgeo , bnijprod , bnijcust 
* * * *
with bn geo
ij  ^0;1`, bn prod
ij cosai , a j , bnijcust cosci , c j 
Equation 4: Business neighborhood
bnij Business neighborhood between i and j
geo Geographical business neighborhood between
bn ij bnijgeo  ^0;1`
i and j
* *
bnijprod Product portfolio overlap between i and j bnijprod cosai , a j 
* *
bnijcust Customer portfolio overlap between i and j bnijcust cosci , c j 
*
ai Bank i's vector of business functions
*
* cj retail, sme, large
cj Vector of customer segments served by i
with retail, sme, large ^0;1`
Table 33: Components of business neighborhood
4.2.3 Cooperative Sourcing
The outsourcing process in this model is represented by taking out the considered
business function with all of its resources and transferring it to another bank in
the coalition. The outsourcing decision ensures optimization of the trade-off
between process cost savings and transaction costs (Apte 1990; Cheon et al.
1995; Holzhäuser et al. 2005, 114; Lacity and Hirschheim 1993a). The new
“source” of activity execution is identified by the index combination km, whereas
k stands for the business function and m is the index of the execution location for
this particular business function type. m=0 represents in-house production and
not joining a coalition. The maximum value for m represents the present number
of existing coalitions for cooperatively sourcing the related business function k.
To simplify the formal representation of the model, we introduce a variable
mktmax as that maximum number of k-related coalitions in period t. Mkmt repre-
sents the set of firms (i.e., cooperative sourcing alliance) which get their k-th
business function provided by the m-th sourcing coalition in period t. Neverthe-
less, one member has to become the insourcer, who physically insources the
process volumes of all other members127.

127
Actually, the model indicates exactly one insourcer for each coalition. Otherwise there would be
multiple coalitions.
Developing a Formal Model for Cooperative Sourcing 231

Figure 96: Cooperative sourcing alliances

The outsourcing relationship is formally described by several binary vari-
ables. First, yikmt = 1 indicates that the i'th firm joins alliance km in period t. In
the following periods, the business function is provided by alliance km, indicated
by zikmt = 1 for each of these following periods t. Finally, binary variable vikmt
determines if bank i is the insourcer of the process volume of alliance km in
period t.
km Index of the m-th coalition providing business function k
max
m kt Number of coalitions serving business function k in period t

M kmt , M kmt Set of coalition members and size of coalition km in period t

y ikmt  ^0;1` Decision variable: bank i outsources business function k to alliance km in period t

z ikmt  ^0;1` Bank i's business function k is operated by coalition km in period t

vikmt  ^0;1` Bank i is insourcer of coalition km in period t

Table 34: Indices, sets, and binary indicator variables

representing cooperative sourcing
4.2.3.1 Transaction Costs Related with Cooperative Sourcing
The following transaction cost categories are based on Albach (1988), who de-
fines a classification of transaction costs (cf. section 2.1.2), and on Loh (1994),
who provides a comprehensive model of costs associated with IT outsourcing.
232 Developing a Formal Model for Cooperative Sourcing

According to (Helper and Levine 1992; Ring and Van de Ven 1994), our
model assumes that the banks participating in a cooperative sourcing alliance
establish an efficient relationship that facilitates high commitment relations
(Helper and Levine 1992) and “produces efficient and equitable solutions to
conflicts” (Ring and Van de Ven 1994, 92). Of course, this is a quite rigid as-
sumption, compared to reality. But, nevertheless these “efficient solutions” are
not costless. Agency problems have to be handled and mechanisms for coordi-
nating and monitoring the partnership have to be established, resulting in transac-
tion costs. The different types of transaction costs relevant to this situation are
adoption costs, negotiation costs, coordination costs, interface costs, and agency
costs. Those are integrated into the model in the following.

4.2.3.1.1 Negotiation Costs

Negotiation costs have to be borne in the initial period of the sourcing deal. The
negotiation costs of a multi-member cooperation depend mainly on the complex-
ity of the negotiation object (i.e. the business function) (Loh 1994), on the num-
ber of members (Beimborn et al. 2004), and on the level of experience the com-
pany had with similar processes in the past, due to learning effects (Tsang 2000).
While complexity measurement is already a very difficult task in physical
production processes (Piller 2000), there are very few suggestions for how to
measure the complexity of business functions (e.g. Piller 2000; Venkatraman
1994, 81; Wood 1986). Our model will use the process cost structure as a proxy
for complexity. Because financial processes regularly do not produce physical
goods, the only applied inputs are IT and HR, but no raw materials which could
strongly differ in price. Because IT and HR are quite homogenous resources
from a cost perspective128, we decided to use process cost parameters as a proxy
for process complexity Fik. Both cost parameters K ikF , cikP are normalized to a
value between 0 and 1 (by dividing them by the corresponding overall maximum
values on a logarithmic base (cf. Equation 5).
§ ·
¨
¨
 
ln K ikF

 
ln c ikP ¸
¸  
§ ln K ikF   ·¸
ln cikP
§ · § · ¨ 
¨¨ ln¨ max ( K ik ) ¸ ln¨ max(cik ) ¸ ¸¸
F P

© © i , k ¹ © i , k ¹¹ ©  
¨ ln K F   ¸¹
ln c P
F ik
2 2
Equation 5: Task complexity

128
Evidently, significant cost differences still exist for the employed resources, e.g. credit adminis-
trator vs. investment banker or high performance transaction systems vs. the “notebook infrastruc-
ture” of a mobile sales representative.
Developing a Formal Model for Cooperative Sourcing 233

Based on (Beimborn et al. 2004), the model assumes total negotiation costs
for managing a coalition to be progressively increasing with the number of
D
members M kmt , i.e. M kmt with D > 1. When broken down to the single mem-
129
ber , this implies a declining function as long as D  2 , which is a valid as-
sumption.
While negotiation costs will increase with the size of the coalition, they de-
crease with increasing outsourcing experience oit of the firm (Cross 1995; Lyles
1988; Tsang 2000), because “joint venturing” can be considered as a skill which
is improved with use (Hirschman 1985). Experience with cooperative sourcing is
operationalized by the total number of sourcing contracts of the bank in former
periods. According to (Ewert and Wagenhofer 2003), the model assumes a de-
creasing impact of the learning effect.
N
The different factors are integrated into the negotiation cost function Cikmt
by multiplying them with a negotiation cost base parameter ckN .
D 1
˜ oit  1
N E
C ikmt c kN ˜ F ik ˜ M kmt with 2 ! D ! 1 and E  0
Equation 6: Negotiation cost function
§ ·
¨ lnK ikF  lncikP  ¸
¨  ¸ § lnK ikF  lncikP  ·
Complexity of the ¨¨ ln§¨ max( K F ) ·¸ ln§¨ max(c P ) ·¸ ¸¸ ¨  ¸
F ik business function ik ik ¨ lnK F  lnc P  ¸
F ik © © i,k ¹ © i,k ¹¹ © ¹
2 2

Number of outsourc-
ing projects firm i cN
oit k Negotiation cost basis
realized up to period
t-1
Table 35: Negotiation cost factors
4.2.3.1.2 Coordination Costs
Subsequently, when the sourcing contract is already functioning, the ex post
costs for coordination take place. These “result from the need to maintain a
greater and more diverse repertoire of cognitive maps, behavioral routines, and
organizational resources for engaging in both cooperative and competitive be-
havior” (Lado et al. 1997, 124). Change requests, renegotiation or further defini-

129
Therefore, we divide the expression by the number of coalition members:
D D 1
M kmt / M kmt M kmt
234 Developing a Formal Model for Cooperative Sourcing

tion of contract items also result in coordination costs during the relationship.
Outsourcers may leave particular items of the contract open to reduce dependen-
cies (Elitzur and Wensley 1997).
Again, coordination costs are expected to rise with the complexity of the
business process (Rouse and Corbitt 2004) and with increasing number of parties
(Olson 1965). As a coalition becomes larger, the “members’ cost-benefit equa-
tion increasingly favors nonparticipation in tasks that create collective benefits”
(Aram 1989, 273).
Coordination costs are modeled quite simply as a relative part J of the ne-
gotiation costs.
C N
Cikmt J ˜ Cikmt with 0  J 1
Equation 7: Coordination cost function
One can argue that the more efforts a firm expends on negotiation the less it
has to expend on ex post coordination, resulting in a reciprocal relation between
both cost categories. But, because we assume efficient relationship management,
the optimal trade-off between negotiation costs and ex post coordination costs
has already been determined. The positive relationship between negotiation costs
and coordination costs is just drawn for reasons of simplification, standing for
the fact that coordination costs rely on the same factors as negotiation costs.
Even in an efficient sourcing partnership, coordination costs are unavoidable
because permanent coordination is necessary to provide a long-term and success-
ful partnership in B2B relationships (Buvik and Gronhaug 2000; Zhang and Liu
2005, 54) (cf. section on relationship theories (2.1.8)). Moreover, TCE assumes
that negotiation between cooperating parties cannot be sufficiently realized ex
ante (Holzhäuser et al. 2005, 114; Williamson 1985, 29).

4.2.3.1.3 Adoption Costs

When it comes to the cooperative sourcing of a particular business function,
success partly depends on the degree of similarity (or standardization) between
the business functions of different firms (Wüllenweber, Beimborn, and Weitzel
2008). The model follows the best-of-breed argument by assuming that the merg-
ing of business functions requires process standardization (Cash and Konsynski
1985; Rouse and Corbitt 2004) – i.e. every member of the alliance accepts the
insourcer’s process – and furthermore, reaching this level of complete standardi-
zation is possible in any case. Adoption costs cover all efforts being necessary in
the initial period to adopt the provider’s process and to incorporate it into the
own business function landscape.
Developing a Formal Model for Cooperative Sourcing 235

The model introduces a similarity degree parameter ]kijt (with ]kijt =[0;1])
which describes the similarity of two business functions k at bank i and bank j in
period t (]kijt = 1 expresses perfect similarity). The higher the similarity between
two business functions, the lower the costs for connecting its own business to the
business function, which is now externally provided. The model assumes that
when a bank has adopted the reference process of the coalition’s insourcer, the
degree of similarity to all remaining actors outside the cooperation shifts to the
corresponding values of the insourcer’s business function. Furthermore, the in-
ternal similarity degrees (between the alliance members’ business functions) are
set to one (and will remain so, if a member leaves). Only if a new bank enters the
coalition and immediately becomes the insourcer, external similarity degrees of
the entrant will be adopted by all members. Consequently, the similarity degrees
will both vary over time and converge into one130.
While the degree of task interdependencies between business functions can
be represented by a vertical relationship (e.g. along a particular business proc-
ess), similarity degree represents a horizontal relationship between equivalent
business functions of two different firms (Figure 97).

Figure 97: Similarity degree (neglecting period index t)

130
Some models make a case that increasing standardization results in decreasing specificity and,
thus, strategic value (e.g. Lammers 2004). In contrast, our model assumes no interrelation be-
tween similarity degree and basic specificity (represented by the residual core competence meas-
ure Oik).
236 Developing a Formal Model for Cooperative Sourcing

AD
For determining the resulting adoption costs Cikmt , which will only occur in
the period when the business function gets outsourced, the model incorporates a
cost function (Equation 8), which is based on the similarity degree ]kijt, on task
complexity F ik (defined above in section 4.2.3.1.1), and on a cost factor base
ckAD , differing for each type of business function. Adoption costs only occur for
the outsourcers but not for the insourcer of a cooperative sourcing coalition. ĵ
represents the insourcer firm.
AD
C ikmt 1  ] ˜ F
kiˆjt ik ˜ c kAD
Equation 8: Adoption cost function
4.2.3.1.4 Interface Costs
Another cost element, which accompanies a sourcing relationship, is interface
costs, covering all costs ensuring the operational interplay between in-house
operated processes and outsourced business functions, such as the maintenance
of technical interfaces, data transfer, human interaction, as well as the loss from
activity-based economies of scope (cf. section 2.1.1.1 and 3.6.2.3), leading to a
decline in performance (Gulati and Singh 1998; Pondy 1970). Thus, interface
costs can be ascribed to both theoretical perspectives of production cost econom-
ics and transaction cost economics. Interface costs are mainly determined by
functional interdependencies. The higher the interdependencies, the higher are
the corresponding periodical interface costs. We assume cost symmetry, i.e. costs
do not depend on which one of two business functions is outsourced.
Maintaining and operating interfaces can require huge efforts and task inter-
dependencies are systematically under-estimated by practitioners (Bahli and
Rivard 2003; Earl 1996; Langlois and Robertson 1992). Interface costs between
two business functions k and l will occur only if both business functions are
executed by different firms, but neither will occur if both business functions are
run internally or if they are being provided by the same insourcer. In order to
consider this relationship, we apply binary variable zikmt, which indicates whether
i's business function k is provided by alliance km in period t, and binary variable
vjkmt, which indicates that bank j is the insourcer of alliance km in period t. Taken
together, the product indicates whether bank j operates business function k for
bank i. Following, the expression v jkmt z ikmt  v jlmt z ilmt will be equal to 1 only
if both business functions are operated by different banks, otherwise it is 0. This
value is multiplied with the task interdependence measure T ikl introduced above
Developing a Formal Model for Cooperative Sourcing 237

(cf. section 4.2.1) and an interface cost basis c IF . Finally, the whole is multi-
plied with binaries aik and ail (cf. Table 30), considering whether k and l are part
of i’s business function portfolio at all.
m ktmax
CitIF ¦¦ ¦¦ a ik ˜ ail ˜ T ikl ˜ c IF ˜ v jkmt zikmt  v jlmt zilmt i, t
k  K l K m 0 j I
l !k j zi

Equation 9: Total interface costs for bank i in period t

C itIF i, t Negotiation costs firm i has to bear when taking part in coalition km in period t
Binary variable which indicates
Binary variable which indicates if i is
vikmt if i is insourcer of coalition km zikmt
member of coalition km in period t
in period t

Degree of task interdependence

T ikl
between k and l of bank i
c IF Interface cost basis

Table 36: Interface cost function and factors

4.2.3.1.5 Agency Costs
Agency costs have been defined in section 2.1.3.1 as the sum of the principal’s
monitoring costs, bonding costs, and the residual loss. The model assumes that
agency costs are related to both task complexity F ik (Cheon et al. 1995; Nam et
al. 1996) and the strategic value of the regarding business function (core compe-
tence measure Oik). Based on agency theory (section 2.1.3) and theory of incom-
plete contracts (section 2.1.4), task complexity is used as a proxy for problems of
measurability and as a driver for inherent contract incompleteness, thus driving
control costs (and risks of not even being able to adequately control the pro-
vider).
Agency costs in this model further depend on the degree of competition be-
tween the coalition members. The higher the competitive part in a coopetition
(for which the competitive neighborhood value between a particular coalition
member and the insourcer ĵ will be given: bniˆj bnikmt ) the higher are the
strategic risk of dependencies being exploited by the coopetitor and the agency
costs to reduce it. The impact of business neighborhood on agency costs will be
limited to a certain proportion bne (bn effect) because outsourcing to a non-
competitor would also induce agency costs (Equation 10).
The effect of business neighborhood bniˆj , task complexity F ik , and strate-
gic value Oik will be monetarized by an agency cost factor base ckAG .
238 Developing a Formal Model for Cooperative Sourcing

ckAG will differ for different business functions k because those are different
regarding their “sensitivity” (more or less customer data is involved etc., for
example).
AG
Cikmt    
ckAG ˜ 1  1  bniˆj bne ˜ F ik ˜ Oik
Equation 10: Agency costs
Once more, it must be argued that agency costs, as explicated by this model,
only represent the difference in agency costs between an intra-organizational
(hierarchy) and an inter-organizational relationship. Obviously, agency costs also
appear if the business function is processed in-house and the firm’s own manag-
ers have to be monitored and motivated to behave cooperatively. Because bnij is
defined to be 0 for i=j, the insourcer does not bear additional agency costs.

4.2.3.1.6 Summary
The total transaction costs TC consist of adoption costs and negotiation costs in
the initiation period of the sourcing relationship W, and of coordination costs,
interface costs and agency costs in the subsequent periods.
TC ikmt W
AD
Cikmt N
 Cikmt and TCikmt !W  CiktIF  Cikmt
C AG
 Cikmt
Equation 11: Transaction cost function
4.2.3.2 Process Cost Effect
In this model, the reason for cooperative sourcing is achieving economies of
scale (through bundling of process volumes) and economies of skill (through
access to dominant process knowledge, i.e. a dominant process cost function of
the insourcer) (Grover et al. 1994; Teng et al. 1995). If several banks decide to
merge their process volumes for a certain business function, one bank becomes
the insourcer and adds the partners’ volume to its own output. For simplification,
the model assumes that the process cost function of the insourcer will remain
stable, although there may be several reasons for a change or shift – either nega-
tively because the bank has to increase capacity, or positively, because the in-
vestment in a new technology, which is efficient for larger process volumes,
becomes favorable. In the model, a coalition will be founded by several banks
with the cost-minimally producing partner becoming the insourcer (i.e., the bank
which can provide cost minimal processing of the aggregated process volume).
M
The total output of alliance km in period t is xkmt ¦ xik . The model assumes
iM kmt

that no market mechanisms (e.g. bidding) take place, i.e. banks do not mutually
Developing a Formal Model for Cooperative Sourcing 239

make and accept offers, considering the costs and surcharges (producer surplus).
PM
Instead, the resulting process costs Ckmt will be allocated to the several members
of the alliance, by applying a certain pre-determined allocation mechanism (de-
scribed by Cikmt
G
) that all members must agree on. In the next chapter, different
allocation mechanisms, such as equal distribution of gains, volume-proportional
cost allocation, or the Shapley allocation (Shapley 1953), will be game-
theoretically tested (section 5.1) in order to determine if they generally lead to
stable coalitions.
Equation 12 formally demonstrates the selection of the optimal insourcer.
§ § ··
PM
Ckmt : min ¨ CikP ¨¨ ¦ xik ¸¸ ¸ min CikP xkmt
iM kmt ¨ ¸ iM kmt
M
  
© © iM kmt ¹ ¹
Equation 12: Process costs of the cooperative sourcing coalition
The total process costs of the coalition have to be completely allocated to the
coalition’s members:
PM G
C kmt ¦C
iM kmt
ikmt

Equation 13: Complete allocation of total coalition process costs

PCSikt describes the process cost savings for bank i's business function k in
period t.
PCSikt CiktP  Cikmt
G
PCSit ¦ PCS ikt
kK
Equation 14: Actor i's process cost savings per business function k and in total
4.2.3.3 Decision Calculus
The decision calculus of a bank to take part in a cooperative sourcing coalition is
based on the comparison of process cost savings (economies of scale and skill) to
transaction costs (including diseconomies of scope in form of interface costs)
(Clemons et al. 1993). In order to evaluate an alliance membership the bank has
to determine the present value of its individual net benefit 3 ikmt * for the sourc-
CoSo
ing contract duration T . The objective does only take monetary arguments
into account131.

131
Some models (e.g. Tsang 2000, Lammers 2005) include the “strategic value” of a decision in the
agent’s objective without explicitly stating how to handle the monetarization. Our model tries to
240 Developing a Formal Model for Cooperative Sourcing

As previously mentioned, the model assumes a risk-neutral decider, optimiz-

ing the expected discounted outcome of the sourcing decision. Operational risks
and environmental uncertainty are taken into account by adopting a risk-adjusted
discount rate rad132.
Figure 98 clarifies the composition of the decision function; the resulting
benefit function is presented in Equation 15.

firm’s decision function

t* = sourcing contract period

transaction costs
t t t*

process cost interface costs agency costs coordination costs

t > t*
savings

t = t* adoption costs negotiation costs

Tikl ]kij Oik Kik, cik Fik bnij oit Mkm

task standardization core process cost task business learning cooperation
interdependence degree competence parameters complexity neighborhood effects size

Figure 98: Decision calculus

prevent this critical issue by solely incorporating cost aspects, as done in many other models (e.g.
Aksin et al. 2004, Knolmayer 1993). This limitation is discussed in section 6.4.
132
The reconsideration of uncertainty within the decision calculus presents a major issue. rad might
be determined via the CAPM (Brealey and Myers 1996), if investment projects with identical risk
structures could be valued by the capital market (Lammers 2004). Nevertheless, the outcomes of a
decision and the associated probabilities must be known. This is a quite rigorous assumption and
furthermore reduces the multidimensional risk construct to an expected value of cost outcomes
“and the problem becomes a straightforward cost trade-off” (Jurison 1995, 243). For a decision
model on IT outsourcing explicitly taking risk into account, nevertheless also in a simplified
qualitative way, see e.g. (Jurison 1995).
Developing a Formal Model for Cooperative Sourcing 241

t * T CoSo 1
3 ikmt * ¦ NPV
W
(cost effect )
,rad
W t*
CoSo
T
§ 1 · ½
¦ ¨¨ C x   C
P G
x M
, xkm˜ ¸ ¾production cost effect
W 1 ©
ik ik ikm t W  ik
1  rad W ¸¹ ¿
N AD ½
C ikmt  Cikmt ¾transaction cost effect in t t *
¿
T CoSo
§ 1 · ½
 ¦ ¨¨ C C
 CikIF(t W )  Cikm
AG
( t W ) ˜  ¸ ¾transaction cost effect in t ! t *
W 1 ©
ikm ( t W )
1  rad W ¸¹ ¿
Equation 15: Inter-temporal monetary benefit of an outsourcing agreement
NPV represents the net present value function, which discounts the cost ar-
IJ,rad

gument from period W to 0 with risk-adjusted calculation discount rate rad.

The first term represents the process cost savings for all periods of the sourc-
ing relationship (beginning with the period after contract closure). The second
term contains transaction costs which occur in the period of contract closure t*
(negotiation costs and adoption costs) and the last term consists of transaction
costs occurring in the following periods133.

4.2.3.4 Constraints
Aside from the decision calculus, the model assumes several constraints which
restrict the banks’ decision scope. First, it is assumed that the process volume
cannot be split and allocated to several sourcing locations. This is, of course, not
a binding constraint, as long as there are no capacity restrictions134.
mktmax

¦z
m 0
ikmt 1 i, k , t

Equation 16: Single sourcing constraint

Second, as argued previously, after joining a cooperative sourcing alliance,
this particular bank is bound to this decision for TCoSo periods. Afterwards, if it
decides to stay in the coalition after re-evaluating its membership, it is again
bound for the same amount of time. In order to formally represent this constraint,

133
Basically, it must be assumed that all costs included in the decision calculus are identical to cash
outflows within the same period, since cash outflows are the base for investment decisions.
134
Capacity constraints are not considered in this model, assuming existing opportunities for acquir-
ing all necessary capacities. Because only production of information goods is taken into account,
this assumption is acceptable.
242 Developing a Formal Model for Cooperative Sourcing

another binary variable uikt has to be introduced, which becomes equal to 1 if

bank i evaluates cooperative sourcing of business function k in period t.
t 1
uikt d 1  ¦y ikmW
and u ikt t y ikmt i, k , m, t
W t T CoSo
Equation 17: Decision to cooperative sourcing bounded to TCoSo periods
Third, the model assumes a bounded decision capacity. A bank cannot
evaluate cooperative sourcing of more than DC of its business functions within a
certain period.

¦u ikt d DC with DC  i, t

kK
Equation 18: Bounded decision capacity
Fourth, as noted above, decision makers have to take the strategic value of
the several business functions and the encapsulated resources and capabilities
into account. Unlike physical assets, competencies are enhanced as they are
applied and, in contrast, they fade if they are not used (Prahalad and Hamel
1990). Therefore, a firm has to ensure that strategically relevant business func-
tions, which contain core competencies, remain inside the organization (cf. sec-
tion on resource-based view and core competence view (2.1.6)). In concordance
with Knolmayer’s formal outsourcing model (Knolmayer 1993), we do not in-
corporate the strategic impact of cooperative sourcing into the decision function
(because it would have to be monetarized in this case), but we assume the resid-
ual core competence, defined in section 4.2.1, to be considered by a strategic
constraint, which restricts the “amount of strategic value” to be outsourced (=
SCi). In contrast, the insourcer of a cooperative sourcing alliance is not affected
by this constraint.
m ktmax

¦ ¦z
k K m 1
ikmt ˜ Oik d SCi  i, t

Equation 19: Strategic constraint

Finally, there might be different reasons for a bank not to outsource a par-
ticular business function which are not captured by the decision function. For
example, in order to account for sales being the intended future core business of
many banks, the model needs a constraint which hinders outsourcing of this
particular business function. The relevant decision variables can be restricted to
0. Nevertheless, although this business function is not explicitly part of the deci-
Developing a Formal Model for Cooperative Sourcing 243

sion problem anymore, it has still to be considered because of the task interde-
pendencies and their impact on interface costs.
uikt 0  k, t zik 0t 1  i, k , t
Equation 20: Explicitly hindering outsourcing of business function k of i

4.3 Centralized Model: Global Optimization

At this point we will introduce a differentiation between centralized and decen-
tralized coordination of the modeled system of banks, business functions, and
cooperative sourcing alliances. Due to externalities, complexity, and bounded
rationality, the actors themselves are expected not to be generally capable of
identifying the optimal cooperative sourcing configuration (i.e. decentralized
coordination). For the purpose of determining resulting inefficiencies, a central
planner, who has complete knowledge about all cost functions and parameters,
can be virtually introduced and assigned to determine the optimal system-wide
sourcing configuration in contrast to the outcome resulting from autonomously
deciding banks. Individuals will always decide in favor of maximizing their local
interests, which does not necessarily lead to an optimal configuration of the
whole system as long as externalities are present. In another context, this ap-
proach was chosen by Weitzel (2004)135.
A global optimization objective is derived by summing up Equation 15 (p.
241) over all actors and business functions and integrating the binary decision
variables (yikmt) (Equation 21). This optimization will ensure maximum cost
savings in the overall system but not necessarily lead to positive net savings for
each bank in the system. Necessary redistributions of savings are not part of this
central model but can be established afterwards to attract all members to make
the globally optimal contribution to the system. In terms of game theory, we can
argue that optimizing this global objective will determine the largest pie but not
help to distribute it.
T
Max 3 ¦¦ ¦ a
t 1 iI kK
ik y ikmt 3 ikmt

Equation 21: Global objective of maximizing cost reduction

Together with the constraints introduced in the last section, a binary non-
linear optimization model (BNLP) for the central planner can be derived, as

135
By comparing a centrally and a decentrally coordinated standardization model, Weitzel explores
the circumstances in which independent agents’ local decisions about adopting a communication
standard lead to an inferior network configuration compared to the optimal solution established
by a central planner (Weitzel 2004; Weitzel et al. 2006).
244 Developing a Formal Model for Cooperative Sourcing

shown in the following. For reasons of complexity reduction, the solution of this
cooperative sourcing problem (CSP) will be static, i.e., solving the CSP leads to
a global cooperative sourcing configuration which is assumed to be optimal for
all periods. Although learning effects from the first sourcing decision will (at
least slightly) reduce negotiation costs and coordination costs of future sourcing
decisions, in this case, it is not included for the purpose of significantly reducing
complexity. Consequently, the scenario consists of two structurally different
periods. In t=0, the cooperative sourcing process takes place, causing adoption
costs and negotiation costs. In t=1…T, the process cost savings as well as coor-
dination costs, interface costs, and agency costs appear and are assumed to be
constant over all periods. Therefore the optimization model only uses decision
variables for t=0 and t=1 (the last index of all variables introduced above is set
either to 0 or 1). Thus, the (binary) decision variables of the optimization model
are yikm0, zikm1, and vikm1 136.
Furthermore, the binary parameter aik has been incorporated into all of the
cost functions, indicating if business function k is part of bank i’s business at all
(cf. section 4.2.1).
In order to solve the CSP, the global objective (Equation 21) has to be con-
verted into a global non-linear decision function. In a first step, the periodical
process cost effect (process cost savings: PCS) is adapted to meet this require-
ment:
K mkmax I § § I ··
¨ z ˜ C P x   v ¨ K F  c P ˜ §¨ x ˜ z ·¸ ¸ ¸ t
1

PCS ¦ ¦¦ aik ˜
¨ ikm1 ik ik ikm1
¨ ik ik ¨ ¦ jk jkm1 ¸
¸
k 1 m 1 i 1
© © ©j1 ¹ ¹ ¸¹
Equation 22: Process cost savings part of the centralized objective
The first term zikm1 ˜ Cik  xik  describes the gross process cost savings when i's
P

business function k is outsourced. The second term

§ § I ··
vikm1 ¨ K ikF  cikP ˜ ¨¨ ¦ x jk ˜ z jkm1 ¸¸ ¸ represents coalition km’s process costs, i.e. cost
¨ ¸
© ©j1 ¹¹
factors of the insourcer (determined by binary indicator variable vikm1) and total
output level of all coalition members. Allocation of costs to the coalition mem-
bers does not play a role in this centralized perspective.
The transaction cost terms are based on the formulas developed in the previ-
ous section. The adoption cost function CAD builds on Equation 8. vikm1 ensures

136
It is evident that we could also remove the t-indices in this case. Nevertheless, the goal was to
thoroughly provide consistency in declaring variables and parameters throughout the whole
model description.
Developing a Formal Model for Cooperative Sourcing 245

that the costs of adapting the insourcer’s process by the other coalition members
are applied.
AD
Cikm 0 aik ˜ yikm 0 ˜ ¦ v jkm1 ˜ 1  9 kij 0  ˜ ckAD ˜ F ik i, k , m
jI
j zi

Equation 23: Adoption costs in the centralized objective

The negotiation cost function CN (Equation 6) is reduced by the learning ef-
fect factor because it is ignored by the centralized model, as explained above.
Negotiation costs are assumed to accrue only in the 0th period. The factor in
brackets determines the size of the coalition. The costs only occur, if i outsources
k (yikm0) and if k is part of i's business (aik) at all.
D 1
N
§ ·
Cikmaik ˜ yikm 0 ˜ ckN ˜ F ik ˜ ¨¨ ¦ z jkm1 ¸¸
0 i, k , m
© jI ¹
Equation 24: Negotiation costs in the centralized objective
In contrast to the negotiation costs, coordination costs CC occur in any of the
following periods. Therefore, yiksm0 is replaced by zikm1.
D 1
C
§ ·
Cikm1 J ˜CN
aik ˜ J ˜ zikm1 ˜ ckN ˜ F ik ˜ ¨¨ ¦ z jkm1 ¸¸ i, k , m
© jI ¹
Equation 25: Coordination costs in the centralized objective
The interface cost function CIF is directly adopted from Equation 9. ziklm
diff
1
, as
a binary substitution variable, becomes equal to 1 if i's business functions k and l
are operated by different banks. This is ensured by a constraint given below
(Equation 29-6). Here, all relationships between two business functions are con-
diff
sidered twice, but ziklm 1
is only equal to 1 in one of both cases.
K mkmax
1 |K |

CiIF
1 ¦ ¦¦¦ a ik ˜ ail ˜ T ikl ˜ c IF ˜ ziklm
diff
1 i
k 1 m 0 l 1 jI
l z k j zi

Equation 26: Interface costs in the centralized objective

As the final part of the transaction costs, agency costs CAG incorporate the
business neighborhood between bank i and the insourcer of the coalition km
bnikm1, which is determined by a further constraint (Equation 29-4).
AG
Cikm 1 aik ˜ zikm1 ˜ ckAG ˜ 1  1  bnikm1  bne ˜ F ik ˜ Oik i, k , m
Equation 27: Agency costs in the centralized objective
246 Developing a Formal Model for Cooperative Sourcing

Finally, the inter-temporal computable global objective takes the following

form:
§ ·
¨ mkmax1 § § § ··· ¸
¨ ¦¦¦ aik ¨ zikm1 ˜ CikP xik   vikm1 ¨ KikF  cikP ˜ ¨ ¦ x jk ˜ z jkm1 ¸ ¸ ¸ ¸
¨ kK m 1 iI ¨© ¨ ¨ ¸ ¸
© © jI ¹ ¹ ¸¹ ¸
¨ ¸
mkmax
¨ 1
IF diff
¸
Max NPV ¨  ¦¦¦¦ aik ˜ ail ˜ T ikl ˜ c ˜ ziklm1 ¸
t 1..T, rad
¨ kK llzKk m 0 iI ¸
¨ ¸
D 1
¨ mkmax
1 § § · ·¸
¨  ¦¦¦ aik zikm1 ¨ ck ˜ 1  1  bnikm1 bne ˜ F ik ˜ Oik  ck ˜ F ik ˜ ¨ ¦ z jkm1 ¸ ˜ J ¸ ¸
AG N

¨ kK m 1 iI ¨ ¨ ¸ ¸¸
© © © jI ¹ ¹¹
mkmax § D 1 ·
1
¨ § · ¸
¦¦¦ aik ¨ yikm0 ¦ v jkm1 ˜ 1  9 kij0 ˜ F ik ˜ ckAD  ckN ˜ Fik ˜ ¨¨ ¦ z jkm1 ¸¸ ˜ yikm0 ¸
kK m 1 iI ¨ jI © jI ¹ ¸
© j zi ¹
Equation 28: Computable Global Objective of the CSP
The upper expression contains and discounts periodical savings and costs
while the lower bracket contains adoption costs and negotiation costs, which
only occur initially. NPV represents the function used to compute the net pre-
t 1..T,rad
sent value for periods t = 1…T with risk-adjusted calculation discount rate rad.
In the following table, all necessary constraints are provided to complete the
cooperative sourcing problem (CSP).
Business function is executed
only in one location (single
m kmax
1 sourcing). m=0 represents own
¦z ikm1 1 i, k (29-1) production, i.e., not taking part
m 0 in cooperative sourcing.
m>0 represents the correspond-
ing coalition km.
yikm0 = zikm1 i, j, k , km (29-2)

(29-3) Only one insourcer in every

¦v
iI
ikm1 ˜ zikm1 1 k, km
coalition km

Business neighborhood is
i, j , k , km (29-4) determined by the competitive
bnij ˜ zikm1 ˜ z jkm1 d bnikm1
degree between outsourcer and
insourcer.
Developing a Formal Model for Cooperative Sourcing 247

mkmax
1

¦¦z ikm1 ˜ 1  v ikm1  ˜ O ik d SC i i (29-5) Strategic constraint

k K m 1

diff
diff Determining z iklm 1
which
ziklm 1 t v jkm1 zikm1  v jlm1 zilm1 (29-6) becomes equal to 1 if bank i's k
i, j , km, k , l z k and l are provided by different
banks137
vikm1 , yikm 0 , z ikm1  ^0;1` i, k , km (29-7) Binary variables

Equation 29: Constraints of the CSP

The third constraint leads to one third of the binary variables becoming ob-
solete. Again, we differentiated between y and z to ensure a consistent use of the
model’s elements.
Solving the CSP, after feeding it with appropriate data, determines which
banks should cooperatively source which business function in order to maximize
global net benefits. Due to the binary and non-linear character of the objective
function, it is quite obvious that the model is far too complex to be solved by
exact algorithms such as Branch&Bound138. Therefore, section 5.2 provides a
heuristic (genetic) algorithm to find at least “good” solutions for the CSP.

4.4 Decentralized Model: Autonomous Actor

Decisions
In this section, the decentralized variant of the model is developed, which, in
contrast to the centralized view, assumes that banks decide autonomously about
cooperatively sourcing their business functions. Decentralized coordination
represents individual and autonomous decision behavior. Agents try to improve
their own situation without considering the performance of the group or the
whole system. In cooperative sourcing they commit to joining a coalition but,
nevertheless, will decide to leave if a more valuable sourcing option can be
reached. They will only enter a coalition if their individual situation can be im-

137
Since the cost minimization routine attempts to minimize z iklmtdiff , “greater than”-constraints are

sufficient. The expression on the right side may become equal to -1, but because all combinations
of k and l are iterated there is a corresponding (more binding) constraint, in which the right side
becomes 1.
138
Cf. section 5.2 for a discussion of the degree of complexity of the cooperative sourcing problem.
248 Developing a Formal Model for Cooperative Sourcing

proved. This may lead to a gap between local and global efficiency – or between
individual consequences and consequences for the overall system139.
In a multi-period scenario, there are several, structurally different decision
steps where the cooperative sourcing model must provide decision functions to
the agents. These are:
o evaluating a new coalition membership
o as a coalition member: voting about the entrance of a new member
o re-evaluating an existing coalition membership compared to the alternatives
of
o backsourcing
o switching to another coalition
In the following, decision functions for all decision points are developed.

Evaluating a new coalition membership

The agent’s benefit function, given by Equation 15, assumes a static coalition
structure, i.e. the rest of the coalition will remain stable during the complete
duration of the contract (all other members will stay in the future and no new
members will join the coalition). Using this benefit function as a decision func-
tion would ignore significant externalities resulting from the others members’
behavior. If a bank joins an already existing cooperative sourcing coalition, all
partners (if we assume all contracts to have the same duration TCoSo) will be al-
lowed to leave the coalition before the new member may leave itself. Because
this can have significant effects on the decider’s cost situation, they must take
these externalities into account ex ante by forming a stochastic assumption about
future changes in the coalition’s configuration. Because we assume risk-neutral
decision makers in this model, the deterministic benefit function (Equation 15)
becomes an expectation function Exp 3 ikmtenter
> @
, determining the expected value of
the individual outcome from entering the coalition. Equation 30 presents this
decision function for bank i, evaluating a possible membership in an already
enter
existing coalition km. If Exp 3 ikmt > @
is positive, the bank decides to join it.

139
Section 5.1 presents a game-theoretical analysis of cooperative sourcing scenarios and determines
the set of imputations in which interests of the group and all individuals fall together.
Developing a Formal Model for Cooperative Sourcing 249

T CoSo
>enter
Exp 3 ikmt @ ¦ §¨¨ Exp>NS   @˜ 1 ·¸¸  C ikm t W
AD
ikmt
N
 Cikmt
© W 1 1  rad  ¹ W

with Exp>NS   @ PCS    C    C

ikm t W ikm t W
IF
ikm t W
C
ikm t W 
AG
 Cikm t W  

¦ 'PCS
jM km
ijkm t W   'TCijkmt W   ˜ p leave
jkm t W 

j zi

Equation 30: Decision function for evaluating a new coalition membership

Expected periodical Exp NSikm t W  > @ consist of process cost savings
P G
PCSikm t W  (= C  C
ik ikm t W 
) minus transaction costs and minus possible
monetary changes caused by any other member j leaving the coalition in later
periods. 'PCSijkm t W  describes the resulting impact on process cost savings
while 'TCijkm t W  represents the corresponding delta in transaction costs. To-
gether with the estimated probability of j’s exit p leave
jkm t W  in a period t+W, this

represents the uncertainty factor resulting from externalities which must be taken
into account.
Because the model assumes bounded rationality, actors do not take into ac-
count the opposite, but much weaker, assumption about potential new coalition
members, although they might have access to relevant information. Since the
bounded rationality concept assumes restricted evaluation and decision capacity
of a decision maker (Simon 1976; Williamson 1975), this will correspond to the
consideration that it would be impractical for a bank to evaluate each market
member regarding its potential membership in the coalition140.
As a further assumption about the decision makers of the model, we adopt
the symmetry principle from game theory, which states that all actors follow the
same rationality concept and furthermore are aware of this fact (Schelling 1961).
The function for determining exit probability p leave
jkm t W 
of partner j in a future
period t+W is adopted and adapted from a formal model on IT standardization
leave
(Weitzel et al. 2003)141. p jkm t W  includes expected additional process cost
>
savings Exp 'PCS leave
j @ which j would gain from leaving the coalition and
140
One could argue that at least general assumptions about the entrance of new alliance members
could and should be made, but it would be difficult to set this assumption on an empirically sound
base. Hence, in this case we adopt a pessimistic and cautious estimator.
141
In this standardization model, p expresses the probability of a partner j’s adoption of the same
standard as actor i, in order to reduce communication costs (e.g. by using the same EDI standard).
Here, p is used reciprocally; it describes the probability of a partner leaving the coalition.
250 Developing a Formal Model for Cooperative Sourcing

switching to another one. From Exp 'PCS leave

j >
, the estimated change of j’s @
transaction costs Exp 'TC > leave
j @ is subtracted and the difference is normalized to
a value between 0 and 1 (i.e. the exit probability). Since both expectation values
can be negative, pleave has to be restricted to this range.
­
°
0 if > j @
Exp 'PCSleave >
d Exp 'TCleave
j @ >
and Exp 'PCSleave
j @ > @
˜ Exp 'TCleave
j t0

pleave
> leave
@
° Exp 'PCSj  Exp 'TCj> leave
@ if > @
Exp 'PCSleave >
! Exp 'TCleave @ >
and Exp 'PCSleave @ > @
˜ Exp 'TCleave !0
®
jkmt W 
° >
Exp 'PCSj leave
@ j j j j

° 1 else
¯

Equation 31: Exit probability pleave

Voting about a new coalition member
If bank i has positively evaluated the possibility of joining alliance km, this still
does not imply that it actually will become a member. First, the existing mem-
bers on their part have to consider about i's membership. The members’ decision
function to evaluate an applicant’s entrance is given by Equation 32. It is as-
sumed that the existing members do not have to pay any additional transaction
costs (for evaluation, negotiation, and adoption) if a new member joins. There-
fore, the decision function can be reduced to a one-period evaluation, consider-
ing the process cost effect (first term) and a possible change in periodical trans-
action costs (second term). Coordination costs will increase in any case, due to
the expanded coalition, but agency costs will only differ if the insourcer changes
(as a result of the increased process volume). If i itself has either been the in-
sourcer or will become the insourcer, the negative resp. positive effect in the cost
difference becomes maximal.
Exp 'NSikmt > G
Cikmt @ >
xkm  x j   Cikmt
G
xkm   'Cijkmt
C AG
 'Cijkmt @ > @
Equation 32: Decision function for alliance member i voting about the entry of a
potential new member j
If the expectation value becomes positive, km’s member i will appreciate the
entry of bank j. Otherwise it declines. The parameteriz model determines under
which circumstances an applicant may join the alliance (unanimity or a certain
pre-specified majority of positive votes).
Developing a Formal Model for Cooperative Sourcing 251

Re-evaluating an existing coalition membership

In later periods of the sourcing partnership, when the sourcing contract termi-
nates (TCoSo), an alliance member has to re-evaluate the situation which might
have been altered due to coalition partners having entered or left the coalition in
the meantime. This re-evaluation takes place in two steps. First, the bank has to
decide if backsourcing has become favorable in comparison to entering a con-
tract renewal for another TCoSo periods. If not, it has to decide whether changing
to another coalition might be dominant. The evaluation functions are given by
Equation 33 and Equation 34.
T CoSo
> stay
Exp 3 ikmt@ ¦§¨¨ Exp>NS   @˜ 1 ·¸¸
ikm t W
©
W 1 1  rad  ¹ W

with Exp>NS   @ OS    C    C
p
ikm t W ikm t W
IF
ikm t W
C
ikmt W 
AG
 Cikmt W 
 ¦ 'PCS
jM km
ijkmt W   'TCijkmt W   ˜ p leave
jkmt W 

j zi

Equation 33: Re-evaluating an existing coalition membership compared to

backsourcing
Re-evaluating a cooperative sourcing partnership, compared to backsourc-
ing, is similar to evaluating a new membership, despite the fact that costs for
negotiation and adoption will not occur again. Therefore, Equation 33 and
Equation 31 differ only in these terms.
Evaluating a change of the coalition is more complex because the cost effect
of entering a hypothetical new coalition has to be estimated and compared with
expected costs from staying in the previous alliance. Therefore, the decision
function compares the estimated net savings from joining an alternative coalition
Exp>3 iknt
alternative
@ with those resulting from staying Exp>NSikmt
stay
W @ (discounted over all

contract periods W = 1...TCoSo). In order to compute Exp 3 iknt

alternative
>
, an alternative @
coalition kn has to be chosen. Based on this alternative coalition, the expected
discounted net outcome can be computed by Equation 31 and can be compared
to an extended stay of TCoSo periods in the current sourcing coalition km
(Equation 33). The algorithm for selecting an alternate alliance is not specified
by the model itself. It just assumes, due to bounded rationality, that only one
existing coalition is used for comparison, instead of all of them. The simulations
in the next chapter will implement various coalition selection algorithms and
compare them.
If the bank decides to change the coalition, it has first to apply in order to
become a member of the alternative alliance kn, which in turn has to be posi-
252 Developing a Formal Model for Cooperative Sourcing

tively answered by the coalition members. The model assumes that the bank will
stay in the current coalition for another TCoSo periods, if the alternative coalition
rejects the application.
Exp '3iktswitch enter
Exp 3 iknt >
stay
 Exp 3 ikmt @ > @ > @
Equation 34: Re-evaluating an existing coalition membership compared to
changing the alliance
After having developed all these decision functions, one aspect remains un-
answered: how does the potential alliance member i estimate the expected differ-
ence in process cost savings Exp>'PCS leave
j @ and transaction costs Exp>'TC leave
j @ of
an alliance partner j in order to determine his exit probability? We assume the
following steps to be conducted:

1. Is the current cooperative sourcing coalition still favorable for partner j

compared to backsourcing? The model assumes that i is capable of deter-
mining the correct answer because it has the necessary information about j’s
cost situation due to their partnership. If i realizes that the alliance has be-
come unfavorable to j, the exit probability is set to 1 and the second step is
skipped.
2. Given the current situation, would it be favorable for partner j to switch to
another coalition? We assume that i will take all necessary information
about the alternate coalition she/he would choose herself/himself when
evaluating her/his own change of coalitions. The expected delta in PCS and
leave
TC is computed to determine p jkm t W  . If the p-formula in Equation 31 is
further developed by inserting the different cost functions, the following
formula results:

p leave
>
Exp 'PCS leave
j @
 Exp 'TC leave
j > @ 1
>
Exp 'TC leave
j @œ
jkmt
>
Exp 'PCS leave
j @ >
Exp 'PCS leave
j @
AD
§
  Tc
F jk ¨¨ 1  ] jˆjkt ˜ k
CoSo
 E§
 
 ckN o jt  1 ¨J §¨ Mˆ knt D 1
  M kmt 
D 1 ·  1 Mˆ
¸
¹ T CoSo

knt 
D 1 · ˆ
  AG ·
¸  bn jknt  bnjkmt bne˜ ck Oik ¸¸
leave © © © ¹ ¹
p 1
 Cˆ
jkmt G G
C jkmt jknt

Equation 35: Explication of probability function p of member j leaving

coalition km
The numerator compares current (periodical) transaction costs to the hypo-
thetical new ones, if j became a member of kn, while the denominator compares
the process cost savings. All variables with an n (and an additional “^”) in the
index represent values of the alternate coalition, assuming that j hypothetically
Developing a Formal Model for Cooperative Sourcing 253

became its member. ĵ represents the current insourcer of kn142. For simplifica-
tion reasons, one-time costs of adoption and negotiation when switching to the
hypothetical new coalition are uniformly distributed over all periods TCoSo.
It should be clarified that the introduced cooperative sourcing model does
not distinguish between cooperative sourcing by establishing a joint venture (i.e.
founding a joint subsidiary) and cooperative sourcing by mutual outsourcing and
insourcing. In fact, the model merges both situations.
The joint venture-like elements are:
o no market pricing mechanisms but, instead, a pre-determined cost allocation
rule which treats all members as equal,
o no profit maximizing but cost minimizing agents,
o negotiation and coordination costs depending on the coalition size,
o all coalition members having a voting right regarding the admission of a
new partner.
The insourcing/outsourcing-like elements are:
o the process cost function of the cost efficient partner taken without altering,
o all coalition members paying adoption costs and agency costs but not the
insourcer, and
o the insourcer being not restricted by the strategic constraint.
Consequently, it can be stated that the model assumes a joint venture part-
nership with the “cost leader,” providing the necessary technology and capacity,
which in turn becomes the insourcer143. Empirical studies found that usually one
of the partners is critical to a multi-firm alliance (e.g. Rouse and Corbitt 2004).

4.5 Extending the Model by Legal and Regulatory

Issues
Section 3.5 gave an overview about legal issues which affect outsourcing deci-
sions in general and in the banking industry in particular. This section provides a
brief overview of how they can be considered in the model.

142
This is a simplification, because j or any of the other existing members of kn could become the
new optimal insourcer as a result of j’s entrance.
143
However, because a kind of partial capital merger (e.g. by exchanging shares) is common in most
of the major outsourcing deals (ITO as well as BPO), outsourcing in a strict sense (i.e. economic
independence of insourcer and outsourcer) happens quite infrequently in the German banking in-
dustry.
254 Developing a Formal Model for Cooperative Sourcing

The first legal aspect which has a major effect on outsourcing decisions in
the banking industry, in particular, is the problem of value-added taxes (VAT).
Because banking services are usually free of VAT, but outsourced services pro-
cured by the bank are basically not, the bank has to take the VAT into account.
Consequently, the outsourcer’s process cost savings will be reduced by VAT in
the model:
PCSikmt CikP xik   (1  VAT ) ˜ Cikmt
G M
xik , xkm  
Equation 36: Considering VAT in the process cost savings
As a regulatory issue, supervisory regulations constrain banks in outsourcing
basic banking activities which have a high strategic impact or high complexity
(cf. section 3.5.1.1, e.g. risk management or granting non-standardized loans).
Therefore, the model has to consider outsourcing restrictions on particular busi-
ness functions. This is simply realized by a binary permission LC to outsource a
certain business function (LC=1) or not (LC=0).
M

¦y ikmt d LC i, k , t LC  ^0;1`

m 1

Equation 37: Business function-based legal outsourcing restriction

Moreover, in Germany the KWG restricts banks from outsourcing more
business functions than the remaining ones (in terms of “importance” and “size”
(BaFin Circular 11/2001 paragraph 17, cf. section 3.5.1.1). In considering the
importance of the strategic constraint, Equation 19 can be limited to outsourcing
less than half of the available core competencies. Size is considered by outsourc-
ing to be a maximum of 50% of the complete business volume, measured by the
sum of original process costs across all business functions.
|M |
1 |M |
1
¦¦ z ikmt  
˜ pik ˜ CikP d
2 kK
 
¦ pik ˜ CikP i, t ¦¦ z ikmt pik Oik d ¦ pik Oik i, t
kK m 1 kK m 1 2 kK
Equation 38: General legal outsourcing restriction, based on volume (left) and on
strategic value (right)
Further legal and regulatory issues can be considered by the parameteriza-
tion of particular transaction cost parameters (esp. legally enforced requirements
to monitor the provider and to enable supervising authorities to obtain access to
any necessary information).
Developing a Formal Model for Cooperative Sourcing 255

4.6 Summary
In this chapter, a formal agent-based model of cooperative sourcing was intro-
duced, which covers the rational and economically reasoned decision behavior of
multiple agents on cooperative sourcing of business functions. Based on different
theories and further assumptions, the decision functions and constraints have
been developed, resulting in a decentralized and centralized variant of the model.
While the first models the firms’ individual decision behavior, the latter repre-
sents a global optimization model for benchmarking the outcome of the indi-
viduals’ actions.
The next section will apply the model to both analytical research and simu-
lation studies, fed by empirical data, presented in section 3.6.
Analytical models and simulation studies are built upon simplifying assump-
tions leading to several limitations and thus need to fulfill several careful steps of
validation (Sargent 1998). Although limitations and conceptual model validity
(whether or not the model fits with existing theory) could already be done at this
stage, we decided to discuss all limitation issues (regarding the model as well as
the approach) and to present all validation steps collected within an specific
section at the end of this research (cf. section 6.3). This section also contains a
table which summarizes the theoretical underpinning of the cooperative sourcing
model (Table 70 on p. 405).
256 Analytical and Simulative Studies

5 Analytical and Simulative

Studies
“Simulation is a way of doing thought experiments.
While the assumptions may be simple the
consequences may not be obvious at all.”
(Axelrod 2000, p. 135)

In this chapter, the previous formal and empirical work will be used for conduct-
ing analytical and simulative studies on cooperative sourcing behavior and the
resulting market effects in order to answer the research questions of this work.
Based on the developed cooperative sourcing model and on game theory, the
first section provides an analytical investigation of different allocation schemes
(section 5.1). In section 5.2, a genetic algorithm is developed to solve the coop-
erative sourcing problem, which was developed in section 4.3 (i.e. global optimi-
zation). The main part of this chapter is formed by comprehensive simulation
studies. Based on the results of the game-theoretical analysis and on the empiri-
cal data which was gathered in section 3.6, the decentralized cooperative sourc-
ing model (section 4.4) will be implemented. Thus, agent behavior under differ-
ent settings can be simulated on the basis of agent-based economics and com-
pared with the results from global optimization (section 5.3). Implications and
limitations of the findings will be discussed in chapter 6.

5.1 Game-Theoretical Analysis of Cooperative

Sourcing144
While cooperative sourcing projects are more or less making promises to all
participating firms, they also pose a challenge to decision makers because com-
mon costs and benefits of the joint “production” make it difficult to agree on
what proportion of the costs will be paid by which firm. At the same time, know-
ing the costs and benefits associated with the project is a precondition for select-
ing partners and to agreeing on the deal. As a consequence, the strategic situation
of cooperative sourcing is even more intricate than in traditional 1:1 relations as

144
An earlier version of this section was published in the proceedings of the 39th Hawaii Interna-
tional Conference on System Sciences (Beimborn et al. 2006b).
Analytical and Simulative Studies 257

now there is the possibility of coalition building. This has a substantial impact on
coalition stability and cost allocation rules.
The cooperative sourcing model in chapter 4 did not answer the question
how costs of a coalition will be allocated to the coalition members. Accordingly,
the research question of this section is
What cost or benefit allocation enables stability of cooperative sourcing
coalitions?
This analysis contributes to the sourcing literature by providing a sound
theoretical foundation for cooperative sourcing and the analysis of the existence
and efficiency of sourcing equilibria. From a managerial perspective, the model
helps managers evaluate cost and benefit allocation rules for their sourcing coali-
tion contracts. While the main goal is to formally prove conditions for the stabil-
ity of cooperative sourcing equilibria based on cooperative game theory, this
section also presents results from a game-theoretical experiment on cooperative
sourcing, which indicates that deciders might be inclined to choose allocation
rules that lead to instable sourcing coalitions.

5.1.1 Basic Concepts from Cooperative Game Theory

While non-cooperative game theory mainly deals with the problem of predicting
allocations, cooperative game theory answers the questions whether or not a
particular coalition will be formed and in which way the coalition will divide
generated benefits among the participating players. Certain assumptions are
made within this theory: first, all contracts are binding and enforceable; second,
utility is transferable without loss (Owen 1995). While the first assumption is
unproblematic as any cross-firm cost allocation requires contracts and can thus
be taken for granted in all outsourcing arrangements, the second assumption is
more problematic. But since a critique of the concept of an exchange economy
and the underlying neo-classical paradigm is not part of this work, we accept this
premise as given. For a discussion, see e.g. (Kelly 1978).
The characteristic function of an n-person game is a real-valued function v
defined on the subsets of N (set of players). It expresses the amount of costs
assigned to a single player and has the following properties:
The costs of the empty coalition are always zero (Equation 39). Further-
more, the subadditivity property must be fulfilled. This means that costs assigned
to a coalition M S1 ‰ S 2 must be lower than the sum of costs of the corre-
sponding sub-coalitions S1 and S 2 .
 258 Analytical and Simulative Studies

Q ‡ 0 vS1 ‰ S 2  d vS1   S 2  if S1 ˆ S 2 ‡
Equation 39: Costs of empty coalition Equation 40: Subadditivity property
An imputation for an n-person game is a vector S satisfying individual ra-
tionality. Individual rationality means that if forming a coalition, no player i will
accept costs Si higher than the costs v(i) she or he has to bear without entering
the coalition. Pareto optimality ensures that all costs of the total coalition v(N)
(i.e., coalition of all players) are divided among the players (Owen 1995).
n

S i d v (i ) (i 1,2,..., n ) ¦S i Q N 
i 1
Equation 41: Individual rationality Equation 42: Pareto optimality
The core is a solution concept for cooperative n-person games which re-
duces the set of possible payoff vectors to a set of all non-dominated imputa-
tions. Every vector within the core is stable, which means that there is no other
coalition in which the players have both the desire and power to change the out-
come of the game. An imputation belongs to the core of a game if it also satisfies
Equation 43.
¦S
iS
i d v( S ) for all S  N

Equation 43: Group rationality

This constraint is also referred to as group rationality (Owen 1995). The
sum of costs assigned to a subset of the coalition members S must not be higher
than the costs occurring if these members form their own sub-coalition. Other-
wise, these players have an incentive to leave the present coalition.
The core may be void (Owen 1995). The resulting allocations of a coopera-
tive game with three players can be represented in the “fundamental triangle of
costs” (Lemaire 1984). Each allocation displayed by the triangle is pareto-
optimal, satisfying Equation 42. Figure 99 shows a triangle with a core at its
center (fulfilling individual rationality = upper border (Equation 41) and group
rationality = lower border (Equation 43) from each of the players’ perspective).
Analytical and Simulative Studies 259

(0; 0; v(N))
x3

Core

x1 x2
(v(N); 0; 0) (0; v(N); 0)
Figure 99: Fundamental triangle of costs
5.1.2 Allocation Mechanisms
Cost allocation problems are usually solved by using “classical” methods, which
are widely spread and easy to understand. These methods can be used ex ante to
suggest different allocations the players can agree upon before establishing a
coalition. We follow the list of allocation mechanisms given in (Lemaire 1984).
The first method is equal allocation of gain. Here, the same amount is sub-
tracted from each player’s process costs, ensuring the same benefits for each
player with v(i) as individual costs of player i, v(N) as the costs assigned to the
full coalition, and ¦ v( j ) as the sum of all players’ individual costs (Lemaire
jN

1984). This solution corresponds to the Nash solution, which maximizes the
product of all members’ benefits (Nash 1950):

1ª º
Si « ¦ v( j )  v ( N ) »
vi 
n ¬ jN ¼
Equation 44: Equal allocation of gain
A second possible allocation would be every player bearing a share of the
coalition’s total costs proportionally to the player’s individual costs. This method
is called proportional cost allocation:
260 Analytical and Simulative Studies

v(i )
Si v( N )
¦ v( j )
jN

Equation 45: Proportional cost allocation

This allocation will be slightly altered in the following analysis, where the
costs are allocated proportionally based on individual process volumes rather
than on individual costs.
Lemaire (1984) discusses two further concepts which are related to the two
mentioned above: equal distribution of non-marginal benefits and proportional
allocation of non-marginal costs. Due to the particular structure of the coopera-
tive sourcing model, the first concept is not applicable, while the second (since
the model assumes linear cost functions) is identical to the proportional cost
allocation.
A completely different allocation scheme has been developed by Shapley,
who proposed a concept that splits the value of a coalition depending on each
player’s bargaining power (Shapley 1953). A basic assumption of what is known
as the Shapley value is that every possible sub-coalition S (with s being the num-
ber of players in S) is equiprobable to be formed. Furthermore, the order of each
player’s entry into a coalition is important. A game with n players has n! differ-
ent possible orders for forming coalitions. The solution given by the Shapley
value will be likely if players are rational and willing to exhaust their bargaining
power completely. Players who do not contribute large amounts to the final solu-
tion will not receive large savings (Rapoport 2001).
s  1! n  s ! >v S   v S  ^i`@
Si ¦
SN n!
Equation 46: Shapley value
The Shapley value is less suitable for determining ex ante allocations but
can be used as a point of reference and for explaining results from negotiating
cost allocations.

5.1.3 Model
In the following, we will use a simplified form of the cooperative sourcing
model, which can be applied to game-theoretical analyses. First, we assume only
one business function which allows us to reduce all variables and parameters by
index k. Furthermore, the model is reduced to a one-period game; thus, index t
can also be neglected.
Analytical and Simulative Studies 261

Fixed costs, marginal costs, and process volume of the

K iF , ciP , xi
investigated activity at bank i
C iP K iF  ciP xi Process cost function
Mm, Mm Set (and number) of firms joining the m-th coalition
Table 37: Basic parameters, indices, and variables
Analogous to the model description in chapter 4, the process costs of the
coalition are formed by the cost-efficient insourcer of the coalition.

C mPM Process costs of cooperation m

§ § ··
¨
C mPM : min ¨ CiP ¨
iM m ¨ ¨ ¦ ¸¸
¸
xi ¸ ¸   
min CiP x mM
iM m
i
© © m ¹¹
M

K mFM , cmPM , xmM Fixed costs, variable costs, and output of coalition Mm
M
CmPM : K mFM  cmPM ˜ xmM with x m ¦x
iM m
i

Table 38: Process costs of coalition Mm

K mFS , cmPS , xmS Fixed costs, variable costs, and output of a sub-coalition Sm
S
CmPS : K mFS  cmPS ˜ xmS with x m ¦x
iS m
i and S m  M m

Table 39: Process costs of a sub-coalition Sm

Now, the allocation mechanisms (Equation 44 to Equation 46) can be refor-
mulated with C iG as the part of the coalition’s costs to be borne by member i as
follows:

1 § ·
CiG CiP  ¨ ¦ C Pj  CmPM ¸
¨ j M ¸
Mm © m ¹
Equation 47: Cost allocation resulting from equal distribution of benefits

xi § K FM ·
CiG CmPM ˜ xi ¨¨ mM  cmPM ¸¸
xmM © xm ¹
Equation 48: Proportional cost allocation
262 Analytical and Simulative Studies

CiG
S m  1!˜ M m  S m !
>K FS
  
 cmPS xmS  K mFS *  cmPS * ˜ xmS  xi @
¦
Sm M m Mm !
m

Equation 49: Shapley cost allocation

The right part of the Shapley value function describes the marginal costs
bank i creates in any sub-coalition of Mm. * represents the cost function parame-
ters after bank i has left the sub-coalition (i.e. the cost function of the new in-
sourcer).
Further, the model will, in a later step, consider transaction costs which rep-
resent the trade-off to process cost savings resulting from cooperative sourcing.
In this reduced version of the cooperative sourcing model, the different transac-
T
tion cost types are aggregated to a single parameter Cim .
The resulting decision function of an actor, determining the effect of taking
part in a cooperative sourcing agreement, includes the actor’s share of process
costs and transaction costs.
In the following section, a game-theoretical analysis shows which allocation
rules lead to stable coalitions.

5.1.4 Analysis
As explained earlier, a cooperative n-person game fulfills the subadditivity prop-
erty, which means that by merging two disjoint coalitions the resulting costs will
fall below the costs of operating separately (cf. Equation 40). The model fulfills
this property because the assumption of linear cost functions leads to decreasing
average costs for every additional unit x due to the entrance of a new coalition
member.
A necessary condition for a stable cost allocation is that the costs assigned to
each bank are situated in the core, which is defined by the following borders:
o The upper border is determined by the bank’s own process costs resulting
from stand-alone in-house production (individual rationality).
o The lower border is determined by the bank’s contribution to the coalition’s
costs. The bank at least has to bear the cost difference between the cost of
the coalition (including itself) and the coalition excluding itself (group ra-
tionality). For more than three coalition members, this rule has to be gener-
alized from one bank to every possible sub-coalition, which also at least has
to make its contribution to the coalition’s total costs.
Formally, for a 3-member coalition the cost interval which is acceptable for
a coalition member i is given by Equation 50, whereas CiG is i's part of the coa-
Analytical and Simulative Studies 263

lition costs, which results from any allocation scheme. K mFM * , cmPM * represent the
cost parameters of the new insourcer for the residual process volume in case i, as
the former insourcer, would leave the coalition.
§ § ·· § § ··
¨ FM PM ¨ ¸ ¸ ¨ FM * PM * ¨ ¸¸ G F P
¨ K m  c m ˜ ¨ ¦ x j ¸ ¸  ¨ K m  c m ˜ ¨ ¦ x j ¸ ¸ d C i d K i  ci xi
¨ ¨ jM m ¸ ¸ ¨ ¨ jM m ¸ ¸
© © ¹¹ © © j zi ¹ ¹
Equation 50: Cost interval from the insourcer’s perspective
If i is not the coalition’s insourcer and if the insourcer would not change if i
left the coalition, Equation 50 can be simplified to Equation 51.
CiG K iF
cmPM xi d CiG d CiF  ciP xi œ cmPM d d  ciP
xi xi
Equation 51: Cost interval from actor i's perspective, if i z insourcer
In the following, we will test several allocation schemes with regard to their
general stability, i.e. analyze whether the allocation result would be in the core
under any parameterization. Due to strongly increasing formal complexity we
will restrict our analysis to 3-member coalitions and only extend it to an n-
member scenario if the analyzed allocation scheme showed to be stable for 3-
member coalitions.

5.1.4.1 Equal Allocation of Gain

For an outsourcer i the following conditions have to be fulfilled (cf. Equation 51)
so that the equal allocation of gain scheme always leads to stable coalitions:

1§ ·
cmPM xi d CiP  
¨ ¦ C Pj  C mPM ¸ d CiF  ciP xi CiP
¨ iM
Mm ¸ 
© m ¹
While the property of subadditivity obviously satisfies the right border, the
left border has to be reformulated as
§ ·
¨ ¦ C Pj  CmPM ¸
¨ iM ¸
© m
Mm
i i m 
¹ d C F  c P  c PM ˜ x
i

For more than 2 banks, the inequation generally does not hold true. For ex-
ample, if bank i has significantly lower fixed costs than the other coalition mem-
264 Analytical and Simulative Studies

bers, only slightly higher variable costs and the same volume, the right side will
become lower than the left side145.
For the insourcer, it would be required to fulfill the following constraints (cf.
Equation 50):

1 § n P ·
K FM
m  
 cmPM ˜ xmM  K mFM *  cmPM * ˜ xmM * d CiP  
¨ ¦ C j  CmPM ¸ d CiP
¨
Mm © j 1 ¸
¹
Since the upper border did not change, we need only to focus on the left
condition. For a 3-bank scenario with bank 1 as insourcer and bank 2 as optimal
insourcer, after bank 1 would have left, the condition to be satisfied is:
K 1
F
  
 c1P ˜ x mM  K 2F  c 2P ˜ x1  x 2 
K F  c1P ˜ x1  K 2F  c 2P ˜ x 2  K 3F  c3P ˜ x3  K 1F  c1P ˜ x mM
d C1P  1
3
  
œ 3 c1P  c2P x2  x3  2 K 2F d  K 3F  c2P ˜ x2  c3P ˜ x3  c1P ˜  x2  x3 
œ 2 K  2c x 2  3c 2P x3 t K 3F  c3P ˜ x3  2c1P ˜  x 2  x3 
F
2
P
2
Since i=2 is the second-best insourcer, the following inequation also holds
true:
 
K 2F  c 2P x 2  x 3 d K 3F  c 3P x 2  x 3  
Thus, it becomes clear that the constraint for ensuring group rationality will
generally not be fulfilled for K 2F  K 3F .

5.1.4.2 Proportional Allocation of Costs

The second analyzed allocation scheme would be every player bearing a share of
the coalition’s total costs proportional to the player’s process volume (Equation
48).
An outsourcer must fulfill the following conditions in order to ensure stable
coalitions (cf. Equation 51):
§ K FM · PM K mFM PM K iF
cmPM ˜ xi d xi ˜ ¨¨ mM  cmPM ¸¸ d K iF  ciP xi Ÿ c m d M  c m d  c iP
© mx ¹ x m x i

The left constraint is obviously satisfied while the right one is fulfilled due
to the subadditivity property of the model (taking part in a coalition always leads
to lower average costs than when processing alone).

145
For example K i
F
^1500, 400,1500`, ciP ^1.0,1.5,1.0`, xi 1000i would lead to an insta-
ble coalition.
Analytical and Simulative Studies 265

This proof implicitly assumes that the insourcer does not change if bank i
leaves the coalition. But, the model always assumes the efficient producing bank
being the insourcer. Consequently, the insourcer might change when the process
volume alters. Thus, the proof is repeated in the following while assuming that, if
bank 1 left the trilateral coalition, bank 3 would become the insourcer, whereas
bank 2 is the insourcer otherwise.
§KF ·
 
K 2F  c 2 x mM  K 3F  c3 x mM  x1 d x1 ˜ ¨¨ M2  c 2P ¸¸ d K iF  ciP xi
© xm ¹
While the right border has again not changed, the lower border is analyzed
below:
x 2  x3 F
K 2  c 2  c3  x 2  x3   K 3F d 0 œ
x mM
x 2  x3 F
K 2  c 2  x 2  x3  d K 3F  c3  x 2  x3  œ
x mM
K 2F K 3F
M
 c2 d  c3
xm x 2  x3
In order to fulfill the assumption that bank 2 dominates bank 3 in the trilat-
eral coalition, the following condition also has to be met:
K 2F K 3F
K 2F  c 2 x mM d K 3F  c3 x mM œ  c 2 d  c3
xmM xmM
M
Since x2  x3 < xm = x1  x 2  x3 , this inequation is more restrictive
than the condition above; thus, the lower border is fulfilled under any given
parameterization.
As a third possible constellation which still has to be tested, we have to as-
sume that i itself is the insourcer. For determining the group rationality constraint
(lower border), another insourcer has to be determined again. The insourcer must
fulfill the following conditions (cf. Equation 50):
xi
K FM
m    
 c mPM ˜ x mM  K mFM *  c mPM * ˜ x mM * d K mFM  c mPM ˜ x mM ˜ 
d K iF  c iP xi 
x mM
As above, ¦x
jM m
j is substituted by x mM and ¦x j by x mM * ( x mM  xi ) .
jM m
j zi

Since i is the insourcer, the following substitutions can be made:

266 Analytical and Simulative Studies

K mFM K iF and c mPM ciP

xi
The right constraint is, of course, always fulfilled: K iF d K iF
x mM
Because of x mM * x mM  xi the left constraint can be reformulated as

x mM *
K mFM ˜  c mPM ˜ x mM *  K mFM *  c mPM * ˜ x mM * d 0
x mM
x mM *
œ K mFM ˜  K mFM * d c mPM *  c mPM  ˜ x mM *
x mM
K mFM K mFM * PM * PM K mFM PM K mFM *
œ  d c m  c m œ  c m d  c mPM *
x mM x mM * x mM x mM *
Since the insourcer provides lowest costs for xmM , the following property
holds true:
K FM K FM *
K mFM  c mPM ˜ x mM d K mFM *  c mPM * ˜ x mM œ mM  c mPM d mM  c mPM *
xm xm
This condition is more restrictive than the lower border of the core. Conse-
quently, it can be stated that a trilateral coalition is always stable if a proportional
cost allocation scheme is adopted.
Since the proportional cost allocation showed to be stable in any case of the
trilateral coalition, the analysis is extended to an n-banks scenario. The proof of
meeting the upper border does not differ because it is determined by individual
process costs. By contrast, the lower border (group rationality) in the n-lateral
coalition must additionally hold true for every possible sub-coalition S m of coa-
lition M m . We distinguish between sub-coalition S m and the remaining set of
actors M m / S m . It has to be proven that every possible sub-coalition S m within
coalition M m has to bear at least the marginal value (cost savings) it attributes to
the coalition (= lower border of the core) in order to ensure that the remainder
M m / S m has no incentive to break out.

xi
K FM
m    ¦ §¨¨ K
 c mPM ˜ x mM  K mFM / S  c mPM / S ˜ xmM / S d FM
m 
 cmPM ˜ x mM ˜
x mM
·
¸
¸ S m  M m
iS m © ¹

xmS

œ K mFM  cmPM ˜ xmM  K mFM / S  cmPM / S ˜ xmM / S d K mFM  cmPM ˜ xmM ˜  xmM
S m  M m
Analytical and Simulative Studies 267

xmS
œ K mFM  cmPM ˜ xmM  K mFM / S  cmPM / S ˜ xmM / S d K mFM ˜  cmPM ˜ xmS S m  M m
xmM

xmS
œ K mFM  cmPM ˜ xmM / S  K mFM / S  cmPM / S ˜ xmM / S d K mFM ˜ S m  M m
xmM

xmS
 
œ K mFM  K mFM / S  cmPM  cmPM / S ˜ xmM / S d K mFM ˜
xmM
S m  M m

xmM / S
œ K mFM ˜
xmM
 
 cmPM  cmPM / S ˜ xmM / S d K mFM / S S m  M m

K mFM PM K mFM / S
œ  c m d  cmPM / S S m  M m
xmM xmM / S
This is always fulfilled due to the linear form of the cost function and the
larger process volume on the left side (monotonically decreasing average costs).
In conclusion, it can be said that the allocation of the coalition costs which is
proportional to the process volume always leads to a stable coalition.

5.1.4.3 Shapley Allocation

Shapley showed that if games are subadditive but not convex, the Shapley value
may fall outside the core of the game (Shapley 1953). Although our model does
not lead to convex games, the particular structure of our model might neverthe-
less lead to the Shapley value always being in the core. In order to test this, we
reformulate Equation 50 to a 3-bank scenario with i=1 as insourcer and examine
the Shapley value for bank i=2. In a first step, we focus on the lower border.
0!˜2! F 1!˜1!
c1 x 2 d
3!

K 2  c2 ˜ x2  
3!
  
Min C1P  x1  x 2 ; C 2P  x1  x 2   C1P  x1  
1!˜1!
3!
   
Min C 2P  x 2  x3 ; C 3P  x 2  x3   C 3P  x3  
2!˜0! F

3!
 
K 1  c1  x1  x 2  x3   Min C1P  x1  x3 ; C 3P  x1  x3  
Assuming C1P ( x )  C 2P  x   C3P x  x leads to
1 P 1 1 1 1 1 1
c1 x2 d
3
 
C2 x2   C1P x1  x2   C1P x1   C2P x2  x3   C3P x3   C1P x1  x2  x3   C1P x1  x3 
6 6 6 6 3 3
1 P 1 1 1
œ 0d C2 x2   C3P  x3   c1P ˜ x2  c2P ˜ x3
2 6 2 6
268 Analytical and Simulative Studies

1 1 1
œ 0 d K 2F  K3F  c2P ˜ x2  c3P x3  c1P x2  c2P x3
3 3 3
Now, one can easily see that, if K3F !! K 2F , the right side of the inequation
will become negative, i.e. the group rationality constraint will be violated.
As a result, it can be said that applying the Shapley allocation to linear cost
functions (with strictly positive fixed costs) does not generally lead to stable
coalitions.
Further, the insourcer must fulfill the following conditions:
K FM
m  
 cmPM ˜ xmM  K mFM *  cmPM * ˜ xmM * d 
( S m  1)!( M m  S m )!
¦ Mm !
>K FS
m   
 cmPS ˜ xmS  K mFS  cmPS ˜ xmS  xi @ d K i
F
 ciP xi 
Sm M m

In the following 3-bank scenario, i=1 is assumed to be the insourcer and i=2
the second best insourcer (after i=1 would have left the coalition) in a 3-bank
scenario.
0!˜2! P 1!˜1!
K 1
F
  
 c1P ˜ xmM  K 2F  c2P ˜ x2  x3  d
3!

C1 x1   
3!
   
Min C1P  x1  x2 ; C 2P x1  x2   C2P x2 

1!˜1! 2!˜0! P

3!
 
Min C1P x1  x3 ; C3P  x1  x3   C3P x3   3!
 
C1 x1  x2  x3   Min C2P x2  x3 ; C3P x2  x3  
Assuming again that C1P ( x)  C2P x   C3P x  x leads to

2 P M 1 P 1 1 1 1 2
3
 
C1 xm d C1 x1   C2P x2   C3P  x3   C1P x1  x2   C1P x1  x3   C2P  x2  x3 
3 6 6 6 6 3
1 P 1 P 1 P 2 P
œ 0 d C2  x2   C3  x3   c1 ˜  x1  x2   c2 ˜ x3
2 6 2 3
1 F x3
œ 0 d K 2  K 3  x2 c2  c1  
F

3 3

4c2  c3  c1P ˜ x1
P P

F F
If K !! K , the right side can become negative, again. Consequently, for
3 2

the insourcer the group rationality constraint is not generally fulfilled by the
Shapley allocation, too.

5.1.4.4 Threshold for Transaction Costs

Incorporating transaction costs, which are not functionally related to the process
costs, circumvents any general conclusions about both the availability of a core
and of the general stability of the investigated allocation mechanisms. Neverthe-
less, a threshold value can be computed, which determines the maximum level of
transaction costs every coalition member is able to bear unless the coalition be-
comes instable (or even inefficient due to an empty core). We assume that trans-
Analytical and Simulative Studies 269

action costs occur at an individual level and are not transferable. Thus, they re-
duce the core by raising the lower borders but are not part of the bargaining pie.
In the following, we restrict our analysis to a trilateral coalition, i z in-
sourcer, and to the proportional cost allocation mechanism because it is the only
one that has proved to be stable. For determining bank i's position in the core,
T
Equation 51 has to be supplemented by Cim (Equation 52):
T
2 ˜ C im CG § K F ·
c mPM xi  2 ˜ C im
T

d C iG d K iF  ciP xi œ c mPM  xi
d i d ¨¨ i  ciP ¸¸
xi © xi ¹
Equation 52: Cost interval from actor i's perspective, incl. transaction costs (i z
insourcer)
T
The level of Cim (relative to process costs) determines the presence of
subadditivity and therefore, according to the outsourcing literature, the advanta-
geousness of cooperative sourcing. In order to join a coalition with a propor-
tional allocation mechanism, the following constraint has to be met (in addition
to a positive core (Equation 52) and the individual rationality border):
T
2 ˜ C im K FM x K FM
c mPM  d mM  c mPM œ C im
T
d i Mm
xi xm 2xm
Equation 53: Threshold for ensuring coalition stability, including transaction
costs for trilateral coalitions
This threshold neglects the fact that transaction costs will also occur in a bi-
lateral coalition. Considering individual transaction costs for managing a bilat-
T ,bilateral
eral coalition C im leads to


T ,trilateral
c mPM xi  2 C im T ,bilateral
 C im  
d C iG d K iF  ciP xi œ 
c mPM 
2C T ,trilateral
im C T ,bilateral
im d K FM
m
 c mPM œ
M
xi x m

T ,trilateral xi K mFM
T ,bilateral
C im d C im 
2 x mM
Equation 54: Threshold for ensuring coalition stability, including transaction
costs for trilateral and bilateral coalitions
270 Analytical and Simulative Studies

5.1.5 Experiment
In January 2005 and December 2005, the model was used to conduct a bargain-
ing game involving student seminar participants in order to test how closely
bargaining results would reach the allocation mechanisms investigated above.
The game consisted of three rounds, each increasing in complexity. In every
round, there were eight games (“tables”) including three “banks” (each consist-
ing of two students) with identical cost structures across the games. In a game,
the a priori information given to each team included the own process cost struc-
ture, production quantity, and the level of transaction costs if joining a coalition.
Each round ran (a maximum of) nine periods, in which each bank sequentially
could make an offer (“want to insource”: price/volume) or a request (“want to
outsource”: price/volume). The game was abandoned after a bilateral or trilateral
coalition had been established. Between the rounds, students changed tables and
positions to ensure that all students only played once against each other.

Round 1 – large core, no transaction costs

In the first setting, process cost parameters that lead to a rather large core were
chosen. This enabled the different parties to exploit a rather wide negotiation
space. Furthermore, no transaction costs occur if a coalition is formed.
Table 40 and Figure 100 show the parameters and the results of the first
round. The optimal insourcer for the total process volume of 600 is bank A (total
process costs to be allocated: 300 + 1.0 * 600 = 900).
Bank A Bank B Bank C
K iF 300 200 100
P
c i
1.0 2.0 3.0

xi 100 200 300

Table 40: Parameterization of experiment – round 1 (A = optimal insourcer)
Analytical and Simulative Studies 271

Cost allocation vectors (A / B / C)

Shapley value: -16.7/333.3/583.3 C: 1000

Proportional allocation: 150/300/450
Equal distribution: 33.3/233.3/633.3 C
(0,0,900)
table 1: -10/340/570
table 2: -60/390/570
table 4: -80/420/550
table 5: 180/300/420
table 6: -50/520/430
table 7: -35/415/520
table 8: 35/400/465
Table 3 cannot
be displayed because C: 300
the players decided on a
0

sub-optimal coalition,

0
20
0

60
B:

leading to total
B:

B:
costs higher than 900.
A:

A:
A:
40

-30
0
0

0
A C: 0 B
(900,0,0) (0,900,0)

Figure 100: Results of experiment – round 1

The cost triangle in Figure 100 shows the resulting cost allocations from the
bargaining games compared with the theoretical allocation mechanisms. All but
one game (table 3) resulted in the optimal coalition (insourcer A) and agreed
with a coalition being in the core. Furthermore, the allocation values of most
insourcers (A) (except table 8) were very close to the allocation result deter-
mined by the Shapley value. All insourcers exploited their bargaining position
quite well. Furthermore, four of these tables also agreed with an allocation quite
close to the Shapley allocation from the perspective of both outsourcers as well.
By contrast, the players at table 5 agreed almost exactly with the proportional
cost allocation.

Round 2 – small core, no coordination costs

In the second round the core was significantly reduced. Negotiators had to act
more cooperatively to even find the optimal insourcer (= C, leading to total proc-
ess costs of 4,700), due to the information asymmetries.
272 Analytical and Simulative Studies

Bank A Bank B Bank C

K iF 500 200 300
P
c i 4.0 5.0 4.0
xi 500 350 250
Table 41: Parameterization of experiment – round 2 (C = optimal insourcer)
C
Cost allocation vectors (A / B / C) (0,0,4700)

Shapley value: 2158/1583/958

A:
Proportional allocation: 2136/1495/1068

0
Equal distribution: 2150/1600/950
0
B:

table 1: 2200/1520/980
table 4: 2125/1500/1075
400

table 5: 2100/1770/830
B:1

table 6: 2350/1550/800 0 5
A:

19
20 0

B:

Tables 2,3,7, and 8 cannot

0

be displayed because
A:

they decided on sub-optimal

25
00

coalitions, leading to total

costs higher than 4700.
C: 1300

C: 800

A B
(4700,0,0) C: 0 (0,4700,0)
Figure 101: Results of experiment – round 2
Only four of eight tables arrived at the optimal solution (Figure 101). Table
2 chose the wrong insourcer while tables 3, 7, and 8 involved an agreement of
two parties to form a bilateral coalition, leaving the third bank out. None of these
can be displayed in the cost triangle because they lead to higher total costs.
The result of table 4 almost exactly matches the proportional allocation. The
insourcers (C) of table 5 and 6 exhausted their bargaining power and pushed the
solution to the border of the core. The insourcer of table 1 also comes close to
the Shapley value.
Analytical and Simulative Studies 273

Round 3 – small core, with transaction costs

In the final round, transaction costs were introduced, increasing with coalition
size (individual transaction costs for bilateral coalition = 100, for trilateral coali-
tion = 150). This again resulted in a small core, which was not found at any of
the tables in this round. Furthermore, due to the relatively high transaction costs,
all of the theoretical allocation schemes discussed above are outside the core.
The triangle in Figure 102 shows the allocation of total costs, i.e. optimal process
costs = 1,475 + 3 * transaction costs (=150) = 1,925. Nevertheless, the transac-
tion costs of 150 per coalition member are not part of the negotiation pie.

Bank A Bank B Bank C

K iF 200 800 100

P
c i 1.5 1.0 2.0

xi 250 100 500

T 100 for each bank when joining a bilateral coalition

C im
150 for each bank when joining a trilateral coalition
Table 42: Parameterization of experiment – round 3 (A = optimal insourcer)
Only three of the eight tables established a trilateral coalition with A as the
optimal insourcer, but, what is even more interesting, is that none of them found
the core (Figure 102). In each of these cases, one player was able to push his part
of the allocated costs below the group rationality constraint. In these cases, the
other players accepted more costs than would occur in a bilateral coalition. At
table 7, for example, banks B and C together bore 1,620, although a bilateral
coalition would have result in costs of only 1,500. Further, table 5 came quite
close to the proportional allocation. It can be stated that the complexity in this
round was too high to enable the players to follow rational strategies.
Furthermore, sometimes, the parties identified the globally optimal solution
but nevertheless – although all parties would have realized cost savings – did not
agree to an offer because the savings were perceived to be too asymmetrically
(“the other bank saves much more than me, therefore, I do not take part”).
274 Analytical and Simulative Studies

Cost allocation vectors (A/B/C)

Shapley value: 408/558/958 C
Proportional allocation: 584/324/1018 (0,0,1925)
Equal distribution: 358/683/883

Table 1: 395/500/1030
Table 5: 555/345/1025
Table 7: 305/550/1070

Tables 2,3,4,6, and 8 cannot C: 1100

be displayed because C: 1000
they decided on sub-optimal
coalitions, leading to total
0
B:

costs higher than 1925.

A:
0
40

0
B:

A:
0

A:
90

425
575
B:

A B
(1925,0,0) C: 0
(0,1925,0)
Figure 102: Results of experiment – round 3
5.1.6 Conclusion
Based on a simplified form of the cooperative sourcing model, this section –
using game theory – analyzed how costs have to be allocated in a cooperative
sourcing coalition to ensure its stability. While the equal allocation of gain, the
proportional allocation of costs, and the Shapley allocation were tested, only the
proportional distribution ensures stable coalitions. Although the other schemes
do not lead to unstable coalitions in all cases, determining them ex ante when
founding a coalition results in the problem that, with new members joining the
coalition in later periods, the allocation scheme would have to be completely
renegotiated.
By contrast, within the experiment most of the participants did not agree
with a cost allocation that is close to this proportional distribution, but instead is
closer to the Shapley value. With increasing coordination difficulty (shrinking
core, advent of transaction costs), the bargaining games increasingly resulted in
inefficient constellations, illustrating the impact of bounded rationality, even in
such simple scenarios. In the last round, the players were no longer able (or
sometimes not willing) to follow a “rational” strategy and instead favored ineffi-
cient outcomes.
Analytical and Simulative Studies 275

The observation that optimal coalitions often did not emerge because parties
deviated from the “rules”, such as common rational behavior (i.e. maximize
savings), leads to the presumption that a common rational behavior among the
agents cannot be guaranteed even in the simplest settings although almost all
microeconomic models are based on this assumption. Future research has to
explore if this result stems from “irrationality” or from the design of the experi-
ment.

5.2 A Genetic Algorithm for Solving the CSP

The centralized variant of the model represents a non-linear and binary (i.e.
combinatorial) optimization problem which – if solved – determines the optimal
cooperative sourcing constellation for a given set of banks and business func-
tions from a centralized perspective (maximizing global net savings). For many
combinatorial problems, there are no algorithms available which provide optimal
solutions in polynomial computing time. Instead, dedicated heuristics are devel-
oped or meta-heuristics (e.g. genetic algorithms, taboo search, simulated anneal-
ing) are adapted to the particular problem structure, in order to find at least
“good” solutions to the given problem.
The cooperative sourcing problem (CSP) developed in section 4.3 shows
structural similarities to another coalition clustering problem – the optimal con-
sortia structure problem (OCSP) (Beimborn et al. 2004, Fladung 2006). The
OCSP determines optimal library consortia for cooperatively procuring elec-
tronic journals from publishers to minimize their procurement and administrative
costs. Fladung (2006) shows that the problem (with realistic size) is not solvable
by exact optimization routines (e.g. Branch&Bound) in a realistic time frame.
Compared with the OCSP, the CSP shows basically the same problem struc-
ture but adds further complexity by not only cooperatively sourcing one business
function (procurement of e-journals) but multiple activities K. Thus, the CSP
structurally represents K interrelated146 OCSPs. Thus, it can be argued that no
exact solution algorithm exists for the CSP, either.
The OCSP was efficiently solved (heuristically) by a genetic algorithm ap-
proach (Beimborn et al. 2007b). Since the OCSP and the CSP show the same
problem structure, it can be argued that the same algorithm can be efficiently
applied to the CSP as well. This section follows (Beimborn et al. 2007b) to de-
velop an adaptation of the genetic algorithm approach to the CSP.

146
The interrelation is given by the interface costs which are determined by the degree of task inter-
dependencies between two business functions (cf. section 4.2.3.1.4). If task interdependencies did
not exist in the CSP, it could be separated to K OCSPs which could be solved separately.
276 Analytical and Simulative Studies

5.2.1 Basics
Genetic algorithms (GAs), as a subtype of population-based metaheuristics (Sil-
ver 2004, 950), try to emulate natural evolutionary processes of biological organ-
isms (Beasley et al. 1993a). In general, GAs work with a population of individu-
als, each representing a feasible solution for the given problem. Each individual
is assigned a ‘fitness score’ (e.g. total net savings of the CSP) representing the
quality of the solution. Individuals with high fitness are given opportunities to
‘reproduce’ with other individuals in the population in order to generate a new
generation of individuals while the least fit members die without reproduction. In
this way, the population is intended to converge to the optimal solution (i.e.
maximum reachable fitness value) of the given problem (Kratica et al. 1998).
In order to implement a GA, a suitable genetic coding for the problem must
be devised. This problem representation is of great importance for the perform-
ance of a GA to solve a given problem (Rothlauf 2006). Thereby, it is assumed
that a potential solution can be represented as a series of values. GAs are often
based on a binary representation of the decision variables (known as genotype
layer, Beasley et al. 1993a, 1993b); the decoded construction of the values is
called phenotype layer (Wendt 1995, 68-69).
Based on the genotype representation of the solution, a specified fitness
function determines the quality of each individual. The selection method chooses
and matches the pairs of individuals which will be crossed while the recombina-
tion strategy (also called crossover) determines the way in which the genetic
codes of the two parents will be mixed to create their descendants (Beasley
1993a).
Finally, the mutation process causes little changes in the genetic code of the
descendants to prevent the premature convergence of the GA to suboptimal solu-
tions (Kratica et al. 1998, Nissen 1995). In order to repair invalid solutions or to
avoid redundant solution representations in the genotype layer, repair operators
have to be implemented (Kratica et al. 1998). The algorithm terminates when a
certain period TGA is reached or when a pre-defined stopping rule has been ful-
filled (e.g. “no solution improvement during the last 20 generations”). Figure 103
depicts the basic steps of GAs in pseudo-code notation.
Analytical and Simulative Studies 277

InitializePopulation();
while not Finish() do {
for i:=1 to Npop do
pi:=Objective:Function(i);
FitnessFunction();
Selection();
Crossover();
Mutation();
Repair ();
}
Figure 103: Basic form of the genetic algorithm (Kratica et al. 1998)
5.2.2 GA Design
In this section, the GA approach is adapted to the particular structure of the CSP.

Genetic representation
As for the OCSP (Beimborn et al. 2007b), a non-binary representation of the
solution structure (genotype) was chosen. The genetic code represents informa-
tion about which banks cooperatively source a certain business function k. The
so-called sourcing entities (being either a coalition or a bank that is not member
of a coalition) receive unique identifiers. The following figure shows a small
example for a problem consisting of six banks and two business functions.
banks
i=1 i=2 i=3 i=4 i=5 i=6

business k=1 1 1 1 2 2 2
functions
k=2 1 2 3 4 5 6

Figure 104: Exemplary genotype representation of an individual

While the first business function is cooperatively sourced by all banks (coa-
lition 1 consists of banks 1, 2, and 3 while coalition 2 consists of banks 4, 5, and
6) the second business function is self-operated by each of the banks. Thus, for
k=1, there are two sourcing entities, while there are six for k=2. The genetic code
of one individual consists of a |I|x|K| matrix with |I| being the number of firms
and |K| being the number of business functions that are part of the problem in-
stance147.

147
This represents the main difference between the cooperative sourcing problem and the optimal
consortia structure problem from (Beimborn et al. 2004; Fladung 2006). The genetic representa-
tion of an OCSP individual comes along with only one row instead of |K| rows.
278 Analytical and Simulative Studies

Mapping this solution to the phenotype (i.e. decision variables of the formal
representation developed in section 4.3) results in the following constellation:
y1110 = z1111 =1 y2110 = z2111 =1 y3110 = z3111 =1 y4120 = z4121 =1 y5120 = z5121 =1 y6120 = z6121 =1
y1200 = z1201 =1 y2200 = z2201 =1 y3200 = z3201 =1 y4200 = z4201 =1 y5200 = z5201 =1 y6200 = z6201 =1
yikm0 = zikm1 = 0 for all other i,k,m
Table 43: Phenotype representation of the solution given in Figure 104148
It should be noted here that the genetic code of the individual does not ex-
plicitly contain information about the coalitions’ insourcers because optimal
insourcers of the coalitions can be computationally derived when the coalition
members are known.

Selection operator
The selection operator decides which pairs of individuals from a created popula-
tion will be crossed. The pairs are selected based on the individuals’ fitness
which is determined by the objective function of the CSP (Equation 28 on p.
246) giving individuals with high fitness a higher probability of being consid-
ered. For this procedure, the binary tournament selection operator is used
(Thierens and Goldberg 1994). First, two individuals are chosen from the popu-
lation at random. The individual with the highest fitness of the two is copied to
an intermediate population (mating pool). This procedure ends when the mating
pool contains the same number of individuals as the size of the population itself.
Thus, in the mating pool, the population has been stochastically filtered from
inferior individuals (Harvey 1994, 301-302). From the mating pool, |I|/2 pairs
are randomly selected to be crossed without taking the fitness into account any-
more, according to (Thierens and Goldberg 1994; Harvey 1994).

Crossover operator
The matched pairs of individuals are crossed with a given crossover probability
pCO to generate two children. If no crossover takes place (1-pCO), both individu-
als unaltered become members of the next generation (Beasley 1993b; Nissen
1995).
During crossover, each item of the genetic code (called alleles) of the de-
scendants is created by copying the corresponding allele from one or the other
parent chosen according to a randomly generated (binary) crossover mask (uni-
form crossover, Beasley et al. 1993b). In order to achieve a structural perpetua-

148
To remind the reader: the binary variables yikm0 and zikm1 are equal to 1 if bank i optimally will
source business function k to coalition km. m=0 stands for not joining any coalition.
Analytical and Simulative Studies 279

tion of the parental code, the standard uniform crossover has to be extended: if
the chosen part of the parental genetic code indicates that the corresponding firm
is part of a coalition, the information about the structure of this particular coali-
tion will be bequeathed.
Figure 105 illustrates the extended uniform crossover procedure for a sce-
nario with six firms and one business function (|K| = 1). If there is a “1” in the
crossover mask, the genetic information of child 1 is copied from the first parent,
while in the case of a “0” in the mask, the gene is copied from the second parent.
For the second child, it is the other way round.
banks
i=1 i=2 i=3 i=4 i=5 i=6

Parent 1 1 2 1 2 1 2

Parent 2 1 1 2 2 1 2

Crossover mask 1 0 1 0 0 1

Child 1 1 2 1 3 1 3

Child 2 1 1 2 2 1 2

Figure 105: Schematic illustration of the extended uniform crossover

(Beimborn et al. 2007b)
The first value in the crossover mask indicates that the first child inherits the
structural information from parent 1. Thus, the first, third and fifth digit – repre-
senting one coalition – of the child’s chromosome are set to “1” in order to con-
serve the structure of this coalition. The second number of the mask shows that
the second digit should be set according to the corresponding allele of parent 2.
In this case only the second allele of the child’s chromosome will be determined
because the first and fifth allele has already been set. Because “1” is already used
in the child’s chromosome, the next free value (“2”) is used here.
The third position of the child’s genetic code is already determined; thus,
this position in the crossover mask will be ignored. The fourth digit of the mask
again indicates to use information from parent 2. Thus, the first child inherits the
structural information that banks 4 and 6 form a coalition. In this way, the whole
chromosome vector of both children can be deduced from the corresponding
280 Analytical and Simulative Studies

parents. If more than one business function is part of the problem, this crossover
procedure will be repeated for each row of the genetic code matrix (cf. Figure
104).

Mutation operator
After generating a new population, the genetic code is slightly modified accord-
ing to a mutation operation. The basic idea is that each digit in the offsprings’
chromosomes is increased or decreased by 1 with a given probability pM. From
four different approaches given in (Schwefel 1981)149, the tests in (Beimborn et
al. 2007b, Fladung 2006) showed a constant mutation probability pM over time to
be most effective. Literature suggests that the initial value should be set to the
reciprocal value of the size of the chromosome (i.e. |I| for the OCSP (Fladung
2004, Beimborn et al. 2007b) and |I|*|K| for the CSP).
Business functions which must not be outsourced for strategic reasons
(Equation 20, p. 243) are marked by a ban flag in the chromosome matrix and
thus are not handled by crossover and mutation operations.

Repair step operator

Finally, the individuals of a new generation have to be repaired for two reasons:
first, due to the crossover operator, there may be redundant solutions which un-
necessarily bloat the solution space (Rothlauf and Goldberg 2003). Second, in
contrast to the OCSP, the CSP contains additional constraints which affect the
optimal solution (strategic constraint (Equation 19, p. 242) and legal constraint
(Equation 38, 254)).
In order to diminish redundancy (first reason), the repair step preserves the
structural integrity of each individual and re-sorts the chromosome strings in
such a way that the digits receive only ascending numbers (by first occurrence)
without vacancies.
Figure 106 shows an example of how this repair operator works: The sec-
ond, fourth, fifth and sixth digits in the unrepaired chromosome have to be modi-
fied so that the string shows an ascending order. Furthermore, the unrepaired
chromosome contains a lack which will be reduced (“4” is not used). Both indi-
viduals represent the same solution of the modeled problem (i.e. same pheno-
type).

149
Other approaches are: linearly decreasing pM over time, hyperbolic decreasing pM over time, and
hybrid procedures with constant probability during the first periods and decreasing probability af-
ter a certain threshold period (Schwefel 1981).
Analytical and Simulative Studies 281

banks
i=1 i=2 i=3 i=4 i=5 i=6

Unrepaired
1 3 1 2 5 2

Identical structure
chromosome

Repaired
1 2 1 3 4 3
chromosome
Figure 106: Schematic illustration of the implemented repair operator
(Beimborn et al. 2007b)
The legal and the strategic constraint of the CSP are regarded as follows. Af-
ter a new individual has been created by the crossover procedure, the GA exam-
ines its compliance with those constraints. If they are violated, one cooperatively
sourced business function is chosen at random and backsourced to in-house
processing. This procedure is repeated until all constraints are met.

5.2.3 Configuration
The development of the genetic algorithm for the OCSP in (Beimborn et al.
2007b; Fladung 2006) was accompanied by comprehensive tests to derive the
optimal configuration. In those works the parameterization given in Table 44 was
finally found to be most effective after applying it to different OCSP scenarios
with different levels of solution difficulty.
Population size = 200 TGA = 100 or 1,000 pCO = .55 pM = .05
Table 44: Optimal GA configuration for the OCSP (Beimborn et al. 2007b)
Genetic algorithms, like every heuristic, provide a trade-off between solu-
tion quality and computation time. The longer a heuristic is allowed to search for
the optimal solution, the better the final result will be. Compared with (Beimborn
et al. 2007b; Fladung 2006), this work on hand differs in the purpose of the GA
application. While the first was intended to optimize a particular real situation by
determining how academic libraries of two German states should be organized in
procurement consortia (given very rich and detailed empirical data), the objec-
tive of this work is to simulate cooperative sourcing behavior in different scenar-
ios. Here, the GA is applied to provide a benchmark from a centralized perspec-
tive in order to identify inefficiencies occurring in systems that are not centrally
organized. Thus, this work has differing requirements regarding the trade-off
between solution quality and computation time. While a single optimization run
282 Analytical and Simulative Studies

of 41 minutes for the OCSP was acceptable (Fladung 2006, 183), for the coop-
erative sourcing studies this would lead to simulations (which require high num-
bers of repetitions) taking years for fulfillment150. Based on the findings in
(Beimborn et al. 2007b, Fladung 2006) and on own performance tests with a
CSP instance consisting of 50 banks with 5 business functions each, the configu-
ration given in Table 45 was chosen to solve the centralized variant of the coop-
erative sourcing model during the simulation studies in the next section.
Population size = 50 TGA = 100 pCO = .55 pM = .001
Table 45: GA configuration for the CSP simulation studies
The GA is restricted to TGA =100 periods (i.e. generations) and to a popula-
tion of 50 individuals. The crossover probability is set to .55 as in (Beimborn et
al. 2007b, Fladung 2006) while the mutation probability is .001 (instead of .002
as the suggested reciprocal value of the problem size). The underlying perform-
ance tests are documented in appendix A1.
During the simulation studies in the next section, the genetic algorithm will
be used to determine the “optimal” global cooperative sourcing configuration
where the results of simulating decentrally and autonomous decision behavior of
the firms can be compared with in order to determine the degree of inefficiency
which occurs in the system. The attribute “optimal” has been put in quotation
marks because the genetic algorithm as a heuristic does not necessarily deter-
mine the true optimal solution and – even if it did – the solution would only be
optimal from a global perspective, but not necessarily from an individual per-
spective.

5.3 Simulation Studies

5.3.1 Agent-based Simulations as Research Approach
Simulation is a rather young field in the social sciences which started growing
fast in the last two decades (Axelrod 1997). “Simulation means driving a model
of a system with suitable inputs and observing the corresponding outputs” (Brat-
ley et al. 1987, ix). There is often no possibility of manipulating a real system in
order to answer what if questions. If this system is modeled appropriately, simu-
lation studies can help to explore the effects of changing system parameters. In

150
Furthermore, it should be noted that the optimized library network in (Fladung 2006) consisted of
20 libraries, while the simulation studies in the subsequent sections are conducted with 100 banks
and 5 business functions (instead of one procurement process). Thus, the problem size is 25 times
larger.
Analytical and Simulative Studies 283

addition, simulations are helpful for making estimates about what happens inside
a system that is difficult to understand, e.g. in a national economy (Bratley et al.
1987, 3).
As a research method, simulations can contribute to both inductive and de-
ductive research approaches (Axelrod 1997). Induction discovers patterns in
empirical data while deduction “involves specifying a set of axioms and proving
consequences that can be derived”, such as the discovery of equilibrium results
in game theory (Axelrod 1997, 24). Simulation, like deduction, starts with a set
of assumptions, but in contrast does not prove theorems. Instead, a simulation
generates data that can be analyzed inductively. However, in contrast to induc-
tion, the data comes from a rigorously specified set of rules rather than from
direct measurement of the real world (Axelrod 2000). “While induction can be
used to find patterns in data and deduction can be used to find consequences of
assumptions, simulation modeling can be used as an aid intuition” (Axelrod
1997, 24). Simulations can often be used when deduction is not possible, e.g. if
many non-linear relationships are present and analytics fail. Highly complex
models can be designed to represent a real-world system. However, while deduc-
tively proved theorems are definitely true, simulation results always depend on
parameter settings. Thus, results are always determined with a degree of confi-
dence and cannot be generalized. Consequently, simulations represent a “second-
best technique” as long as deductive approaches can be applied (Axelrod
2000)151.
In economics, simulations are applied to find the determinants of economic
development processes and to identify conditions for reaching certain equilibria
(Medeiros Rivero et al. 1999). Although there are deductive approaches to an-
swer these types of questions (e.g., based on game theory), the models underly-
ing those approaches have to remain very abstract, simple, and small. By con-
trast, economic systems are made up of economic agents with complex behavior
which interact dynamically with each other. Thus, the resulting dynamics of the
system often are unpredictable (Flake 1998). Medeiros Rivero (1999) states that
“not only the structure of the economic system emerges from the individuals’
behavior, but that the agents’ behavior is influenced by the structure” as well.
The dichotomy between complexity in real systems and the high level of abstrac-
tion in analytical models can be bridged in parts by simulations which are flexi-
ble enough to capture higher levels of formal complexity while still providing
explanation power (Medeiros Rivero et al. 1999).

151
A detailed positioning of the simulations approach from a philosophy of science perspective is
given by Axelrod (1987).
284 Analytical and Simulative Studies

One possible way of going about this is formulating differential non-linear

equation models (via econometrics) and solving them by applying numerical
simulations (Bossel 2004, Medeiros Rivero et al. 1999). This method is very
common when handling complex models of economies (Bossel 2004). Short-
comings of this approach are still difficult analytics (derivation of the differential
equations), changes in the actors’ behavior leading to the necessity of re-deriving
the equation system, the assumption of homogeneity of the agents (at least within
their groups), and missing ability to incorporate agents’ expectations about the
future (Medeiro Rivero 1999).
A second strand of simulations-based research handles these problems by
explicitly considering the micro-economic structure of economics systems and
developing models consisting of virtual agents with individual decision behavior
and learning capabilities. Methodologically, this approach has been called agent-
based modeling (ABM, Axelrod 1997) or – when reduced to the economic do-
main – the paradigm of agent-based computational economics (ACE, Tesfatsion
2002a+b, 2006). It “is the computational study of economies modeled as evolv-
ing systems of autonomous interacting agents" (Tesfatsion 2002b), each being
represented as a piece of software. Agents in ACE models can be described as
reactive and interactive goal-directed entities, strategically aware of both com-
petitive and cooperative possibilities with other agents, who are able to form and
utilize expectations about the future (Franklin 1997; Tesfatsion 2006). “The
modeled economic system must be able to develop over time solely on the basis
of agent interactions, without further interventions from the modeler” (Tesfatsion
2006, 8).
This type of computational model allows to simultaneously analyze the im-
pact of a variety of local and global influences on system behavior that is other-
wise very difficult to accomplish. "One principal concern of ACE researchers is
to understand why certain global regularities have been observed to evolve and
persist in decentralized market economies despite the absence of top-down plan-
ning and control […]. The challenge is to demonstrate constructively how these
global regularities might arise from the bottom up, through the repeated local
interactions of autonomous agents" (Tesfatsion 2002b). “How can economic
systems be more fully understood through a systematic examination of their
potential dynamical behaviors under alternatively specified initial conditions?”
(Tesfatsion 2006, 9) This methodological approach thus focuses on how struc-
tures emerge in decentralized networks (bottom-up, “decentralized”) rather than
being explicitly planned and rationally implemented (top-down, “centralized”).
One can therefore argue that the aim of agent-based modeling follows the per-
spective of Nobel laureate Thomas Schelling in his work Micromotives and
Macrobehavior (Schelling 1978).
Analytical and Simulative Studies 285

Of course, even highly complex ACE models still represent an abstract pic-
ture of reality. It is not aimed (and not possible) to provide an accurate represen-
tation. “Instead, the goal of agent-based modeling is to enrich our understanding
of fundamental processes that may appear in a variety of applications” (Axelrod
1997, 25). Models of dynamic systems are not developed to act as forecast tools.
Their objective is to investigate the spectrum of development potentialities which
allows them to offer qualitative decision support in steering the system towards
desired states (Bossel 2004, 110).
Nevertheless, this comes with the strong requirement to validate simulation
models. ACE simulations generate outcome distributions for theoretical eco-
nomic systems. These outcome distributions often suggest multiple equilibria. In
contrast, the real world represents only one instance of the modeled system. Even
if the simulations perfectly reproduced the real system, it would be impossible to
verify this accuracy using common statistical methods (Tesfatsion 2006, 12).
Sargent (1998), Bratley (1987), and Naylor et al. (1967) suggest multistage vali-
dation approaches which are applied to this research work in a separate section in
chapter 6.

5.3.2 Simulation Procedure

In the following, the process of simulating the cooperative sourcing model will
be described.
The scenario is set in an initial period t=0, i.e. all banks are instantiated and
all parameters are determined. After starting the simulation process, banks will
jointly evaluate potential cooperative sourcing contracts with duration TCoSo to
found a coalition or they will individually request membership in an already
existing coalition for one or more of their business functions. The evaluations are
based on the decision functions given in section 4.4.
After a bank has agreed to join a cooperative sourcing coalition, it is bound
to its decision for TCoSo periods. Pre-terminations, although existent in reality, are
not possible. After the contract has terminated, the bank reevaluates its sourcing
decision and can become a member in a more beneficial coalition or backsource
the business function. In the first case, it requests participation in another exist-
ing coalition, whereupon this coalition then votes about the new member’s en-
trance.
Figure 107 gives a small example. It shows three banks in a market segment
that (in t=1) are able to decide cooperatively about sourcing their business func-
tion k=1. Banks 1 and 3 agree to form a coalition while bank 2 does not partici-
pate. Because bank 1 has the cost-efficient business function it becomes the
insourcer of coalition km = 11, represented by a highlighted margin. In t=2, bank
286 Analytical and Simulative Studies

2 requests membership in this coalition. In accordance with the request, the coa-
lition agrees to incorporate bank 2.

Figure 107: Exemplary cooperative sourcing process

After the sourcing contract between bank 1 and 3 has ended in period t=6,
both reevaluate their sourcing decision. They can choose between backsourcing,
requesting membership in another coalition, or renewing the contract. If both
banks leave the coalition, the coalition will be dissolved. If only one bank leaves,
the remaining partners will have to determine a new insourcer.
The simulation is controlled by three parameters:
o initCoalSize determines the (maximum) size of a newly formed coalition. In
a certain period, this number of banks is randomly selected and requested to
form a joint coalition with regard to a certain business function. If one bank
decides against it, the bank is omitted, and the decision process is repeated
with the remaining actors. If it is not possible to establish a coalition with
only three banks, the process is completely canceled.
o coalBuildingFreq determines how often (in terms of number of periods) this
coalition building procedure is repeated.
o maxCoalBuilding determines the maximum number of coalitions (for each
business function k) that will be created within one period. This upper bor-
der is used only for the first period (t=1). In subsequent periods, it is halved.
In all periods, banks which have not outsourced their business function can
apply for entering any existing coalition. Furthermore, all banks which have
expiring coalition agreements can backsource or change the coalition. In order to
select an adequate coalition for requesting membership, different selection crite-
Analytical and Simulative Studies 287

ria are possible, such as random selection, size (in terms of number of actors or
process volume), or efficiency. For the latter, the bank evaluates its financial
consequences which would occur if joining the new coalition (evaluation func-
tion given by Equation 31, p. 250). Finally, it chooses the coalition which prom-
ises highest cost savings. During the simulation studies, the different selection
mechanisms are tested (cf. section 5.3.4.6).
The simulation process usually will be terminated after a certain number of
periods T and then be repeated multiple times with the same parameter setting to
achieve more reliable results of the stochastic system behavior (DoD 1993).
Afterwards, one or more parameters are varied and the same procedure is re-
peated for measuring the impact of the parameter on the system behavior (sensi-
tivity analysis). The complete simulation process is visualized by Figure 108.
After a scenario has been simulated over a certain number of periods, the
genetic algorithm determines the optimal cooperative sourcing constellation of
the banking network from a centralized perspective (maximizing global net sav-
ings).

Implementation
The simulation model and the genetic algorithm have been implemented in
JAVA 6. The main simulation routines consist of five classes, amounting to
2,370 lines of code. Furthermore, the implementation consists of the genetic
algorithm (3 classes, 2,185 lines) and 5 supporting classes (1,013 lines) which
provide data input/output routines as well as random value generation and com-
putation of statistical measures. Eclipse 3.1.2 was used as integrated develop-
ment environment (IDE). Apart from the Java libraries which come along with
Sun’s Java Runtime Environment, several packages from Apache were used to
develop interfaces for reading data from and writing results into Excel spread-
sheets (Jakarta POI152), to conduct statistical analyses, and to generate random
values from empirical distributions (Jakarta Commons-Math153). Uniform ran-
dom numbers are generated by the Colt Random Number Generator from W.
Hoschek at CERN154 while normal random numbers are generated by inversion
(Bratley et al. 1987, 147). Microsoft Excel 2003 and SPSS 11.5 were used for
deriving the parameterization (section 5.3.3) and for analyzing and visualizing
the results (section 5.3.4).

152
http://jakarta.apache.org/poi/ (as of 2006-07-13)
153
http://jakarta.apache.org/commons/math/ (as of 2006-07-13)
154
http://hoschek.home.cern.ch/hoschek/colt/ (as of 2006-07-13)
288 Analytical and Simulative Studies

Figure 108: Simulation process

Analytical and Simulative Studies 289

5.3.3 Parameterization
In order to give the simulation studies a firm basis, it is necessary to collect as
much pertinent data as possible from the system to be simulated (DoD 1993).
Due to a lack of available general banking data, the simulation studies in
section 5.3.4 will focus on a dedicated part of the banking business. Since all of
the studies introduced in section 3.6 focused on a particular part of the credit
business, the simulation studies will cover this domain of granting and adminis-
tering SME loans. Possible generalizations are discussed in the last chapter.
Basically, there are two different approaches to parameterize simulations
from available empirical data:
1. Data set-based (horizontal): The virtual actors in the simulation model rep-
resent real-world actors. Each data set is transformed into the parameter set
of a single simulated agent. For example, we derive bank size and process
cost allocation to the different business functions, their individual strategic
value etc. from one single data set for one of the simulated banks.
2. Distribution-based (vertical): For each of the applied empirical indicators (or
aggregates), a distribution over all data sets is estimated which leads to the
distribution of the corresponding global simulation parameter. During the
different simulation runs, random values are generated based on those de-
rived distributions. Correlations in empirical data can partly be taken into
account by conditional or multivariate distributions (e.g. larger banks have
larger credit volume and differing process cost structures, cf. section
3.6.2.5).
Arguments for the first approach are:
o The simulation would reflect one particular “real-world” situation (as long
as empirical data is assumed to be a “true” reflection of the real world).
o All statistical relationships (correlations) existing in the data are appropri-
ately considered. The second approach usually will not be able to consider
all correlations due to the numerical complexity.
The second approach has the following advantages:
o Subjectivity of the respondents when giving qualitative answers can be lev-
eled as long as there is no structural bias. The answers are reflected against
the remaining data. Thus, a simulation based on empirical distributions is
more robust.
o In contrast to representing one particular real-world situation as in the data
set-based approach, the distribution-based approach considers the general in-
terrelations between system elements and enables differing scenarios. In
290 Analytical and Simulative Studies

other words, simulations are frequently repeated to get more reliable results.
All virtual agents will be completely re-parameterized after each simulation
run. By contrast, the first approach leaves only limited space to vary the set-
ting between different simulation runs.
o In cases where not all parameters can be determined empirically, estimations
and computational proxies have to be included to complete parameteriza-
tion. In these cases, the distribution-based approach is more consistent be-
cause it considers distributions (empirical as well as estimated) for all pa-
rameters. This leads to the results being more robust.
For our simulation studies, we chose the second approach because the simu-
lations do not follow the aim to explicitly forecast market developments what
would not be a realistic objective for simulations of organizational decisions.
Instead, the simulations aim to identify structural effects. This requires a broad
statistical base rather than fixing more individual parameters ex ante than neces-
sary.
The main part of the parameterization is derived from the empirical results
(section 3.6). In areas where no empirical data was available (e.g. transaction
costs), parameterization is based on assumptions which will be altered for sensi-
tivity analyses during the simulation studies. In the following sections, the
parameterization for the simulation studies is derived from the empirical data.
Appendix A2 provides tables which summarize the results.

5.3.3.1 Demographics
In order to be able to handle the computational complexity of the simulations, we
will implement a set of |I| = 100 banks which are implemented in a way to struc-
turally represent the empirically investigated part of the largest 1,000 banks of
the German banking industry (cf. section 3.6) and covering the SME credit busi-
ness with five business functions following the reference SME credit process
underlying the empirical studies S1 and S2 (cf. Figure 33, p. 161) => |K| = 5.
All modeled banks are assumed to be active in the SME credit business and
currently operating all of the five process steps (fully integrated, cf. section
1.5.4) => a ik 1 i, k .
Next, we introduce a parameter sizei which represents a standardized value
to describe the size of a bank, determining process volumes and process cost
structures. sizei is defined between 0.0 and 1.0 (1.0 represents the maximum
Analytical and Simulative Studies 291

stated number of loans in stock) and follows the “number of loans in stock”
distribution (Figure 38) of the respondents of the S2 study155.
In order to parameterize the process volume of the back-office process steps
processing/servicing (xi3) and risk monitoring (xi4), sizei is multiplied with the
maximum stated number of loans in stock (30,000) and the result is stochasti-
cized by normally distributed noise with a variation coefficient156 of vc =.2: xi3 =
xi4 ~ ND(P = sizei * 30,000, V= .2*P).
For sales/preparation (xi1) and assessment/decision (xi2), assumptions have
to be made either about the acquisition rate or about the average loan contract
duration to estimate the number of new contracts per year. Since no data was
available, we assume the acquisition rate to be 15% of the back-office process
volume (loans in stock) xi3 and xi4 and vary this value during sensitivity analyses.
For the process volume of workout, we use a medium value between the
lower and upper border in Figure 46 on page 167 (i.e. between percentage of
“not good” loans and failed loans) (xi5): For each bank which took part in the
survey, the middle of the corridor between “not good” loans and failed loans is
computed. From these values the mean value (20.17%) and standard deviation
(9.74%) are taken for computing normally distributed process volumes of work-
out157 based on the administration volume xi3: xi5 ~ ND(P = xi3 * .202, V= xi3 *
.097).
The business neighborhood bnij can not be parameterized based on empirical
data. Since the simulations will provide a partial analysis (SME credit business),
we only model banks which are active in this business. Therefore, they all are
competitors from a customer segment perspective. Furthermore, they all offer the
same product (SME loan). Regarding the geographical perspective, we assume
smaller banks serving smaller geographical domains. Thus, there is a smaller
business neighborhood from a geographical perspective between smaller banks.
By contrast, the neighborhood between large and small banks is high because the
large institutes usually serve the same region as the considered small bank.
Finally, some global parameters have to be set:
o The risk-adjusted discount rate is set to rad = .05.

155
Number of SME loans in stock is highly correlated with firm size (in total assets). The S2 sample
further showed to be representative for the largest 1000 German banks in terms of firm size. (Cf.
section 3.6.1.)
156
Variation coefficient (vc) = V/P.
157
The moderate correlation between firm size and fraction of failed loans (cf. footnote 90 on p. 170)
was ignored because the fraction of failed loans already showed a low bandwidth compared with
the computed corridor values.
292 Analytical and Simulative Studies

o The duration of an outsourcing contract TCoSo is varied between 2 and 10

periods, reflecting real outsourcing contracts usually running between 2 and
10 years (Lacity and Willcocks 1998; Quélin and Duhamel 2003, TPI out-
sourcing index158). In the first simulations, the parameter will be set to a me-
dium value of TCoSo=5 periods.
o The strategic constraint SCi (cf. Equation 19, p. 242), which restricts the
outsourcing of business functions representing high strategic value will be
set to SCi = 3 for all i unless otherwise noted. Each bank can outsource
business functions with core competence measures Oik (defined between 0
and 1 and parameterized in the next section) as long as the sum of out-
sourced competencies does not exceed SCi. Moreover, 90% of the banks are
determined ex ante not to outsource sales (Equation 20 (p. 243)), reasoned
by the particularities of this process step and following the empirical results
on the banks’ future business strategy (Figure 75 on p. 201).
o The decision capacity DC (Equation 18, p. 242) will be assumed to be not
restricting.
o The value-added tax rate is set to VAT = .19159 in cases where the simulation
takes VAT into account.
o The legal constraint to not outsource more business functions than the re-
maining ones in terms of “importance” and “size” (BaFin circular 11/2001
par. 17, cf. section 3.5.1.1 and Equation 38 on p. 254) is not taken into ac-
count because this does not make sense if only a part of the banking is cov-
ered by the simulation studies.

5.3.3.2 Business Function Properties and Process Costs

Process cost parameterization is derived in several steps from the empirical
analyses in section 3.6.2.5 (visualized by Figure 109). According to the data, we
first take the empirically gathered distribution of total process costs for a single
loan (cf. Table 20, p. 189). Outliers have been removed following the same rule
as in section 3.6.2.5 (cf. footnote 109 on p. 195). Based on the empirical distribu-
tion function, the total process costs per unit are randomly generated for each
bank.
Second, because the relative allocation of process costs to the five subproc-
esses is distributed almost normally (cf. section 3.6.2.5) and independently of
firm size160 or number of loans in stock161, the computed process cost value is

158
http://www.tpi.net/knowledgecenter/tpiindex/ (as of 15 Aug 2007).
159
VAT rate in Germany since 01 Jan 2007.
160
Only relative cost allocation for risk monitoring/management is moderately correlated with total
assets (Pearson correlation = .227, p<.05).
Analytical and Simulative Studies 293

allocated to the five process steps based on normal distributions following means
and standard deviations of the empirical distributions (distribution parameters
taken from the right columns of Table 19, p. 188).
Third, the derived unit costs for single process steps are split into fixed and
variable parts, based on the analysis in 3.6.2.5. The parameterization has to take
into account that
o fixed/variable cost ratios do not follow a normal (or other standard) distribu-
tion,
o cost ratios are moderately correlated with number of loans in stock, and
o for strengthening reliability, three different cost allocation schemes A, B,
and C have been used to estimate the cost ratios (cf. Table 22 and Table 23
on pp. 195+195).

Random value PC
for each bank i
1 based on empirical distribution
of average process costs

Allocating PC to business functions

PCSales= PC * S PCDecision= PC * S PCProcessing= PC * S PCRiskMon= PC * S PCWorkout= PC * S

2 with S ~ ND(.32, .12) with S ~ ND(.16, .10) with S ~ ND(.29, .13) with S ~ ND(.12, .07) with S ~ ND(.11, .06)

analogous analogous analogous analogous

to to to to
3 3 3 3 3

Separate PCProcessing into Det. fixed/variable Random value sizei

avg. fixed and variable cost ratio (empirical distribution
costs, based on cost ratio based on sizei of loans in stock)
g
essin
4 K
f * x Proc c
avg
„Stochastize“ fixed „Stochastize“ variable
Legend:
5 cost parameter based cost parameter based
S ~ ND(P, V):
on normal distribution on normal distribution
with vc = .1 with vc = .1 S is normally
distributed with
0
expectation P
1

and st. dev. V

vc: variation coefficient
Kf Final parameterization c … = step

Figure 109: Computation of process cost parameters

In order to consider these conditions, all empirically measured cost ratio
values (from A, B, and C) regarding a particular business function are put into a

161
Only relative cost allocation for workout correlates moderately with the number of loans in stock
(Pearson correlation = .289, p<.01).
294 Analytical and Simulative Studies

“lottery wheel”, stored together with the (normalized162) related number of loans
in stock. In order to take the correlation into account, values are clustered into
ten different equally large groups regarding number of loans in stock, separated
by deciles. Now, a concrete cost ratio parameter of a simulated bank’s business
function is determined by choosing a value (with replacement) from the subset
which belongs to the bank’s size. Based on the ratio value and on the unit costs
derived in step 2, variable and fixed unit costs are calculated (step 3 in Figure
109).
The total fixed costs are determined by multiplying average fixed costs per
unit with the corresponding process volume (step 4). Finally, in the fifth step the
resulting cost parameters are “stochasticized” by a normal distribution with a
variation coefficient of .1. The following table compares computed results (1,000
generated random values for each business function) with empirical data (based
on cost ratios). While standard deviation is slightly lower in the simulated data,
mean values match quite well.
Sales/ Assessment/ Processing/ Risk
Workout
preparation decision servicing monitoring
sim. emp. sim. emp. sim. emp. sim. emp. sim. emp.
mean .158 .140 .285 .274 .209 .214 .618 .652 .320 .300
st. dev. .138 .141 .258 .300 .196 .206 .631 .715 .254 .261
median .112 .080 .212 .170 .196 .206 .631 .715 .254 .261
skew 1.16 1.29 1.74 1.76 1.37 1.30 1.61 2.09 1.27 1.36
Table 46: Comparison of statistical properties of simulated and empirical cost
ratio distributions for each business function163
Apart from process costs, further properties of the investigated business
functions are relevant to the parameterization.
First, similarity between business functions of different banks is needed for
computing adoption costs as part of the transaction costs. Section 3.6.2.4 de-
scribes the empirical results of the SME credit process survey S1 regarding the
respondents’ perceived similarity among SME credit processes of different
banks. As already argued in section 3.4.3, the credit business in the past has had
no need to adopt standardized procedures and underlying data standards as for
example in the payments processing domain. Consequently, the survey showed
that the use of standard data formats is rather uncommon in the credit business
(Figure 66, p. 186). Nevertheless, many of the respondents stated that there is the

162
scaled to values between 0.0 and 1.0
163
A more detailed presentation of the empirical data can be found in Table 25 on p. 198.
Analytical and Simulative Studies 295

potential to standardize most parts of the SME credit process (Figure 64, p. 185).
We will use those perceived assessments to parameterize the model’s similarity
measure ]ijk0 for the start configuration (period t=0) of the simulation studies.
Since standardization potential does not describe similarity but only potential
similarity and due to missing process standardization in the industry, the empiri-
cal distributions will be used as trend proxies by normalizing ]ijk0 to values be-
tween 0.0 and only 0.5 (instead of 1.0).
The correlation of standardization potential of workout with firm size (Pear-
son correlation = -.245) will be neglected. For the simulation studies we assume
]ijk0 to be normally distributed164, following means and standard deviations of the
empirical distributions (normalized) (Table 47).

Sales/ Assessment/ Processing/ Risk

Workout
preparation decision servicing monitoring

P]ijk0  .189 .168 .359 .253 .291

V]ijk0  .179 .171 .119 .160 .143

Table 47: Parameterization of ]ijk0 (similarity measure)

Second, task interdependence between two business functions determines
the height of potential synergy loss from outsourcing one of the two. In the em-
pirical studies, a mixed picture occurred at the overall level (Figure 56, p. 179 to
Figure 58, p. 180). In summary, about one-third of the respondents evaluated
task interdependence to be rather low, a further third saw problems with selective
sourcing and the remainder was either indifferent or had no opinion. At business
function level (Figure 59, p. 180), relative differences for different activity pairs
occurred. Highest synergy loss is seen if either sales or assessment/decision is
outsourced. The data presented by Figure 59 are used to parameterize the task
interdependence measure Tikl. The parameter is again assumed to be normally
distributed for the simulation studies and – as defined in chapter 4 – restricted to
values between 0.0 and .99. After downscaling the Likert scale values (between
1 and 5, cf. Figure 59) to this range, the following parameter setting is derived
(Table 48 – P and V of Tikl are given for each activity pair):

164
Empirical distributions are slightly left-skewed for sales/preparation and assessment/decision and
more or less right-skewed for the remaining business functions.
296 Analytical and Simulative Studies

Assessment/ Processing/ Risk

P(Tikl),V(Tikl) Workout
decision servicing monitoring
Sales/
.855, .212 .500*, .280* .630, .276 .435, .283
preparation
Assessment/
.615, .281 .500*, .280* .400*, .280*
decision
Processing/
.513, .281 .100*, .280*
servicing
Risk
.485, .286
monitoring
* = These values have not been requested in the survey. Means are estimated by case study
interviewees and have to be varied during sensitivity analyses. Standard deviation is deter-
mined according to the deviation of the survey-based parameters.
Table 48: Parameterization of Tikl (task interdependence measure)
Third, in order to parameterize the core competence measure Oik, the respon-
dents were asked whether the different steps of the credit process represent a
core competence of their bank (cf. Figure 52). Again, the data has been normal-
ized to values between 0.0 and 1.0 but because of stronger skewness it cannot be
assumed to be normally distributed. Instead, a constant interpolation of the em-
pirical distribution (Bratley et al. 1987) is used to generate random numbers165
and normalized to the range defined for Oik. In addition, because answers regard-
ing core competence were significantly correlated among each other, the follow-
ing correlations have been considered when computing the parameterization166
(Figure 140).
Correlation between Pearson correlation
Oi1 (Sales/preparation) Ok2 (Assessment/decision) .415 (p<.01)
Oi2 (Assessment/decision) Oi3 (Processing/servicing) .305 (p<.01)
Oi2 (Assessment/decision) Oi4 (Risk monitoring) .428 (p<.01)
Oi3 (Processing/servicing) Ok5 (Workout) .477 (p<.01)
Table 49: Correlations between Oik and Oil

165
In order to receive continuous values from the discrete 5-Likert scale indicators, these are sto-
chasticized by applying a uniform distribution: If the generated random value is 3, then the result
is equally distributed between 2.5 and 3.5, implicitly generating a cascade distribution function.
166
Consideration of correlations has been realized as follows: With a probability p = r(Oik,Oil), Oik is
set equal to Oil or vice versa.
Analytical and Simulative Studies 297

5.3.3.3 Transaction Costs

Similarity measures, task interdependence measures, and core competency pa-
rameters represent business function properties which flow into the bundle of
transaction cost functions. These further require cost factors to monetarize the
effects of process similarity or task interdependence, for example, and to relate
them to process cost savings in the decision function. While cIF (interface cost
factor) is globally defined, c kAD (adoption cost factor), c kN (negotiation/coordina-
tion cost factor), and c kAG (agency cost factor) are defined for each type of k.
Transaction costs can hardly be estimated based on empirical data; therefore
the parameters will be strongly varied during simulation studies for analyzing the
sensitivity of the simulation results of transaction costs. The parameterization
given here will only be an initial setting. Variations of the parameters during the
simulation are described in the results section when applied.
As a starting point, Barthélemy (2001) will be consulted, who found nego-
tiation costs (CN) of IT outsourcing deals to be (on average) 3% of the contract
volume CG167 while periodical costs for coordination (CC) and control (CAG) to-
gether represent another 8% (cf. section 2.1.2.2). Since CC was defined as J CN
(with 0< J<1), this leads to the following (average) relation:
CikG ˜ .03 CikN
CikG ˜ .08 CikC  CikAG with CikC JCikN
Ÿ 2.67 CikN CikC  CikAG Ÿ 2.67 CikN J CikN  CikAG Ÿ CikAG 2.67  J CikN
Equation 55: Relation between different transaction costs, based on (Barthélemy
2001) (simplified indices)
Since the contract volume CikG is not determined ex ante, the variable costs
of business function k are assumed as its lower border (because major parts of
the fixed costs are intended to be saved by joining the coalition). Thus, following
Barthélemy (2001), c kN is determined as 3% of the average (i.e. over all actors)
variable costs for each business function k in a first step.
The resulting parameters are given in Table 51. In order to complete the
parameterization of the negotiation and coordination cost functions, D, E, andJ
are set to the values given in Table 50:

167
CG = the costs periodically paid for the service, cf. section 4.2.3.2.
298 Analytical and Simulative Studies

D E J
1.25 -.9 .3
Table 50: Parameterization of negotiation and coordination cost functions168
If the ratio parameter between coordination costs C ikC and negotiation costs
CikN is set to J =.3, agency costs CikAG represent 237% of the negotiation costs
AG
(cf. Equation 55). Due to the different function types, the cost factor c k has to
N
be set much higher than c . The agency cost function includes three parameters
k
and variables defined to values between 0.0 and 1.0 which are multiplicatively
interrelated – while the negotiation cost function has only one. If all of these
were symmetrically distributed (average = .5), then c kAG would have to be set to
4* c kN *2.37 to lead to agency costs being 237% as high as negotiation costs169.
Nevertheless, this does not take into account the different risk levels of out-
sourcing different business functions of the credit business, which are relevant to
agency costs. Thus, for agency cost factor c kAG we chose the empirical results
regarding the question on how high the strategic risk from outsourcing a particu-
lar process step to another bank would be (Figure 84, p. 208). The means of the
answers are transformed to the values given in Table 51.
N
Thus, we first add c k over all k and multiply the result with 4*2.37. Then,
the product is split into the five business functions, following the ratios between
means of the answers in Figure 84. The resulting values are given in Table 51.
Transaction Sales/ Assessment/ Processing Risk
Workout
cost factor preparation decision /servicing monitoring
N N N N N
c k c1
N
6,740 c 2 3,506 c 3 40,773 c 4 17,034 c 5 3,095
AG
c k c1
AG
178,996 c AG
2 173,984 c AG
3 98,090 c AG
4 135,142 c AG
5 88,782
AD AD AD AD AD
c k c 1
AD
10,000 c 10,000 c
2
10,000 c
3 10,000 c4 10,000 5

Table 51: Parameterization of transaction cost factors [monetary units]

168
To remind the reader: D is the progressive effect of the coalition size on negotiation and coordi-
nation costs) while E represents the declining effect of learning effects.J describes the ratio be-
tween negotiation and coordination costs (cf. Equation 6 and Equation 7).
169
One stochastic parameter with expectation = .5 would be 4 times higher than the expected product
of three parameters having the same expectation (.53=.125).
Analytical and Simulative Studies 299

No empirical data was available for determining adoption cost factor c kAD ;
even Barthélemy was not able to gather adoption (or “transition”) costs in his
ITO survey. Differences in the process design as a driver of adoption costs have
already been considered in the similarity measure ]ijk, parameterized above.
Moreover, task complexity F ik has been already captured by the function itself.
Therefore, c kAD is initially set to 10,000 for all business functions and varied
during sensitivity analyses.
Finally, the interface cost factor also is set independently of (non-available)
empirical data. Interface costs do not occur for each business function, but for
each pair of business functions (if one of both is outsourced); thus, the initial
parameter is set to only cIF=5,000.
5.3.3.4 Simulation and Optimization Control
Apart from the model parameterization, the simulation control as well as the
genetic algorithm configuration have to be parameterized. The simulation is
started by trying to form coalitions in the first period which then are tested on
stability (cf. section 5.3.2). This coalition setting is repeated after a certain num-
ber of periods. The parameterization, which showed to be useful during test runs,
is as follows:
initCoalSize 9 Initial size of created coalitions
Repeat rate (number of periods) for crea-
coalBuildingFreq 5
tion of new coalitions
Maximum number of coalitions created per
maxCoalBuilding 4
business function k in one period
Table 52: Parameterization of simulation control
The following configuration was chosen for the genetic algorithm (cf. sec-
tion 5.2.3):
50 Population size
pCo = .55 Crossover probability
pM = .001 Mutation probability
TGA = 100 Maximum number of generations
Pre-termination after 50 periods, if best found solution did not change
Table 53: Parameterization of the genetic algorithm (cf. section 5.2.3)
300 Analytical and Simulative Studies

Appendix A2 provides tables which summarize the whole parameterization

of the model and of the simulation routines which have been used in the follow-
ing simulation studies.

5.3.4 Simulation Results

This section presents the results from the simulation analyses and follows the
structure displayed in Figure 110.
1. Basic Setting
Analyzing the basic setting
in view of the following three dimensions:
• Monetary effects
• Market structure effects
• Decision behavior

2. Variation of Process Cost Structure 3. Variation of Transaction Costs

What is the impact of the following variations How do the transaction costs affect the results
on the three dimensions? in view of the three dimensions?
• Variation of coalition process cost structure • Variation of transaction cost parameters
• Variation of the ratio between fixed/variable costs • Variation of further parameters in the TC functions
• Homogenization of the fixed/variable costs ratio • Impact of task complexity

Dimensions
of
4. Variation of Further Decision Calculus Parameters
analysis
What is the impact of further decision-relevant factors
In all steps, the on the three dimensions?
analysis is in view
• Level of risk-adjusted calculation discount rate
of the following • Value-added tax
dimensions: • Strategic constraint
• Monetary effects • Heterogeneity of the decision-relevant parameters
• Market structure
effects
• Decision behavior

5. Impact of Demographic Properties 6. Impact of Simulation Control Param.

How do the results react to variations of : What is the impact of varying:
• System size (number of firms) • Initial coalition size
• Business neighborhood • Maximum number of initially formed coalitions
• Heterogeneity of firm sizes • Choice of coalition selection mechanism

7. Reasons for Inefficiency

How can the different reasons for inefficiency
be summarized?
• Start-up problem
• Individual lock-in
• System-wide lock-in

Figure 110: Overview of the simulation studies (section 5.3.4)

Section 5.3.4.1 starts with an in-depth analysis of the basic parameterization
along the following dimensions:
Analytical and Simulative Studies 301

1. The degree of cooperative sourcing and the related monetary results such as
process cost savings (PCS), transaction costs (TC), and net savings (NS) are
analyzed from a global perspective and from an individual perspective.
2. In addition, the resulting market structure effects in terms of market concen-
tration, number of coalitions, and resulting business models are investigated.
3. The third dimension focuses on the dynamics leading to those results. How
often do banks outsource, switch coalitions, or backsource the business
function? The ratio between switching, backsourcing, and outsourcing is
taken as a proxy for behavioral uncertainty from the banks’ perspective be-
cause it reflects the volatility of the decision-relevant environment (see fur-
ther explanations below).
Finally, the simulation results are compared with the results from centrally
optimizing the cooperative sourcing configuration of the system by applying the
genetic algorithm.
In sections 5.3.4.2 and 5.3.4.3, the process cost structure and the level of
transaction costs are varied in order to investigate the effects of different cost
situations on the three dimensions. Section 5.3.4.4 completes the picture of deci-
sion-relevant factors by analyzing the impact of calculation discount rate, VAT,
the strategic constraint, and of general parameter heterogeneity.
In a next step, the stability of the found results is tested against variations of
the basic demographic properties (number of simulated firms, degree of business
neighborhood, heterogeneity of firm sizes) (section 5.3.4.5) and of artificial
parameters to control the simulation procedure (initCoalSize, maxCoalBuilding,
coalition building mechanism) (section 5.3.4.6). Finally, section 5.3.4.7 summa-
rizes and complements the analysis with a particular focus on sources of ineffi-
ciency. Each section concludes with a short summary.

5.3.4.1 Results from the Basic Setting

This section describes the simulation results from instantiating the cooperative
sourcing model with the parameterization derived from the German SME credit
business (sections 3.6 and 5.3.3). The results in this section are derived from
1,000 simulation runs with identical parameterization (unless otherwise noted).
The aim of this section is to analyze the system behavior and the resulting global
monetary effects and market structures. In the subsequent sections, the parame-
terization will be varied (ceteris paribus analyses) to test the impact of the vari-
ous model elements on the results found in this section.
The first figure shows the global monetary effect of cooperative sourcing ac-
tivities over time. The left axis shows the cumulated absolute effect, while the
right axis shows the same value related to the original process costs. Further-
 302 Analytical and Simulative Studies

more, the dashed graph gives the relative average individual net savings (only
right ordinate).
60%
120,000,000

rel. global PCS, NS, TCp and

50%
abs. global PCS, NS, TCp

100,000,000

rel. individual NS
[monetary units]

40%
80,000,000

30%
60,000,000

40,000,000 20%

20,000,000 10%

0 0%
0 50 100 period 150 200 250
global PCS global NS rel. ind. NS global TCp
Figure 111: Periodical average process cost savings (PCS), transaction costs
(TCp), and net savings (NS) over time
In total, the 100 modeled banks save 55% or about €117 million of the
original process costs per period (periodical net savings (NS) = periodical proc-
ess cost savings (PCS) minus periodical transaction costs (TCp), which consist
of coordination costs, agency costs, and interface costs170) by cooperative sourc-
ing. The resulting transaction costs are around 6% of the process cost savings
(PCS). This is almost analogous to empirical results of IT outsourcing research
in the banking industry (Ang and Straub 1998).
Although savings of 55% of the overall process costs seem to be a very high
figure, one should be aware that this represents the aggregate savings, not the
average savings per actor. Due to the high heterogeneity in stated process costs
(cf. section 3.6.2.1), there are banks with very inefficient processes which can
save much money through cooperative sourcing in the simulations171. This gen-
erates the high aggregate (relative) savings of the whole system. A breakdown of
the savings to the individual banks results in average individual savings of 34.4%
after 250 periods (dashed line in the figure above).

170
Equation 11 and Equation 14 on pp. 238-239. We chose to present the results on a periodical
basis instead of showing the net present values of particular sourcing contracts (Equation 15 on p.
241) because setting the sourcing contract duration as an extra parameter would make a compara-
tive analysis much more complex for the reader. Thus, the one-time transaction costs for negotia-
tion and adoption are not included in the TC displayed in the diagrams.
171
In reality, banks will probably first optimize their processes internally before outsourcing them
(although this has not necessarily been the case in the past). Obviously, inhouse optimization re-
duces the relative advantageousness of cooperative sourcing.
 Analytical and Simulative Studies 303

Although the graphs in Figure 111 show no major increases after the first
100 periods, the simulations still show some dynamics over a long timeframe.
Due to the many complexities that the banks are faced with, caused by external-
ities and also by simulation design, the banks need quite a long time to find the
optimal coalition, from their perspective, which is stable and offers the expected
savings. Thus, the concept of periods is not directly transferable into reality.
Figure 113 shows the distribution of the number of periods needed to reach the
stationary state which occurs after 173 periods on average. Around 7% of the
simulation runs did not converge into a stationary state at all172. The number of
periods needed does not correlate with global monetary results (such as PCS and
NS).
100%

80%
relative frequency

60%

40%

20%
(mean = 173 periods, 500 simulation runs)

0%
0 100 200 300 400 500 600 700
periods
Figure 112: Number of periods needed to reach stationary state
Naturally, results from simulation studies show a certain bandwidth of varia-
tion. The following figure shows the range of results for total net savings (NS) in
every period. The standard deviation (ranges around the mean) as well as the
minimally and maximally obtained values resulting from 1,000 simulation runs
are depicted in the diagram.
Although there are quite large deviations from the mean, fitting tests showed
that the results are almost lognormally distributed. Continuously, over time,
about 73% of the simulation runs stay within the inner displayed range [mean–
sd, mean+sd]. The deviations are strongly influenced by high variations in the
(empirically conducted) process cost parameters. When the variation of the rela-
tive savings (NS/original process costs) is measured instead, the final resulting

172
Of the 500 simulation runs, 93% of the runs were computed to be 664 periods in maximum
duration. The remaining 7% of the simulation runs did not converge, even after 2,000 periods, be-
cause the system was locked in a dynamic equilibrium: due to a certain constellation of expected
NPVs, a small number of banks cyclically outsource and backsource the same business function.
 304 Analytical and Simulative Studies

relative savings of 55% (cf. Figure 111, right scale) show a quite low standard
variation of only 6.1 percentage points.
300,000,000

250,000,000
max(NS)
net savings (NS) [monetary units]

mean(NS)+sd(NS)
200,000,000 mean(NS)
mean(sd)-sd(NS)
min(NS)
150,000,000

100,000,000

50,000,000

0
0 25 50 75 100 125 150 175 200 225 250
period
Figure 113: Global net savings (NS): mean and spread (standard deviation and
min/max values) from 100 simulation runs
In order to look at the cost savings in greater detail, we analyzed the differ-
ent business functions’ contribution to overall net benefits. A breakdown of the
average individual savings of 34.4% into the different business functions results
in the average net savings given in the following table. Since interface costs,
which are part of the transaction costs and thus reduce the net savings, cannot be
dedicated to a single business function, they are omitted in this analysis. The
highest savings are generated from cooperatively sourcing processing/servicing
and risk monitoring. Workout offers some savings potential, too, while the front
and middle office functions show no significant cost savings.
Business Sales/ Assessment/ Processing/ Risk
Workout
function preparation decision servicing monitoring
Average net
0.1% 2.2% 41.1% 35.3% 16.0%
savings per bank
Table 54: Average net savings at business function level
(without considering interface costs)
Because there is a strong asymmetry in terms of bank sizes and thus process
volumes, the next figures show the results in more detail by grouping the banks
into ten equally large clusters ordered by firm size. The larger the group index is,
the larger the size of the associated banks. Figure 114 shows the absolute savings
broken down into business functions and bank size groups. The box plots repre-
Analytical and Simulative Studies 305

sent the quartiles (boxes) and total ranges (antennas) of the absolute net savings
for each business function and bank size group.
20000000
Avg. absolute net savings per business function

15000000

10000000

5000000

0
10
9
8
7
6
firm size group 5 5
4 4
3 3
2 2
1 1 business function
Figure 114: Distribution of absolute net savings per business function (box plot
representation without outliers and extreme values173)
Obviously, the main contribution, by far, to overall net benefits is given by
the processing/servicing business function. The larger the firm is, the higher the
net savings are. Nevertheless, processing/servicing also promises the highest
savings for the smallest banks and is often the first or only one which is out-
sourced by those, in particular. Figure 115 shows the relative composition of the
overall net benefits174.

173
Outliers and extreme values, here and in all of the following box plot charts, represent values
which fall short of or exceed the .25 or the .75 quartile by more than 2 or 3 times the inter-quartile
distance.
174
Once again, the interface costs CIF are disregarded, because they cannot be clearly assigned to a
particular business function (cf. section 4.2.3.1.4).
 306 Analytical and Simulative Studies
Avg. contribution to total net savings

100%

80%
Workout
of the bank

60%
Risk monitoring
Processing/servicing
40% Assessm./decision
Sales/preparation
20%

0%
1 2 3 4 5 6 7 8 9 10
bank size group
Figure 115: Average contribution of outsourcing a particular business function
to total net savings (per bank), related to bank size
For the smallest banks (small bank size group indices), the relative contribu-
tion provided by savings from cooperatively sourcing the processing/servicing
function is highest. Since many small banks outsource only this process step, in
these cases it accounts for 100% of the net savings. Apart from process-
ing/servicing, the only business function which also significantly contributes to a
positive net benefit is risk monitoring while the remaining three business func-
tions play only a secondary role.
What are the resulting “market” effects from cooperatively sourcing the dif-
ferent business functions, now? Figure 116 shows the increase of market concen-
tration for the different process parts. The left diagram represents the actor-based
Herfindahl index (HFact)175 and measures the size of coalitions in terms of mem-
ber counts. The right diagram demonstrates the aggregate process volume of the
different coalitions (HFvol). Due to the asymmetric distribution of process vol-
umes throughout the system of banks, the concentration values are higher.

175
The Herfindahl index measures the market concentration by summing the squared market shares.
In this work we use the relative size of a coalition – either in terms of number of actors or process
volume – as market share:
2 2
mktmax
2 § ¦ xik
mktmax
· § § · ·
§ M kmt · M 0 ¨ iM kmt ¸ |I | ¨
¨x ¸ ¸
HFktact ¦ ¨ ¸  km HFktvol ¦ ¨ ¸  ¦ ¨ z ik 0t ¨ ik ¸
¸ . Single banks
¨ I ¸ I2 ¦ x ¦ x ¸
m 1© ¹ m 1¨ ik ¸ i 1 ¨ ¨ ik ¸
© iI ¹ © © iI ¹ ¹
are counted as solitary “market members” (represented by the second term in each of the equa-
tions, with M km0 being the number of actors not being involved in cooperative sourcing regard-
ing k in t.
Analytical and Simulative Studies 307

Figure 116: Average “market concentration” for the provision of the different
business functions, based on actor-based (left) and process volume-
based (right) Herfindahl index
Both diagrams show moderate to strong consolidation tendencies for the dif-
ferent back-office functions (processing/servicing, risk monitoring, workout),
while there are no or almost no activities for the first two business functions.
HFact = .2 for processing/servicing means that if all coalitions had an equal size
and all banks were members of one of them, the 100 banks would organize
themselves in 5 coalitions. From a process-volume perspective, HFvol = .5 means
that if the coalitions had the same process volumes, there would be only 2 coali-
tions left in the market, operating equally large process volumes. Interestingly,
the only graph showing significant increases over the whole observed time frame
of 250 periods is the actor-based concentration measure for processing/servicing.
Although the volume-based Herfindahl index reached a stable level after 150
periods, there appear to be small banks which still enter into coalitions or switch
to larger coalitions. Their small process volumes have no significant impact on
both global monetary savings (Figure 111) and volume-based market concentra-
tion (Figure 116, right).
Due to asymmetries in the banks’ process volumes (huge differences be-
tween a few large and many small banks), additional measures were incorporated
in order to obtain a more comprehensive picture of the resulting market structure.
Figure 117 shows the average number of banks that are members of any coalition
(left) and the average number of coalitions in the system (right).
 308 Analytical and Simulative Studies
avg. num ber of banks in coalitions

80 8

avg. number of coalitions

70 7

60 6

50 5

40 4

30 3

20 2

10 1

0 0
0 50 100 150 200 250 0 50 100 150 200 250
period period
Sales/Prep. Assessm./Decision Processing/Servicing Risk monitoring Workout

Figure 117: Avg. number of actors in coalitions and number of coalitions over
time
In the left diagram, it can be seen that, in the long term, the majority of
banks (74%) join a processing/servicing coalition, followed by risk monitoring
(61%). Furthermore, regarding processing/servicing, the number of coalitions
increases greatly in early periods and decreases slightly afterwards. Thus, we can
argue that, in contrast to risk monitoring there is an over-reaction in the first
periods. Joining any coalition seems to be beneficial from the perspective of
banks and thus many small coalitions are established in early periods, which are
not efficient in comparison with later consolidation. Together with a slightly
increasing number of cooperatively sourcing banks (Figure 117 – left) the con-
solidation separates into fewer but larger coalitions in later periods, leading to
the long-term increase of market concentration (Figure 116 – left) for this par-
ticular business function.
This effect cannot be observed for the remaining business functions. Never-
theless, there are also strong cooperative sourcing activities for risk monitoring.
In regards to workout, there are some consolidation activities (on average only
1.4 coalitions exist in the long term), but the majority of banks decide not to take
part in any coalition. Because these banks are also considered in the Herfindahl
index computation, the concentration measures remain at a relatively low level.
In order to receive greater insight into the heterogeneity in coalition size,
Figure 118 shows the common concentration measures CR1 (left) and CR3 (right)
which, in this work, represent the “market share” of the largest coalition and the
three largest coalitions in terms of number of members (cf. Hannan and McDow-
ell 1984).
Analytical and Simulative Studies 309

Figure 118: Average “market share” of the largest coalition (CR1) resp. the three
largest coalitions (CR3) in terms of number of members176
In correlation with the highest level of cooperative sourcing activities, the
largest coalition can be found for processing/servicing, followed by risk monitor-
ing. When comparing both diagrams, the second and third largest coalitions
already prove to be much smaller for all business functions. If we compare
Figure 118 (right) with Figure 117 (left), it becomes clear that the remaining
coalitions are very small. For processing/servicing as well as for risk monitoring
it can be seen that only 11-14 coalition members are not members of one of the
three largest coalitions. If there are, on average, 6-7 coalitions in the long term
(Figure 117, right), these 11-14 coalition members separate into 3-4 coalitions.
Range of services offered
Business model Outsourcing
to other banks
1 – (Traditional) fully
None None
integrated bank
2 – Fully integrated bank
Every business function None
with service provision
Every business function One back-office function (either process-
3 – Selective outsourcing
which is not outsourced ing/servicing, risk monitoring, or workout)
Every business function
4 – Major outsourcing Two of three back-office functions
which is not outsourced
Sales/preparation and All back-office functions (process-
5 – Sales bank
assessment/decision ing/servicing, risk monitoring, and workout)
6 – Pure sales bank Sales/preparation Everything except sales/preparation
7 – Processing service
Back-office functions Sales and assessment/decision
provider
Table 55: Business models, tracked in the simulation studies (cf. Figure 80)

176
2 indicates that 20% of all actors are members of the largest (left) resp. of the three largest (right)
coalitions.
310 Analytical and Simulative Studies

In the empirical survey (section 3.6.3.1), we analyzed the potential fre-

quency of new business models in terms of sourcing configurations within the
SME credit business. Figure 119 shows the corresponding results of the simula-
tion studies. The used classification of business models was already derived in
section 3.6.3.1 (Figure 80 on p. 206):
Figure 119 provides both the average distribution (left chart) and the range
of relative frequencies within 100 simulation runs in the form of box plots. Per-
fectly in line with the empirical results, around one-fifth of the banks do not take
part in any coalition and remain on the sourcing configuration of the traditional
universal bank. At least an additional five percent decide to offer their services to
other banks and to become the insourcer of a cooperative sourcing coalition. The
most common sourcing strategy is to outsource two business functions of the
back office (i.e., processing/servicing, risk monitoring, and workout) while se-
lectively outsourcing only one business function or outsourcing the whole back
office (to become a sales bank) are also common strategies. The pure sales bank,
which also outsources the assessment/decision step, can not be found. Further-
more, the pure processing service provider is not a valid option, either, and can
only be found very seldom177.
60
traditional fully
frequency [in %]

integrated bank 50
15% 21%
fully integrated bank 40
w ith service provision
30
5% selective outsourcer
20
major outsourcer 10
35%
24%
0
sales bank
traditional fully selective major sales bank pure sales processing
fully integrated outsourcer outsourcer bank service
integrated bank w ith provider
bank service
provision
average distribution business model
Figure 119: Frequency of business models (left: average distribution, right: box
plot representation), based on 100 simulation runs
In order to look more closely at the underlying dynamics (i.e. third dimen-
sion of analysis), we tracked the actors’ decisions over time. During the simula-
tion, banks have to evaluate (new) and to re-evaluate (existing) cooperative
sourcing coalition memberships. Their evaluations can result in outsourcing178,

177
It should be once more noted that this does not imply that a bank is unwilling to establish a credit
factory. However, in this analysis, that would mean offering a bank’s own services to other banks
(if the credit factory is created to serve third parties). In this analysis, we examine existing banks
(or their virtual agent representatives) to investigate whether each bank (and not the credit fac-
tory) decides not to do anymore (e.g.) sales activities.
178
The term outsourcing is used here and in the following to describe that a bank enters a coopera-
tive sourcing coalition while its business function has previously run internally. Entering a coali-
 Analytical and Simulative Studies 311

switching of coalitions, and backsourcing. The average frequency of these ac-

tions is depicted over time in Figure 120 (over all actors and business functions).
The oscillations result from the sourcing contract duration179.
35

30

25

20
frequency

15 outsourcing
switching
backsourding
10

0
0 20 40 60 80 100 120 140
period

Figure 120: Average frequency of outsourcing, switching, and backsourcing

actions over time (counted over all actors and business functions)
One can see that the outsourcing activities are strongest in the very begin-
ning and are reduced to very low frequencies after some periods of time. Obvi-
ously, the monetary results and market concentration over time (shown in the
figures above) are primarily affected by outsourcing in the first 20 periods while
later increases are mainly induced by switching activities because banks that
have already outsourced business functions try to further improve their cost
situation.
While outsourcing activities correspond positively with firm size (i.e. banks
with larger credit process volumes outsource their processes more frequently)180,
the frequency of switching and backsourcing actions is completely independent
from firm size.

tion, however, can also imply that the bank will become the insourcer of the coalition’s process
volume.
179
After closing a sourcing contract, the parties must not leave the coalition for TCoSo = 5 periods (cf.
section 5.3.3.1).
180
Spearman correlation = .326 (p<.01), Pearson correlation = .171 (p<.01).
 312 Analytical and Simulative Studies

Although the average frequency of switching and backsourcing displayed in

Figure 120 is low, much higher values can be reached in some simulation runs.
Coalition switching can occur up to 40 times per period while backsourcing
happens in a maximum of about 10 cases (over all 100 actors and 5 business
functions). The almost constant frequency of switching actions in the long term
shows once again that the system needs a long time to terminate in a stationary
state (cf. Figure 112). However, actions in late periods have almost no more
impact on the global monetary measures (cf. Figure 111). Nevertheless, those
few banks still try to improve their individual benefit situation.
While the outsourcing activities for each business function can be read from
the diagrams above (Figure 117, in particular), the following charts show the
switching and backsourcing dynamics for the back-office business functions in
detail181. The diagrams show how frequently which proportion of banks switches
coalitions or even backsources it within the first 150 periods.
20 12
18
10
rel. number of actors [in %]

16 Backsourcing processing/servicing
rel. number of actors [in %]

Switching processing/servicing
14 Switching risk monitoring Backsourcing risk monitoring
8 Backsourcing workout
Switching workout
12

10 6
8
4
6

4
2
2

0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6
number of switching actions number of backsourcing actions

Figure 121: Number of switching and backsourcing actions regarding back-

office functions (within the first 150 periods)
Processing/servicing and risk monitoring show a quite similar frequency of
switching and backsourcing activities. About 10% of the simulated banks back-
source their back-office functions once and about 2% backsource them twice
(after outsourcing them again). The frequency of switching activities is higher;
about one-sixth of the banks changes the coalitions once and around 7% do it
twice during the simulation runtime of 150 periods. Backsourcing and switching
activities of the workout business function show much lower activities, which
relates to the lower frequency of outsourcing activities displayed in Figure 117.
The next figure presents the preceding step of decision making. The graphs
depict the cumulated number of positive and negative evaluations of new coali-

181
Since the outsourcing activities for sales/preparation and assessment/decision are already mini-
mal, we have disregarded the analysis of switching and backsourcing activities for those business
functions.
 Analytical and Simulative Studies 313

tion memberships as well as re-evaluations of existing memberships resulting

from the same simulation runs as the previous figure. Evaluations will only take
place if outsourcing of the specific business function is not explicitly restricted
(cf. Equation 16 on p. 241) and as long as the bounded decision capacity (BDC)
is not exhausted (cf. Equation 18 on p. 242).
180

160

140

120
frequency

100

positive evaluations
80
negative evaluations
positive re-evaluations
60
negative re-evaluations

40

20

0
0 20 40 60 80 100 120 140
period
Figure 122: Evaluations and re-evaluations over time (counted over all actors
and business functions; average from 100 simulation runs)
The diagram shows that the number of evaluations and re-evaluations with
positive results decreases over time while negative evaluation results demon-
strate increasing tendencies. Thus, the market converges into a stable structure.
The high number of negative evaluations results from the fact that a bank can
evaluate more than one potential coalition membership during a single period.
The strong increase in negative results of examining new evaluations during very
early periods indicates that the actors are very quickly locked into their coalition,
i.e. switching the coalition would often be unfavorable because the additional
savings are expected to be too low to exceed the transaction costs for switching.
The following diagram breaks down the results (cumulated over 150 peri-
ods) into the different business functions and shows the absolute and relative
average frequencies of positive and negative evaluations. The left side depicts
the average absolute number of evaluations per period, while the right side
shows the same results in form of relative frequencies.
 314 Analytical and Simulative Studies

40 0.6% 0.4% 3.5%

absolute frequency per period

1.24 100%
35 12.4% 11.0%

30 3.92 80% 38.0%

relative frequency
25 0.10
10.36 60%
20 99.4% 99.6% 96.5%
34.40 87.6% 89.0%
15 27.82 40%
24.25 62.0%
10 16.88 20%
5 0.01
0 0%
1.23

l
n

ut
t

l
g
n

g
g

ra
u

on
on

io

in
io

in

rin
rin

ko
ko

ve
t
at

vic
ic

isi
ra
isi

to
it o

or
or

O
rv
ar

ec
pa
ec

er

W
on
W
on
se
ep

s
./d
e
./ d

m
g/
m

pr
g/
Pr

in
m

s/

sk
in

k
s/

ss

ss
ss

le
ss

Ri
le

Ri

se
Sa

e
se

e
Sa

oc
positive evaluation results

As
oc

positive evaluation results per period

As

Pr
Pr

business function negative evaluation results per period business function negative evaluation results

Figure 123: Average frequency of positive and negative evaluation results

(over all actors; average from 100 simulation runs)
The most frequent evaluation activities can be observed for workout, fol-
lowed by risk monitoring/management, processing/servicing, and assess-
ment/decision. In contrast, the absolute number of evaluations for
sales/preparation is naturally low because 90% of the banks decided ex ante to
not allow outsourcing of this business function for strategic reasons. The left
diagram shows that, on average, in each period between 24 and 36 actors (de-
pending on the particular business function) evaluate a new coalition member-
ship in each period. Since most of them result in negative decisions, they repeat-
edly consider cooperative sourcing with another coalition in the subsequent peri-
ods.
With regard to the amount of positive evaluations, the picture partly turns
around. Processing/servicing shows the highest frequency of positive evalua-
tions, followed by risk monitoring and workout. This can be explained by the
fact that positive evaluations will often lead to outsourcing (see below), which
makes further evaluations obsolete. Nevertheless, this does not explain another
discrepancy in the left chart. Both assessment/decision and workout show a very
low degree of positive evaluations but differ significantly in the overall number
of evaluations. This difference reveals that, in contrast to credit application as-
sessment and decision, many banks want to basically get rid of the workout func-
tion but often do not find a “positive business case”.
However, positive evaluations do not necessarily lead to cooperative sourc-
ing. This would only be possible if the coalition which was evaluated by the
potential new member also agrees to incorporate the new bank. This is not neces-
sarily the case, since the existing members may fear negative consequences and
therefore vote against the new member. Figure 124 shows that the number of
positive and negative votes does not change significantly over time.
Analytical and Simulative Studies 315

100%
frequency

50%

negative votes
positive votes
0%
1 21 41 61 81 101 121 141
period
Figure 124: Frequency of positive vs. negative votes over time
When focusing on the different business functions some differences are dis-
closed (no figure): while the results are rather balanced for risk monitoring (on
average 58% positive votes), assessment/decision (50%), and workout (39%),
sales/preparation and processing/servicing show diverging extreme situations:
In the first, only about 2% of the membership requests are accepted while in the
latter, almost all (98%) of the voting results are positive (no figure).
Interestingly, there is no structural change of the picture when we alter the
voting mechanism. While a unanimity principle was used for this analysis (all
members have to agree to the new member), we also tested a two-third majority.
The almost perfect similarity of the results shows that the members of a coalition
have rather homogeneous preferences regarding the entrance of a new member.
As aggregation of the evaluation results above, Figure 125 shows how many
evaluations are made by how many actors (given as relative frequency) during
the first 150 periods (cumulated over all 5 business functions). Once again, the
figure distinguishes between evaluations (of new membership) (upper diagrams)
and re-evaluations (of existing memberships) (lower) and between positive (left)
and negative (right) results.
 316 Analytical and Simulative Studies

20%

12%

15%

8%

frequency
frequency

10%

4%
5%

0% 0%
0 50 100 150 200 0 200 400 600
number of positive evaluations number of negative evaluations
per actor over all business functions and 150 periods per actor over all business functions and 150 periods

20%

30%

15%
frequency
fequency

20%
10%

10%
5%

0% 0%
25 50 75 100 125 0 25 50 75
number of positive re-evaluations number of negative re-evaluations
per actor over all business functions and 150 periods per actor over all business functions and 150 periods

Figure 125: Number and results of evaluations and re-evaluations per actor over
the first 150 periods (cumulated over all business functions)
When comparing the diagrams, the number of positive evaluations and nega-
tive re-evaluations per actor proves to be much lower than in the corresponding
opposite diagram. This corresponds to the results of the activity analysis above
(Figure 120). Outsourcing activities are not very common (low frequency of
positive evaluations), but switching or backsourcing is even less common, dem-
onstrating satisfaction with the cooperative sourcing coalition (low frequency of
negative re-evaluations). Additionally, it has been found that the frequency of
Analytical and Simulative Studies 317

negative evaluations corresponds negatively to firm size: The larger the firm is,
the less it will decline a coalition membership182.
There can be different reasons for negative evaluations of potential coalition
memberships. First, outsourcing can be hindered by the strategic constraint,
which restricts the number of tasks being outsourced (cf. section 4.2.3.4,
Equation 19)183. On average, this factor is responsible for negative evaluation
results in only 7.6% of all negative cases (Figure 126)184. Second, the most
common case is a negative net present value estimation of the cooperative sourc-
ing option (92.4% on average)185.
4.3% 3.9% 8.3%
100% 7.6%
90% 18.5% 15.2%
80%
relative frequency

70%
60%
50% 95.7% 96.1% 92.4%
84.8% 91.7%
40% 81.5%
30%
20%
10%
0%
g

ut
n

ng

ll
on

ri n

ra
io

ko
ci
at

isi

ve
ito

or
vi
ar

ec

O
on
er

W
ep

./d

/s

m
pr

ng
m

k
s/

ss

is
si

business function
le

R
es
se
Sa

oc
As

binding strategic constraint

Pr

Exp(NPV) < 0

Figure 126: Reasons for negative evaluation results

Now the question is which part of the estimated net present value function
(Equation 30, p. 249) is primarily responsible for the negative result. The analy-
sis first examines how often a negative NPV results from forming expectations.
If the net present value which is computed based on the assumption of a static
coalition structure (Equation 15, p. 241; referred to as static NPV in the follow-
ing) and the final expected NPV (Equation 31, p. 250) are compared, the fraction
of negative expectations where the static NPV is positive represents the effect of
uncertainty about the coalition partners’ future actions and thus the impact of

182
Pearson correlation = -.346 (p<.01).
183
Note that the other constraint which restricts outsourcing, being the bounded decision capacity
(BDC), per definition does not lead to evaluations at all. Therefore, neither positive nor negative
evaluation results occur, in this case.
184
These simulation results as well as the following analyses of the reasons for negative evaluation
results are based on 25 simulation runs with 150 periods.
185
Furthermore, legal constraints could lead to a negative evaluation, although this is not considered
within these simulations. As already noted in section 5.3.3, considering the legal constraint, not to
outsource more business activities than to keep inhouse (cf. section 3.5.1.1), will not make sense
as long as the simulations cover only a part of the banking business.
 318 Analytical and Simulative Studies

externalities on the expectation value. Overall, in the basic parameter setting, 6-

16% of the negatively evaluated coalition memberships are induced by uncer-
tainty over time, with a slight decrease of uncertainty over time (Figure 127 –
left). The right diagram reveals some more details by providing the values sepa-
rately for the different business functions (on average over time). Process-
ing/servicing and risk monitoring show the highest proportions of positive static
NPV (up to 21.5%). There is higher uncertainty when more banks outsource the
particular business function, and thus there are more dynamics present in this
particular market.
4.3% 1.2% 2.5%
100% 100%
14.8% 9.7%
90% 90% 21.5%
80% 80%
relative frequency

relative frequency

70% 70%
60% 60%
50% 50% 95.7% 98.8% 97.5%
85.2% 90.3%
40% 40% 78.5%
30% 30%
20% positive static NPV 20%
10% negative static NPV 10%
0% 0%
1 21 41 61 81 101 121 141

ut
n

ll
g
n

rin

ra
io

in
io

ko
period
at

ve
ic

ito
s

or
i

rv
ar

ec

O
on

W
se
ep

./d

m
g/
pr

business function
m

k
s in
s/

ss

is
le

R
es
se
Sa

static NPV < 0 static NPV > 0

oc
As

Pr

Figure 127: Proportion of positive and negative static NPV

when expected NPV is negative
In the majority of cases, where the static NPV is already negative – i.e.
transaction costs would outweigh the process cost savings for a given coalition
structure – the question arises about which part of the transaction costs is primar-
ily responsible for this negative result. Figure 128 shows for the different busi-
ness functions, whether the periodical net savings (PCS minus agency costs,
interface costs, and coordination costs) are already negative or whether they are
positive, but diminished by start-up transaction costs for negotiation and adop-
tion. The left diagram shows the average number of occurrences of negative
static NPV evaluations, distinguishing between negative and positive periodical
net savings. The right diagram shows the same data in the form of relative fre-
quencies.
As Figure 128 shows, an overwhelming part of the periodical net savings for
most of the business functions is already negative, while for the minority of the
business functions, the NPV becomes negative because the setup transaction
costs (negotiation and adoption) exceed the net present value of the positive net
savings during the contract duration.
 Analytical and Simulative Studies 319
avg. number of actors in coalitions

100 10
90 decentralized 9 decentralized

avg. number of coalitions

80 centralized 8 centralized
70 7
60 6
50 5
40 4
30 3
20 2
10 1
0 0

ut
ut

ng

g
on
n

g
ng
n

ko
io

rin
ko
tio

rin
io

ci
isi
at
ci

or
to
cis

or
ra

vi
to

ec
ar
vi

W
i
er
W

on
i
pa

er

ep
e

on

./d

/s
./d

m
/s
re

pr
m

ng
m
ng
m

isk
/p

s/

ss

si
isk
ss
s

le
si

es

R
se
le

Sa
es

R
se

business function

oc
Sa

As
business function
oc
As

Pr
Pr

Figure 128: Proportion of positive and negative periodical net savings NS

when static NPV is negative (over 100 actors)
Thus, in the majority of cases, periodical transaction costs would exceed
process cost savings if the bank would enter into the evaluated coalition. Never-
theless, there are again significant differences between the different business
functions. The ratio between process cost savings and transaction costs proves to
be higher for highly automated processes of processing/servicing and risk moni-
toring, leading to a higher fraction of positive NS.
Then, we went into more detail in order to analyze the main reason for the
negative periodical net savings. When analyzing the cases in which the periodi-
cal net savings are negative, Figure 129 demonstrates the relative level of differ-
ent transaction cost types compared with process cost savings (PCS). Ratio val-
ues larger than 1.0 show that the PCS are already exhausted by this single trans-
action cost element. The mean values, standard deviation, and the fraction of
ratio values above 1.0 are given above the boxes.
Interface costs (CIF) and agency costs (CAG) have a huge negative impact on
the business functions sales/preparation, assessment/decision, and workout.
Coordination costs (CC) do not play a major role (please note the difference in
ordinate scaling) and have the greatest negative impact on processing/servicing.
Overall, the impact of the different transaction cost types is rather balanced for
processing/servicing, while risk management is more strongly influenced by
agency costs, followed by interface costs.
320 Analytical and Simulative Studies

IF AG
C / PCS C / PCS

mean=4.98 mean=4.94 mean=.82 mean=.88 mean=2.83 mean=1.56 mean=5.42 mean=1.03 mean=1.77 mean=2.67
40 40
sd=5.51 sd=6.34 sd=.87 sd=.95 sd=3.30 sd=2.42 sd=7.75 sd=1.07 sd=1.72 sd=3.38
35 71.3%>1 73.1%>1 31.4%>1 33.3%>1 66.2%>1 35 42.0%>1 58.3%>1 39.6%>1 63.9%>1 60.2%>1

/ PCS ratio
30 30
C / PCS ratio

25 25
20 20
15 15

AG
IF

10 10

C
5 5
0 0

g
g

.
.

ut
g
ut
g

.
.

ep
ep

ec

in
ec

in

rin
rin

ko
ko
vic

vic
Pr
Pr

./D
./D

ito
i to

or
or

s/
s/

er
er

on
on

W
m
m
le

le
/S

/S
ss
ss

m
Sa
Sa

s.
s.

se
se

sk
sk
es

es
As

As
Ri

Ri
business function business function
oc

oc
Pr

Pr

management
monitoring/
Risk
C

decision
Assessment/
C / PCS

servicing
Processing/
preparation
Sales/

Workout
Business function
mean=.29 mean=.22 mean=.77 mean=.31 mean=.21
4.5 sd=.37 sd=.29 sd=.78 sd=.32 sd=.25
4 12.0%>1 12.1%>1 29.5%>1 13.1%>1 9.6%>1
3.5
C / PCS ratio

Total no. of cases (in 25 sim.

3 5,553 74,478 71,246 107,930 162,946
runs with 150 periods)
2.5
2 Fraction of CIF 1.5% 8.7% 7.8% 8.3% 9.2%
1.5 extreme values
C

1 CAG 4.4% 3.8% 2.1% 3.0% 3.7%

(not considered
0.5 in the box plots) CC 3.9% 8.2% 6.7% 7.3% 8.1%
0
g
.

ut
g
.
ep

ec

in

rin

ko
vic
Pr

./D

ito

or
s/

er

on

W
m
le

/S
ss

m
Sa

s.
se

k
es

s
As

Ri

business function
oc
Pr

Figure 129: Ratio between different transaction cost types and process cost
savings in negatively evaluated periodical net savings (table gives
information about the sample size and the fraction of excluded
extreme values for all diagrams.)
The graph in Figure 130 summarizes the course of the preceding analysis of
the actors’ decisions and evaluations related with cooperative sourcing. For the
business function processing/servicing, the established frequencies of different
reasons for not participating in a cooperative sourcing coalition are depicted. For
this business function, we can conclude that, at least in the basic parameter set-
ting, cooperative sourcing activities are inhibited in the overwhelming majority
of cases by their own economic calculus (too low expected savings) and only
rarely by negative votes from the potential coalition partners. Reasons for insuf-
ficient expected savings are too high transaction costs and uncertainty about the
partners’ future behavior (externalities). All of the given numbers are taken from
the figures above, thus we have forbeared from presenting the graph for all busi-
ness functions.
Analytical and Simulative Studies 321

cooperative
sourcing
voting result:
positive

97.9% uncertainty about

potential partners’
behavior
evaluation result:
static NPV
positive start-up transaction
2.1% positive
costs outweigh
evaluate 38.0% voting 21.5% positive periodical
cooperative sourcing result: negative periodical savings
of net savings
processing/servicing no cooperative expected NPV positive
62.0% sourcing negative 78.5%
evaluation 95.7% static 31.6% interface costs
result: negative NPV negative outweigh
analyze reasons process
no cooperative 31.4% cost savings
4.3% 69.4%
sourcing strategic coord. costs
constraint periodical
outweigh
binding no coop. sourcing net savings 29.5% process
due to strategic reasons negative cost savings

Note: values represent average frequencies (over time). 39.6% agency costs
Usually, values depend at least to a small degree on the stage of the simulation process. outweigh
Results are only valid for processing/servicing and for the given basic parameterization. no disjoint process
cost savings
alternatives!

Figure 130: Overview of the preceding analysis,

exemplary for processing/servicing
Actually, these results may only occur due to the specific parameter setting.
Hence, sensitivity analyses in the later sections will test whether (and how) co-
operative sourcing activities will change in varied settings.
However, before different setscrews in the model parameterization are al-
tered, the economic results from the banks’ cooperative behavior will be com-
pared with a centralized optimization of the system. In section 4.3, we developed
a centralized version of the cooperative sourcing model which determines a cost-
efficient cooperative sourcing configuration from a global perspective. By solv-
ing this cooperative sourcing problem (CSP), instantiated with the same parame-
terization as the simulations, the degree of inefficiency, resulting from decentral-
ized behavior in the presence of bounded rationality and externalities, can be
measured. In order to solve this combinatorial problem, a genetic algorithm was
developed, configured (section 5.2), and applied for obtaining the following
comparative results.
Table 56 compares the global monetary results from centralized coordina-
tion with decentralized coordination.
 322 Analytical and Simulative Studies

Relative
Decentralized coordination
Centralized coordination difference of
(after 250 periods)
means
mean sd mean sd [in %]
PCS absolute 123m MU 19.5m MU 167m MU 19m MU 36%
TCp absolute 7.5m MU .5m MU 18.1m MU .6m MU 141%
absolute 115m MU 19.5m MU 148m MU 19m MU
NS rel. to org. 28%
55% 6.1 p.p. 70.4% 3.4 p.p.
process costs
Table 56: Comparative results of decentralized and centralized coordination186
On average, the solutions from the centralized coordination result in about
28% higher global net savings. The process cost savings are even higher (36%),
but are offset by much higher transaction costs. The solution quality of the GA
proves to be very consistent (standard deviation of only 3.4 percentage points).
The corresponding effects on the market structure are depicted in the follow-
ing charts:
HFvol (process volume-based Herf. index)
HFact (actor-based Herfindahl index)

0.45 0.70
decentralized decentralized
0.40 0.60
centralized
0.35 centralized
0.50
0.30
0.25 0.40
0.20 0.30
0.15
0.20
0.10
0.05 0.10

0.00 0.00
n

ng

ut
n

on
t

io
ng

rin
on
io

ko
rin

ici
at
ko

isi
ic i
at

it o
isi

or
ar

rv
it o

or

ec
ar

rv

se

on
ec

W
ep
on

W
se

./d
ep

m
g/
./ d

pr
m

m
g/
pr

sin
s/

k
m

ss
sin

s
s/

sk

le

es

Ri
ss

se
le

Sa
Ri

business function
es

oc
se
Sa

As

business function
oc

Pr
As

Pr

Figure 131: Average “market concentration” for the provision of the different
business functions, based on actor-based and process volume-based
Herfindahl index (decentralized coordination results from the 250th
period)
While the actor-based market concentration is much higher in a centrally
coordinated system for all business functions (Figure 131, left), the Herfindahl
index, based on process volumes (Figure 131, right), shows only slight devia-
tions of the back office and even slightly lower values for the two right back-
office business functions, indicating a slightly more homogenous coalition struc-
ture. Figure 132 shows the corresponding numbers of coalitions (right) and ac-
tors in coalitions (left).

186
“Xm MU” stands for “X million monetary units”.
Analytical and Simulative Studies 323

100 10
avg. number of actors in coalitions

90 decentralized 9 decentralized

avg. number of coalitions

80 centralized 8 centralized
70 7
60 6
50 5
40 4
30 3
20 2
10 1
0 0

g
.

ut
ut

g
.

on
ep
g
g
ep

in
n

rin
ko

ko
in
in
io

v ic
isi
pr
pr

it o
c
cis

or

or
to

s/

ec
vi

er
s/

on
W

W
i
er
e

le
on
le

s
./d
./d

Sa
/s

m
g/
Sa

m
ng
m

sin

k
isk

ss

s
ss

si

es

Ri
se
es

R
se

business function business function

oc
As
oc
As

Pr
Pr

Figure 132: Average number of actors in coalitions and number of coalitions

(decentralized coordination results from the 250th period)
In the centralized “benchmark” world, about 90% of banks cooperatively
source all business functions, except sales/preparation, and organize themselves
in 6 to 9 coalitions per business function. Compared with the decentralized sys-
tem, much more cooperative sourcing activities take place in the middle office
(assessment/decision) (Figure 132). In the decentralized scenario, this business
function plays almost no role in cooperative sourcing strategies.
In the case of centralized coordination, it is interesting to see how many
banks are confronted with negative resulting consequences. Centralized coordi-
nation only maximizes system-wide net savings without ensuring positive net
benefits for each of the individuals. The question is whether negative conse-
quences are the primary reason for the decentralized coordinated system not
arriving at the centralized solution or whether uncertainty and bounded rational-
ity are more responsible. The following table breaks down the results into the
different possible reasons for individual deviation.
324 Analytical and Simulative Studies

Table 57: Deviations between centralized and decentralized coordination and

sources for reason (average values from 200 simulation runs)
The first row shows the average difference in number of banks joining a co-
operative sourcing coalition from centralized vs. decentralized coordination. This
difference is the net effect of banks that did not join a coalition, despite the prob-
able recommendation of centralized coordination (3rd row) and vice versa (2nd
row). The first group is split further into banks that would actually face negative
monetary consequences (4th row) (negative NS) and banks that would not face
those (5th row) if following the suggestions of the central optimizer. Particularly
in the cases of processing/servicing and risk monitoring, the proportion of banks
that would achieve negative net savings, when following the suggestions of cen-
tralized coordination, is only around one-fourth (23.7% and 26.6%, respectively).
For workout, it would be 43.3% and for assessment/decision, it would be 48.6%.
These negative individual benefits provide an first reason for the missing
outsourcing activities in the decentralized scenario. The second explanation for
this inefficiency dilemma, from a global perspective, lies in the behavioral uncer-
tainty and interdependencies (network effects). Since banks cannot assume a
stable coalition as guaranteed in the long term, they decide not to outsource their
business functions. Thus, in this case, the global inefficiency dilemma also repre-
sents an individual inefficiency dilemma for these banks, too, since they would
save costs if deciding for cooperative sourcing.
Analytical and Simulative Studies 325

As a summary of this section the following key findings can be noted:

o In the basic parameter setting, the system of banks demonstrates moderate to
high cooperative sourcing activities for the back-office business functions
which lead to significant net savings. From an individual perspective, the
largest banks show the highest savings.
o The resulting coalitions differ strongly in size.
o Depending on the business function characteristics, different patterns of the
system dynamics can be found (e.g. over-reaction for processing/servicing).
o The greatest part of the savings is already achieved in the first periods (inde-
pendently of whether the system shows over-reaction or not). In later peri-
ods, only marginal additional savings can be achieved from individual opti-
mization of own coalition memberships.
o Members of a coalition have a homogeneous attitude towards an extension
of their coalition by new applicants.
o Depending on the characteristics of the particular business function, the
different types of transaction costs have a discriminative impact on the NPV
of a cooperative sourcing evaluation.
o The evaluation of a coalition membership is partially influenced by behav-
ioral uncertainty (network effects) – this uncertainty is the stronger, the
more banks have outsourced the particular business function, due to the re-
sulting higher dynamics.
o Centralized coordination, i.e. optimization of the whole system from a cen-
tralized perspective, leads to significantly greater savings, but also leads to a
certain fraction of firms which achieve negative monetary results. The opti-
mization leads to higher market concentration but does not lead to monopo-
lization, as one would expect. On average, 6 to 9 coalitions serve the differ-
ent business functions.
o The divergence of the monetary results of centralized vs. decentralized co-
ordination represents a global inefficiency dilemma. In cases where banks
decide not to outsource because they cannot assume a coalition to be stable
(at least during their contractually agreed membership) this represents an in-
dividual inefficiency dilemma for the banks themselves, too. The strength of
this effect, which occurred in up to three-quarters of the banks, depends
again on the characteristics of the respective business function.
One objective of the following sections is to test the stability of the estab-
lished results in regards to the different assumptions made in the parameteriza-
tion section (5.3.3). When developing a simulation model, artificial components
have to be included and assumptions about parameter settings and distributions
have partially to be made without an empirical basis being available.
 326 Analytical and Simulative Studies

5.3.4.2 Analysis of Process Cost Structure Variations

In this section, two analyses are conducted. First, we argue that larger scales
resulting from cooperative sourcing will often lead to a change of technology
becoming more favorable, which leads to higher fixed costs but lower variable
(unit) costs (cf. section 2.1.1.3). Second, the basic process cost structure will be
affected by altering the ratio between fixed and variable costs. Although these
ratio parameters were extracted from empirical data in section 5.3.3.2, this pro-
cedure was based on several assumptions which will be relaxed in the following
to test for the sensitivity of the results on those assumptions.

Variation of Coalition Process Cost Structure

In order to test the effect of technology changes when establishing a cooperative
sourcing coalition, the following simulations alter fixed and variable cost pa-
rameters of the coalition’s (i.e. the insourcer’s) process cost function. It can be
argued that larger scales will usually lead to a higher degree of automation being
beneficial. In the process cost function, this can be represented by increased
fixed costs and decreased variable costs. In the following, the impact of these
changes on global monetary measures and market structure is analyzed187.
The following figure shows the average global net savings (NS) relative to
the original process costs (which have not changed, compared with the basic
simulations above). Each data point in the figure represents an average of 150
simulations. Fixed costs KF and variable costs cP of coalitions’ insourcers have
been varied by the factors alter_KF and alter_cP in the following ranges: al-
ter_KF = [1.0, 25.0], alter_cP = [1.0, .1].

187
The reader should be aware that the basic fixed and variable process cost parameters are not
altered in this step. Only when a bank becomes the insourcer of a coalition’s process volume, the
parameters will be changed.
Analytical and Simulative Studies 327


 
 
     
  
   
        
  
 
   
 
        
    

    
0.6 
 



  
   

   
   
               
  
     
        
       
  
  
     
      
 
       
        
       
   
  
     
         
  
       
0.5  
 

 
  
 
  
relative global net savings

   
  
 
 
 


 

0.4 


 




0.3


0.2 


25.0
20.0 0.2
15.0 0.4
10.0 0.6
alter_KF 5.0 0.8
1.0 alter_cP

Figure 133: Relative global net savings in relation to alter_KF and alter_cP
The chart shows that the monetary results are not sensitive to changes in the
coalition process cost structure. The original situation is demonstrated by the
corner in the front (alter_KF = 1.0, alter_cP = 1.0). Even if cP remained constant
(alter_cP = 1.0), KF would have at least to be quintupled to lower the net savings.
Furthermore, a slight reduction in cP would keep the net savings at their level,
even if fixed costs were to increase drastically.
The next figure shows the counterpart of centralized coordination (i.e. re-
sults from the GA optimization). Here, the results are completely insensitive to
the increase in fixed costs in a sourcing coalition. Instead, a slight decrease in
variable costs in cooperative sourcing leads to improved relative global net sav-
ings (80% instead of the original 76% at alter_cP = 1.0).
328 Analytical and Simulative Studies
 
 
 

 
 

   

 

 
  
   
  
   
   
    
     
      
 
0.80 



 





















       
       
        
           
     
       
        
        
         
   
  
 
  
         
   
       
 
      
       
      

from centralized optimization

   
 
 
 

relative global net savings



0.60 





0.40

0.20

0.00
25.0
20.0
15.0 0.2
10.0 0.4
0.6
alter_KF 5.0 0.8
1.0 alter_cP
Figure 134: Relative net savings resulting from centralized coordination
in relation to alter_KF and alter_cP 188
Centralized coordination does not account for individual consequences.
Hence, Figure 135 investigates the number of banks which face a monetary loss
due to centralized configuration changes. According to the last column of Table
57 (p. 324), the charts distinguish between deviation types I and II189 and cumu-
late the frequencies of all five business functions.

188
Due to the significantly longer computation times required by the genetic algorithm, the central-
ized coordination results are based on a smaller number of repetitions. Here, each data point
represents the average of 25 simulation runs.
189
These are banks that have cooperatively sourced but should not (I), from centralized perspective,
and vice versa (II).
 Analytical and Simulative Studies 329
 
Frequency (over all banks and business functions)

 
   
 
     
  

20 









   



    
      

  
 
 
  
       
300
           
      
   
       
    
    
     
       
      

        
15
  

   
       


   
        
          
        
     
     

       
 
              
    
       
     

            



 







200    
  

  
           
      
   
         
10       
   

    
      
      
  
    
   
  
     
  
 
   

  

100
5




deviation type I deviation type II

0 0
25.0 0.2 25.0
20.0 20.0 0.2
15.0 0.4 15.0 0.4
10.0 0.6 10.0 0.6
alter_KF 5.0 0.8 alter_cP alter_KF 5.0 0.8 alter_cP
1.0 1.0

Figure 135: Deviations between centralized and decentralized coordination

(frequency over all banks and business functions (total of 500))
As can be seen, deviation type I, although quite irrelevant in regards to the
absolute frequency190, increases significantly with a reduction of alter_cP. Devia-
tion type II shows less intense and opposite effects. While the average of 217
deviations (cf. Table 57) is only slightly reduced by changing alter_cP, higher
fixed costs lead to a significantly greater number of deviations (only as long as
alter_cP is not reduced). The main source of increasing type II deviations is an
increasing number of banks which receive negative net savings when following
the result of centralized optimization. Higher fixed costs in a coalition lead to
greater asymmetry and thus to a greater opportunity to bundle process volumes
but also increases the probability of instability (i.e. insufficient benefits for some
of the participants). Here, the result of the game-theoretical stability analysis is
not maintained. If the fixed costs increase upon entering the coalition, the pro-
portional allocation mechanism no longer ensures positive benefits for all mem-
bers, even if no transaction costs are apparent (cf. section 5.1).
Shifting the focus to the resulting market structure (Figure 136) provides a
possible explanation for the quite insensitive reactions of the global monetary
values found in the decentralized coordination (Figure 133). For the major busi-
ness functions, which are cooperatively sourced (processing/servicing and risk
monitoring), the market concentration (measured by HFact) increases rather sig-
nificantly with increasing fixed costs, and decreases again with very high fixed
costs. On the other hand, it decreases slightly with decreasing cP (Figure 136,
upper charts). By contrast, for assessment/decision, there is a slight ascent with
decreasing variable costs (and almost no correlation with alter_KF). For workout,
190
There are only between 10 and 20 deviations per 500 business functions (note the differently
scaled z-axes).
 330 Analytical and Simulative Studies

there is also an increase in market concentration with decreasing alter_cP, but,

moreover, a strong drop with increasing alter_KF and high alter_cP.


 
 

0.40





 


Processing/ 0.40 Risk monitoring
 

  





servicing
 
 
   
0.30 
      0.30 

 

 
    

   

      
   
    
     
  
 


    
   
                  

          
             
   
  
 
     
  
 
   
  
   
 
        
         
    
      
0.20

  





 0.20 
 

 
  




 
         
             
         
           
        
         

    
    
     


 
     

       
         
   

0.10
    
Actor-based Herfindahl index

  
0.10 
       
   

 

      
   
    
    

0.00 0.00
25.0 25.0 0.2
20.0 0.2 20.0 0.4
15.0 0.4 15.0
0.6 10.0 0.6
10.0
alter_KF 5.0 0.8 alter_cP alter_KF 5.0
1.0
0.8 alter_cP
1.0 




  

  
  




 
 
 

Workout

0.10 

 
 
0.10  

Assessment/
  
  
 
 

  
     
    


decision
   
     
   

  
 
0.08 0.08
   

 



  
    
      
   
  
    
  
   
   

   
   
0.06

 


0.06   


 
   

  
        
    
    
   
   
  
 
   
  
     
    
0.04
 
0.04
  
    
    
     
       
     
     
    
   
  
   

    

 
      
   
    
0.02
  
0.02 

   
    






    
   
   
        
      
  
 


      
  
        
    
25.0 

 
  

0.2 25.0
20.0 

 
20.0 0.2
15.0
 
 0.4 0.4
 
0.6 15.0
10.0 10.0 0.6
alter_KF 5.0
1.0
0.8 alter_cP alter_KF 5.0 0.8 alter_cP
1.0

Figure 136: Actor-based Herfindahl index for different business functions

in relation to alter_KF and alter_cP 191
For processing/servicing and risk monitoring, it can be argued that the banks
react to the changes in situation of process cost structure by forming larger or
smaller coalitions. Since the high fixed costs only occur within a coalition (and
only once per coalition), a stronger market consolidation neutralizes this effect.
Consequently, the global net savings do not change significantly, even if the
parameterization is altered greatly. The drop in HFact at very high KF and high cP
can be traced to the fact that, in these areas, the advantageousness of cooperative
sourcing diminishes because the cost structure becomes unfavorable. A sharp
increase of fixed costs has to lead to a large decrease in variable costs; otherwise

191
Process volume-based Herfindahl index leads to the same pictures at a higher overall level, due to
the asymmetric distribution of process volumes across the banking industry. Results for the sales
function have been omitted due to a lack of cooperative sourcing activities.
Analytical and Simulative Studies 331

the technology change would be inefficient (except in cases where the previously
used technology is bounded in capacity).
The various effects, which are presented in the lower diagrams, can be ex-
plained as follows:
o Assessment/decision: As there are almost no cooperative sourcing activities
in the original setting, changing the cost structure – if it has any effect at all
– will only lead to more activities and thus higher market concentration.
This occurs with reducing cP.
o Workout: The process characteristics cause workout to be positioned be-
tween the frequently outsourced and the almost never outsourced business
functions. Thus, the lower right chart aggregates the effects which can be
observed in the remaining three diagrams. At high cP, cooperative sourcing
becomes unfavorable for most of the banks (drop of Herfindahl index). If
only increasing KF minimally, the system behavior proves to be equal to as-
sessment/decision. But, when reducing cP, cooperative sourcing activities
and the formation of larger coalitions can be significantly advanced, even
when KF is also increased, according to the upper diagrams. Hence, it can be
concluded that the banks react very sensitively to cost changes regarding the
workout sourcing strategy.
In order to support this argumentation, Figure 137 shows the average num-
ber of banks participating in cooperative sourcing, and the number of coalitions
for all of the back-office functions.
The market structures which result from centralized coordination are quite
similar (no figures): HFact for processing/servicing and risk management shows
almost the same results as in Figure 136 (upper diagrams) while the levels for
assessment/decision and workout are higher (between .2 and .3) for all combina-
tions of alter_KF and alter_cP. For all business functions, the market concentra-
tion decreases slightly with a decrease in alter_cP but does not show significant
changes with an increase in alter_KF.
 332 Analytical and Simulative Studies

Number of banks in coalitions Number of coalitions



 
   
  



   
   
      
     
 
12
   
   
  
     
    
       
   
75
  
      
Processing/servicing

 
      
     
      
        
   
 
 
 
     
  
 
    
         
  
    
    




8 




 
 




50 

 








     
    

  
    
    
  
 
   
  
   
   

  
  

       
4 

 
25 


 
     
  
  



  
    
  
  
  
      
  
0 0  
 
 



25.0 0.2 25.0 0.2

20.0 20.0
15.0 0.4 15.0 0.4
10.0 0.6 10.0 0.6
alter_KF 5.0
1.0
0.8 alter_cP 12
alter_KF 5.0
1.0
0.8 alter_cP
75 

   
     
      
    
   
               


 
  
  
Risk monitoring

       
   
    
              
      
        
 

   




 
  

8 
50       
      

       
    
  

  
       
    
       
    
   
  
 
   
     

 
     
      

     
25  4  
     
   
 

 
  
    
 

     
    
 
   
   
   
   
      
    
 
  
 
     
  
  
0 0  
 
  
 
25.0 0.2 25.0
20.0 20.0 0.2
15.0 0.4 0.4
0.6 15.0
10.0 12 10.0 0.6
alter_KF 5.0
1.0
0.8
alter_cP alter_KF 5.0 0.8 alter_cP
75 1.0

8
50  



  
Workout

  
   
     

       
  
    
     
     

      
  
     
 
  
   
 


      



 
     

4 
           
 

25  
  



 
    
   


        
    
        
    
        

  

        
     
     
        
     
   
     
    
   
    
     
      

0 
 

    


 0 

 
  
 
 

 
      
    
25.0
    
  
25.0 0.2
 
0.2
20.0 20.0
0.4 15.0 0.4
15.0 0.6 0.6
10.0 10.0
0.8 5.0 0.8
alter_KF 5.0
1.0 alter_cP alter_KF 1.0 alter_cP
Figure 137: Average number of coalitions and number of banks in coalitions in
relation to alter_KF and alter_cP
The change in “market structures” is related to a diffusion change of the dif-
ferent business models. For the decentralized coordination scenario, the charts of
Figure 138 show how the absolute frequency of business models changes in the
sensitivity analysis.
Analytical and Simulative Studies 333

60 20

50


number of banks
number of banks

 Traditional 15
Fully integrated bank
40 
fully integrated with service provision






bank 10
30 
 


 
  
       
 
         
    
     
       
      

  
 
20 


  
      


      
 
  
  
 




 
 
           
 

  

  


   
  

 
   
5  






  

 

        
     

  
               
 


    
    
10 
          
 
      

   
 





 

  
    
     

   
       
  
     
       
     
  
     
   
25.0 0.2 25.0    
 
20.0 20.0
  
  0.2
15.0 0.4 0.4
60 0.6 15.0
10.0 10.0 0.6
alter_KF 5.0
1.0
0.8
alter_cP 60 alter_KF 5.0
1.0
0.8 alter_cP

  
   
number of banks

     
      
           
        
        
   
 
40 
  
    
 


   

 

   

40
 
  
           
   

          
           
             


          
    
 

           
              

       
 
              
                
       
 
   
          

 
       
       
     
     
       
  
  
  
   
 

20 




 


 
20 


 
 

Major




Selective 

outsourcer



outsourcer
0 0
25.0 0.2 25.0
20.0 20.0 0.2
15.0 0.4 0.4
0.6 15.0
10.0 10.0 0.6
alter_KF 5.0
1.0
0.8 alter_cP alter_KF 5.0 0.8 alter_cP
1.0
60
number of banks

40 Sales bank



 
  
 
 
 
 
   
     
     
20  

  



 

  
   
   
      
     
   
  
   
   
 
 

 

      
  
   


  
  
0
 


  

 
25.0   
 


20.0 0.2
15.0 0.4
10.0 0.6
5.0 0.8
alter_KF 1.0 alter_cP
Figure 138: Frequency of business models in relation to alter_KF and alter_cP
(please note the differently scaled ordinate in the upper right chart)
While the frequency of the traditional fully integrated bank decreases with a
decrease in alter_cP but greatly increases with higher fixed costs, integrated
banks with additional third-party offerings (fully integrated bank with service
provision) show a (weaker) opposite trend. A similar pair-like relationship exists
334 Analytical and Simulative Studies

for selective and major outsourcers as well as for sales banks. With increasing
fixed costs, outsourcing becomes less favorable, therefore, major outsourcers and
sales banks transform to only selective outsourcers. With a slight decrease in
alter_cP, this effect diminishes. Furthermore, decreased variable costs lead to a
slightly higher fraction of sales banks, in particular.
The following tables summarize the monetary and market structure results
from the sensitivity analysis regarding alter_KF and alter_cP. Under both decen-
tralized and centralized coordination, the deciders utilize a changed market struc-
ture and thus are able to achieve the same global net savings of process costs as
in the original setting.
Increase of alter_KF Decrease of alter_cP
Decentralized Centralized Decentralized Centralized
coordination coordination coordination coordination
Negative impact only
Relative Very weak
for strong increase and No impact No impact
NS positive impact
at alter_cP = 1.0
Table 58: Sensitivity analysis regarding alter_KF and alter_cP
Increase of alter_KF Decrease of alter_cP
Assessm./ Process./ Risk Assessm./ Process./ Risk
Workout Workout
decision servicing monitoring decision servicing monitoring
dec cen dec cen dec cen dec cen dec cen dec cen dec cen dec cen
Herfin-
= = n = np = p = (n) = p p p p (p) =
dahl
Number
of coali- (p) (p) (p) (p) (p) (p) (p) (p) (n) = n (n) n (n) (n) p
tions
Number
of banks
(p) = p = p = (p) = n = n = (n) = n =
in coali-
tions
Table 59: Sensitivity analysis regarding alter_KF and alter_cP (cont.)192
(dec/cen = de/centralized coordination result)
As a final step, we analyzed the dynamics of the system, i.e. how the level of
activities is affected. As in the previous section, the number of outsourcing,

192
The sales/preparation function is omitted due to the absence of cooperative sourcing activities.
np indicates an initial increase and later decrease with additional variation of the parameter. Ar-
rows in brackets represent slight but insignificant changes of the analyzed measure.
Analytical and Simulative Studies 335

switching, and backsourcing actions within a time frame of 150 periods was
counted for each bank.
In order to be able to estimate the actors’ uncertainty about the others’ be-
havior, the number of switching and backsourcing actions must be set in relation
to outsourcing activities. Higher ratios mean that the banks have to test for more
sourcing configurations unless they find their local optimum193. Furthermore,
additional activities in the system lead to more changes of the decider’s envi-
ronment, which, in turn, aggravate the search for the optimal coalition. There-
fore, these ratios are used as proxies for behavioral uncertainty from the indi-
viduals’ perspective in the following.
The following figure shows the results. Ceteris paribus variations of either
alter_KF or alter_cP lead to a decrease of these “corrective measures” (i.e. less
switching and backsourcing after outsourcing a business function), while simul-
taneous variation keeps the levels constant, leading to the depicted saddle-shaped
graphs.
1.0
4.0 switching actions 
backsourcing actions
per outsourcing action per outsourcing action

0.8
3.0 


  
   
 
 
  
  
    
  0.5
2.0 




 





     
     
    
    

  
 
         
        

   
 


        
        
    
       


 
    

     
  
       

 
 
  
 

         
              

 
0.3
    
  
    
1.0  
        

   
         
 

            

  
  
     
         
    
  
       
        
      
      
       
 



 

   
   
   
 
   
       
       
     
   
       
             
0.0   

 

0.0 


25.0 25.0 20.0 0.4 0.2

20.0 15.0 0.4 0.2 15.0 10.0 0.8 0.6
10.0 5.0 0.8 0.6 5.0 1.0
1.0
alter_KF alter_cP alter_KF alter_cP

Figure 139: Number of switching and backsourcing actions relative to

outsourcing actions194

193
The local optimum is defined as the firm’s finally chosen sourcing strategy (insourcing vs. coop-
erative sourcing, choice of coalition). Noticeably, this may not be the optimal strategy even from
the firm’s individual perspective, but there is not better solution found (maybe due to transaction
costs for switching: CAD and CN). This definition is taken from the discipline of (mathematical)
optimization.
194
Each data point represents the average of 200 actors of all sizes. This simplified analysis, which is
based on the average values of all banks, neglects the asymmetry in firm sizes, which might have
an additional impact. However, regression analyses which test for moderating effects of firm size
on the impact of alter_KF or alter_cP variation were shown to be insignificant.
336 Analytical and Simulative Studies

Both changes – increased coalition fixed costs as well as reduced coalition

variable costs – lead to a more certain decision basis – making either cooperative
sourcing or not cooperative sourcing more favorable for all of the parties.

Sensitivity Analysis of Basic Process Cost Structure

The process cost parameterization derived in section 5.3.3 is based on several
assumptions about both the distribution of process costs to business functions
and the ratio of fixed and variable costs. In order to relax these assumptions and
to analyze their impact on the results found above, the ratio between fixed and
variable costs will be varied. In Table 46 (p. 294) this ratio was determined from
empirical data for each of the five business functions. The following figures
display the results from gradually varying all ratio parameters by a factor al-
ter_costRatio to both the quartered (alter_costRatio = .25) and the fourfold value
(alter_costRatio = 4.0) of the original setting.
Figure 140 shows the level of relative net savings (in relation to original
process costs) in the left chart (for both centralized and decentralized coordina-
tion) and the market concentration resulting from the decentralized coordination.
The vertical line marks the original setting (alter_costRatio=1.0)
0.8 0.25
avg. actor-based Herfindahl index
relative total net savings (NS)

0.20
0.6

0.15
0.4

0.10

0.2
centralized coordination 0.05
decentralized coordination

0 0.00
0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75 0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75
alter_costRatio alter_costRatio
Sales/Prep. Assessm./Decision Processing/Servicing
Risk monitoring Workout

Figure 140: Impact of process cost ratio variation on global net savings (left)
and market concentration (right, only decentralized coordination)195
From these charts, it can be determined that the simulation results do not re-
act considerably to a change in the basic process cost structure. A decrease in the
cost ratios (i.e. decreasing the level of fixed costs compared with variable costs)
will lead to slightly less cooperative sourcing activities and lower global net
benefits, but only from a decentralized perspective. The net savings from central-
ized coordination (left chart, upper graph) do almost not react to the decrease of

195
Empirical cost ratios were varied by factor alter_costRatio = [.25, 4.0] in steps of .125 with 200
simulation repetitions for determining each average point of the plotted graphs (32 x 200 = 6,400
simulation runs in total).
 Analytical and Simulative Studies 337

the cost ratios; the optimization routine counterbalances the parameter change by
slightly increasing the market concentration (no figure, local elasticities196 for
HFact between -.08 and .00). Increasing the ratio has no structural impact under
both coordination mechanisms. The local elasticities are .07 and .00 for NS from
decentralized and centralized coordination and between .02 and .35 for the Her-
findahl indices of the different business functions.
Although the market concentration remains stable for alter_costRatio
around and above 1.0, this does not apply to the underlying measures. As the
next figure shows, there is a moderate variation in both the number of coalitions
and the number of banks in coalitions (local elasticities are between .16 and .82
for the numbers of banks in coalitions and between .19 and .82 for the numbers
of coalitions).
90 10
a v e r a g e n u m b e r o f b a n k s in

average number of coalitions

80 9
70 8
60 7
c o a litio n s

50 6
40 5
4
30
3
20
2
10
1
0
0
0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75 0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75
alter_costRatio alter_costRatio

Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 141: Average number of coalitions (right) and banks in coalitions (left)
(decentralized coordination)
Finally, Figure 142 shows how the frequency of the resulting business mod-
els changes. The elasticities are given in the diagram.

196
Local elasticities are estimated as follows: For five different ranges between data points i and j
around the original parameter value (in this case: alter_costRatio = 1.0), the arc elasticity is com-
puted, which is the quotient of differences of endogenous variable y and exogenous variable x di-
y j  yi xi , j
vided by the means of values within the ranges: H iarc
,j ˜ . The final estimated elasticity
x j  xi yi , j
value is formed by the average of the five computed arc elasticities in order to flatten possible ir-
regularities in the local environment.
338 Analytical and Simulative Studies

45
major outsourcer
avg. frequency of business models [in %]

40

35
H=.09

30
H=-.04 selective outsourcer
25

20
H=-.57
sales bank
15
H=.47 "traditional" fully integrated bank
10

5 fully integrated bank with service provision

0
H=.31 PSP

0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75

alter_costRatio
Figure 142: Average frequency of business models
(decentralized coordination)
Generally, the number of traditional fully integrated banks decreases with
higher relative fixed costs, while the number of banks which have outsourced
more than one of their back-office functions (major outsourcer and sales bank)
and the number of fully integrated banks with service provision to third parties
increase. The frequency of the remaining business models does not change.
Apart from an absolute variation of the ratios between fixed and variable
costs, an analysis of the impact of the heterogeneity is also necessary. As expli-
cated in the previous section, the heterogeneity of the process cost level and
structure throughout the empirical sample is responsible for the high global proc-
ess cost savings. Therefore, the variation in cost ratios has been incrementally
reduced by converging the values of the empirical random number seed with the
corresponding means (of each business function). The distance between the
original ratio value and mean is reduced by a factor homogenize_costRatio which
is between 0.0 (i.e. original values) and 1.0 (all ratio values = mean).
As a result it has been found that similarly to the absolute variation of cost
ratios, monetary aggregate results do not differ at all when varying homoge-
nize_costRatio between 0.0 and 1.0. Furthermore, in this analysis, the market
concentration and frequency of business models also does not change (therefore
no figures are presented). The results from the basic simulations in section
5.3.4.1 remain completely stable with regard to the heterogeneity of cost ratios
throughout the system of banks. This can be explained by the fact that a suffi-
ciently large system (i.e. number of simulated banks) compensates for internal
heterogeneity and produces the same picture (cf. section 5.3.4.4).
Analytical and Simulative Studies 339

In section 3.6.2.5 (Table 23), three alternative cost application schemes were
applied to break down the overall process costs, which have been derived by the
empirical study, to the five business functions. While the previous simulation
results are based on cost ratios determined by all three schemes, we also tested
the impact of applying only one of them197. This test is necessary to validate the
selected approaches for deriving simulation parameters from empirical data.
The Kruskal-Wallis test was used for testing the similarity of the results.
Testing the global monetary measures (original process costs, process cost sav-
ings, periodical transaction costs, absolute and relative net savings) shows almost
no differences when applying only one of the three alternative cost application
schemes instead of all of them. In most cases, the results differ by less than 1%
from the simulations above. The only measure which shows any deviation is the
global periodical transaction costs TCp: when applying cost allocation scheme C,
the overall transaction costs are up to 6% higher than in the simulations above.
By contrast, allocation schemes A and B do not lead to significant deviations.
This is interesting because the cost allocation schemes only handle process costs
and not transaction costs. Allocation scheme C leads to more cooperative sourc-
ing activities, although the resulting net savings do not differ from the remaining
allocation schemes. Consequently, the altered ratio of fixed and variable costs
forces the banks to find another trade-off between process cost savings and
transaction costs in order to reach the same net savings as before.
Moreover, the use of only one instead of all three cost allocation schemes
leads to slightly higher HFact in most cases (except for workout in general and for
risk monitoring under cost allocation scheme A where lower concentrations
prove to be more apparent). Using only one allocation mechanism leads to higher
homogeneity of process cost structures between the banks. Thus, banks will most
likely tend to agree to a cooperative sourcing coalition because the structure of
their potential savings is also more similar.
The results of this section on process cost structure variation can be summa-
rized as follows:
o The monetary results do not react significantly to all tested changes of the
process cost structure under both decentralized and centralized coordination.
o Changes in a coalition’s process cost structure prompted by switching to a
dominant technology (leading to increased fixed costs and reduced variable
costs) affect the system behavior. Higher fixed costs will either lead to larger
coalitions (processing/servicing, risk monitoring) or to a drop in cooperative
197
For this purpose, the sensitivity analysis concerning the cost ratio, conducted above, was repeated
with cost parameterization data based solely on cost allocation schemes A, B, and C. In total, 3 x
32 x 150 (= 14,400) simulation runs were conducted.
340 Analytical and Simulative Studies

sourcing activities (workout). Reduced variable costs have only a small im-
pact on the system behavior, but little reductions already compensate for
sharply rising fixed costs and thus stabilize the degree of cooperative sourc-
ing throughout the system.
o Reduced variable costs in coalitions lead to a higher frequency of sales
banks which outsource their whole back office.
o A univariate change in coalition process cost structure (increased fixed costs
or reduced variable costs) reduces behavioral uncertainty and leads to the
banks finding their local optimum more quickly (either insourcing or the op-
timal coalition from their perspective).
o An increase in the ratio of fixed to variable costs (alter_costRatio) in the
basic parameterization leads to no structural changes in the results. Only de-
creasing the ratio, and thus making fixed costs less relevant, leads to a slight
reduction of cooperative sourcing activities and related monetary savings.
o Homogenizing the ratio of fixed to variable costs in the basic parameteriza-
tion does not lead to changes in the basic results.
o Using only one instead of all three cost allocation algorithms developed in
section 3.6.2.5 leads to greater homogeneity of the process cost structures
throughout the system and, as a consequence, to slightly more cooperative
sourcing activities.

5.3.4.3 The Impact of Transaction Costs

After analyzing the impact of the process cost structure, in this section the focus
shifts to the impact of the “counterpart”, i.e. transaction costs involved in the
outsourcing relationship.

Impact of Increasing Overall Transaction Costs

The model contains five different types of transaction costs. In order to analyze
the impact of the overall level of transaction costs, all transaction cost parameters
(cAD, cAG, cN, and cIF) are simultaneously increased by the same factor al-
ter_TCcoefficients.
Figure 143 shows the impact of transaction costs on monetary results by
varying alter_TCcoefficients between 1.0 (i.e. original parameterization) and
20.0 (i.e. thirtyfold increase of original transaction cost factors). The local elas-
ticities for the original value of alter_TCcoefficients = 1.0 are given in Table 60.
 Analytical and Simulative Studies 341

decentralized coordination centralized coordination

90% 90%
rel. global PCS 80% rel. global PCS

average relative PCS, NS, TCp

80%
average relative PCS, NS, TCp

rel. global NS rel. global NS

70% 70%
rel. individual NS rel. global TCp
60% 60%
rel. global TCp
50% 50%

40% 40%
30% 30%

20% 20%

10% 10%

0% 0%
1 6 11 16 1 6 11 16
alter_TCcoefficients alter_TCcoefficients

Figure 143: Relative average process cost savings, net savings, and transaction
cost resulting from decentralized (l.) vs. centralized coordination (r.)
For decentralized coordination (left diagram), the effect on global savings
(PCS and NS) is quite linear. From an individual perspective (avg. NS), the drop
in savings is much greater. A variation in transaction cost factors by factor 4
already cuts the benefits of cooperative sourcing in half. The actors’ decision
behavior is heavily influenced by transaction costs although they still remain at a
low level below 8% of the overall process costs.
Elasticities at Global Global Indiv. Global NS
Global NS
alter_TCcoefficients = 1.0 PCS TCp NS with-out CC
Decentralized coordination -.18 .48 -.24 -.62 -.23
Centralized coordination -.00 1.09 -.22 N/A -.17
Table 60: Local elasticities for monetary simulation results
By contrast, the centralized coordination (right diagram) holds high process
cost savings constant over a certain range by accepting higher transaction costs
to optimize the net savings. After quadrupling the transaction cost parameters,
the transaction costs become too high and lead to a drop in cooperative sourcing
activities.
This stability of cooperative sourcing activities for small increases of trans-
action cost parameters under centralized coordination comes with at a high price.
As the next figure shows, the amount of banks which would experience an indi-
vidual loss when following the sourcing configuration suggested by centralized
coordination sharply increases to over 30% before the reduction of sourcing
activities reduces this amount again.
342 Analytical and Simulative Studies

100%

90% positive individual net savings

negative individual net savings
average proportion of banks

80%

70%

60%

50%
40%

30%
20%
10%
0%
1 6 11 16
alter_TCcoefficients

Figure 144: Proportion of banks with (strictly) positive vs. negative individual
net savings resulting from centralized coordination
In comparing the resulting transaction costs and net savings from centralized
and decentralized coordination, a compatibility problem becomes evident. Parts
of the transaction costs (negotiation and coordination costs) are influenced by
learning effects. Thus, in the stationary state, the agents have accumulated learn-
ing effects which significantly reduce the transaction costs. In centralized coor-
dination, there is no possibility of reducing transaction costs by learning effects
because the whole system is optimized and configured only once. In order to
enable the comparison of the resulting monetary consequences, the following
figure shows, once more, the global net savings from both coordination mecha-
nisms, but without coordination costs (negotiation costs occur only in the trans-
fer period and are not part of NS per definition).
80%
rel. avg. net savings without coordination costs

70% centralized coordination

decentralized coordination
60%

50%

40%

30%

20%

10%

0%
1 6 11 16
alter_TCcoefficients
Figure 145: Comparison of relative average net savings, excluding coordination
costs (CC), for centralized and decentralized coordination
 Analytical and Simulative Studies 343

Together with the previous diagrams above, the figure shows that differ-
ences between both coordination mechanisms diminish after increasing the
transaction cost parameters by a factor of ten or more. Then, cooperative sourc-
ing no longer plays a major role in either system.
After analyzing the monetary results, Figure 146 displays the corresponding
market concentrations (HFact). For some business functions, the centralized co-
ordination (right chart) even increases market concentration to counteract the
increasing transaction costs. By contrast, for decentralized coordination (left
chart), the market concentration declines directly with only slightly increasing
transaction costs (Figure 146 – left). Workout shows the greatest drop, i.e., reacts
most sensitively to increasing transaction costs.
decentralized coordination centralized coordination
0.50 0.50
0.45 0.45
acto r-b ased H erfin d ah l in d ex
acto r-b ased H erfin d ah l in d ex

0.40 0.40
0.35 0.35
0.30 0.30
0.25 0.25
0.20 0.20
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
1 6 11 16 1 6 11 16
alter_TCcoefficients alter_TCcoefficients
Sales/Prep. Assessm./Decision Processing/Servicing Risk monitoring Workout

Figure 146: Average market concentration resulting from decentralized (left) vs.
centralized coordination (right)
Elasticities at Sales/ Assessm. Processing Risk moni-
Workout
alter_TCcoefficients = 1.0 prep. /decision /servicing toring
Decentralized coordination -.04 -1.22 -.36 -.74 -1.20
Centralized coordination .01 -.37 -.16 .23 -.04
Table 61: Local elasticities for market concentration results
In correlation to the results above, higher transaction costs lead to a decrease
in individual activities in finding an optimal coalition from an individual per-
spective. Since higher transaction costs hinder the change in coalitions, banks are
locked into a local optimum much more quickly. Thus, the behavioral uncer-
tainty, as defined above, significantly declines when the transaction cost factors
are only slightly increased. The following figure shows the ratio of switching,
backsourcing, and outsourcing, which is defined above as proxy for uncertainty,
accumulated over all five business functions.
344 Analytical and Simulative Studies

2
cum. switching/outsourcing actions
cum. backsourcing/outsourcing actions
1,5
ratio

0,5

0
1 6 11 16
alter_TCcoefficients
Figure 147: Switching and backsourcing actions relative to outsourcing actions,
aggregated over all business functions198
Finally, Figure 148 provides the frequencies of the different business models
resulting from the simulations (decentralized coordination) – analogous to Figure
142 in the section on process cost variation. Even slight variations in the transac-
tion costs have a strong impact on the market configuration (also reflected by the
elasticities in Table 62). While the number of selective outsourcers remains sta-
ble within a wide range, the frequency of major outsourcers and sales banks
drops in favor of the fully integrated bank.
100
relative frequency of business models [in %]

90
"traditional" fully integrated bank
80

70

60

50

40

30
selective outsourcer
20
major outsourcer
10
sales bank
0
fully integrated
1 6 11 16
bank with
service provision alter_TCcoefficients
Figure 148: Average frequency of business models
(decentralized coordination)

198
I.e. sum of all switching actions (resp. backsourcing) divided by sum of all outsourcing actions.
Analytical and Simulative Studies 345

“Traditional” Fully integrated

Selective Major
Business model fully inte- bank with service Sales bank
outsourcer outsourcer
grated bank provision
Elasticities at
1.23 -.41 -.10 -.43 -1.03
alter_TCcoefficients = 1.0
Table 62: Local elasticities for business models
Overall, the simulations show results that are sensitive to transaction cost
variations. Nevertheless, the sensitivity analysis provides a huge bandwidth of
parameter variation. Although this is necessary because empirical transaction
costs can only be estimated on a very weak base, high variations of all transac-
tion costs by more than a factor of ten seem to be rather unrealistic.

Influence of Single Transaction Cost Parameters

In the next step, we analyze the comparative impact of varying single transaction
cost factors – either cIF, cAD, cAG, or cN.
Figure 149 shows the impact of singular increases in each parameter on rela-
tive global process cost savings (upper diagrams) and relative global net savings
(lower diagrams) for decentralized (left) and centralized (right) coordination. The
left sides of the graphs represent the common initial conditions while each dif-
ferent curve progression represents a discriminative impact of one of the TC
parameters.
In the simulations (left diagrams), an increase of cN (negotia-
tion/coordination costs) has the strongest negative impact while adoption costs,
since they occur only in the first period, have none at all (as in centralized coor-
dination). From a net savings perspective, the effect of coordination costs, inter-
face costs, and agency costs shows to be quite similar over a wide range (with
agency costs having a slightly greater effect) before the impact of cN becomes the
largest, again. On the right side (results from the optimization), the effect of
agency costs clearly dominates.
 346 Analytical and Simulative Studies

90% 90%
relative average process cost savings

80% 80%

relative average process cost savings

(decentralized coordination)

70% 70%

(centralized coordination)
60% 60%

50% 50%

40% 40%

30% 30%
cIF
20% 20% cAD
cAG
10% 10%
cN
0% 0%
0 10 20 30 40 50 0 10 20 30 40 50

factorial change of transaction cost parameters factorial change of transaction cost parameters

90% 90%

80% 80%
(decentralized coordination)
relative average net savings

relative average net savings

70% 70%
(centralized coordination)

60% 60%

50% 50%

40% cIF
40%
cAD
cAG
30% 30% cN

20%
20%

10%
10%
0%
0%
0 10 20 30 40 50
0 10 20 30 40 50
factorial change of transaction cost parameters
factorial change of transaction cost parameters

Figure 149: Variation in process cost savings (upper diagrams) and net savings
(lower diagrams) when altering a single transaction cost factor –
decentralized (left) vs. centralized (right) coordination
It seems that the optimal solution is almost solely influenced by a variation
in agency costs, while the PCS graphs remain constant even when increasing
(either) interface cost, coordination cost, or adoption cost parameters by a factor
of up to 40. Nevertheless, they clearly have an impact on the net savings (lower
right diagram), which is linear because the number of banks in coalitions almost
barely changes. The increase in negotiation costs is compensated by an increase
in coalitions and a decrease in coalition size in order to keep the negotiation and
coordination costs low (no figure). The resulting market concentration is only
shown for processing/servicing (Figure 150)199.

199
In order to not overload this work with simulation results, we decided against displaying all
market structure results (HFact, number of coalitions, number of banks in coalitions) for all busi-
ness functions.
 Analytical and Simulative Studies 347

decentralized coordination centralized coordination

0.5 0.7

0.6
cIF
0.4
actor-based Herfindahl index

actor-based Herfindahl index

cAD
(processing/servicing)

(processing/servicing)
0.5
cAG
0.3 cN
0.4

cIF
0.3
0.2 cAD
0.2 cAG

0.1 cN
0.1

0 0
0 10 20 30 40 50 0 10 20 30 40 50
factorial change of transaction cost parameters factorial change of transaction cost parameters
act
Figure 150: Variation of market concentration (HF , exemplarily for
processing/servicing) when varying a single transaction cost factor –
decentralized (left) vs. centralized (right) coordination
The initial increases in some graphs result from the number of coalitions de-
creasing faster than the number of banks in coalitions. For example, slight in-
creases in cIF under decentralized coordination (left diagram) lead to a drop in
the number of coalitions while the number of cooperatively sourcing banks de-
creases much more gradually. Surprisingly, interface costs, although not func-
tionally related to coalition size, drive the market concentration towards larger
coalitions. This can be explained by a stronger consolidation to a smaller number
of service providers which not only provide one business function but offer a
broader portfolio of services (e.g. all back-office functions). Thus, the interface
costs between those business functions operated by the same insourcer can be
saved, although all are outsourced200.
This effect of the reduction in the number of coalitions is much stronger in
the centralized setting (right diagram). For three out of the four transaction cost
parameters, the market concentration soars with only marginal variations in the
parameters.
Finally, Figure 151 shows the system dynamics results. While the ratio of
switching and outsourcing actions (left chart) shows no structural differences
among the various transaction cost factors, the relative level of backsourcing
initially increases when raising the agency cost factor but drops when raising the
other factors.

200
Cf. (TPI 2007) on “Multi-process BPO“, which describes the same strategy and, interestingly, is
an emerging trend in the global BPO market.
 348 Analytical and Simulative Studies

Cumulative switching / outsourcing actions Cumulative backsourcing / outsourcing actions

2.5 0.4
cIF
cIF
2.0 cAD
cAD
cAG 0.3
cAG
cN

ratio
1.5 cN
ratio

0.2
1.0

0.1
0.5

0.0 0.0
0 10 20 30 40 50 0 10 20 30 40 50
factorial change of transaction cost parameters factorial change of transaction cost parameters

Figure 151: Number of switching and backsourcing actions relative to

outsourcing actions, aggregated over all business functions
Consequently, higher agency costs, which are triggered by the business
neighborhood and therefore by the composition (which parties) of the particular
coalition, will force banks to backsource their business function rather than sim-
ply switching coalitions. Compared with the other transaction cost types, agency
costs are mostly influenced by other firms’ behavior and therefore the behavioral
uncertainty increases with rising agency costs.

Impact of Additional Parameters of the Transaction Cost Function

Several parameters of the transaction cost function had to be determined without
empirical foundation. Hence, the primary aim of this section is to test for the
stability of the previous results.
Parameter D is an exponent in both the negotiation and coordination cost
function and increases these transaction costs with coalition size (cf. section
4.2.3.1.1). Comparable to Fladung (2006), who determined the optimal size of
procurement consortia of academic libraries, this coefficient is altered between
1.0 (i.e. linear impact of coalition size on total negotiation/coordination costs)
and 2.0 (highly progressive impact)201. Figure 152 aggregates the monetary re-
sults (upper diagram) and the market structure effects (lower diagrams) (only
decentralized coordination). Thus far, the simulations have been conducted with
D=1.25 (represented by the vertical line).

201
Please note once more that the progressive impact results for the overall coalition’s negotiation
costs while the average function (i.e. costs per participating bank) declines at least for small coali-
tion sizes.
 Analytical and Simulative Studies 349

70%

60%
average relative PCS, TCp, NS

50%

40%

30%
relative global PCS
20% relative global NS
relative individual NS
10% relative global TCp

0%
1 1.2 1.4 1.6 1.8 2
D
80 14
avg. number of banks in coalitions

avg. number of coalitions

70 12
60
10
50
8
40
6
30
4
20
10 2

0 0
1 1.2 1.4 1.6 1.8 2 1 1.2 1.4 1.6 1.8 2
D D

Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 152: Monetary results, number of coalitions and number of banks in

coalitions when varying D (only decentralized coordination)
The global monetary results decrease moderately when altering D (upper
diagram). Furthermore, the savings are obtained in different ways by increasing
the number of coalitions and thus reducing the coalition size. As a result, the
transaction costs can be kept constant but the savings decrease due to lower
economies of scale. This effect is aggravated by fewer banks being involved in
cooperative sourcing.
Higher values for D do not only lead to smaller coalitions but also to fewer
activities. As consolidation becomes less favorable and switching becomes more
costly, banks will tend to stay in their small coalitions. Figure 153 shows that the
level of switching and backsourcing actions in relation to outsourcing drops
significantly. The banks reduce their searching activities due to higher switching
costs and because the system has been more stable in earlier periods.
350 Analytical and Simulative Studies

2.5
relative switching/outsourcing actions
relative backsourcing/outsourcing actions
2.0

1.5
ratio

1.0

0.5

0.0
1 1.2 1.4 D 1.6 1.8 2
Figure 153: Number of switching and backsourcing actions relative to
outsourcing actions, aggregated over all business functions
Next, we focus on the impact of learning effects. Parameter E introduces
learning effects in outsourcing management, by decreasing negotiation and coor-
dination costs, based on experiences from former outsourcing projects (cf.
Equation 6 on p. 233). In the simulations above, E is set to -.9. For the sensitivity
analyses, it has been varied between -.25 and -1.5, representing smaller and
greater learning effects. One hypothesis is that greater learning effects will en-
courage banks to search longer for their individually optimal sourcing configura-
tion because switching the coalition will become relatively cheaper with each
step. This longer search process should result ex post in higher net savings from
both lower coordination costs and higher process cost savings of finding a supe-
rior sourcing configuration.
By contrast, the sensitivity analysis reveals that PCS, NS, and TCp do not
change much when altering E (no figures). The global NS are constantly around
55% at E between -.4 and -1.5. But, the average individual net savings increase
with rising learning effects – from 28% for E = -.4 to 36% for E= -1.5 (no fig-
ure). A further reduction of learning effects to E = -.0 (no learning effect any-
more) leads to a global NS of 48% and an average individual NS of 25%.
Lower learning effects lead to a slight decrease in cooperatively sourcing
banks. By contrast, the number of coalitions increases sharply, leading to a lower
market concentration (no figures). The latter effect, nevertheless, can be primar-
ily attributed to the progressive impact of coalition size on coordination costs.
Since both factors are multiplicatively interrelated, lower learning effects in-
crease the impact of coalition size on coordination costs. Therefore, the impact of
E is indirect but has similar consequences on the market concentration as D. If
 Analytical and Simulative Studies 351

the elasticities of both parameters are compared (Table 63, p. 352), the same
signs show up in most cases, with E having a significantly weaker impact.
While the second part of the hypothesis formulated above (positive impact
of stronger learning effects on NS) must be disregarded at least for the given
parameter setting (although varying negotiation/coordination costs has the
strongest impact on NS and thus on the cooperative sourcing decision calculus,
cf. Figure 149 – left), the question remains unanswered as to whether more and
longer switching activities take place when learning effects are greater. Higher
learning effects should lead to a more dynamic system which represents a more
volatile decision environment and thus higher behavioral uncertainty from the
individual’s perspective.
As Figure 154 (left chart) shows, there is, in fact, a significant influence of
learning effects. The number of switching activities (shown exemplarily but
representatively for processing/servicing) increases much more strongly than the
number of outsourcing actions, when reducing the learning effect parameter (i.e.
raising the impact of learning effects on transaction costs).
180 200
periods needed to reach stationary

160 outsourcing actions 180

(processing/servicing function)

switching actions 160

140
backsourcing actions
number of actions

140
120
120
100
state

100
80
80
60 60
40 40
20 20
0 0
-0.25 -0.5 -0.75 -1 -1.25 -1.5 -0.25 -0.5 -0.75 -1 -1.25 -1.5
E E

Figure 154: Average number of outsourcing, switching, and backsourcing

actions in regards to processing/servicing (left) and avg. number of
periods required to reach the stationary state (right) when altering E
Furthermore, stronger learning effects lead to a longer timeframe of system
dynamics, as proposed (Figure 154 – right). Interestingly is, however, that the
number of periods prior to reaching the stationary state only increases to a cer-
tain point and then slightly decreases. One explanation is that learning effects are
realized very quickly with a highly negative E, i.e. negotiation and coordination
costs drop significantly even after only one or two completed outsourcing pro-
jects. This leads to earlier and to more simultaneous actions and thus reduces the
overall time of system dynamics.
As a third parameter, the impact of J is analyzed. J is part of the coordina-
tion cost function and describes the ratio of periodical coordination costs to setup
negotiation costs. A variation of J between .0 and .5 does not lead to a change in
352 Analytical and Simulative Studies

the monetary values, as can be seen with the very low elasticities given in Table
63 (the original parameter setting was J=.3). Since the coalition size has an im-
pact on negotiation and coordination costs, the increased coordination costs are
compensated for by forming smaller (and more) coalitions (no figures). The level
of dynamics is not affected by altering J within this range.
In summary, Table 63 compares the elasticities of the monetary and market
structure results from varying D, E, and J.
Rel. monetary results Avg. number of banks in coalitions Average number of coalitions
Ind.
PCS NS TCp k=1 k=2 k=3 k=4 k=5 k=1 k=2 k=3 k=4 k=5
NS
D -.44 -.46 -.96 -.35 -1.05 .39 -.30 -.23 -.36 -.61 .57 2.09 1.70 .90
E -.02 -.03 -.19 .02 .38 -.19 -.10 -.13 .01 -.87 .04 .91 .33 .63
J -.03 -.03 -.04 .03 .31 .25 -.07 -.01 -.05 .15 .35 .25 .07 .20
Table 63: Elasticities resulting from variation of D, E, and J
Apart from the previously investigated parameters, the task complexity
measure Fik is part of some of the transaction cost functions. In the following, the
consequence of the assumption that task complexity is a source of transaction
costs will be tested. The results of the overall transaction cost parameter varia-
tion (alter_TCcoefficients) above will be compared with similar simulations
without considering task complexity in the transaction cost functions202. Figure
155 shows the differences (in percentage points) in monetary results between the
scenario without taking task complexity into consideration and the original sce-
nario above.
Disregarding task complexity basically leads to no significant differences in
the original level of transaction costs (alter_TCcoefficients = 1.0). If the transac-
tion cost parameters are sharply increased, additional net savings of up to 13
percentage points could be reached. Consequently, a more homogenous business
process structure than estimated from empirical data or a lower (than assumed)
influence of task complexity on transaction costs in the formal model structure
would have a positive monetary impact. Therefore, the model development and
parameterization can be interpreted as a “careful” drawing of reality in regard to
this issue of process complexity (which is hardly operationalizable and measur-
able in reality).

202
Fik has been set to a constant average value for all banks i and business functions k.
Analytical and Simulative Studies 353

14%
Difference of rel. PCS
difference between scenario without and with

12%
Difference of rel. NS
task complexity [in percentage points]

Difference of rel. TCp

10%

8%

6%

4%

2%

0%
0 10 20 30 40 50
-2%
alter_TCcoefficients
Figure 155: Difference between scenario without and with considering task
complexity (only decentralized coordination) in average PCS, NS,
and TCp
The corresponding market concentration shows almost no changes at all.
The delta of the Herfindahl index is less than .03 for processing/servicing and
less than .015 for the remaining business functions within the whole range of
alter_TCcoefficients. The additional net savings are obtained by slightly more
banks joining coalitions and by a corresponding larger number of coalitions
themselves.
In conclusion, the following results can be derived from this section:
o The simulations and the centralized coordination results are sensitive to
increasing transaction costs. While the global savings are almost linearly re-
lated to the level of transaction cost factors, the individual savings drop
much more significantly. The relative frequency of coalition changes falls in
a similar manner.
o The centralized coordination accepts higher transaction costs up to a certain
point without reducing the number of cooperatively sourcing banks. As a
consequence, the process cost savings can be kept constant, but the number
of actors that would achieve a financial loss increases.
o The monetary results are most strongly affected by cN (negotia-
tion/coordination costs) in the decentralized setting and by cAG (agency
costs) under centralized coordination. Higher cIF (interface costs) leads to
higher consolidation tendencies and to more sales banks keeping interface
costs low because sales banks can receive an integrated portfolio of back-
354 Analytical and Simulative Studies

office functions from a single provider. A slight to medium increase in cAG

leads to higher behavioral uncertainty for a certain range and thus to a dis-
proportionate increase of backsourcing activities.
o Altering D has a slight negative impact on the monetary results and a sig-
nificantly negative impact on the coalition size because it disproportionately
increases negotiation and coordination costs along with increasing coalition
size. Increasing these costs also significantly reduces the system dynamics.
o The impact of learning effects (ß) on the monetary results is marginal. How-
ever, greater learning effects lead to higher dynamics in the system. Banks
change coalitions more often, in order to find their local optimum. Conse-
quently, learning effects have primarily negative consequences from an
overall system perspective.
o The impact of J (ratio between coordination and negotiation costs) on the
monetary results is marginal. Higher coordination costs lead to smaller but
more coalitions.
o Disregarding the task complexity has a profound and entirely positive im-
pact on the monetary results. Therefore, taking task complexity into account
leads to a more cautious analysis of effects resulting from the system behav-
ior.

5.3.4.4 Variation of Additional Decision Calculus Parameters

In the previous section, we analyzed the system behavior in regards to process
cost parameters and transaction cost parameters. There are some more parame-
ters which directly influence the decision calculus: first, the risk-adjusted dis-
count rate (rad), second, the value-added tax (VAT), third, the strategic con-
straint (SC), and finally, the heterogeneity of all of the parameter distributions
(variation coefficients, vc).

Impact of the Risk-Adjusted Discount Rate

The effect of rad clearly depends on both the contract duration (TCoSo) and the
setup transaction costs in the transfer period (negotiation costs CN and adoption
costs CAD). Since an examination of the resulting effect of simultaneously alter-
ing three parameters would be too complex for a comprehensible presentation,
the analysis has been limited to the interplay of rad and TCoSo at two different
levels of the transaction cost basis (original level and tenfold increased level) in
the following diagrams.
The calculation discount rate is adjusted between rad=[.00, .10] while the
original setting was rad=.05. As Figure 156 shows, the global relative net sav-
ings are not influenced by a change in the discount rate within this range.
Analytical and Simulative Studies 355

 
 
 

 
    
relative global net savings (NS)

relative global net savings (NS)

 
     
 

high transaction costs

 
  
 

  

0.50

0.50  


 



 



0.40 0.40


0.30 0.30 





   
  
  
  
 
0.20 0.20   

 
 
 
 


0.10 0.10  

low transaction costs

 




0.00 



0.02 10 
10
0.02
0.04 8 0.04
 
 8

0.06 6 0.06 6
rad 0.08
0.10 2
4
TCoSo rad 0.08
0.10 2
4
TCoSo
Figure 156: Impact of rad and TCoSo on relative global net savings (NS) with
different levels of transaction costs (alter_TCcoefficients = 1 (left)
and = 10 (right))203
At tenfold transaction costs in the right diagram (alter_TCcoefficients=10), a
contract of short duration leads to low or no cooperative sourcing activities,
because the initial transaction costs cannot be amortized sufficiently. However, a
variation of rad has no impact. Corresponding results can be found for the mar-
ket effects. Market concentration, as well as the number of banks in coalitions,
does not significantly change when altering rad (no figures). Consequently, the
simulation results in the previous chapters can be evaluated as stable in regards
to the discount rate parameterization.

Impact of Value-Added Tax

Value-added tax (VAT) is often argued by practitioners to be a major outsourc-
ing inhibitor for banks because, in their view, it presents direct additional trans-
action costs related with procuring the formerly insourced services from an ex-
ternal vendor. Since the impact of VAT largely depends on the process cost
structure and the possible outsourcing savings resulting thereof, in the following,
the impact of VAT is analyzed based on an alteration of the process cost struc-
ture (alter_costRatio).
First, we investigate how VAT reduces savings and activities when the ratio
between fixed and variable costs is changed (alter_costRatio, cf. section 5.3.4.2).
As Figure 157 (left) shows, the impact of VAT=19% on global and average
individual net savings is rather low.

203
Each data point in the charts represents the average of 100 simulation runs.
 356 Analytical and Simulative Studies

70% 90

80
60%
70

number of actors in coalitions

50%
60
relative NS

40% 50

40
30%
30
20% rel. avg. global NS (VAT=0%) processing/servicing (VAT=0%)
rel. avg. global NS (VAT=19%) 20 processing/servicing (VAT=19%)
risk monitoring (VAT=0%)
rel. avg. individual NS (VAT=0%)
10% risk monitoring (VAT=19%)
rel. avg. individual NS (VAT=19%) 10
workout (VAT=0%)
workout (VAT=19%)
0% 0
0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75 0.25 0.75 1.25 1.75 2.25 2.75 3.25 3.75
alter_costRatio alter_costRatio

Figure 157: Impact of VAT on net savings (left) and number of banks in
coalitions (right) at different levels of alter_costRatio
Depending on the level of alter_costRatio, the relative drecrease is between
.5% and 8.7% for global net savings and between 2.5% and 11.5% for individual
net savings. Correspondingly, the number of banks involved in cooperative
sourcing (shown for the back-office functions in the right diagram) decreases by
less than 13% for all business functions. The higher the fraction of fixed costs is
(i.e. high alter_costRatio), the higher are the individual savings and the less the
taxes reduce cooperative sourcing activities and benefits.
In a second analysis, the fixed/variable cost ratios are again gradually ho-
mogenized, i.e. the ratios of the different banks move closer to the mean (ho-
mogenize_costRatio). In section 5.3.4.2, it was already stated that this homogeni-
zation procedure has no impact on both monetary results and market structure.
Figure 158 shows that this also applies to a scenario with VAT. Although all
firms have similar fixed/variable cost ratios (but not the same absolute process
costs), the relatively small impact of VAT, which is shown above, remains con-
stant and leads to a quite consistent reduction of around 5% for global savings
and around 9% for individual net savings on average.
60% 80

70
50%
number of banks in coalitions

60
40%
50
relative NS

30% 40

30
20%
processing/servicing (VAT=0%)
rel. avg. global NS (VAT=0%) 20 processing/servicing (VAT=19%)
rel. avg. global NS (VAT=19%) risk monitoring (VAT=0%)
10%
rel. avg. individual NS (VAT=0%) 10 risk monitoring (VAT=19%)
rel. avg. individual NS (VAT=19%) workout (VAT=0%)
workout (VAT=19%)
0% 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
homogenize_costRatio
homogenize_costRatio

Figure 158: Impact of VAT on net savings (left) and number of banks in
coalitions (right) for different levels of homogenize_costRatio
Analytical and Simulative Studies 357

The reason for this rather minor impact from introducing a value-added tax
lies primarily in the rather high level of heterogeneity in total process costs for
the banks. If a bank has high process costs and gets rid of them by outsourcing,
the savings will be high enough that VAT will not affect the decision. The VAT
has been critical for only a minority and those firms do not decide to outsource
anymore. Banks with more beneficial cost structures did not even outsource
before introducing VAT. Moreover, a certain number of the cooperatively sourc-
ing banks takes over the role of the insourcer. Consequently, this group does not
pay VAT at all and further reduces the average impact of VAT.

Impact of the Strategic Constraint

The strategic constraint incorporates the RBV suggestion that strategically rele-
vant business functions (covering the related resources and capabilities) should
not be outsourced. Every business function of each bank bears a core compe-
tence value Oik between 0 and 1 (based on the empirical data) which expresses
the strategic relevance. Thus far, the strategic constraint has been set to a rather
high value of SCi = 3 for all banks. In the following figure, it is reduced in steps
of .25 to 1.5204. Figure 159 shows the results of the global net savings from de-
centralized and centralized coordination.
80%

70%

60%
relative global net savings

50%

40%

30%

20%

relative NS from centralized coordination

10%
relative NS from decentralized coordination
0%
3 2.75 2.5 2.25 2 1.75 1.5
SC
Figure 159: Decrease in global net savings while reducing SC
As one would anticipate, the monetary results of the restriction are quite lin-
ear. However, the centralized coordination reacts less, in relative terms, com-
pared with the decentralized coordinated scenario. While the lower graph drops

204
Consequently, a reverse abscissa scale was chosen in the following charts. 150 simulation runs
have been conducted at each step of the reduction.
 358 Analytical and Simulative Studies

by 54%205, net savings resulting from centralized coordination decrease only by

35%.
While banks themselves simply reduce their cooperative sourcing activities,
the optimization algorithm partially compensates for the constraint by establish-
ing more and smaller numbers of coalitions because the coalition insourcer is
defined as not outsourcing its business function. Thus, the insourcer can join a
coalition without losing core competencies. The following figure demonstrates
this effect by focusing solely on the example of processing/servicing.
0.50
decentralized coordination
centralized coordination
0.40
actor-based Herfindahl index

0.30

0.20

0.10

0.00
3 2.75 2.5 2.25 2 1.75 1.5
SC
100 20
decentralized coordination decentralized coordination
90 18
centralized coordination centralized coordination
80 16
number of banks in coalitions

70 14
number of coalitions

60 12

50 10

40 8

30 6

20 4

2
10
0
0
3 2.75 2.5 2.25 2 1.75 1.5
3 2.75 2.5 2.25 2 1.75 1.5 SC
SC

act
Figure 160: HF , number of banks in coalitions and of coalitions for pro-
cessing/servicing for both coordination schemes while reducing SC
The market concentration decreases faster under centralized coordination
(upper chart), but not primarily because banks do not cooperatively source any-
more (lower left), but because the coalitions are split into smaller ones (lower
right). This helps more actors to stay in coalitions than under decentralized coor-

205
Although not displayed, the average individual savings decreased by the same factor from 34% to
16%.
Analytical and Simulative Studies 359

dination206. Structurally similar pictures exist for the remaining business func-
tions.

Impact of Parameter Heterogeneity

For many parameters, the simulations assume a normal distribution with a certain
variation coefficient vc (i.e. ratio between standard deviation and mean). While it
can sometimes be derived from the empirical data, this is not possible in every
case and thus leads to the choice of an initial value of vc = .1207.
Following, all originally set variation coefficients are simultaneously altered
by a multiplicative factor alter_vc = [0.0, 2.0]208 to test for their impact on both
position and deviation of the basic results. While the minimum value of alter_vc
= 0.0 represents perfect homogeneity for all parameters which are assumed to be
normally distributed, alter_vc = 2.0 represents a doubling of the heterogeneity of
the hitherto simulations.
The following figure shows that the global monetary results are positively
influenced by varying the overall range of parameter variation.
80% 7%
standard deviation of relative monetary results

70% 6%
average relative monetary results

60%
(PCS, NS, TCp) [% points]

5%
(PCS, NS, TCp)

50%
4%
40%
3%
30% rel. global PCS
rel. global PCS
rel. global NS
rel. global NS 2%
20% rel. individual NS
rel. individual NS rel. global TCp
10% 1%
rel. global TCp

0% 0%
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
alter_vc alter_vc

Figure 161: Mean (left) and standard deviation (right) of global monetary results
(PCS, NS, TCp) when varying all variation coefficients
simultaneously by factor alter_vc
A reduction in the standard deviation has a slightly greater impact than an
increase of the standard deviation. The standard deviation increases proportion-
ally to the mean so that the variation coefficient of the results remains almost
constant at .1. The lower impact of increasing the variation, in comparison to
decreasing it, shows that the original scenario already had a high internal hetero-
geneity. Nevertheless, if we consider that alter_vc = .0 represents not any varia-
tion in the affected parameters, the monetary results do not react significantly.

206
Thereby, the proportion of banks which confront negative monetary consequences of following
the centralized coordination only increases from 10% to 16%.
207
The relevant parameters can be found in the Appendix A2.
208
In steps of .05, with 150 simulation runs for each step.
360 Analytical and Simulative Studies

Corresponding results can be found for the market structure (no figures).
While the market concentration decreases slightly for processing/servicing and
workout, an opposite (and greater) effect can be seen for assessment/decision.
The impact on risk monitoring is non-existent. The elasticities are provided in
the following table.
Rel. monetary results Actor-based Herfindahl index Average number of banks in coalitions
PCS NS Ind. TCp k=1 k=2 k=3 k=4 k=5 k=1 k=2 k=3 k=4 k=5
NS
.27 .30 .31 -.13 .13 .47 -.20 .01 -.11 .03 .64 -.07 -.01 .01
Table 64: Elasticities resulting from varying variation coefficients
The system dynamics are only slightly affected (no figure). The ratios of
switching, backsourcing, and outsourcing actions decrease only marginally with
higher heterogeneity.
Consequently, it can be argued that the selected size of the model instance
proves to be large enough to attain the random number seed derived from the
empirical foundation. A simulated system size of 100 banks balances, at least,
the degree of heterogeneity we tested for, and delivers inherently stable results.
Areas in which only few outsourcing activities are present (assessment/decision
in particular) react more sensitively to more heterogeneity because more extreme
values of the cost parameters will cause it to overcome the start-up problem more
easily. If there is one bank with an outstanding advantageous cost structure, the
establishment of a cooperative sourcing coalition will be more likely.
In summary, this section revealed the following findings:
o The discount rate rad, considered in the actors’ decision calculus, has no
impact on the simulation results (monetary as well as market structure).
o Only at high transaction costs does the contract duration TCoSo have at least
some impact on the results, because high setup transaction costs for negotia-
tion and adoption can be amortized only if the contract duration is suffi-
ciently long.
o Due to rather high process cost heterogeneity, an introduction of value-
added tax (VAT) does not lead to a strong reduction of cooperative sourcing
activities and resulting net savings.
o The impact of a more restrictive strategic constraint which hinders outsourc-
ing of strategically important business functions differs in the opposite coor-
dination mechanisms. While autonomous banks simply reduce cooperative
sourcing activities, the centralized coordination restructures the system to-
ward more and smaller-sized coalitions where more banks can act as in-
 Analytical and Simulative Studies 361

sourcers, thus keeping their strategically important business functions in-

house.
o Varying the spread of the random number seed used in the simulations leads
to only moderate changes in the results. The simulated system seems to have
the necessary size in order to deliver inherently stable results.

5.3.4.5 Impact of Demographic Properties

In this section, the basic results are tested for stability regarding demographic
parameters of the model such as number of banks in the overall system, business
neighborhood, or heterogeneity in process volumes.

Impact of System Size

For computational reasons the number of banks simultaneously simulated in a
closed system was set to I = 100. The question is whether or not the results re-
main stable when altering this number. For the following analysis, I is altered in
intervals of 20, from 40 to 160209. This ensures that all scenarios show the same
distribution of firm sizes.
The following figures show the average results. In the left diagram of Figure
162, it can be seen that the global and average individual monetary results do not
change at all when altering I. The vertical line again represents the original set-
ting with 100 banks. Nevertheless, the market concentration, measured by the
actor-based Herfindahl index, varies significantly (right diagram).
70% 0.40

60% 0.35
relative monetary results

actor based Herfindahl index

0.30
50%
0.25
40%
0.20
30%
relative global PCS 0.15
20% rel. global TC
rel. global NS 0.10
10% rel. individual NS 0.05
0% 0.00
40 60 80 100 120 140 160 40 60 80 100 120 140 160
I I
Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 162: Effect of system size (number of banks) on monetary results (left)
and market concentration (right)

209
There are 200 simulation runs for each setting. The runtime (T) was increased from T=250 to
T=400 for the largest scenario in order to take into account longer search activities.
 362 Analytical and Simulative Studies

As the left diagram of Figure 163 shows, this is not caused by a decreasing
number of actors participating in cooperative sourcing activities. This relative
number remains quite stable for all business functions.
80% 0.08
rel. number of banks in coalitions

normalized number of coalitions

70% 0.07

60% 0.06

50% 0.05

40% 0.04

30% 0.03

20% 0.02

10% 0.01

0% 0
40 60 80 100 120 140 160 40 60 80 100 120 140 160

I I

Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 163: Effect of system size (number of banks) on market structure

Instead, a first reason for this effect, partially, is an increase in the number of
existing coalitions. In the right diagram, the number of coalitions for each busi-
ness function is displayed on a normalized scale, i.e. the number of coalitions is
divided by the number of banks in order to neutralize the effect from an increas-
ing system size. Interestingly, the relative number of coalitions increases for
processing/servicing and also slightly for risk monitoring but decreases for
workout and assessment/decision.
A second reason, which is primarily responsible for the reduction of market
concentration, in larger systems (I > 100), in particular, is the decreasing hetero-
geneity of coalition size. The market shares of the largest (CR1) and three largest
(CR3) coalitions decline, more or less, for all back-office functions (Figure 164).
70% 0.40

60% 0.35
relative monetary results

actor-based Herfindahl index

0.30
50%
0.25
40%
0.20
30%
relative global PCS 0.15
20% rel. global TC
rel. global NS 0.10
10% rel. individual NS 0.05
0% 0.00
40 60 80 100 120 140 160 40 60 80 100 120 140 160
I I
Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 164: Effect of system size (number of banks) on the market share of the
largest coalition (left) and the three largest (right) coalitions
Analytical and Simulative Studies 363

This can be explained once more by the basic heterogeneity in firm sizes.
Since economies or scale decrease only disproportionately with larger coalitions,
very large coalitions (in absolute sizes) will become unfavorable due to the
higher coordination costs and the only minimal additional economies of scale.
Thus, the relative size of the largest coalitions will drop.
The third dimension of analysis focuses on the dynamics which provide an
estimation of the behavioral uncertainty from the actors’ perspective. The rela-
tive number of switching actions relative to outsourcing activities almost doubles
when quadrupling the system size from I=40 to I=160 (no figure). A larger sys-
tem leads to a higher absolute number of coalitions, which in turn leads to more
opportunities for the actors and thus to higher dynamics (as well as higher behav-
ioral uncertainty).

Impact of Business Neighborhood

Another demographic property, considered in the model, is the business
neighborhood. Although the simulations abstracted from the sectorization, which
is typical in Germany, the model is able to take competitive relationships into
account. In the simulations, these were parameterized based on firm sizes (cf.
section 5.3.3.1). In the following comparative simulations, the business
neighborhood is “disabled”, representing a solely cooperative environment with-
out any competition between the banks in the system. In Figure 165, the transac-
tion costs are varied (analogous to section 5.3.4.3) in the original parameter set-
ting and in an alterative scenario without taking the business neighborhood into
account (bnij = .0 for all inter-bank relationships). Since bn determines the level
of agency costs it is important to vary the transaction cost parameters as well, in
order to measure the trade-off.
60%
relative global net savings (bn = .0)
relative global and individual NS

relative global net savings (bn > .0)

50%
relative individual net savings (bn = .0)
relative individual net savings (bn > .0)
40%

30%

20%

10%

0%
1 6 11 16
alter_TCcoefficients
Figure 165: Global and indiv. savings with and without business neighborhood
 364 Analytical and Simulative Studies

The figure shows that the business neighborhood has a visible impact on the
monetary results (global and individual NS). In the initial situation (al-
ter_TCcoefficients = 1.0 on the left margin of the diagram), the exclusion of
business neighborhoods increases the net savings by about 3.5%. This difference
increases up to 145% for alter_TCcoefficients = 14.
There appear to be strong differences between the different business func-
tions. As the results in Figure 166 show, the market concentration reacts greatly
to risk monitoring (right chart) and workout (no figure), but only weakly to proc-
essing/servicing (left). In the original situation (alter_TCcoefficients = 1.0), the
relative difference is 65.6% (risk monitoring), 39.4% (workout), and 4.1% (proc-
essing/servicing).
processing/servicing risk monitoring
0.25 0.25
bn = .0 bn = .0
actor-based Herfindahl index

actor-based Herfindahl index

bn > .0 bn > .0
0.20 0.20

0.15 0.15

0.10 0.10

0.05 0.05

0.00 0.00
1 6 11 16 1 6 11 16
alter_TCcoefficients alter_TCcoefficients

Figure 166: Market concentration for processing/servicing and risk monitoring

with and without business neighborhood
Although the market concentration shows structural differences, neither the
number of coalitions nor the number of banks in coalitions differs significantly
between the scenarios with and without including business neighborhood (no
figures). Instead, the differing market concentration results from a much greater
heterogeneity in coalition size. While the largest coalition (CR1) for risk monitor-
ing consists of 29 banks, on average, in the original setting, it shoots up to 42
banks when rejecting business neighborhood. By contrast, the largest coalition
for processing/servicing does not change at all, and for workout, it increases
from 27 to 33. One possible explanation is the specific process cost structure of
risk monitoring, which consists of more fixed costs than that of the other busi-
ness functions. Since business neighborhood increases in line with the degree of
competition between the coalition members (and therefore in line with the coali-
tion size in a statistical meaning), fewer large coalitions are found in competitive
environments than in competition-free environments. For example, this confirms,
the empirical observation that cooperative sourcing scenarios are older and larger
in the public savings and the cooperative sector where almost no competition
between the sector members existed in the past.
Analytical and Simulative Studies 365

There is another observation that supports this finding. In Figure 167, the
proxy for behavioral uncertainty is displayed (number of switching and back-
sourcing actions relative to outsourcing actions).
1.8
cum. switching/outsourcing actions (bn > .0)
1.6 cum. switching/outsourcing actions (bn = .0)
1.4 cum. backsourcing/outsourcing actions (bn > .0)
cum. backsourcing/outsourcing actions (bn = .0)
1.2

1.0
ratio

0.8

0.6

0.4

0.2

0.0
1 6 11 16
alter_TCcoefficients
Figure 167: Number of switching and backsourcing actions relative to
outsourcing actions, aggregated over all business functions
Excluding the business neighborhood leads to a massive decline in these
measures. Dynamics of revising cooperative sourcing decisions are significantly
reduced, leading to a more stable environment and thus a more stable decision
base for the individual bank, which is consequently less likely to cancel a coali-
tion membership in order to search for more beneficial cooperative sourcing
opportunities.

Impact of Heterogeneity in Firm Sizes

Another demographic parameter which may be significantly responsible for the
simulation results is the huge heterogeneity in firm size, which determines the
process volumes in our model. Since the sample underlying the parameterization
setting consists of the 1,000 largest German banks, there are obviously large
deviations. Hence, in the following, we analyze how the results change when the
heterogeneity is reduced. Similar to the homogenization of the seed of process
cost ratios (cf. section 5.3.4.2), the process volume deciles gradually converge
toward the mean, by applying a parameter homogenize_procVol 210. This parame-

210
Because of the high asymmetry of the process volume distribution across the banks, we did not
choose the median as a reference value, but rather the mean. This has the additional benefit that
the overall number of loans being processed and administered in the system remains statistically
 366 Analytical and Simulative Studies

ter is equal to 0 in the original setting and equal to 1 if all deciles are similar to
the original mean (i.e. all banks have the same process volume in a particular
business function).
70% 0.5

60%
0.4

actor-based Herfindahl index

relative monetary results

50%

0.3
40%

30% relative global PCS

0.2
relative global NS
20% relative individual NS
relative global TC 0.1
10%

0% 0.0
0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
homogenize_procVol homogenize_procVol
Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 168: Monetary results and market concentration effects when

homogenizing the process volumes across the banks
In the left diagram of Figure 168, it can be seen that the global monetary re-
sults increase only slightly while the average individual net savings rise strongly
from 35% to 56% when equalizing the banks’ process volumes. As the right
diagram shows, this is caused by a large increase in market concentration for
most of the business functions. Here, the greatest relative increase takes place for
workout and assessment/decision. Additionally, it should be noted that, with
increasing homogenization, the displayed actor-based Herfindahl index becomes
similar to the volume-based market concentration, which was additionally meas-
ured in the basic setting (cf. Figure 116, p. 307).
The following figure shows the corresponding market structure effects.

constant. The distances between deciles and means were multiplicatively reduced by homoge-
nize_procVol in steps of .25 with 200 simulation runs for each step.
 Analytical and Simulative Studies 367

100 10

90 9
number of actors in coalitions

80 8

number of coalitions
70 7

60 6

50 5

40 4

30 3

20 2

10 1

0 0
0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1
homogenize_procVol homogenize_procVol
Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 169: Number of actors in coalitions and number of coalitions in the

system when homogenizing the process volumes across the banks
As can be seen in the left diagram, the number of cooperatively sourcing
banks increases for workout and assessment/decision, while it just initially in-
creases and then remains rather constant for the frequently cooperatively sourced
processing/servicing and risk monitoring. This increase results in a “temporary”
increase of the numbers of coalitions (right diagram) which drops significantly
with stronger homogenization. Finally, when each bank has identical process
volumes, the number of coalitions reduces to exactly one for each business func-
tion. This explains the strongly rising market concentration shown above.
Finally, regarding the third dimension of analysis, the following diagram
shows the corresponding dynamics of switching and backsourcing (in relation to
outsourcing activities).
5
cum. switching/outsourcing actions
com. backsourcing/outsourcing actions
4

3
ratio

0
0 0.25 0.5 0.75 1
homogenize_procVol

Figure 170: Number of switching and backsourcing actions relative to

outsourcing actions, aggregated over all business functions
368 Analytical and Simulative Studies

In particular, the relative number of switching actions increases greatly with

more homogeneous process volumes, up to a certain point, and then drops again
in the situation of equal process volumes of all banks. The more equal the proc-
ess volumes are, the more active the banks are in searching for their individual
optimal coalition, i.e. there are high possibilities of changing the coalition and,
therefore, high uncertainty, due to the very volatile environment. As shown
above, these high levels of switching dynamics lead to a strong consolidation of
the system. By contrast, if all banks have identical pre-conditions (homoge-
nize_procVol), this effect will be reversed and all banks will move directly to-
wards the final resulting optimal (and only surviving) coalition.
With this considerable effect of volume homogenization on the system be-
havior we can conclude that an asymmetrical volume distribution leads to less
cooperative sourcing activity and to less consolidation. For large banks with high
process volume, cooperative sourcing will often not generate enough additional
economies of scale, while the small banks would have to bear too high relative
transaction costs in cases where they have already optimized their internal cost
structure. Although these arguments have their sources in production cost eco-
nomics and transaction cost economics, the effect is also in accordance with
newer findings of the network effect theory, which argue that returns from using
the same good (or, here, joining the same coalition) do not constantly increase
with every new participant.
The following results can be summarized:
o The monetary results are stable in regards to the number of simulated firms
(I). In larger systems, market concentration will decrease slightly due to less
heterogeneity in the coalition sizes but not to a larger number of coalitions.
Furthermore, larger systems lead to higher dynamics (i.e. stronger coalition
switching activities).
o The business neighborhood (bn) has a significant influence on the results,
leading to lower market concentration for certain business functions (par-
ticularly risk monitoring and workout). In the absence of competitive ele-
ments between the banks, the resulting coalitions for these particular busi-
ness functions would be noticeably larger. Moreover, the dynamics of
switching and backsourcing are strongly reduced which leads to a more sta-
ble coalition structure and thus a more stable decision base for the individual
banks.
o A system with less heterogeneity in firm size leads to higher relative indi-
vidual savings from cooperative sourcing, greater dynamics and stronger
market consolidation. If the pre-conditions for all banks are identical in
terms of process volume, the market consolidates very quickly into a single
Analytical and Simulative Studies 369

coalition. In the original scenario with high asymmetries, the perceived co-
operative sourcing benefits for both very large and very small firms are
lower than in a virtually homogenized world.

5.3.4.6 Impact of Simulation Control Parameters

The simulation procedures are controlled by a few parameters which may have
an impact on the simulation results. The starting point for a new coalition is
always a central mechanism (not to be mistaken for centralized coordination),
which randomly chooses banks to form a coalition. Nevertheless, the coalition
will be created only if all banks agree to it. This mechanism is controlled by two
parameters: the first determines the initial coalition size (initCoalSize) while the
second determines the maximum number of new coalitions created in a certain
time period (maxCoalBuilding).

Effect of Initial Coalition Size

The simulations above have been conducted with initCoalSize = 9. This does not
mean that every coalition starts with nine members. When forming a coalition,
all potential members will evaluate the advantageousness of joining it. If one
bank decides against it, the evaluation will be repeated with the remaining banks.
Only if all partners agree with the coalition, will the coalition actually be estab-
lished. Thus, the resulting initial size is always d initCoalSize. In the following,
initCoalSize is varied from 4 to 19211. Figure 171 shows the monetary and market
structure results after 200 periods. The previously used value of initCoalSize=9
is marked by the vertical line.
While the monetary results (upper diagram) are not influenced, the numbers
of banks for some business functions (processing/servicing and risk monitoring)
increase slightly to initCoalSize=9 before they fall slightly again (lower left dia-
gram). The numbers of coalitions show similar but larger peak curves (lower
right). While the upward part of the concave graphs can be explained specifically
by the increasing initial coalition size, which leads to more coalitions surviving
until T = 200, the downward side results from coalitions becoming larger and,
given a quite stable number of banks, less frequent. As a consequence, the mar-
ket concentration almost remains stable (no diagram).

211
In intervals of one, with 150 simulation runs for each step.
 370 Analytical and Simulative Studies

70%

60%

and average individual NS

relative PCS, NS, TCp, 50%

40%

30%
PCS
20% TCp
NS
10% avg. ind. NS

0%
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
initCoalSize

80 8

70 7
number of banks in coalitions

60 6
number of coalitions

50 5

40 4

30 3

20 2

10 1

0 0
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
initCoalSize initCoalSize
Sales/preparation Assessm./decision Processing/servicing Risk monitoring Workout

Figure 171: Variation of monetary and market structure results from altering the
initial coalition size initCoalSize (results from T = 200)
The level of dynamics, as the third dimension of analysis, remains constant
as well (no figure). The relative number of switching and backsourcing actions to
outsourcing actions does not change significantly. Consequently, this simulation
control parameter does not affect system dynamics and behavioral uncertainty
from the individuals’ perspective.

Effect of Maximum Number of Initial Coalitions

The second parameter, maxCoalBuilding, was originally set at a low value of 4.
Remember that this means that the simulation control will try to establish four
coalitions in the first period (per business function), while the number is reduced
to the half for the following periods212. In the following, this parameter is in-
creased to 20.
The simulations reveal that the level of maxCoalBuilding has no impact on
the final monetary results and market concentration (HFact) after 200 simulation
periods (no figures). A sufficient number of periods obviously clearly lead to

212
Moreover, the coalition building procedure is applied only to every 5th period (section 5.3.3.4).
 Analytical and Simulative Studies 371

neutralizing any delayed effect from less initial coalitions. Differences appear
only in very early periods (t = 1–20). This indicates that, when maxCoalBuilding
is increased to high values, the actual number of coalitions does not increase
beyond a certain level because the number of banks willing to join a coalition (at
least for the given initial coalition size) remains constant.
A fascinating result is found in the number of coalitions for risk monitoring
over time. Above, we discovered structural differences in the progression of the
number of coalitions between processing/servicing (which showed an over-
reaction with subsequent decrease) and risk monitoring (which showed a mo-
notonously increasing path). As Figure 172 (right diagram) shows, the same
result of processing/servicing can also be found in risk monitoring, if the level of
maxCoalBuilding is increased to values higher than 10. Furthermore, in contrast
to processing/servicing, the “over-reaction” occurs in the very first periods while
it takes more time for processing/servicing (Figure 172 – left diagram). How-
ever, even for the latter, the effect is accelerated and aggravated by higher max-
CoalBuilding.
Processing/servicing Risk monitoring
10 10
9 9
8 8
7
num ber of coalitions

7
number of coalitions

6
6
5
5
4
maxCoalBuilding = 20 4
maxCoalBuilding = 20
3 maxCoalBuilding = 16
3 maxCoalBuilding = 16
2 maxCoalBuilding = 12
2 maxCoalBuilding = 12
maxCoalBuilding = 8
1 maxCoalBuilding = 8
maxCoalBuilding = 4 1
maxCoalBuilding = 4
0
0
0 50 100 150 200
t 0 50 100 150 200
t

Figure 172: Number of coalitions for processing/servicing and risk monitoring

over time, while varying the maxCoalBuilding parameter
After dropping slightly, the risk monitoring graphs show another slight in-
crease, which is structurally equivalent to the first results. Consequently, the
increase of maxCoalBuilding triggers a short-term over-reaction213 which is ad-
justed by consolidations. In later periods, the slight increase in actors in coali-
tions (no figure but equivalent to the results in section 5.3.4.1) leads to an in-
crease in the number of coalitions, once again. However, over the entire lifetime
of the system, the dynamics (number of outsourcing, switching, and backsourc-
ing actions) do not differ among the different scenarios.

213
The diagrams depict average values. In single runs, the reactions were much stronger.
372 Analytical and Simulative Studies

Effect of Coalition Selection Mechanism

If a sourcing contract terminates (or if a bank does no cooperative sourcing at
all), the bank can evaluate the membership in another coalition which might
represent a more beneficial sourcing location. Thus far, banks selected a poten-
tially new coalition based on efficiency, i.e. the bank evaluates all relevant coali-
tions based on the expected net benefits from joining it (selection mechanism
“efficiency”) and requests membership in the coalition which promises highest
net benefits. Simpler mechanisms would entail choosing a coalition based on the
number of current members (“size”), current total process volume (“volume”), or
randomly (“random”). Consequently, the variation of selection mechanisms
relaxes the hitherto existing assumption about the actors’ information setting or
bounded rationality and tests the consequences for the simulation results.
The following table shows that no significant differences in global monetary
values between these four selection mechanisms are ultimately found, after a
sufficient number of periods. As shown in the simulations above, the number of
coalitions in the system is not very high. Thus, large deviations from applying
other selection mechanisms do not appear.
Coalition selection mechanism: efficiency size volume random
Monetary PCS 57.6% 59.7% 58.1% 59.2%
results NS 54.0% 56.2% 54.7% 55.5%
Activities for outsourcing actions 97 96 94 169
processing/ switching actions 166 170 238 694
servicing backsourcing actions 25 23 21 94
Activities for outsourcing actions 88 88 79 197
risk switching actions 211 160 189 704
monitoring backsourcing actions 27 26 19 137
Table 65: Monetary results and activities resulting from different coalition
selection mechanisms (average results after 400 periods)
But, the different scenarios need a significantly different number of activi-
ties in order to reach these results (therefore a longer simulation horizon of
T=400 was chosen). On the one hand, this causes more setup transaction costs
not covered in the periodical NS, but also lowers periodical transaction costs by
learning effects in scenarios where more outsourcing and switching actions oc-
cur.
One can hypothesize that greater dynamics (in the random scenario, in par-
ticular) lead to a slower adjustment process of the final values in the table above.
In contrast, Figure 173 shows that there are no significant differences over time
either. Entering any coalition as a first step is highly beneficial and any addi-
tional adjustments only lead to marginal market improvements.
 Analytical and Simulative Studies 373

60%

50%
relative global net savings (NS)

40%

30%
efficiency
size
20%
volume
random
10%

0%
0 50 100 150 200 250 300 350 400
t

Figure 173: Evolution of NS for different coalition selection mechanisms

The almost identical global net savings are obtained in different ways within
the four scenarios. Figure 173 shows the actor-based Herfindahl index and the
number of coalitions over time, specifically for processing/servicing and risk
monitoring.
processing/servicing risk monitoring
0.25 0.25
actor-based Herfindahl index
actor-based Herfindahl index

0.20 0.20

0.15 0.15

0.10 0.10

efficiency efficiency
0.05 size 0.05 size
volume volume
random random
0.00 0.00
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800
t t
processing/servicing risk monitoring
8 8

7 7

6 6
num ber of coalitions
num ber of c oalitions

5 5

4 4

3 3
efficiency efficiency
2 2
size size
1 volume 1 volume
random random
0 0
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800
t t

Figure 174: Actor-based Herfindahl index (le.) and number of coalitions (ri.) for
processing/servicing (up.) and risk monitoring (lo.) over time
374 Analytical and Simulative Studies

The size scenario shows the highest market concentrations for both business
functions. This results from a smaller number of coalitions. In contrast, random
selection leads to comparably lower market concentration for process-
ing/servicing (upper left diagram), although it leads to even fewer coalitions. The
number of actors in coalitions (not displayed) increases to a certain level during
the first 200 periods and then remains almost constant with no significant differ-
ences between the four scenarios. Therefore, changes in market concentration are
almost solely influenced by the number of coalitions and heterogeneity in size.
For example, the random and volume scenarios show the same market concentra-
tion for risk monitoring, but differ strongly in the number of coalitions. Conse-
quently, the volume scenario shows a comparatively greater heterogeneity in
coalition size.
The reason why the random scenario leads to the lowest number of coali-
tions is revealed in the greater dynamics. Since the random selection mechanism
will less frequently lead to a local optimum, from the individual’s perspective,
which locks it to the current coalition. Many more switching actions take place,
which lead to a stronger shakeout and thus to fewer coalitions with high global
net savings. Of course, this cannot be viewed as an argument for higher effi-
ciency. Taking into account the transaction costs for switching, there would be
high pitfalls related with these high dynamics as well.
The results of varying the simulation control parameters can be summarized
as follows:
o Varying the initial coalition size (initCoalSize) leads to a larger number of
coalitions at smaller initCoalSize values, but to a reduced number for larger
parameter settings. The effect on the monetary results and on market con-
centration is marginal.
o Increasing the initial number of coalitions (maxCoalBuilding) does not af-
fect the eventually resulting monetary results and market structures, but ac-
celerates the consolidation process. Furthermore, it leads to a temporary
over-reaction in terms of number of coalitions.
o The choice of the coalition selection mechanism has no structural impact on
the monetary results but leads to different market consolidations. The ran-
dom selection of coalitions leads to the fewest number of coalitions because
the banks get less or later caught in a local optimum since improvement of
the individual situation happens more gradually.
 Analytical and Simulative Studies 375

5.3.4.7 Lock-in to Effects

This final section of the simulation studies summarizes and complements the
previous results in regards to the lock-in of banks into local optima which leads
to an overall loss in efficiency from a global perspective.
While the simulation results prove to be rather stable in regards to process
cost parameterization and disclosure of high potential benefits from cooperative
sourcing (section 5.3.4.2), behavioral uncertainty, resulting from externalities,
and transaction cost parameters have a strong negative impact (section 5.3.4.3).
Thus, banks often outsource their business functions, but depending on the level
of transaction costs, are supposed to remain in the first entered coalition because
further expected process cost savings are too low and the costs of switching are
too high.
In order to analyze the strength and characteristics of this lock-in effect, in
the following the length of a sourcing deal TCoSo is varied because a longer time-
frame helps both to amortize the switching costs (transaction costs for negotia-
tion CN and adoption CAD) and to reduce the impact of externalities (due to
stronger discounting). To open the analysis, Figure 175 shows the finally result-
ing relative individual net savings separately for all of the back-office functions.
Since there is interdependence between the term of the contract and the setup
transaction costs for negotiation and adoption, which have to be amortized dur-
ing the term of the contract, we analyze the impact of TCoSo at different levels of
cN and cAD (simultaneously varied by an alter_TCcoefficients_setup factor in
seven steps from 1 to 22).
relative indiv. NS processing/servicing 
relative indiv. NS risk monitoring 
relative indiv. NS workout 
   
   
 
   
  
  
  
   
 
     
    
     
 
      

0.40
  
   
   
   
   
 

  



 

0.15 



 
0.30
 
  
  
 
   
  

0.30  
 


 
  
 
  
    

 
 0.10
 0.20  


0.20  



  
 
 
 
  
  0.05 
 0.10  

0.10  
  

 



0.00 0.00  
0.00




10 10 10
5 5 
8 alter_ 5 
8
8
alter_ 10 
6 alter_ 10 6 TCco 10

6
TCco 15 TCco 15 4 effic 15 4
e fficie 4 oSo e fficie 20 oSo ien oSo
nts_
20 2 TC nts_
2
TC ts_se 20 2
TC
setu setu tup
p p

Figure 175: Relative individual net savings for the different back-office
functions
As expected, increasing the contract duration leads to higher individual net
savings. For all back-office functions, structurally similar increases in the aver-
age net savings correspond with increasing TCoSo, with workout showing the
highest marginal increases at all levels of setup transaction costs, followed by
376 Analytical and Simulative Studies

risk monitoring. The following table shows the results from a linear regression
which approximates the results from Figure 175. The larger b1 is, the more sensi-
tively the savings react to the contract duration.
b1 b2 (impact of al-
r2
(impact of TCoSo) ter_TCcoefficients_setup)
Processing/servicing .262 -.909 .895
Risk monitoring .374 -.896 .942
Workout .458 -.837 .910
Table 66: Results of linear regression analysis – dependence of relative indi-
vidual net savings on contract duration and setup transaction costs214
This positive relationship might seem to contradict the existing literature on
the subject, which states that short-term contracts tend to be more successful than
long-term agreements (Lacity and Willcocks 1998, Hartzel and Nightingale
2005). There are two arguments to solve this divergence: First, these papers
argue that short-term contracts, which have to be more frequently renegotiated,
can take into account changes of requirements, e.g. caused by environmental
dynamics. The simulation model covers the renegotiation of contract details by
incorporating periodical coordination costs. Since the insourcer has the same
interests in effective and efficient process management as the outsourcer (be-
cause both receive the same services), long-term contracts in cooperative sourc-
ing must deal with fewer incentive problems than in traditional outsourcing.
Second, empirical studies often do not (and are not able to) take costs for renego-
tiation into account, since they only measure a particular point in time. Nonethe-
less, these sometimes have a strong negative impact on the overall efficiency of
the sourcing solution.
The question now is: does a longer contract term raise market dynamics be-
cause, as argued above, there will be more time for amortizing switching costs,
or does it reduce it, due to decreased behavioral uncertainty and a lower impact
of externalities?
Figure 176 shows the number of switching (left) and backsourcing (right)
actions related to the number of outsourcing actions for all three back-office
functions. 2-dimensional figures with indifference curves for different levels of
the setup transaction cost factors have been preferred for the following analyses
in order to reduce visual complexity. The legends show the corresponding levels
of alter_TCcoefficients_setup.
The diagrams show structural differences between the three business func-
tions. For processing/servicing, the graphs (upper diagrams) show an initial peak
214
All parameters are significant with p< 0.001.
 Analytical and Simulative Studies 377

at short contract durations before decreasing, except for very high transaction
costs where a slightly and continuously increasing relationship can be found. For
risk monitoring (middle diagrams), the increase into the peak lasts much longer,
but nevertheless the graphs drop again at high contract durations. By contrast,
workout (lower diagrams) shows significant and continuously rising relative
switching and backsourcing activities.
switching/outsourcing actions backsourcing/outsourcing actions
processing/servicing: switching/outsourcing actions 1 processing/servicing: backsourcing/outsourcing actions
2.5 1.5 0.3
2
3
0.25
2 5
10
22
0.2
1.5
ratio
ratio

0.15

1
1
0.1 1.5
2
0.5 3
0.05 5
10
22
0 0
2 3 4 5 6 TCoSo 7 8 9 10 2 3 4 5 6 TCoSo 7 8 9 10

risk monitoring: switching/outsourcing actions risk monitoring: backsourcing/outsourcing actions

2.5 0.3

0.25
2

0.2
1.5
ratio

ratio

0.15
1
0.1

0.5
0.05

0 0
2 3 4 5 6 CoSo
7 8 9 10 2 3 4 5 6 7 8 9 10
T CoSo
T

workout: switching/outsourcing actions workout: backsourcing/outsourcing actions

0.7 0.3
1 1
1.5 1.5
0.6 2 0.25 2
3 3
0.5 5
5
10 0.2
10
22
0.4 22
ratio
ratio

0.15
0.3
0.1
0.2

0.05
0.1

0 0
2 3 4 5 6 CoSo
7 8 9 10 2 3 4 5 6 CoSo
7 8 9 10
T T

Figure 176: Number of switching (left) and backsourcing (right) actions related
to outsourcing activities for processing/servicing (upper diagrams),
risk monitoring (middle) and workout (lower) for different setup
transaction costs under variations of contract duration TCoSo (legends
give levels of alter_TCcoefficients_setup)
378 Analytical and Simulative Studies

An explanation for the different results can be provided by comparing the

market structures for the different business functions. As the previous sections
showed, processing/servicing has the highest number of cooperatively sourcing
actors, followed by risk monitoring, and workout, further behind. This results
from a different level of cooperative sourcing because actors see fewer potential
savings in workout. Consequently, for workout, longer contract duration first of
all helps to overcome the initial start-up problem, which is also reflected by the
strongest relative increase of net savings in Figure 175. Nevertheless, the market
remains less mature; still much fewer banks decide to cooperatively source
workout compared with processing/servicing. Furthermore, as previously shown
in section 5.3.4.1, the smaller process volumes contain higher possible econo-
mies of scale, which lead to a higher consolidation over time and therefore in-
creasing switching activities until they reach this state.
By contrast, the large process volumes, the already exploited process cost
savings, and the rather small proportion of fixed costs in processing/servicing
will reveal no further individual benefits from additional economies of scale if
the actor switches the coalition. Therefore, longer contract duration does not lead
to greater switching opportunities. Moreover, longer contract duration has the
effect of stabilizing the market by giving higher behavioral certainty to all par-
ties. The dynamics decrease and the impact of externalities is reduced (due to
stronger discounting).
At very high transaction costs, the effect is reversed. Since high transaction
costs lead to a much lower number of cooperatively sourcing actors, the same
argumentation holds here as for workout.
For risk monitoring, both effects can be observed, depending on the contract
duration. Risk monitoring contains a larger proportion of fixed costs, which leads
to higher potential economies of scale, even if the bank has already entered a
coalition despite the large process volume. Thus, an increase at small TCoSo has a
positive impact on system dynamics (and also results in a stronger increase of
individual net savings in this range of TCoSo, cf. Figure 175). By contrast, at high
TCoSo, the effect of increasing stability, which leads to higher behavioral cer-
tainty, becomes dominant.
This stabilizing effect is strengthened by the fact that the selection of a coali-
tion, based on the efficiency criterion, leads to a quite efficient configuration in
very early phases. If we used the other extreme in coalition selection mecha-
nisms, i.e. random selection (cf. section 5.3.4.6), all charts of Figure 176 above
would show increasing trends. When banks join coalitions randomly, there is a
much higher probability that there is another coalition which promises higher net
savings.
 Analytical and Simulative Studies 379

One should be aware that the dynamics measures, shown in the charts
above, always represent two sides of the coin: on the one hand, higher dynamics
represent a more volatile market where actors search longer in order to gain
higher cost savings. On the other hand, higher dynamics represent higher behav-
ioral uncertainty for the single actor. He does not only have the opportunity to
switch coalitions or to backsource but is also forced to do so since the coalition
structure changes, affecting his own situation. The resulting transaction costs for
more frequent coalition changes must not be disregarded (although they are not
included in the periodical net savings (NS), which is used as the primary qualita-
tive output measure in this work).
In the previous sections, the degree of inefficiency was measured by com-
paring the simulation results with centralized optimization. Consequently, does
the dual effect of a longer contract term (i.e. easier amortization and, to a degree,
less market dynamics) increase or decrease system efficiency?
The left diagram of Figure 177 shows the basic ratio of average individual
net savings resulting from decentralized vs. centralized coordination while the
right diagram presents a corrected ratio which takes into account that centralized
coordination leads to a certain proportion of banks which achieves negative
monetary results215. Furthermore, the coordination costs are left out again to
ensure comparability (cf. Figure 145 on p. 342).
ratio global NS without CC (decentralized /

0.8 1.4
ratio global NS without C (decentralized /
centralized coordination), corrected

1.2
centralized coordination)

0.6
1.0
C

0.4 1 0.8
1.5 1
2
1.5
3 0.6
2
0.2 5
3
10 5
22 0.4
10
22
0 0.2
2 3 4 5 6 7 8 9 10 2 3 4 5 6 7 8 9 10
CoSo
CoSo
T T
C
Figure 177: Ratio between individual NS without C resulting from
decentralized vs. centralized coordination – in the right chart
corrected by the proportion of banks with non-negative net savings
achieving from centralized coordination
As the charts show, the ratio between net savings from decentralized coordi-
nation and centralized coordination is, depending on the level of setup transac-

215
In order to obtain an accurate ratio measure, the net savings resulting from centralized coordina-
tion are multiplied with the proportion of banks which do not show negative savings, because it is
implausible and irrational, from an individual perspective, to accept negative monetary results in
a decentrally coordinated system. This bias is stronger the higher the transaction costs are.
 380 Analytical and Simulative Studies

tion costs, more or less significantly positively affected by the contract duration.
This positive relationship does not change when considering the proportion of
banks receiving negative consequences from centralized coordination (right
chart).
Since the centralized coordination takes the contract duration into account
by optimizing the global net present value of the sourcing decision, this positive
relationship between efficiency and contract duration indicates that centralized
coordination overcomes some form of lock-in which is not covered by the analy-
sis of system dynamics above.
If the market structures of decentralized and centralized coordination are
compared in the same way, strong differences can be found both in the number
of cooperatively sourcing actors and resulting coalitions. Figure 178 compares
the number of actors in coalitions. This is again done by computing the (cor-
rected216) ratio between the result from decentralized and centralized coordina-
tion. Thus, a ratio of 1.0 would represent a similar structure in both “worlds”.
processing/servicing risk monitoring workout 1
1 1 1
1.5
0.9 0.9 0.9
2
0.8 0.8 0.8
3
0.7 0.7 0.7
5
0.6 0.6 0.6
1 10
ratio

ratio
ratio

0.5 0.5 0.5

1.5 22
0.4 2 0.4 0.4
0.3 3 0.3 0.3
0.2
5 1 1.5 2 3
0.2 0.2
10
0.1 22 0.1 5 10 22 0.1
0 0 0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
TCoSo TCoSo TCoSo

Figure 178: Ratio of actors in coalitions resulting

from decentralized / centralized coordination
The figures clearly show that the percentage of cooperatively sourcing ac-
tors remains somewhere between 10% (processing/servicing) and almost 65%
(workout) below the parity ratio of 1.0. The effect of switching costs on this
result (i.e. relative distance between indifference curves) is strongest for process-
ing/servicing and workout, followed by risk monitoring. It is reduced by extend-
ing the time for amortization, as shown by the converging graphs.
These results show that there is a problem with motivating certain actors to
take part in cooperative sourcing (i.e. start-up problem), although they would not
even encounter negative savings since the results have already been corrected by
this issue. This corresponds to the findings at the end of section 5.3.4.1. This
problem is more prevalent when the potential absolute savings are lower, due to
lower process volumes (workout). In contrast, risk monitoring reacts comparably

216
Again, the values resulting from centralized coordination are multiplied with the fraction of banks
which do not show negative savings.
 Analytical and Simulative Studies 381

less sensitively than processing/servicing or workout because it contains a cost

structure which is strongly driven by fixed costs.
Beside the number of actors involved in cooperative sourcing, the number of
resulting coalitions is revealing (Figure 179). Corresponding to Figure 178, the
diagrams show the ratio of the number of coalitions resulting from decentralized
and centralized coordination217.
processing/servicing risk monitoring workout
4 2 5 1 1.5 2
1.8 4.5
3 5 10
3.5 22
1.6 4
3
1.4 3.5
2.5 1.2 3
ratio
ratio

2 1 2.5
0.8 2
1.5
0.6 1.5
1
1 1.5 2 3 0.4 1 1.5 2 3 1
0.5 0.2 5 10 22 0.5
5 10 22
0 0 0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
TCoSo TCoSo TCoSo

Figure 179: Ratio of number of coalitions resulting from decentralized /

centralized coordination
In almost all cases, the resulting ratio is larger than 1.0, representing a situa-
tion in which decentralized coordination leads to more coalitions and thus lesser
consolidation than for central optimization. In more mature markets (process-
ing/servicing and risk monitoring), this effect is further aggravated by longer
contract duration (i.e. increasing graphs) and, at least in the case of process-
ing/servicing, with higher setup transaction costs.
This shows a third reason for inefficiencies. While the results above showed
that there is almost no individual lock-in for processing/servicing, in particular,
and although actors which are subject to a start-up problem (and thus are not
member of any coalition) are disregarded here, the part of the system which is
involved in cooperative sourcing still does not reach the same level of consolida-
tion as the central optimization routine. This third effect can be described as a
further lock-in effect which results from the fact that not individuals but rather
groups of individuals (i.e. coalitions) do not switch (i.e. merge) coalitions. While
the effect of individual lock-in is driven by individual behavioral uncertainty and
individual switching costs, the complex network of driving and inhibiting factors
leads to a group-based or system-wide lock-in which cannot be assigned to and
resolved by singular individual decisions.

217
Again, the results from centralized coordination are corrected by the proportion of actors not
being worse off. Moreover, the ratio is adjusted because of the fact that less decentralized coordi-
nation leads to fewer actors in coalitions. To isolate this effect, the decentralized result of the
number of coalitions is divided by the ratio of actors in coalitions (centralized / decentralized re-
sult). This leads to the same calculatory basis of actors forming the differing number of coalitions
in the two governance settings.
382 Analytical and Simulative Studies

This effect is driven by the contract duration because the design of the coop-
erative sourcing model allows for only individual decisions and not for group
(i.e. coalition-wide) decisions. As explained above, higher TCoSo leads to less
uncertainty about the partners’ behavior and thus stabilizes the system by reduc-
ing dynamics (cf. decreasing dynamics for processing/servicing (Figure 176), in
particular, which showed the most consistent results here). By contrast, since this
stabilizing effect did not occur (predominantly) in the relatively immature market
for workout, the graphs show the opposite direction in Figure 179 as well.
Finally, we can argue that lock-in effects are reduced by two different fac-
tors which reduce the setup transaction costs.
First, negotiation costs decrease with increasing experience in outsourcing
projects. Designing outsourcing contracts and agreeing to join a coalition will
become more controllable and calculable with increasing sourcing management
capabilities of the firm.
Second, the adoption costs of merging business functions will decrease as
well because they are partially determined by the degree of similarity of the
different partners’ business functions. Cooperative sourcing activities will lead to
higher similarity (or “process standardization”) throughout the system and thus
to less switching costs.
During the simulations, however, the average degree of similarity through-
out the whole system of modeled banks and business functions increases only
insignificantly. The model assumes perfect process similarity for all members of
the same coalition, i.e. the business functions become “standardized” (which
causes adoption costs). However, since an individual’s path towards a stationary
state usually does not consist of many coalition switches, and because several
actors do not take part in cooperative sourcing at all (cf. section 5.3.4.1), the
average similarity increases only marginally throughout the whole system.
This final section can be summarized as follows:
o System-wide inefficiency, which represents inferior monetary results from a
global perspective, can be ascribed to three causes:
o For various reasons, the bank does not engage in cooperative sourc-
ing at all (i.e. start-up problem). These reasons can be negative
benefits, the strategic constraint, uncertainty about the potential
partners’ behavior, or a refusal from the coalition’s side (negative
voting result, cf. section 5.3.4.1).
o After entering a coalition, the actor is locked in because costs for
switching to a more beneficial coalition exceed expected additional
savings. This is especially the case if the early decision was already
based on efficiency criteria.
Analytical and Simulative Studies 383

o The whole system is locked into an inferior solution because struc-

tures have been formed which might be inefficient (e.g. too many
coalitions). A change cannot be derived by individual switching
decisions but only by merging coalitions, for example.
o The following factors can reduce lock-in effects:
o Learning effects (regarding the negotiation and coordination of out-
sourcing relationships) reduce costs of switching the coalition.
o Increasing the similarity of business functions (process standardiza-
tion) throughout the system, resulting from cooperative sourcing,
leads to lower adoption costs and thus to less costly switching ac-
tivities. Nevertheless, the simulations showed no sufficient process
standardization throughout the system.
The following chapter starts with a summary which also provides an overall
aggregation of the simulation study results presented in this and the previous
sections. Based on these findings, contributions to theory and practitioners are
derived, followed by a discussion of the limitations and a validation of the ap-
proach.
384 Conclusion

6 Conclusion
“… relationships between alliance partners entail both cooperation and competition. Thus, winning
through alliances is, to a significant degree, also a matter of winning within one’s own alliances.”
(Doz and Hamel 1998, p. 55)

This concluding chapter starts with a summary of the findings gathered in this
work (section 6.1). Based on the results, the contributions for research and impli-
cations for practitioners are derived (6.2). The subsequent sections focus on
validating the chosen research approach (6.3) as well as on the related limita-
tions, which have been accepted in this research (6.4). While certain limitations
always exist in research, some of them provide promising directions for future
research, highlighted in section 6.5.

6.1 Summary of the Findings

Cooperative Sourcing as Governance Mode
The object of this research was cooperative sourcing, which describes the merg-
ing of primary business functions by several firms. This merging can take place
either in one of the partner firms or in a joint subsidiary which will become the
insourcer while the remainder of the coalition members will act as outsourcers.
Consequently, cooperative sourcing as research domain is part of the outsourcing
research.
In addition or in contrast to traditional outsourcing relationships, cooperative
sourcing has the following unique characteristics:
o The outsourcer and insourcer of the sourcing relationship are not pre-defined
by their distinct overall business models.
o Since cooperative sourcing is a close form of horizontal interorganizational
cooperation, it inherently represents a form of coopetition. Coalition partners
usually are (potential) competitors.
o Aligned interests: Since the insourcer firm provides not only services for
other firms but also for its own needs, it has the same interests in effective
and efficient process management as the outsourcers. Thus, from this point
of view, there is a lower incentive for moral hazard and resulting service de-
basement problems.
Conclusion 385

o In contrast to traditional outsourcing relationships, the benefits are allocated

by negotiation rather than by market mechanism.
o Conceptually, cooperative sourcing is a multilateral instead of a bilateral
agreement.
o As a consequence, the strategic situation of cooperative sourcing is even
more intricate than in traditional 1:1 outsourcing relations as now there is
the possibility of coalition building. This has a substantial impact on coali-
tion stability and cost allocation rules (cf. game-theoretical analysis in sec-
tion 5.1).
Research Question 1: Drivers and Inhibitors of Cooperative Sourcing
The first research question of this work asked for the drivers and inhibitors
which affect cooperative sourcing activities. Based on an extensive theoretical
discussion as well as on a literature analysis in the IT outsourcing and BPO re-
search domain, several factors have been extracted and classified in a theoretical
frame, summarized in Table 67 and Table 68.
Driver Theoretical argument Theory
Cost reduction Economies of scale and skill, PCE,
network effects, business strategy NET, PPF
Capital reduction / cost variabilization Economies of scale PCE
Task performance improvement Resource gaps RDT
Overall business performance improvement, Economies of skill, agency costs PCE, PAT
improvement of control
Strategic access to superior skills and re- Resource gaps, environmental RDT
sources, reduction of technological risks uncertainty
Increased technological and process flexibility Resource gaps RDT
Core competence focus / increased business Resource properties RBV/CCV
flexibility
Table 67: Cooperative sourcing drivers (summary of Figure 6, Figure 8, Table
2, and Table 3)
386 Conclusion

Inhibitor Theoretical argument Theory

Expected “start-up” costs Transaction costs TCE
Hidden “start-up” costs Transaction costs, lack of expertise, task inter- TCE, PCE
dependencies
Contractual amendments Uncertainty, complexity TCE, TIC
Unexpected disputes and litiga- Measurement problems, lack of expertise, poor PAT,
tion relational governance RBV, RT
Cost escalation, loss of control Measurement problems, insufficient capabilities PAT, PCE,
of the provider (no economies of skill), uncer- TCE, NET
tainty, no critical mass reached
Loss of quality Insufficient capabilities of the provider (no PCE, TIC,
economies of skill), task complexity, measure- PAT, RT
ment problems, incomplete contracts, dis-
economies of scope, poor relational governance
Loss of operational flexibility Necessary standardization for achieving econo- PCE, PPF
mies of scale, outsourcer cannot differentiate
Confidentiality / security prob- Moral hazard, opportunistic behavior, transac- PAT, TCE
lems tion risk
Loss of own skills, becoming Uncertainty about future strategic importance of CCV/RBV,
dependent on provider outsourced activity, external resources become RDT
more valuable
Lock-in, loss of strategic flexi- Loss of skills, switching costs, small number of RBV,TCE,
bility suppliers PPF,NET
Table 68: Cooperative sourcing inhibitors (summary of Figure 6, Figure 9,
Table 2, and Table 5)
While many of the arguments originally stem from IT outsourcing research,
the following arguments are specific to BPO or cooperative sourcing:
o BPO is supposed to lead to less (hidden) transaction costs. It might be less
difficult to take whole business functions out of the firm since the close rela-
tionship between IT and business has not to be cut interorganizationally
(TCE perspective).
o By contrast, selectively outsourcing parts of a business process leads to
vertical diseconomies of scope. In the empirical study, these showed to be a
substantial inhibitor for BPO potential (section 3.6.3.2) (PCE perspective).
o Furthermore, BPO requires standardization of business functions, which is a
weak concept (Ungan 2006) compared with “IT standards” or “standard
software”. A lack of missing process standardization can be a significant in-
hibitor of BPO (Petzel 2003; Rouse and Corbitt 2004; Wüllenweber, Beim-
born, and Weitzel 2008).
Conclusion 387

o The strategic argument of access to specialized resources is seen as less

relevant in cooperative sourcing since the firms see themselves core compe-
tent in providing their primary business functions (which are the object of
cooperative sourcing) (RDT perspective). Nevertheless, there can be large
differences in economies of skill, as the cost differences in the SME credit
process showed.
o The already mentioned alignment of interests between insourcer and out-
sourcers should lead to fewer problems related to opportunistic behavior of
the insourcer.
The credit business in the German banking industry was chosen as applica-
tion domain for empirical evidence. Although most banks have extensive experi-
ence with (IT) outsourcing, BPO activities emerged very late, compared with
other industries. Furthermore, huge differences can be found between the differ-
ent business segments. While payments and securities processing are served by
an established provider market in many countries, the credit processing market is
still very immature in Germany and German banks are quite reluctant in adopting
this option, compared to other countries (e.g. USA, UK, or the Netherlands).
Explanations for low dynamics in the credit business are manifold and stem
from different theoretical perspectives:
o Lack of standardization: Compared with payments and securities processing,
there is no need for inter-bank coordination to create common standards in
the credit business (Bongartz 2004, Focke et al. 2004). As a consequence,
activities cannot directly be bundled to achieve economies of scale (PCE
perspective).
o Critical mass argument: As long as the credit factories have not been able to
attract a sufficient number of clients and to aggregate their clients’ processes
to a unified standardized process platform, they cannot offer services at a
sufficiently low price and thus cannot attract further clients (NET + PCE
perspective).
o As a consequence, the proof of relative cost superiority of cooperative
sourcing is still missing (Focke et al. 2004). Credit factories often cannot of-
fer dominant process costs (PCE perspective) or banks themselves often do
not know their own process costs in order to be able to evaluate make-vs.-
buy options (Figure 87).
o Cultural barriers and historical path dependency: Many German banks still
cannot imagine outsourcing parts of their business processes. The credit
business (mortgaging, in particular) as whole is seen as a core competence
(cf. empirical results in section 3.6.2.1) (RBV/CCV perspective). In other
countries, not only the processing and servicing is provided by external par-
388 Conclusion

ties, but furthermore the refinancing (e.g. on the capital market in the USA
and the UK) or even sales (e.g. brokers in the Netherlands) is provided out-
side the bank. German banks often evaluate the agency risks as too high, in
particular for outsourcing sales (PAT perspective).
o Integration requirements: Since the credit business consists of making and
communicating more or less complex decisions, real-time integration be-
tween the involved parties has to be realized, ensured by a sophisticated ser-
vice level management (Focke et al. 2004; Krichel and Schwind 2003). This
induces transaction costs (TCE perspective).
o Legal issues (section 3.5) such as the German Civil Code (BGB), section
613a and the VAT problem in banking often hinder the efficiency of out-
sourcing options.
Nevertheless, observations of the German banking industry show a slow but
upward trend. The empirical studies (section 3.6) showed the following:

o One-third of the study participants confirm that there is potential for credit
factories in terms of economies of skill (Figure 53). Further, the majority of
the banks argue that credit factories could achieve economies of scale
(Figure 89 and Figure 90) and see potential for process standardization of at
least parts of the credit process (Figure 64) (PCE perspective: economies of
skill or scale, NET perspective). Experts expect that the low profitability
which is common to the SME loans business will be tackled by higher stan-
dardization and more cooperative sourcing.
o Driven by the “industrialization” hype, banks start to think in business func-
tions instead of whole business segments when it comes to evaluating re-
sources and capabilities (PPF perspective (value chain concept) and
RBV/CCV perspective). Thus, selective outsourcing becomes an option.
The studies showed that banks are averse to cooperative sourcing in the
credit business in general but showed a less negative attitude when considering it
at the business function level: up to 46% of the responding banks could imagine
outsourcing of single back-office functions to another bank.

Research Question 2: Stability of Cooperative Sourcing Coalitions

Based on a simplified form of the cooperative sourcing model, we analyzed how
the evolving process costs have to be allocated in a cooperative sourcing coali-
tion in order to ensure its stability, i.e. to ensure the participation of all potential
members. While the equal allocation of gain as well as the proportional alloca-
tion of costs and the Shapley allocation were tested, only the proportional alloca-
tion ensures stable coalitions. Although the other schemes do not lead to unstable
Conclusion 389

coalitions in all cases, determining them ex ante when founding a coalition re-
sults in the problem that, with new members joining the coalition in later periods,
the allocation scheme would have to be completely renegotiated.
Moreover, in the case of additional coordination costs, a potential coopera-
tion first has to be tested for the existence of a core, i.e. whether the rationality
criteria lead to any cost allocation which can be accepted by all potential coop-
eration members.
Although proportional cost allocation showed to be the only stable alloca-
tion in the test, the experimental results showed that most participants did not
want a cost allocation which is close to it, but preferred the Shapley value in-
stead. With added coordination difficulty (shrinking core, advent of transaction
costs), the bargaining games increasingly resulted in inefficient constellations,
giving the impression of the impact of bounded rationality or even the occur-
rence of “irrationality”, even in such simple scenarios.

Research Question 3: Market Effects

Based on the derived cooperative sourcing drivers and inhibitors, the third re-
search question focused on the resulting market effects from cooperative sourc-
ing activities in a system of independent firms. In order to shed light on this
question and to provide a theoretical contribution towards cooperative sourcing,
a mathematical, agent-based model of a system of firms, which autonomously
decide about sourcing their activities, was developed. The model was fed with
empirical data and used in simulation studies on system dynamics and resulting
market structures.
The analysis of market effects followed three different dimensions. First, the
general degree of cooperative sourcing and the resulting monetary consequences
for the overall system and individuals were measured; second, the resulting mar-
ket structures in terms of market concentration and heterogeneity of coalitions
were captured while, third, the dynamics during the system’s path to a stationary
state were tracked (outsourcing, switching, and backsourcing activities).
The following figure summarizes the empirical characteristics of the ana-
lyzed business functions of the SME credit process which were found to be rele-
vant in the simulations and, thus, allow the possibility of generalizing the find-
ings:
390 Conclusion

Business function of the SME credit process

Sales/ Assessm. Processing Risk
Workout
preparation /decision servicing monitoring

Core Figure 53
competence

Strategic
risk Figure 91

Standardization
potential Figure 68

Task Figure 55 and

complexity simulation studies

Ratio between fixed Process cost

and variable costs analysis (Fig. 79)

Level of Process cost analysis

fixed costs (section 3.6.2.5)

Level of Process cost analysis

variable costs (section 3.6.2.5)

Process Empirical results and

volume simulation parameterization

Figure 180: Business function properties (empirical status quo)

While sales/preparation and assessment/decision are evaluated as a core
competence and as strategically sensitive by most banks, processing/servicing
and workout represent the opposite pole with risk monitoring/management being
in the center. Accordingly, the industry-wide standardization potential is re-
versely evaluated. Task complexity is quite balanced while the relative fraction
of fixed costs is (rather) high for risk monitoring/management and workout. In
absolute numbers, fixed costs are highest for processing/servicing and risk moni-
toring while variable costs are highest for sales/preparation and process-
ing/servicing. The process volume is highest for processing/servicing and risk
monitoring/management because these business functions at least partially oper-
ate on all loans in stock.
With this classification in mind, the results from the simulation studies can
be interpreted in a more generic manner.
Based on the empirical data, the simulations found moderate to high coop-
erative sourcing activities of the less strategically relevant business functions.
From a qualitative perspective, these results almost matched the empirical results
where the respondents were asked for the optimal cooperative sourcing configu-
Conclusion 391

ration of their credit process. In the simulations as well as in the empirical study,
processing/servicing showed to be the most frequent candidate. This result was
followed by workout in the empirical study and risk monitoring/management in
the simulation studies, each followed by the other respective business function.
Sales/preparation and assessment/decision played no significant role in both
investigations.
From a quantitative perspective, the level of cooperative sourcing activities
nevertheless differed quite strongly between the simulation studies and the em-
pirical results – even when only asking for optimal instead of real sourcing be-
havior in the survey. This can partially be explained by the high strategically
reasoned resentments against outsourcing in reality. In the simulation studies,
this was resembled by narrowing the strategic constraint which drastically re-
duced the cooperative sourcing activities.
In the simulation studies, the cooperative sourcing activities led to signifi-
cant process cost savings and – after subtracting transaction costs – net savings.
For business functions with lower process volumes, diseconomies of scope (or
interface costs) had a strong negative impact and agency costs inhibited out-
sourcing of processes with higher strategic relevance and with higher complex-
ity. Overall, the net savings reacted most sensitively to changes of the negotia-
tion and coordination cost factor (cN).
The monetary results of the simulation studies usually reacted rather insensi-
tively towards most changes of the parameterization – e.g. when changing the
process cost structure, introducing value-added tax, altering the discount rate, or
other basic parameters. By contrast, when decreasing the ratio between fixed and
variable process costs (i.e. less economies of scale, the system reacted quite
sensitively by reducing cooperative sourcing activities. More homogeneous
process cost structures218 interestingly had the opposite effect (more homogene-
ous decisions led to easier and more frequent agreements towards forming a
coalition). However, the strongest effect was obviously driven by a variation of
the transaction cost level. While the global savings were almost linearly related
to this variation, the average individual savings dropped a lot more.
The benefits from cooperative sourcing turned out to be greater when firms
were of a similar size. In a heterogeneous system, large banks with high process
volumes would often not be able to generate enough additional economies of
scale from cooperative sourcing while small banks would have to bear too high
relative transaction costs.

218
Business functions of different firms which have more similar ratios between fixed and variable
costs.
392 Conclusion

The simulations were accompanied by an optimization routine which deter-

mined the resulting cooperative sourcing configuration for the whole system of
simulated firms from a global efficiency perspective. This centralized coordina-
tion achieved significantly higher savings by increasing the cooperative sourcing
rate (esp. regarding business functions with lower process level but not too high
strategic relevance) but also burdened some actors by way of additional costs
instead of cost savings. In the decentrally coordinated setting of autonomously
acting firms this was not possible since firms optimize their net present value and
cannot negotiate transfer payments in the model.
Furthermore, when varying transaction costs, the centralized coordination
accepted higher transaction costs up to a certain degree without reducing the
number of cooperatively sourcing banks. As a consequence, the process cost
savings could be kept constant but the number of actors that would achieve a
financial loss increased strongly.
As the second dimension, the resulting market structure effects were ana-
lyzed. First of all, a strong heterogeneity was found. Although the markets did
not tend to be monopolistic, the resulting coalitions differed very strongly in size.
This structure was also determined by the centralized coordination.
Most banks only outsourced one or two of their back-office functions to
keep some competencies inhouse even in the back office. This led to the “major
outsourcer” being the most favored business model in the basic setting.
If the fixed costs in coalitions increase and the variable unit costs decrease
due to a technology change (dynamic economies of skill), larger coalitions will
occur in mature markets with high cooperative sourcing rates and large process
volumes while cooperative sourcing activities will decline, by contrast, in less
developed markets with low process volumes. The decrease of variable costs had
only a slight impact on the resulting market structure, but even little reductions
already compensated for strongly rising fixed costs and thus stabilized the over-
all rate of cooperative sourcing regarding all business functions. Reduced vari-
able costs or increased fixed costs further led to a higher frequency of sales
banks (i.e. outsourcing the whole back office).
As already noted, increasing transaction costs obviously lead to lower coop-
erative sourcing rates. Nevertheless, some further interesting effects could be
found. First, increasing the interface cost base led to higher consolidation ten-
dencies and to more insourcers of multiple processes in order to keep the result-
ing interface costs low since these players offer an integrated portfolio of back-
office functions which captures economies of scope between these activities. By
contrast, higher coordination costs led to smaller and more coalitions. Similar
results could be found when considering the business neighborhood (competitive
elements increase agency costs); particularly, sensitive and more complex busi-
Conclusion 393

ness functions such as risk monitoring and workout were cooperatively sourced
in smaller coalitions to reduce cooperation with competitors.
Under centralized coordination, slight increases of transaction costs gener-
ally were answered by stronger consolidation (except for negotiation and coordi-
nation costs which over-proportionally increase with larger coalitions).
As a further decision determinant, the strategic constraint, which ensures
that strategically relevant core competencies are not outsourced, was reduced.
While the autonomously acting firms in the simulations simply reduced their
cooperative sourcing activities (coming closer to the empirical results), central-
ized coordination restructured the system towards more and smaller coalitions
where more banks can act as insourcers and thus keep their strategically impor-
tant business functions inhouse. Hence, bounded rationality of the decentrally
deciding actors led to a stronger negative impact when an additional component
had to be taken into account in the decision calculus.
In systems with lower heterogeneity in terms of firm size (i.e. process vol-
ume), there was stronger market consolidation. In the extreme case of all firms
having identical process volumes, the market consolidated very quickly towards
one single coalition.
In some scenarios, the system showed temporary over-reactions in terms of
the number of coalitions. When the benefits of cooperative sourcing were obvi-
ous, many coalitions were found in the very first periods with a subsequent con-
solidation process following in later periods. By contrast, for other business
functions, monotonously increasing concentration curves evolved over time with
the number of cooperatively sourcing firms and the number of coalitions increas-
ing rather proportionally.
During the simulations, different assumptions about bounded rationality of
the deciders were tested which had no significant impact on the monetary results
but led to different degrees in market concentration. Interestingly, the “simplest”
decision calculi when selecting a coalition (i.e. random selection and selecting a
coalition based on the number of members (“size”)) eventually led to the lowest
number of coalitions and to the highest market concentration because the banks
did not get caught too soon in a local optimum where switching the coalition to
improve the individual situation becomes inefficient.
The third dimension of analysis focused on the individual behavior of the
firms during the simulations. The rate of outsourcing, coalition switching, and
backsourcing actions was tracked and aggregated to ratio measures representing
behavioral uncertainty: in order to be able to estimate the actors’ uncertainty
about the others’ behavior, the number of switching and backsourcing actions
was set in relation to outsourcing activities. Higher values indicate that the banks
394 Conclusion

have to test more sourcing configurations unless they find their local optimum219.
More activities in the system lead to more changes of the decider’s environment
which in turn aggravates the search for the optimal coalition. It was mentioned
that these measures always represent two sides of the coin: first, higher dynamics
represent a more agile market where actors search longer to gain higher cost
savings; second, higher dynamics also lead to higher behavioral uncertainty of
the single actor which forces him – and not only gives him the opportunity – to
switch coalitions or to backsource – which in turn leads to high cumulated
switching costs.
During the simulations, outsourcing activities predominantly occurred in
early periods. Dynamics in later periods consisted almost entirely of switching
coalitions and – to a lower degree – of backsourcing formerly outsourced busi-
ness functions. The degree of dynamics differs between the different business
functions. While processing/servicing showed the most frequent outsourcing
activities, risk monitoring/management showed more frequent switching behav-
ior. The higher ratio of fixed costs which embraces higher potential economies of
scale leads to a stronger consolidation process (e.g. securities and payments
processing, c.f. section 3.4.2) and a longer search process for the optimal coali-
tion from an individual perspective.
Although the major part of monetary savings was achieved in the first peri-
ods, the individual switching behavior could take a long time before the overall
system reaches a stationary state. Some individuals searched for the optimal
coalition for a long time although the benefits were only marginal and further-
more induced negative externalities for those firms which were left in the former
coalition.
The more the firms had outsourced a particular business function, the higher
were the dynamics in the system and thus the behavioral uncertainty for the indi-
vidual firm. Since the evaluation of a coalition membership is influenced by this
behavioral uncertainty (network effects), higher dynamics led to more frequent
sub-optimal decisions which in turn led either to a lock-in to sub-optimal local
optima or to a longer and repeated search process for a more beneficial coalition.
When transaction costs were increased, the dynamics measures (and thus
behavioral uncertainty) consequently decreased. The consequence is dichoto-
mous: on the one hand, higher transaction costs are salutary because they stabi-
lize the market and reduce switching costs; on the other hand, they might rather
lock-in the market in a globally sub-optimal configuration.

219
The firm’s ultimately chosen sourcing strategy (insourcing vs. cooperative sourcing, choice of
coalition) was defined as the local optimum.
Conclusion 395

There is one exception: when increasing the agency cost basis, which is part
of the transaction costs, the ratio of backsourcing to outsourcing actions in-
creased up to a certain degree (Figure 151, p. 348). Higher agency costs, which
are triggered by the business neighborhood and therefore by the composition of
the particular coalition (which parties are involved), forced firms to backsource
their business function rather than only to switch the coalition. The agency costs
were mostly influenced by other firms’ behavior and therefore the behavioral
uncertainty increased with rising agency costs. A corresponding result could be
found when neglecting the business neighborhood. In this case the dynamics of
switching and backsourcing were strongly reduced, which in turn led to a more
stable decision basis for the firms.
It was already mentioned above that the market structure becomes much
more consolidated when the system consists of more homogenous firms in terms
of process volume. When assimilating the process volumes stepwise, the fre-
quency of switching actions increased strongly up to a certain point and then
finally dropped again to the situation of perfectly equal process volumes. The
more homogeneous the process volumes, the more active were the firms in
searching for their individual optimal coalition, i.e. there were high opportunities
for changing the coalition and, as a result, high uncertainty due to the very vola-
tile environment. These high switching dynamics led to a strong consolidation of
the system. However, when all firms had identical volumes in the extreme case,
this effect reversed, leading to all banks directly moving into the only and finally
remaining coalition.
The progress of the dynamics measures often showed structural differences
between the different business functions. When varying the contract duration at
different transaction cost levels in the final section of the simulation studies
(5.3.4.7), the different process characteristics led to apparently opposite results.
For business functions with high process volumes, already high outsourcing
rates, and rather low fixed costs (processing/servicing), longer contract duration
did not lead to any new opportunities, since economies of scale had already been
exhausted and were not very high at all. Therefore, increasing contract duration
in this case led to more stability, decreasing dynamics, and reduced behavioral
uncertainty. With opposite characteristics (workout), potential economies of
scale are high but absolute process cost savings are low due to only small process
volumes. Hence, longer contract duration helped to amortize setup transaction
costs for setting up or joining a coalition: dynamics increased. As risk monitor-
ing/management had mixed process properties (high outsourcing rate, large
process volume, but also high fixed costs), the progress of dynamics showed a
mixed picture as well.
396 Conclusion

Finally, by aggregating the results of the three dimensions of analysis –

monetary, market structure, and individual behavior – the following can be con-
cluded: The divergence in the monetary results from centralized vs. decentralized
coordination represents a global inefficiency dilemma. In general, the missing
cooperative sourcing activities can be reasoned in negative individual monetary
results or in behavioral uncertainty which again can be traced back to the exis-
tence of externalities (network effects). Both reasons represent two fundamen-
tally different arguments (Weitzel 2004, Weitzel, Beimborn, and König 2006).
While the first is “only” an inefficiency dilemma from the overall system’s per-
spective, the second directly represents an inefficiency dilemma also to the indi-
vidual decider.
In any case, there is a problem of motivating the firm to take part in coop-
erative sourcing. While in the first case, a monetary transfer is inevitable because
the firm has no economic incentive to take part in a coalition, the second prob-
lem can be solved by several contractual mechanisms or further governance
instruments220.
The dilemmas will occur less frequently in the cooperative sourcing context
than in other domains, e.g. in the IT standardization domain, where these prob-
lems already have been analyzed (Weitzel 2004). Since the gross benefits from
joining a coalition will occur for each member (economies of scale and/or
economies of skill) and can easily be negotiated in a way that all members will
have a net benefit (because the allocation of coalition costs has to be negotiated
either way – in contrast to e.g. the adoption of IT standards), both problems have
a lower impact. This was confirmed by the simulation results. The problem of
behavioral uncertainty usually arose in the second step after the firm had already
entered a coalition. Due to the dynamics, the stability of another, more beneficial
coalition cannot be guaranteed. Since substantial cost savings have already been
exploited in this state, the impact of externalities will be much stronger in this
“second” step and lead to a lock-in of the actor (individual lock-in). Furthermore,
high dynamics may lead to a globally inefficient coalition structure (e.g. many
small coalitions) which can no longer be resolved by individual decisions any-
more (system lock-in).
As already summarized at the very end of the simulation studies section,
these problems are reduced by learning effects (regarding the negotiation and
management of outsourcing relationships) and increasing process standardization
resulting from cooperative sourcing activities or other measures (e.g. the adop-
tion of standard software).

220
See (Weitzel 2004) for a discussion within a different but related context (IT standardization).
Conclusion 397

As a counter-argument, the simulations also showed that process standardi-

zation which only results from cooperative sourcing activities might be insuffi-
cient to overcome a sub-optimal system structure of too small coalitions or clus-
ters. When standardization takes place only in sourcing coalitions, an industry-
wide homogenization will fail to appear and a potential migration or merger to
larger clusters may even become more aggravated.

6.2 Contributions
The results of this research work offer several contributions to existing theory
(section 6.2.1) and implications for practitioners who have to evaluate sourcing
opportunities (6.2.2).

6.2.1 Implications for Theory

Generally, theoretical research contributions can be distinguished by their aim of
either testing theories or developing theories. The predominant part of this work
followed a theory development perspective; only section 3.6.3.2, which provided
a positivist approach on testing different production cost-theoretical arguments
towards the potential of BPO, represented a part of theory testing.
In this thesis, the concept of cooperative sourcing was theoretically devel-
oped and empirically analyzed. As an add-on to the existing theoretical view of
outsourcing research, the concept of externalities was integrated by incorporating
network effect theory. The dependence of sourcing decisions on other entities’
activities adds a further component of complexity in form of behavioral uncer-
tainty to the already complex outsourcing decision. In accordance with newer
arguments of the network effect theory (Weitzel et al. 2000) it was shown that
externalities – in a positive (generating additional scale economies) as well as in
a negative (increasing coordination efforts and agency costs) occurrence –
strongly depend on who (and not only on how many entities) enters the coalition.
For example, while systems with low heterogeneity in terms of firm size led to
higher relative individual savings from cooperative sourcing in the simulation
studies, in a heterogeneous system large banks with high process volumes were
often not able to generate sufficient additional economies of scale from coopera-
tive sourcing, while small banks would have to bear too high relative transaction
costs. Although these arguments are in accordance with PCE and TCE, the effect
also resembles newer findings of the network effect theory which argue that
returns from using the same good (here: joining the same coalition) do not con-
stantly increase with every new participant (Weitzel et al. 2000).
As conceptual contribution, a formal agent-based model was developed
which captures the theoretical arguments for and against cooperative sourcing
398 Conclusion

and integrates them into the agents’ decision function. Thus, this model repre-
sents a tool for analytically and simulatively analyzing cooperative sourcing
behavior which substantially extends earlier works with regard to both the com-
plexity of the decision basis (i.e. captured decision relevant determinants) and
the modeled unrestricted set of interacting and autonomously deciding individu-
als. The theory-based model helps to explain and to anticipate structural effects
from a system’s cooperative sourcing dynamics by integrating microbehavior
(decisions of the individuals) and macrobehavior (resulting market effects)
(Schelling 1978).
Apart from capturing the complexity and the externalities which underlie a
cooperative sourcing decision, a further contribution is the incorporation of co-
opetition into the decision calculus. Cooperative sourcing of parts of the value
chain will lead to the cooperation of competitors because similar parts of the
value chain indicate that the firms are members of the same industry and there-
fore – at least potentially – competitors. The model and its application offer a
way of quantifying the synergies and strategic risks of coopetive situations
thereby providing an initial approach for measuring competitiveness between
firms and measuring its impact on cooperative sourcing activities (i.e. operation-
alizing coopetition). For example, the simulations showed market structures to be
more consolidated in non-competitive environments. This matches the empirical
observation that cooperative sourcing activities are older and larger in the public
savings sector and the cooperative sector where almost no competition between
the members took place in the past.
As a further contribution, this work helps to “dynamize” the transaction cost
view. Burr (2003) criticized TCE as too “static” to be helpful in organization
analyses. The formal representation of transaction costs and its consideration in
multi-period simulations which enable the disclosing of variations of the level
and impact of different transaction costs over time in different governance modes
helps to dynamize the transaction cost-based view of governance and to give an
answer to Burr’s criticism.
The main contribution from simulation studies can be summarized as the
identification and separation of three different inefficiency sources in coopera-
tive sourcing (Table 69):
Conclusion 399

Inefficiency
Description Reasons
source
Firm is reluctant to cooperatively No sufficient process cost savings (PCE: econo-
Start-up
source business functions or does mies of scale + skill)
problem
not find a coalition to join Too high transaction costs, agency costs, com-
petitive degree, or diseconomies of scope (TCE,
TIC, PAT, PCE)
Uncertainty about the partners’ behavior and thus
Individual Firm is reluctant to switch to the savings (NET: externalities)
lock-in superior coalition Coalition members discard “application” (too
high additional transaction costs (TCE), too low
additional economies of scale (PCE))
Strategic constraint (only start-up problem)
No firm does find a superior coali- Whole system runs into inefficient coalition
System- tion to switch to, although another structure, due to decentralized decision making
wide lock- market structure would be more and due to “too fast” cooperative sourcing activi-
in beneficial from a global perspec- ties (in the simulations, over-reaction sometimes
tive (i.e. higher aggregate benefits) lead to a more fragmented system configuration)
Table 69: Sources for global and/or local cooperative sourcing inefficiencies
In particular, the differentiation of lock-in into individual and system-wide
lock-in, which cannot be resolved by individual decisions, represents a signifi-
cant contribution to theory, which, furthermore, is not solely restricted to the
domain of cooperative sourcing but can also be tested in other contexts which
contain network effects and multiple choice221.
Apart from these structural inefficiency sources, the dynamics of switching
coalitions represent a further behavioral inefficiency source. The higher the
dynamics in the system, the higher is the behavioral uncertainty for the individu-
ally deciding firm. As a consequence, sub-optimal decisions can occur which
either lead to lock-in to sub-optimal local optima or to a longer and repeated
search process (inducing higher cumulated transaction costs) for the “optimal”
coalition. This effect is aggravated by bounded rationality as could be seen in the
effect from altering the coalition selection mechanism (from “efficiency” to
“size” and further to “random”, cf. section 5.3.4.6) – although “more” bounded
rationality partially was able to overcome lock-in effects and lead to slightly
higher efficiency in the final state of the system. As conclusion, we can state that
stronger bounded rationality in a system with externalities leads to higher “be-
havioral inefficiency” (i.e. costly dynamics) but can also lower the resulting
structural inefficiencies (i.e. sub-optimal coalition structure due to lock-in).

221
System-wide lock-in can only occur if the actors can choose from multiple options (e.g. coali-
tions), apart from the decision to do nothing.
400 Conclusion

Of course, this result depends on the way bounded rationality is operational-

ized, i.e. where it affects the individual decision calculus. When the strategic
constraint was aggravated and forced the actors to take more care of their core
competencies, bounded rationality had the opposite effect. Additionally incorpo-
rating this restriction into the decision calculus led to more inefficient results,
compared with the centralized coordination result where the constraint was met
by smaller coalitions instead of less cooperative sourcing activities.
As a final theoretical contribution, the game-theoretical analysis of coalition
stability should be emphasized. Out of several tested cost allocation mechanisms,
only the “proportional cost allocation” could be proven to be stable in any case,
independent of the coalition members’ or applicants’ process costs structures.
Although the other schemes do not necessarily lead to unstable coalitions, ex
ante determining them when founding a coalition bears the problem that, with
new members joining the coalition in later periods, the allocation scheme might
have to be renegotiated between all members. Thus, this analysis contributes to
the sourcing literature not only by extending the traditional 1:1 view on outsourc-
ing to sourcing networks but also by formally introducing the argument of how
cost and benefit allocation affect coalition stability. This therefore provides a
sound theoretical foundation for cooperative sourcing and the analysis of exis-
tence and efficiency of sourcing equilibria.

6.2.2 Managerial Implications

Beside the contributions to research and theory development, the results of this
work yield several implications for practitioners who are responsible for coop-
erative sourcing decisions and activities.
The cooperative sourcing model and the simulation routines provide a valu-
able tool for evaluating cooperative sourcing decisions as long as some parame-
ters can be quantified by the decider and valid ranges for sensitivity analyses can
be specified for the remainder. Although the simulations were based on data
from the SME credit business, the results can be transferred to other banking
business segments along the different process characteristics of the modeled
business functions (cf. the summary section 6.1):
o Depending on the general properties of the focal business function, the re-
sults suggest that the strongest potential not yet exploited derives from util-
izing superior capabilities of partners (economies of skill). As indicated by
the dichotomy between the direct answers of the respondents in the surveys
(3.6.3.1) and the positivist test of PCE arguments (3.6.3.2), additional
economies of scale might often be over-estimated and economies of skill
might be under-estimated by managers.
Conclusion 401

Moreover, the large differences between the effects regarding the finan-
cial outcomes and the corresponding market structures in the simulation
studies show that the size of coalitions has only a minor effect on the overall
cost savings. This again supports the finding that economies of scale play
only a secondary role compared with economies of skill. As long as players
with superior capabilities take over the role of insourcers, significant cost
savings can be realized. By contrast, economies of scale often will be ex-
ploited rather quickly, due to the already high process volumes within the
banks (depending on the cost structure). Of course, if the business function’s
process cost structure predominantly consists of fixed costs, this argumenta-
tion does not hold anymore. In this case, strong outsourcing and switching
activities can be found in both simulations (e.g. strong switching behavior
for risk monitoring) and in reality (e.g. in the payments processing domain).
o For the business functions of the credit process, the survey results showed
rather low actual and potential cooperative sourcing activities. These could
be emulated by the simulations when the strategic constraint was narrowed.
Therefore, we can derive the general finding that – if it emerges – the struc-
ture of cooperative sourcing markets will consist of a rather large number of
small coalitions. Thus, banks can either become insourcers or control their
strategically important business functions more easily. The different regional
cooperative sourcing activities in the public savings banks sector – while, by
contrast, the large national player in the cooperative sector has so far been
unable to attract a sufficient number of credit cooperatives – gives evidence
to this argument.
This “small cluster approach” was also suggested by the centralized
perspective: Stronger strategic boundaries often led to the case that banks
did not enter any coalition although the global optimization showed that or-
ganizing themselves in small coalitions would have been efficient. Hence,
communication between potential partners might be necessary to overcome
the reluctance and to establish small clusters to skim the major part of real-
izable economies of scale.
o A demand for system-wide communication will also be necessary in the case
of a system-wide lock-in: although many banks might have agreed to coop-
erative sourcing there might be too many coalitions that are too small, re-
quiring a certain form of centralized coordination to enforce a more consoli-
dated structure since individual firms are reluctant to change coalitions.
o The simulations showed several decision-relevant factors to be critical with
regard to the system’s cooperative sourcing activities and subsequent dy-
namics. Potential members have to discuss how to control them, in terms of
reducing variance and influencing them in the desired direction:
402 Conclusion

o The homogeneity of process cost structures increases cooperative sourc-

ing activities and thus potential benefits. Banks will more likely agree to
a cooperative sourcing coalition if the potential savings are more
equally distributed.
o The differential impact of different types of transaction costs should be
explicitly evaluated and considered. For example, during the outsourc-
ing process, adoption costs for migration might be much lower than ne-
gotiation costs, although they are more obvious.
o In particular, diseconomies of scope (operationalized as interface costs)
have to be carefully estimated, especially when the firm follows an or-
ganizational strategy of modularizing all of its activities. Economies of
scope (e.g. from shared resources) may already get lost by this intra-
organizational pre-outsourcing measure if not making “clean cuts”
(Beimborn et al. 2005b).
Current trends show that firms have started to consider task inter-
dependencies and increasingly move to “multi-process BPO” where
task interdependencies can otherwise cause problems. In this case, mul-
tiple interrelated processes are outsourced to the same vendor (TPI
2007). This strategy preserves activity-based economies of scope.
o The simulation results for changing coalition process cost structures
(due to technological change) show that technology investments by the
insourcer strongly help to stabilize the coalition. Even low reductions of
variable costs justify investments because the latter scales with the
number of members. In the simulations, this had further a positive effect
on coalition size. An analysis of requirements towards banking IT sys-
tems used in cooperative sourcing coalitions is given by Nitz et al.
(2004), for example.
o Although the sourcing activities in the simulations reacted quite insensi-
tively to changes of the process cost structures in general (e.g. com-
pared to varying transaction costs), they must not be neglected in the
evaluations. The example of extreme descent of workout outsourcing
activities when only slightly increasing the coalitions’ fixed costs ex-
emplifies this very well.
o A cooperative sourcing bandwagon can lead industry to higher effi-
ciency: From a global perspective, cooperative sourcing activities lead
to process standardization. Generally, standardization leads to decreas-
ing adoption costs and thus outsourcing and switching activities becom-
ing less costly. Market dynamics and thus structural efficiency will in-
crease. However, if standardization only takes place in coalition clus-
Conclusion 403

ters, these effects will fail to appear and may also lead to more rigid oli-
gopolistic structures.
The major implication of this research is to focus on the particular aspect of
externalities as part of the cooperative sourcing decision calculus. As can be seen
in the dynamic market for securities processing where switching of coalitions is
not uncommon (and easy due to the high degree of standardization, cf. section
3.4.2), the decision of partners or potentially new coalition entrants can strongly
affect the own cost situation. When firms cannot assume a coalition to be stable
in the long term, they might decide against outsourcing their business functions
(leading to the start-up problem becoming an individual inefficiency dilemma for
them). From a complementary perspective, too many activities may also be prob-
lematic. If a firm changes the coalition, it might achieve more or less higher
benefits but will also induce negative externalities both for the firms remaining
in the former coalition and for the firms which have not yet entered any coali-
tion. This, as a consequence, can affect the firm’s own situation. Thus, the “take-
away” is that own cooperative sourcing behavior always induces a feedback
loop.
Finally, this work provides decision support for the distribution of a coali-
tion’s monetary benefits. The game-theoretical analysis of cost allocation
mechanisms helps managers to evaluate cost and benefit allocation rules in order
to install stable sourcing coalition contracts. While the main goal of the analysis
was the formal proof of conditions of the stability of cooperative sourcing equi-
libria, the section also contained results from a game-theoretical experiment on
cooperative sourcing that indicates that, even in a basically cooperative context,
deciders might be inclined to choose “non-rational” allocation rules leading to
instable sourcing coalitions. The missing predictability of a fundamental and
consistent behavioral pattern, which underlies most agent-based microeconomic
models, is a fundamental problem in reality. Managers should direct attention to
the often unexpected implications of choice of cost/benefit sharing rules in a
cooperative sourcing coalition.

6.3 Validation of the Research Approach

Using computer laboratories has great advantages, as it allows the handling of
highly complex, dynamic, and discrete choice problems as well as the analysis of
simultaneous parameter variations. Methodologically, this approach has been
called the paradigm of agent-based computational economics (ACE) (Tesfatsion
2002a).
As shown in this work, computational models make it possible to analyze
the impact of a variety of local and global influences on system behavior simul-
404 Conclusion

taneously in a way that is otherwise very difficult to accomplish. ACE focuses on

how structures emerge in decentrally organized systems in contrast to being
globally planned and rationally implemented. But this comes with the require-
ment that simulation models must somehow be validated. The role of validation
is to demonstrate that the model is a reasonable representation of the actual sys-
tem. Drawing on Sargent (1998) and Naylor/Finger (1967), the following three-
stage validation approach is used:
o Conceptual model validity: do model assumptions conform to theory and
observations?
o Model verification: is the implementation of the conceptual model into
source code correct?
o Operational validity: are the simulation results reasonable and how do they
fit the real world?
Conceptual Model Validity: Do Model Assumptions Conform to Theory and
Observations?
This step focuses on internal coherence in that causal relations are “reasonable”
and in line with existing theory. The model elements were explicitly derived
from the literature. The following table summarizes the model’s embedding into
the existing literature on outsourcing or on the underlying theories to show the
conceptual validity of the developed model. Literature has also been referred to
during the model development in chapter 4.
Model Assumption or concept (theory) Literature
element
Actor Bounded rationality assumption (e.g. TCE) (Simon 1976; Williamson 1975)
Risk-neutral decision behavior (covering the assump- (DeAngelo 1981; Ewert/Wagenhofer 2003)
tion of spanning and competitivity)
All actors follow the same rationality concept and are (Schelling 1961)
aware of this fact
Decision Coop. sourcing as hybrid governance mode (TCE) (Lammers 2004; Roy/Aubert 2002)
calculus Outsourcing decision based on trade-off between (Clemons et al. 1993)
process costs and transaction costs (PCE, TCE), cost
optimization strategy (PPF)
Insourcing (of partner’s process volumes) vs. out- (Petzel 2003; Roy/Aubert 2002)
sourcing handled as structural equivalents
Outsourcing as investment decision (based on NPV) (Lammers 2004; Lancellotti et al. 2003)
Utilizing superior resources (RDT, PCE) (Grover et al. 1994; Teng et al. 1995)
Considering externalities (NET) Own theory-based contribution
Business Activity-based view (PPF) (Knolmayer 1993; Lacity/Willcocks 2003;
functions Lammers 2004; Porter 1985)
Process Linear cost functions No empirical validation, but used in other ana-
costs lytical models (Barua et al. 1991; Gal-Or 1983;
Quan et al. 2003; Thatcher/Oliver 2001)
Conclusion 405

(or Economies of scale and skill from cooperative (Aubert et al. 1996a; Hamel 1991)
produc- sourcing (PCE)
tion costs) Constant process volume over time (Anderson/Parker 2002)
Trans- Transaction costs split up into costs for negotiation, (Albach 1988; Loh 1994)
action coordination, adoption, interface, and agency (TCE)
costs (in Assumption of efficient relationship building and (Helper/Levine 1992; Ring/Van de Ven 1994)
general) management
Nego- Progressively increasing with growing coalition (Beimborn et al. 2004; Fladung 2006)
tiation Decreasing with rising number of outsourcing Experiences with interorg. relationships de-
and projects (learning effects) crease TC (Alchian 1984; Lacity et al. 1996;
coordina- Tsang 2000). Formal representation of learning
tion costs effects adopted from (Ewert/Wagenhofer 2003)
Coordination costs appear in each outsourcing period (Buvik/Gronhaug 2000; Williamson 1985;
Zhang/Liu 2005)
Increasing with task complexity (more complex (Loh 1994; Rouse/Corbitt 2004)
contracts required) (TCE, TIC, PAT)
Adoption Outsourcer’s organization has to adopt the in- (Cash/Konsynski 1985; Rouse/Corbitt 2004)
costs sourcer’s process (process standardization) (TCE)
Process similarity reduces adoption costs (Porter/Fuller 1986; Wüllenweber et al. 2008)
Complexity drives adoption costs (Aubert/Patry 1998; Aubert et al. 1996b)
Interface Task interdependence parameter (vertical economies (Bruch 1998; Knolmayer 1993; Langlois/Ro-
costs of scope) (TCE, PCE) bertson 1992; Meyer/Schumacher 2003;
Roy/Aubert 2002). Relatedness as risk factor
and driver for coordination effort (Bahli/Rivard
2004; Earl 1996; Lacity et al. 1996). Cost of
building, maintaining, operating interfaces
(Bahli/Rivard 2003; Porter/Millar 1985).
Problem of technological (“asset”) indivisibility
(Teece 1980b) or skill indivisibility (Hamel
1991; Hitt et al. 1993; Lei/Hitt 1995). Impact of
task interdependencies in outsourcing decisions
(Lacity et al. 1996; Roy/Aubert 2002). Formal
representation from (Knolmayer 1993)
Agency Increasing with task complexity (inherent contract Complexity drives control costs, agency costs in
costs completeness and bad measurability of provider general due to inherent contract incompleteness
performance) (TCE, TIC, PAT, RT) and bad measurability of provider performance
(Cheon et al. 1995; Lacity/Hirschheim 1993a;
Nam et al. 1996; Poppo/Zenger 1998)
Increasing with degree of competition between Own assumption
insourcer and outsourcer (coopetition)
Strategic Business neighborhood dimensions (PPF) (Porter 1985)
risk Strategic risk from cooperation (Afuah 2000; Beasley et al. 2004; Lacity/Will-
cocks 1995; Nueno/Oosterveld 1988; Oliver
1990; Ring/Van de Ven 1994)
Strategic Save a sufficient stack of core competencies in-house (Prahalad and Hamel 1990)
constraint (CCV, RBV) Formal representation from (Knolmayer 1993)
Table 70: Conceptual model validation
406 Conclusion

Model Verification: Is the Implementation of the Conceptual Model Correct?

In order to ensure a correct implementation of the simulation model, structured
walks through the program code have been carried out. Parts of the implementa-
tion were done by two different developers with cross-checking the routines.
In his simulations handbook, Bratley (1987, p. 9) further calls for (a) modu-
lar testing (Does each subroutine fulfill its task correctly?), (b) sensitivity testing
(Is the behavior of the model sensible to ceteris paribus parameter changes?),
and (c) stress testing (Do strange parameter values “blow up” the model in an
understandable fashion?).
The different subroutines of the simulation tool (e.g. decision calculus, sin-
gle cost functions, coalition entering and exiting, genetic algorithm with all of its
subroutines) have each been tested by tracking the output for a known set of
input parameters. Sensitivity analyses have partially been documented in the
simulation results section (5.3.4). In cases, where the model reacted non-sensibly
to parameter changes, additional data was extracted during the execution of the
sub-routines to ensure a valid execution and the correct implementation of the
formal model and in order to explain those results. Furthermore, for stress testing
the sensitivity analyses often covered a much wider (i.e. unrealistic) range of
parameter values than documented in the simulation results section.

Operational Validity: Are the Simulation Results Reasonable and How do they
Fit Reality?
For validating simulation models, this is the most problematic step because
simulation results should reflect reality when feeding the simulation routines
with real data (Bossel 2004, 61).
In economic models, this aim usually cannot be fulfilled (as Bossel’s macro-
economic simulations themselves cannot, either). Simulating organizational
behavior is a very ambitious task and each model of the deciders’ decision calcu-
lus can only be a very weak and imperfect abstraction. Bratley summarizes the
following types of approximation which represent the main sources for diver-
gences between simulation results and the real world (summarized from Bratley
1987, pp. 9-10):
o Restricting the boundary of the model and ignoring everything outside that
is not an explicit input.
o Neglecting factors believed to be unimportant (see also Foss et al. 1995, 11).
o Functional approximation: highly nonlinear functions, themselves often
approximate, are approximated by simpler ones.
Conclusion 407

o Distributional approximation: real-world problem distributions, which are

known only approximately, are frequently approximated by simple distribu-
tions (e.g. normal distribution).
o Independence: various stochastic parameters are assumed to be statistically
independent.
o Stationarity: assuming that parameters and other features of the system do
not vary over time.
o State assumption: the past is statistically irrelevant for predicting the future
behavior.
o Aggregation: the most pervasive type of approximation is aggregation. Sev-
eral of something are treated as one:
o Temporal aggregation: all events occurring during a period are as-
sumed to occur simultaneously.
o Cross sectional aggregation: multiple organizational units are
treated as one.
o Resource aggregation: a production system is treated as one unit.
o Properties aggregation: individuals are aggregated in classes of in-
dividuals.
As with most simulations and analytical approaches, this work also had to
apply almost all of these approximation steps. Nevertheless, since we used an
ACE approach, the system behavior could be handled more realistically than in
“traditional” simulation studies which are based on differential equations. Thus,
no approximation was necessary in an essential part of the model (i.e. how indi-
vidual decision behavior leads to the system’s degree of outsourcing activities)
and more opportunities for incorporating heterogeneity and individual properties
have been available.
In our case, a direct matching between the real-world behavior and the simu-
lation studies was neither possible nor intended. Due to the still poor empirical
basis which was available to feed the simulations, a sufficient representation of
reality or a forecast of future real-world behavior is mostly impossible. Instead,
the simulations intended to uncover structural effects which help us to under-
stand phenomena and to get sensitized for potential problems in a cooperative
sourcing scenario. Thus, the step of operational validation can be reduced to the
question if the results are reasonable.
In the results section (5.3.4) and in the previous sections of this chapter, sev-
eral arguments have already been discussed to support the operational validity.
Furthermore, the basic discrepancy between the moderate to high cooperative
sourcing activities in the model versus the rather low (even the only intended)
rate of outsourcing in reality could be explained by a more restrictive “strategic
constraint” and higher transaction costs than in the basic simulation setting.
408 Conclusion

When varying the strategic constraint (which obviously could not be parameter-
ized based on empirical data), different market configurations could be found
(including the reality). Additionally, the empirical studies showed that many
practitioners perceived potential cost savings to be almost non-existent, although
the survey uncovered large process cost differences between the different banks
(which were considered in the simulations). In accordance with the simulation
studies, financial service industries in other countries show that cooperative
sourcing activities are possible and beneficial in the credit processing domain.
Beside some key figures which match with empirical data (relative level of
transaction costs, frequency of different business models, relative frequency of
sourcing activities), the robustness of the results, in particular, can be used as a
further argument for operational validity:
o Varying the artificial parameters of the simulation control to establish new
coalitions (initCoalSize, maxCoalBuilding, cf. section 5.3.2) did not signifi-
cantly change the resulting monetary effects and market structures.
o A variation of the spread of the process cost ratios (fixed / variable costs),
which have been estimated from empirical data by using three different ap-
proaches (cf. section 3.6.2.5), did not lead to a change of the simulation re-
sults.
o Varying the spread of the overall seed of random numbers used in the simu-
lations did not lead to a significant change of the results.
The two latter arguments additionally show that the simulated system had
the necessary size to deliver inherently stable results despite the high number of
stochastic variables.

6.4 Limitations
In addition to the general limitations of simulation studies which have been dis-
cussed in the previous section, this section discusses concrete limitations of the
developed model and the chosen approach.

Model Limitations
This section reflects the limitations of the cooperative sourcing model (CSM),
developed in chapter 4. Some simplifications were made that helped to signifi-
cantly reduce mathematical complexity and/or that did not affect factors that
were primarily relevant in order to answer the research questions. Nevertheless,
tackling some of the limitations is a promising starting point for future research.
The first limitation of the model which has to be mentioned is the simplified
market structure. The model only takes firms of the same industry into account;
Conclusion 409

it allows neither the appearance of new entrants nor does it consider different
forms of outsourcing such as outsourcing to providers from other industries or
multi-step subcontracting. Furthermore, the firms are modeled as cost-
minimizing but not profit-maximizing agents. Consequently, it was not necessary
to implement market mechanisms explicitly. It was assumed that banks follow
their business model of purchasing banking services and do not see insourcing as
an explicit revenue source, i.e. business model, but do this primarily for cost-
cutting reasons and thus becoming more profitable. Consequently, the model also
does not distinguish different strategy types in a Porterian sense but assumes that
all banks try to optimize costs in processing their business functions. Strategies
of technology or innovation leadership leading to differentiation are not handled
by the model.
Regarding the business functions, the activity-based perspective of the CSM
neither explicitly considers business processes or a product-oriented perspective
nor does it follow a hierarchical understanding of business functions (different
levels of granularity and interdependencies, e.g. as in (Beimborn et al. 2005b)).
Nevertheless, the model implicitly captures the aggregate processual relation-
ships between business functions by the task interdependence measure. More-
over, it can be adopted to different application domains and thus to different
levels of business function granularity – e.g. either covering the whole credit
business or only tasks of the processing/servicing process step. Interdependen-
cies to other (not modeled) business functions could be considered by incorporat-
ing a virtual “residual business function” that represents all inhouse activities
which are not part of the modeled domain. All task interdependencies can then
be connected to this proxy.
Focusing on the cost structures, the assumption of linear cost functions
represents a limitation since it is not based on empirical evidence, although
managerial cost accounting approaches usually make the same assumption.
There would be arguments for assuming a declining cost function which not only
leads to economies of scale from fixed cost reduction but also from the func-
tion’s shape. For example, if process volume increases, certain investments in IT
are necessary. This stepwise function can be approximated either by a linear
relationship (as done in the CSM) or – more likely – by a declining function
because basic infrastructural investments have been made and therefore the addi-
tional investments decrease. As a consequence, the results of the optimization
and simulations would tend towards slightly larger coalitions.
A further issue is the constancy of parameters and costs during a sourcing
relationship. There are outsourcing decision models which handle the volatility
of process volumes over time (e.g. Lammers 2005). Cost variabilization instead
of cost reduction is a major advantage of outsourcing often cited by practitioners
410 Conclusion

(Accenture 2002; Alexander and Young 1996; Lacity et al. 1996). However, in
each outsourcing decision, assumptions need to be made about the future process
volume; therefore a constant value seems to be a simple but valid proxy for the
objectives of our model development. Moreover, the model can easily be ex-
tended in order to consider more complex estimators about future process vol-
ume progress.
Another problem is the stability of the process cost function when joining a
sourcing relationship. Usually, learning effects would cause average process
costs to decrease over time (dynamic economies of skill, cf. section 2.1.1) (Lam-
berti and Pöhler 2004) but volatile process volumes may also cause them to in-
crease. Volume peaks force a sourcing provider to increase capacity, which can-
not be cut back as fast as it is installed. This leads to a feather-like upward shift
of the process costs (ratchet effect). Although it would not be a major problem to
extend the model by these effects, a much more sophisticated parameterization
would be necessary which would require a non-realizable amount of empirical
work in order to adequately estimate these parameters. As a first answer, the
simulations in section 5.3.4.2 considered variations of the process cost function
parameters when the cooperative sourcing coalition is established.
Changing the focus to the constancy of transaction costs, theory of incom-
plete contracts and relationship theories argue, on the other hand, that agency
costs and coordination costs are substituted by norms such as trust over time (cf.
section 2.1.8) (Tsang 2000). Langlois (1992) – in contrast to other authors – even
assumes transaction costs to be only temporarily existent. Therefore, the corre-
sponding functions should lead to decreasing transaction costs throughout the
sourcing duration TCoSo. However, to avoid additional cut-rate parameters, we
decided against it but implemented learning effects regarding the outsourcing
process itself, i.e. former outsourcing experiences reduce transaction costs of
future ones. Moreover, in reality transaction costs are commonly estimated on a
very weak basis and often under-estimated (Barthélemy 2001; Bettis et al. 1992;
Dibbern et al. 2003). Hence, the simulations particularly considered wide varia-
tions of the transaction cost factors during sensitivity analyses.
Agency costs are modeled very primitively, compared with the mathemati-
cal models which directly focus on this particular issue and on the principal-
agent relationship. The main issue is that the basic driver for agency costs, i.e.
the level of interest divergences between principal and agent, is not explicitly
modeled. However, because we focus on cooperative sourcing – i.e. process
volumes of principal (outsourcer) and agent (insourcer) are assumed to be han-
dled by the same resources, which in turn leads to comparably low interest di-
vergences – the model omits this factor.
Conclusion 411

As a further point of critique regarding the decision calculus, the selection of

potential coalitions has to be considered. Due to its level of abstraction, the
model cannot incorporate important arguments from relationship theories for the
formation of coalitions such as preexisting friendship ties, institutional mandate,
resource dependency, prior economic relationships, social ties, or the mediating
activities of third parties (“venture capitalists, corporate sponsors, or investment
bankers who act as ‘cupids’”) (Ring and Van de Ven 1994, 100-101).
Finally, the major shortcoming of the decision calculus is the purely eco-
nomically reasoned decision function with regard to the non-consideration of
non-economic outsourcing drivers and inhibitors in the firm’s objective. While
some models include the “strategic value” of a decision or other strategic argu-
ments – such as the goal of reaching a stronger focus on core competencies (Pra-
halad and Hamel 1990) or the objective to raise market power by bundling re-
sources (Porter 1985) – in the agent’s objective without explicitly stating how to
handle the monetarization (e.g. Tsang 2000, Lammers 2005), our model tries to
avoid this critical issue. As in many other models (e.g. (Aksin et al. 2004, Knol-
mayer 1993), we chose a purely economically reasoned decision function and
captured strategic issues partially by the strategic constraint which restricts out-
sourcing decisions based on strategic arguments (cf. section 4.2.3.4). Although it
is often argued that costs are no longer the primary argument for outsourcing,
deciders usually have to calculate a positive business case before suggesting the
outsourcing of a certain business function and Lacity and Willcocks (1995) argue
that economic decisions usually also follow an economic rationale (p. 204). Fur-
thermore, studies showed high correlations between cost efficiency from out-
sourcing and non-financial advantages such as focusing on core competencies
(Miranda and Kim 2004).

Limitations of the Approach

As main approach of this research, we applied a combination of agent-based
computational simulations based on empirical data gathered from the SME credit
business of the German banking industry.
The developed model provides higher complexity than earlier formal models
on outsourcing decision behavior to enable the incorporation of empirical data.
Nevertheless, the approach certainly cannot fully represent the behavior of com-
plex organizations.
The high complexity of the model prohibited conducting analytical equilib-
rium analyses to answer the third research question (“What are the resulting
market effects from cooperative sourcing behavior?”). As a consequence, the
simulative ACE approach was applied. Moreover, no exact optimization routine
412 Conclusion

could be used to determine optimal sourcing configurations as an efficiency

benchmark for the simulation results. Instead, a genetic algorithm was applied.
The incorporation of empirical data has its own weaknesses. Due to strong
restrictions in the questionnaire design, the data which was requested from the
respondents was insufficient for parameterizing all necessary parts of the model
and our approach for determining fixed and variable process costs also has its
own weaknesses, as discussed in section 3.6.2.5. Nevertheless, as a pretest had
shown, quantitative parts of the needed data such as process cost functions (e.g.
proportion of fixed and variable costs) for the different business functions were
generally not available in the banks at all; similar results occurred for transaction
costs resulting from BPO. Only very few banks provided (at least aggregate)
quantitative data in the questionnaire. Case study results partially helped to fill
these “data holes” and sensitivity analyses were used to increase the reliability of
the results.
Quantitative data can only be interpreted very carefully. Requesting cost
data etc. cannot ensure that all respondents refer to the same cost basis (e.g.
which part of the process, which products). Moreover, we then cannot expect
that firms outsource parts of the business processes without internally optimizing
(although this is not a seldom occurrence). Consequently, the high heterogeneity
of actual process costs which lead to high savings from cooperative sourcing in
the simulation studies can only represent an upper border. Thus, we did not try to
interpret and transfer the monetary findings back from the simulations to real
world.
Despite the limitations, this research – the formal development as well as the
simulation studies – proposed an exploratory contribution towards theory devel-
opment and derived implications for practitioners. The aim was not to directly
reproduce or forecast the dynamics of the German banking industry but rather to
provide an analysis of typical behavioral and market structure effects. The sum-
mary section of this chapter gave an overview of how the simulation results can
be generalized, based on some basic business function properties, towards busi-
ness domains other than the one investigated empirically.
Nevertheless, the limitations as well as the issues left open by this work of-
fer a wide range of opportunities for future research.

6.5 Further Research

Extending the Cooperative Sourcing Model
The cooperative sourcing model (CSM) developed in chapter 4 and the corre-
sponding implementation as simulation model have a modular structure which
Conclusion 413

offers the opportunity to exchange parts without affecting its integrity and appli-
cability. The following items represent some ideas of promising model exten-
sions:
o Considering additional cost components: For example, the model does not
cover possible changes of the bank’s credit risk or market risk structure. In
the special case of outsourcing the refinancing of a particular credit product
(i.e. selling the credit), the credit risk structure and, thus, the capital costs
will change. Risk costs are a substantial part of the overall credit costs (cf.
Figure 68). Holzhäuser et al. (2005) provide a BPO decision support model
which takes this effect into account. The decision calculus of their model
can be incorporated into the CSM. Another argument is the trade-off be-
tween risk costs and process costs which should be optimized (Hölzer 2004,
236). By increasing automation (often related to cooperative sourcing),
process costs will be partially reduced at the cost of increasing credit risk
(less strict control by humans leading to less effective risk monitoring) (Ku-
ritzkes et al. 2000, 46).
o In IT outsourcing research, imitative behavior has often been cited as an
important driver of outsourcing decisions (Lacity and Hirschheim 1993b;
Loh and Venkatraman 1992b; Rouse and Corbitt 2004). In order to capture
this argument, the concept of (bounded) rational choice, as implemented in
the CSM, can be replaced by some form of adaptive behavior as is in the co-
operation theory, for example (Axelrod 1984, 1987, 2000).
o In its internal structure, the CSM can be extended by an activities x re-
sources matrix which links the business functions to underlying required re-
sources. In this way, economies of scope from resource sharing can be more
precisely modeled and considered in outsourcing decisions. For example, an
IT system might represent a resource used for different activities. Thus, the
outsourcing of a particular business function will not necessarily lead to the
divestment of the related resources.
o A further promising extension of the agents’ decision functions is the con-
cept of reputation. It is often observed that reputation and trust acquire fun-
damental importance in B2B relationships. According to Mui et al. (2002b),
reputation is a “perception that an agent creates through past actions about
its intentions and norms” and trust is a “subjective expectation an agent has
about another’s future behavior based on the history of their encounters”. It
can be shown that reputation reduces the complexity of the decision process
through a better estimation of the likelihood of cooperation failure (Marsh
1992). Reputation has already been formalized (e.g. Carter et al. 2002) and
has been repeatedly incorporated into multi-agent systems (see (Mui et al.
2002a) for an overview). Thus, concepts from the relationship theories (sec-
414 Conclusion

tion 2.1.8) have already been integrated into formal simulation models.
However, although reputation has a significant influence on economic deci-
sions, only few research projects have addressed the impact of reputation on
B2B cooperation (Franke et al. 2005b).
o As a more general enhancement, the model can be extended by providing a
wider range of possible governance modes. While the CSM only allows co-
operative sourcing and pure insourcing, outsourcing to other industries and
divestments can be incorporated, for example following the alternatives of
Lammers’ formal sourcing governance decision model (Lammers 2004).
This would require the implementation of market and negotiation mecha-
nisms and ultimately lead to a unified micro-economically founded instru-
ment for investigating sourcing economies.

Game-Theoretical Analysis of Conditions for Stable Coalitions

In section 5.1, a game-theoretical analysis was conducted which analyzed the
pre-conditions for stable coalitions. It determined which ex ante defined cost
allocation mechanisms always result in all potential coalition members entering
the coalition. While the analysis used the core as concept of stable allocation
vectors, further research will have to test the cost allocation mechanisms against
other concepts of defining stable imputations such as stable sets and the kernel
(Osborne and Rubinstein 1994). In a next step, further cost allocation mecha-
nisms can be tested: Lemaire (1984) refers to possible alternatives (e.g. the nu-
cleolus (Schmeidler 1969), the proportional nucleolus (Young et al. 1980), or
simple allocation rules from corporate cost accounting). In a further step, empiri-
cal evidence should be collected from cooperative sourcing cases.
The game-theoretical experiment showed counter-intuitive behavior in some
cases. The situation that players do not accept cost savings because other parties
would benefit “more” can also be observed in real cooperative settings (Weitzel
2003) and runs contrary to the basic behavioral assumptions of most economic
models. Future research has to refine and repeat the experiments with different
incentive mechanisms and a more thoroughly developed experiment design in
order to ensure reliable results. Furthermore, the groups of players involved in
the experiment should have different degrees of experience (e.g. students vs.
managers). If in these very simple settings, where players have to take into ac-
count only four parameters and apparently have the same objective, inefficient
solutions are still knowingly accepted, which means that the consequences for
economic modeling have to be reconsidered.
Conclusion 415

Further Research on Cooperative Sourcing in General

Finally, two distinct topics should be highlighted which will be major problem
areas and thus represent great research opportunities within the field of coopera-
tive sourcing in the next years.
The first topic is IT-driven and a current hype term in the practitioners’ and
academics’ world: service-oriented architectures (SOA). These are widely dis-
cussed and firms start to rebuild their IT infrastructures based on the SOA para-
digm although research has so far failed to offer adequate methods for actually
evaluating the benefits from investments in SOA.
Extrapolating the SOA trends to the future will lead to some major conse-
quences for cooperative sourcing. First, units of sourcing decisions (i.e. the busi-
ness function) are assumed to become more granular (Hagel 2007). Second,
encapsulation of activities at a granular level will facilitate their standardization
(Richter et al. 2006). Thus, cooperative sourcing and switching behavior will
become cheaper and easier, leading to more dynamic and competitive markets.
By contrast, this more modular and dynamic smart sourcing (Fairchild 2003;
Prokein and Faupel 2006) will also require sophisticated sourcing portfolio man-
agement mechanisms in order to be able to handle this increased organizational
complexity.
The related issue of standardizing business activities or business processes
is also a major field of promising research. While IT standardization research has
recently become an accepted domain in the IS community – as reflected in up-
coming conferences, journals (e.g. JITSR – Journal of IT Standards & Standardi-
zation Research) and special issues (e.g. MIS Quarterly in 2006) – the topic of
process standardization is still diffuse and unattended (Ungan 2006) although it
has strong relevance for reaching firm-internal cost efficiency (Hammer and
Stanton 1999; Manrodt and Vitasek 2004) and even more for realizing BPO
(Davenport 2005; Wüllenweber et al. 2008). How can process standardization be
classified and measured? What is then the impact of process standardization or
the adoption of process standards on value generation in general and on coopera-
tive sourcing in particular? Adopting the model developed in this thesis can lead
to a framework for tackling these questions.
As the second major field of further research, relationship governance has
already become a focal point in outsourcing research in general (Goles and Chin
2005). Although we argued that interests of insourcers and outsourcers will be
more congruent in cooperative sourcing than in traditional outsourcing relation-
ships there is still much room for conflict in the operational business. One indica-
tor is the low success of current credit outsourcing activities in Germany (Figure
77 on p. 202). Within the last 8-10 years, publications on “partner-based out-
sourcing” (Willcocks and Kern 1998) have appeared and the importance of out-
416 Conclusion

sourcing relationship management has been emphasized. Firms realized that a

contract-based relationship is not sufficient to lead to a successful symbiosis
(Lee et al. 2003; Willcocks and Kern 1998; cf. also sections 2.1.4 and 2.1.8 on
the theory of incomplete contracts and relationship theories).
In the cooperative sourcing domain, this leads to the questions of how to ef-
fectively design a cooperative sourcing governance mode and how to establish a
high relationship quality and stabilize it in the long term. The side conditions are
that business functions become more granular and markets become more dy-
namic, resulting in less stable coalition structures and higher frequencies of coa-
lition changes, and thus more decisions to be made. Moreover, coopetition is
likely to be involved (Afuah 2000). Underlying research tasks are to specify how
relationship quality can be measured and evaluated. First attempts in the IT out-
sourcing context can be found in (Beimborn and Blumenberg 2007; Goles and
Chin 2005; Kern and Willcocks 2002; Kern 1997; Lee and Kim 1999; Marcolin
and Ross 2005).
Taking all these promising research areas into account, cooperative sourcing
in particular and business process outsourcing in general still show to be a very
young and unexplored research field. This work tried to shed some light on es-
sential research questions – the reasons and the outcome facets of cooperative
sourcing. Apart from dedicated and more sophisticated advancements in these
areas, which hopefully will be made by future researchers, many other research
questions have remained open regarding the exploration of an exciting phenome-
non.
References 417

References

Aareal: Der Kreditservicer – Unternehmenspräsentation. Aareal Hypothekenmanagement Informati-

onsforum Kreditservicing, Frankfurt am Main, 16 Nov 2005.
Accenture: Outsourcing 2007. Accenture, Kronberg, 2002.
Adams, R.M., Bauer, P.W., and Sickles, R.C.: Scope and scale economies in Federal Reserve pay-
ment processing. Federal Reserve Bank of Cleveland, Cleveland (OH), 2002.
Ade, B., and Moormann, J.: Dekonstruktion der Kreditwertschöpfungskette. In: Sourcing in der
Bankwirtschaft, W. Achenbach, J. Moormann and H. Schober (eds.), Bankakademie-Verlag,
Frankfurt am Main, 2004, pp. 153-174.
Adelman, M.A.: Concept and statistical measurement of vertical integration. In: Business concentra-
tion and price policy, G.J. Stigler (ed.), Princeton University Press, Princeton (NJ), 1955.
Afuah, A.: How much do your co-opetitors' capabilities matter in the face of technological change.
In: Strategic Management Journal (21:3) 2000, pp. 397-404.
Afuah, A.: Redefining firm boundaries in the face of the Internet: are firms really shrinking? In:
Academy of Management Review (28:1) 2003, pp. 34-53.
Akerlof, G.A.: The market for "lemons": quality uncertainty and the market mechanism. In: Quar-
terly Journal of Economics (84:3) 1970, pp. 488-500.
Aksin, O.Z., Vericourt, F.d., and Karaesmen, F.: Call center outsourcing contract design and choice.
Fuqua School of Business, Duke University, Durham (NC), 2004,
http://faculty.fuqua.duke.edu/ ~fdv1/bio/OUT31_10_04.pdf (as of 30 June 2006).
Alavi, M., Carlson, P., and Brooke, G.: The ecology of MIS research: a twenty year status review.
10th International Conference on Information Systems, Boston (MA), 1989, pp. 363-375.
Albach, H.: Kosten, Transaktionen und externe Effekte im betrieblichen Rechnungswesen. In: Zeit-
schrift für Betriebswirtschaft (58:11) 1988, pp. 1143-1170.
Alchian, A.: Uncertainty, evolution and economic theory. In: Journal of Political Economy (58:2)
1950, pp. 211-221.
Alchian, A., and Demsetz, H.: Production, information costs, and economic organization. In: Ameri-
can Economic Review (62:5) 1972, pp. 777-795.
Alchian, A., and Woodward, S.: The firm is dead: long live the firm. A review of Oliver E. William-
son's "The economic institutions of capitalism". In: Journal of Economic Literature (26:1)
1988, pp. 65-79.
Alchian, A.A.: Specificity, specialization and coalitions. In: Journal of Institutional and Theoretical
Economics (140:1) 1984, pp. 34-49.
Aldrich, H.: Resource dependence and interorganizational relations: relations between local employ-
ment service office and social services sector organizations. In: Administration and Society
(7:4) 1976, pp. 419-455.
Alexander, M., and Young, D.: Strategic outsourcing. In: Long Range Planning (29:1) 1996, pp. 116-
119.
Allen, S., and Chandrashekar, A.: Outsourcing services: the contract is just the beginning. In: Busi-
ness Horizons (43:2) 2000, pp. 25-34.
Allison, G.: Essence of decision. Little, Brown & Organization, Boston, MA, 1971.
Alms, W.: Einsparpotentiale durch Business-Process- und IT-Outsourcing. In: Börsenzeitung, Son-
derbeilage "IT-Zukunft der Banken", 15 Mar 2003.
Alstyne, M.v., and Brynjolfsson, E.: The net effect: modeling and measuring the integration of
electronic communities. MIT Sloan School, Cambridge (MA), 1997.
418 References

Alt, R., and Zerndt, T.: Gestaltung und Bewertung – Sourcing-Modelle in der Finanzbranche. In: Die
Bank (45:8) 2005a, pp. 66-70.
Alt, R., and Zerndt, T.: Process Sourcing bei Banken: Die nächste Vernetzungsstufe. In: Business
Intelligence Magazine (1) 2005b, pp. 50-53.
Amel, D., Barnes, C., Panetta, F., and Salleo, C.: Consolidation and efficiency in the financial sector:
a review of the international evidence. In: Journal of Banking and Finance (28:10) 2004, pp.
2493-2519.
Amit, R., and Schoemaker, P.: Strategic assets and organizational rent. In: Strategic Management
Journal (14) 1993, pp. 33-46.
Anderson, J., and Narus, J.: A model of distributor firm and manufacturer firm working partnerships.
In: Journal of Marketing (54:1) 1990, pp. 42-58.
Anderson, E.G., and Parker, G.G.: The effect of learning on the make/buy decision. In: Production
and Operations Management (11:3) 2002, pp. 313-339.
Ang, S.: Towards conceptual clarity of outsourcing. In: Business process reengineering: information
systems opportunities and challenges, B.C. Glasson (ed.), Elsevier Science (North-Holland),
Amsterdam, 1994, pp. 113-126.
Ang, S., and Cummings, L.: Strategic response to institutional influence on information systems
outsourcing. In: Organization Science (8:3) 1997, pp. 235-256.
Ang, S., and Straub, D.W.: Production and transaction economies and IS outsourcing: a study of the
U.S. banking industry. In: MIS Quarterly (22:4) 1998, pp. 535-552.
Angermüller, N.O., Eichhorn, M., and Ramke, T.: MaRisk – Der Nebel lichtet sich. In: Zeitschrift für
das gesamte Kreditwesen (8), 15 Apr 2005, pp. 396-398.
Anupindi, R., and Bassok, Y.: Centralization of stocks: manufacturer vs. retailers. In: Management
Science (45:2) 1999, pp. 178-191.
Apte, U.M.: Global outsourcing of information systems and processing services. In: The Information
Society (7) 1990, pp. 287-303.
Apte, U.M., Sobol, M.G., Hanaoka, S., and Shimada, T.: IS outsourcing practices in the USA, Japan,
and Finland: a comparative study. In: Journal of Information Technology (12:4) 1997, pp. 289-
304.
Aram, J.D.: The paradox of interdependent relations in the field of social issues management. In:
Academy of Management Review (14:2) 1989, pp. 266-283.
Arrow, K.: Classificatory notes on the production and transmission of technological knowledge. In:
American Economic Review (59:2) 1969, pp. 29-35.
Arthur, W.B.: Competing technologies and lock-in by historical small events: the dynamics of alloca-
tion under increasing returns. International Institute for Applied Sciences, Laxenburg, Austria,
1983.
Arthur, W.B.: Competing technologies, increasing returns, and lock-in by historical events. In:
Economic Journal (99:March) 1989, pp. 116-131.
Artz, K.W., and Brush, T.H.: Asset specificity, uncertainty, and relational norms: an examination of
coordination costs in collaborative strategic alliances. In: Journal of Economic Behavior and
Organization (41:4) 2000, pp. 337-362.
Astley, W.G.: Administrative science as socially constructed truth. In: Administrative Science Quar-
terly (30:4) 1985, pp. 497-513.
Aubert, B.A., and Patry, M., Rivard, S.: Assessing the risk of IT outsourcing. 31st Hawaii Interna-
tional Conference on System Sciences (HICSS-31), Honolulu (HI), 1998.
Aubert, B.A., Dussault, S., Rivard, S., and Patry, M.: Managing the risk of IT outsourcing. 32nd
Hawaii International Conference on System Sciences (HICSS-32), Wailea (HI), USA, 1999.
Aubert, B.A., Houde, J.-F., Patry, M., and Rivard, S.: Characteristics of IT outsourcing contracts. 36th
Hawaii International Conference on System Sciences (HICSS-36), Waikoloa, Big Island (HI),
USA, 2003.
References 419

Aubert, B.A., Patry, M., and Rivard, S.: The outsourcing of IT: autonomous versus systemic activi-
ties. 28th Annual Meeting of the Decision Sciences Institute, San Diego (CA), 1997, pp. 809-
812.
Aubert, B.A., Patry, M., and Rivard, S.: A framework for information technology outsourcing risk
management. In: The DATA BASE for Advances in Information Systems (36:4) 2005, pp. 9-
28.
Aubert, B.A., Patry, M., Rivard, S., and Smith, H.: IT outsourcing risk management at British Petro-
leum. 34th Hawaii International Conference on System Sciences (HICSS-34), Wailea, Maui
(HI), USA, 2000.
Aubert, B.A., Rivard, S., and Patry, M.: Development of measures to assess dimensions of IS opera-
tion transactions. In: Omega Journal (24:6) 1996a, pp. 661-680.
Aubert, B.A., Rivard, S., and Patry, M.: A transaction cost approach to outsourcing behavior: some
empirical evidence. In: Information and Management (30:2) 1996b, pp. 51-64.
Aubert, B.A., Rivard, S., and Patry, M.: Managing IT Outsourcing risk: lessons learned. In: Informa-
tion Systems Outsourcing: enduring themes, emergent patterns and future directions, Hirsch-
heim, R.A., Heinzl, A., Dibbern, J. (ed.), Springer, Heidelberg et al., 2002, pp. 155-176.
Auguste, B.G., Hao, Y., Singer, M., and Wiegand, M.: The other side of outsourcing. In: McKinsey
Quarterly (2002:1) 2002, pp. 52-63.
Aumann, R.J., and Maschler, M.B.: Repeated games with incomplete information. The MIT Press,
Cambridge (MA), 1995.
Axelrod, R.: The evolution of cooperation. Basic Books, New York (NY), 1984.
Axelrod, R.: The complexity of cooperation. Princeton University Press, Princeton (NJ), 1987.
Axelrod, R.: Advancing the art of simulation in the social sciences. In: Simulating social phenomena,
R. Conte, R. Hegselmann and P. Terna (eds.), Springer, Heidelberg et al., 1997.
Axelrod, R.: On six advances in cooperation theory. In: Analyse & Kritik (22:1) 2000, pp. 130-151.
Axelrod, R., and Dion, D.: The further evolution of cooperation. In: Science (242:4884), 09 Dec
1988, pp. 1385-1390.
Axelrod, R., and Hamilton, W.D.: The evolution of cooperation. In: Science (211:4488), 27 Mar
1981, pp. 1390-1396.
Bacharach, S.B.: Organizational theories: some criteria for evaluation. In: Academy of Management
Review (14:4) 1989, pp. 496-515.
BaFin: Outsourcing of operational areas to another enterprise pursuant to section 25a(2) of the Bank-
ing Act. Circular 11/2001, Federal Financial Supervisory Authority (BaFin), Bonn, Germany,
2001a, http://www.bafin.de/rundschreiben/93_2001/rs11_01en.htm (as of 03 Sep 2006).
BaFin: Bundesaufsichtsamt für das Kreditwesen gibt Standards für die Auslagerung wesentlicher
Bereiche von Kredit- und Finanzdienstleistungsinstituten auf andere Unternehmen bekannt.
Federal Financial Supervisory Authority (BaFin), Bonn, Germany, 2001b,
http://www.bafin.de/presse/pm01/ba_011206.htm (as of 30 June 2003).
BaFin: Mindestanforderungen an das Kreditgeschäft der Kreditinstitute (MaK). Circular 34/2002,
Federal Financial Supervisory Authority (BaFin), Bonn, Germany, 20 Feb 2002. In German:
http://www.bafin.de/rundschreiben/92_2002/r280202.htm (as of 17 July 2006).
BaFin: Vermerk: Kreditfabriken – Aufsichtliche Rahmenbedingungen und Anforderungen. Federal
Financial Supervisory Authority (BaFin), Bonn, Germany, 12 Dec 2003. In German:
http://www.bafin.de/sonstiges/031212_kreditfabriken.htm (as of 17 July 2006).
BaFin: Minimum requirements for risk management (MaRisk). Circular 18/2005, Federal Financial
Supervisory Authority (BaFin), Bonn, Germany, 04 May 2006. Final version in German:
http://www.bafin.de/marisk/060504_rs.htm. Preliminary version in English:
http://www.bundesbank.de/download/bankenaufsicht/pdf/051220_en.pdf (as of 17 July 2006).
Bagozzi, R.P., and Yi, Y.: On the evaluation of structural equation models. In: Journal of the Acad-
emy of Marketing Science (16:1) 1988, pp. 74-94.
420 References

Bahli, B., and Rivard, S.: A validation of measures associated with the risk factors in information
technology outsourcing. 36th Hawaii International Conference on System Sciences (HICSS-
36), Waikoloa, Big Island (HI), USA, 2003.
Bahli, B., and Rivard, S.: Validating measures of information technology outsourcing risk factors. In:
Omega Journal (33:2) 2004, pp. 175-187.
Baird, I.S., and Thomas, H.: Toward a contingency model of strategic risk taking. In: Academy of
Management Review (10:2) 1985, pp. 230-245.
Baker, G., Gibbons, R., and Murphey, K.J.: Relational contracts and the theory of the firm. In: Quar-
terly Journal of Economics (117:1) 2002, pp. 39-84.
Bakos, J.Y., and Brynjolfsson, E.: Information technology, incentives, and the optimal number of
suppliers. In: Journal of Management Information Systems (10:2) 1993a, pp. 37-54.
Bakos, J.Y., and Brynjolfsson, E.: From vendors to partners: Information technology and incomplete
contracts in buyer supplier relationships. In: Journal of Organizational Computing (3:3) 1993b,
pp. 301-328.
Banker, R.D., Kalvenes, J., and Patterson, R.A.: Information technology, contract completeness, and
buyer-supplier relationships. 21st International Conference on Information Systems (ICIS),
Brisbane, Australia, 2000, pp. 218-228.
Barney, J.B.: Firm resources and sustained competitive advantage. In: Journal of Management (17)
1991, pp. 99-120.
Barney, J.B.: How a firm's capabilities affect boundary decisions. In: Sloan Management Review
(42:3) 1999, pp. 137-145.
Barney, J.B.: Is the resource-based "view" a useful perspective for strategic management research?
Yes. In: Academy of Management Review (26:1) 2001, pp. 41-56.
Barney, J.B., Wright, P., and Ketchen Jr., D.J.: The resource-based view of the firm: ten years after
1991. In: Journal of Management (27:6) 2001, pp. 625-641.
Bartell, S.M.: Information systems outsourcing: a literature review and agenda for research. In:
International Journal of Industrial Theory and Behavior (1:1) 1998, pp. 17-44.
Barth, T.: Outsourcing unternehmensnaher Dienstleistungen. Europäischer Verlag, Stuttgart, Germa-
ny, 2003.
Barthélemy, J.: The hidden costs of outsourcing. In: Sloan Management Review (42:3) 2001, pp. 60-
69.
Barthélemy, J., and Geyer, D.: IT outsourcing: findings from an empirical survey in France and
Germany. In: European Management Journal (19:2) 2000, pp. 195-202.
Barua, A., Kriebel, C.H., and Mukhopadhyay, T.: An economic analysis of strategic information
technology investments. In: MIS Quarterly (15:3) 1991, pp. 3-23.
Barzel, Y.: Measurement cost and the organization of markets. In: Journal of Law and Economics
(25:1) 1982, pp. 27-48.
Bátiz-Lazo, B., and Wood, D.: Management of core capabilities in Mexican and European banks. In:
International Journal of Service Industry Management (10:5) 1999, pp. 430-448.
Bauer, S.: Auswirkungen der Informationstechnologie auf die vertikale Integration von Unterneh-
men. Peter Lang, Berlin, 1997.
Baumol, W.J., Panzar, J.C., and Willig, R.D.: Contestable market and the theory of industry struc-
ture. Harcourt Brace Jovanovich, New York (NY), 1982.
Bausch, H.-H., Gschrey, E., Kollbach, W., Petersen, H.-G., Schorr, G., Wick, P., Osterkamp, S., and
Spanier, G.: Die Kreditprüfung bei Kreditinstituten. (5th ed.) Deutscher Genossenschaftsver-
lag, Wiesbaden, 2004.
BDB: Überblick über das Bankgewerbe in der Europäischen Union – Mengengerüst, Bundesverband
deutscher Banken, 2005a. http://www.bankenverband.de/pic/artikelpic/122005/EU- 2005-11-
mengengeruest.pdf (as of 02 Feb 2007).
References 421

BDB: Überblick über das Bankgewerbe in der Europäischen Union – Ertragskennzahlen, Bundesver-
band deutscher Banken, 2005b.http://www.bankenverband.de/pic/artikelpic/122005/
EU-2005-11-ertragskennzahlen.pdf (as of 02 Feb 2007).
BDB: Überblick über das Bankgewerbe in der Europäischen Union – Mengengerüst, Bundesverband
deutscher Banken, 2006. http://www.bankenverband.de/pic/artikelpic/022007/EU-2006-07-
mengengeruest.pdf (as of 06 Mar 2008).
Beasley, D., Bull, D.R., and Martin, R.R.: An overview of genetic algorithms: part 1, fundamentals.
In: University Computing (15:2) 1993a, pp. 58-69.
Beasley, D., Bull, D.R., and Martin, R.R.: An overview of genetic algorithms: part 2, research topics.
In: University Computing (15:4) 1993b, pp. 170-181.
Beasley, M., Bradford, M., and Pagach, D.: Outsourcing? At your own risk. In: Strategic Finance
(2:7) 2004, pp. 23-29.
Beaumont, N., and Costa, C.: Information technology outsourcing in Australia. In: Information
Resources Management Journal (15:3) 2002, pp. 14-31.
Beck, R., Beimborn, D., and Weitzel, T.: The mobile standards battle. 36th Hawaii International
Conference on System Sciences (HICSS-36), Waikoloa, Big Island (HI), USA, 2003.
Beimborn, D.: A simulative analysis of causes and effects of cooperative business process outsourc-
ing in the banking industry. IRMA International Conference (Doctoral Symposium), San
Diego, 2005.
Beimborn, D.: A model for simulation analyses of cooperative sourcing in the banking industry. 39th
Hawaii International Conference on System Sciences, Kauai, Hawaii, 2006.
Beimborn, D., and Blumenberg, S.: How to measure relationships – Merging alignment and out-
sourcing research towards a unified relationship quality construct. 13 th Americas Conference
on Information Systems, Keystone (CO), USA, 2007.
Beimborn, D., Fladung, R.B., and König, W.: An optimization framework for efficient information
supply in the academic sector. 8th Pacific-Asia Conference on Information Systems (PACIS),
Shanghai, China, 2004.
Beimborn, D., Franke, J., and Weitzel, T.: Drivers and inhibitors for outsourcing financial processes
– a comparative survey of economies of scale, scope, and skill. 11th Americas Conference on
Information Systems (AMCIS), Omaha (NE), USA, 2005a.
Beimborn, D., Martin, S.F., and Homann, U.: Capability-oriented modeling of the firm. IPSI 2005,
Amalfi, Italy, 2005b.
Beimborn, D., Franke, J., Wagner, H.-T., Weitzel, T.: The impact of outsourcing on IT business
alignment and flexibility : a survey the German banking industry. 12th Americas Conference on
Information Systems (AMCIS), Acapulco, Mexico, 2006a.
Beimborn, D., Lamberti, H.-J., and Weitzel, T.: Game theoretical analysis of cooperative sourcing
scenarios. 39th Hawaii International Conference on System Sciences (HICSS-39), Koloa, Kauai
(HI), USA, 2006b.
Beimborn, D., Franke, J., Wagner, H.-T., Weitzel, T.: The influence of alignment on the post-
implementation success of a core banking information system: an embedded case study. 40th
Hawaii International Conference on System Sciences (HICSS-40), Waikoloa, Big Island (HI),
USA, 2007a.
Beimborn, D., Fladung, R.B., and Rothlauf, F.: How to Configure Cost-optimal Procurement Consor-
tia for Academic Libraries. 13th Americas Conference on Information Systems, Keystone (CO),
USA, 2007b.
Beimborn, D., and Weitzel, T.: Web Services und Service-orientierte IT-Architekturen. In: Das
Wirtschaftsstudium (WISU), (11) 2003, pp. 1360-1364
Benbasat, I., and Weber, R.: Research commentary: rethinking 'diversity' in information systems
research. In: Information Systems Research (7:4) 1996, pp. 389-399.
422 References

Benna, R., Heydolph, M., and Mitsche, T.: Erfolgreiches Retail Banking durch Disaggregation der
Wertschöpfungsketten. In: Die Bank (45:2) 2003, pp. 91-93.
Benson, J.K.: The interorganizational network as a political economy. In: Administrative Science
Quarterly (20:6) 1975, pp. 229-249.
Berens, W., Knappkötter, R., Segbers, K., Siemes, K., and Ullrich, T.: Der Firmenkredit – Kreditpro-
zess, Informationsbeschaffung und Kundenorientierung. In: Zeitschrift für das gesamte Kre-
ditwesen (15), 2005, pp. 784-789.
Berg, S.A., Forsund, F.R., and Jansen, E.S.: Technical efficiency of Norwegian banks: the non-
parametric approach to efficiency measurement. In: Journal of Productivity Analysis
(2:supplement) 1992, pp. 127-142.
Berg, S.A., and Friedman, P.: Impacts of domestic joint ventures on industrial rates of return. A
pooled cross-sectional analysis. In: Review of Economics and Statistics (63:2) 1981, pp. 293-
298.
Berger, A.N., Demsetz, R.S., and Strahan, P.E.: The consolidation of the financial services industry:
causes, consequences, and implications for the future. In: Journal of Banking and Finance
(23:2-3) 1999, pp. 637-653.
Berger, A.N., and Humphrey, D.B.: The dominance of inefficiencies over scale and product mix
economies in banking. In: Journal of Monetary Economics (28:1) 1991, pp. 117-148.
Berger, A.N., and Humphrey, D.B.: Efficiency of financial institutions: international survey and
directions of future research. In: European Journal of Operational Research (98:2) 1997, pp.
175-212.
Berger, A.N., Humphrey, D.B., and Pulley, L.B.: Do consumers pay for one-stop banking? Evidence
from an alternative revenue function. In: Journal of Banking and Finance (20:9) 1996, pp.
1601-1621.
Berger, A.N., and Mester, L.J.: Inside the black box: what explains differences in the efficiencies of
financial institutions? In: Journal of Banking and Finance (21:7) 1997, pp. 895-947.
Bergeron, F., Raymond, L., and Rivard, S. "Ideal patterns of strategic alignment and business per-
formance," Information & Management (41:8) 2004, pp 1003-1020.
Besen, S.M., and Farrell, J.: Choosing how to compete strategies and tactics in standardization. In:
Journal of Economic Perspectives (8:2) 1994, pp. 117-131.
Bettis, R., Bradley, S., and Hamel, G.: Outsourcing and industrial decline. In: Academy of Manage-
ment Executive (6:1) 1992, pp. 7-22.
Bettis, R.A.: Strategic management and the straightjacket: an editorial essay. In: Organization Sci-
ence (2:3) 1991, pp. 315-319.
BIS: Operational risk transfer across financial sectors. Risk Management Sub-Group of the Basel
Committee on Banking Supervision, Bank for International Settlements, Basel, Switzerland,
2003. http://www.bis.org/publ/joint06.pdf (as of 13 Aug 2006).
BIS: Basel II: International convergence of capital measurement and capital standards. Basel Com-
mittee on Banking Supervision, Bank for International Settlements, Basel, Switzerland, 2004.
http://www.bis.org/publ/bcbs118.pdf (as of 13 Aug 2006).
BIS: Outsourcing in financial services. The Joint Forum of the Basel Committee on Banking Super-
vision, Bank for International Settlements, Basel, Switzerland, 2005. http://www.bis.org/
publ/joint12.pdf (as of 11 Nov 2005).
Bloch, A.: Telematics, inter-organizations and economic performance. FAST Occasional Papers, no.
195, Paris, 1987.
Bloch, M., and Spang, S.: Reaping the benefits of business-process outsourcing. In: McKinsey on IT
(Fall), 2003, pp. 10-16.
Blum, C., and Roli, A.: Metaheuristics in combinatorial optimization: overview and conceptual
comparison. In: ACM Computing Surveys (35:3) 2003, pp. 268-308.
References 423

Bongartz, U.: Transaktionsbanking quo vadis? In: Management von Transaktionsbanken, H.-J.
Lamberti, A. Marlière and A. Pöhler (eds.), Springer, Heidelberg et al., 2004, pp. 39-57.
Boot, A.W.A.: European lessons on consolidation in banking. In: Journal of Banking and Finance
(23:2-4), pp. 609-613.
Börner, C.J.: Die Konzentration im Bankwesen – Ursachen und Folgen. In: Finanzplatz Deutschland
an der Schwelle zum 21. Jahrhundert, H.E. Büschgen (ed.), Fritz Knapp, Frankfurt am Main,
1998.
Bösch, M.: Schneller, effizienter und billiger. In: Bankmagazin (48:3) 1999, p. 33.
Bossel, H.: Systeme, Dynamik, Simulation – Modellbildung, Analyse und Simulation komplexer
Systeme. Books on Demand, Norderstedt, Germany, 2004.
Braeutigam, R.R., and Daughety, A.F.: On the estimation of returns to scale using variable cost
functions. In: Economic Letters (11:1) 1983, pp. 25-31.
Brandenberger, S.: Investment Engineering, Intermediation und Produktgestaltung in der Vermö-
gensberatung, Paul Haupt, Bern, 1995.
Brandenburger, A.M., and Nalebuff, B.J.: Co-opetition. Currency-Doubleday, New York (NY), 1996.
Bratley, P., Fox, B., and Schrage, L.: A guide to simulation. Springer, Heidelberg et al., 1987.
Braun, M.: Vom IT- zum Business Process Outsourcing. In: Bankmagazin (53:2) 2004, p. 22.
Brealey, R.A., and Myers, S.C.: Principles of Corporate Finance. (5th ed.) McGraw-Hill, New York
(NY), 1996.
Bruch, H.: Outsourcing – Konzepte und Strategien, Chancen und Risiken. Gabler, Wiesbaden, 1998.
Brynjolfsson, E., and Hitt, L.: Paradox lost? Firm-level evidence on the returns to information sys-
tems spending. In: Management Science (42:4) 1996, pp. 541-559.
Buckley, P.J., and Casson, M.: A theory of cooperation in international business. In: Co-operative
strategies in international business, F.J. Contractor and P. Lorange (eds.), Lexington Books,
Lexington (MA), 1988.
Buhl, H.U., Kreyer, N., and Wehrmann, A.: Go east? Bewertung und Gestaltung von IT-Sourcing-
Strategien in der Finanzdienstleistungsbranche. Institute of Business Administration, Augsburg
University, Augsburg, Germany, 2005.
Burr, W.: Fundierung von Leistungstiefenentscheidungen auf der Basis modifizierter Transaktions-
kostenansätze. In: Zeitschrift für betriebswirtschaftliche Forschung (55:3) 2003, pp. 112-135.
Büschgen, H.E.: Bankbetriebslehre. (3 ed.) Gustav Fischer, Jena, Stuttgart, 1994.
Buse, H.P.: Kooperationen. In: Betriebswirtschaftslehre der Mittel- und Kleinbetriebe: Größenspezi-
fische Probleme und Möglichkeiten ihrer Lösung, H.-C. Pfohl (ed.), Erich Schmidt, Berlin,
1997, pp. 442-477.
Buvik, A., and Gronhaug, K.: Inter-firm dependence, environmental uncertainty and vertical coordi-
nation in industrial buyer-seller relationships. In: Omega Journal (28:4) 2000, pp. 445-454.
Caballero, R.J., and Hammour, M.L.: The "Fundamental Transformation" in macroeconomics. In:
American Economic Review (86:2) 1996, pp. 181-186.
Cachon, G., and Harker, P.: Competition and outsourcing with scale economies. In: Management
Science (48:10) 2002, pp. 1314-1333.
Caldwell, B. and McGee, M. (1997) Outsourcing backlash, InformationWeek Online. 29 Sep 1997,
http://www.informationweek.com/650/50iuout.htm (as of 04 Mar 2007).
Canals, J.: Competitive strategies in European banking. Oxford University Press, Oxford, UK, 1994.
Cantwell, J.: The theory of technological competence and its application to international production.
In: Foreign investment, technology and economic growth, D. McFetridge (ed.), University of
Calgary Press, Calgary, Canada, 1991, pp. 33-70.
Capron, L., and Hulland, J.: Redeployment of brands, sales forces, and general marketing manage-
ment expertise following horizontal acquisitions: a resource-based view. In: Journal of Market-
ing (63:2) 1999, pp. 41-54.
424 References

Carter, J., Bitting, E., and Ghorbani, A.A.: Reputation formalization within information sharing
multiagent architectures. In: Computational Intelligence (18:4) 2002, pp. 515-534.
Cash, J.I.J., and Konsynski, B.R.: IS redraws competitive boundaries. In: Harvard Business Review
(63:2) 1985, pp. 134-142.
Caves, R.E., and Bradburd, R.M.: The empirical determinants of vertical integration. In: Journal of
Economic Behavior and Organization (9:3) 1988, pp. 265-279.
Chabrow, E.: The IBM-Amex Deal: a radical change in IT outsourcing. In: Information Week, In-
formation Week, 2002, http://www.informationweek.com/story/IWK20020301S0039 (as of 05
July 2006).
Chalos, P., and Sung, J.: Outsourcing decisions and managerial incentives. In: Decision Sciences
(29:4) 1998, pp. 901-918.
Chan, Y.E., Huff, A.S., Barclay, D.W., and Copeland, D.G.: Business strategic orientation, informa-
tion systems strategic orientation, and strategic alignment. In: Information Systems Research
(8:2) 1997, pp 125-150.
Chang, J.C.-J., and King, W.R.: Measuring the performance of information systems: a functional
scorecard. In: Journal of Management Information Systems (22:1) 2005, pp 85-115.
Chaudhury, A., Nam, K., and Rao, H.R.: Management of information systems outsourcing: a bidding
perspective. In: Journal of Management Information Systems (12:2) 1995, pp. 131-159.
Chen, S.: A new paradigm for knowledge-based competitions: building an industry through knowl-
edge sharing. In: Technology Analysis & Strategic Management (9:4) 1997, pp. 437-452.
Cheon, M.J., Grover, V., and Teng, J.T.C.: Theoretical perspectives on the outsourcing of informa-
tion systems. In: Journal of Information Technology (10:4) 1995, pp. 209-219.
Chiles, T.H., and McMackin, J.F.: Integrating variable risk preferences, trust, and transaction cost
economics. In: Academy of Management Review (21:1) 1996, pp. 73-100.
Chin, W.W.: The partial least squares approach for structural equation modelling. In: Modern Meth-
ods for Business Research, G.A. Marcoulides (ed.), Lawrence Erlbaum Associates, Mahwah
(NJ), USA, 1998, pp. 295-336.
Chua, W.: Radical developments in accounting thought. In: The Accounting Review (61:4) 1986, pp.
601-632.
Ciborra, C.U.: Reframing the role of computers in organizations – The transaction costs approach. In:
Office Technology and People (3:1) 1987, pp. 17-38.
Clark, T.D.J., Zmud, R.W., and McGray, G.E.: The outsourcing of information services: transform-
ing the nature of business in the information industry. In: Journal of Information Technology
(10:4) 1995, pp. 221-237.
Clemons, E.K.: Evaluation of strategic investments in information technology. In: Communications
of the ACM (34:1) 1991, pp. 24-36.
Clemons, E.K., and Reddi, S.P.: The impact of IT on the degree of outsourcing, the number of sup-
pliers, and the duration of contracts. 27th Hawaii International Conference on System Sciences
(HICSS-27), Wailea, Maui (HI), USA, 1994, pp. 855-864.
Clemons, E.K., Reddi, S.P., and Row, M.C.: The impact of information technology on the organiza-
tion of economic activity: the 'Move to the Middle' hypothesis. In: Journal of Management In-
formation Systems (10:2) 1993, pp. 9-35.
Clemons, E.K., and Row, M.C.: Sustaining IT advantage: the role of structural differences. In: MIS
Quarterly (15:3) 1991, pp. 275-292.
Clever, N.: Arbeitsrechtliche Fragen beim Outsourcing. In: Outsourcing und (Organic) Insourcing in
der Kreditwirtschaft, V.M. Jorczyk, J. Diehlmann, A. Troll, W. Frank, H. Zerwas, A. Ober-
mann, S. Rieck, E. Jaster, S. Neubert and M. Hammer (eds.), VÖB-Service, Bonn, 2004, pp.
223-245.
Coase, R.: The nature of the firm. In: Economica (4:16) 1937, pp. 386-405.
Coleman, J.: Foundations of social theory. Harvard University Press, Cambridge (MA), 1990.
References 425

Coleman, J.S., Menzel, H., and Katz, E.: The diffusion of an innovation among physicians. In: So-
ciometry (20:4) 1957, pp. 253-270.
Collis, D.J.: A resource-based analysis of global competition: the case of the bearings industry. In:
Strategic Management Journal (12:Winter Special Issue) 1996, pp. 49-68.
Commons, J.R.: Institutional economics. University of Wisconsin Press, Madison (WI), 1934.
Conway, R.W.: The manufacturing progress function. In: Journal of Industrial Engineering (10:1)
1959, pp. 39-53.
Corbett, M. F.; Outsourcing failures? Corbett & Associates, Ltd. http://www.firmbuilder.com/ arti-
cles/ 19/ 48/388/ (as of 01 Apr 2003).
Corbitt, B., and Rouse, A.: Minimizing risks in IT outsourcing: choosing target services. 7th Pacific
Asia Conference on Information Systems (PACIS), Adelaide, Australia, 2003, pp. 927-940.
Corts, K.S., and Singh, J.: The effect of repeated interaction on contract choice: evidence from off-
shore drilling. In: Journal of Law, Economics and Organization (20:1) 2004, pp. 230-260.
Cross, J.: IT outsourcing: British Petroleum's competitive approach. In: Harvard Business Review
(73:3) 1995, pp. 94-102.
Crux, A., and Schwilling, A.: Business Reengineering – Ein Ansatz der Roland Berger & Partner
GmbH. In: Prozessmanagement und Reengineering – Die Praxis im deutschsprachigen Raum,
M. Nippa and A. Picot (eds.), Campus, Frankfurt, New York (NY), 1996, pp. 206-223.
Cullen, S., and Willcocks, L.: Intelligent IT-Outsourcing – Eight building blocks to success. Butter-
worth-Heinemann, Oxford, UK, 2003.
Cunningham, P.A., and Fröschl, F.: Outsourcing: Strategische Bewertung einer Informationsdien-
stleistung. Frankfurter Allgemeine Zeitung, Frankfurt am Main, 1995.
Currie, W., Desai, B., Khan, N., Wang, X., and Weerakkody, V.: Vendor strategies for business
process and applications outsourcing: recent findings from field research. 36th Hawaii Interna-
tional Conference on System Sciences (HICSS-36), Waikoloa, Big Island (HI), USA, 2003.
Currie, W., and Seltsikas, P.: Exploring the supply side of IT outsourcing: evaluation the emerging
role of ASPs. In: European Journal of Information Systems (10) 2001, pp. 123-134.
Currie, W., and Willcocks, L.: Using multiple suppliers to mitigate the risk of IT outsourcing at ICI
and Wessex Water. In: Journal of Information Systems (4:4) 1998, pp. 226-236.
D'Aveni, R., and Ravenscraft, D.: Economies of integration versus bureaucracy costs: does vertical
integration improve performance? In: Academy of Management Journal (37:5) 1994, pp. 1167-
1206.
Davenport, T.H.: The coming commoditization of processes. In: Harvard Business Review (83:6)
2005, pp. 100-108.
Davenport, T.H., and Short, J.E.: The new industrial engineering: information technology and busi-
ness process redesign. In: Sloan Management Review (31:4) 1990, pp. 11-27.
Davies, S.W., and Morris, C.: A new index of vertical integration: some estimates for UK manufac-
turing. In: International Journal of Industrial Organization (13:2) 1995, pp. 151-177.
Davydov, M.M.: Corporate portals and e-business integration. A manager's guide. McGraw-Hill,
New York (NY), 2001.
Day, G.S.: The capabilities of market-driven organizations. In: Journal of Marketing (58:4) 1994, pp.
37-52.
Dayasindhu, N.: Information technology enabled process outsourcing and reengineering: case study
of a mortgage bank. 10th Americas Conference on Information Systems (AMCIS), New York
(NY), 2004.
DBB: Ertragslage deutscher Kreditinstitute. Deutsche Bundesbank, Frankfurt am Main, 2002.
DBB: Monatsbericht August 2004. Deutsche Bundesbank, Frankfurt am Main, 2004.
DBB: Time series database. Deutsche Bundesbank (German Central Bank), 2005,
http://www.bundesbank.de /statistik/statistik_zeitreihen.en.php (as of 15 Sep 2006).
426 References

DBB: Statistiken über den Zahlungsverkehr in Deutschland 2001-2005. Deutsche Bundesbank (Ger-
man Central Bank), Frankfurt am Main, Germany, 2006, http://www.bundesbank.de/ downlo-
ad/zahlungsverkehr/zv_statistik.pdf (as of 02 Jan 2007).
DBB: Bankenstatistik Januar 2008 – Statistisches Beiheft zum Monatsbericht. Deutsche Bundesbank,
Frankfurt am Main, 2008.
De Looff, L.A.: Information systems outsourcing decision making: a framework, organizational
theories, and case studies. In: Journal of Information Technology (10:4) 1995, pp. 281-297.
DeAngelo, H.: Competition and unanimity. In: American Economic Review (71:1) 1981, pp. 18-27.
Demski, J.S., and Sappington, D.E.M.: Sourcing with unverifiable performance information. In:
Journal of Accounting Research (31:1) 1993, pp. 1-20.
Dernbach, W.: Geschäftsprozessoptimierung – Der neue Weg zur marktorientierten Unternehmens-
organisation. In: Prozessmanagement und Reengineering – Die Praxis im deutschsprachigen
Raum, M. Nippa and A. Picot (eds.), Campus, Frankfurt am Main, New York (NY), 1996, pp.
187-205.
Dess, G.G., Ireland, R.D., and Hitt, M.A.: Industry effects and strategic management research. In:
Journal of Management (16:1) 1990, pp. 7-27.
Dess, G.G., Rasheed, A., McLaughlin, K., and Priem, R.: The new corporate architecture. In: Acad-
emy of Management Executive (9:3) 1995, pp. 7-20.
DeYoung, R., Hughes, J.P., and Moon, C.-G.: Regulatory covenant enforcement and the efficiency of
risk-taking at U.S. commercial banks. In: Journal of Economics and Business (53:2-3) 2001,
pp. 255-282.
Diamantopoulos, A., and Winklhofer, H.M.: Index construction with formative indicators: an alterna-
tive to scale development. In: Journal of Marketing Research (38:2) 2001, pp. 269-277.
Dibbern, J.: The sourcing of application software services. Springer Physica, Heidelberg et al., 2004.
Dibbern, J., Goles, T., Hirschheim, R., and Jayatilaka, B.: Information systems outsourcing: a survey
and analysis of the literature. In: The DATA BASE for Advantages in Information Systems
(35:4) 2004, pp. 6-102.
Dibbern, J., and Heinzl, A.: Outsourcing der Informationsverarbeitung im Mittelstand: Test eines
multitheoretischen Kausalmodells. In: Wirtschaftsinformatik (43:4) 2001, pp. 339-350.
Dibbern, J., Heinzl, A., and Leibbrand, S.: Interpretation des Sourcings der Informationsverarbeitung:
Hintergründe und Grenzen ökonomischer Einflussgrößen. In: Wirtschaftsinformatik (45:5)
2003, pp. 533-540.
Dierickx, I., and Cool, K.: Asset stock accumulation and sustainability of competitive advantage. In:
Management Science (35:12) 1989, pp. 1504-1511.
Dillmann, R., and Sioulvegas, N.: Outsourcing von Beschaffungsprozessen. In: Information Man-
agement & Consulting (18:3) 2003, pp. 31-37.
DiMaggio, P.J.: Interest and agency in institutional theory. In: Institutional patterns and organiza-
tions, L.G. Zucker (ed.), Ballinger, Cambridge, MA, 1988, pp. 3-22.
Disher, C., Teschner, C., Kaul, A., and Wright, M.: Next generation outsourcing and offshoring –
Capturing value in financial services. Booz Allen Hamilton, McLean (VA), USA, 2004.
Dittrich, J., and Braun, M.: Business Process Outsourcing: Ein Entscheidungsleitfaden für das Out-
und Insourcing von Geschäftsprozessen. Schäffer-Poeschel, Stuttgart, Germany, 2004.
DoD: Functional process simulation – A guidebook. U.S. Department of Defense, Washington, D.C.,
1993. http://www.defenselink.mil/nii/bpr/bprcd/0062c5.htm (as of 18 Apr 2005).
Dombret, A.R.: Retail Banking im 21. Jahrhundert – Die Reorganisation der Wertschöpfungskette.
Forschungsgesellschaft für Genossenschaftswesen, Münster, 2004.
Dombret, A.R., and Kern, H.J.: European retail banks – An endangered species? Monitor Group,
Rothschild, John Wiley, New York (NY), 2003.
Dorn, P.: Selling one's birthright. In: Information Week (241), 16 Oct 1989, p. 52.
References 427

Dowling, M.J., Roering, W.D., Carlin, B.A., and Wisnieski, J.: Multifaceted relationships under
coopetition: description and theory. In: Journal of Management Inquiry (5:2) 1996, pp. 155-
167.
Doz, Y.L., and Hamel, G.: Alliance advantage – The art of creating value through partnering. Har-
vard Business School Press, Boston (MA), 1998
Droge, C., Jayaram, J., and Vickery, S.K.: The effects of internal versus external integration practices
on time-based performance and overall firm performance. In: Journal of Operations Manage-
ment (22:6) 2004, pp. 557-573.
Drori, O.: Using text analysis to inform clients of the subject of a document. Informing Science +
Information Technology Education Joint Conference (InSITE), Pori, Finland, 2003.
Dubin, R.: Theory building. Free Press, New York (NY), 1969.
Dubin, R.: Theory building in applied area. In: Handbook of Industrial and Organizational Psychol-
ogy, M. Dunnette (ed.), Rand McNally, Chicago (IL), 1976, pp. 17-40.
Duhan, S., Levy, M., and Powell, P.: Information systems strategies in knowledge-based SMEs: the
role of core competencies. In: European Journal of Information Systems (10:1) 2001, pp. 25-
40.
Dwyer, R., Schurr, P., and Oh, S.: Developing buyer-seller relationships. In: Journal of Marketing
(51:2) 1987, pp. 11-27.
Earl, M.J.: The risks of outsourcing IT. In: Sloan Management Review (37:3) 1996, pp. 26-32.
EC: Consultation paper on modernising Value Added Tax obligations for financial services and
insurances. European Commission, Directorate General, Taxation and Customs Union, Brus-
sels, Belgium, 2006. http://ec.europa.eu/taxation_customs/resources/documents/common/ con-
sultations/tax/modernising_VAT_en.pdf (as of 22 Sep 2006).
ECB: EU Banking Sector Stability. European Central Bank, Frankfurt/Main, 2007.
Eichelmann, T.: Strategien für deutsche Banken im Kontext des Jahres 2004. Guest presentation at
Friedrich Alexander University Erlangen-Nuremberg, Roland Berger, 24 Jan 2004. www.prof-
gerke.de/__Download/Bankman_Vortrag_Berger_WS0304.pdf (as of 23 Aug 2006).
Eichelmann, T., Schneidereit, F., and Dosis, D.: Verringerung der Wertschöpfungstiefe – Analyse
und Ausblick. In: Sourcing in der Bankwirtschaft, W. Achenbach, J. Moormann and H. Scho-
ber (eds.), Bankakademie-Verlag, Frankfurt am Main, 2004, pp. 325-342.
Eikebrokk, T.R., and Olsen, D.H.: Co-opetition and e-business success in SMEs: an empirical inves-
tigation of European SMEs. 38th Hawaii International Conference on System Sciences (HICSS-
38), Waikoloa, Big Island (HI), USA, 2005.
Eisenhardt, K.M.: Agency theory: an assessment and review. In: Academy of Management Review
(14:1) 1989, pp. 57-76.
Elitzur, R., and Wensley, A.: Game theory as a tool for understanding information services outsourc-
ing. In: Journal of Information Technology (12:1) 1997, pp. 45-60.
Emery, F.E., and Trist, E.L.: The causal texture of organizational environments. In: Human Relations
(18:1) 1965, pp. 21-32.
Emmelhainz, M.A.: Electronic Data Interchange: does it change the purchasing process? In: Journal
of Purchasing and Materials Management (23:4) 1987, pp. 2-8.
Englert, J.: Internetbasierte Unternehmenskooperationen als Wettbewerbsfaktor für den deutschen
Mittelstand. Dissertation Thesis, Albert Ludwigs University, Freiburg, Germany, 2000.
Engstler, M., and Vocke, C.: Bank&Zukunft 2004-2005. Fraunhofer Institut für Arbeitswirtschaft
und Organisation, Frankfurt am Main, 2004.
Epple, D.L., and Devadas, R.: Organization learning curves: a method for investigating intra-plant
transfer of knowledge acquired through learning by doing. In: Organization Science (2:1) 1991,
pp. 58-70.
Essig, M.: Cooperative sourcing: strategies and tactics of consortium purchasing. 4th IFPMM Sum-
mer School on Advanced Purchasing and Supply Research, Salzburg, Austria, 1998, pp. 9-25.
428 References

Ewert, R., and Wagenhofer, A.: Interne Unternehmensrechnung. (5th ed.) Springer, Heidelberg, 2003.
Fairchild, A.M.: Enabling usage-based IT costing in the banking sector. In: Electronic Journal of
Information Systems Evaluation (6:2) 2003, pp. 87-94.
Farrell, J., and Saloner, G.: Installed base and compatibility: innovation, product preannouncements,
and predation. In: American Economic Review (76:5) 1986, pp. 940-955.
Favero, C.A., and Papi, L.: Technical efficiency and scale efficiency in the Italian banking sector: a
non-parametric approach. In: Applied Economics (27:4) 1995, pp. 385-395.
FBE: General statistics on the European financial sector. European Banking Federation, Brussels,
Belgium, 2003.
Feenstra, R.C., and Hanson, G.H.: Ownership and control in outsourcing to China: estimating the
property rights theory of the firm. In: The Quarterly Journal of Economics (120:2) 2003, pp.
729-761.
Feeny, D.F., and Willcocks, L.P.: Core IS capabilities for exploiting information technology. In:
Sloan Management Review (39:3) 1998b, pp. 9-21.
Feeny, D.F., Lacity, M.C., and Willcocks, L.P.: Business process outsourcing: the promise of the
"Enterprise Partnership" model. Templeton College, Oxford, UK, 2003.
Feeny, D.F., Lacity, M.C., and Willcocks, L.P.: Taking the measure of outsourcing providers. In:
Sloan Management Review (46:3) 2005, pp. 41-48.
Fehr, B.: etb wickelt nun für Sparda-Banken ab. In: Frankfurter Allgemeine Zeitung, 30 Sep 2002, p.
21.
Fehr, B., and Mussler, H.: Transaktionsbank der drei Großbanken gescheitert. In: Frankfurter Allge-
meine Zeitung, 03 May 2003, p. 12.
Finken, T.: Entscheidungskriterien für das Outsourcing der betrieblichen Datenverarbeitung. Shaker,
Aachen, 1997.
Flake, G.W.: The computational beauty of nature: computer explorations of fractals, chaos, complex
systems, and adaptation. MIT Press, Cambridge (MA), 1998.
Fladung, R.B.: Scientific communication – Economic analysis of the electronic journal market.
Dissertation thesis, J.W. Goethe University, Frankfurt am Main, 2006.
Flannery, M.J.: Retail bank deposits as quasi-fixed factors of production. In: American Economic
Review (72:3) 1982, pp. 527-536.
Flesch, J.R.: DG-Bank erwartet Aufsplittung des Bankenmarktes: Spezialisierte Vertriebs-, Portfolio-
und Produktionsinstitute decken die Wertschöpfungskette ab. In: Börsenzeitung, 10 Aug 2000,
p. 8.
Focke, H., Kremlicka, R., Freudenstein, G., Gröflin, J., Pratz, A., Röckemann, C., and West, A.:
Tendenz steigend: Transaction Banking auf dem Weg zu Service und Innovation, A.T. Kear-
ney Transaction-Banking-Studie 2004. A.T. Kearney, Frankfurt am Main, Munich, 2004.
Fontenot, R., and Wilson, E.: Relational exchange: a review of selected models for prediction matrix
of relationship activities. In: Journal of Business Research (39:1) 1997, pp. 5-12.
Ford, D., and Farmer, D.: Make or buy – A key strategic issue. In: Long Range Planning (19:5) 1986,
pp. 54-62.
Foss, N.J.: The resource-based perspective: an assessment and diagnosis of problems. In: Scandina-
vian Journal of Management (14:3) 1998, pp. 133-149.
Foss, N.J., and Eriksen, B.: Competitive advantage and industry capabilities. In: Resource-based and
evolutionary theories of the firm: towards a synthesis, C.A. Montgomery (ed.), Kluwer, Boston
(MA), 1995, pp. 43-70.
Foss, N.J., Knudsen, C., and Montgomery, C.A.: An exploration of common ground: integrating
evolutionary and strategic theories of the firm. In: Resource-based and evolutionary theories of
the firm – Towards a synthesis, C.A. Montgomery (ed.), Kluwer, Boston (MA), 1995, pp. 1-18.
Frank, W.: Bankaufsichtsrechtliche Aspekte beim Outsourcing. In: Outsourcing und (Organic) In-
sourcing in der Kreditwirtschaft, V.M. Jorczyk, J. Diehlmann, A. Troll, W. Frank, H. Zerwas,
References 429

A. Obermann, S. Rieck, E. Jaster, S. Neubert and M. Hammer (eds.), VÖB-Service, Bonn,

2004, pp. 29-61.
Franke, J., and Gewald, H.: A comparison of risks in information technology outsourcing and busi-
ness process outsourcing. 11th Americas Conference on Information Systems (AMCIS), Omaha
(NE), USA, 2005.
Franke, J., Wagner, H.-T., and Weitzel, T.: The role of IT business alignment for value creation: a
multiple case study among German banks. 26th International Conference in Information Sys-
tems (ICIS05), Las Vegas (NV), 2005a.
Franke, J., Stockheim, T., and König, W.: The impact of reputation on supply chains. In: Journal of
Information Systems and e-Business Management (3:4) 2005b, pp. 323-341.
Franklin, S.: Autonomous agents as embodied AI. In: Cybernetics and Systems (28:6) 1997, pp. 499-
520.
Freedman, R.: Building the IT consulting practice. John Wiley, New York (NY), 2002.
Friend, M., Lucacs, M., and Snowdon, J.: European business process outsourcing forecast and analy-
sis 2001-2006. IDC, Framingham (MA), USA, 2002.
Fritsch, M.: Cooperation and efficiency of regional R&D activities. In: Cambridge Journal of Eco-
nomics (28:6) 2004, pp. 829-846.
Frohmüller, K.P.: Die Outsourcing-Diskussion in der Banking-Community. Commerzbank AG, E-
Finance Lab Jour Fixe at Johann Wolfgang Goethe-University, Frankfurt am Main, 2005.
Fröschl, F.: Vom IuK-Outsourcing zum Business Process Outsourcing. In: Wirtschaftsinformatik
(41:5) 1999, pp. 458-460.
FSO: Datenreport 2004. Bundeszentrale für politische Bildung, Bonn, Germany, 2004.
Furubotn, E.G., and Richter, R.: Institutions and economic theory: the contribution of the new institu-
tional economics. The University of Michigan Press, Ann Arbor (MI), USA, 1998.
Gal-Or, E.: Quality and quality competition. In: Bell Journal of Economics (14:2) 1983, pp. 590-600.
Gallivan, M.J., and Oh, W.: Analyzing IS outsourcing relationships as alliances among multiple
clients and vendors. 32nd Hawaii International Conference on System Sciences (HICSS-32),
Wailea, Maui (HI), USA, 1999.
Ganesh, J., and Moitra, D.: An empirical examination of the determinants of successful transition
management in offshore business process outsourcing. 10th Americas Conference on Informa-
tion Systems (AMCIS), New York (NY), 2004.
Gartner Group: Gartner's BPO model. Gartner Group, Stamford (CT), USA, 2002.
Gartner Group: Vendors seek clear role in SMB market. Gartner Group, Stamford (CT), USA, 2004.
Gaumert, U.: Grundsätze ordnungsgemäßen Ratings (GoR): Basel II- und MaK-konforme Organisa-
tion des Kreditgeschäfts. Bank-Verlag, Cologne, 2005.
Gebauer, J.: Virtual organizations from an economic perspective. 4th European Conference on Infor-
mation Systems (ECIS), Lisbon, Portugal, 1996.
Gellrich, T.: Schaffen Outsourcing-Ankündigungen Shareholder-Value? Spring Symposium of the E-
Finance Lab, 2004. http://www.efinancelab.com/events/conf/archive2004/tagung02_2004/ vor-
trag_gellrich.pdf (as of 17 July 2006).
Gellrich, T., Hackethal, A., and Holzhäuser, M.: Vertical integration and bank performance. IN-
FORMS Annual Meeting, New Orleans (LA), 2005.
Gewald, H., and Dibbern, J.: The influential role of perceived risks versus perceived benefits in the
acceptance of business process outsourcing: empirical evidence from the German banking in-
dustry. Working paper 2005-9, E-Finance Lab, J.W. Goethe University, Frankfurt am Main,
2005.
Gewald, H., and Hinz, D.: A framework for classifying the operational risks of outsourcing. 8th
Pacific-Asia Conference on Information Systems (PACIS), Shanghai, China, 2004.
Gewald, H., and Lammers, M.: What is core? San Diego International Systems Conference (SISC),
San Diego (CA), USA, 2005.
430 References

Geyer, G., and Venen, A.: Ökonomische Prozesse – Globalisierung und Transformation: Eine institu-
tionenökonomische Analyse aus der Perspektive des institutionellen Wandels und der Transak-
tionskostenökonomik. LIT Verlag, Hamburg, 2001.
Gietzmann, M.: Incomplete contracts and the make or buy decision: governance design and attainable
flexibility. In: Accounting, Organizations, and Society (21:6) 1996, pp. 611-626.
Gilley, K.M., and Rasheed, A.: Making more by doing less: an analysis of outsourcing and its effects
on firm performance. In: Journal of Management (26:4) 2000, pp. 763-790.
Glöckeler, W.: Die Wertschöpfung von Kreditinstituten: Untersuchungen zu einem Merkmal für die
Messung von Unternehmenskonzentration. Berlin, 1975.
Goldenberg, J., and Efroni, S.: Using cellular automata modeling of emergence of innovations. In:
Technological forecasting and social change (68:3) 2001, pp. 293-308.
Goles, T.: The impact of client-vendor relationship on outsourcing success. Ph.D. thesis, University
of Houston, Houston (TX), 2001.
Goles, T.: Vendor capabilities and outsourcing success: a resource-based view. In: Wirtschaftsinfor-
matik (45:2) 2003, pp. 199-206.
Goles, T., and Chin, W.: Relational exchange theory and IS outsourcing: developing a scale to meas-
ure relationship factors. In: Information Systems Outsourcing, R.A. Hirschheim, A. Heinzl and
J. Dibbern (eds.), Springer, Heidelberg et al., 2002, pp. 77-109.
Goles, T., and Chin, W.W.: Information systems outsourcing relationship factors: detailed conceptu-
alization and initial evidence. In: The DATA BASE for Advances in Information Systems
(36:4) 2005, pp. 47-67.
Gomber, P., König, W., Beimborn, D., Franke, J., Wagner, H.-T., and Weitzel, T.: Kritische Erfolgs-
faktoren in Finanzprozessen – IT-Management und Alignment im Kreditprozess der 1.000
größten Banken in Deutschland. E-Finance Lab, Frankfurt am Main, 2006.
Gopal, A., Bostrom, R.P., and Chin, W.W.: Applying adaptive structuration theory to investigate the
process of group support systems use. In: Journal of Management Information Systems (9:3)
1993, pp 45-69.
Gort, M.: Diversification and integration in American industry. Princeton University Press, Princeton
(NJ), 1962.
Goshal, S., and Moran, P.: A critique of the transaction cost theory. In: Academy of Management
Review (12:1) 1996, pp. 13-47.
Götz, O., and Liehr-Gobbers, K.: Analyse von Strukturgleichungsmodellen mit Hilfe der Partial-
Least-Squares(PLS)-Methode. In: Die Betriebswirtschaft (64:6) 2004, pp. 714-738.
Graband, T.: Norisbank-Filialen – Als Testfelder unverzichtbar. In: bank & markt (33:6) 2004, pp.
25-28.
Grant, R.M.: The resource-based theory of competitive advantage: implications for strategy formula-
tion. In: California Management Review (33:3) 1991, pp. 114-135.
Grant, R.M.: Prospering in dynamically-competitive environments: organizational capability as
knowledge integration. In: Organization Science (7:4) 1996a, pp. 375-387.
Grant, R.M.: Toward a knowledge-based theory of the firm. In: Strategic Management Journal
(17:Winter Special Issue) 1996b, pp. 109-122.
Greaver, M.H.: Strategic outsourcing: a structured approach to outsourcing decisions and initiatives.
American Management Association, New York (NY), 1999.
Grebe, M., Kottmann, D., and Nettesheim: Business Process Outsourcing – aber richtig! In: Informa-
tion Management & Consulting (18:3) 2003, pp. 24-30.
Griffiths, G.H., and Finlay, P.N.: IS-enabled sustainable competitive advantage in financial services,
retailing and manufacturing. In: Journal of Strategic Information Systems (13:1) 2004, pp. 29-
59.
Chandler, A.: What is a firm? A historical perspective. In: European Economic Review (36:2-3)
1992, pp. 483-492.
References 431

Grill, H., and Perczynski, H.: Wirtschaftslehre des Kreditwesens. (38 ed.) Deutscher Sparkassen-
Verlag, Stuttgart, 2004.
Grossman, S., and Hart, O.: The costs and benefits of ownership: a theory of vertical and lateral
integration. In: Journal of Political Economy (94) 1986, pp. 691-719.
Grover, V., Cheon, M.J., and Teng, J.T.C.: An evaluation of the impact of corporate strategy and the
role of information technology on IS functional outsourcing. In: European Journal of Informa-
tion Systems (3:3) 1994, pp. 179-190.
Grover, V., Cheon, M.J., and Teng, J.T.C.: The effect of service quality and partnership on the out-
sourcing of information systems functions. In: Journal of Management Information Systems
(12:4) 1996, pp. 89-116.
Grover, V., Teng, J.T.C., and Fiedler, K.D.: Investigating the role of information technology in
building buyer-supplier relationships. In: Journal of the Association for Information Systems
(3:7), 245 2003, p. 217.
Gulati, R., and Singh, H.: The architecture of cooperation: managing coordination costs and appro-
priation concerns in strategic alliances. In: Administrative Science Quarterly (43:4) 1998, pp.
781-814.
Gurbaxani, V., and Whang, S.: The impact of information systems on organizations and markets. In:
Communications of the ACM (34:1) 1991, pp. 60-73.
Hackethal, A.: German banks – A declining industry? Johann Wolfgang Goethe University, Frankfurt
am Main, 2003.
Hagel, J.: The business potential of SOA, bpm.com, 23 Jan 2007. http://www.soa-world.com/ Fea-
turePrint.asp?FeatureId=222 (as of 01 Feb 2007).
Halvey, J.K., and Melby, B.M.: Information technology outsourcing transactions: processes, strate-
gies, and contracts. John Wiley, New York (NY), 1996.
Halvey, J.K., and Murphy Melby, B.: Business process outsourcing – Process, strategies and con-
tracts. John Wiley, New York (NY), 2000.
Hamel, G.: Competition for competence and interpartner learning within international strategic
alliances. In: Strategic Management Journal (12:Special Issue) 1991, pp. 83-104.
Hamel, G., and Doz, Y.L.: Collaborate with your competitors – And win. In: Harvard Business
Review (67:1) 1989, pp. 133-159.
Hammer, M., and Stanton, S.: How process enterprises really work. In: Harvard Business Review
(77:6) 1999, pp. 108-117.
Hammer, M., and Champy, J.: Reengineering the corporation. B&T, New York (NY), 1993.
Hamoir, O., McCamish, C., Nierderkorn, M., and Thiersch, C.: Europe's banks: verging on merging.
McKinsey Quarterly (3), 2002, pp. 116-125.
Hancox, M., and Hackney, R.: Information technology outsourcing: conceptualizing practice in the
public and private sector. 32nd Hawaii International Conference on System Sciences (HICSS-
32), Wailea, Maui (HI), USA, 1999.
Hannan, T.H., and McDowell, J.M.: Market concentration and the diffusion of new technology in the
banking industry. In: The Review of Economics and Statistics (66:4) 1984, pp. 686-691.
Hansen, G., and Wernerfelt, B.: Determinants of firm performance: the relative importance of eco-
nomic and organizational factors. In: Strategic Management Journal (10) 1989, pp. 399-411.
Hansmann, K.-W., and Ringle, C.M.: SmartPLS Manual. Hamburg, 2004. http://www.ibl-unihh.de/
manual.pdf (as of 01 Mar 2007).
Harrigan, K.R.: Exit barriers and vertical integration. In: Academy of Management Journal (28:3)
1985, pp. 686-697.
Harris, M., and Raviv, A.: Optimal incentive contracts with imperfect information. In: Journal of
Economic Theory (20:2) 1979, pp. 231-259.
Hart, O.: Incomplete contracts and the theory of the firm. In: Journal of Law, Economics, and Or-
ganization (4:1) 1988, pp. 119-139.
432 References

Hart, O., and Moore, J.: Property rights and the nature of the firm. In: Journal of Political Economy
(98:6) 1990, pp. 1119-1158.
Hartzel, K.S., and Nightingale, J.P.: Digital offshore outsourcing: the hidden transaction costs. 16th
IRMA International Conference, San Diego (CA), 2005, pp. 764-766.
Harvey, I.: Evolutionary robotics and SAGA: the case for hill crawling and tournament selection. 3rd
Artificial Life Conference, Santa Fe (NM), 1994, pp. 299-326.
Haykek, F.A.v.: Individualism and economic order, University of Chicago Press, Chicago, 1948.
Heinzl, A.: Die Ausgliederung der betrieblichen Datenverarbeitung – Eine empirische Untersuchung
der Motive, Formen und Auswirkungen. (2nd ed.) Schäffer-Poeschel, Stuttgart, 1993.
Helper, S., and Levine, D.I.: Long-term supplier relations and product-market structure. In: Journal
of Law, Economics, and Organization (8:3) 1992, pp. 561-581.
Henderson, B.D.: Die Erfahrungskurve in der Unternehmensstrategie. (2 ed.) Campus, Frankfurt am
Main, 1984.
Henderson, J.: Plugging into strategic partnerships: the critical IS connection. In: Sloan Management
Review (31:3) 1990, pp. 7-18.
Henderson, J., and Venkatraman, N.: Strategic alignment: a model for organizational transformation
through information technology. In: Transforming organizations, T. Kochan and M. Unseem
(eds.), Oxford University Press, New York (NY), 1992, pp. 97-117.
Herrmann, J.: Business Process Outsourcing – Heilsamer Zwang zur Flexibilität. In: FIN.KOM (1),
2004, pp. 9-11.
Hertel, S.: Marktsituation Servicing und Anforderungen an Servicing-Dienstleister. plenum Mana-
gement Consulting GmbH, Wiesbaden, Germany, Aareal HM Kundeninformationsforum,
Frankfurt am Main, 2004.
Hevner, A.R., March, S.T., Park, J., and Ram, S.: Design science in information systems research. In:
MIS Quarterly (28:1) 2004, pp. 75-105.
Heymann, T.: Outsourcing als Form der Kooperation. In: Computer und Recht (1), 2000, p. 23.
Hill, C.W.L.: Cooperation, opportunism, and the invisible hand: implications for transaction cost
theory. In: Academy of Management Review (15:3) 1990, pp. 500-514.
Hippel, E.v.: Cooperation between rivals: informal know-how trading. In: Industrial dynamics, B.
Carlsson (ed.), Kluwer, Dordrecht, Netherlands, 1989, pp. 157-175.
Hirsch, P., and Friedman, R.: Collaboration or paradigm shift? Economic vs. behavioral thinking
about policy? In: Best papers proceedings, J. Pearce and R. Robinson (eds.), Academy of Man-
agement, Chicago (IL), 1986, pp. 31-35.
Hirsch, P., Michaels, S., and Friedman, R.: "Dirty hands" versus "clean models": is sociology in
danger of being seduced by economics? In: Theory and Society (16:3) 1987, pp. 317-336.
Hirschheim, R.: Interview with Rudy Hirschheim on "perception on information systems outsourc-
ing". In: Wirtschaftsinformatik (45:2) 2003, pp. 111-114.
Hirschheim, R., and Dibbern, J.: Information systems outsourcing in the new economy – an introduc-
tion. In: Hirschheim, R. (ed.): Information systems outsourcing: enduring themes, emergent
patterns, and future directions, Springer, Heidelberg et al., 2002, pp. 3-23
Hirschheim, R., and Lacity, M.C.: The myths and realities of information technology insourcing. In:
Communications of the ACM (43:2) 2000, pp. 99-107.
Hirschman, A.O.: Against parsimony: three ways of complicating some categories of economic
discourse. In: Economics and Philosophy (1:1) 1985, pp. 7-21.
Hitt, M., Hoskisson, R.E., and Nixon, R.D.: A midrange theory of interfunctional integration, its
antecedents and outcomes. In: Journal of Engineering and Technology Management (10) 1993,
pp. 161-185.
Hoetker, G.: Do modular product lead to modular organizations? University of Illinois at Urbana-
Champaign, 2002.
References 433

Holland, C.P., and Lockett, G.: Strategic choice and inter-organizational information systems. 27th
Hawaii International Conference on System Sciences (HICSS-27), Wailea, Maui (HI), USA,
1994, pp. 405-413.
Holmström, B.: Moral hazard and observability. In: Bell Journal of Economics (10:1) 1979, pp. 74-
91.
Holtmann, C.-F., and Kleinheyer, N.: Strategische Schritte zur Errichtung einer Kreditfabrik. In:
Kreditwesen (10), 2002, pp. 477-483.
Hölzer, A.: Rethinking the Credit Process – Neue Wege bei der Kreditprozessgestaltung (I). In: Die
Bank (4), 2004, pp. 234-243.
Holzhäuser, M., Lammers, M., and Schwarze, F.: Integrated decision model for credit product out-
sourcing. In: Wirtschaftsinformatik (47:2) 2005, pp. 109-117.
Hoopes, D.G., Madsen, T., and Walker, G.: Guest editors' introduction to the special issue: Why is
there a resource-based view? Towards a theory of competitive heterogeneity. In: Strategic
Management Journal (24:10) 2003, pp. 889-902.
Hoppen, N., Beimborn, D., and König, W.: The impact of software patents on the structure of the
software market – A simulation model. 11th European Conference on Information Systems
(ECIS), Naples, Italy, 2003.
Hoppenstedt, D.H.: Die Strategie der Sparkassen-Finanzgruppe in globalisierten Märkten. In: Hand-
buch Veränderungsmanagement und Restrukturierung, N. Kleinheyer and E. Priewasser (eds.),
Frankfurt, 2000, pp. 81-91.
Hu, Q., Saunders, C.S., and Gebelt, M.: Research report: diffusion of information systems outsourc-
ing: a reevaluation of influence sources. In: Information Systems Research (8:3) 1997, pp. 288-
301.
Huber, R.L.: How Continental Bank outsourced its crown jewels. In: Harvard Business Review
(71:1) 1993, pp. 121-129.
Hughes, J.P.: Incorporating risk into the analysis of production. In: Atlantic Economic Journal (27:1)
1999, pp. 1-23.
Hughes, J.P., Mester, L.J., and Moon, C.-G.: Are scale economies in banking elusive or illusive?
Evidence obtained by incorporating capital structure and risk-taking into models of bank pro-
duction. In: Journal of Banking and Finance (25:12) 2001, pp. 2169-2208.
Hulland, J.: Use of Partial Least Squares (PLS) in strategic management research. A review of four
recent studies. In: Strategic Management Journal (20:2) 1999, pp. 195-204.
Hunt, S.D.: Modern marketing theory – Critical issues in the philosophy of marketing science. South-
Western Publishing, Cincinnati (OH), 1991.
Hunter, W.C., and Timme, S.: Core deposits and physical capital: a reexamination of bank scale
economies and efficiency with quasi-fixed inputs. In: Journal of Money, Credit, and Banking
(27:1) 1995, pp. 165-185.
Hunter, W.C., and Timme, S.G.: Technological change and production economies in large U.S.
commercial banks. In: Journal of Business (64:3) 1991, pp. 339-362.
Hunter, W.C., Timme, S.G., and Yang, W.K.: An examination of cost subadditivity and multiproduct
production in large U.S. commercial banks. In: Journal of Money, Credit, and Banking (22:4)
1990, pp. 504-525.
IBM: European wealth and private banking – Industry survey 2003. IBM Business Consulting Ser-
vices, Armonk (NJ), USA, 2003.
Itschert, J., and ul-Haq, R.: International banking strategic alliances: reflections on BNP/Dresdner.
Palgrave Macmillan, New York (NY), 2003.
Jäger-Goy, H.: Das Controlling des Outsourcings von IV-Leistungen. In: Arbeitspapiere WI, Mainz,
Germany, 1998, http://geb.uni-giessen.de/geb/volltexte/2004/1679/pdf/Apap_WI_1998_06.pdf
(as of 28 Apr 2005).
434 References

Jasny, R.: Wertschöpfungsketten bei Banken. In: Management der Wertschöpfungsketten in Banken,
L.P. Marighetti, R. Jasny, A. Herrmann and F. Huber (eds.), Gabler, Wiesbaden, 2001, pp. 25-
35.
Jayatilaka, B., Schwarz, A., and Hirschheim, R.: Determinants of ASP choice: an integrated perspec-
tive. In: European Journal of Information Systems (12:3) 2003, pp. 210-224.
Jensen, M.C.: Organization theory and methodology. In: Accounting Review (58:2) 1983, pp. 319-
339.
Jensen, M.C., and Meckling, W.H.: Theory of the firm: managerial behaviour, agency costs and
ownership structure. In: Journal of Financial Economics (3:4) 1976, pp. 305-360.
Jepperson, R.L.: Institutions, institutional effects, and institutionalism. In: The new institutionalism in
organizational analysis, W.W. Powell and P.J. DiMaggio (eds.), University of Chicago Press,
Chicago (IL), 1991, pp. 143-163.
Johnston, R., and Lawrence, P.R.: Beyond vertical integration – The rise of the value-adding partner-
ship. In: Harvard Business Review (66:4) 1988, pp. 94-101.
Jones, W.P., and Furnas, G.W.: Pictures of relevance: a geometric analysis of similarity measures. In:
Journal of the American society for information science (38:6) 1987, pp. 420-442.
Joos-Sachse, T.: Controlling, Kostenrechnung und Kostenmanagement. Gabler, Wiesbaden, 2002.
Jorczyk, V.M.: Optimierungsmöglichkeiten durch Organschaftsstrukturen. In: Outsourcing und
(Organic) Insourcing in der Kreditwirtschaft, V.M. Jorczyk, J. Diehlmann, A. Troll, W. Frank,
H. Zerwas, A. Obermann, S. Rieck, E. Jaster, S. Neubert and M. Hammer (eds.), VÖB-Service,
Bonn, 2004, pp. 195-222.
JPMorganChase: JPMorgan Chase and IBM announce changes to technology infrastructure. Press
release, 2004. http://investor.shareholder.com/jpmorganchase/press/releasedetail.cfm?Release
ID=143884 &ReleaseType=Current (as of 18 July 2006).
Jurison, J.: The role of risk and return in information technology outsourcing decisions. In: Journal of
Information Technology (10:4) 1995, pp. 239-247.
Kakabadse, A., and Kakabadse, N.: Trends in outsourcing. In: European Management Journal (20:2)
2002, pp. 189-198.
Kassner, S.: Erfolgsfaktor für Kreditinstitute im Wettbewerb. In: Presentations at the Bankakademie
Conference Sourcing, Frankfurt am Main, 2004.
Katre, A.U.: Governance: building successful outsourcing engagements. Kanbay, 2005.
http://www.kanbay.com/executive_connections11.asp (as of 02 June 2006).
Katz, M.L., and Shapiro, C.: Network externalities, competition, and compatibility. In: American
Economic Review (75:3) 1985, pp. 424-440.
Kavan, C., Saunders, C.S., and Nelson, R.: [email protected]. In: Business Horizons (42:5) 1999,
pp. 73-82.
Kelly, J.S.: Arrow impossibility theorems. Academic Press, New York, 1978.
Kern, T.: The gestalt of an information technology outsourcing relationship: an exploratory analysis.
18th International Conference on Information Systems (ICIS), Atlanta (GA), 1997.
Kern, T., and Willcocks, L.P.: Exploring relationships in information technology outsourcing: the
interaction approach. In: European Journal of Information Systems (11) 2002, pp. 3-19.
Ketterer, K.-H., and Ohmayer, E.: Die Transaktionsbank – Ein neuer Banktyp entsteht. In: Karlsruher
Transfer (17:29) 2003, pp. 7-12.
Khosrowpour, M., Subramanian, G., and Gunterman, J.: Outsourcing organizational benefits and
potential problems. In: Managing information technology investments with outsourcing, M.
Khosrowpour (ed.), Idea Group Publishing, Harrisburg, London, 1995, pp. 244-268.
Kiely, T.: Business process: consider outsourcing. In: Harvard Business Review (75:3) 1997, pp. 11-
12.
King, W.R., and Grover, V.: The strategic use of information resources: an exploratory study. In:
IEEE Transactions in Engineering Management (38:4) 1991, pp. 293-305.
References 435

King, W.R., and Malhotra, Y.: Developing a framework for analyzing IS sourcing. In: Information
and Management (37:6) 2000, pp. 323-334.
Klaemmt, T.: Outsourcing im Personalbereich. In: Information Management & Consulting (18:3)
2003, pp. 42-45.
Klein, B., Crawford, R., and Alchian, A.: Vertical integration, appropriable rents and the competitive
contracting process. In: Journal of Law and Economics (21:2) 1978, pp. 297-326.
Klein, E.: Outsourcing in Banken. IDC Finance Konferenz, IDC, Frankfurt am Main, 2004.
Klein, H.K., and Myers, M.D.: A set of principles for conducting and evaluating interpretive field
studies in information systems. In: MIS Quarterly (23:1) 1999, pp. 67-94.
Klepper, R.: The management of partnering development in I/S outsourcing. In: Journal of Informa-
tion Technology (10:4) 1995, pp. 249-258.
Klotz, D.E., and Chatterjee, K.: Dual sourcing in repeated procurement competitions. In: Manage-
ment Science (41:8) 1995, pp. 1317-1327.
Knaese, B.: Kernkompetenzen im strategischen Management von Banken: Der "Resource-based
view" in Kreditinstituten. Deutscher Universitäts-Verlag, Wiesbaden, Germany, 1996.
Knolmayer, G.: Modelle zur Unterstützung von Outsourcing-Entscheidungen. In: Wirtschaftsinfor-
matik '93, K. Kurbel (ed.), Springer, Heidelberg et al., 1993, pp. 70-83.
Knolmayer, G.: Zur Berücksichtigung von Transaktions- und Koordinationskosten in Entschei-
dungsmodellen für Make-or-Buy-Probleme. In: Betriebswirtschaftliche Forschung und Praxis
(46:4) 1994, pp. 316-332.
Knowledgestorm: Business process outsourcing: enabling information delivery. Knowledgestorm,
Alpharetta (GA), USA, 2004.
Knudsen, C.: Theories of the firm, strategic management, and leadership. In: Resource-based and
evolutionary theories of the firm: towards a synthesis, C.A. Montgomery (ed.), Kluwer, Boston
(MA), 1995, pp. 179-217.
Koch, M., and Rill, R.: Serviceorientierte Architekturen bei Finanzdienstleistern. ibi research GmbH,
Regensburg, Germany, 2005.
Koetter, M., Nestmann, T., Stolz, S., and Wedow, M.: Structures and trends in German banking.
Working paper No. 1225, Kiel Institute for World Economics, 2004. http://www.ifw-
kiel.de/pub/kap/2004/kap1225.pdf (as of 17 July 2006).
Kogut, B.: Joint ventures: theoretical and empirical perspectives. In: Strategic Management Journal
(9:4), 332 1988, p. 319.
Köhler, P., and Walter, N.: Die technische Umsetzung der "Basel "-Vorschriften verursacht Ausga-
ben in Milliardenhöhe. In: Handelsblatt, 15 Oct 2002, p. 23.
König, W., and Beimborn, D.: Zur Integrationstiefe in der zwischenbetrieblichen Zusammenarbeit.
In: The digital economy – Anspruch und Wirklichkeit, Festschrift für Beat F. Schmid, Sta-
noevska-Slabeva (ed.), Springer, Heidelberg et al., 2004, pp. 199-213.
König, W., and Beimborn, D.: Sourcing-Trends im KMU-Kreditgeschäft der deutschen Banken. In:
Outsourcing in Banken, B. Kaib (ed.), Gabler, 2nd ed., 2008, pp.183-209.
Kooymans, R.R.: The outsourcing of corporate real estate management – How do corporate real
estate units and outsource service providers view each other and the management issues? Paci-
fic Rim Real Estate Society Conference, Sydney, 2000.
Kopper, H.: Perspektiven der Universalbanken. In: Finanzplatz Deutschland an der Schwelle zum 21.
Jahrhundert, H.E. Büschgen (ed.), Fritz Knapp, Frankfurt am Main, 1998.
Kotabe, M.: Global sourcing strategy: R&D, manufacturing, and marketing interfaces. Quorum, New
York (NY), 1992.
Kotabe, M., and Murray, J.: Linking product and process innovations and modes of international
sourcing in global competition: a case of foreign multinational firms. In: Journal of Interna-
tional Business Studies (21:3) 1990, pp. 383-408.
436 References

Kotter, J.P.: Managing external dependence. In: Academy of Management Review (4:1) 1979, pp.
87-92.
Krabichler, T., and Krauß, I.: Konsolidierung im europäischen Bankenmarkt. ibi – Institut für Bank-
informatik und Bankstrategie an der Universität Regensburg, Regensburg, 2003.
Krahnen, J.-P.: Sunk Costs und Unternehmensfinanzierung. Gabler, Wiesbaden, 1991.
Kratica, J., Filipovic, V., and Tosic, D.: Solving the uncapacitated warehouse location problem by
SGA with ADD-heuristic. 15th International Conference on Material Handling and Warehou-
sing, Belgrade, 1998.
Krawietz, P., Lehmann, K., and Druba, S.: Leistungsvergleich der Serviceprovider für Kreditbackof-
fice-Dienstleistungen in Deutschland. BearingPoint, Frankfurt am Main, 2003.
Kreditwerk: Statusbericht aus der größten deutschen Kreditfabrik. Presentation on the “Hypotheken-
und Kreditprocessing 2006” symposium, Frankfurt, 28 Sep 2006. http://www.kreditwerk.de/i-
apps/kreditwerk/web-kw.nsf/acrobat/2006_09_28_Fachkonferenz_IuF_Statusbericht.pdf/ $fi-
le/2006_09_28_Fachkonferenz_IuF_ Statusbericht.pdf (as of 31 Jan 2007).
Kreikebaum, H.: Organisationsmanagement internationaler Unternehmungen. Grundlagen und neue
Strukturen. Gabler, Wiesbaden, Germany, 1998.
Kreps, D.M., Milgrom, P., Roberts, J., and Wilson, R.: Rational cooperation in the finitely repeated
prisoner's dilemma. In: Journal of Economic Theory (27:2) 1982, pp. 245-252.
Krichel, M., and Schwind, K.: Kreditfabriken im Baufinanzierungsgeschäft – Processing im Span-
nungsfeld von Mensch und Technik. In: Die Bank (43:11) 2003, pp. 49-51.
Krüger, W., and Homp, C.: Kernkompetenz-Management: Steigerung von Flexibilität und Schlag-
kraft im Wettbewerb. Gabler, Wiesbaden, Germany, 1997.
Kshetri, N., and Williamson, N.: The osmosis model for studying offshore business process outsourc-
ing. 10th Americas Conference on Information Systems (AMCIS), New York (NY), 2004.
Kuritzkes, A., Harris, S., and Strothe, G.: Redesign des Kreditprozesses. In: Die Bank (40:1) 2000,
pp. 42-47.
Lacity, M.C., and Hirschheim, R.: The information systems outsourcing bandwagon. In: Sloan Man-
agement Review (35:1) 1993a, pp. 73-86.
Lacity, M.C., and Hirschheim, R.: Information systems outsourcing: myths, metaphor and realities.
John Wiley, New York (NY), 1993b.
Lacity, M.C., and Willcocks, L.P.: Interpreting information technology sourcing decisions from a
transaction cost perspective: findings and critique. In: Accounting, Management and Informa-
tion Technologies (5:3/4) 1995, pp. 203-244.
Lacity, M.C., and Willcocks, L.P.: An empirical investigation of information technology sourcing
practices: lessons from experience. In: MIS Quarterly (22:3) 1998, pp. 363-408.
Lacity, M.C., and Willcocks, L.P.: IT sourcing reflections – Lessons for customers and suppliers. In:
Wirtschaftsinformatik (45:2) 2003, pp. 115-125.
Lacity, M.C., Hirschheim, R., and Willcocks, L.P.: Realizing outsourcing expectations. In: Informa-
tion Systems Management (11:4) 1994, pp. 363-408.
Lacity, M.C., Willcocks, L.P., and Feeny, D.F.: IT outsourcing: maximize flexibility and control. In:
Harvard Business Review (73:3) 1995, pp. 84-93.
Lacity, M.C., Willcocks, L.P., and Feeny, D.F.: The value of selective IT sourcing. In: Sloan Man-
agement Review (37:3) 1996, pp. 13-25.
Lado, A.A., Boyd, N.G., and Hanlon, S.C.: Competition, cooperation, and the search for economic
rents: a syncretic model. In: Academy of Management Review (22:1) 1997, pp. 110-141.
Laffer, A.B.: Vertical integration by corporations. In: The Review of Economics and Statistics (51:1)
1969, pp. 91-93.
Lambert, R.: Long-term contracts and moral hazard. In: Bell Journal of Economics (14:2) 1983, pp.
441-452.
References 437

Lamberti, H.-J.: Neue Herausforderungen für Banken und Aufsicht. In: Zeitschrift für das gesamte
Kreditwesen (10), 2005, pp. 510-522.
Lamberti, H.-J., and Pöhler, A.: Die Industrialisierung des Backoffice am Beispiel der etb. In: Man-
agement von Transaktionsbanken, H.-J. Lamberti, A. Marlière and A. Pöhler (eds.), Springer,
Heidelberg et al., 2004, pp. 3-38.
Lammers, M.: Make, buy or share – Combining resource based view, transaction cost economics and
production economies to a sourcing framework. In: Wirtschaftsinformatik (46:3) 2004, pp.
204-212.
Lammers, M.: Sourcing decision making in the banking industry. Pro Business, Berlin, 2005.
Lammers, M., Löhndorf, N., and Weitzel, T.: Strategic sourcing in banking – A framework. 12th
European Conference on Information Systems (ECIS), Turku, Finland, 2004.
Lancellotti, R., Schein, O., Spang, S., and Stadler, V.: ICT and operations outsourcing in banking –
Insights from an interview-based pan-European survey. In: Wirtschaftsinformatik (45:2) 2003,
pp. 131-141.
Landis, K.M., Mishra, S., and Porrello, K.: Calling a change in the outsourcing market. Deloitte
Consulting, New York (NY), 2005.
Landry, M., and Banville, C.: A disciplined methodological pluralism for MIS research. In: Account-
ing, Management and Information Technologies (2:2) 1992, pp. 77-97.
Lang, G., and Welzel, P.: Technology and cost efficiency in universal banking: a "thick frontier"-
analysis of the German banking industry. In: Journal of Productivity Analysis (10:1) 1998, pp.
63-84.
Langlois, R.: Transaction-cost economics in real time. In: Industrial and Corporate Change (1:1)
1992, pp. 99-127.
Langlois, R.: Capabilities and coherence in firms and markets. In: Resource-based and evolutionary
theories of the firm: towards a synthesis, C.A. Montgomery (ed.), Kluwer, Boston (MA), 1995,
pp. 71-100.
Langlois, R.N., and Robertson, P.L.: Networks and innovation in a modular system: lessons from the
microcomputer and stereo component industries. In: Research Policy (21:4) 1992, pp. 297-313.
Large, J.: How networks net business. In: Management Today, February 1987, pp. 86-94.
Laudan, L.: Science and values: an essay on the aims of science and their role in scientific debate.
University of California Press, Berkeley (CA), USA, 1984.
Lebart, L., Salem, A., and Berry, L.: Exploring textual data. Kluwer, Dordrecht, Netherlands, 1997.
Lee, H.L., and Whang, S.: The impact of the secondary market on the supply chain. In: Management
Science (48:6) 2002, pp. 719-731.
Lee, J.-N.: Outsourcing reference list. www.is.cityu.edu.hk/staff/isjnlee/out_frame_tot.htm (as of 20
Mar 2005).
Lee, J.-N.: The impact of knowledge sharing, organizational capability and partnership quality on IS
outsourcing success. In: Information and Management (38:5) 2001, pp. 323-335.
Lee, J.-N., Huynh, M.Q., Chi-Wai, K.R., and Pi, S.-M.: The evolution of outsourcing research: what
is the next issue? 33rd Hawaii International Conference on System Sciences (HICSS-33),
Wailea, Maui (HI), USA, 2000.
Lee, J.-N., Huynh, M.Q., Kwok, R.C.-W., and Pi, S.-M.: IT Outsourcing evolution – Past, present,
and future. In: Communications of the ACM (46:5) 2003, pp. 84-89.
Lee, J.-N., and Kim, Y.-G.: Effect of partnership quality on IS outsourcing success: conceptual
framework and empirical validation. In: Journal of Management Information Systems (15:4)
1999, pp. 29-61.
Lee, J.-N., and Kim, Y.-G.: Exploring a causal model for the understanding of outsourcing partner-
ship. 36th Hawaii International Conference on System Sciences (HICSS-36), Waikoloa, Big Is-
land (HI), USA, 2003.
438 References

Lee, J.-N., Miranda, S.M., and Kim, M.: IT outsourcing strategies: universalistic, contingency, and
configurational explanations of success. In: Information Systems Research (15:2) 2004, pp.
110-131.
Lehmann, K.: Kreditbackoffice-Bearbeitung in Deutschland – Erste Ergebnisse. UNISYS, Aareal
Hypothekenmanagement Informationsforum Kreditservicing 2005, Frankfurt am Main, 2005-
11-16.
Lehnsdorf, L., and Schneider, J.: Das Rundschreiben des Bundesaufsichtsamtes für Kreditwesen zur
Auslagerung von wesentlichen Bereichen von Kredit- und Finanzdienstleistungsinstituten auf
andere Unternehmen gemäß §25a Abs. 2KWG. In: WM Wertpapiermitteilungen (39), 2002,
pp. 1949-1956.
Lei, D., and Hitt, M.: Strategic restructuring and outsourcing: the effect of mergers and acquisitions
and LBOs on building firm skills and capabilities. In: Journal of Management (21:5) 1995, pp.
835-859.
Leimstoll, U., and Schubert, P.: Integration von Business Software – Eine Studie zum aktuellen Stand
in Schweizer KMU. 7. Internationale Tagung Wirtschaftinformatik 2005, Bamberg, Germany,
2005, pp. 983-1002.
Leland, H.E., and Pyle, D.H.: Informational asymmetries, financial structure, and financial interme-
diation. In: The Journal of Finance (32:2) 1977, pp. 371-387.
Lemaire, J.: An application of game theory: cost allocation. In: ASTIN Bulletin (14:1) 1984, pp. 61-
81.
Leontieff, W.W.: The structure of American economy 1919-1939. Oxford University Press, New
York (NY), 1951.
Levina, N., and Ross, J.W.: From the vendor's perspective: exploring the value proposition in infor-
mation technology outsourcing. In: MIS Quarterly (27:3) 2003, pp. 331-364.
Levinthal, D.: A survey of agency models of organization. In: Journal of Economic Behavior and
Organization (9:2) 1988, pp. 153-186.
Lewis, T., and Sappington, D.E.M.: Technological change and the boundaries of the firm. In: Ameri-
can Economic Review (81:4) 1991, pp. 887-900.
Li, F.: The geography of business information. John Wiley, Chichester, 1995.
Licci, C.: Kann die Industrie den Banken ein Vorbild sein? In: Kreditwesen (7), 2003, pp. 42-44.
Liebowitz, S.J., and Margolis, S.E.: Path dependence, lock-in, and history. In: Journal of Law, Eco-
nomics and Organization (11:4) 1995, pp. 205-225.
Liebowitz, S.J., and Margolis, S.E.: Winners, losers & Microsoft: competition and antitrust in high
technology. The Independent Institute, Oakland (CA), USA, 1999.
Lindblom, C.E.: Alternatives to validity: some thoughts suggested by Campbell's guidelines. In:
Science Communication (8:3) 1987, pp. 509-520.
Linder, J., Cantrell, S., and Christ, S.: Business process outsourcing big bang: creating value in an
expanding universe. Accenture Institute for Strategic Change, Chicago (IL), 2002.
Linn, N.: Das Geheimnis des Erfolges – Was Hochleistungsbanken auszeichnet. In: Frankfurter
Allgemeine Zeitung (F.A.Z.), 06 Apr 2005, p. B5.
Lippman, S.A., and McCardle, K.F.: The competitive newsboy. In: Operations Research (45:1) 1997,
pp. 54-65.
Loetto, T., Remy, R., and Rothe, A.: Basel II – Ratingverfahren der Landesbanken. In: HMD - Praxis
der Wirtschaftsinformatik (233) 2003, pp. 33-42.
Logan, M.: Using agency theory to design successful outsourcing relationships. In: International
Journal of Logistics Management (11:2) 2000, pp. 21-32.
Loh, L.: An organizational-economic blueprint for information technology outsourcing: concepts and
evidence. 15th International Conference on Information Systems (ICIS), Vancouver, Canada,
1994, pp. 73-89.
References 439

Loh, L., and Venkatraman, N.: Information technology outsourcing as an administrative innovation:
imitative behavior as an explanation of the diffusion pattern. MIT Sloan School of Manage-
ment, Cambridge (MA), USA, 1991.
Loh, L., and Venkatraman, N.: Determinants of information technology outsourcing: a cross-
sectional analysis. In: Journal of Management Information Systems (9:1) 1992a, pp. 7-24.
Loh, L., and Venkatraman, N.: Diffusion of IT outsourcing: influence sources and the Kodak effect.
In: Information Systems Research (3:4) 1992b, pp. 334-358.
Loh, L., and Venkatraman, N.: An empirical study of information technology outsourcing: benefits,
risks, and performance. 16th International Conference on Information Systems (ICIS), Amster-
dam, Netherlands, 1995, pp. 277-288.
Luiz de Medeiros Rivero, S., Storb, B.H., and Wazlawick, R.S.: Economic theory, anticipatory
systems, and artificial adaptative agents. In: Brazilian Economic Journal on Economics (2:2)
1999.
Lyles, M.A.: Learning among joint venture sophisticated firms. In: Management International Re-
view (28:Special Issue) 1988, pp. 85-97.
Maasjost, W.: Business Reengineering als Methode. In: Outsourcing in der Informationstechnologie
– Eine strategische Management-Entscheidung, J. Berg and H. Gräber (eds.), Campus, Frank-
furt am Main, 1995, pp. 51-59.
Macaulay, S.: Non-contractual relations in business: a preliminary study. In: American Sociological
Review (28:1) 1963, pp. 55-67.
Macneil, I.R.: The many futures of contracts. In: Southern California Law Review (47:5) 1974, pp.
691-816.
Maddigan, R.J.: The measurement of vertical integration. In: The Review of Economics and Statistics
(63:3) 1981, pp. 328-335.
Maddigan, R.J., and Zaima, J.K.: The profitability of vertical integration. In: Managerial and Deci-
sion Economics (6:3) 1985, pp. 178-179.
Mahoney, J.T., and Pandian, R.: The resource-based view within the conversation of strategic man-
agement. In: Strategic Management Journal (13:5) 1992, pp. 363-380.
Mahr, H.: Arbeitsrechtliche Grundlagen. In: IT-Outsourcing – Eine Darstellung aus rechtlicher,
technischer, wirtschaftlicher und vertraglicher Sicht, P. Bräutigam (ed.), Erich Schmidt, Berlin,
2004, pp. 367-427.
Malkamäki, M.: Are there economies of scale in stock exchange activities? Bank of Finland, Hel-
sinki, 1999.
Malone, T.W., and Crowston, K.: What is coordination theory and how can it help design cooperative
work systems. Third Conference on Computer-supported Cooperative Work, Los Angeles
(CA), 1990, pp. 357-370.
Malone, T.W., and Crowston, K.: The interdisciplinary theory of coordination. In: Computing Sur-
veys (26:1) 1994, pp. 87-119.
Malone, T.W., and Rockart, J.F.: Computers, networks, and the corporation, in: Scientific American
(265:3) 1991, pp. 128-136.
Malterud, K.: Qualitative research: standards, challenges, and guidelines. In: The Lancet (358:9280),
11 Aug 2001, pp. 483-488.
Mani, D., Barua, A., and Whinston, A.B.: Successfully governing business process outsourcing
relationships. In: MIS Quarterly Executive (5:1) 2006, pp. 15-29.
Manrodt, K.B., and Vitasek, K.: Global process standardization: a case study. In: Journal of Business
Logistics (25:1) 2004, pp. 1-23.
Marcolin, B.L.: Spiraling effect of IS outsourcing contract interpretations. In: Information Systems
Outsourcing – Enduring themes, emergent patterns and future directions, R.A. Hirschheim, A.
Heinzl and J. Dibbern (eds.), Springer, Heidelberg et al., 2002, pp. 277-310.
440 References

Marcolin, B.L., and Ross, A.: Complexities in IS sourcing: equifinality and relationship management.
In: The DATA BASE for Advances in Information Systems (36:4) 2005, pp. 29-46.
Margulius, D.L.: Betting on BTO. In: CIO Magazine (20), 15 Oct 2003,
Marighetti, L.P., Herrmann, A., and Hänsler, N.: Herausforderungen an das Management von Wert-
schöpfungsketten. In: Management der Wertschöpfungsketten in Banken, L.P. Marighetti, R.
Jasny, A. Herrmann and F. Huber (eds.), Gabler, Wiesbaden, 2001, pp. 13-23.
Marlière, A. Die Grenzen von Banken, Dissertation thesis, St. Gallen, Switzerland, 2002.
Marlière, A.: Effizientes Management von Sourcing-Aktivitäten durch Service Level Agreements.
NetCo AG, Aareal HM Kundeninformationsforum, Frankfurt am Main, 2004a.
Marlière, A.: Service Level Management – Geordnete Liefer- und Leistungsbeziehungen in fünf
Schritten. In: Materna Monitor (3), 2004b, pp. 9-12.
Marsh, S.: Trust and reliance in multi-agent systems: a preliminary report. Department of Computer
Science and Mathematics, University of Stirling, 1992.
Martin, S.: Causes and effects of vertical integration. In: Journal of Applied Economics (18:7) 1986,
pp. 737-755.
Martinsons, M.G.: Outsourcing information systems: a strategic partnership with risks. In: Long
Range Planning (26:3) 1993, pp. 18-25.
Matiaske, W., and Mellewigt, T.: Motive, Erfolge und Risiken des Outsourcings – Befunde und
Defizite der empirischen Outsourcing-Forschung. In: Zeitschrift für Betriebswirtschaft (72:6)
2002, pp. 641-659.
McCarthy, J.C.: BPO's fragmented future. Forrester Research, Cambridge (MA), USA, 2003.
McConnell, J., and Nantell, J.: Common stock returns and corporate combinations. The case of joint
ventures. In: Journal of Finance (40:2) 1985, pp. 519-536.
McDougall: New deals may be start of something big. In: Information Week, November 18 2002,
McFarlan, F., and Nolan, R.: How to manage an IS outsourcing alliance. In: Sloan Management
Review (36:2) 1995, pp. 9-23.
McGovern, J., Sims, O., Jain, A., and Little, M.: Enterprise Service Oriented Architectures: concepts,
challenges, recommendations. Springer, Heidelberg et al., 2006
McLellan, K.L., Marcolin, B.L., and Beamish, P.W.: Financial and strategic motivations behind IS
outsourcing. In: Journal of Information Technology (10:4) 1995, pp. 299-321.
Medeiros Rivero, S.L.d., Storb, B.H., and Wazlawick, R.S.: Economic theory, anticipatory systems,
and artificial adaptative agents. In: Brazilian Economic Journal of Economics (online journal)
(2:2) 1999.
Menner, S.: Umsatzsteuerfragen bei TX-Bankleistungen. In: Management von Transaktionsbanken,
H.-J. Lamberti, A. Marlière and A. Pöhler (eds.), Springer, Heidelberg et al., 2004, pp. 357-
373.
Mensik, M.: Should companies outsource Sarbanes-Oxley compliance? In: Outsourcing Essentials
(2:3) 2004.
Mertens, P., Griese, J., Ehrenberg, D.: Virtuelle Unternehmen und Informationsverarbeitung. Sprin-
ger, Heidelberg et al., 1998.
Meyer, M., and Schumacher, J.: Outsourcing von CRM-Teilprozessen an Betreiber von Internet-
marktplätzen – Möglichkeiten und Vorteile für Anbieter und Betreiber. In: Wirtschaftsinforma-
tik (45:2) 2003, pp. 165-175.
Middendorf, J., and Göttlicher, C.: Visionen für den Transaction Banking-Markt. BearingPoint,
Frankfurt am Main, 2003.
Milgrom, P., and Holmström, B.: Multi-task principal-agent analyses: incentive contracts, asset
ownership and job design. In: Journal of Law, Economics and Organization (7:Special Issue)
1991, pp. 24-52.
Milgrom, P., and Roberts, J.: Economics, organization and management. Prentice-Hall, Englewood
Cliffs (NJ), USA, 1992.
References 441

Mintzberg, H.: Policy as a field of management theory. In: Academy of Management Review (2:1)
1977, pp. 88-103.
Miranda, S.M., and Kim, Y.: IT outsourcing strategies: universalistic, contingency and configura-
tional explanations of success. In: Information Systems Research (15:2) 2004, pp. 110-131.
Mohr, J., and Spekman, R.: Characteristics of partnership success: partnership attributes, communica-
tion behavior, and conflict resolution techniques. In: Strategic Management Journal (15:2)
1994, pp. 135-152.
Monteverde, K.: Technical dialog as an incentive for vertical integration in the semiconductor indus-
try. In: Management Science (41:10) 1995, pp. 1624-1638.
Monteverde, K., and Teece, D.: Supplier switching costs and vertical integration in the automobile
industry. In: The Bell Journal of Economics (13:1) 1982, pp. 206-213.
Montgomery, C.A., and Wernerfelt, B.: Diversification, Ricardian rents, and Tobin's q. In: RAND
Journal of Economics (19:4) 1988, pp. 623-632.
Moore, J.: The firm as a collection of assets. In: European Economic Review (36:2-3) 1992, pp. 493-
507.
Moormann, J., and Möbius, D.: Wertschöpfungsmanagement in Banken. Bankakademie-Verlag,
Frankfurt am Main, 2004.
Mosakowski, E., and McKelvey, B.: Predicting rent generation in competence-based competition. In:
Competence-based strategic management, A. Heene and R. Sanchez (eds.), John Wiley, New
York (NY), 1997, pp. 65-85.
Müller, D.: Business Process Outsourcing legt in Deutschland deutlich zu. ZDNET, 2005.
http://www.zdnet.de/itmanager/strategie/0,39023331,39130714,00.htm (as of 25 Feb 2005).
Mui, L., Halberstadt, A., and Mohtashemi, M.: Notions of reputation in multi-agents systems: a
review. First International Joint Conference on Autonomous Agents and Multiagent Systems,
ACM Press (New York), Bologna, Italy, 2002a.
Mui, L., Mohtashemi, M., and Halberstadt, A.: A computational model of trust and reputation. 35th
Hawaii International Conference on System Sciences (HICSS-35), Waikoloa, Big Island (HI),
USA, 2002b.
Mummert: Branchenkompass 2005: Kreditinstitute – Aktuelle Entscheiderbefragung. Mummert
Consulting, steria, F.A.Z.-Institut, Frankfurt am Main, 2005.
Murray, J.D., and White, R.W.: Economies of scale and economies of scope in multiproduct financial
institutions: a study of British Columbia credit unions. In: Journal of Finance (38:3) 1983, pp.
887-890.
N.N.: Outsourcing 2007. Accenture, Kronberg, Germany, 2002a.
N.N.: The value of business process outsourcing. CSC, Falls Church (VA), USA, 2002b.
N.N.: Top 1,000 banks. In: The Banker (7) 2002c, pp. 216-249.
N.N.: Sparkasse Bremen mit "Kreditfabrik. In: Frankfurter Allgemeine Zeitung (F.A.Z.), 26 Feb
2003, p. 13.
N.N.: Dresdner Bank unterstützt Kreditfabrik. In: Frankfurter Allgemeine Zeitung (F.A.Z.), 28 Aug
2004, p. 10.
N.N.: Aareal verkauft Kreditfabrik. In: Frankfurter Allgemeine Zeitung (F.A.Z.), 15 Dec 2005a, p.
18.
N.N.: Gespräche über Kreditfabrik. In: Frankfurter Allgemeine Zeitung (F.A.Z.), 26 Apr 2005b, p.12.
N.N.: Übernahme der Eurohypo verunsichert Commerzbank-Anleger. In: Frankfurter Allgemeine
Zeitung (F.A.Z.), 15 Nov 2005c, p. 14.
Nadig, L.: Prozesskostenrechnung in Theorie und Praxis. Schulthess, Zürich, 2000.
Nag, B.: Business process outsourcing: impact and implications. In: Bulletin on Asia-Pacific Perspec-
tives, 2004, pp. 59-73. http://www.unescap.org/pdd/publications/bulletin04-05/bulletin04-
05_ch4.pdf (as of 13 July 2006).
442 References

Nagengast, J.: Outsourcing von Dienstleistungen industrieller Unternehmen. Dr. Kovac, Hamburg,
1997.
Nam, K., Rajagopalan, S., Rao, R., and Chaudhury, A.: A two-level investigation of information
systems. In: Communications of the ACM (39:7) 1996, pp. 37-44.
Namasivayam, S.: Profiting from business process outsourcing. In: IT Professional (1), 2004, pp. 12-
18.
Narus, J., and Anderson, J.: Distribution contributions to partnerships with manufacturers. In: Busi-
ness Horizons (30:1) 1987, pp. 34-42.
Nash, J.F.: The bargaining problem. In: Econometrica (18) 1950, pp. 155-162.
Naylor, T.H., Finger, J.M., McKenney, J.L., Schrank, W.E., and Holt, C.C.: Verification of computer
models. In: Management Science (14:2) 1967, pp. B92-B106.
Nees, F.: Produktivität wird wichtiger als Funktionalität. In: Geldinstitute (4), 2005, p. 16.
Nelson, P., Richmond, W., and Seidman, A.: Two dimensions of software acquisition. In: Communi-
cations of the ACM (39:7) 1996, pp. 29-35.
Nelson, R.L.: Concentration in the manufacturing industries of the United States. Yale University
Press, New Haven (CT), USA, 1963.
Neymann, A.: Bounded complexity justifies cooperation in the finitely repeated prisoners' dilemma.
In: Economic Letters (19:3) 1985, pp. 227-229.
Nickerson, J.A., Hamilton, B.H., and Tetsuo, W.: Market position, resource profile, and governance:
linking Porter and Williamson in the context of international courier and small package ser-
vices in Japan. In: Strategic Management Journal (22:3) 2001, pp. 251-273.
Nissen, V.: Evolutionäre Algorithmen. In: Wirtschaftsinformatik (37:4) 1995, pp. 393-397.
Nitz, R., Fürst, M., and Gutzwiller, T.: Vorgehen bei der Evaluation von IT-Plattformen für Transak-
tionsbanken. In: Management von Transaktionsbanken, H.-J. Lamberti, A. Marlière and A.
Pöhler (eds.), Springer, Heidelberg et al., 2004, pp. 263-297.
Nueno, P., and Oosterveld, J.: Managing technology alliances. In: Long Range Planning (21:3) 1988,
pp. 11-17.
Nunnally, J.C.: Psychometric theory. McGraw Hill, New York, 1978.
Oh, W., and Gallivan, M.J.: An empirical assessment of transaction risks of IT outsourcing arrange-
ments: An event study. 37th Hawaii International Conference on System Sciences (HICSS-37),
Waikoloa, Big Island (HI), USA, 2004.
Ohmae, K.: The global logic of strategic alliance. In: Harvard Business Review (67:2) 1989, pp. 143-
154.
Oliver, C.: Determinants of interorganizational relationships: integration and future directions. In:
Academy of Management Review (15:2) 1990, pp. 241-265.
Olson, M.: The logic of collective action. Harvard University Press, Cambridge (MA), 1965.
Onvista: Onvista Aktienanalyse. 2004. http://www.onvista.de (as of 20 Oct 2006).
Orlikowski, W.J., and Baroudi, J.J.: Studying information technology in organizations: research
approaches and assumptions. In: Information Systems Research (2:1) 1991, pp. 1-28.
Orton, J.D., and Weick, K.E.: Loosely coupled systems: a reconceptualization. In: Academy of
Management Review (15:2) 1990, pp. 203-223.
Osband, K., and Reichelstein, S.: Information-eliciting compensation schemes. In: Journal of Public
Economics) 1985, pp. 177-192.
Osborn, R.N., and Hagedoorn, J.: The institutionalization and evolutionary dynamics of interorgani-
zational alliances and networks. In: Academy of Management Journal (40:2) 1997, pp. 261-
278.
Osborne, M., and Rubinstein, A.: A course in game theory. MIT Press, Cambridge (MA), 1994.
Owen, G.: Game theory. (3rd ed.) Academic Press, New York (NY), 1995.
Palvia, P.C.: A dialectic view of information systems outsourcing: pros and cons. In: Information
Management (29:5) 1995, pp. 265-275.
References 443

Panzar, J.C., and Willig, R.D.: Economies of scale and economies of scope in multi-output produc-
tion. In: Quarterly Journal of Economics (91:3) 1977, pp. 481-494.
Panzar, J.C., and Willig, R.D.: Economies of scope. In: American Economic Review (71:2) 1981, pp.
268-272.
Parker, P.M.: Aggregate diffusion forecasting models in marketing: a critical review. In: International
Journal of Forecasting (10:2) 1994, pp. 353-380.
Parkhe, A.: Messy research, methodological predispositions, and theory development in international
joint ventures. In: Academy of Management Review (18:2) 1993, pp. 227-268.
Parsian, A., Nematollahi, N., Allen, L., and Rai, A.: Operational efficiency in banking: an interna-
tional comparison. In: Journal of Banking and Finance (20:4) 1996, pp. 655-672.
Peak, D., Windsow, J., and Conover, J.: Risks and effects of IS/IT outsourcing: a securities market
assessment. In: Journal of Information Technology Cases and Applications (4:1) 2002, pp. 6-
33.
Peng, Y., and Wenhua, H.: A framework of total performance improvement and transaction cost-
driven business process outsourcing strategy. 8th Pacific-Asia Conference on Information Sys-
tems (PACIS), Shanghai, China, 2004.
Penrose, E.T.: The theory of the growth of the firm. Oxford University Press, Oxford, UK, 1959.
Perrow, C.: Complex organizations. Random House, New York (NY), 1986.
Peteraf, M.A.: The cornerstones of competitive advantage: a resource-based view. In: Strategic
Management Journal (14:3) 1993, pp. 179-191.
Petry, T., and Rohn, H.: Deconstruction in der Bankenbranche. In: Zeitschrift Führung + Organisati-
on (74:5) 2005, pp. 265-272.
Petzel, E.: Effizienzsteigerung durch Outsourcing? In: S-Management Praxis (45:6) 2003, pp. 61-72.
Pfannenstein, L.L., and Ray, J.T.: Offshore outsourcing: current and future effects on American IT
industry. In: Information Systems Management (21:4) 2004, pp. 72-80.
Pfeffer, J.: Organizations and organization theory. Pitman, Boston (MA), 1982.
Pfeffer, J., and Salancik, G.: The external control of organizations: a resource dependence perspec-
tive. Harper & Row, New York (NY), 1978.
Pfeiffer, W., Weiß, E., and Strubl, C.: Systemwirtschaftlichkeit. (2nd ed.) Vandenhoek & Ruprecht,
Göttingen, 1999.
Picot, A.: Transaktionskostenansatz in der Organisationstheorie: Stand der Diskussion und Aussage-
wert. In: Die Betriebswirtschaft (42:2) 1982, pp. 267-284.
Picot, A.: Ein neuer Ansatz zur Gestaltung der Leistungstiefe. In: Zeitschrift für betriebswirtschaftli-
che Forschung (43:4) 1991, pp. 336-357.
Picot, A.: Marktorientierte Gestaltung der Leistungstiefe. In: Marktnahe Produktion, R. Reichwald
(ed.), Gabler, Wiesbaden, 1992, pp. 103-124.
Picot, A., and Dietl, H.: Transaktionskostentheorie. In: Wirtschaftswissenschaftliches Studium (19:4)
1990, pp. 178-184.
Picot, A., Dietl, H., and Franck, E.: Organisation: Eine ökonomische Perspektive. (2 ed.) Schäffer-
Poeschel, Stuttgart, 1999.
Picot, A., Laub, U., and Schneider, D.: Innovative Unternehmensgründungen – Eine ökonomisch-
empirische Analyse. Springer, Heidelberg et al., 1989.
Picot, A., Reichwald, R., and Wigand, R.T.: Die grenzenlose Unternehmung – Information, Organi-
sation und Management. Springer, Heidelberg et al., 2001.
Pieske, R.: Hypotheken-Portfolios effizient verwalten. In: Versicherungswirtschaft (21:17) 2005, pp.
1300-1302.
Piller, T.: Mass Customization – Ein wettbewerbsstrategisches Konzept im Informationszeitalter.
Gabler, Wiesbaden, Germany, 2000.
444 References

Pilling, B.K., Crosby, L.A., and Jackson, D.W.: Relational bonds in industrial exchange: an experi-
mental test of the transaction cost economic framework. In: Journal of Business Research
(30:3) 1994, pp. 237-251.
Pisano, G.P.: The R&D boundaries of the firm: an empirical analysis. In: Administrative Science
Quarterly (35:1) 1990, pp. 153-176.
Plambeck, E.L., and Taylor, T.A.: Sell the plant? The impact of contract manufacturing on innova-
tion, capacity and profitability. In: Management Science (51:1) 2001, pp. 133-150.
Platzer, A., and Riess, W.: Finanzierung über Kredite. In: Die neue Unternehmensfinanzierung, W.
Stadler (ed.), Redline Wirtschaftsverlag, Frankfurt am Main, 2004, pp. 154-168.
Polster, D.: Verbundexterne Kooperation von Genossenschaftsbanken – Möglichkeiten, Grenzen,
Alternativen. Westfälische Wilhelms-Universität, Münster, 2001.
Pondy, L.R.: Toward a theory of internal resource allocation. In: Power in organizations, N.Z. Mayer
(ed.), Vanderbilt University Press, Nashville (TN), USA, 1970, pp. 270-311.
Poppo, L., and Lacity, M.C.: The normative value of transaction cost economics: what managers
have learned about TCE principles in the IT context. In: Information Systems Outsourcing –
Enduring themes, emergent patterns and future directions, R. Hirschheim, A. Heinzl and J.
Dibbern (eds.), Springer, Heidelberg et al., 2002, pp. 253-276.
Poppo, L., and Zenger, T.R.: Testing alternative theories of the firm: transaction cost, knowledge-
based, and measurement explanations for make-or-buy decisions in information services. In:
Strategic Management Journal (19:9) 1998, pp. 853-877.
Poppo, L., and Zenger, T.R.: Do formal contracts and relational governance function as substitutes or
complements? In: Strategic Management Journal (23:8) 2002, pp. 707-725.
Porter, M.E.: Competitive Advantage. Creating and sustaining superior performance. Free Press,
New York (NY), 1985.
Porter, M.E.: Towards a dynamic theory of strategy. In: Strategic Management Journal (12:4) 1991,
pp. 95-117.
Porter, M.E.: What is strategy? In: Harvard Business Review (74:6) 1996, pp. 61-78.
Porter, M.E., and Fuller, M.B.: Coalitions and global strategy. In: Competition in global industries,
M.E. Porter (ed.), Harvard Business School Press, Boston (MA), 1986, pp. 315-344.
Porter, M.E., and Millar, V.E.: How information gives you competitive advantage. In: Harvard
Business Review (63:4) 1985, pp. 149-174.
Powell, W.W.: Expanding the scope of institutional analysis. In: The new institutionalism in organ-
izational analysis, W.W. Powell and P.J. DiMaggio (eds.), University of Chicago Press, Chi-
cago (IL), 1991, pp. 183-203.
Prahalad, C.K., and Hamel, G.: The core competence of the corporation. In: Harvard Business Re-
view (68:3) 1990, pp. 79-93.
Priem, R., and Butler, J.E.: Is the resource-based "view" a useful perspective for strategic manage-
ment research? In: Academy of Management Review (26:1) 2001, pp. 22-40.
Prokein, O., and Faupel, T.: Using web services for intercompany cooperation – an empirical study
within the German industry. 39th Hawaii International Conference on System Sciences
(HICSS-39), Koloa, Kauai (HI), USA, 2006.
Pujals, G.: The inaugural EUROFRAME – EFN spring 2005 report "economic assessment of the
Euro area: forecasts and policy analysis," appendix 7: offshore outsourcing in the EU financial
services industry. The EUROFRAME Network, Dublin, Ireland, 2005. http://www. eu-
roframe.org/fileadmin/user_upload/euroframe/efn/spring2005/appendix7_pujals.pdf (as of 13
July 2006).
Quan, J., Hu, Q., and Hart, P.J.: Information technology investments and firms' performance – A
duopoly perspective. In: Journal of Management Information Systems (20:3) 2003, pp. 121-
158.
References 445

Quélin, B., and Duhamel, F.: Bringing together strategic outsourcing and corporate strategy: out-
sourcing motives and risks. In: European Management Journal (21:5) 2003, pp. 647-661.
Quinn, J.B.: Intelligent enterprise: a knowledge and service based paradigm for industry. Free Press,
New York (NY), 1992.
Quinn, J.B., Doorley, T.L., and Paquette, P.C.: Technology in services: rethinking strategic focus. In:
Sloan Management Review (31:2) 1990, pp. 79-87.
Quinn, J.B., and Hilmer, F.G.: Strategic outsourcing. In: Sloan Management Review (35:4) 1994, pp.
43-55.
Radecki, L., and Wenninger, J.: Industry structure: electronic delivery's potential effects on retail
banking. In: Journal of Retail Banking Services (19:4) 1997.
Radner, R.: Can bounded rationality resolve the prisoner's dilemma? Harvard University, Cambridge
(MA), 1979.
Rampl, D.: Dienstleistungsfabrik – Vorwort. In: Managementkompass, Mummert Consulting und
F.A.Z. Institut, 2003, pp. 2-3.
Rapoport, A.: N-person game theory: concepts and applications. Dover Publications, Inc., Mineola
(NY), USA, 2001.
Rasche, C.: Wettbewerbsvorteile durch Kernkompetenzen. Ein ressourcenorientierter Ansatz. Dt.
Universitätsverlag, Wiesbaden, 1994.
Ray, G., Barney, J.B., and Muhanna, W.A.: Capabilities, business processes, and competitive advan-
tage: choosing the dependent variable in empirical tests of the resource-based view. In: Strate-
gic Management Journal (25:1) 2004, pp. 23-37.
Rebouillon, J., and Matheis, J.: Outsourcing – Strategische Option im Back-Office. In: Management
von Transaktionsbanken, H.-J. Lamberti, A. Marlière and A. Pöhler (eds.), Springer, Heidel-
berg et al., 2004, pp. 331-356.
Reed, R., and DeFillippi, R.: Casual ambiguity, barrier to imitation and sustainable competitive
advantage. In: Academy of Management Review (15:1) 1990, pp. 80-102.
Reed, R., and Fronmueller, M.P.: Vertical integration: a comparative analysis of performance and
risk. In: Managerial and Decision Economics (11:3) 1990, pp. 177-185.
Reich, R., and Mankin, E.G.: Joint ventures with Japan give away our future. In: Harvard Business
Review (64:2) 1986, pp. 78-90.
Reinecke, S.: Marketing von IT-Outsourcing-Kooperationen aus Anbietersicht. Dissertation thesis,
University of St. Gallen, Switzerland, 1996.
Reponen, T.: Outsourcing or insourcing? 14th International Conference on Information Systems
(ICIS), Orlando (FL), USA, 1993, pp. 103-116.
Rettig, D.: Wie die "Kreditfabrik" Dampf machen soll. Manager-Magazin.de, 2004.
http://www.manager-magazin.de/unternehmen/mittelstand/0,2828,315897,00.html (as of 08
Feb 2006).
Richter, J.-P., Haller, H., and Schrey, P.: Serviceorientierte Architektur. In: Informatik Spektrum
(28:5) 2006, pp. 413-416.
Richter, R.: Institutionen ökonomisch analysiert. J.C.B. Mohr, Tübingen, 1994.
Richter, R., and Furubotn, E.: Neue Institutionenökonomik: Eine Einführung und kritische Würdi-
gung. JCB Mohr, Tübingen, 1996.
Riedl, R.: Begriffliche Grundlagen des Business Process Outsourcing. In: Information Management
& Consulting (18:3) 2003, pp. 6-10.
Riera, A., Davies, R., Wurzel, G., and Schwarz, J.E.: Banking à la Nike and Dell: achieving scale
without acquisition premiums. Boston Consulting Group, 2003. http://www.bcg.com/ publica-
tions/publication_view.jsp?pubID=878 (as of 29 Oct 2004).
Ring, P., and Van de Ven, A.: Developmental processes of cooperative interorganizational relation-
ships. In: Academy of Management Review (19:1) 1994, pp. 90-118.
446 References

Robey, D.: Research commentary: diversity in information systems research: threat, promise, and
responsibility. In: Information Systems Research (7:4) 1996, pp. 400-408.
Rogerson, W.: The structure of wage contracts in repeated agency models. Institute of Mathematical
Studies in the Social Sciences, Stanford University, Stanford (CA), 1982.
Rösemeier, J.: Kreditfabriken in der Sparkassenorganisation. Hamburger Sparkasse, Aareal Hypothe-
kenmanagement Informationsforum Kreditservicing, Frankfurt am Main, 16 Nov 2005.
Ross, J.W., Beath, C.M., and Goodhue, D.L.: Develop long-term competitiveness through IT assets.
In: Sloan Management Review (38:1) 1996, pp. 31-42.
Ross, S.: The economic theory of agency: the principal's problem. In: American Economic Review
(68:2) 1973, pp. 134-139.
Rothlauf, F.: Representations for genetic and evolutionary algorithms. Springer, Heidelberg et al.,
2006.
Rothlauf, F., and Goldberg, D.E.: Redundant representations in evolutionary computation. In: Evolu-
tionary Computation (11:4) 2003, pp. 381-415.
Rouse, A.C.: Revisiting IT outsourcing risks: Analysis of a survey of Australia's Top 1000 organiza-
tions. 14th Australasian Conference on Information Systems (ACIS), Perth, Australia, 2003.
Rouse, A.C., and Corbitt, B.: IT-supported business process outsourcing (BPO): the good, the bad,
and the ugly. 8th Pacific-Asia Conference on Information Systems (PACIS), Shanghai, China,
2004.
Roy, V., and Aubert, B.A.: A resource-based analysis of IT sourcing. In: ACM SIGMIS Database
(33:2) 2002, pp. 29-40.
Rudner, R.S.: Philosophy of social science. Prentice-Hall, Englewood Cliffs, 1966.
Rumelt, R.P.: Towards a strategic theory of the firm. Conference on Non-Traditional Approaches to
Policy Research, Los Angeles, 1981.
Rumelt, R.P.: Inertia and transformation. In: Resource-based and evolutionary theories of the firm:
towards a synthesis, C.A. Montgomery (ed.), Kluwer, Boston (MA), 1995, pp. 101-133.
Rusch, G.K.: Abschied vom Do-it-yourself-Prinzip. In: Information Management & Consulting
(18:3) 2003, pp. 12-16.
Salmony, M.: Veränderungen der Wirtschaftsstrukturen durch E-Commerce. In: Electronic Business
in Europa – Internationales, Europäisches und Deutsches Online-Recht, T. Bettinger, W.
Büchner, J. Drexl and M. Lehmann (eds.), Manz, Munich, 2002.
Salton, G.E., and McGill, M.: Introduction to modern information retrieval. McGraw-Hill, New York
(NY), 1983.
Salton, G.E.: The Smart Retrieval System – Experiments in automatic document processing. Pren-
tice-Hall, Inc., Englewood Cliffs (NJ), USA, 1971.
Samuelson, P.: Economics. (10th ed.) McGraw-Hill Kogakush, New York (NY), 1976.
Sanchez, R., Heene, A., and Thomas, H.: Introduction: towards the theory and practice of compe-
tence-based competition. Pergamon Press, Oxford, UK, 1996.
Sappington, D.: Incentives in principal-agent relationships. In: Journal of Economic Perspectives
(5:2) 1991, pp. 29-40.
Sargent, R.G.: Verification and validation of simulation models. Winter Simulation Conference,
Washington D.C., 1998.
Sarkar, S., and Gosh, D.: Contractor accreditation: a probabilistic model. In: Decision Sciences (28:2)
1997, pp. 235-259.
Saunders, C.S., Gebelt, M., and Hu, Q.: Achieving success in information systems outsourcing. In:
California Management Review (39:2) 1997, pp. 63-79.
Sauter, W.: Grundlagen des Bankgeschäfts. Bankakademie-Verlag, Frankfurt am Main, 2002.
Schaaf, J.: Königsweg durchs Minenfeld? IT-Outsourcing. In: Die Bank (44:8) 2004, pp. 70-73.
Schäper, C.: Entstehung und Erfolg zwischenbetrieblicher Kooperation: Möglichkeiten öffentlicher
Förderung. Gabler, Wiesbaden, Germany, 1996.
References 447

Schelling, T.: Choices and consequences: perspectives of an errant economist. Harvard University
Press, Cambridge (MA) 1961.
Schelling, T.: Micromotives and macrobehavior. Norton, New York (NY), 1978.
Schmeidler, D.: The Nucleolus of a characteristic function. In: SIAM Journal for Applied Mathemat-
ics (17:6) 1969, pp. 1163-1170
Schmiedel, H., Malkamäki, M., and Tarkka, J.: Economies of scale and technological development in
securities depository and settlement systems. Bank of Finland, Helsinki, 2002.
Schneider Traylor, P.: Outsourcing. In: CFO IT (7:4), 2003, pp. 24-25.
Scholl, R.: BPO Users in large enterprises become more demanding. Gartner Group, Stamford (CT),
USA, 2003.
Schott, E.: Markt und Geschäftsbeziehung beim Outsourcing: Eine marketingorientierte Analyse für
die Informationsverarbeitung. Deutscher Universitäts-Verlag, Wiesbaden, 1997.
Schrauth, K.-D.: Die Erwartungen einer Landesbank an die TxB. In: Management von Transaktions-
banken, H.-J. Lamberti, A. Marlière and A. Pöhler (eds.), Springer, Heidelberg et al., 2004, pp.
59-70.
Schrey, J.: Vertragsrechtliche Aspekte beim IT-Outsourcing. In: IT-Outsourcing in der Praxis –
Strategien, Projektmanagement, Wirtschaftlichkeit, T. Gründer (ed.), Erich Schmidt Verlag,
Berlin, 2004, pp. 345-347.
Schulte-Noelle, H.: Renaissance der Allfinanz-Vision. In: Zeitschrift für Organisation (67:6) 1998,
pp. 324-327.
Schulte, M., and Horsch, A.: Wertorientierte Banksteuerung II: Risikomanagement. Bankakademie-
Verlag, Frankfurt am Main, 2002.
Schwefel, H.-P.: Numerical optimization of computer models. John Wiley, Chichester, 1981.
Selten, R.: Re-examination of the perfectness concept for equilibrium points in extensive games. In:
International Journal of Game Theory (4) 1975, pp. 25-55.
Sethi, V., and King, W.: Development of measures to access the extent to which an information
technology application provides competitive advantage. In: Management Science (40:12) 1994,
pp. 1601-1627.
Shapley, L.: A value for n-person games. In: Contributions to the theory of games, H.W. Kuhn and
A.W. Tucker (eds.), Princeton University Press, Princeton (NJ), 1953, pp. 307-317.
Sharma, R., and Yetton, P.: Interorganizational cooperation to develop information systems. 16th
International Conference on Information Systems (ICIS), Cleveland (OH), 1996, pp. 122-132.
Shavell, S.: Risk sharing and incentives in the principal and agent relationship. In: Bell Journal of
Economics (10:1) 1979, pp. 55-73.
Silver, E.A.: An overview of heuristic solution methods. In: Journal of the Operational Research
Society (55) 2004, pp. 936-956.
Shimpi, P.A.: Integrating corporate risk management. Texere, New York (NY), 2001.
Simon, H.A.: A behavioral model of rational choice. In: Quarterly Journal of Economics (69:1) 1955,
pp. 99-118.
Simon, H.A.: The new science of management decision. Harper, New York (NY), 1960.
Simon, H.A.: Administrative behavior. (3 ed.) Free Press, New York (NY), 1976.
Simon, H.A.: Preismanagement. (2 ed.) Gabler, Wiesbaden, 1992.
Simon, S.: Übernahme von Personal und Einrichtungen. In: IT-Outsourcing in der Praxis – Strate-
gien, Projektmanagement, Wirtschaftlichkeit, T. Gründer (ed.), Erich Schmidt Verlag, Berlin,
2004, pp. 358-370.
Sjurts, I., and Stieglitz, N.: Outsourcing aus kostenrechnerischer, transaktionskosten- und strategie-
orientierter Sicht. In: Sourcing in der Bankwirtschaft, W. Achenbach, J. Moormann and H.
Schober (eds.), Bankakademie-Verlag, Frankfurt am Main, 2004.
Slater, G., and Spencer, D.A.: The uncertain foundations of transaction cost economics. In: Journal of
Economic Issues (34:1) 2000, pp. 61-87.
448 References

Slaughter, S., and Ang, S.: Employment outsourcing in information systems. In: Communications of
the ACM (39:7) 1996, pp. 47-54.
Smith, M.A., and Kumar, R.L.: A theory of application service provider (ASP) use from a client
perspective. In: Information & Management (41:8) 2004, pp. 977-1002.
Smith, M., Mitra, S., and Narasimhan, S.: Information systems outsourcing: a study of pre-event firm
characteristics. In: Journal of Management Information Systems (15:2) 1998, pp. 61-93.
Snow, C.C., Miles, R.E., and Coleman, H.J.: Managing the 21st century network organizations, in:
Organizational Dynamics (20:3) 1992, pp. 5-20.
Snowdon, J.: BPO adoption patterns and buying intentions: IDC's 2003 BPO survey. IDC, Framing-
ham (MA), USA, 2004.
Söbbing, T.: Handbuch IT-Outsourcing: Rechtliche, strategische und steuerliche Fragen. mitp-
Verlag, Bonn, 2002.
Spence, M.A.: Job market signaling. In: Quarterly Journal of Economics (87:3) 1973, pp. 355-374.
Spence, M.A., and Zeckhauser, R.: Insurance, information and individual action. In: American Eco-
nomic Review (61:2) 1971, pp. 380-387.
Spencer, B.J.: International outsourcing and incomplete contracts. In: Canadian Journal of Economics
(38:4) 2005, pp. 1107-1135.
Sperber, B., and Günther, A.: Joint Ventures im Transaction Banking. In: Management von Transak-
tionsbanken, H.-J. Lamberti, A. Marlière and A. Pöhler (eds.), Springer, Heidelberg et al.,
2004, pp. 179-201.
Spiegel, T.: Prozessanalyse in Dienstleistungsunternehmen. Deutscher Universitätsverlag, Wies-
baden, Germany, 2002.
Spulber, D.F.: Monopoly pricing of capacity usage under asymmetric information. In: Journal of
Industrial Organization (41:3) 1993, pp. 241-257.
Sridhar, S.S., and Balachandran, B.V.: Incomplete information, task assignment, and managerial
control systems. In: Management Science (43:6) 1997, pp. 764-778.
Steffens, U.: Chancen und Risiken der deutschen Banking&Finance-Branche – Eine strategische
Analyse. In: Strategisches Management in Banken, U. Steffens and W. Achenbach (eds.),
Bankakademie-Verlag, Frankfurt am Main, 2002.
Stevensen, H.H.: Defining corporate strengths and weaknesses. In: Sloan Management Review (17:3)
1976, pp. 51-68.
Stewart, K., Straub, D.W., and Weill, P.: Strategic control of IT resources: a test of resource-based
theory in the context of selective IT outsourcing. Center for Information Systems at the Sloan
School of Management, Cambridge (MA), 2002.
Stiglitz, J.E., and Weiss, A.: Credit rationing in markets with imperfect information. In: American
Economic Review (71:3) 1981, pp. 393-410.
Strassmann, P.A.: The real cost of OA. In: Management of Information Systems, P. Gray, W.R.
King, E.R. McLean and R.J. Watson (eds.), The Dryden Press, Hillsdale (IL), USA, 1989, pp.
95-105.
Stump, R.L., and Sriram, V.: Employing information technology in purchasing: buyer-supplier
relationships and size of the supplier base. In: Industrial Marketing Management (26:2) 1997,
pp. 127-136.
Suyter, A.: Aufsichtliche Normen im Bankgeschäft und ihre Erweiterung um Mindestanforderungen
an das Kreditgeschäft (MaK). In: WM Wertpapiermitteilungen (20), 2002, pp. 991-998.
Sydow, J.: Strategische Netzwerke. Gabler, Wiesbaden, Germany, 1992.
Szivek, E.: Sourcing aus aufsichtsrechtlicher Sicht. In: Sourcing in der Bankwirtschaft, W. Achen-
bach, J. Moormann and H. Schober (eds.), Bankakademie-Verlag, Frankfurt am Main, 2004,
pp. 45-66.
Teece, D.: The diffusion of an administrative innovation. In: Management Science (26:5) 1980a, pp.
464-470.
References 449

Teece, D.: Economies of scope and the scope of the enterprise. In: Journal of Economic Behavior and
Organization (1) 1980b, pp. 223-247.
Teng, J.T.C., Cheon, M.J., and Grover, V.: Decisions to outsource information systems functions:
testing a strategy-theoretic discrepancy model. In: Decision Sciences (26:1) 1995, pp. 75-103.
Tesfatsion, L.: Agent-based computational economics: Growing economies from the bottom up. In:
Artificial Life (8) 2002a, pp. 55-82.
Tesfatsion, L.: Agent-based computational economics, 2002b. http://www.econ.iastate.edu/tesfatsi/
ace.htm (as of 30 Oct 2005).
Tesfatsion, L.: Agent-based computational economics: a constructive approach to economic theory.
In: Handbook of Computational Economics, L. Tesfatsion and K.L. Judd (eds.), North-
Holland, Amsterdam, 2006.
Thatcher, M.E., and Oliver, J.R.: The impact of technology investments on a firm's production effi-
ciency, product quality, and productivity. In: Journal of Management Information Systems
(18:2) 2001, pp. 17-45.
Theewen: Problemkredite und die "Mindestanforderungen an das Kreditgeschäft der Kreditinstitute"
– Workout, Outsourcing oder Bad Bank. In: WM Wertpapiermitteilungen (3), 2004, pp. 105-
114.
Thierens, D., and Goldberg, D.: Convergence models of genetic algorithm selection schemes. In:
Parallel problem solving from nature, Y. Davidor, H.-P. Schwefel and R. Manner (eds.), Sprin-
ger, Berlin, 1994, pp. 119-129.
Thompson, J.D.: Organizations in action. McGraw-Hill, New York (NY), 1967.
Thorelli, H.B.: Between markets and hierarchies. In: Strategic Management Journal (7:1) 1986, pp.
37-51.
Tornbohm, C.: Market trends: business process outsourcing, Western Europe, 2003-2008. Gartner
Group, Stamford (CT), USA, 2005.
Tortosa-Ausina, E.: Bank cost efficiency and output specification. In: Journal of Productivity Analy-
sis (18:3) 2002, pp. 199-222.
TPI: The TPI index – An informed view of the state of the global commercial outsourcing market,
Technology Partners International, 10 Jan 2007, http://www.tpi.net/pdf/index/
4Q06%20TPI%20Index%20Presentation.pdf (as of 20 Feb 2007).
Tsai, W.: Social structure of 'coopetition'. Within a multiunit organization: coordination, competition,
and intraorganizational knowledge sharing. In: Organization Science (13:2) 2002, pp. 179-190.
Tsang, E.W.K.: Transaction cost and resource-based explanation of joint ventures. In: Organization
Studies (21:1) 2000, pp. 215-242.
Tucker, I.B., and Wilder, R.P.: Trends in vertical integration in the U.S. manufacturing sector. In:
Journal of Industrial Economics (26:1) 1977, pp. 81-94.
U.S.A.: The Sarbanes-Oxley Act. Senate and House of Representatives of the U.S.A., 2002.
http://www.sec.gov/about/laws/soa2002.pdf. (as of 11 July 2006).
Ulrich, D., and Barney, J.B.: Perspectives in organizations: resource dependence, efficiency, and
population. In: Academy of Management Review (9:3) 1984, pp. 471-481.
Ungan, M.C.: Standardization through process documentation. In: Business Process Management
Journal (12:2) 2006, pp. 135-148.
Van de Ven, A.H., Delbecg, A.H., and Koenig, R.: Determinants of coordination modes within
organizations. In: American Sociological Review (41:2) 1976, pp. 322-338.
Van der Meer-Kooistra, J., and Vosselman, E.G.: Management control of interfirm transactional
relationships: the case of industrial renovation and maintenance. In: Accounting, Organizations
and Society (25:1) 2000, pp. 51-77.
Van der Vegt, G., Emans, B., and Van de Vliert, E.: Motivating effects of task and outcome interde-
pendence in work teams. In: Group and Organization Management (23:2) 1998, pp. 124-143.
450 References

Van Gils, M.R.: Interorganizational relations and networks. In: Handbook of Work and Organiza-
tional Psychology, P.J.D. Drenth, H. Thierry, P.J. Willems and C.J. de Wolff (eds.), Routledge,
London, 1984, pp. 1073-1100.
Van Mieghem, J.A.: Coordinating investment, production, and subcontracting. In: Management
Science (45:7) 1999, pp. 954-971.
Vander Vennet, R.: Causes and consequences of EU bank takeovers. In: The changing European
financial landscape, S. Eijffinger, K. Koedijk, M. Pagano and R. Portes (eds.), CEPR, Brussels,
Belgium, 1999, pp. 45-61.
Vander Vennet, R.: Cost and profit efficiency of financial conglomerates and universal banks in
Europe. In: Journal of Money, Credit, and Banking (34:1) 2002, pp. 254-282.
Venkatraman, N.: IT-induced business reconfiguration. In: The corporation of the 1990s – Informa-
tion technology and organizational transformation, M.S. Scott Morton (ed.), Oxford University
Press, New York (NY), 1991, pp. 122-158.
Venkatraman, N.: IT-enabled business transformation – From automation to business scope redefini-
tion. In: Sloan Management Review (35:2) 1994, pp. 73-87.
Venkatraman, N., and Camillus, J.C.: Exploring the concept of "fit" in strategic management. In:
Academy of Management Review (9:3), 1984, pp. 513-525
Venkatraman, N., and Loh, L.: Strategic issues in information technology sourcing: patterns, perspec-
tives, and prescriptions. Center for Information Systems Research, Sloan School of Manage-
ment, Cambridge (MA), 1993.
Wade, M., and Hulland, J.: Review: The resource-based view and information systems research:
review, extension, and suggestions for future research. In: MIS Quarterly (28:1) 2004, pp. 107-
142.
Wagner, H.-T., Beimborn, D., Franke, J., and Weitzel, T.: IT business alignment and IT usage in
operational processes: a retail banking case. 39th Hawaii International Conference on System
Sciences (HICSS-39), Koloa, Kauai (HI), USA, 2006.
Wahrenburg, M., König, W., Beimborn, D., Franke, J., Gellrich, T., Holzhäuser, M., Schwarze, F.,
and Weitzel, T.: Kreditprozess-Management. Books on Demand, Norderstedt, Germany, 2005.
Walker, G., and Weber, D.: A transaction cost approach to make-or-buy decisions. In: Administrative
Science Quarterly (29:3) 1984, pp. 373-391.
Walsham, G.: Interpreting information systems in organizations, Wiley, Chichester, UK, 1993.
Walter, H.: Optimierung von Wertschöpfungsketten bei Privatkundenbanken. In: Management der
Wertschöpfungsketten in Banken, L.P. Marighetti, R. Jasny, A. Herrmann and F. Huber (eds.),
Gabler, Wiesbaden, 2001, pp. 39-49.
Wang, E.T.G.: Transaction attributes and software outsourcing success: an empirical investigation of
transaction cost theory. In: Information Systems Journal (12:2) 2002, pp. 153-181.
Wang, E.T.G., Barron, T., and Seidmann, A.: Contracting structures for custom software develop-
ment: the impacts of efficient investments and imperfect knowledge on internal development
and outsourcing. In: Management Science (43:12) 1997, pp. 1726-1744.
Weber, M.: Bankenmarkt Deutschland: II. Perspektiven. In: Die Bank (42:7) 2002, pp. 456-459.
Webster, F.E.: The changing role of marketing in the corporation. In: Journal of Marketing (56:4)
1992, pp. 1-17.
Weill, P., and Ross, J.: A matrixed approach to designing IT governance. In: Sloan Management
Review (46:2) 2005, pp. 26-34.
Weisser, N.: Das Phänomen der falschen Zahl. In: Die Bank (44:12) 2004, pp. 48-51.
Weitzel, T. A network ROI. MISQ Academic Workshop on ICT Standardization at the 24th Interna-
tional Conference on Information Systems (ICIS), Seattle (WA), USA, 2003.
Weitzel, T.: The economics of standards in information networks. Springer Physica, Heidelberg et
al., 2004.
References 451

Weitzel, T., Beimborn, D., and Franke, J.: Outsourcing the financial chain: an empirical analysis of
sourcing and partnering potentials. 10th Americas Conference on Information Systems (AM-
CIS), Best Paper Award in SIGISO (outsourcing track), New York (NY), 2004.
Weitzel, T., Beimborn, D., and König, W.: Coordination in networks: an economic equilibrium
analysis. In: Information Systems and e-Business Management (1:2) 2003, pp. 189-211.
Weitzel, T., Beimborn, D., and König, W.: A unified model of standard diffusion: the impact of
standardization cost, network effects, and network topology. In: MIS Quarterly (30:special is-
sue) 2006, pp. 489-514.
Weitzel, T., Westarp, F.v., and Wendt, O.: Reconsidering network effect theory. 8th European Con-
ference on Information Systems (ECIS 2000), Vienna, Austria, 2000.
Wendt, O.: Tourenplanung durch Einsatz naturanaloger Verfahren. Deutscher Universitäts-Verlag,
Wiesbaden, Germany, 1995.
Wernerfelt, B.: A resource-based view of the firm. In: Strategic Management Journal (5:2) 1984, pp.
171-180.
Wernthaler, G.: Outsourcing des Zahlungsverkehrs am Beispiel der Naspa. In: Sourcing in der
Bankwirtschaft, W. Achenbach, J. Moormann and H. Schober (eds.), Bankakademie-Verlag,
Frankfurt am Main, 2004, pp. 139-152.
Westarp, F.v.: Modeling software markets: empirical analysis, network simulations, and marketing
implications. Springer, Heidelberg et al., 2003.
Whang, S.: Contracting for software development. In: Management Science (38:3) 1992, pp. 307-
324.
Wibbelsman, D., and Maiero, T.: Cosourcing. Outsourcing, Cosourcing and Insourcing Conference,
University of California in Berkeley (CA), USA, 1994.
Wieland, J.: Kooperationsökonomie: Die Ökonomie der Diversität, Abhängigkeit und Atmosphäre.
In: Konkurrenz und Kooperation, S.A. Jansen and S. Schleissing (eds.), Metropolis, Marburg,
Germany, 2000, pp. 103-128.
Wiese, H.: Netzeffekte und Kompatibilität – Eine theoretische und simulationsgeleitete Analyse zur
Absatzpolitik für Netzeffekt-Güter. Poeschel, Stuttgart, 1990.
Wilder, R.P., and Tucker, I.B.: Trends in vertical integration: Reply. In: Journal of Industrial Eco-
nomics (32:3) 1984, pp. 391-392.
Willcocks, L.P., Fitzgerald, G., and Lacity, M.C.: To outsource IT or not? Recent research on eco-
nomics and evaluation practice. In: European Journal of Information Systems (7:1) 1996, pp.
29-45.
Willcocks, L.P., Hindle, J., Feeny, D.F., and Lacity, M.C.: IT and business process outsourcing: the
knowledge potential. In: Information Systems Management (21:4) 2004, pp. 7-15.
Willcocks, L.P., and Kern, T.: IT outsourcing as strategic partnering: The case of the UK inland
revenue. In: European Journal of Information Systems (7:1) 1998, pp. 29-45.
Willcocks, L.P., and Lacity, M.C.: Strategic sourcing of information systems. Wiley, Chichester,
1998.
Williams, A.: Application management outsourcing versus insourcing. 2001. http://artw-
consulting.com/ArtW%20Consulting/BlueStar_Whitepaper.htm (as of 17 Aug 2005).
Williamson, O.E.: The vertical integration of production: market failure considerations. In: American
Economic Review (61:2) 1971, pp. 112-123.
Williamson, O.E.: Markets and hierarchies: analysis and antitrust implications. A study in the eco-
nomics of internal organization. Free Press, London, New York, 1975.
Williamson, O.E.: Transaction cost economics: the governance of contractual relations. In: Journal of
Law and Economics (22:2) 1979, pp. 233-261.
Williamson, O.E.: The modern corporation: origins, evolution, attributes. In: Journal of Economic
Literature (19:4) 1981, pp. 1537-1568.
Williamson, O.E.: The economic institutions of capitalism. Free Press, New York (NY), 1985.
452 References

Williamson, O.E.: Corporate finance and corporate governance. In: Journal of Finance (43:3) 1988,
pp. 567-591.
Williamson, O.E.: Transaction cost economics. In: Handbook of industrial economics, R. Schmalen-
see, Willig, R.D. (ed.), North-Holland, Amsterdam, 1989, pp. 135-182.
Williamson, O.E.: Comparative economic organization: the analysis of discrete structural alterna-
tives. In: Administrative Science Quarterly (36:2) 1991, pp. 269-296.
Williamson, O.E., Wachter, M.L., and Harris, J.E.: Understanding the Employment Relation: The
Analysis of idiosyncratic exchange. In: Bell Journal of Economics (6:1) 1975, pp. 205-280.
Wöhe, G.: Einführung in die Allgemeine Betriebswirtschaftslehre. (17 ed.) Franz Vahlen, Munich,
1990.
Wold, H.: Partial Least Squares. In: Encyclopedia of statistical sciences, S. Kotz and N.L. Johnson
(eds.), John Wiley, New York (NY), 1985, pp. 581-591.
Wood, R.E.: Task complexity: definition of the construct. In: Organizational Behavior and Human
Decision Processes (37:1) 1986, pp. 60-82.
Wüllenweber, K., Beimborn, D., and Weitzel, T.: The impact of process standardization on business
process outsourcing success. Forthcoming in: Information Systems Frontiers, 2008.
Wullenkord, A., Kiefer, A., and Sure, M.: Business Process Outsourcing: Ein Leitfaden zur Kosten-
senkung und Effizienzsteigerung im Rechnungs- und Personalwesen. Franz Vahlen, Munich,
2005.
Wybo, D.M., and Goodhue, L.D.: Using interdependence as a predictor of data standards: theoretical
and measurement issues. In: Information and Management (29:6) 1995, pp. 317-329.
Yang, M.C., Wood III, W.H., and Cutkosky, M.R.: Design information retrieval: a thesauri-based
approach for reuse of informal design information. In: Engineering with Computers (21:2)
2005, pp. 177-192.
Young, H., Okada, N., and Hashimoto, T.: Cost allocation in water resources development. A case
study of Sweden. IIASA Research report RR-80-032, Laxenburg, 1980.
Zadeh, L.T.: Fuzzy sets. In: Journal of Information and Control (8:3) 1965, pp. 338-353.
Zaheer, A., and Venkatraman, N.: Determinants of electronic integration in the insurance industry: an
empirical test. In: Management Science (40:5) 1994, pp. 549-566.
Zajac, E.J., and Olson, C.: From transaction cost to transaction value analysis: implications for the
study of interorganizational strategies. In: Journal of Management Studies (30:1) 1993, pp.
131-145.
Zanthier, U.v., and Gärtner, C.: Mindestanforderungen an das Kreditgeschäft der Kreditinstitute. Cap
Gemini Consulting, 2003. http://www.de.capgemini.com/servlet/PB/show/1006063/ MaK.pdf
(as of 16 Aug 2005).
Zeithaml, C.P., and Zeithaml, V.A.: Environmental management: revising the marketing perspective.
In: Journal of Marketing (48:2) 1984, pp. 46-53.
Zhang, G., and Liu, Z.: Managing coordination costs of interorganizational relationships: an analysis
of determinants. 5th International Conference on Electronic Business (ICEB), Hong Kong,
2005, pp. 54-58.
Appendix 453

Appendix

A1 – Performance Tests of the Genetic Algorithm

For optimally configuring the genetic algorithm that accompanied the simulation
studies in chapter 5, six test problems of the cooperative sourcing problem with
different solution difficulty have been developed based on the parameterization
given in section 5.3.3. The problem size was reduced from 100 to 50 banks. The
test problems differ in the level of transaction costs. The higher the transaction
cost factors are, the less favorable is cooperative sourcing. Thus, based on the
parameterization from section 5.3.3, all transaction cost factors (cN, cIF, cAG, cAD)
were successively increased by 50% for generating a new scenario. The follow-
ing scenarios occurred:
Scenario A Scenario B Scenario C Scenario D Scenario E Scenario F
TC factors TC factors TC factors TC factors TC factors TC factors
* 1.5 * 2.25 * 3.375 * 5.063 * 7.594 * 11.391
Table 71: Test scenarios (increasing “difficulty” from left to right)
In a first step, nine different GA configurations were tested starting with the
configuration from (Beimborn et al. 2007b, Fladung 2006) (cf. Table 44, p. 281,
and first row in Table 72 below). The objective was to strongly reduce comput-
ing time while minimally reducing solution quality (cf. section 5.2.3).
The different configurations are listed in Table 72. While the first configura-
tion represents the configuration used in (Beimborn et al. 2007b), the second
configuration was designed to create benchmarks. A double runtime was ex-
pected to generate superior solutions (as also shown by Fladung (2006)). Con-
figurations 3 to 10 represent the alternative configurations to be tested against the
original configuration (1) and the benchmark (2). Especially, the variation of
crossover probability pCO and mutation probability pM was tested.
Confi- population
pCO pM TGA Annotations
guration size
Original setting from (Beimborn et al. 2007b)
1 100 .55 .002 100
except preTermination = 50
2 50 .55 .002 200 Increased runtime for benchmarking
3 50 .55 .002 100 Testing effect of reduced population size
4 100 .65 .002 100 Testing effect of increased pCO
454 Appendix

Testing effect of reduced population size and in-

5 50 .65 .002 100
creased pCO
Testing effect of increased initial pM compared with
6 50 .55 .005 100
configuration 3
Testing effect of increased pCO and initial pM com-
7 50 .65 .005 100
pared with configuration 3
Testing effect of reduced pCO compared with con-
8 50 .45 .002 100
figuration 3
Testing effect of reduced initial pM compared with
9 50 .55 .001 100
configuration 3
Table 72: Test configurations of the GA222
Primary set screws to reduce computation time – which is a critical objective
for simulation studies – are population size and runtime TGA. Thus, the maximum
runtime to be used should not exceed 100 periods. Moreover, the number of
individuals per generation was halved to 50.
Each of the test scenarios was solved 50 times with each GA configuration
in order to receive a statistically sound basis. The best overall solution was set as
the benchmark. As measure for effectiveness, the ratio between the average solu-
tion quality of each GA configuration and the benchmark (best overall found
solution) was used. The following table shows the results with the best values in
each column marked in bold.
Configuration Scenario A Scenario B Scenario C Scenario D Scenario E Scenario F
1 97.9% 97.8% 96.1% 96.1% 95.4% 90.2%
2 98.8% 99.0% 97.4% 97.1% 96.6% 93.5%
3 98.3% 98.2% 96.5% 96.3% 95.1% 89.5%
4 97.1% 96.9% 95.0% 93.9% 91.9% 88.4%
5 97.7% 97.4% 95.5% 94.6% 92.3% 88.8%
6 97.3% 97.2% 95.1% 94.2% 93.1% 86.4%
7 96.9% 96.6% 94.5% 93.2% 90.3% 86.5%
8 98.4% 98.3% 96.5% 96.5% 95.8% 87.4%
9 98.6% 98.7% 97.1% 96.7% 95.4% 89.3%
Table 73: Solution quality of tested GA configurations
The GA configurations differed only marginally in solution quality. Overall,
configuration 9 showed the best average performance (best values for scenarios

222
Explanation of parameters (cf. section 5.2): pCO = crossover probability, pM = mutation probabil-
ity. TGA = last generation.
Appendix 455

A, B, C, and D) (after configuration 2). For difficult problems E and F, configu-

ration 8 (which has a higher mutation probability) and 3 performed slightly bet-
ter.
For simulation studies, it is further important that an optimization routine
performs well in each run and not only on average. For all of the configurations
and tested scenarios it was measured how close the runs were to the benchmark.
Table 74 presents the results.

Con- Scenario A Scenario B Scenario C Scenario D Scenario E Scenario F

<10%

<10%

<10%

<10%

<10%
<20%
fig.
<1%
<2%

<5%
<1%
<2%
<5%

<1%
<2%
<5%

<1%
<2%
<5%

<1%
<2%
<5%

<1%
<2%
<5%
1 2 22 50 2 19 50 50 0 0 50 50 0 0 46 50 0 1 29 50 0 0 2 24 49
2 19 45 50 24 48 50 50 2 11 50 50 2 7 50 50 6 10 42 50 0 2 20 36 50
3 5 36 50 3 30 50 50 0 0 50 50 0 2 49 50 0 1 27 50 5 6 4 22 50
4 0 7 50 0 0 50 50 0 0 50 50 0 0 7 50 0 0 1 15 0 0 0 23 48
5 1 19 50 0 50 50 50 0 2 50 50 0 1 16 50 1 1 1 18 0 0 4 11 48
6 0 19 50 1 5 50 50 1 1 50 50 0 0 10 50 0 1 4 29 0 0 2 10 48
7 0 4 50 0 1 50 50 0 1 50 50 0 0 6 50 0 0 0 8 0 0 0 6 49
8 8 36 50 9 32 50 50 0 0 46 50 0 3 46 50 2 3 34 50 0 0 6 16 42
9 10 43 50 16 43 50 50 0 7 50 50 1 5 49 50 0 1 31 50 0 0 2 24 50
Table 74: Number of computed solutions (from 50) which did not deviate
from the benchmark by more than 1%, 2%, 5%, and 10% (and 20%
for scenario F)
For the less difficult problems A-C, all configurations reached a solution
which was less than 5% away from the benchmark in all cases. For D and E,
some of them could satisfy at least the 10% threshold in all cases. Out of all
configurations with a runtime of no more than 100 generations (all except con-
figuration 2), configuration 8 and 9 usually showed the best values.
As a further test, the relative spans between the best and worst solution
found for each GA configuration were compared. This bandwidth as well as the
variation coefficient (i.e. normalizing standard deviation to the mean) showed
very small variations for all of the configurations (Table 75). The worst solution
found often did not fall below the best solution (found by the same GA configu-
ration) by more than 5%. Especially configurations 1, 2, 8, and 9 showed good
values.
456 Appendix

Configuration Scenario A Scenario B Scenario C Scenario D Scenario E Scenario F

1 .005, 2% .005, 2% .005, 2% .008, 3% .012, 5% .035, 13.1%
2 .006, 2% .006, 3% .009, 4% .010, 5% .016, 6% .044, 13.6%
3 .005, 4% .005, 5% .007, 5% .008, 7% .015, 13% .043, 15.9%
4 .007, 3% .008, 3% .008, 5% .014, 7% .030, 17% .036, 14.0%
5 .006, 3% .008, 4% .009, 4% .016, 7% .034, 17% .043, 17.3%
6 .009, 4% .011, 5% .014, 7% .017, 7% .044, 21% .045, 16.1%
7 .009, 4% .011. 5% .012, 6% .019, 8% .032, 12% .039, 14.1%
8 .007, 3% .007, 3% .008, 4% .011, 5% .015, 7% .066, 21.7%
9 .005, 2% .005, 2% .008, 3% .010, 5% .012, 6% .034, 12.4%
Table 75: Variation coefficient (first value) and relative span of found
solutions (second value) 223
Based on these results, configuration 9 was selected as the adequate configu-
ration for a simulation-accompanying genetic algorithm in this work.
In a second step, it was analyzed whether computation time can further be
reduced. We compared the solution quality increase over time for this particular
configuration.
GA configuration 9

1
current optimal solution / final optimal solution

0.98

0.96

0.94
Scenario A
Scenario B
0.92
Scenario C
Scenario D
0.9 Scenario E
Scenario F

0.88

0.86
0 10 20 30 40 50 60 70 80 90 100
generations

Figure 181: Average solution quality over time for GA configuration 9 and
different problem scenarios

223
Relative span = (maximum value – minimum value) / maximum value
Appendix 457

As Figure 181 shows, 98% of the objective value of the final solution (100th
generation) are already reached in the 50th period for most scenarios. Only for
very difficult problems (Scenario F), this is not the case. Reverse tests showed
that the other GA configurations showed no better results in the 50th generation,
either. Thus, we finally decided to use configuration 9 with reduced TGA = 50 in
the simulation studies. If transaction costs are very high compared with process
cost savings (i.e. very difficult optimization problems), the runtime is extended
to 100 periods.

A2 – Parameterization of the Simulation Studies

The following tables summarize the parameterization of the simulation studies
which was derived in section 5.3.3.
Demographics
Number of business
|I| = 100 Number of actors |K| = 5
functions
Availability of business
TCoSo =5 Contract duration aik 1 i, k
function k in bank i
Business neighborhood
sizei = based on number bnijcust 1.0  i, j (customer-based and
of loans in stock distri- Size of bank i
bution of S2 (Figure 38) bn geo
ij max sizei , size j  geographical) degree
between i and j
Risk-adjusted dis-
rad = .05 SCi = 3.0  i Strategic constraint
count rate
VAT =.0 or .19 Value-added tax
Table 76: Parameterization of simulation studies – demographics
Transaction costs
Transaction Sales/ Assessment/ Processing Risk
Workout
cost factor preparation decision /servicing monitoring
N N
c k c 1
N
6,740 c 2N 3,506 c 3N 40,773 c N
4 17 ,034 c 5 3,095
AG AG AG
c k c1
AG
178,996 c AG
2 173,984 c 3 98,090 c AG
4 135,142 c 5 88,782

c kAD c1AD 10 ,000 c 2AD 10 ,000 c 3AD 10,000 c 4AD 10 ,000 c 5AD 10,000
Misc.: cIF = 5,000 D = 1.25 E = -.9 J = .3
Table 77: Parameterization of simulation studies – transaction costs
458 Appendix

Business function characteristics

xi3 = xi4 ~ ND(P = sizei xi5 = ~ ND(P = xi3 * .202,
xi1 = xi2 = .15 * xi3 Process volumes
* 30,000, V= .2*P) V= xi3 * .097)
P
K ik and c ik are based on empirical data (cf. procedure in Figure 109): total
F

process costs are allocated to the different process steps and further sepa- Process costs
rated into fixed and variable costs based on process cost volumes and
process cost ratios.
(Residual) core
Oik follows empirical distributions (Figure 52) and correlations (Table 47)
competencies
Similarity Sales/ Assessm./ Processing/ Risk
Workout
degree preparation decision servicing monitoring Similarity degree ]ijk0
PP]ijk0) .189 .168 .359 .253 .291 (normally distributed)
VV(] ijk0) .179 .171 .119 .160 .143
Assessm./ Processing/ Risk
P(Tikl), V(Tikl) Workout
decision servicing monitoring
Sales/
.855, .212 .500, .280 .630, .276 .435, .283
preparation
Assessm./ Modularity degree Tikl
.615, .281 .500, .280 .400, .280
decision (normally distributed)
Processing/
.513, .281 .100, .280
servicing
Risk
.485, .286
monitoring
Table78: Parameterization of simulation studies – business function
characteristics
Simulation control and genetic algorithm
Initial size of created Repetition rate for creation
initCoalSize = 9 coalBuildingFreq = 5
coalitions of new coalitions
Maximum number of coalitions created per business function k in one
maxCoalBuilding = 4
period
Population of the Maximum number of gen-
Population size = 50 TGA = 100
genetic algorithm erations
Crossover probability Mutation probability of
pCo = .55 pM = .001
of genetic algorithm genetic algorithm
Table 79: Parameterization of simulation studies – simulation and GA control