Chapter 10

Forecasting
Chapter takeaways
After completion of this chapter the reader would be able to
1. Appreciate how a realistic forecast paves the way for an effective production plan-
ning function.
2. Understand the difference between long-term and short-term forecasting and
apply that concept for day-to-day planning.
3. Distinguish between qualitative and quantitative forecasting.
4. Appreciate how the aided judgment helps when planning the basis for the day-to-
day planning.
5. Understand the concept of product life cycle and how it affects production
planning.
6. Realize that overmeticulous forecasting also costs money, and develop and select
optimum methodology on a cost-benefit analysis basis.

10.1 Introduction
Forecasting is a management technique to estimate the sales of a product or
service in physical units for a fixed future period, under the proposed or
existing market plan. It assumes a set of economic and other forces outside
the organization for which the forecast is made. It determines how to allocate
their budgets for an upcoming period of time. It also provides an important
benchmark for firms to determine their future operations, policy planning,
and reengineering if necessary. The dynamic changes in the quantity or qual-
ity of products and/or services require a change in the organization structure.

10.2 The need for forecasting

Today, all organizations operate in an atmosphere of uncertainty. Due to the
rapid changes in technology, an organization’s involvement in the national
scenario of the economic, social, and political changes depends very signifi-
cantly on the forecasting of not only their sales but also the other forces

Production Planning and Control. DOI: https://doi.org/10.1016/B978-0-12-818364-9.00010-X

Copyright © 2019 BSP Books Pvt. Ltd. Published by Elsevier Inc. All rights reserved. 141
142 Production Planning and Control

effecting their operations. Decisions that are made by organizations today

will affect future outcomes. It is essential to obtain an estimate of the
changes as accurately as possible for companies to survive and to strive for
operational excellence. This necessitates the need for companies to develop
reliable forecasting systems.
1. A majority of industrial activities depend upon future sales because no
sales means no production.
2. Forecasting plays an important role for the development of future plans.
3. Forecasting is used for decision-making investments in plants.
4. It is used for planning marketing strategy and programming.
5. Scheduling of production activities depends on forecasting.
6. Forecasting is needed for effective manpower planning.
7. Forecasting facilitates planning for effective utilization of the plant and
machinery.
8. Forecasting is used for material planning and replenishment action
(making available the right material at the right time in the right
quantities).
9. Forecasting provides information on the relationship between the
demands for different products.
10. Forecasting is used for product design and development.
11. Forecasting helps company expansions and diversification.

10.3 Definitions of forecasting

Following are various definitions available on forecasting:
Forecasting is not a guess work. It works on the basis of a large number of
data by using statistical techniques and analyses by use of historic data to
determine the direction of future trends.
Investopedia

Forecasting is a planning tool that helps management in its attempts to cope

with the uncertainty of the future, relying mainly on data from the past and
present and analysis of trends.
Business Dictionary

Forecasting is the process of making predictions of the future based on past

and present data and analysis of trends.
Wikipedia.
 Forecasting Chapter | 10 143

Forecasting is an estimate of sales in physical units or monetary values for a

specified period and proposed marketing plan or program under the assured
set of economic and other factors outside the organization for which the fore-
cast is made.
American Management Association

Forecasting is about predicting the future as accurately as possible, given all

of the information available, including historical data and knowledge of any
future events that might impact the forecasts. It is a common statistical task in
business, where it helps to inform decisions about the scheduling of production,
transportation and personnel, and provides a guide to long-term strategic
planning.
www.otexts.org

Forecasting is a decision-making tool used by many businesses to help in bud-

geting, planning, and estimating future growth. In the simplest terms, forecast-
ing is the attempt to predict future outcomes based on past events and
management insight.
http://www.vanguardsw.com/business-forecasting

Forecasting is to calculate or predict (some future event or condition) usually

as a result of study and analysis of available pertinent data.
http://www.merriam-webster.com/dictionary/forecast

Primary forecasting techniques help organizations plan for the future. Some
are based on subjective criteria and often amount to little more than wild
guesses or wishful thinking. Others are based on measurable, historical quanti-
tative data and are given more credence by outside parties, such as analysts
and potential investors. While no forecasting tool can predict the future with
complete certainty, they remain essential in estimating an organization’s for-
ward prospects.
http://smallbusiness.chron.com/four-primary-forecasting-techniques

Forecasting can be broadly considered as a method or a technique for estimat-

ing many future aspects of a business or other operation. The goal of forecast-
ing is to come as close to possible to an accurate picture of the future.
http://www.inc.com/encyclopedia/forecasting.html
144 Production Planning and Control

10.4 Basic steps of forecasting

Following are the basic steps in the forecasting process:
1. Identify and determine the objectives, the purpose, and need for the
forecast.
2. Establish a time horizon.
3. Categorize the products.
4. Select a forecasting technique.
5. Identify and select the definite variables that affect the demand.
6. Gather and analyze data.
7. Make the forecast.
8. Make trend correction, etc., where relevant.
9. Monitor the forecast.

10.5 Characteristics of a good forecast

William J. Stevenson, author of Operations Management, lists the following
characteristics of a good forecast:
1. Accurate—Some degree of accuracy should be determined to make a
comparison with other alternative forecasts.
2. Reliable—The user should establish some degree of confidence.
3. Timely—It is necessary to set a certain amount of time in response to
make changes if necessary.
4. Easy to use and understand—Users should find the forecast
comfortable to work with.
5. Cost-effective—The cost should not be higher than the benefits obtained
from the forecast.

10.6 Short-term, medium-term, and long-term forecasts

Forecasting is an integral part of the decision-making activities of manage-
ment, as it plays an important role in many areas of a company. Modern
organizations require short-term, medium-term, and long-term forecasts,
depending on the specific application.
Short-term forecasts (up to 1 year) are needed for the scheduling of per-
sonnel, production, and transportation. As part of the scheduling process,
forecasts of demand are also required.
Medium-term forecasts (up to 5 years) are needed to determine future
resource requirements in order to purchase raw materials, hire personnel, or
buy machinery and equipment.
Long-term forecasts (over 5 years) are used in strategic planning. Such
decisions must take account of market opportunities, environmental factors,
and internal resources.
 Forecasting Chapter | 10 145

10.7 Techniques of forecasting

There can basically three categories of the forecasting methods or techniques
as illustrated in Fig. 10.1:
G Qualitative, also called as judgmental methods
G Quantitative methods
G Time series analysis methods
Each of these categories is detailed in the subsequent paragraphs.

10.8 Qualitative forecasting methods

Qualitative forecasting methods are based on the opinion and judgment of
consumers and field-wise experts. These techniques are subjective and used
when all required data is not available. Intermediate- or long-range decisions
can be made, but not short-range or immediate decisions, like shop floor

FIGURE 10.1 Categories of the forecasting methods.

146 Production Planning and Control

planning. The following are some of the qualitative forecasting methods,

which are described more in detail in paragraphs:
1. Opinion survey
2. Customer and distributor survey
3. Aided judgment
4. Judgmental bootstrapping
5. Jury of executive opinion
6. Delphi technique
7. Prediction markets
8. Marketing trials
9. Market research
10. Simulated interaction

10.9 Quantitative forecasting methods

Quantitative forecasting methods simulate the future data as a function of the
past data. They are objective and are appropriate to use when past numerical
data is available, and the trend is assumed to continue in the future to a rea-
sonable extent. Examples of quantitative forecasting methods are weighted
moving averages, simple exponential smoothing, seasonal indexes, etc.,
which can be applied to short- or intermediate-range decisions. The follow-
ing are some of the quantitative forecasting methods:
1. Discrete event simulation
2. Group method of data handling (GMDH)
3. Reference class forecasting
4. Quantitative analogies
5. Game theory
6. Rule-based forecasting
7. Data mining
8. Conjoint analysis
9. Causal models
10. Segmentation
11. Cross-sectional forecasting
Chapter 9, Correlation and probability theory, on the fundamentals of sta-
tistics will illustrate how statistics forms the basis for trendsetting and moving
annual total, etc., which are described in more detail in the next chapter on
trend analysis. Cited here are some of these quantitative forecasting methods:
1. Trend projection methods
2. Extrapolation
3. Cyclic trend projection
4. Moving average method
5. Exponential smoothing method
 Forecasting Chapter | 10 147

10.10 Detailed explanation of the forecasting techniques

The following paragraphs explain each of three categories of these listed
forecasting techniques.

10.11 Qualitative or judgmental forecasting methods

10.11.1 Opinion survey
Pre-poll and exit surveys are used when the media conducts surveys among
the public chosen at random to get their opinions about who will win and
how many votes they may get in an election. This principle can be applied
to businesses conducting similar surveys among the public about specific
marketing aspects of certain products. Opinions are collected from prospec-
tive buyers on what they buy and why they want to buy certain products.
The statistical principle of sampling makes it possible to estimate how the
general population would respond to the product. This is relatively simple
and is the most commonly adapted method used by startups. This method is
sometimes referred to as intentions and expectations surveys.

10.11.2 Aided judgment

When a startup company in a new industry lacks data about the market,
aided judgment, based on feedback from potential customers, experienced
sales staff, and industry experts, can either be aided by using only intuition
and personal judgment or aided by more specific analysis of the available
information, depending upon the situation. Comparing the fuel economy data
of a certain model car with those of other models of the same make as well
as those of other makes and models of similar-sized engines is an example
of this.

10.11.3 Judgmental bootstrapping

The term bootstrapping originated in the early 19th century in the United
States, and it usually refers to a self-starting process that is supposed to pro-
ceed without external input. In business, it refers to starting a business with-
out external help or capital. Here the entrepreneur makes the decisions by
himself, in a manner similar to unaided judgment. A majority of the startups
that are now started by young entrepreneurs, who use their judgement and
intuition in deciding their business activities, fall under this category.

10.11.4 Jury of executive opinion

NASDAQ (National Association of Securities Dealers Automated
Quotations) defines jury of executive opinion as a method of forecasting
148 Production Planning and Control

using a composite forecast prepared by a number of individual experts. The

experts form their own opinions initially from the data given and then revise
their opinions according to others’ opinions. Finally, each independent jury’s
final opinions are listed, analyzed, and combined.
Business Dictionary gives a more elaborate explanatory definition to the
concept of jury of executive opinion as a method of integrating, combining,
and averaging views of several executives regarding a specific decision or
forecast. The general practice is to bring together top executives from
finance, resources, marketing, purchasing, and production, etc., who provide
their background information, experiences, and opinions to the board of
directors. This exercise usually leads to a quicker (and often more reliable)
result without the use of elaborate data manipulation and other statistical
tools and techniques, where applicable.

10.11.5 Delphi technique

Delphi technique, whose name is derived from the Oracle of Delphi, is based
on the principle that forecasts from a structured group of individuals are
more accurate than those from unstructured groups, and that group judg-
ments are more valid than individual judgments. A panel of experts answer
prepared questionnaires in two or more rounds. After each round, these anon-
ymous answers together with the reasons offered by them for their forecasts
are summarized and given to the experts, who revise their earlier estimates
in light of those of other members of the panel, and give more logical, more
accurate figures. This is similar to the jury of executive opinion method,
except that the experts answer the questions individually in two or three
rounds.

10.11.6 Prediction markets

Prediction markets are speculations made for the purpose of making predic-
tions of future demands. Prediction of market prices generally is close to the
mean belief of market participants. This forecasting is generally adapted for
the share market and pre-poll election results. The fluctuation of the market
prices can indicate what the public thinks of the probability of the item and
consequently the fluctuations in the probable sales of the item.

10.11.7 Marketing trials

If a product is absolutely new to the market, it is difficult to know the custo-
mers’ acceptability of the product since past data is not available. In such
cases, the product is produced first as a trial batch and is sent to the market
for actual trial, with requisite advertisement, promotion, etc., and then
 Forecasting Chapter | 10 149

depending upon its success during the pilot trial, further batches are released.
This method is normally followed for items like cosmetics and toothpaste.

10.11.8 Market research

Before releasing new products into the market, organizations conduct market
research to collect required data and conduct surveys for information on fac-
tors that influence the product demand, like location, demography, nature of
the consumption, buyer perception, buyer income levels, acceptable price
structure, etc. The data collection and analysis may comprise one or more of
the methods previously detailed and would be more target-oriented. This
method can also take advantage of specialist consultants in the field of mar-
ket research, who have the specified experience and large databanks.

10.11.9 Simulated interaction

Simulated interaction in forecasting is similar to the decisions that people
will make in conflict situations such as buyer-seller negotiations, employer-
union disputes, commercial competition, hostile takeover bids, civil unrest,
international trade negotiations, etc. The primary reason simulated inter-
action works is that it gets a person to think like the other party. That is, it
puts them in the other person’s shoes. This method can be compared to
brain-storming, the creativity tool, described in Chapter 7, Terminology used
in Japanese management practices.

10.12 Quantitative forecasting techniques

10.12.1 Discrete event simulation
Discrete event simulation (DES) is the process of codifying the behavior of a
complex system as an ordered sequence of well-defined events. Each event
occurs at a particular instant in time and marks a change of state in the sys-
tem. This is basically used to monitor and predict the behavior of invest-
ments like the stock market, but this tool is now increasingly used to predict
the market for industrial goods.

10.12.2 Group method of data handling

GMDH is a family of inductive algorithms for computer-based mathematical
modeling of multiparametric datasets that features fully automatic structural
and parametric optimization of models. It is now increasingly applied for
data mining and knowledge discovery in forecasting. Using a computer aims
to minimize the influence of the forecaster on the modeling.
150 Production Planning and Control

10.12.3 Reference class forecasting

Reference class forecasting, or comparison class forecasting, is the method
of predicting the future through looking at similar past situations and their
outcomes. It involves the following three steps:
1. Identify a reference class of past, similar projects.
2. Establish a probability distribution for the selected reference class for the
parameter that is being forecast.
3. Compare the specific project with the reference class distribution in order
to establish the most likely outcome for the specific project.

10.12.4 Quantitative analogies

Quantitative analogy is a forecasting method that assumes that two different
kinds of phenomena share the same model of behavior. For example, one
way to predict the sales of a new product is to choose an existing product
that is similar to the new product in terms of the expected demand pattern
for the product sales. Qualitative analogies are suitable for sketch under-
standing systems because they highlight important relationships while leav-
ing out details that are nonessential for conceptual understanding.

10.12.5 Game theory

Game theory, also called interactive decision theory, is the study of strategic
decision-making by creating mathematical models of conflict and coopera-
tion among intelligent rational decision-makers. Developed during the 1950s,
it was originally used in economics, political science, and psychology but
has now become an umbrella term for the science of logical decision-making
in business. The decision tree, described in Chapter 12, Decision theory, is
an off-shoot of game theory.

10.12.6 Data mining

Data mining is an analytic process designed to explore large amounts of
business or market-related data to identify consistent patterns and systematic
relationships among variables and then to validate the findings by applying
the detected patterns to new subsets of data. This is performed in three
stages:
G The initial exploration, which includes data preparation, data transforma-
tions, and selection of subsets in data
G Model building or pattern identification with validation/verification for
competitive evaluation of models
 Forecasting Chapter | 10 151

G Using the model selected as best in the previous stage and applying it to
new data in order to generate predictions or estimates of the expected
outcome

10.12.7 Conjoint analysis

Originated by Paul Green and V. Srinivasan of Stanford University from
mathematical psychology, conjoint analysis is a statistical technique used in
market research to determine how people value different attributes (feature,
function, benefits) that make up an individual product or service. It deter-
mines which combination of a limited number of attributes influences the
respondent’s choice or decision-making. It is also used in testing customer
acceptance of new product designs or the appeal of advertisements to the
customers and also in service design.

10.12.8 Causal models

Business Dictionary defines forecasting as estimating the cause-and-effect
relationship based on the assumption that the variable to be forecast
(dependent variable) has a cause-and-effect relationship with one or more
other (independent) variables. In other words, the information available in
the value of one variable enables us to forecast the value of another
variable.
For example, information about temperature variations would help us to
forecast the sales of electric fans, or some information about the international
cricket match series would help us to forecast the sale of T-shirts and similar
sports goods. Several informal methods used in causal forecasting use the
judgment of the forecaster.
Some forecasts take account of past relationships among variables. If one
variable has, for example, been approximately linearly related to another for
a long period of time, it may be appropriate to extrapolate such a relationship
into the future.
The causal methods can either be by regressive or autoregressive
analysis:
G Regression analysis includes a large group of methods for predicting
future values of a variable using information about other variables. These
methods include both parametric (linear or nonlinear) and nonparametric
techniques.
G Autoregressive moving average with exogenous inputs (ARMAX).
Quantitative forecasting models are often judged against each other by
comparing their in-sample or out-of-sample mean square error, although
some researchers have advised against this.
152 Production Planning and Control

10.12.9 Segmentation
As the name implies, segmentation classifies the customers into A, B, or C
groups according to their buying pattern and then analyzes them based on
the group or the segment. This is somewhat similar to the A, B, and C
grouping of inventory items or the classic 80-20 rule of the economic seg-
mentation of the general population. Segmentation is essentially the identifi-
cation of subsets of buyers within a market that share similar needs and
demonstrate similar buyer behavior.
The customers are divided into segments based on their buying behavior,
and it might happen that a group includes seemingly different customers
from manufacturing, retail, and oil and gas, but their buying behavior would
be the same.

10.12.10 Cross-sectional forecasting

On several occasions, we may use the information on the cases that we have
observed to predict the value of something we have not observed. This is
mostly used in share market analysis for determining the equity premium by
using general statistical methods for testing stock-return predictability based
on endogenous variables. A similar concept can also be applied to forecast
the sales time series with the cross-sectional price of risk, which is strongly
correlated with the market’s yield measures.

10.13 Life cycle effect on forecasting

A few decades ago, once a product was introduced in the market, it might
have had a long life cycle of 3 5 years for general products and 10 years
for automobiles; that is, the customers were buying the same products repeat-
edly for long periods. But today, the typical product life cycle in many mar-
kets is a matter of months, thanks to competitive pressures and the speed of
technological development, as well as the organizations’ commitment to
delight their customers.
Consequently, to predict sales of mature products in such dynamic mar-
kets, companies should address the transitional phases of a product’s life
from introduction to maturity. Fig. 10.2 illustrates the life cycle, and the data
needed, the recasting methods, and the time horizon needed for the forecast-
ing are detailed in Table 10.1.

10.14 Forecasting errors

Forecast error is the difference between the actual and the predicted or fore-
cast value. This error measurement plays a critical role in tracking forecast
 Forecasting Chapter | 10 153

FIGURE 10.2 Product life cycle.

TABLE 10.1 Product life cycle on the forecasting methods.

Phase A. B. C. D.

Introduction Growth Maturity Decline phase

phase phase phase
Data No data Some Considerable Abundant data
available data data available available, but useful
except available on demand only for new
qualitative or for trend, products
subjective analysis inventory
estimates levels, etc.
Forecast Market All the Market Continued use of
methods surveys, methods surveys, still the methods
judgement discussed useful, coupled discussed basically
methods, in this with regression for extending the
Delphi, chapter analysis maturity and
historical decline period
Tracking Tracking of
analogy, etc.
of product history
product is essential.
history is
essential
Horizon Very long time Fairly Short-term Short-term
time for horizon long time forecasts, with forecasts. Time to
forecasts horizon long-term change over
projections
154 Production Planning and Control

accuracy, monitoring for perfections, and benchmarking the forecasting pro-

cess. Normally, this error is expressed by the following four measures:
1. The MAPE (Mean Absolute Percent Error), which measures the size of
the error in percentage terms
2. The MAD (Mean Absolute Deviation), which measures the size of the
error in absolute units.
3. The GMRAE (Geometric Mean Relative Absolute Error), which mea-
sures the out-of-sample forecast performance. It is calculated using the
relative error between the naı̈ve model (i.e., next period’s forecast) and
this period’s actual
4. The SMAPE (Symmetric Mean Absolute Percentage Error) is a variation
on the MAPE as a ratio between the average of the absolute value of the
actual and the absolute value of the forecast.

10.15 Costs of forecasting

Like any other activity, forecasting too costs money. However, organizations
that pay no attention to forecasting and presume that whatever happened ear-
lier will continue to do so as it is in future would end up in paying a heavy
price in the form of excessive labor, material, capital costs, and lost revenues
(Fig. 10.3).
Note that this cost curve is similar to the cost of optimal preventive main-
tenance activity illustrated in Fig. 8.1 of the book Maintenance Engineering
and Management—Precepts and Practices, or the cost of maintaining the
quality conformance level illustrated in Fig. 8.4 of the book Total Quality
Management, an Integrated Approach, or maintaining the optimal safety
level illustrated in Fig. 12.4 of the book Professional Ethics and Human

FIGURE 10.3 Cost of forecasting.

 Forecasting Chapter | 10 155

Values. All these curves refer in general to the selection of the optimal level
of an activity based on the cost-benefit analysis.

10.16 Tracking signals in forecasting

A tracking signal is an automatic indication of variation of actual forecasts
in relation to sales, inventory, or anything pertaining to an organization’s
future demand. It monitors and warns when there are unexpected departures
of outcomes from the forecasts.
A tracking signal can be defined as the ratio of the cumulative sum of the
deviations between the estimated forecasts and the actual values to the mean
absolute deviation. In 1967, Trigg and Leach proposed an adaptive method
of forecasting where the tracking signal ratio as above is used as the smooth-
ing parameter in simple exponential smoothing.

10.17 International symposia on forecasting

The International Institute of Forecasters (IIF) conducts International
Symposia on Forecasting (ISF) annually, mostly during spring, and are held
globally. The 34th ISF of June 2014 was held at Rotterdam, Netherlands,
and the 35th ISF in June 2015 was held at Riverside, California. The 36th
ISF was held June 2016 at the Palace of La Magdalena in Santander, Spain.
The 35th ISF listed the following sessions as illustrating how these sym-
posia widen the field of forecasting activity not only for the industrial goods
but other activities. This list gives us an idea of the latest trends and concepts
in forecasting:
1. Climate predictability
2. Forecasting electricity demand
3. Prediction of business cycles in real time
4. Financial market volatility
5. Forecast optimality
6. Early warning signals
7. Tourism forecasting
8. Information technology trends
9. Macroeconomic trends
10. History of prediction science

10.18 Conclusion
As evident from the discussions, forecasting helps not only the overall orga-
nization but also helps the production planning department in effectively
planning the day-to-day production scheduling and controlling. It helps in
other vital planning dilemmas like economic batch quantity, make or buy
decisions, and production scheduling.