6471 - (International Series in Management Scien PDF

Techniques in
Corporate Manpower Planning

International Series in
Management Science/Operations Research
Series Editor:
James P. Ignizio
The Pennsylvania State University, U.S.A.
Advisory Editors:
Thomas Saaty
University of Pittsburgh, U.S.A.
Katsundo Hitomi
Kyoto University, Japan
H.-J. Zimmermann
R WTH Aachen, West Germany
B.H.P. Rivett
University of Sussex, England
c. J. Verhoeven
Eindhoven University of Technology
Techniques in
Corporate Manpower Planning
Methods and Applications
Kluwer • Nijhoff Publishing

Boston/The Hague/London
DISTRIBUTORS FOR NORTH AMERICA:
Kluwer Boston, Inc.
190 Old Derby Street
Hingham, Massachusetts 02043, U.S.A.
DISTRIBUTORS OUTSIDE NORTH AMERICA:

Kluwer Academic Publishers Groups
Distribution Centre
P.O. Box 322
3300 AH Dordrecht, The Netherlands
Library of Congress Cataloging in Publication Data

Verhoeven, C. J.
Techniques in corporate manpower planning.
(International series in management science! oper-
ations research)
Bibliography: p.
Includes index.
I. Manpower planning. I. Title. II. Series.
HF5549.5.M3V47 658.3'01 81-12341
AACR2
ISBN-13: 978-94-011-7416-9 e-ISBN-13: 978-94-011-7414-5
DOl: 10.1007/978-94-011-7414-5
Copyright © 1982 by C. J. Verhoeven

Softcover reprint of the hardcover 1st edition 1982
No part of this book may be reproduced in any form by print, photoprint, microfilm, or any
other means without written permission from the publisher.
CONTENTS
Preface vii
Acknowledgments xi
1 Corporate Manpower Planning 1

Interpretations of Corporate Manpower Planning 2
Importance of Manpower Planning 5
Manpower Planning and Its Relation to Corporate
Planning 12
The Multicategory Approach to Manpower Planning 19
Typologies of Organizations with Respect to
Manpower Planning 21
Goals and Restrictions of This Book 34
2 Manpower Requirement Forecasting 38

Statistical Techniques for Personnel Demand Forecasting 39
Subjective Methods for Manpower Requirement Forecasting 48
The Practice of Manpower Requirement Forecasting 51
Manpower Requirement Forecasting Procedure 56
v
vi CONTENTS
3 Manpower Availability Forecasting 60

Mathematical Models for Manpower Availability Forecasting 61
Discussion of the Models 77
Recommended Models for Manpower Planning 82
Manpower Planning Systems 83
4 FORMASY: An Interactive Manpower Planning System 85

FORMASY 85
Examples of FORMASY Models 87
Options of the FORMASY System 93
Estimation of Transition Fractions 106
Computational Efficiency of FORMASY 107
Usefulness of FORMASY 108
5 The Use of FORMASY 110

The Matching Process and FORMASY 111
Applications of FORMASY in Designing Career and
Recruitment Policies 117
Level of Aggregation 130
Planning Horizon 135
Requirements for Personnel Registration 140
Requirements for the Personnel Department 142
Concluding Remarks 142
6 A Case Study 144

Manpower Planning within the Royal Netherlands Air Force 144
Model Structure 146
Long-Term Planning 148
Medium-Term Planning 153
Short-Term Planning 155
Detailed Planning 155
Summary 157
Appendix 159
References 179
Index 185
PREFACE
The title of this book is Techniques in Corporate Manpower Planning:

Methods and Applications. Manpower planning, also called personnel plan-
ning, implies the analysis of possible discrepancies in the future between
personnel demand and supply. Personnel demand will also be called person-
nel requirement; and personnel supply, personnel availability. The notion
of corporate manpower planning refers to the planning of personnel on the
level of an industrial or governmental organization. As such, it does not
stand for manpower planning for branches of industries or labor market
studies of countries or international communities.
One type of manpower planning is the planning of short-term succes-
sions of managers or the assignment planning of positions for individual
employees for the next year. In fact, this type of short-term manpower plan-
ning is always executed, whether formally or informally, centrally or other-
wise. Another type of manpower planning, however, may be executed to
match the requirement for and availability of personnel for the medium and
long term. This type of manpower planning considers groups of employees
rather than individuals. Our goal is to consider medium- and long-term
manpower planning for groups of employees. We call this the multicategory
vii
Vlll PREFACE
approach to manpower planning. In our view, this medium- and long-term

personnel planning provides the conditions for individual manpower plan-
ning or for personnel development.
The discrepancies mentioned above between manpower requirement and
availability refer, for instance, to a shortage of highly qualified personnel, a
surplus of employees in lower grades, and so on. Such possible discrep-
ancies can be forecasted by the estimation of the future requirement for and
availability of personnel. Then one may assess the impact of alternative pol-
icies-for example, changes in the degree of automation, recruitment of dif-
ferent numbers or types of personnel, changes in career opportunities, and
the like-on the evolution of manpower requirement and availability. Al-
ternative policies may lead to management decisions. Since employees often
stay with an organization for a long time, management decisions concerning
manpower produce a long-term impact on it. Moreover, some time may
pass before decisions with respect to the personnel become effective. If, for
instance, the organization needs employees with much experience, then it
has to recruit them many years in advance. However, decisions with respect
to manpower flows, such as changes in recruitment and increase or decrease
of average time until promotion, have consequences not only for the organi-
zation but also for the employees. In our view, medium- and long-term
manpower planning is important to dissolve discrepancies between man-
power requirement and availability; but the impact of alternative policies on
the interests of the employees has also to be considered.
In this book we consider techniques for medium- and long-term corpo-
rate manpower planning-in particular, techniques that demonstrate the
impact of managerial policies on the evolution of manpower requirement
and availability. Such techniques are needed in order to forecast quantita-
tive and qualitative manpower requirement and availability. Also, they
must offer the possibility of considering the impact of alternative policies
on demand and supply of personnel, career opportunities, salary costs, and
so on. Management needs a decision support system that facilitates the ap-
plication of such instruments. Again, these techniques must offer the op-
portunity to analyze possible discrepancies with respect to manpower and to
evaluate the impact of alternative policies. To this effect, we have developed
the manpower planning system FORMASY, which is also described in this
book. The application of FORMASY in some large Dutch organizations
has already yielded much insight into the applicability of manpower plan-
ning instruments.
In our view of manpower planning, one first has to estimate both future
manpower requirement and future manpower availability. A matching pro-
cess follows in which these forecasts are compared and analyzed for possi-
ble discrepancies. In this phase, the impact of alternative policies on the
PREFACE ix
manpower requirement and availability forecasts must be assessed in order

to solve the discrepancies. The techniques for executing this manpower
planning process are the subject of this book. The chapters correspond to
the different phases of our manpower planning approach.
In chapter 1 we introduce medium- and long-term corporate manpower
planning. The importance of this type of personnel planning is discussed, as
well as the different manpower planning activities that are part of it. In
addition, we consider the type of organization and the type of personnel fCir
which these manpower planning activities are important. We also discuss
why groups of employees rather than individual employees are considered.
In this chapter we also explain in some detail the goals and restrictions im-
posed on this book.
In chapter 2 we investigate manpower requirement forecasting. Statis-
tical and subjective methods are described, and an analysis is given of the
way some Dutch organizations try to forecast their manpower requirement.
Finally, we outline a procedure for manpower requirement forecasting.
Chapter 3 is devoted to some mathematical models for manpower avail-
ability forecasting as found in the literature (see, for example, Bartholomew
1973; Bartholomew and Forbes 1979; Grinold and Marshall 1977). It has
been shown that planning systems based on such models can be used to de-
scribe the evolution of personnel size, to analyze career prospects, and so
on. However, these methods are not extensively used or reported on.
In chapter 4 we introduce our manpower planning system, called
FORMASY, which is based on Markov models. This manpower planning
instrument is an approach to manpower availability forecasting and the
matching process that uses an interactive computer program system.
FORMASY can be used by members of a personnel or planning depart-
ment. The fact that FORMASY can be used by experts on manpower plan-
ning adds to the applicability of this system.
In chapter 5 we consider the use of FORMASY for the matching of man-
power requirement and availability. We explain how it can be used to evalu-
ate alternative policies on recruitment, career opportunities, and so on, with
respect to their impact on the evolution of manpower availability. We also
discuss the level of aggregation (the categories of personnel) that can be
chosen and the planning horizon length that is useful. Moreover, we make
some remarks on the requirements of the introduction of such manpower
planning systems for the registration of personnel and for the organization
of the personnel department.
In chapter 6 we illustrate the applicability of FORMASY by presenting at
case study made within a large Dutch organization.
Finally, in the appendix we give part of the processing list of an applica··
tion of FORMASY on a general purpose computer.
ACKNOWLEDGMENTS
This book has grown out of my research for the interuniversity working
group "Dynamic Programming Applications" of Eindhoven University of
Technology and the Graduate School of Management in Delft. The goal of
this working group is the improvement of the applicability of mathematical
methods, such as dynamic programming and forecasting methods based on
Markov models. As application areas, manpower planning and production
and inventory planning have been chosen. I am indebted to the members of
this working group and, in particular, to Dr. J. Wijngaard, who partici-
pated in the research. Likewise, I am grateful for the many discussions held
with Prof. Dr. P. M. Bagchus.
I want to express my gratitude also to my colleagues who offered sugges-
tions for improvement and to the staff members of some Dutch organiza-
tions who helped to form our ideas on manpower planning and cooperated
in the application. I want to mention here, in particular, Philips Ltd. (re-
search department of the directorate of Social Affairs), Royal Netherlands
Airforce (manpower planning department), Ministry of the Interior (per-
sonnel department of the directorate of Police), and the Public Works
Office (personnel department).
xi
xii ACKNOWLEDGMENTS
The research was performed under the supervision of Prof. Dr. J.

Wessels. I want to thank him for his contributions in the research and for
his stimulation. Also, I am grateful for the contributions of Prof. Dr. W.
Monhemius.
1 CORPORATE MANPOWER
PLANNING
Corporate manpower planning is the planning of personnel for large or-

ganizations, such as industrial firms, public services, and hospitals. Man-
power planning is the process of forecasting manpower requirement and
manpower availability and matching this demand and supply. Instruments
based on mathematical models can be useful for this forecasting as well as
for the analysis of discrepancies between personnel demand and supply.
Mathematical models for manpower planning were used first to forecast
the evolution of personnel distribution (Seal 1945; Vajda 1947). Personnel
distribution indicates the number of employees on different levels of the or-
ganization corresponding to grade, rank, salary, and the like. Computer-
aided systems, based on mathematical models, have been developed to de-
scribe this evolution. Many publications on this subject appeared in the
seventies (see, for instance, Bartholomew and Forbes 1979; Bryant and
Niehaus 1978; Grinold and Marshall 1977. The publications mainly consid-
ered the mathematical aspects of manpower planning instruments.
In this chapter we explain our view on medium-term and long-term cor-
porate manpower planning. This view has proceeded from our research on
Dutch organizations. Of course, social and legal circumstances influence
the goals and restrictions of manpower planning. Examples of such circum-
2 CHAPTER 1
stances are labor laws, the impossibility of involuntary dismissal or transfer

of employees, equal employment opportunities, and so on. We believe,
however, that the instruments described in this book are also suitable in
other countries.
INTERPRETATIONS OF CORPORATE
MANPOWER PLANNING
In this section we introduce some views on corporate manpower planning

found in previous publications. In most interpretations only the interests of
the organization are considered, namely. the matching of personnel supply
and demand. Only a few interpretations consider-besides the interests of
the organization-the interests of the employees. These interests may be the
ambition to do a specific type of work, to earn a certain salary, and the like.
We will present such an interpretation of the tasks of manpower planning in
this chapter. We will also compare the definitions of manpower planning in
some detail and give our own view of manpower planning.
Interpretations considering only organization interests
In the view of Vajda (1970) "manpower planning is concerned with arrang-

ing for the necessary number of suitable people to be allocated to various
jobs, usually in a hierarchical structure."
An analogous interpretation has been proposed by Geisler (1967). He
states that manpower planning is the process by which a firm ensures the
availability of the right number of people and the right kind of people, at
the right place and the right time, doing things for which they are econom-
ically most useful. This process includes not only forecasting and control-
ling but also the development and implementation of instruments to provide
top management with the necessary information.
Bowey (1977) considers manpower planning as "a strategy for matching
future manpower numbers and skills with organizational activities."
Kiiching (1973) regards manpower planning as the set of activities to be
executed by the company in order to make available the necessary number
of personnel with respect to the profitability or cost-minimizing principle.
Casson (1978) mentions two kinds of activities under the heading "man-
power planning":
CORPORATE MANPOWER PLANNING 3
1. A regular monitoring activity, through which manpower and the re-

lationship of manpower to the business can be better understood, as-
sessed, and controlled; problems can be highlighted; and a base can
be established from which one can respond to unforeseen events,
2. An investigative activity, by which the manpower implications of
particular problems and changing situations-for example, the re-
structuring of the business-can be explored and the effects of
alternative policies and actions can be investigated.
An interpretation considering the interests of employees also
Here we consider an interpretation of the tasks of manpower planning that

pays attention to the interests and ambitions of the individual employees.
Wachter (1974) points out that manpower planning is useful only if the
employees can take part in the process of goal setting. Based on this consid-
eration, the following formulation of the tasks of manpower planning has
been developed by Gohl and Opelland (1978) (see Schmidt, Hagenbruck,
and Samann 1975): "Manpower planning is an integrated part of corporate
planning in order to keep available a necessary number of employees with
qualifications and dispositions-well fitting to the work tasks-under the
assumption of an equivalent handling of the corporate goal to achieve an
economic design of work and realization of individual goals."
Discussion
In our opinion, the realization of individual interests should be an essential

consideration for manpower planning. Such individual interests may con-
cern salary level, career opportunities, work circumstances, protection from
dismissal or involuntary transfer, and so on. If, while doing manpower
planning, one does not consider whether such goals can be achieved now or
in the future, this may have negative consequences both for the organiza-
tion and for the individual employees. A consequence for the organization
may be that the best employees leave the organization or that the perfor-
mance of the employees deteriorates. The consequence for the employees
may be that their goals are not achieved. Of course, such considerations ap-
ply in a different degree for different countries or organizations. In some
countries-for instance, in the Netherlands-the commitment of many or-
ganizations toward their personnel is high, which means that protection
4 CHAPTER 1
from dismissal exists as well as good career opportunities, good work cir-
cumstances, and the like. This is partly in contrast to the United States,
where people can be fired on short notice in many organizations.
We will discuss the commitment of organizations toward their personnel
in more detail later in this chapter. Here we want to emphasize the necessity
of considering individual prospects and interests when developing man-
power policies. Therefore, in our opinion, the definitions of manpower
planning that consider only organization interests are insufficient. We are in
sympathy with the interpretation that incorporates individual interests in
manpower planning.
Notice, moreover, the correspondence among the interpretations consid-
ering only organization interests, except the one given by Casson. These
definitions emphasize the search for manpower policies such that the or-
ganization's demand for personnel (qualitatively and quantitatively) will be
answered. These interpretations emphasize organization interests but do not
discern the importance of including individual expectations and interests.
Casson describes the activities belonging to manpower planning, particular-
ly a regular monitoring activity and an investigative activity for particular
problems. His interpretation stresses the importance of developing man-
power policies and, implicitly, of developing instruments for the analysis of
manpower policies. In fact, this opens the way to considering individual
interests.
Our own view on manpower planning
In our view, manpower planning is a set of activities designed to keep avail-

able the necessary numbers of employees with the necessary qualifications
in order to realize the organization's goals while taking into consideration
the interests of the individual employees. This definition of manpower plan-
ning is in fact a reformulation of the interpretation of Gohl and Opelland.
Consider our definition somewhat further. The necessary qualifications
indicate the abilities required of employees for the performance of tasks
within the organization. Such abilities may be expressed, for instance, by
the training, experience, or age of the employee. We will discuss the charac-
teristics of employees that are important for manpower planning somewhat
further in chapter 3.
The implication of this definition of manpower planning is that several
manpower planning activities that may have a regular monitoring or inves-
tigative nature have to be executed (see Casson 1978). These manpower
planning activities (career planning, recruitment planning, training and
development planning, and allocation planning) will be described later in

this chapter.
This definition brings us to the distinction of the following phases in the
manpower planning process:
The forecasting of manpower requirement (both qualitatively and quan-

titatively),
The forecasting of manpower availability (both qualitatively and quanti-
tatively),
The matching of manpower requirement and availability.
The distinction-which will be sustained throughout the book-follows

particularly from the importance of considering the organization's commit-
ment to its personnel. The distinction enables the organization to analyze
possible discrepancies between manpower requirement and availability
forecasts and to develop alternative manpower policies. Thus both the or-
ganization's interests and the goals and interests of the employees can be
considered.
If one interprets the word planning in a narrow sense, then only the third
phase would deserve the label manpower planning. However, we will use
the notion of manpower planning in a broader sense and include both fore-
casting and matching of personnel requirement and availability.
We will restrict ourselves to medium- and long-term manpower planning.
In this book we will interpret the time indications of short, medium, and
long in the following sense: in short-term planning one considers a horizon
of less than one year; in medium-term planning one considers roughly one
to ten years ahead; and in long-term planning more than ten years ahead are
considered.
IMPORTANCE OF MANPOWER PLANNING
In this section we describe some circumstances that emphasize the impor-

tance of medium- and long-term manpower planning.
Let us discuss first a very simple case. Consider an organization in which
the manpower requirement will remain constant over a very long time (both
qualitatively and quantitatively). Moreover, the manpower availability has
been constant over a very long time. We will call such an organization a
stable organization. In that case, medium- and long-term manpower plan-
ning for groups of personnel (multicategory manpower planning) is not
necessary since the matching of manpower requirement and availability can
6 CHAPTER 1
be executed easily. In fact, the only manpower planning activities to be con-

sidered are the allocation of jobs to the individual employees, personnel de-
velopment insofar as the succession of management is concerned, and possi-
bly job rotation for staff functions. However, these subjects are beyond the
scope of this book.
Consider, on the other hand, an unstable organization in which, for in-
stance, the manpower requirement is strongly varying. In that case, a match
of manpower requirement and availability will not automatically exist.
Thus future manpower requirement and availability have to be estimated,
and policies must be analyzed in order to prevent possible discrepancies.
Usually the ideal situation of stability does not occur, which implies that
some medium- and long-term manpower planning activities are always
needed. The following discussion focuses on some circumstances that call
for manpower planning.
Circumstances calling for manpower planning
Decreasing growth of organizations. Some organizations have known sev-

eral years of growth leading to the engagement of many-usually young
-recruits. These recruits could be offered a decent career because the
growth generated a sufficient number of functions on a higher organiza-
tionallevel. Of course, this does not imply that all recruits would reach the
highest organizational level (expressed, for instance, by salary, responsibil-
ity, or required qualifications); but for people with high potential the prob-
ability of achieving a high position has increased compared with the situa-
tion of no growth or even of decreasing size. If such an organization is no
longer growing, then average career prospects-in the'sense of the positions
that can be attained and the time in which this may happen-will diminish,
at least if the wastage fractions remain constant. Wastage refers to the em-
ployees who leave the organization because of retirement, dismissal, death,
and so on.
Figure 1.1 illustrates the effect mentioned above by showing an example
of the age distribution of employees in the actual situation after several
years of growth (a); then the forecasted age distribution about ten years
later if the growth were to. continue (b); and the distribution if the growth
were to stop (c). Recruitment occurs in the younger age groups, and the re-
tirement age is 65.
Figure 1.1(c) shows how the age distribution will be disturbed if the
growth of an organization suddenly stops. Since recruitment will decrease,
the fraction of experienced employees will increase, which-paradox-
age age age
65
60
50
40
30
20
no. of no. of no. of

employees employees employees
(a) (b) (c)
FIGURE 1.1. Actual age distribution after several years of growth (a); the fore-
casted distribution about ten years later if the growth continues (b); the forecasted
age distribution about ten years later if the growth stops (c). It is assumed that the
retirement age is 65 and recruitment occurs in the younger age groups.
ically-may not be a desirable development (for instance, for a research lab

requiring many young engineers). The poor career prospects may imply that
employees cannot be promoted even though they are qualified. So there are
unequal opportunities for the employees who entered the organization at
different times. These effects affect the individual employees. However,
poor career prospects will also affect the organization, since some employ-
ees may become frustrated, the better people may leave, different require-
ments arise for training and for the type of people recruited, the qualitative
structure of the organization (that is, the employees with their experience,
training, and so on) may change in an undesirable direction, and the like.
Moreover, some organizations have a high commitment to their personnel,
as in some Dutch public services, where sudden changes of career oppor-
tunities are not allowed. Examples of the effects of decreasing growth are
given in chapter 5, and in Van der Beek, Verhoeven, and Wessels (1978) and
Wijngaard (1978).
8 CHAPTER 1
Another example of decreasing growth of an organization is encountered

if an organization plans to transfer its activities to a different location. This
may not cause a problem if the geographical mobility of the employees is
high. If their mobility is low, however, the matching becomes difficult:
overstaffing may occur in the old location and understaffing in the new
location. Moreover, if the employees who will not be transferred are highly
qualified (experienced), then there will also be development and training
problems in the new location.
Irregular age structure. In some organizations the actual age distribu-

tion of the employees is irregular because of manpower policies in the
past. Such an irregular age distribution may occur if the total organiza-
tion size has changed much in the past and if recruitments take place in
only a few age classes (as is the case in many organizations-for in-
stance, in defense organizations and police forces). If an organization has
had a period of growth followed by several years of roughly constant size
and has grown again recently, the result may be an age distribution with
relatively many older employees and many younger employees and only a
relatively small number of employees in the intermediate age groups (see
figure 1.2).
Examples of irregular age distributions are given in chapter 6 as well as in
Van Meeteren (1978) and Verhoeven (1981).
Particularly in organizations with recruitment only in the lower age
classes, the age of an employee can be seen as an indication of qualification
or experience (as in the police forces). An irregular age distribution as in fig-
ure 1.2 then implies the following:
1. The distribution of the qualifications of the employees is irregular;

2. The numbers of recruits in the future will vary heavily.
Figure 1.2 shows that in the actual situation many employees are highly
qualified and many employees have little experience, whereas only a few
people belong to the intermediate class. In the near future, many employees
will retire, which means that a shortage of highly qualified personnel will
arise. If recruitment takes place only in the young age classes, then the ca-
reer prospects of the employees should correspond to age. This may lead to
difficulties if the situation in figure 1.2 occurs. On the one hand, the em-
ployees in the intermediate classes have good career prospects. These em-
ployees may be promoted because of many retirements, even though these
employees may not yet have the required qualifications. On the other hand,
the younger employees will have poor career prospects if the total number
of employees remains constant from now on.
age
60
50
40
30
20
no. of
employees
FIGURE 1.2. Actual age distribution for an
organization after several years of growth fol-
lowed by several years of about constant size
and again several years of growth. Recruit-
ment occurs in only a few (young) age classes.
The high number of retirements in subsequent years will lead to the re-
cruitment of many young people if the number of employees remains con-
stant. Then, after several years, the recruitments will decrease if the re-
tirements decrease. If this process continues, an irregular age distribution
will continue for an extremely long time, since the wave in the age distribu-
tion propagates itself in the manpower system.
In a situation of irregular age distribution, where age is so closely related
to the qualifications of the employees, one has to forecast future manpower
requirement as well as availability, both qualitatively and quantitatively.
Alternative policies have to be developed in order to obtain a better age dis-
tribution-for instance, by recruitment in different age groups, flexible re-
10 CHAPTER 1
tirement ages, training and development, together with changes in career

prospects for the lower-level employees, and so on. Investigations of the age
distribution also help in studying the impact of previous recruitment pol-
icies.
Salary structures. The salary costs for the organization depend heavily on
the evolution of the personnel distribution, particularly on the number of
employees in different grades and ranks; but they may also depend on the
evolution of the age distribution. Medium- and long-term forecasts of sal-
ary costs are useful in the case of an irregular age distribution or an irregu-
lar distribution of any other salary-determining factor. We will consider
here the impact of an irregular age distribution on medium- and long-term
salary costs. If the actual age distribution is characterized by many young
employees and a relatively small number of older people (see figure 1.3 [a])
and if the organization will be of constant size in the future, the age distri-
bution-several years later-may have the shape of figure 1.3 (b).
age age
no. of no. of
employees employees
(a) (b)
FIGURE 1.3. Actual age distribution (a) and the forecasted age distribution after
about twenty years if the organization remains at about constant size (b).
In the case of figure 1.3 (a) the salary costs will be relatively low, since
most employees are in the younger age groups, compared with the situation
of figure 1.3 (b), which will result about twenty years later. Because of the
relevance of this uncontrollable evolution for financial planning, it is im-
portant to estimate future manpower availability such that future salary
costs can be determined.
Medium- and long-term manpower planning is particularly important
for the financial planning of labor-intensive organizations, such as public
services, banks, and insurance companies. The forecasting of salary costs
will be discussed in greater detail in chapter 4. An example is given in
Verhoeven, Wessels, and Wijngaard (1979).
Increasing complexity of organizations. Organizations have become more

complex in time. The importance of planning in large and complex
organizations has been described by Koontz and O'Donnell (1974),
among others. Manpower planning, in particular, also becomes more im-
portant. In some cases, the content of personnel tasks has changed
(specialization); in others, the number of hierarchical levels has been in-
creased, technology has become complex, and so on. The increasing com-
plexity of organizations leads to the employment of differently skilled
and trained personnel.
On the other hand, the functional and geographic mobility of employees
is often low. Therefore, the qualitative as well as the quantitative require-
ment and availability of personnel have to be estimated in order to deter-
mine necessary recruitment, development and training, reformulation of
tasks, and the like.
Top management of large and complex organizations is interested in
roughly controlling manpower distribution and manpower policies, since
one is not able to influence manpower distribution or policies in a very
detailed way. As was mentioned earlier, manpower distribution indicates
the numbers of employees at different levels of the organization (corre-
sponding to salary, grade, age, and so on). Manpower policies are partic-
ularly useful here not only with respect to recruitment and training, but
also with respect to career opportunities in that unequal career prospects
may be avoided for employees in different locations and/or divisions of
the organization. Avoiding unequal career prospects is important be-
cause, among other reasons, it helps to ensure that the abilities of em-
ployees are used in the best way and will be used so in the future. On a
lower level of the organization, management will be interested more in
the impact of the manpower policies for their part of the organization (a
more detailed approach).
12 CHAPTER 1
Length of the feedback cycle of manpower policies. Another reason for

the importance of medium- and long-term manpower planning is the length
of the feedback cycle of manpower policies. If one notices, for instance, a
lack of experienced employees in the organization, this cannot always be
remedied in the short term. In the case of a shortage of experienced employ-
ees, one may hire people from other organizations who have the required
experience, but the hiring and working-in of these people might take about
two years. Another possibility for remedying the discrepancy would be the
development and training of employees (see Barber 1968), but this might
take five to ten years. This example illustrates the importance of monitoring
the evolution of manpower requirement and availability, both qualitatively
and quantitatively, and the signaling and analyzing of possible discrep-
ancies in time.
The manpower decision-making process. Decisions with respect to man-

power policies have a great impact on the organization and on individual
employees. Therefore, in large and complex organizations such impacts
have to be studied in detail. Moreover, many deliberations on various levels
of the organization will precede a decision (see, for example, Likert 1961).
Deliberations within top management, middle management, work councils,
and the like, increase the demand for information about the consequences
of alternative manpower policies on the career opportunities of personnel,
necessary capacity of training institutions, finance, and so on. This infor-
mation can be obtained by forecasting the requirement and availability of
personnel, both qualitatively and quantitatively. Then possible discrep-
ancies have to be determined and alternative policies designed to dissolve
these discrepancies. Examples of manpower forecasting and the design of
alternative policies are given in Verhoeven, Wessels, and Wijngaard (1979)
and in chapter 5. The importance of making carefully considered decisions
and the long time required for deliberations imply the necessity of taking
relatively long views.
MANPOWER PLANNING AND ITS RELATION

TO CORPORATE PLANNING
In this section we consider the relations between manpower planning and

the planning of other organization resources. We also describe the man-
power planning process itself in greater detail.
Manpower planning and organizational planning
The goal of organizational or corporate planning is the matching of re-

quired and available resources (finance, raw materials, equipment, and per-
sonnel). This resource planning covers the determination of necessary re-
source requirements, plans for the acquisition or generation (supply) of
resources, and, finally, the allocation of the resources (see Ackoff 1970). A
change in the level of one of the resources has consequences for the level of
the other resources. For the planning of each of the resources, therefore,
one needs information about the other resources (figure 1.4).
The purchase of equipment in the form of new types of machines, for in-
stance, permits the processing of more raw materials; but it also has finan-
cial consequences and may lead to a requirement for different numbers and
types of personnel. If people with certain qualifications were needed and re-
cruited in the past, then this may restrict investment in equipment, and it
also has financial consequences. Similar considerations hold for the impact
of changes in career policies, training and development programs, and
allocation patterns on the organization's resources, and conversely. Notice,
however, that the importance of such impacts depends on the type of or-
ganization: for processing industries, the consequences for the planning of
raw materials will be important, whereas these resources in effect do noll
exist for banks, and so on.
FIGURE 1.4. Information flows for the planning of

the organization's resources.
14 CHAPTER 1
Factors influencing manpower requirement and availability
Particularly manpower requirement depends on the levels of other re-

sources. The organization plan provides future organization activities in the
sense of levels and types of production or services. Moreover, this plan indi-
cates (or can be translated into) the necessary numbers and types of re-
sources to execute these activities. Thus the relationship of manpower re-
quirement, qualitatively and quantitatively, to the levels of other resources
is also included in this plan. But as we have just noted, the level of every re-
source depends on the level of any of the other resources. Therefore, when
laying an organization plan, there has to be feedback with respect to all re-
sources. The starting point for the organization plan, which is therefore the
basis for the manpower requirement, is the forecasted demand for the prod-
ucts or services of the organization, which in turn depends on such factors
as the market share and the economic situation. One further variable influ-
encing manpower demand is innovation with respect to products or services
and with respect to production methods (for example, automation). Notice
that the organization cannot control some variables, such as the economic
situation, but it can partly influence others, such as market share, level of
automation, and the market in which the organization operates. We will de-
fine external variables as those variables that are out of the organization's
control.
Manpower availability in subsequent years depends on the actual man-
power distribution and the existing manpower policies (concerning careers,
training, and so on), which may be explicitly formulated or which may exist
implicitly. Furthermore, it depends on external variables, such as the labor
market situation, the attitude of trade unions, social influences, and so on.
The values of these variables determine the evolution of internal manpower
availability-that is, the numbers and qualifications of available employees
within the organization.
Figure 1.5 expresses the influences of the various variables on manpower
availability and requirement. The activities plan indicates the future pro-
duction level or amount of services to be delivered by the organization. This
plan thus incorporates the necessary levels and types of resources, including
finance, raw materials, and equipment.
The manpower planning process, now, is the forecasting of manpower
requirement and availability as well as the development of alternative pol-
icies in order to dissolve remaining discrepancies. This may imply changes
in manpower requirement, in manpower availability, or in both. In the fol-
lowing discussion we consider manpower planning activities that can help to
match manpower requirement and availability.
-
INTERNAL MANPOWER
I-IANPOWER MANPOWER REQUIREMENT - ACTIVITIES
POLICIES AVAILABILITY PLAN
FIGURE 1.5. Influences of the various variables on manpower availability and

manpower requirement and the feedback.
Manpower planning activities
In principle, a match of manpower requirement and availability can be ob-

tained not only by adapting either the requirement or the availability but
also by adapting both. Which one primarily will be adapted depends on the
type of organization. In a capital-intensive industry, such as a processing in-
dustry, the manpower requirement will be less adaptable than in labor-in-
tensive organizations, such as public services.
The matching process results in a manpower plan that contains a career
plan, a recruitment plan, a development and training plan, and an allo-
cation plan. We will call manpower planning activities the development of
such plans and also the development of alternative policies resulting in
adaptations of these plans. Thus we distinguish the following manpower
planning activities:
Career planning,
Recruitment planning,
Development and training planning,
Allocation planning.
16 CHAPTER 1
Career planning. Career planning is the development of career policies.

The goal of these policies is to have available for the organization at the
right time the properly qualified employees, while the employees have the
opportunity to perform tasks that correspond best to their abilities and ex-
pectations. (Here we consider career planning for groups of employees, not
for individual employees.)
The career of a group of employees can be expressed by several mea-
sures, such as the average time until promotion, the fraction of employees
promoted, and the like. The combination of function level and function se-
niority is often a good measure for the experience of employees. Here, func-
tion seniority (or grade seniority) is defined as the number of years spent in
a certain function level (or grade, rank, and so on). Engineers, for instance,
with a high function seniority on a high function level will be better quali-
fied-on the average-than recently graduated engineers. Since there may
be many differences in qualification or experience among employees with
the same function seniority, such measures for experience are valid only for
groups of employees (average numbers) and not for individuals.
In some organizations explicitly formulated career plans exist, as in a de-
fense organization where the percentage of employees who will be promoted
is given for every grade seniority:
Grade Seniority: 1 2 3 4 5 6 7 8 9 10 11 ~ 12
,J. ,J. ,J. ,J. ,J. ,J. ,J. ,J. ,J. ,J.,J. ,J.
Percentages: 0 0 0 0 0 0 0 35 56 65 0 0
The example shows part of a career scheme indicating the promotion per-
centages for every grade seniority (in years) with respect to promotion op-
portunities from the grade "major" to the grade "lieutenant colonel."
Thus, for instance, 56 percent of the employees who now have grade senior-
ity 9 will be promoted this year. In chapter 6 we will consider the role of
manpower planning for such an organization in more detail.
In other organizations less strictly formalized career schemes exist or ca-
reer policies are not formulated at all. In the central research laboratory of a
large industry, for instance, the promotion possibilities are not formulated
explicitly in terms of grade seniority. However, one feels that the young re-
cruited engineers should be offered the opportunity to continue their career
in a manufacturing division after a number of years in the research labora-
tory. By the career policies for these engineers, one tries to achieve this goal,
but these policies are not explicitly formulated. We will illustrate this exam-
ple in some more detail in chapter 5.
As has been said, we are interested in career planning for groups of em-
ployees; however, the overall planning career prospects may have conse-
quences for the individual employee. If the average time until promotion
from a given function level to the next higher level is lengthened, this may
imply that all employees will have to spend more time on the function level;
but it may also imply that the better employees will be promoted according
to historical career prospects, whereas some employees may not be pro-
moted at all. One goal of career planning is to make such choices. This
example illustrates that career planning for groups of employees provides
the conditions for personnel development, which, in fact, arranges the func-
tions the individual employee will occupy in the future ..
Recruitment planning. Recruitment planning is the development of re-

cruitment policies. The goal of these policies is to obtain a reasonable filling
of vacancies, both qualitatively and quantitatively, in the medium and long
term.
Recruitment policies can be expressed by the numbers of employees of
relevant types to be recruited. The type of personnel indicates, for instance,
the qualifications (experience), training, or ages of the employees.
Some circumstances that can make recruitment planning important are
labor market conditions (overemployment), restricted absorption capacity
of the organization, and restricted possibility of involuntary dismissal of
employees (which is the case in most West European countries).
Notice that there exists a link between recruitment planning and career
planning. In the case of vacancies one may also decide to recruit (promote)
employees from a 19wer level in the organization, which is sometimes called
internal recruitment. Such policies have, of course, a great impact on the ca-
reer opportunities of both lower-level and higher-level employees. Similar
considerations hold for the relation of recruitment planning and training
and development planning as well as allocation planning. We will discuss
these relations in greater detail in chapter 5.
Development and training planning. Development and training planning

is the design of policies to prepare employees for the positions they will oc-
cupy in the organization in the medium and long term.
Development and training policies can be expressed by the number and
types of employees who should have the opportuniity for additional train-
ing and by the required scope of this training. These policies in particular
emphasize the importance of matching qualitative manpower requirement
and availability. Development and training are important, for instance, if
the organization decides to automate; they are also important for the suc-
cession of management. Management training is important for qualified
employees who will occupy management positions in the future. In chapter
18 CHAPTER I
4 we will give some examples of groups of employees for whom develop-

ment and training playa significant role.
Circumstances that call for development and training planning are
limited capacities of training institutions, high required qualifications, em-
ployees' lack of willingness to be trained, and so on. Notice the relation to
career planning, where development and training can be necessary to qual-
ify for promotion.
Allocation planning. Allocation planning is the design of allocation pol-

icies for the medium and long term. Such policies indicate which type of
functions will be assigned to different groups of employees (rotation
schemes). These functions exist on the same organizational level. Thus
for allocation planning we consider the horizontal movements of employ-
ees.
Allocation planning can be important for those functions that require
highly qualified personnel, since rotation schemes may offer the opportu-
nity for employees to obtain the required qualifications during their em-
ployment in different functions on the same hierarchical level.
Consider the following example of an allocation policy:
Level 2
Levell
Employees who will obtain functions on the second level must first spend
two years in A-type functions, three years in B-type functions, and five
years in C-type functions on the first level.
Circumstances that call for allocation planning are high required qualifi-
cations, low functional mobility, and restricted numbers of suitable func-
tions for allocation to employees who will occupy these positions for only a
short time. In chapter 5 we will consider allocation planning in greater de-
tail. The importance of allocation planning will also be demonstrated in the
case study (chapter 6).
Here again, we notice the relation of allocation planning to career plan-
ning. Career opportunities may depend on the willingness of employees to
participate in rotation schemes. A similar relation exists between allocation
planning and training and development planning. Training and develop-
ment may depend on the willingness of the organization to change alloca-
tion patterns.
Discussion
We have seen that all manpower planning activities are interrelated. All ac-
tivities may be useful for matching manpower requirement and availability,
but sometimes one activity will be more relevant than others. The impor-
tance of an activity is determined by the corresponding restrictions with re-
spect to the matching. If, for instance, discrepancies with respect to training
and development of employees are forecasted, then it is important to de-
velop (or change) training and development policies.
We have argued that policies have to be developed for matching man-
power requirement and availability. In the discussion above we described
the manpower planning activities that will result in the design of such pol-
icies. Actions that may follow from these policies are, for instance, in the
case of an expected shortage of personnel:
External recruitment,
Changes in career policies,
Initiatives with respect to training and development,
Changes in allocation patterns (job rotation),
Subcontracting of work (changes in organization plans),
Changes in manpower requirement (accelerating automation),
and in the case of an expected surplus of personnel:
Dismissal (or stimulating turnover in various ways),

Contracting additional work,
Changes in allocation patterns in order to decrease the number of over-
occupied functions,
Changes in career policies.
Such actions will be discussed extensively in chapter 5.
THE MULTICATEGORY APPROACH TO

MANPOWER PLANNING
In this section we discuss why groups of employees are considered in me-

dium- and long-term manpower planning (multicategory planning) and we
explain the relationship between this type of planning and manpower plan-
ning for individual employees.
20 CHAPTER 1
Different aggregation levels can be chosen in planning. In financial plan-

ning, for instance, the investment level for the next ten years may be ex-
pressed in investment amounts for periods of two years (strong aggrega-
tion). The investment level for the next four years may be expressed in
amounts of investment for each year per operation area (some aggregation).
And the investments for next year may be indicated for each project expli-
citly (no aggregation).
In manpower planning, the level of aggregation corresponds to the de-
gree of heterogeneity of the considered personnel group. If manpower plan-
ning is executed for the total number of employees, a heterogeneous group,
one has the strongest degree of aggregation. Some aggregation exists if
manpower plans are considered for groups of employees, such as top-level
engineers, middle-level engineers, and lower-level engineers. Manpower
planning for individual employees-that is, for homogeneous "groups"-is
an example of no aggregation.
The choice of degree of aggregation depends on the following factors:
The desired degree of detail in the planning,

The practicability of the detail in the planning.
Desired degree of detail
The desired degree of detail is determined by the planning horizon and by

the organizational level on which the planning is executed. For a long plan-
ning horizon, one is not interested in a high degree of detail. For short plan-
ning horizons, a higher degree of detail is desired. In chapter 5 we will con-
sider the planning horizon in more detail; examples of the relation between
the planning horizon and the desired degree of detail are given in chapter 6.
If the organizational level on which the planning is executed is such that a
great distance exists between that level and the level of the considered per-
sonnel, there will be a low desired degree of detail. In many organizations
top management is not interested in executing detailed manpower planning
for all employees. At the top level only manpower planning with some de-
gree of aggregation will be relevant (see, for instance, Crichton 1968; Sayles
and Strauss 1977). Thus in large organizations manpower planning may be
relevant on several organizational levels, and the degree of aggregation in-
creases by increasing organizational level. In chapter 4 we will give some
examples of the level of aggregation for different organizational levels of
the planners.
The desired degree of detail may also depend on the considered personnel
group. At the top level of the organization, management development will
often be executed, which means that for a particular group of employees in-
dividual manpower planning is executed (that is, forecasting of medium-
and long-term requirements for management, the potentials of successors,
the selection of managers, necessary training and development, and so on).
Management development has been described, for instance, by Taylor and
Lippitt (1975). Further references can be found in Margerison and Hunter
(1978).
Practicability of the detail
The practicability of the detail in the planning depends on the available in-
formation. The available information is determined by the organizational
level on which the planning is executed. If a great distance exists between
that level and the level of the considered personnel, then there will probably
be insufficient information to make detailed planning possible. At the top
level of large organizations, for instance, it will be impossible to forecast in
the medium and long term which individual employees will be promoted or
will leave the organization, which functions will arise, and so on.
The level of aggregation will be discussed in greater detail in chapter 5.
Thus if some aggregation is used in manpower planning, then the personnel
is classified in groups, and the planning is based on this classification. From
the above it follows that in medium- and long-term manpower planning
some degree of aggregation will always be desired and/or necessary. This il-
lustrates the relevance of the development of instruments for manpower
planning based on groups of employees. The multicategory models for
manpower planning discussed in this book are applicable, in principle, for
all aggregation levels except, of course, for the individual level.
TYPOLOGIES OF ORGANIZATIONS WITH RESPECT TO

MANPOWER PLANNING
In this section a typology of organizations with respect to manpower plan-

ning problems is presented. The terms typology and classification are inter-
changeable. This typology has been designed to give some idea about the
types of organization, personnel, and problems for which manpower plan-
ning activities (that is, career, recruitment, development and training, and
22 CHAPTER 1
allocation planning) are important. We will call this a manpower planning

typology of organizations. We will use the term organization to refer not
only to the organization or suborganization (division) but also to the con-
sidered personnel group of the organization. The relevant manpower plan-
ning activities depend on the type of organization and on the considered
personnel group of the organization. We consider only relatively large and
structured organizations, such as large industrial firms, public services, and
hospitals, rather than small units, such as small workshop enterprises.
Circumstances indicating the relevance of manpower

planning activities
Obviously a certain amount of manpower planning is always executed, in

particular for decisions on which functions have to be allocated to which
employees in the short term. However, as we have stated before, organiza-
tions are also interested in medium- and long-term personnel planning. The
two main reasons to do medium- and long-term manpower planning are:
High rigidity and high variability of manpower requirement,

High rigidity and high variability of manpower availability.
Rigidity means that manpower requirement or availability cannot be in-

fluenced easily. High rigidity of manpower requirement occurs, for in-
stance, if the organization wants to accelerate automation but cannot do so
with the actual manpower distribution. In the case of high rigidity of man-
power requirement and/or availability, it is important to execute manpower
planning activities.
Variability means that manpower requirement or availability changes
over time. If we use the term variability, then we mean predictable variabil-
ity (that is, there are fluctuations, but these fluctuations can be forecasted).
High variability may also have occurred in the past. In the case of high vari-
ability of manpower availability and/or requirement, it is useful to execute
manpower planning activities. If, however, there are many fluctuations
with a high degree of uncertainty, then manpower planning is less useful. In
our discussion of aggregation level and planning horizon in chapter 5, we
will consider this in greater detail.
The terms rigidity and variability may refer to qualitative and/or quanti-
tative factors.
The manpower planning activities we consider in this section have al-
ready been described. They include:
Career planning (vertical movements),

Recruitment planning,
Training and development planning,
Allocation planning (horizontal movements).
We will discuss the relevance of these activities for groups of personnel and
for the medium and long term.
In order to identify the types of organization and personnel for which
manpower planning activities are necessary, one has to investigate the cir-
cumstances that cause rigidity and variability of manpower requirement and
availability.
Causes oj manpower requirement rigidity. A high degree of rigidity of

manpower requirement is caused by:
1. A low degree of controllability of the market by the organization,

2. A low degree of freedom with respect to technology and/or organi-
zation structure.
A low degree of controllability of the market by the organization means

that the organization can hardly influence the consumer market-in which
it operates-in quality and/or in quantity. In this case, the organization
would have difficulty in changing its market share or in starting other
products or services. Manpower planning is particularly important in the
case of a low degree of controllability of the market and, at the same time,
a high degree of variability of the manpower requirement. Market is used
here in a very broad sense, such that the outlet of services is also included.
Some aspects of the controllability of the market by the organization are
described, for instance, in Needham (1972) and in Sayles and Strauss
(1977, p. 157).
A low degree of freedom with respect to technology and organization
structure means that the organization is hardly able to change the technol-
ogy or the organization structure. The technology is defined as the tech-
niques and technical processes the organization applies to change inputs
into outputs. A change of technology may lead to a requirement for differ-
ent numbers and/or types of employees. The organization structure indi-
cates the numbers and types of functions and the corresponding hierarchical
levels in the organization. If the technology and/or the organization struc-
ture can hardly be changed, then manpower planning is particularly impor-
tant, and that is certainly so if-at the same time-a high degree of variabil-
ity of the manpower requirement exists.
24 CHAPTER 1
Causes oj manpower requirement variability. A high degree of variability

of manpower requirement is caused by:
1. A high degree of market variability,

2. A high degree of variability of technology and/or organization
structure.
A high degree of market variability means that the consumer market-in

which the organization operates-changes over time in quantity and/or
quality. The quantitative changes indicate the changes in the demanded
numbers of products or services. The qualitative changes indicate the de-
mand by customers for other types of products or services in the same prod-
uct or service class. The impact of market fluctuations on manpower re-
quirement has been described, for instance, in Sayles and Strauss (1977).
A high degree of variability of technology and organization structure
means that technology and organization structure change over time. As-
pects of changes in technology and organization structure are discussed in
Crichton (1968) and in Sayles and Strauss (1977).
Causes oj manpower availability rigidity. A high degree of rigidity of

manpower availability is caused by:
1. A high degree of uncontrollability of employees' characteristics,

2. Overemployment in relevant segments of the labor market,
3. High commitment of the organization toward its personnel.
A high degree of uncontrollability of employees' characteristics is caused

by the fact that some characteristics, such as age or potential, cannot be
changed. An exception to this rule exists if, for instance, older employees
can be dismissed and, at the same time, younger people can be recruited.
Other characteristics, such as experience or mobility (functional and/or
geographic), can be partly influenced (sometimes). A lack of mobility im-
plies a high degree of rigidity.
In a situation of overemployment in relevant segments of the labor mar-
ket, it will be hard to recruit employees with the required qualifications (see,
for example, Sayles and Strauss 1977, pp. 169-70).
Commitment with respect to personnel is the level of responsibility the
organization accepts for its employees' welfare. High commitment is ex-
pressed by, for instance, the obligation of offering careers to the personnel,
protection from dismissal or involuntary transfer, and the like. A high level
of commitment may arise from the type of organization (if, for instance,
highly qualified personnel are employed), from external social control, or

voluntarily.
Causes of manpower availability variability. A high degree of variability

of manpower availability is caused by:
1. An irregular distribution of employees with respect to the age or type

of personnel,
2. Fluctuations in labor market conditions in the past and in the future.
If, for instance, the organization employs many older employees, then
there will be many retirements in subsequent years. If the age distribution is
irregular (as in figure 1.2), then the number of retirements will change
heavily over time. An example of irregular age distribution will be given in
chapter 6.
An irregular distribution of employees with respect to types of personnel
exists, for instance, if recruitment has been constant over time but the frac-
tion of highly trained recruits has been fluctuating.
Fluctuations in labor market conditions in the past and in the future may
cause fluctuations in turnover. In the case of overemployment, turnover
may be high; in the case of unemployment, turnover will be low (see, for in-
stance, Sayles and Strauss 1977, pp. 52-53).
In summary, high levels of rigidity and variability of manpower require-

ment and availability indicate, in our view, the relevance of manpower plan-
ning activities. Particularly in the case of both high rigidity and high vari-
ability, the importance of manpower planning is clear. We will use these
rigidity and variability concepts for our manpower planning typology.
Existing typologies of organizations
In previous works many classifications (or typologies) of organizations have

been developed. We will present some of these classifications here in order
to investigate whether they are useful as a typology of organizations for
manpower planning.
Woodward (1958) investigated the organization structure in manufactur-
ing firms. She developed a classification of technologies from which a
typology of organizations can be deduced based on "technical complexity."
The degree of technical complexity is "the extent to which the production
process is controllable and its result predictable." A high degree of techni-
26 CHAPTER 1
cal complexity occurs, for instance, in the processing industry. In organiza-

tions based on small batch and unit production, the degree of technical
complexity is low.
Another well-known classification was made by Etzioni (1968). His
typology is based on two elements: the "power applied by the organization
to lower participants" and the "degree of involvement in the organization
by lower participants."
The typology of Blau and Scott (1970) is based on the "cui bono" prin-
ciple. This means that the organization is typified by the category of persons
who are the prime beneficiaries of the organization's operations. The cat-
egories of persons are: (1) the members of the organization, (2) the owners
or managers, (3) the clients, and (4) the public at large. Four types of
organizations then result: (1) mutual-benefit organizations, (2) business
concerns, (3) service organizations, and (4) commonweal organizations, re-
spectively. The police organization is an example of the fourth type of
organization in the ideal situation (but not in a police state).
Harvey (1968) classifies the industrial organizations according to "tech-
nical specificity." The degree of technical specificity is indicated by the
number of product changes in the last ten years. Harvey distinguishes three
classes:
1. Technically specific organizations using sophisticated technologies

(for example, the oil industry),
2. Technically diffuse organizations with a wide variety of products,
many new products on the market, or manufacturing according to
customer specifications (for example, the electronics industry),
3. Technically intermediate industries.
Botter (1975) classifies industrial organizations according to:
1. Their place in the industry column corresponding to the flow of

goods,
2. The specificity of the assortment (the assortment is specific if no
standardizations have been achieved-unit and small-series produc-
tion).
Note that some organizations (for instance, organizations buying raw ma-
terials on one hand and selling machinery on the other hand) fill several ele-
ments of the industry column.
Katz and Kahn (1967) distinguish first-order factors that describe the
genotypic function of the organization and second-order factors that are
generally related to these basic functions. Genotypic functions refer to the

type of activity the organization is engaged in: (1) productive or economic
functions, (2) maintenance functions, (3) adoptive functions, and (4) man-
agerial or political functions, concerned with coordination and control of
resources and people. The second-order characteristics are: (1) the nature of
the through-put, (2) the processes for ensuring the continuous input of
human personnel, (3) the nature of bureaucratic structure, and (4) the sys-
tem's type of equilibrium.
Pugh, Hickson, and Hinings (1969) used the following empirical factors
as a basis for a classification of organization structures:
1. The degree to which the activities are structured,

2. The concentration of authority,
3. The degree of line control of workflow.
The organizations have been classified according to the extent to which

these factors occurred in the organizations that Pugh, Hickson, and
Hinings considered. The following types of organization result, then, corre-
sponding to their characteristics: "full bureaucracies, nascent full bureau-
cracies, workflow bureaucracies, personnel bureaucracies, implicitly struc-
tured bureaucracies."
Other classifications of organizations are given in Blau and Scott (1970),
Jackson and Morgan (1978), and Parsons (1960).
Discussion of existing typologies of organizations
Let us investigate whether the essential factors for manpower planning,

rigidity, and variability can be found in the classifications given above.
The rigidity of manpower requirement recurs more or less in the techni-
cal complexity (Woodward), the specificity of the assortment (Botter), and
some characteristics of Katz and Kahn. The variability of the manpower re-
quirement can be found in the technical specificity (Harvey). In particular
the factors concerning manpower availability do not occur explicitly in the
existing classifications (except, possibly, the commitment of the organiza-
tion, which may be expressed in the cui bono principle of Blau and Scott).
In fact, the existing classifications have not been developed for man-
power planning. Therefore, it is not surprising that essential factors for
manpower planning do not recur explicitly in these classifications. For this
reason we will now describe a typology of organizations with respect to
manpower planning.
28 CHAPTER 1
A manpower planning typology of organizations
The characteristics according to which the organizations can be classified in

order to determine which manpower planning activities are important will
be called dimensions.
The dimensions we will use here are rather broad. If the typology is ap-
plied to an organization in order to investigate which manpower planning
activities are important, then an operationalization of the dimensions is
necessary. Later in this chapter we will give some indicators that can be used
to determine roughly the place of an organization within our typology.
In our typology the organizations are classified according to the follow-
ing dimensions:
1. The commitment of the organization with respect to its personnel

(called commitment),
2. The inflexibility of the organization with respect to the market
(called inflexibility),
3. The specificity of the personnel of the considered (sub-) organization
(called specificity).
Here we will consider these dimensions in greater detail and will describe
their relation to rigidity and variability.
Commitment. The dimension of commitment indicates the obligations the

organization feels with respect to its personnel. Commitment may concern
different aspects: job security, offering of a career, and so on. If the com-
mitment is high, then career planning is usually important.
A high degree of commitment causes a high degree of rigidity of man-
power availability, since job security and the offer of a career imply that the
numbers and types of employees cannot be changed easily.
The degree of commitment may be different for different countries. In
the United States, for instance, the restriction that employees cannot be dis-
missed involuntarily is usually less strictly enforced than in the Netherlands
or in Scandinavian countries.
The degree of commitment may also be different for different (sub-)
organizations or personnel groups within a given country. In Dutch public
services, for instance, the degree of commitment is high: careers are usually
offered to the personnel, employment is ensured, and so on. On the other
hand, in some smaller private industries the degree of commitment is rela-
tively low. In some organizations, career prospects are explicitly formulated
for highly qualified employees, but not for the other employees.
In away, the degree of commitment indicates the necessity of taking into

account individual expectations together with the organization's goals.
Inflexibility. The degree of inflexibility is the degree to which the organi-

zation has to follow fluctuations in the consumer market. The degree of in-
flexibility also indicates the degree to which the organization is unable to
change its production level autonomously or is unable to start other prod-
ucts or services. In fact, this dimension combines the rigidity and the
variability of the manpower requirement. A high degree of inflexibility
corresponds to a high degree of rigidity and variability of manpower re··
quirement. Thus it indicates the importance of manpower planning activi-
ties.
If a high degree of inflexibility occurs with respect to production level,
then recruitment planning and allocation planning will usually be impor-
tant. This may occur, for instance, in a situation of high competition. If, on
the other hand, the inflexibility exists with respect to starting new types of
products or services, then different numbers of personnel with different
qualifications may be required. Thus recruitment planning and training and
development planning as well as allocation planning may be important.
The degree of inflexibility may be different for different groups of per-
sonnel within a single organization. The degree of inflexibility of an indus-
trial firm's staff department, for instance, may be low compared with the
production units if the size of the staff department is independent of fluc-
tuations in the consumer market.
Specificity. The degree of specificity of the personnel of the considered

(sub-) organization is the degree to which the personnel require certain qual-
ifications (experience). A high degree of personnel specificity indicates that
the employees can hardly perform tasks other than those they perform now.
A high degree of specificity indicates the importance of manpower plan-
ning.
Two types of specificity can be distinguished:
Internal specificity,
External specificity.
Personnel with a high degree of internal specificity are those who perform
tasks that require particular qualifications but that also occur in other
organizations. In many organizations, sales managers form an example of
this type of personnel. Personnel with a high degree of external specificity
are those who perform tasks that require particular qualifications, which
30 CHAPTER 1
are not required by tasks in other organizations. Army officers are an exam-
ple of such a personnel group.
A high degree of specificity is related to a high degree of rigidity and vari-
ability of manpower availability.
Discussion of our typology
The typology given above has been developed for manpower planning pur-
poses. It classifies organizations (or suborganizations or personnel groups
of a [sub-] organization). The way of distinguishing personnel groups,
hence the level of aggregation, will be discussed in some detail in chapter 5.
Here we will classify some personnel groups of organizations according
to our typology and will explain the important manpower planning activi-
ties. We will indicate whether the personnel group has the dimensions to a
high or low degree. It should be remarked here that the typology is only a
rough classification. In some instances, only the quantitative aspects of a di-
mension can be found in an organization to a high degree (such as market
flexibility).
Table 1.1 shows the theoretical classification of some organizations and
personnel groups. Notice that the classification of the examples has been
executed according to prevailing standards in relevant Dutch organizations.
It is not maintained that, for example, the group of sales managers in all
organizations is of the second type. The eight types of organizations will be
discussed in some detail later in this chapter.
As we mentioned before, our purpose is to give an idea of the relevant
manpower planning activities for different types of (sub-) organizations or
personnel groups. We will give here some indicators with which the place of
an organization within our typology can be determined roughly.
In our typology, one has to indicate whether the considered organization
has the different dimensions to a high or low degree. In practice, one may
investigate which manpower planning activities are important by means of a
score based on indicators. A high score means that the considered dimen-
sion is present to a high degree. We will now introduce some indicators for
the dimensions of commitment, inflexibility, and specificity.
Indicators for commitment. High commitment of an organization to its

personnel is indicated when:
The personnel may not be transferred to other locations or functions in-

voluntarily (legal standing);
Table 1.1. Classification of Organizations
Dimensions a Manpower
Type of Planning
Organization Specificity Inflexibility Commitment Activitiesb Examples
+ + + A,B,C,D Policy and

army officers
2 + + B,C,D Sales managers
3 + + A,C,D Engineers in
research lab
of large pri-
vate industry
4 + C,D Hospital nurs-
ing personnel
5 + + A,B Postal workers
6 + X Female cashiers
in large re-
tailing firms
7 + A Secretaries in
public ser-
vices
8 X Gatemen in
factories
a + indicates that the dimension occurs to a high degree; - indicates that the dimension
occurs to a low degree.
b A indicates career planning activities; B, recruitment planning; C, training and
development planning; D, allocation planning; and X, no activities are necessary.
The organization will not dismiss its employees when manpower require-
ment is smaller than manpower availability Gob security);
The organization offers a career to its personnel.
Indicators for inflexibility. High inflexibility of an organization is indi-

cated when:
Many changes in the total number of demanded products or services are
expected (and/or have occurred in the last few years);
The market share of the organization will probably change many times
unintentionally (and/or has often changed unintentionally in the last
few years);
A large number of product changes or technology changes (new ma-
chines, changes in labor productivity) are expected to occur (and/or
have occurred in the last few years).
32 CHAPTER 1
Indicators for specificity. High specificity of an organization is indicated

when:
There is a low degree of internal functional mobility of the employees
(that is, a large fraction of the employees cannot be transferred to
other functions because of specialization, for instance);
There is a low degree of external functional mobility of the employees
(that is, the employees do not have many opportunities to obtain simi-
lar functions in other organizations);
Recruits need a long training period;
Turnover rates are very low;
Overemployment exists for the personnel group.
Discussion of the types of organization
We will now discuss in some detail the types of organization and the exam-
ples given in table 1.1.
Type 1 is characterized by high specificity, high inflexibility, and high
commitment. Examples of this type are police officers and army officers.
For these groups of personnel, the degree of commitment is high: there is
job security (though employees can be transferred to other locations), em-
ployees are offered a career, and so on. External specificity is high, since
this type of employee is not found in other organizations. Medium-and
long-term recruitment planning is useful to determine the required capaci-
ties of the training institutions and to change an irregular age distribution.
Inflexibility is high, since manpower requirement may change drastically in
the course of time (political influences). Thus, career planning is useful
(high commitment), as are allocation planning (assignment of locations to
the employees), recruitment planning, and development and training plan-
ning.
Type 2 is characterized by high specificity, high inflexibility, and low
commitment. An example of this type is the category of sales managers.
This personnel group has a high degree of internal specificity, since very
specific qualifications are required, but there are sales managers in a great
variety of organizations. The requirement for sales managers depends
heavily on the fluctuations of the consumer market, particularly the quan-
titative changes. Because of competition, market variability may be high. In
these circumstances development and training planning, recruitment plan-
ning, and allocation planning will be important. Career planning may not
be required, except for some very large industries, since commitment is
often low for this personnel group.
Type .3 is characterized by high specificity, low inflexibility, and high

commitment. An example of this type is the category of engineers in are··
search laboratory of a large private industry. The manpower requirement of
the industrial research lab depends on changes in technology and on qualita-
tive changes in the consuming market (demand for new products). The high
degree of specificity is caused by the required qualifications. However,
young engineers can be recruited from the laboratory market without many
problems. The high commitment of this organization with respect to these
highly qualified engineers is indicated by the obligation of offering func-
tions on a higher organization level for experienced personnel some time
after their recruitment. Career planning is therefore necessary (specificity,
commitment). Because of the qualification demands, one should also exe-
cute training and development planning. Allocation planning may help to
obtain the right type of specialists in the functions they perform best and
feel comfortable with (also in the medium and long term!).
Type 4 is characterized by high specificity, low inflexibility, and low
commitment. An example of this type is the nursing personnel of a hospital.
The degree of specificity is high, since the employees have obtained a spe-
cific training. Medium- and long-term forecasts of the numbers and types of
patients will not show many fluctuations, which causes a low degree of in-
flexibility. The degree of commitment is low and turnover is high, so career
planning is not important. Allocation planning may be useful insofar as the
succession of wardmasters is concerned. Allocation patterns may be de-
veloped, then, for the assignment of nurses to wards for periods of several
months. Training and development planning is relevant because of the re-
quired qualifications.
Type 5 is characterized by low specificity, high inflexibility, and high
commitment. An example of this type is the category of postal workers of
the Netherlands Post Office. This public service has a high degree of com-
mitment. The employees cannot be transferred involuntarily to other loca-
tions (they can only be transferred to similar functions in their actualloca-
tions), careers are offered to all employees, and so on. Though the total
working program does not change much in time, the inflexibility is high,
since automation is accelerated. The degree of specificity is low. Only a
short training period is required, since external mobility is high and so is in-
ternal functional mobility (many types of unskilled work can be done by
these employees). Geographic internal mobility is low (fixed locations). Im-
portant manpower planning activities may be career planning, because of
high commitment, and recruitment planning, because of high inflexibility.
Type 6 is characterized by low specificity, high inflexibility, and low
commitment. An example of this type is the category of female cashiers of a
large retailing firm. The cashiers can be recruited directly from the labor
34 CHAPTER 1
market, since not much experience (qualification) is required. Because of

strong competition in shopping firms, the inflexibility is high (high variabil-
ity). Turnover rates are usually high. Therefore, recruitment planning is not
useful. The commitment is relatively low, since no career prospects are of-
fered and sometimes even no job security exists. Usually no transfers to
other affiliations occur, so allocation planning is not important. Consider-
ing affiliation managers of such a retailing firm, we have an example of type
2.
Type 7 is characterized by low specificity, low inflexibility, and high
commitment. An example of this type is the category of secretaries in a pub-
lic service. Though these employees are highly trained, the degree of exter-
nal specificity is low, since secretaries can be found in most organizations.
The degree of commitment is usually high in public services, and the inflexi-
bility can be low, at least in some public services where the working pro-
gram does not change much in time and the automation is not accelerated.
Thus the only relevant manpower planning activity may be career planning.
This planning activity gains relevance with decreasing turnover rates.
Type 8 is characterized by low specificity, low inflexibility, and low com-
mitment. An example of this type is the category of gatemen of factories.
Commitment for these employees is low (except for some very large indus-
tries); no career opportunities exist. These functions do not depend on any
market conditions and are not specific because they can be done by many
unskilled employees. Recruitment from the labor market is directly possible
without training, and so on. Thus no manpower planning activities for this
personnel group are necessary.
Notice that, for the examples above, we indicated the medium- and long-
term activities that are desirable because of the type of the organization.
However, there may be other reasons for doing some form of medium- or
long-term manpower planning-for instance, in order to determine future
personnel costs. It should also be remarked that short-term manpower plan-
ning can be desirable too. Such short-term planning can be the assignment
of employees to different functions. However, we will not consider short-
term manpower planning in this book.
GOALS AND RESTRICTIONS OF THIS BOOK
The main theme of this book is the treatment of techniques for corporate
manpower planning, with special attention to the applicability and applica-
tions of such instruments. Manpower planning concerns the regular activity
of monitoring manpower requirement, manpower availability, and the

matching process, as well as the activity of investigating particular problems
with respect to the personnel. An example of such problems is the effect of a
low degree of geographic mobility of employees on the transfer of em-
ployees to other locations.
We consider medium-term and long-term manpower planning for large
organizations. This type of manpower planning may be characterized by the
following phases:
1.Forecasting the requirement for personnel (both qualitatively and

quantitatively),
2. Forecasting the availability of personnel (both qualitatively and
quantitatively),
3. Matching personnel requirement and availability.
According to our insight into the manpower planning process, the re-
quirement for personnel has to be known first (to some extent). After the
manpower availability has been forecasted, the matching can be executed
through the development of alternative manpower policies and alternative
activity programs. Our intention is to emphasize, in particular, the tech··
niques for the forecasting of manpower availability and for the dev:,elop-
ment of alternative policies.
Let us consider the phases of manpower planning in greater detail.
Forecasting the requirement for personnel
Forecasting the personnel requirement is a less-developed area of manpower

planning. General forecasting techniques can be used, particularly for the
estimation of future manpower demand. However, this demand may be in-
fluenced by many factors that cannot be expressed in exact quantitative
data, such as future labor laws, motivation of the employees, and degree of
automation. Other factors may be forecasted rather well, such as the de-
mand for products or services. A good starting point for obtaining the
necessary information concerning future developments lies in the part of the
corporate plan describing the level of the activities (production, amount of
services, and so on) for the medium and long term. The next phase will be
the translation of this corporate production plan into a personnel require-
ment forecast. Because of large differences in organization policies concern-
ing manpower, the nature of the production processes, organization struc-
tures, and so on, there are, in our opinion, no general quantitative methods
36 CHAPTER 1
for manpower requirement forecasting applicable to all organizations and

all groups of personnel.
We will develop a rough outline for manpower requirement forecasting
in chapter 2.
Forecasting the availability of personnel
The availability of personnel with different qualifications in the future de-

pends on such factors as on the actual personnel structure (the number of
employees, their ages, qualifications, schooling, and so on), recruitments,
promotion, turnover, and training and schooling. Extrapolation of trends
concerning the development of these factors in the past gives a forecast of
manpower availability. However, it may be important to assess the impact
of changes in the numbers of recruits, career prospects, training programs,
and so on, on the future manpower supply, as will be explained below.
Therefore, methods for manpower availability forecasting should offer the
possibility of evaluating such consequences. This turns out to be an impor-
tant condition for the applicability of manpower availability forecasting
methods (chapter 3).
We have developed a manpower planning system that forecasts the con-
sequences of alternative manpower policies (concerning recruitment, career
prospects, and so on) for the evolution of personnel distribution. This in-
strument has been developed to be used by personnel managers (chapter 4).
Matching personnel requirement and availability
The final phase of manpower planning relates availability and requirement

to each other. One has to search for manpower policies (with respect to re-
cruitment, career patterns, training and development, and allocation) and
for corresponding activities plans. This is called the process of matching
manpower requirement and availability. In our approach, the matching
process requires a very flexible use of manpower availability and require-
ment forecasting instruments (chapter 5).
The manpower availability forecasting system FORMASY, which we
used in our applications (see, for instance, Van der Beek, Verhoeven and
Wessels 1978; Verhoeven and Wijngaard 1978; chapter 6, this volume), will
get special attention in this book. FORMASY, which helps in the design of
manpower policies, may be seen as a decision support system. Decision sup-
port systems for other fields of applications are described in, for instance,
McCosh and Scott Morton (1978). In our view, the attention we have paid
to applicability and applications of the techniques fills part of the gap be-
tween the theory and the practice of manpower planning that exists at the
moment.
The magnitude of the manpower planning area puts some restrictions on
this book. First, we consider manpower planning instruments only for a
large organization interested in manpower planning for groups of personnel
(multicategory approach). Second, we deal with medium- and long-term
matching for groups of personnel, and we consider the activities of career
planning, recruitment planning, development and training planning, and
allocation planning. We emphasize the importance of quantitative methods
and the development and application of interactive manpower planning sys-
tems.
In this book the organization activities plan and the environmental fac-
tors (labor market conditions, social laws, and so on), as well as the attitude
of employees, are assumed to be given and uncontrollable by the manpower
planning results. We will use such data, but we will not study the data sepa-
rately. Moreover, organization structures, reward systems, and psychQlogi-
cal and social aspects will not be drawn into the scope of this book.
2 MANPOWER REQUIRE;MENT
FORECASTING
The estimation of future personnel requirement is a very important and also

a very difficult phase of the manpower planning process. In this chapter we
consider manpower requirement forecasting, assuming that it is indepen-
dent of the manpower availability forecast. In this way one obtains a good
starting point for the matching phase, which may result in adaptations of
both manpower requirement and availability forecasts. In chapter 5 the
matching phase will be considered in some detail.
Future manpower requirement depends mainly on the organization
activities plan (see figure 1.5). Of course, many factors will influence the
operations of the organization, such as changes in the demand for products
or services, technology, the economic situation. The degree to which these
factors playa role depends on the type of organization and the personnel
group considered. In manufacturing industries, for instance, changes of
technology and competition effects will be essential. For public services,
however, government policies may be much more important. Direct person-
nel strength-that is, the number of employees directly involved in produc-
tion or in the services delivered by the organization-will usually depend
more on competition and on technology changes than is the case with the
number of supporting staff. Therefore, no general method for manpower
38
MANPOWER REQUIREMENT FORECASTING 39
requirement forecasting exists that is valid for all organizations and all per-
sonnel categories. Some authors have proposed general statistical tech-
niques and have applied them to requirement forecasting. We will describe
these statistical techniques for manpower requirement forecasting. Further-
more, subjective and informal methods may be useful to forecast man-
power requirement. Such methods are discussed in this chapter. We also
consider how manpower requirement forecasting is executed in a number of
large Dutch organizations, thus illustrating the extent to which the methods
previously mentioned are used in practice. Finally, we will outline a proce-
dure for manpower requirement forecasting.
We will use the terms demand, requirement, and need for personnel as
synonyms. These terms refer to the necessary numbers and types of person-
nel in the organization. As mentioned in chapter 1, we will consider catego-
ries of personnel instead of individuals.
So far, not many publications have appeared on manpower requirement
forecasting (see Bowey 1977). Also, reports of regularly used requirement
forecasting methods are exceptional. Bartholomew and Forbes (1979) have
described some requirement forecasting techniques.
STATISTICAL TECHNIQUES FOR PERSONNEL

DEMAND FORECASTING
In this section we consider some models for manpower requirement fore-

casting based on statistical techniques. We distinguish the following types
(see Van Winkel 1979):
Models based on extrapolation of manpower trends (univariate models),

Models relating manpower requirement to organizational and environ-
mental circumstances (multivariate models).
These types of models will be discussed below in some detail. A model is

defined here as a mechanism that determines the manpower requirement.
Models based on extrapolation of manpower trends

(univariate models)
The models discussed here reflect only the evolution of manpower require-
ments and do not consider explicitly the influence of various factors on the
manpower requirement. These models can be divided into two groups:
40 CHAPTER 2
1. Models that include a deterministic trend,

2. Models that include a stochastic trend.
Denote by M t the manpower requirement at time t. It is supposed that now

we live at time 0 and that manpower requirements M t for t = -k,
-k + 1, ...... ,-1,0 are given.
Models that include a determi1'!istic trend. In these models the manpower

requirement forecast for a specific category of personnel is found by extrap-
olating the relationship between manpower requirement and time. As an
example we consider a linear model-that is, we suppose that M t can be
written in the following form:
Mt =a+ b . t + ft, (2.1)
where a and b are constant but as yet unknown; a and b have to be esti-
mated from the past. The ft are uncorrelated random variables with zero
means.
Estimates of a and b c~n be obtained by using the weighted regression
criterion-that is, {j and b are chosen as estimates if they minimize
o
t =E_it(a + b· t - Mt)2, (2.2)
where 0 < {3 ::;: 1. By using the discount factor {3 < 1, one indicates that re-
cent realizations of manpower requirement are more relevant than previous
realizations.
Now the predictor for the requirement at time t becomes (see figure 2.1):
NIt = a+ b· t. (2.3)
The exponential model has the form:
Mt = a . eb · t + ft, (2.4)
where a and b again are obtained by smoothing of past data and ft has the
characteristics mentioned above.
The predictor for the manpower requirement at time t is then:
(2.5)
where {j and b are the estimated values of the coefficients.
Linear and nonlinear models are described extensively in Goldfeld and
Quandt (1972).
An essential problem concerning the application of this type of model is,
How can the manpower requirement in the past be known? We will meet
manpower
requirement
time
o
FIGURE 2.1. The linear model for manpower requirement fore-
casting. In the model described in the text and in this figure, a
linear relationship between manpower requirement and time has
been assumed. Fitting of the data may also be based on other rela-
tions-for instance, on an exponential growth curve.
this question in several other types of statistical models. One possible value
for the manpower requirement in the past is the manpower size in the past,
possibly adapted for existing vacancies and/or the need for more personnel
as it was felt by management.
The extrapolations for this type of model are useful only for short- or
medium-term forecasts, since the modeled type of trend may be valid only
for a small number of time periods. Exactly for this reason the weighted re-
gression criterion is used. If the annual growth rate has been high in recent
years and one assumes that this trend will hold for a long time period, then
the manpower requirement estimates would become impossibly high. An
example of this effect has been given by Bartholomew, Hopes, and Smith
(1976). One should also notice that cyclical effects may occur in the histor-
ical data corresponding to a trade cycle. In fact, the cyclical effect is a good
example of a deterministic trend. An example of this type of extrapolation
method is given by Rowntree and Stewart (1976) who investigated the man-
power requirement for the four types of work of the Land Registry of the
British Civil Service.
Models that include a stochastic trend. We will consider here autoregres-

sive models as examples of a model that includes a stochastic trend. In such
a model the manpower requirement is-apart from noise effects-explicitly
determined by past values of the manpower requirement.
42 CHAPTER 2
An autoregressive model of order k has the following form:

Mt = (XI • Mt- I + (X2 • M t - 2 + ... + (Xk • M t - k + Et, (2.6)
where k denotes the number of past periods that directly influence the man-
power requirement. The coefficients (XI, (X2, ••• ,(Xk denote the weights for
the historical values of the manpower requirement. These weights are esti-
mated by fitting to past data. The error term is represented by Et, which has
the usual characteristics:
IE(E t ) = 0 and IE(Et, Et - c) 1a

= O2
f
forC
forC
i= 0,
= O.
(2.7)
IE(Et)indicates the statistical expectation of Et.

Since the predictors do not have a simple form, we will not give them
here, but we refer to Van Winkel (1979). This type of forecasting method
has been described extensively in Box and Jenkins (1970) and in Brown
(1963). Further explanations concerning the autoregressive model for man-
power requirement forecasting are given in Bartholomew and Forbes
(1979). Applications of this type of model for a firm have been given by
Young and Vassiliou (1974) and by Cameron and Nash (1974).
Discussion of univariate models
In the univariate models described above, the value of manpower require-

ment is supposed to depend only on past values of the manpower require-
ment. Thus the model assumes that an upward trend in the past will con-
tinue. This is also the case if management expects that in subsequent years
the organization's activities will decrease and labor productivity will in-
crease. Formulated in a different way, it is unsatisfactory that management
policies and the opinions of management with respect to future develop-
ments and influences from the outside are not considered in these models.
This implies that these models may not be appropriate for manpower re-
quirement forecasting since it is not likely that management will accept
methods that do not reflect its opinions. The role that these models can play
in manpower requirement forecasting may be considered, however, as pre-
senting an instrument for stimulating discussion with respect to the evolu-
tion of the manpower requirement.
Note also that if different types of manpower are required in a fixed pro-
portion, then it is sufficient to have a single model for requirement forecast-
ing; otherwise, for each type of manpower a different model is required.
Application of these models may be limited by the lack of data concern-
ing manpower requirement in the past. Usually the total manpower size of
past years will be known, but the question remains whether this manpower
size reflects the real manpower requirement or not.
In our view, the models may be useful only if a rough estimate of the
total manpower requirement of a (sub-) organization is required. If detailed
estimates concerning smaller units of the organization (or for different
types of personnel) are demanded, then it may be better to investigate which
factors determine the manpower requirement (see the discussion below). In
any case, the forecasts obtained with univariate models have to be com-
pared with the views of management and adapted afterward.
Models relating manpower requirement to organizational and

environmental circumstances (multivariate models)
Here we consider the models for manpower requirement forecasting that ex-
plicitly formulate the manpower requirement as a function of various vari-
ables, such as the organization's activities, technology, and so on. We dis-
tinguish the following types of models:
1.A simple model based on the current value of the manpower require-
ment and the expected change in output level,
2. A more sophisticated model based on the current and past values of
the manpower requirement and the changes in output levels,
3. Regression models,
4. Economic models.
The first two model types use time-dependent coefficients; the last two
model types use time-independent coefficients. For the first two models,
one uses the output levels of the organization and forecasts for these output
levels.
Define:
Yt == output level at time t (expressed in number of products or ser-

vices);
Yt == forecast at time 0 for the output level at time t.
A simple model based on the current value oj the manpower requirement

and the expected change in output level. In the model, the manpower re-
quirement is assumed to be proportional to the output level (that is, the
number of products or services delivered by the organization):
Yt
M t =Mo . - . (2.8)
Yo
44 CHAPTER 2
The predictor for the manpower requirement is then:
M- t =Mo . -
1't . (2.9)
Yo
Some remarks have to be made on this model. First, the actual man-
power requirement has to be estimated. Second, the output level has to be
forecasted. Because this is the only factor considered for the manpower re-
quirement, these forecasts have to be relatively precise. Third, it is assumed
that economies of scale-which means that an increase in the output level
leads to less than proportional increase in the manpower requirement-will
not occur. Notice that this model can be applied to every type of product or
service if the actual manpower requirement for every product or service as
well as the expected change in output level can be determined. The model
has been tested for the British engineering industry by Bosworth, Evans,
and Lindley (1974).
A more sophisticated model based on the current and past values oj the
manpower requirements and the changes in output levels. This model can
be formulated as:
M t =Mo- · Y t + (Mo
--- M_-1). Yt • (2.10)
Yo Yo Y- 1
This model, which has been developed by Ghosh (1960), can be considered
as an extension of the first model we described. The changes in labor pro-
ductivity in this model are included in the forecasts of the manpower re-
quirement. The predictor then becomes:
M- t = -
Mo. Y- t + (Mo
- - M_
- -1) . Y- t • (2.11)
Yo Yo Y- 1
As we already mentioned with regard to formula (2.9), an important
problem is the estimation of future output levels. This is also true for for-
mula (2.11). Therefore, we will consider this question somewhat further
here.
Future output levels may be forecasted by applying forecasting tech-
niques such as the methods described in our discussion of univariate mod-
els. However, if it is relevant to consider the influence of such factors as
technology changes on the relation of output level and manpower require-
ment, then a more detailed analysis is required. Often such technology
changes can be predicted, since there exists a time lag between the introduc-
tion of a new technology and its adoption. Together with the growth curve
of a product or service (figure 2.2), the data with respect to the impact of
technology changes can give important information for the output levd esti-
cumulative
sales
---------------
o~~~------------------------~ time
-I~
~troduc-~~age ofgrowth stage of
tory stage rapid saturation
FIGURE 2.2. Growth curve showing the life cycle of a product or service.
mates and for the relation of manpower requirement and output level for a
specific group of products and service. Purkiss (1974) applied this method
of output level estimation for the British Steel Industry.
Regression models. In manpower requirement regression models, a rela-

tionship is formulated for the manpower requirement and a number of vari-
ables that are assumed to determine the personnel demand. If the man-
power requirement depends on q variables, which depend on time, then we
obtain the following model:
(2.12)
where Xlt, X 2t , ••• , X qt are the values of the considered explanatory vari-
ables at time t, f:t indicates the error with the usual characteristics (formula
[2.7]), and ao, aJ, ... , a q are the coefficients for the constant term and
the explanatory variables, respectively.
The coefficients can be estimated now from the application of formula
(2.12) on historical data-that is, on the realizations of M I , Xlt, X 2t , ••• ,
X qt for t = -k, -k + 1, ... , O. If the values of the explanatory variables
can be forecasted, then the future manpower requirement can be forecasted
by:
NIt = ao + al . XIt + a2 • X2t + ... + a q • Xql , (2.13)
where Xit , is the forecast of variable Xi for time t at time O. Notice that the
problem of forecasting the explanatory variables XJ, X 2 , ••• , Xq arises.
46 CHAPTER 2
An application of regression models for manpower requirement forecast-

ing has been given by Drui (1963). He determined the manpower require-
ment for the finance department of a firm. The explanatory variables he
used were the total company employment, the total numbers of direct per-
sonnel, sales, and the number of processed reports.
Lapp and Thompson (1977) forecasted the demand for engineers in
Ontario with the gross national product level as the only explanatory vari-
able.
One problem deals with the interpretation of the resulting equation. The
value and even the sign of the regression coefficients may be affected if
some relevant explanatory variable is omitted.
These problems are discussed in detail in Bartholomew and Forbes (1979)
and in Draper and Smith (1966).
Notice that (2.12) might also contain lagged variables.
Economic models. Economic models for manpower requirement forecast-

ing can be based on the specific form of the production function. The out-
put level of the production process is supposed to be determined by the in-
put levels of labor and capital according to the Cobb-Douglas production
function (see Klein 1962):
(2.14)
where c is a constant, K t is the amount of capital used in period t (the value
of the capital stock), Ut is assumed to be a log-normally distributed error
term, P t is the output level expressed in value added, and M t is the amount
of labor used in period t.
The variables a and (3 express the output elasticities with respect to labor
and capital-that is, the marginal increase of the output level because of a
small amount of extra input of labor or capital, respectively. It generally
holds that lal ~ 1 and 1(31 ~ 1 and, in the case of complementarity of labor
and capital, a + (3 = 1.
Taking logarithms of formula (2.14), we obtain:
log P t = log c + a log M t + (3 log K t + log Ut , (2.15)
which is an ordinary regression equation.
A modified version of formula (2.15) includes the impact of technology
improvements by a rate e yt on the output level. In that case the relationship
for the manpower requirement becomes (see formula [2.15]):
-1 l' 1 (3 1
10gMt = -log c - -. t + -log P t - -log K t - -logUt • (2.16)
a a a a a
The predictor for the manpower requirement at time 0 for time t follows
from:
~ 1 'Y 1 ~ {3 ~
log M t = - - log c - - . t + - log P t - - log K t • (2.17)
a a a a
Application of these models implies the necessity of forecasting the output
level and capital stock level.
In formula (2.16) one may also include other variables, such as the num-
bers of hours actually worked or a variable expressing the capital utiliza-
tion. Estimations of such relationships are given by Fair (1969) and by
Wabe (1974), who considers the British engineering industry.
Discussion of multivariate models
Here we discuss the following aspects of multivariate models:
1. Reality of the model,

2. Application to numbers of employees,
3. Forecasting of variables,
4. Estimation of manpower requirement in the past.
Reality of the model. In our view, the multivariate models are more useful
for manpower requirement forecasting than the univariate models. The fact
that multivariate models can include those factors that determine the man-
power requirement underlines the reality of these models. Naturally it is
essential to use the right explanatory variables and also a sufficient number
of variables in order to forecast the personnel requirement accurately. The
usefulness of these models is also illustrated by the possibility of incorporat-
ing management policies with respect to the future activities of the organiza-
tion (output levels) and, for instance, automation decisions.
Application to numbers of employees. So far we have considered quanti-

tative manpower requirement forecasting. The described models have to be
applied to numbers of employees. It should be remarked, however, that the
personnel requirement is investigated for groups of functions with similar
characteristics (required qualifications, types of work, and so on). By ap-
plying the models mentioned above to such groups of functions, it is possi-
ble to consider the forecast of the required number of employees with the
required qualifications. It should be remarked that it may be difficult to
48 CHAPTER 2
determine for all groups of functions which variables influence the man-
power requirement. In the multivariate models, the relationship between
personnel demand and, for instance, output level or used amount of capital
will not always be known. For large groups of direct personnel or for indi-
rect personnel with activities connected with the output level (such as mar-
keting employees), the multivariate models may be particularly useful.
However, for small groups of employees (or functions) or for certain
categories of indirect personnel, such as department heads, possible changes
in organizational structure will have consequences for the manpower re-
quirement and also for many other variables. Thus the application of
mathematical models will be less useful then, and there will be a greater
need for the opinions of management for manpower requirement forecast-
ing.
Forecasting oj explanatory variables. For all multivariate models, fore-

casts of the explanatory variables have to be available (except, of course, if
lagged variables with a sufficiently large time lag are used). The extent to
which such forecasts can be made will depend heavily on the type of orga-
nization and on the personnel group considered. Estimating the output lev-
els may be relatively easy for a stable organization, but it is difficult for an
organization for which the development of the market cannot be fore-
casted. Technology changes have a great impact on personnel demand.
However, sometimes it is difficult to forecast these changes as weB as to
determine their impact. Probably the forecasting of the explanatory vari-
ables is one of the important problems in the application of multivariate
models.
Estimation oj manpower requirement in the past. In many organizations

there is a lack of historical data with respect to the manpower requirement
level in the past and the values of the explanatory variables in the past. The
question whether the manpower distribution in the past can be used as an
estimate for the manpower requirement has already been discussed. In our
view, the knowledge and views of management are necessary for the deter-
mination of past manpower demand levels.
SUBJECTIVE METHODS FOR MANPOWER

REQUIREMENT FORECASTING
Subjective methods are those forecasting techniques that directly use the
experience and judgments of experts. The application of subjective methods
for manpower requirement forecasting may be useful not only to estimate

directly the future manpower requirement, but also to forecast explanatory
variables by collecting the views of management on the organization's
activities in the future (the types and numbers of output), technology
changes, organization structure changes, and the like. We distinguish the
following methods:
1. Collecting of management opinions,

2. Delphi method,
3. Work-study.
Collecting of management opinions
A fast and relatively simple way to obtain forecasts concerning manpower

requirements is to ask management which developments it expects in the
future. Management is best aware of changes in the workload, possible
automation projects, changes in organizational structure, and so on. It can
translate these developments into required numbers and types of employees.
These forecasts can be made on top-management level for the total orga-
nization or on the level of division managers for their departments. In the
latter case, if the figures have been given without commitment, the forecasts
might be overestimated in order to ensure a certain manpower strength in
the future (wishful thinking). On the other hand, the division managers usu-
ally know very well the specific qualities of their departments and the ex-
pected developments concerning personal demands, which makes the exer-
cise recommendable. A possible disadvantage of this procedure is the time
managers have to spend on it.
More information on this method for manpower requirement forecasting
is given in Lawrence (1975).
Delphi method
The delphi method, originally developed by the Rand Corporation in the

late 1940s (see, for instance, Gordon and Helmer 1964), consists of the col-
lection of experts' information in a structured way. The delphi method is a
special procedure for obtaining management opinions. A research team for-
mulates some questions concerning future development with respect to a
certain subject. This subject is, in our case, the manpower requirement. The
questions may concern, for instance, the expected evolution of labor pro-
50 CHAPTER 2
ductivity, technology changes, output levels, and the like. The participants
in this procedure answer the questionnaire anonymously: each participant
knows neither the other participants nor their answers.
In every following round, each participant obtains as information the
answers given by the group in the previous round, including the most im-
portant arguments as well as the response he gave himself. This procedure is
continued until the answers agree. It is expected that convergence of opin-
ion occurs if every expert is confronted with the opinions of his colleagues.
In summary, the main characteristics of this method are:
Anonymity of the participants to prevent domination by some experts,

Feedback of information to increase the quality of the responses.
An advantage of the delphi technique is that the experts do not need to be

present at the same time and they can study the problem in detail. On the
other hand, participants should not take too much time to answer the ques-
tions, which must be formulated clearly, and the number of participants
should not be too large (twenty to thirty persons). The number of rounds
until agreement is achieved will usually be two or three.
More information on this technique is given by Milkovitch, Annoni, and
Mahoney (1972), who applied this technique for manpower requirement
forecasting.
Work-study method
A very detailed approach to manpower requirement estimation is the work-

study technique. When applying this technique, one considers the different
functions in relation to the organization's activities plan.
The background for the use of this method for manpower requirement
forecasting is the possibility that the contents of some functions have to be
changed if the output levels change, if technology changes, and so on. It
should be remarked that we consider work-study for groups of functions
and not for functions individually. In fact, work-study is a very detailed
approach that requires a thorough knowledge of the production process (or
the process of service rendering) and of the organizational structure. Also,
the expected output level has to be given. Based on work-study principles,
standards may be developed for the numbers and qualifications of employ-
ees that are necessary to perform the various tasks. Of course, these stan-
dards have to be tested continually on their up-to-dateness with respect to
the efficiency and performance of the organization. The development of
standards and the grouping of tasks depend partly on human judgment, and
for this reason work-study is considered a subjective method. However, its
subjectivity is less obvious than in the first two methods we discussed, since
technical norms will also play an important role here.
Halpern (1977) describes an application of work-study techniques for
manpower requirement forecasting with the British Civil Service.
Discussion of subjective methods
The methods just considered may have some advantages beyond the models
described at the beginning of the chapter. In particular, the fact that man-
agement is directly involved in the manpower requirement forecasting pro-
cedure, instead of confronted with relatively untransparent models, might
be considered advantageous. Subjective methods are very useful if the
qualitative demand for manpower is investigated. Work-study techniques
can be applied, in particular, to small groups of functions, since these meth-
ods are much more detailed.
The application of subjective methods can be executed in several ways.
With these methods one may estimate, for instance, the expected change of
technology, the expected change of the future workload, or the expected
organizational change. Also combinations of these types of changes can be
investigated. Combinations of statistical and sUbjective methods are also
possible-for example, the application of regression analysis where the ex-
pected values of the explanatory variables are obtained by means of collect-
ing management opinions. Or extrapolation of the workloads and applica-
tions of work-study can be combined to determine standards for certain
activities.
Often, alternative scenarios for future developments are used. In that
case, alternatives are formulated on the basic scenario. This may result, for
instance, in scenarios based on upper estimates of variables and in scenarios
based on lower estimates. Thus different views of management can be pro-
jected in order to stimulate discussion.
THE PRACTICE OF MANPOWER REQUIREMENT

FORECASTING
We have described some methods for manpower requirement forecasting

that may be useful in practice. However, it is not reported that most meth-
ods mentioned in literature are used regularly for manpower requirement
52 CHAPTER 2
forecasting. One gets the impression that the methods follow from research
by operations research staff members and that the results have not been
applied by the manpower planning or personnel department. Therefore, it is
interesting to investigate which methods are really used for manpower re-
quirement forecasting in an organization. Of course, a complete survey of
manpower requirement forecasting procedures and the degree of acceptance
of several methods in large organizations cannot be given here; but we will
discuss some aspects of manpower requirement forecasting in a number of
large Dutch organizations, both industries and public services. The results
have been obtained by interviews with personnel managers and members of
personnel or planning departments.
In fact, when we asked the personnel managers of large organizations
how manpower requirement estimates were obtained, the usual answer was,
This activity is not done at all, or at least not in a systematic and formal
way. One gets the impression that the organization runs as it is, and man-
power demand forecasting is not considered very important. However, it is
interesting to consider that personnel numbers and qualifications change in
every organization. The questions are now, On which arguments are the
decisions concerning changes in personnel demands built and which meth-
ods have been used to obtain the necessary information? Of course, the per-
sonnel group under scrutiny and the type of organization play an important
role in manpower requirement forecasting (remember the manpower plan-
ning typology of chapter 1). We will now consider some practical cases of
manpower demand forecasting.
Cases of manpower requirement forecasting
Case 1. In a defense organization in the Netherlands, a very detailed man-

power requirement estimation procedure is applied. In this organization a
function registration system is used in which all functions are denoted by
numbers indicating the activity group to which it belongs; for every activity
group, the department to which it belongs; and for every department, the
number of each function. Moreover, some extra information about the
function may be included. For all functions, also called positions, the corre-
sponding rank and a function name are attached. This registration system is
continually assessed and analyzed by means of a function analysis proce-
dure in which all functions are considered individually and the level and
numbers of tasks for that function are judged. This detailed function regis-
tration system plays an important role in manpower requirement estima-
tion.
In a rolling corporate planning procedure, every two years a new ten-year

plan for the entire organization is laid. Rolling planning means that after
each planning period a new plan is made for up-dated figures. In the mean-
time, the plan is adjusted because of the governmental budgets proposed
every year. The corporate plan indicates organization developments, the
moments of introduction of new weapons systems, and the years in which
obsolete equipment will be replaced. In this long-term plan, the personnel
planning and tp.e planning of any of the other resources (finance, equip-
ment, and materials) are linked together: besides the investments, the mate··
rials and personnel costs are also considered. The corporate plan has to
correspond to the organization's goals (political goals). These goals are
translated into an operational plan representing the tasks to be executed.
From this operational plan the materials plan, the personnel plan, and the
finance plan are deduced in such a way that interaction among these plans is
possible (feedback), which results in readjustments of the original plans
while satisfying the organizational goals. These adjustments may be neces-
sary, for instance, because of changes in the political situation, the technol-
ogy, the labor market, and, particularly, the available budgets.
Furthermore, manpower requirements depend on possible changes in the
organizational structure that result from frequent efficiency studies. How-
ever, changes in the organizational structure are usually a consequence of
changes in the levels and types of activities and of technological improve-
ments.
In summary, in the short and medium term, if technology is not changing
much, manpower requirements depend especially on the intended activities.
In the long term technological improvements also have to be considered.
Work-study techniques are often employed to determine standards for
the relationship between the number and quality of necessary employees
and the workload. The personnel consequences of the introduction of new
technologies (weapons systems, machinery, and the like) are usually rather
well known because of foreign experiences, particularly from the United
States.
Manpower requirement estimation is done for ten years in the future,
function by function, in this defense organization; it is necessary to obtain a
detailed matching of functions and employees. Because of the specificity of
this organization (and of its employees), this matching is very important
and may have consequences for the number and qualifications of recruits,
career policies, education and training programs, and allocation patterns.
A detailed matching is necessary since an military units have a fixed orga-
nizational structure. Personnel requirements can be forecasted reasonably
for a period of about ten years since the introduction of new weapons sys-
tems can be foreseen and takes some time.
54 CHAPTER 2
Case 2. The second case is the directorate for the local police forces in the
Netherlands. This public service has no function registration system, and
manpower requirements are not estimated in a formal and structural way.
As in the first case, the police officers have had special education and train-
ing and can be regarded as being externally specific in the sense of the typol-
ogy discussed in chapter 1. Though the technology is not changing much or
irregularly, manpower demand estimation is necessary because of the speci-
ficity of the personnel and the changes of the workload. In practice, how-
ever, the strength of every local police force depends mainly on the number
of inhabitants of the municipality and on the allowed strength of the entire
police force, which is determined by political priorities. This distribution
policy is not necessarily a good one, since the need for policemen may also
depend on other local factors, and it is questionable whether the local work-
loads depend directly on the numbers of inhabitants. The local workloads
do have some influence on the allowed strength of the police force, since
some municipalities obtained an additional number of policemen because of
the local situation. Such assignments, however, are not based on a thorough
investigation of the local situation but on signals coming from the local
police management.
The directorate has developed a number of general organization charts in
which a reasonable function distribution for a local police force is given.
Such organization charts, depending on the number of inhabitants, are con-
sidered as guidelines for the municipalities in building the organizational
structure.
For a few years now, planners have tried to estimate real personnel
demands. In the large cities a part of the workload is measured by the num-
ber of requests for police assistance. Of course, many other tasks have to be
performed, so this cannot be the only explanatory variable. Therefore,
other investigations are carried out in which the attempt is to translate the
organizational goals into tasks and, afterward, to find some relationship be-
tween the number of necessary policemen for every task and the service ren-
dered to the public (for instance, a relationship between the fraction of
solved offenses of certain types and the numbers and qualifications of po-
licemen whose task is to solve these offenses).
Further investigations are set up in order to achieve a better organiza-
tional structure. From the top level a work-study project (see "Project
Group Organization Structure" 1979) has begun in which the times police-
men spend on all kinds of activities have been summed up. Some centraliza-
tion is arising since several police forces, not belonging to the local police,
have been formed for special tasks. On the other hand, some decentraliza-
tion takes place within the forces of the big cities since the work of police-
men on the beats gets more attention.
In summary, on the top level, manpower requirement forecasting is not a

part of the decision process; but the importance of the subject has become
noticeable as some experiments have been started recently. On lower levels
in the big cities, the importance was noticed earlier, since problems had
arisen with respect to allowed numbers of policemen on one hand and in-
creasing demand for police help on the other hand. Investigations with re-
spect to manpower requirements in the big cities have resulted in several
internal reports. Moreover, it is possible to investigate the necessary qualifi-
cations for policemen by executing manpower requirement forecasting.
Case 3. As a third example of manpower requirement planning, we will

consider a division of a large multinational private company. The person-
nel department of this division is involved in personnel demand forecast-
ing for highly educated personnel, mainly employees with an education in
advanced technology or with an academic degree. This activity is done
for every unit in the organization (development group, mechanization
group, production group, and so on) and concerns about two thousand
employees. Every year a questionnaire is sent to the plant managers and
the department heads. The questions concern expected developments in
the next four years with respect to production size, technology, and orga-
nizational changes (such as centralization/decentralization and make or
buy decisions). The managers are asked to translate the answers into a
forecast of manpower requirements in their department for all categories
of functions corresponding to pay classes. Moreover, some questions are
posed with respect to the qualitative filling of the organization at this
moment (the distribution of employees according to age, experience, nec-
essary education in the future), the number of vacancies, and so on. This
procedure has existed for about two years, and its goal is to gain more
insight into possible future discrepancies between demand and supply for
personnel. The decisions based on this approach concern possible trans-
fers of employees to other divisions, necessary recruitments, promotions,
or decisions for training.
Manpower requirement forecasting in a formal way is not done for the
lower-level employees because recruitments are possible if necessary; career
paths are short; turnover fractions are relatively high, so that demand and
supply can be matched more easily; and the workload is relatively stable. As
we discussed in chapter 1, medium- and long-term manpower planning is
not important in such situations.
In fact, the management opinion procedure applied by this organization
division seems to work quite well. A possible development in the future may
consist of efforts to determine the relationship between workload and the
required numbers of employees.
56 CHAPTER 2
Discussion
Many organizations have a procedure for manpower demand estimation

corresponding to the outline of the last case above. However, this procedure
is not often a standard procedure. The management itself notices continu-
ally whether a shortage or a surplus of personnel will arise in the near
future. If there is an expected discrepancy, management will ask for actions
such as recruitment, transfer, promotion, training, and so on. Thus the
principle is applied that the manager who does not complain will probably
have the right numbers and types of employees. The lack of standard proce-
dures for manpower requireIl)ent forecasting is also illustrated by the lack
of reports on standard procedures. An exception is, for instance, the man-
power planning system applied by Hoesch (see Beth-Von der Warth and
Steinecke 1980).
A difference between the first and the last case discussed above is the
applied degree of detail. In the defense organization the manpower require-
ment estimation is executed function by function, while in the industrial
organization the manpower requirements are forecasted roughly for pay
classes. Also, the length of the planning horizon differs. In the defense
organization a horizon often years is chosen, whereas in the industrial orga-
nization only a four-year planning horizon is used. These differences mainly
proceed from the degree to which future developments with respect to the
activities program can be forecasted.
The cases we discussed indicate that manpower requirement forecasting
is important. However, its relevance depends on the type of organization
and on the personnel group considered. In the following section we will
roughly indicate for which groups of personnel the manpower requirement
forecasting is particularly important.
MANPOWER REQUIREMENT FORECASTING

PROCEDURE
Outline
1. As a first step, one should estimate the actual manpower require-

ment in order to get a good starting point. The actual manpower require-
ment may follow from the organization chart or the function registration
system. Thus the available numbers and types of functions are known.
Since the multicategory approach is applied, these functions have to be clus-
tered such that numbers of function groups result. Now for every function
group, the amount and types of actual activities should be described, as well
as the experience and qualifications necessary to perform these activities.
Usually the function description indicates the type of work for every func-
tion. For certain functions, the yearly workload (the numbers and types of
tasks to be executed) may be hard to determine. Particularly for some cate-
gories of indirect personnel, this estimation may be difficult.
It should be remarked that the clustering has to be executed in such a
way that manpower requirement and manpower availability estimates can
be compared with each other. Thus one must be able to identify the char-
acteristics of the functions with the characteristics of the employees; for
instance, function level with grade, required experience with grade senior-
ity, required qualifications with the training and development of the
employee.
2. In the second step, one compares the actual formal manpower re-
quirement, which has been obtained in step 1, with the actual manpower
availability. The comparison of actual manpower requirement and avail-
ability gives insight into the number of vacancies (or into the surplus of per-
sonnel). Also, possible discrepancies between the function requirements and
the qualifications of the employees can be found.
3. In the third step, one confronts the foregoing results with the opin-
ions of management (for instance, of the department heads). The confron-
tation of the results of the previous steps with the opinions of management
indicates whether the actual formal manpower requirement, as estimated in
the first step, agrees with the real personnel demand in the view of manage-
ment. Based on this comparison-which should take into account possible
disturbing influences by seasonal effects, the trade cycle, and so on-the
actual manpower requirement can be determined.
4. In the next step, the output levels for all activities have to be fore-
casted. The output level for each organizational activity can be forecasted
according to the medium-term organization plan. Since these plans are
approved by management, the forecasted output levels are, in fact, the pro-
duction (or service) targets for the medium term. If the organization plan is
not sufficiently detailed, extrapolation techniques may be helpful here.
S. In the fifth step, the output levels for all activities have to be trans-
lated into workload forecasts for each function group. The manpower re-
quirement forecasts have to be given for each function group. Thus the
forecasted output levels for the different activities have to be converted into
workload forecasts for the function groups. In these workloads, the sea-
sonal effects, economies of scale, or other disturbing factors have to be in-
corporated.
58 CHAPTER 2
6. In the next step, the workloads have to be adapted for possible orga-
nizational changes and overhead activities. Organizational changes and
overhead activities may lead to a change of the workload for several func-
tion groups. In this step, the opinion of management is again particularly
important since the determination of overhead activities will be rather diffi-
cult. Estimation of the level of activities is particularly relevant for those
departments that perform tasks depending in size on the number of direct
personnel (for example, the personnel department) or even on the amount
of products or services to be delivered (for example, the marketing depart-
ment). In the last case, the number of employees in the indirect depart-
ment-indirect in the sense that the products or services are not directly
handled-depends on the output level.
7. In the seventh step, the forecasted workload is translated into man-
power requirement forecasts. For the conversion of workload into man-
power requirement, standards must be available. Applications of work-
study techniques can give these standards. If the standards have been
adapted to technology improvements, then for direct personnel it is rela-
tively easy to forecast the manpower requirements. For the indirect person-
nel it is recommended to let management do the conversion since it is
difficult to develop standards for the relation of workload and personnel
demand for indirect personnel. The use of data with respect to the numbers
of indirect employees and the workload in the past for these employees may
stimulate discussion.
Discussion
The outline of a manpower requirement forecasting procedure is only a

rough outline. The application of this procedure and, in particular, the ne-
cessity of executing each step depend on the type of organization and the
personnel group considered. Organizations may be in different circum-
stances with respect to the importance of manpower planning. Sometimes
manpower planning has to be executed only roughly; sometimes a very
detailed approach is required. If manpower requirement forecasting is exe-
cuted because particular problems have to be considered, then some aspects
should have particular attention. For instance, if the introduction of large
automation projects is discussed, the impact of technology changes on the
required training and development of employees will be very important for
manpower requirement forecasting.
The importance of manpower requirement forecasting also depends on
the type of personnel. Usually the forecasting of personnel demand in the
long term for direct employees with a low degree of training is important for
the estimation of future salary costs and for the estimation of the total num-
ber of employees. The last-mentioned forecast may be important in negotia-
tions with the unions with respect to future employment levels and, for
example, in housing decisions. For direct personnel, a detailed considera-
tion is usually not needed. This cannot be said for direct employees with
long career paths or if a high degree of commitment exists.
The considerations above indicate that medium- and long-term man-
power demand estimations are particularly useful if matching problems in
the future between supply and demand may possibly occur. Indications for
such situations are, for example, sudden changes in the output level, auto-
mation projects, and transfer of production to other locations. Short-term
manpower requirement estimation should always be executed, since the
work has to be done and one has to foresee certain peaks of the workload.
The planning period that is considered, therefore, depends on the type of
product or service in question and on the organizational structure. These
aspects will not be discussed here further.
Finally, in our view, it is essential that management be involved in the
manpower requirement estimation procedure. It is insufficient to use only
statistical techniques. We believe that the results of these techniques may
help in stimulating discussion, but collecting the opinions of management is
at least as important.
3 MANPOWER AVAILABILITY
FORECASTING
In this chapter we consider mathematical models for forecasting the evolu-

tion of manpower distribution and for their possible use in manpower plan-
ning. Manpower availability forecasting is a somewhat confusing term,
since we will not only consider mathematical models with which manpower
availability can be forecasted, but we will also describe some mathematical
models that combine the forecasting of manpower availability and the
matching with manpower requirement. For all these models we use the term
manpower availability forecasting models.
The first mathematical models for medium- and long-term manpower
availability forecasting were given by Seal (1945) and Vajda (1947). Many
early papers on manpower availability forecasting consider, in particular,
the effect of wastage, which has a considerable impact on the evolution of
personnel distribution (see, for example, Si1cock 1954). The first applica-
tions of Markov models were described in the papers of Young and Almond
(1961) and Gani (1963). From that time on, many mathematical models
have been developed for manpower planning. Markov models, renewal
models (see Bartholomew 1973, 1976), and optimization models (see
Charnes, Cooper, and Niehaus 1978) have received the most attention. All
these models describe the evolution of manpower distribution. The impor-
60
MANPOWER A VAILABILITY FORECASTING 61
tant variables for this evolution are recruitment, withdrawal (wastage), and
internal transfer (promotion or another change of position).
MATHEMATICAL MODELS FOR MANPOWER

A VAILABILITY FORECASTING
In this section we describe some mathematical models for manpower avail-

ability forecasting, including the Markov model (and the longitudinal
model), the renewal model, and the optimization model. Notice that in
some types of models forecasting of availability and matching are com-
bined.
For all these models we assume that the categorization of employees (that
is, the choice of the categories in which the employees are classified) has
been given in advance. The problem of categorization will be discussed sep-
arately.
Markov models
The basic model. The object of a Markov model is to forecast the number
of employees in each of the categories at equidistant points in time
(I = 0,1,2, ... ). In a Markov model the number of employees who make a
transition from one category to another in a given period is assumed to be a
constant fraction of the size of the first-mentioned category at the beginning
of the period. These fractions are called Iransition fraclions. The actual
number of employees in all categories, the transition fractions, and the
future recruitment in all categories have to be given. The future number of
employees in all categories (the future manpower distribution) is forecasted.
Define:
ni(t) == the number of employees in category i at time I;

Pji == the transition fraction from category j to category i;
ri(/) == the number of recruits in category i in the time interval (t - 1,1].
Then we have:
K
n·(t) =
I
E n·(1 -
j=l}
1)· p.
jl
+ r(t)
"
for i,j = 1,2, ... ,K,
for I = 1,2, ... , (3.1)
where K is the number of categories.

62 CHAPTER 3
If there are employees leaving the organization from category i, the fol-
lowing formula is valid:
K
.E Pij < 1.
J =1
(3.2)
We can also write formula (3.1) in vector notation. Define:
that is, the row vector of the number of employees at time t;
that is, the row vector of the number of recruits in the time interval
(t-l,t];
P!! P!2 P!K
P == P2! P22 P2K
PK! PKK
that is, the matrix of transition fractions for the different categories. Thus a
different formulation of formula (3.1) is:
N(t) = N(t - 1) . P + R(t), for t = 1,2, .... (3.3)
Markov models are also called push models, since the transition fractions
are given and the future distribution of manpower over the categories
(called manpower distribution) is determined by the values of these transi-
tion fractions. Thus the employees are "pushed" through the manpower
system.
EXAMPLE 3.1. Consider a manpower system with three categories. The
transition matrix Pis:
123
1
P = 23 [ .~ :~ .~].
o 0 .6
The actual manpower distribution is given in table 3.1. Suppose we recruit
eighty employees in category 1 every year. Table 3.2 shows the manpower
distribution over the categories based on the Markov model.
MANPOWER AVAILABILITY FORECASTING 63
Table 3.1. Actual Manpower

Distribution over the Categories
for Example 3.1
Category
1 2 3 Total
t =0 140 100 60 300
In table 3.2 the manpower distribution is forecasted for eight years

ahead. The long-term distribution of employees is forecasted over the cate-
gories with the recruitment of eighty employees per year in category 1 and
with transition fractions as given in the matrix P. The first column for each
category (E) represents the number of new entrants, the second column (8)
indicates the number of employees who stay in the category, and the third
column (T) gives the total number of employees in the category. All num-
bers have been computed exactly and rounded off afterward to the nearest
integer values.
Markov models formulated in this way are deterministic models. If all
flows (recruitment, wastage, promotion, or internal transfer) are control-
lable and predictable, then it is not necessary to have a stochastic model.
However, a deterministic model is rather unnatural in situations where the
Table 3.2. Forecast of the Manpower Distribution for Example 3.1
Category
1 2 3
E S T E S T E S T Total
t=0 140 100 60 300
t= 1 80 + 84 = 164 42 + 40 = 82 30 + 36 = 66 312
t=2 80 + 98 = 178 49 + 33 = 82 25 + 40 = 64 324
t=3 80 + 107 = 187 54 + 33 = 86 25 + 39 = 63 336
t=4 80 + 112 = 192 56 + 35 = 91 26 + 38 = 64 347
t=5 80 + 115 = 195 85 + 36 = 94 27 + 38 = 65 354
t=6 80+117=197 59 + 38 = 96 29 + 39 = 67 360
t=7 80 + 118 = 198 59 + 39 = 98 29 + 40 = 69 365
t=8 80 + 119 = 199 59 + 39 = 99 29 + 42 = 71 369
00 80 + 120 = 200 60+40=100 30 + 45 = 75 375

64 CHAPTER 3
flows are to a certain extent unpredictable and uncontrollable. In most or-

ganizations, wastage is not exactly controllable or predictable. A realistic
stochastic model for wastage is a model where, to each category i, a wastage
probability Pio is assigned. Each person in category i has each year a proba-
bility Pio of leaving the organization. The expected number of employees
leaving from category i is then ni . PiO, where ni is the number of employees
in that category. So we obtain similar forecasting expressions as in formulas
(3.1) and (3.3), but now the forecasts are statistical expectations. If all
transitions (promotion, was~age, recruitment, and internal transfer) are
modeled in this way, we have a situation where the evolution of each indi-
vidual employee through the manpower system is described by a Markov
chain.
In fact, this purely stochastic situation and the purely deterministic situa-
tion are only extreme cases. It is also possible that some transitions are sto-
chastic (such as wastage), while other transitions are deterministic (such as
promotion).
Essential for a Markov model is that the number of employees who make
a transition from category i to category j in a given period is independent of
the evolution of the employees in previous years and depends only on the
number of employees in category i at the beginning of the period. Markov
models have been described in Bartholomew and Forbes (1979) and Young
and Almond (1961).
Steady-state distribution. Markov models can be used to forecast the

distribution of employees over the distinguished categories. The fore-
casted manpower distribution indicates, then, the consequences of wast-
age, promotion policies, and recruitment policies. If recruitment is con-
stant in time, then it is possible to calculate the stationary or steady-state
distribution. The stationary distribution is the long-term forecasted num-
ber of employees in all categories if recruitments, as well as promotion
and wastage fractions, are constant.
Define:
ni == the number of employees in category i in the steady-state situation;

ri == the recruitment in category i.
In the steady-state situation we have:

K
~
n·I
~ ~
= 1 n··
= j i.J J
p JI.. + r·I' for i = 1,2, ... ,K, (3.4)
MANPOWER A VAILABILlTY FORECASTING 65
where
Iii = lim ni(t),
(-->co
for i = 1, 2, ... ,K. (3.5)
The value of this steady-state distribution lies in the possibility of com-

paring the long-term manpower availability forecast with the long-term
manpower requirement forecast. In this way one may investigate whether
manpower policies based on medium-term forecasts are acceptable in the
long term. In chapter 4 we will discuss the stationary distribution for the
Markov model in greater detail.
Determination of transition probabilities. The way to determine the tran-

sition fractions (or transition probabilities, respectively) depends on the
situation in which the Markov model is used. In the purely deterministic
situation, it is simple to observe these fractions, since the transition frac-
tions are fixed and deterministic.
In the stochastic situation, however, the transition probabilities have to
be estimated. In order to obtain a first estimate of the transition probabili-
ties, one may use historical data. Suppose that data are available for the pe-
riods from t = - T until t = O.
Define:
mij(t) == the number of employees who made a transition from category i

to category j in the time interval (t - 1,t].
The transition probabilities Pi) can be estimated now by:

A 1:7 = - T+ 1 mi;(t)
for i = 1,2, ... ,K, (3.6)
Pij = i.o(=
.... 0 ( '
-T+lnit-1) for j = 0,1,2, ... ,K,
where PiO denotes the wastage probability from category i.

In formula (3.6) all previous years have equal weights. If one assumes,
for instance, that the number of transitions in recent years is more impor-
tant for the estimation of the transition probabilities, then one may intro-
duce weights into formula (3.6).
Usually T should not be chosen too large, since historical data from
many years ago may not be appropriate to obtain reliable transition proba-
bilities. Moreover, the quality of the forecast is not much improved by tak-
ing many years of historical data. In the discussion below we will consider
66 CHAPTER 3
the influence of the estimation of the transition probabilities on the quality

of the forecasts in some detail. From our experience, it appears to be suffi-
cient in general to take T = 3 or T = 4.
Another possibility for determining transition probabilities is the use of
estimated transition probabilities from similar organizations for similar
categories of personnel. It may also be useful to estimate the transition
probabilities for a given personnel group from the transitions of employees
on a higher level of aggregation within the organization-for instance, the
estimation of wastage probabilities of highly trained engineers based on the
wastage for all engineers in the organization.
Quality of the forecasts. Bartholomew (1975) describes the following

causes of prediction errors:
1. Statistical causes,
2. Estimation causes,
3. Specification causes.
Statistical causes of prediction errors occur in the stochastic situation.

Under the condition that all employees move independently through the
categories, we have the following formula for the covariance of the num-
bers in category i and j at time t, according to Bartholomew (1973, pp.
27-28):
K K
cov{ni(t), nit)} = kE
=, i
E
=, hi' Pij . cov{nk(t - 1), nf(t - I)}
K
+ k~ ,(oij . hi - Pki . hi) . IEnk(t - 1),
for i,j = 1,2, ... ,K,

for t = 1,2, ... , (3.7)
where
J1 iii = j;
oij = lO otherwise.
The covariances are equal to zero for t = O. The variance ~(t), which indi-
cates the quality of the forecast of the number of employees in category i at
time t, is now given by:
(3.8)
-
Particularly important in judging the quality of the forecast is the varia-

tion coefficient given by:
-.J Vi(t)
(3.9)
IEni(t)"
Under the assumption of independent behavior of all employees, the varia-
tion coefficients are relatively small-if the content of the categories is not
too small (see, for example, Wessels and Van Nunen 1976). In the deter-
ministic situation, where the transition fractions are fixed, the variances are
equal to zero. Reality will be between these extremes.
Estimation causes of prediction error may also occur in the stochastic
situation when estimating transition probabilities. In the situation of com-
plete independent behavior of the employees, the expected number of em-
ployees IEmij (1) going from category i to category j in the time interval (0,1]
is estimated by ni(O) . Pij. Thus the quadratic difference of this estimated
value with the theoretically best forecast is:
nr(O)· [Pij - Pij]2. (3.10)
If Pij is estimated according to historical data (see formula [3.6]), then the
expectation of this quadratic difference (given the numbers ni[t] and Pi) is
equal to
n·2(0) . Pij(1 - Pij) , (3.11)

, E~= -T+ Ini(t - 1)
since
o
t=
E
-T+ 1
n-(t -
I
1) • p ..
IJ
has a binomial distribution
B(t =
E
-T+ 1
ni(t - 1), pij).
Since the variance of mij(l) is equal to ni(O) . Pij(1 - Pij), the influence of
the estimation of the transition probability on the quality of the forecasts is
small if
o
t =
E
-T+ 1
n-(t -
I
1)
is substantially larger than ni(O). As we mentioned earlier, it is usually suf-

ficient to take T = 3 or T = 4. In the deterministic situation, there is no in-
fluence of the estimate on the quality of the forecasts.
68 CHAPTER 3
Specification causes of prediction errors may occur from the model

chosen. This cannot be tested in general. It is very important to define the
categories of personnel correctly, such that all relevant characteristics of
employees are included in the model. The definition of the categories will be
considered in some detail later in this chapter. The quality of the forecast
has been discussed in Bartholomew (1973), Van der Beek (1977 b), and
Wessels and Van Nunen (1976).
The longitudinal model. In Markov models the evolution of manpower

distribution over a number of categories is considered. These models are
called cross-sectional models. A specific type of Markov model is the longi-
tudinal model, in which the evolution of a group of employees entering the
system at the same time is observed. The recruits of each year are distrib-
uted over the different groups, which are called cohorts. For every cohort
different transition fractions are used. The usefulness of this type of model
lies in the fact that it is possible to analyze the situation in which the evolu-
tion of the employees depends on the type of cohort to which they belong.
In longitudinal models the cohorts enter the system and disappear from it
after some time. The number of employees in the various categories is then
the sum of the contributions of each of the cohorts to these categories.
Define:
C == the number of cohorts;

gc(t) == the number of employees of cohort c entering at time t;
ni(t) == the total number of employees in category i at time t;
Pic(U) == the fraction of employees entering cohort c at time t who are
counted in category i at time t + u;
T == the maximum number of periods an individual is counted in the
model.
The following data are given:

C,T,gc(t) and Pic(U), for all c = 1,2, ... ,C; i = 1,2, ... ,K; and
t = 1,2, ....
The following formula can be given:
7 C
ni(t) E
= u=O E Pic(U) . gc(t -
c=l
u),
for i = 1,2, ... ,K; t = 1,2, .... (3.12)
Obviously longitudinal models are equal to Markov models with an ex-
tended number of categories. If, in the Markov model, types of employees
are distinguished for which different transition fractions are valid, a longi-
MANPOWER A VAILABILITY FORECASTING 69
tudinal model is obtained. Therefore, we will not discuss longitudinal

models separately in the remainder of this book. Longitudinal models are
described extensively in Grinold and Marshall (1977).
Renewal models
The basic model. The object of a renewal model is to forecast the man-
power flows that are necessary to obtain given numbers of employees in aU
categories. As in Markov models, the situation is observed at equidistant
points in time. The actual manpower distribution, the desired manpower
distribution in the future, and wastage fractions have to be known. Promo-
tion flows and recruitment are determined by filling the vacancies. There-
fore, renewal models are also called pull models. Notice that in this type of
model the availability forecasting and the matching process are integrated.
We will consider renewal models only for the deterministic situation.
This means that wastage is also considered a deterministic flow. This is, in
general, a limitation of the model. With respect to the quality of the fore-
cast, causes of prediction errors can be distinguished that are similar to
those given in the discussion of the Markov model.
Consider as an example a strictly hierarchical manpower system, where
the filling of vacancies in all categories is possible only from the category
below or by recruitment in category 1 (figure 3.1). The flows from category
i to category j in the time interval (t - I,t] are indicated by mij(t), and re-
cruitment in category 1 in the time interval (t - I,t] is denoted by r,(t).
Define:
ni(t) == the desired number of employees in category i at time t;

ni(O) == the actual number of employees in category i at time t = 0;
mij(t) == the number of employees who make a transition from category i
to category j in the time interval (t - 1, t];
ri(t) == the number of recruits in category i in the time interval (t - I,t];
DX_l,C(t) ~o(t)
cat. cat. cat. cat.

1---91
,--K-_l~ K_l,K(t)L-K_~
FIGURE 3.1. A strictly hierarchical manpower system with K categories.

70 CHAPTER 3
where i = 1,2, ... ,K; j = 0,1,2, ... ,K; t = 1,2, ... , and miO(t) indi-
cates wastage from category i during the time interval (t - 1,t].
Now, mij(t) and rj(t) have to be computed for i,j, = 1,2, ... ,K and
t = 1,2, ... , while nj(t) is given for i = 1,2, ... ,K and t = 0,1,2,
.... Wastage follows from:
for i = 1,2, ... ,K,

for t = 1,2, ... , (3.13)
where Pio is the given wastage' fraction from category i (as in the Markov
model for the deterministic situation).
The promotion flow from category K - 1 to category K follows from
formula (3.14):
fort= 1,2, .... (3.14)
REMARK. Formula (3.14) is valid only if nK(t) - nK(t - 1) + mKO(t) ;:: 0,

since otherwise the desired category size would be exceeded. If the desired
category size has decreased more than the wastage flow, the following adap-
tation of formula (3.14) is necessary:
for t = 1,2, ....

(3.15)
In addition, formula (3.14) is valid only if mK-I,K(t):::; nK_I(t - 1) +
-mK _ 1,0(t), since otherwise more employees would be promoted from
category K - 1 than the number of employees in that category. Thus the
following adaptation of formula (3.14) may be given:
mK - I,K(t) = min{nK(t) - nK(t - 1) + mKO(t), nK _ I(t - 1)
-mK-I,O(t)}. (3.16)
A possible remedy would be the filling of a remaining shortage of em-

ployees in category K by the promotion of employees from category K - 2.
Integration of formulas (3.15) and (3.16) gives:
mK - I,K(t) = max{O, min[nK(t) - nK(t - 1) + mKO(t),nK _ l(t - 1)

- mK-I,O(t)]}. (3.17)
If mK _ I,K(t) is determined, then the transition flow from category K - 2

to category K - 1 is computed, and so on. Thus the following formula can
be given (where exceeding of the bounds as mentioned in the remark above
is excluded):
MANPOWER A VAILABIUTY FORECASTING 71
mdf) == n2(f) - n2(f - 1) + m2o(f) + m23(f),

'I (f) == nl (f) - nl (f - 1) + m!O(t) + mdt),
forf== 1,2,.... (3.18)
Renewal models are described in Bartholomew (1973) and Bartholomew
and Forbes (1979).
EXAMPLE 3.2. Consider the manpower system of example 3.1. The
actual manpower distribution is given (table 3.1), as well as the constant
wastage fractions for categories 1,2, and 3, which are 0.1,0.3, and 0.4, re-
spectively. One wants to obtain a given manpower distribution in the fol-
lowing eight years. The question is, How many recruits are needed and what
will be the numbers of promotions from each category? We will consider
two cases here:
1. The total manpower size increases by 3% each year, whereas the dis-
tribution of the employees over the categories remains unchanged
(we call this the dynamic case);
2. The total manpower size as well as the distribution of employees
over the categories does not change (we call this the static case).
In the dynamic case, application of formulas (3.13), (3.14), and (3.18)

for f == 1,2, ... ,8 gives the results presented in table 3.3. Thus the desired
Table 3.3. Forecast for Renewal Model (Dynamic Case) of Manpower Flows
Category
1 2 3
E S T E S T E S T Total
t=0 140 100 60 300
t=1 77 + 67 = 144 59 + 44 = 103 26 + 36 = 62 309

1= 2 79 + 69 = 149 61 + 46 = 106 27 + 37 = 64 318
t=3 82 + 71 = 153 62 + 47 = 109 27 + 38 = 66 328
t=4 84 + 73 = 158 64 + 48 = 113 28 + 39 = 68 338
1=5 87 + 76 = 162 66 + 50 = 116 29 + 41 = 70 348
t =6 89 + 78 = 167 68 + 51 = 119 30 + 42 = 72 358
t =7 92 + 80 = 172 70 + 53 = 123 31 + 43 = 74 369
t=8 95 + 83 = 177 72 + 55 = 127 32 + 44 = 76 380
72 CHAPTER 3
manpower distribution and the numbers of employees who leave the cate-
gories (wastage) determine the forecasted numbers of promotions and re-
cruitments.
Table 3.3 presents the forecast for the renewal model (dynamic case) of
the manpower flows for eight years ahead with constant wastage fractions.
The first column for each category (E) represents the number of new en-
trants, the second column (S) indicates the number of employees who stay
in the category, and the third column (T) gives the total number of em-
ployees. All numbers have been computed exactly and rounded off after-
ward to the nearest integer values. Recall that the desired category size in-
creases by 3070 a year.
In the static case, where the manpower distribution is constant and equal
to the actual level, we obtain a steady-state behavior of the system. The
number of promotions, wastage, and recruitment are constant, then, for
each year, starting from the given wastage fractions. The numbers of
promotions and recruitments as well as the manpower distribution are given
in table 3.4. The table shows the forecast for the renewal model (static case)
of the manpower flows. Wastage fractions and the desired manpower distri-
bution are assumed to be constant. The first column for each category (E)
represents the number of new entrants, the second column (8) indicates the
number of employees who stay in the category, and the third column (T)
gives the total number of employees.
Table 3.4 also demonstrates that constant recruitment of 68 people and
promotion of 54 and 24 employees from categories 1 and 2, respectively,
gives the stationary distribution. Thus if the transition matrix of the Mar-
kov model were changed and 68 people were recruited each year, the fore-
cast of the Markov model would give results (stationary distribution) simi-
lar to those shown in table 3.4.
Table 3.4. Forecast for Renewal Model (Static Case) of Manpower Flows
Category
J 2 3
E s T E s T E s T Total
t = 0 140 100 so ' 300
t =1 68 + 72 = 140 54 + 46 = 100 24 + 36 = 60 300

The change of the transition matrix yields:

2 3
1 [ .514 .386 0]
P = 23 0 .46 .24.
o 0 .6
The promotion fractions from categories 1 and 2 are 54/100 = .386 and
24/100 = .24, respectively.
Steady-state distribution. Renewal models can be used to forecast the

transitions of employees in order to obtain a given future manpower distri-
bution. If the wastage fractions are constant, the policy with respect
to promotion and recruitment that maintains a given manpower distribu-
tion can be computed easily. For this situation we have the following form-
ulas:
mK - I,K(t) = mKO(t),
mi,i + I (t) = mi + I,O(t) + mi + I,i + 2(t), for i = 1,2, ... ,K - 2,
rl (t) = mlO(t) + mu(t), (3.19)
where t = 1,2, ....
Recall from our remark on formula (3.14) that formulas (3.19) are valid
only if the bounds on the number of employees in each category are not ex-
ceeded. Otherwise, modifications of such formulas are necessary, for in-
stance, as suggested in the remark above.
Estimation of input data. The input data that have to be estimated con-
cern the wastage fractions and desired future manpower distribution. The
wastage fractions can be estimated in a way similar to the estimation of the
wastage probabilities PiO in the Markov model (formula [3.6]).
The estimation of the desired future manpower distribution can be exe-
cuted by using the forecasting techniques for manpower requirement de-
scribed in chapter 2.
Extension of the strictly hierarchical renewal model. In the preceding dis-

cussion we considered a renewal model in which the filling of vacancies in
all categories is possible only from the category below or by recruitment in
category 1. However, if vacancies can also be filled from another category
or by recruitment, then the number of promotions is not determined
uniquely by the desired manpower distribution. Consider, for instance, a
manpower system where external recruitment is also allowed in category 2
(figure 3.2).
74 CHAPTER 3
cat. ffi12 (t) cat.

2 ----
'ffi 23 (t)
cat.
K-l
cat.
FIGURE 3.2. A manpower system with recruitment in two categories. The nota-
tion corresponds to the notation in figure 3.1.
Now the relation for mdt) in formula (3.18) has to be replaced by:
(3.20)
In fact, this means that one has to give a rule to distribute the total inflow in
category 2 over the promotion flow from category 1 and the recruitment
flow in category 2. Possible rules are:
Recruit as many as possible from category 1;

Recruit from category 1 and from the external category in a fixed
proportion.
This will be discussed in greater detail later in this chapter.
Optimization models
The basic model. As we have seen, the use of Markov models can show
the impact of constant transition fractions on the evolution of manpower
distribution. In the discussion above we considered renewal models with
which the impact of a given manpower distribution on the numbers of tran-
sitions can be computed. Optimization models for manpower planning can
do both. These models combine the forecasting of manpower availability
and the matching with manpower requirement.
Many formulations of optimization models are possible. We will con-
sider here some examples that can be found (somewhat modified) in the
literature (see, for instance, Grinold 1978; Uebe et al. 1978). Consider as an
example a strictly hierarchical manpower system (see figure 3.2).
Define:
ni(t) == the number of employees in category i at time t;

ni(t) == the minimally required number of employees in category i at
time t;
n7(t) == the maximally allowed number of employees in category i at

time t;
Pio == the wastage fraction from category i;
Pij == the minimal promotion fraction from category i to category
j;
Pij == the maximal promotion fraction from category i to category
j;
mki (t) == the number of employees who make a transition from category
k to category f in the time interval (t - l,t];
d ki == the costs for an employee who makes a transition from
category k to f;
Ci == the yearly salary costs for an employee in category i;
for i,j, = 1,2, ... ,K; for k,f = 0,1,2, ... ,K; for t = 1,2, ... ,T. mkO(t)
indicates the wastage from category k in the period (t - l,t], whereas mOk(t)
indicates the recruitment in category k in this period. The planning horizon
is denoted by T.
Suppose ni(t), n;(t), ni(O), PiO, Pij, pij, dki , and Ci are given for
i,j = 1,2, ... ,K; for k,f = 0,1,2, ... ,K; for t = 1,2, ... ,T. The flows
mki (t) and the number of employees in all categories ni(t) have to be chosen
now such that a given criterion is optimal.
One possible objective is to minimize:
TIK K K )
t~l i~l Ci' ni(t) + k~O i~O dki . mki(t) . (3.21)
Restrictions with respect to the numbers of promotions are:

ni(t - 1) . Pij ::;; mij(t) ::;; ni(t - 1) . pij, for i,j = 1,2, ... ,K, (3.22)
for t = 1,2, ... ,T.
Restrictions with respect to the manpower distribution are:
for i,j = 1,2, ... ,K, (3.23)
for t = 1,2, ... ,T.
If (a) the conditions in formulas (3.22) and (3.23) hold, (b) the defining rela-
tions in formulas (3.14) and (3.15) are given, and (c) all variables are non-
negative, then minimization of formula (3.21) yields future flows and man-
power distribution.
In optimization models a finite horizon is usually chosen. One may also
compute the steady-state manpower distribution and the corresponding
numbers of employees who make a transition. We will not discuss this in
more detail.
76 CHAPTER 3
Since all flows are deterministic, the statistical causes of prediction errors
need not be discussed here. For the other causes of prediction errors we re-
fer to the earlier subsection "Quality of the Forecasts."
Optimization models are described in Grinold and Marshall (1977),
Smith (1970), and Vajda (1978).
Goal programming. A goal programming model is a specific type of opti-

mization model. In a goal programming model a number of goals are
established that the decisionmaker desires to achieve. Such goals are, for in-
stance, minimum total salary costs, avoiding overemployment, a certain
employment of minorities, and so on.
An example of a goal programming model is obtained if one wishes
-apart from the minimization of formula (3.21)-to minimize the devia-
tions of the promotion flows from certain desired promotion flows in order
to obtain a stable number of promotions. Then one has to add to the objec-
tive function in formula (3.21) the following term:
PI . E E E
T K K
f=li=lj=1
I m·{t) - n{t -
IJ I
1) . p ...
IJ
I (3.24)
If one also wants to minimize the deviations of the manpower distribu-

tion with some desired manpower distribution, the following term must be
added:
P2 . t~l i~1 I ni(t) - n~(t)I, (3.25)
where n~(t) == the desired number of employees in category i at time t. PI

and P2 are the weights assigned to each goal by the decisionmaker.
A solution to the programming problem may lead to an adaptation of the
priorities of the decisionmaker, whereafter the new goal programming
problem can be solved until acceptable values of the decision variables are
obtained. The technique of goal programming has been described, for in-
stance, in Ignizio (1976). Applications of goal programming for manpower
planning are given in Charnes, Cooper, and Niehaus (1978) and in Kahalas
and Gray (1976).
Aggregation. If categories are partly defined by characteristics like age,

length of service, and so on, the goals have to be formulated, in general, on
aggregated flows or aggregated categories. The number of people in a certain
grade may be important, but not the distribution of these employees over age
and length of service. In the same way, the distribution over age of the num-
bers of employees promoted from a certain grade is not important.
Thus if C = {cc} is the collection of all categories after aggregation

over age, then we obtain for formulas (3.24) and (3.25) the following form-
ulas:
t
t=I
EEI.E
C k
.E
E Cf J E Ck
1
mij(t)-.E ni(t-l)·PCfCkI,
1 E Cf
(3.26)
t E I.E
t = l e i ECf
ni(t)-n~£(t)I, (3.27)
where n~c (t) == the total desired number of employees in all categories of Cf,
and PCCCk == the desired transition fraction from aggregated category Cc to
aggregated category Ck.
Estimation of important data. As in renewal models, we assume that the

expected wastage will be realized in the future. Thus the described optimiza-
tion models presume a deterministic situation.
The input data that have to be estimated depend on the exact formula-
tion of the model. For the basic model, for instance, wastage fractions have
to be estimated, as well as the upper and lower bounds for the transition
fractions and for the numbers of employees in all categories. This will be
discussed now in some detail.
DISCUSSION' OF THE MODELS
To determine whether the models described above are appropriate for the
manpower planning approach we suggested in chapter 1, we will discuss the
following features of the models:
1. Definition of the categories-that is, which characteristics of em-

ployees can be included in the models?
2. Usefulness for the design of manpower policies-that is, can these
models show the impact of alternative manpower policies?
3. Computational aspects of the models.
Definition of the categories
In this discussion we will consider which characteristics of employees should

be included in the definition of categories and which models are appropriate
for such a category definition.
78 CHAPTER 3
The definition of the categories has to satisfy the following conditions:
1. The characteristics included in the models have to be such that the

requirement and the availability of personnel can be matched;
2. The wastage of personnel is forecasted adequately;
3. Alternative career policies can be designed.
We will now consider these conditions in some detail.

As we mentioned in chapter 1 with respect to multicategory models, the
degree of aggregation and thus the definition of the categories depend on
the organization level where the planning is executed. If management is in-
terested, for instance, in manpower planning for highly trained engineers,
relevant characteristics in manpower requirement may be the function level,
the training and development level, and the experience of the engineers.
These characteristics also have to be included, then, in the manpower avail-
ability forecasting model. In chapter 4 we will give some examples of rele-
vant characteristics for different aggregation levels.
The models should also offer the opportunity to include the characteris-
tics of employees that are correlated to wastage, such as age, length of ser-
vice (sometimes), or sex (sometimes). In order to obtain reliable forecasts
for the numbers of employees who will leave the organization, it may be
useful to distinguish retirements, being strictly age dependent, and other
reasons for withdrawal or wastage. Wastage for the other reasons will from
now on be called turnover.
A general characteristic of wastage patterns is that turnover decreases
with the age of the employee (except for high age groups because of illness),
with length of service (except for high length of service because of illness),
and increasing skill or salary. In addition, wastage is higher for women then
for men. More about this subject can be found in Bartholomew (1976),
Bowey (1969), and Tuggle (1978).
As mentioned in chapter 1, the design of alternative career policies has a
great impact both on the organization and on the individual employees. The
career opportunities of the employees occupy a special place in manpower
planning, as we discussed in our view on manpower planning (chapter 1).
Therefore, the model one would like to use for manpower availability fore-
casting has to be such that career policies can be made visible and the design
of alternative career policies is facilitated. Relevant characteristics of em-
ployees with respect to career prospects may be function level (grade) and
grade seniority, which indicates the number of years an employee has spent
in his or her current grade.
Let us now consider whether the different models are flexible with re-
spect to the category definition. In Markov models the categories can be
chosen such that several characteristics of employees are included in the
model. Such an extended Markov model may contain, for instance, the
characteristics of grade and grade seniority. In particular, career policies
can be represented very well in such a model, since career opportunities may
be expressed by the average grade seniority at which promotion occurs. Of
course, the Markov model is more appropriate for organizations with a high
degree of commitment than for other organizations. This can also be illus-
trated by the type of organizations where Markov models have been applied
(see chapters 4 and 6, this volume; Verhoeven, Wessels, and Wijngaard
1979).
Wastage is assumed to be a push flow in a Markov model, which is a real-
istic assumption. Moreover, the age of the employees may be included in the
model, which permits the separate forecasting of retirement and turnover.
It should be remarked that extension of the number of categories implies
that many transition fractions have to be given; but many of these are trivial
(for instance the yearly transitions to a higher age or grade seniority). Fur-
thermore, it is not necessary to estimate these fractions separately. One can
use a certain level of aggregation in the estimation, for instance, by sum-
marizing over the age classes. Examples of Markov models with extended
categories will be given in chapter 4 and have also been given by Bartholo-
mew and Forbes (1979).
In renewal models it may be difficult to include addition~l characteristics
of employees. If one desires to take into account the age of the employees,
one has to indicate in advance from which age classes a possible vacancy
will be filled. Also, if one wants to consider the grade seniority in order to
design alternative career policies, one has to indicate from which grade se-
niorities a possible vacancy will be filled. Thus renewal models are not very
appropriate in indicating the impact of career policies. In fact, the pull ele-
ment emphasizes the restrictions of the allowed manpower distribution. Re-
newal models may be particularly useful in organizations where the design
of alternative career policies is less important. Many applications of these
models can be found in the Civil Service in Great Britain. It should be re-
marked that the introduction of certain characteristics such as age and
grade seniority in a renewal model gives the combination of a pull and a
push element. On one hand, one has to indicate how vacancies have to be
filled (pull); on the other hand, the age and grade seniority of employees in-
creases each year by one (push). Examples of renewal models with extended
categories are given by Nash and Goddard (1978) and by Sherlock (1978).
80 CHAPTER 3
In optimization models it is also possible to include additional character-

istics, but the size of the models will then become very large. This is dis-
cussed in some detail later in this chapter. In our view, an extension of the
number of categories in optimization models is not always easy, since it may
not be possible to translate the preferences of management with respect to
detailed manpower flows in costs and restrictions of the model. For in-
stance, if the grade seniority of employees is included in the model, it may
be difficult to give minim~m and maximum promotion fractions for all
grade seniorities. Therefore, optimization models may be particularly use-
ful for organizations with strictly formulated goals and restrictions. Appli-
cations can be found especially in defense organizations in the United
States. There, the goals are mainly formulated in the numbers of employees
in certain aggregated categories and also in equal employment terms.
Usefulness for the design of manpower policies
Markov models are essentially based on career policies and are therefore
very applicable for comparing the impact of alternative career policies. Re-
newal models are essentially based on manpower requirements and there-
fore show the consequences of given requirements.
Both types of models can be useful for the matching process: in push
models one can vary the career policies until the computed availabilities
agree with the requirements; in pull models one can vary the requirements
until the resulting career policies are acceptable. So the choice for one type
of model is not a very fundamental one, since both types of models can be
used for the same purpose. The main point is that these models can show
the impact of alternative policies.
In principle, this is also the case for optimization models. By choosing
the right restrictions, the optimization model can be very close to either a
Markov model, if there are strict promotion restrictions, or to a renewal
model, if the allowed number of employees is strictly formulated. However,
it is very difficult to translate the preferences of management with respect to
manpower flows into costs and restrictions of a linear programming model.
This is especially true since costs of some aspects are essentially difficult to
compare (for example, the salaries and deviations in the objective function
of formula [3.21] combined with formula [3.24]) and since many costs are
essentially nonlinear (for example, deviations become important only if
they surpass some bound or if they occur together with some other feature).
Moreover, the number of situations that can arise is so large and the situa-
tions are so different that it is infeasible to check all these situations with re-
spect to preference.
It is difficult not only to construct a linear programming model, but also
to evaluate the results. The usual postoptimal analysis does not give suffi-
cient information about the sensitivity of the solution with respect to the
conditions and costs involved. Notice that the costs of deviations of the
goals are partly artificial.
It is quite possible, and not simply verifiable, that many costs are caused
by some combination of conditions that are in fact not strictly needed (for
another discussion of this aspect, see Wessels and Van Nunen 1976). If
goals and restrictions are often changed, the use of optimization models be-
comes very expensive.
Computational aspects
Here we will make some remarks on computational aspects (computation

time and storage capacity) of the manpower availability forecasting models.
For Markov and renewal models the computation times are usually rela-
tively low. Consider, for instance, a manpower system with five grades and
ten grade seniorities (see figure 3.3). A category is thus defined by grade and
grade seniority.
Thus for the Markov model the computation time depends on the num-
ber of categories and on the length of the planning horizon. A complete run
for this example with a planning horizon of ten years took only a few sec-
onds process time on a Burroughs B7700 computer. For the renewal model
similar considerations can be given, assuming that the manpower distribu-
tion over the grades has been given, as well as the distributions of the input
in each of the grades over the grade seniorities.
It should be remarked that computation time increases with the number
of categories and the number of transitions. If, in the example above, age is
included instead of grade seniority, the number of categories may be about
5 x 50 = 250. However, a reduction of the numbers of categories and tran-
sitions is often possible, since some combinations of grade and age will not
occur in practice. This will be discussed in greater detail in chapter 4.
Usually the solution of an optimization model takes relatively much
computation time. For the example of figure 3.3 and for an optimization
model (see formulas [3.21] through [3.23]), we have about nine hundred
variables and fourteen hundred contraints. The restrictions are then set on
the numbers of employees in each grade and on the promotion flows from
82 CHAPTER 3
B-B-B---- ----EJ ----+ grade seniority
FIGURE 3.3. Schematic representation of a manpower system with five grades and
ten grade seniorities in all grades. Wastage is not represented here. From all categories
in the first four grades, there are two flows out of the category (promotion and wast-
age) that have to be computed. The third transition out of a category is straightfor-
ward, since grade seniority increases by one each year. In category 50 there is only one
flow out of this category (wastage).
each category, while the planning horizon is ten years. Solution of this
problem takes about three minutes process time on a Burroughs B7700 com-
puter.
Similar conclusions can be drawn with respect to the required storage
capacity. In Markov models, it is important to store the transition matrix
efficiently. It is sufficient to store the elements that can be positive. In re-
newal models no transition matrix is necessary, but the manpower require-
ment and the priority rules for the filling of vacancies have to be stored. The
size of optimization models is relatively large, as we mentioned above.
Moreover, the optimization packages themselves require storage capacity.
RECOMMENDED MODELS FOR

MANPOWER PLANNING
Manpower planning models have to satisfy the conditions mentioned

above. Based on this discussion, we do not recommend optimization models
for medium- and long-term manpower planning. Simply said, it is not clear
in these models which judgment has been built in. Moreover, because of the
large numbers of variables and constraints, changing policies may lead to
many changes of parameter values.
In our view, the most appropriate approach for manpower planning is
mathematically a very simple one: present the impact of alternative poli-
cies to management, and let management do the evaluation. This can be
achieved by designing a manpower planning system that facilitates analysis
of policy changes and provides management with the forecasted impact of
these changes. We will discuss this in greater detail below.
Thus, in our view, Markov and renewal models are the most valuable
models for manpower planning. Of these two models we prefer a Markov
model as the basis for a manpower planning system. The Markov model fits
best the manpower planning approach we introduced in chapter 1, in which
three phases can be distinguished: (1) the forecasting of manpower require-
ment, (2) the forecasting of manpower availability, and (3) the matching of
manpower requirement and availability. The Markov model can be used for
pure manpower availability forecasting. Alternative parameter values for
recruitment and promotion lead to changes in manpower availability, which
may agree better with the manpower requirement. Desirable adaptations of
manpower requirement also become apparent in this way. Moreover, appli-
cation of Markov models permits the direct consideration of individual
prospects since the transitions are assumed to be push flows. Thus existing
career policies are embedded directly in the forecasting procedure. In re-
newal and optimization models, manpower availability forecasting and the
matching process interfere with each other and, particularly in renewal
models, the individual prospects of employees are not considered explicitly.
MANPOWER PLANNING SYSTEMS
In the discussion above we chose the Markov model as the preferred man-
power availability forecasting model. A manpower availability forecasting
model, however, is only part of a manpower planning system. Here we will
discuss which demands should be met by a manpower planning system.
Manpower planning systems have to support management in making the
right decisions concerning manpower policies. The underlying model has to
be understandable by management. Moreover, a manpower planning sys-
tem should help in the design of policies by presenting the forecasted impact
of alternative policies to management in a meaningful way. One is not in-
terested in the consequences of one particular set of input data. Manage-
ment must be able to change recruitment policies, career patterns, and so
84 CHAPTER 3
on, in an easy way and has to obtain the results of such changes immediately
in a conveniently arranged form. This means that an interactive computer
program is needed that can be used, for instance, via a typewriter terminal.
Alternative policies can be evaluated immediately, then, and one has the
opportunity to base changes in parameter values on previous results. For
practical reasons, interactive planning systems must have low computation
times, which again makes the application of optimization models unattrac-
tive.
The impact of a given policy has to be evaluated by experts (a personnel
manager or, for instance, a staff member of the personnel or planning divi-
sion). This means that manpower planning systems using a forecasting
model of the evolution of the manpower distribution have to be made such
that they can be used by these experts. Thus the codes for the operation of
the system (change of policies, calculation of the results, and the like) have
to be expressed in manpower planning terms. In our view, human assess-
ment of the alternatives is absolutely necessary for manpower planning sys-
tems.
In summary, a manpower planning system should meet the following
conditions:
It has to be based on a model in which the evolution of manpower avail-

ability is described adequately and in which the decision variables are
indicated clearly;
Input and output have to be organized in such a way that the system can
be used by experts for manpower planning rather than experts for
computer software;
It has to contain many options to easily implement alternative policies;
It has to present the results in a conveniently arranged form;
No criteria have to be built in for evaluation of the results, which is done
by management;
It has to be an interactive planning system.
Keen and Scott Morton (1978) describe the conditions that have to be satis-
fied by such decision support systems for other areas of application. In
chapter 4 we will describe the manpower planning system FORMASY,
which we claim satisfies the conditions mentioned above.
FORMASY: 4
An interactive manpower planning system
In chapter 3 we argued that Markov models can serve as a suitable basis for
a manpower planning system. When applying such models, it is possible to
evaluate the impact of alternative policies on the evolution of manpower
size and distribution by changing transition fractions and/or recruitment.
We also listed some practical conditions that a manpower planning system
should satisfy. In this chapter we present an interactive manpower planning
system based on a Markov model that satisfies these conditions. This plan-
ning system is called FORMASY, which stands for: FOrecasting and
Recruitment in a MAnpower SYstem.
FORMASY
The basis of FORMASY is a Markov model for the evolution of manpower

distribution. The categories in which employees are classified are defined by
four labels: g, q, e, and s. These labels can be applied with some imagina-
tion, but their primary purpose will be indicated below. Furthermore, the
age of an employee in the category (g,q,e,s) is considered. Thus the follow-
ing characteristics of employees are incorporated into the model:
85
86 CHAPTER 4
g is the grade or function level (g = 1,2, ... ,0);

q is a variable index (q = 1,2, ... ,Q);
e is a variable index (e = 1,2, ... ,E);
s indicates the grade seniority-that is, the number of years the employee
has already spent in his or her current grade (s = 1,2, ... ,S);
a is the age of the employee.
The variable indices q and e can be used, for instance, to indicate the em-
ployee's qualification or experience, level of training and development, po-
tential abilities, mobility, and so on. The variable a (age) is used only to pre-
dict retirement. So if age is also an explanatory factor for promotion or
turnover, then one of the variable indices should be reserved for age.
It should be remarked that the flexibility of FORMASY with respect to
category definition is a main advantage over existing manpower planning
systems such as MANPLAN (see Nash and Goddard 1978) or CAMP (see
Dudding and Piskor 1978); see also Huisjes, Van Tuyll van Serooskerken,
and Wijbenga (1977) and Siebelt, Taal, and Van Wijk (1977).
In the remainder of this chapter we use the terms category for a group of
employees defined by (g,q,e,s) and class for a group defined by (g,q,e).
Thus in every class several grade seniorities may occur.
In a model with the variables indicated above, an employee in category
(g,q,e,s) usually makes a transition to (g,q,e,s + 1). If his grade changes in
a certain year to g', then his next category is (g',q,e,I). Of course, every
year his age increases by one. Sometimes the variable indices may also
change. If e represents, for instance, the training level of the employee, then
an employee who has finished a training program but stays in his current
grade moves from category (g,q,e,s) to (g,q,e + 1,s + 1). This transition
structure makes it possible to construct efficient computer programs and
facilitates the input of the transition fractions, since only the possible transi-
tions have to be considered. We will discuss this aspect in some detail later
in this chapter (see also Verhoeven, Wessels, and Wijngaard 1979).
It has appeared in practice that a Markov model with a category defini-
tion as given above is rich enough to serve as the basis for a manpower plan-
ning system. On the other hand, such a system is not too complex with re-
spect to data collection and computational efficiency, as will be seen in sub-
sequent sections.
FORMASY was written originally in BEA (Burrough's Extended Algol),
which is a version of ALGOL 60, and has been implemented on the Bur-
roughs B7700 computer at Eindhoven University of Technology (see Ver-
hoeven 1977). Translations have been made, for instance, in FORTRAN,
FORMASY: AN INTERACTIVE MANPOWER PLANNING SYSTEM 87
ALGOL, and BASIC, and implementation is accomplished at this time on

several other computer systems in large Dutch organizations.
EXAMPLES OF FORMASY MODELS
In this section we will give some examples of manpower planning problems

within Dutch organizations. For every example, we indicate how the model
structure can be chosen and implemented in FORMASY.
Example A
In the Netherlands Public Works Office, the engineers of three different

training levels have their own grade system. For all three groups there are
five grades, after lumping together of some of the top grades in which only
a small number of employees can be found. This aggregation is allowed be-
cause employees in these top grades cannot be promoted to other grade sys-
tems and will leave the grades only by wastage.
For the two higher-level groups of engineers, there is no need for extra
indices. The groups are homogeneous with respect to training, and the expe-
rience of the employees is indicated properly by grade and grade seniority.
The lower-level group of engineers (called surveyors) is a less homoge-
neously trained group of employees; people in the same grade, have had dif-
ferent types of experience. The age of the employees is included in order to
predict retirements.
One is interested in career opportunities within these grade systems as
well as in the development of manpower supply and in matching this supply
with the manpower requirement.
We will describe now the models for the three groups of engineers.
1. For top- and middle-level engineers, we obtain a model with G = 5,
Q = 1, E = 1, and S = 10 and 13, respectively (see figures 4.1 and 4.2).
2. For the surveyors, four levels of training appear to be relevant. This
leads to a model with G = 5, Q = 4, E = 1, and S = 12.
Some lumping of categories is possible here, since the fact that older peo-
ple have reached a certain grade implies that their experience compensates
for a lack of theoretical knowledge. Thus career prospects are assumed to
be equal for experienced employees and highly trained employees. Exclud-
ing the class of former employees, this clustering leads to 13 x 12 categories
instead of 20 x 12 (see figure 4.3).
grade
1 2 3 4 5
grade seniority
10
FIGURE 4.1. The positions or categories for top-level engineers of example A.l
with the transition possibilities. Position 51 denotes the class of former employees.
1 2 3 4 5
_ _ _?) grade
FIGURE 4.2. A simplified version of figure 4.1, where grade seniority re-
finement has been suppressed and the class of former employees has been de-
leted.
level
of
training
4
GJ )
3 [2J )
2
0 )
1
GJ
1 2 3 4 5
~
grade
FIGURE 4.3. Classes for the surveyors of example A.2 and the transition pos-
sibilities. As in figure 4.2, grade seniority and the class of former employees
have been deleted.
This example will be discussed in greater detail in chapter 5 and has been
described extensively in Van der Beek (1977a), Van der Beek, Verhoeven,
and de Waij (1978), and Van der Beek, Verhoeven, and Wessels (1978).
Example B
1. In the Netherlands, the Ministry of the Interior has a directorate of

police for the local police forces. With respect to the future availability of
management, one is interested in the career of individual employees and in
the necessary recruitment for the medium and long term. The Markov
model in this example has essentially five grades (as in the foregoing exam-
ple, the top grade in the model aggregates some sparsely occupied actual top
grades), but the lowest grade is subdivided into three subgrades (see figure
4.4). In two of these subgrades recruitment is possible. The difficulty now is
that for the promotion to the next main grade, one counts the total number
of years spent in the lowest main grade. So the number of years spent in the
last subgrade has no influence. This can be accounted for in the model by
using one of the variable indices to indicate the subgrades. Grade seniority
then indicates the number of years spent in the main grade. Age is again
used only for predicting retirement.
90 CHAPTER 4
subgrade 2
r 1
1 2 3 4 5
--~) grade
FIGURE 4.4. The classes for members of the police management in example
B.l and their transition possibilities (the class of former members has been de-
leted).
In this way we obtain a model with G = 5; Q = 3; E = 1; and S = 10 (see

figure 4.4).
This example has been described in Van Meeteren (1978) and, roughly, in
Verhoeven (1979).
2. FORMASY has also been used on the local police force level for po-
licemen with lower rank. The characteristics in this model are grade, grade
seniority, and sex, since female police officers have much higher wastage
rates. One is especially interested here in career opportunities and in the
necessary recruitment. Moreover, one is interested in the consequences of
changes in the ratio of male and female recruits. More female recruits im-
prove the career possibilities of male employees because of the higher wast-
age rates for women.
In this model G = 5; Q = 2; E = 1; and S = 13 (see figure 4.5).
This application has been described by Van Meeteren (1979).
Example C
In a large industrial firm one is interested in the development of the total

manpower size, the necessary recruitment, and the future distribution of
employees over some main grades. These main grades define the function
levels of the employees. Promotion within this model is determined by
sex 2
1
1 0--~
1 2 3 4 5
--~) grade
FIGURE 4.5. The classes for policemen with lower rank, where, in the two low-
est grades, the sex characteristic has also been included.
grade and age whereas the wastage fractions depend on grade, age, and
length of service. It turns out that wastage fractions are high for short
length of service. Therefore, three levels of length of service have been in-
cluded in the model (0-1 year, 1-2 years, more than 2 years). For this situa-
tion, age and length of service can be used as variable indices, whereas grade
seniority has not been included in the model.
In this model G = 5; Q = 3; E = 49 (ages 16-64 years), and S = 1 (see
figure 4.6).
3
length of
service
r
2
2 3 4 5
_ _~) grade
FIGURE 4.6. The classes and transitions in the industrial firm of example C (now
the age and the class of former employees have been deleted). In this example class is
defined by grade and length of service.
92 CHAPTER 4
This example is discussed, for instance, in Verhoeven and Wijngaard

(1978).
Example D
In the Staff department of a large industrial firm, one wants to obtain a

given number of employees at certain professional levels and stable recruit-
ment policies. In the model ten grades are distinguished. Promotion is de-
termined mainly by grade seniority (8 = 10) and professional level. One dis-
tinguishes three professional levels, of which the highest is equivalent to a
completed university training (master's or doctor's degree). A person's pro-
fessionallevel can change when formal degrees are obtained or when one's
professional level is reestimated internally.
In the model obtained in this way, G = 10; Q = 3; E = 1; and 8 = 10 (see
figure 4.7).
In the situation demonstrated in this example, an even more refined
model has also been used that incorporates an age index. For this index we
used the second variable index. This index indicates the age of the em-
ployee-namely, e = 1 if the age is between 16 and 34, e = 2 if the age is
between 35 and 44, and e = 3 if the age is between 45 and 65. It will be clear
that in a Markov model such a rough age index cannot indicate the age
classes properly since one does not know exactly when the index has to be
professional level
1
1 2 3 4 5 6 789 10
~ grade
FIGURE 4.7. The classes and transition possibilities in example D (the class of
former employees has been deleted).
changed if the exact age has not been included. However, such a rough in-
dex can be of help if some promotion probabilities are influenced partly by
age. This example has been described in Kessels (1974) and in Esser (1975).
OPTIONS OF THE FORMASY SYSTEM
By using FORMASY, valuable information can be obtained for the design

of manpower policies. In this section we describe in detail the options of the
program system that make it possible to obtain this information. The types
of manpower planning problems for which these options can be used will be
discussed in chapter 5.
We will distinguish the following options:
1. Forecasts with respect to manpower distribution over the categories

(these options provide the forecasts of manpower distribution, the
statistical errors for these forecasts, the steady-state distribution of
employees, the age distributions, and the forecasts of salary costs),
2. Data with respect to average career schemes and career planning,
3. Simulation possibilities of the model (changes of transition frac-
tions, recruitment).
The tables and figures in this section have been taken from the applica-
tion for middle-level engineers (example A.I).
Forecast of manpower distribution
The forecasted manpower distribution over the categories is computed for

an arbitrary planning period chosen by the user. The following data are
needed for this forecast:
Actual manpower distribution over the various categories, including age,

Promotion and turnover fractions (assumed to be constant) for all cate-
gories,
Number of recruits and their distribution over the various categories for
each future year within the planning horizon.
The forecasting procedure for the number of employees in each class

when grade seniority, age, and retirement are not taken into account was
94 CHAPTER 4
described by formula (3.1). The introduction of grade seniority and age is

straightforward. All promotions take place to grade seniority one, and
every year-in the case of no promotion-the grade seniority of the em-
ployee increases by one. Each year, age increases by one until the retirement
age is reached. Later in this chapter we will describe the computation
method in more detail.
The forecasts are printed in the form of a table or a histogram, which can
be given for each of the included characteristics of employees (grade, quali-
fication, grade seniority, and so on). Examples are given in tables 4.1 and
4.2; a histogram is presented in table 4.3. In table 4.3 the forecasted number
of employees in each grade is reflected by two lines indicating the number of
the grade. Each line represents the total number of employees in the grade,
and one figure stands for ten employees. The scaling factor is chosen auto-
matically. Such a histogram can be printed for each year of the planning
period.
Statistical errors
Another option of FORMASY gives the statistical errors of the forecasts. In

chapter 3 we discussed the quality of the forecasts for Markov models. In
the stochastic situation, the standard deviations of the forecasted numbers
of employees in the grades give some information about the dispersion of
the forecasts. Only in the stochastic situation is it useful to compute the
standard deviations, since these are zero in the deterministic situation.
FORMASY can give the standard deviations for the stochastic situation.
However, recruitment is supposed to be deterministic. Table 4.4 i1l1dicates
the standard deviations of the forecasts of table 4.1 in integer numbers.
In fact, the computational effort to calculate these standard deviations is
relatively high when age and grade seniorities are included as characteris-
tics. However, one may omit the first term after the equation sign of for-
mula (3.7) to obtain an upper bound for the standard deviations. The con-
tribution of the covariances stems from the fact that it is possible to reach
category j from category i along different paths. But the numbers of em-
ployees following these different paths are always negatively correlated
since each employee can follow only one path. Thus, deleting the covari-
ances gives an upper bound of the standard deviation. This upper bound
can be computed much faster and requires less storage capacity, whereas the
deviations from the real covariances are usually relatively small. However,
we will not discuss this point further here.
Table 4.1. Actual Manpower Distribution over Grades and Forecasts for Planning Period
1978-1982
MANPOWER'DISTRIBUTION IN PLANNING PERI(¥)
••••••••••••••••••••••••••••••••••••••••
GRADE
YEAR * TA TAl THA THAi THABD * TOTAL
-----------------------------------------------------
1977 * 238 547 376 180 70 1411
*
• • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • i' . . . . . . . . . . . . . . .
1978 * 208 531 394 184 75 * 1392

1979 * 190 504 412 190 75 * 1371
1980 * 184 425 200 47~~79 * 1361
1981 * 182 440 436 210 81 * 1349
1982 * 182 410 448 222 86 * 1348
************************************************************
Table 4.2. Forecasted Manpower Distribution over Grades
and Grade Seniorities in 1980, Based on 1977 Data
Gf,. f.,EN"I 01·( I TY TA TAl THA THAt THABD
,.,:I. ** 6::) ~.)~) 64 34

:n
1:5
..~.
.:> **
**
::;4
:37
61.
64
68
66 3()
1. ~.)
:1.4
4
** 1.:5 70
::,9
4;! 1.6 4
.. '
f::'
** 7 44 2:'~ 7
l)
** :3 47 30
n
1:7; 4
'.7
n ** 4 ;.~ :~~ 1. ()
.,
7
** :1. H!
:3:5
;!.O 10 :~
** :1. 2!7j
9 {,
'1,0
'1 1 **
**
()
0
1 ::~
6
:1. ~:;
8
7
6
2
1
".-,
l:'
1,\ **
**
()
() ~~2
:~
-,
1 ".
::~
2
~~
1
Table 4.3. Histogram of Actual Number of Employees in 1977

YEAI~ 1977
----A----A----A----A----O-·---A----A----A----A----A----A----A
TA 111111111111111111111111 238
T,) 111111111111111111.111111 238
TA1. 2222222222222222222222222222222222222222222222222222222 547
TAl 2?222222222222~:)2222222222222222222222222222222222222 22 547
THA 33333~J3333333333333333333333333333333 376
'THO 333333333333333333333333333J3333333333 376
THAt 444444444444444444 180
THAl 444444444444444444 180
THAIHl ~:. :=-; ~~, ~.)~:; ~:'I~) 70
'1 HA!3lI ~5:7i ~:i ~'; ~, ~:i ~) 70
Table 4.4. Standard Deviations of Forecasted Numbers of

Employees
STANDARD DEVIATIONS OF THE FORECASTS (EXACT)
••••••••••••••••• + ••••••••••••• + ••••••••••••
GRADE
YEAR * TA TAl THA THAi THABD
* 77
1978 11 10 7 5
* 7
1979 13 1 ') 9 6
** 7
1980
1981
14
15
14
15
10
11
7
7
1982
* 7 15 15 12 8
--------------------------------------------
The standard deviations indicate that the considered number of employ-

ees may not be too small. In our experience, about thirty employees in each
grade is sufficient to give an indication of the evolution of manpower distri-
bution over the grades.
Steady-state distribution of employees
Another option of FORMASY is the computation of the expected number

of employees in each grade in the long run. This is called the steady-state
distribution, or stationary distribution. We have described this already in
chapter 3. The steady-state I;listribution shows the consequences of constant
policies with respect to promotion and recruitment on the very long term.
When age is included in the model, the steady-state distribution is equal to
the forecast for year T, where T is such that the recruits of the first year of
the planning period have already retired.
With FORMASY one may also compute the number of recruits neces-
sary to keep the total expected number of employees constant in the long
run under the same conditions of constant policies. First compute the conse-
quences of an arbitrary recruitment policy with the desired relation between
recruitment in the different classes. Then change the total forecasted num-
ber of employees while maintaining the relative distribution over the grades
by changing the recruitment, but maintain the relative distribution of re-
cruitment over the classes.
Define:
ii(O) == total number of employees at time zero;

ii(/) == total number of employees at time 1 under the arbitrary recruit-
ment policy r;
rj == yearly recruitment in class i.
Now, the constant recruitment rj needed to keep the total number of em-
ployees-as well as the distribution of employees over the classes-constant
in the long run can be computed by:
_ ii(Q)
rj = rj· ii(/) , (4.1)
where i = 1,2, . . . ,K and 1 indicates the time in which the steady-state

situation would be obtained.
An example of the steady-state distribution computed with FORMASY
is given in table 4.5. The first forecast shows the steady-state distribution
for the given constant recruitment, and the second forecast shows the distri-
Table 4.5. Steady-State Manpower Distribution over Grades
STEADY-STATE MANPOWER DISTRIBUTION FOR LAST CHOSEN RECRUITMENT
••••••••••••••••••••••••••••••••••••••••• t •••••••• + •••••••••••
YEAR 2020
GRADE
TA TAl THA THAl THABD TOTAL
TOTAL: 180 313 311 208 116 ** 1128
NECESSARY RECRUITMENT TO MAINTAIN THE TOTAL NUMBER OF EMPLOYEES

CLASS 1: 82
THE MANPOWER DISTRIBUTION WILL BE THEN
GRADE
TA TAl THA THAl THABD TOTAL
TOTAL: 227 394 392 263 147 ** 1422
bution for the changed constant recruitment of eighty-two employees. The

total forecasted number of employees is then close to the actual number of
employees in 1977.
Forecast of age distribution
One point of interest in manpower planning is the change in the age distri-
bution of the employees. Although probably no one can determine exactly
what an ideal age distribution will be for an organization, it is obvious that
a balanced age distribution is of great interest because of the continuity of
experience and knowledge within the organization. This continuity may be
secured by stable recruitment policies. In the preceding examples we men-
tioned the importance of this stability (example B). Some aspects of the age
distribution will be considered in detail in chapter 5.
With the help of FORMASY one can show the actual distribution of the
employees over the grades and ages (input data) and also the forecasted dis-
tribution. Moreover, for all grades the average ages and the corresponding
standard deviations may be computed. In table 4.6 this is illustrated. Each
figure in the table denotes the number of the grade, while the number of fig-
ures on each line shows the number of employees in the corresponding age
class. The scaling factor (here, 4) is chosen automatically: 1 indicates grade
TA, 2 = TAl, 3 = THA, 4 = THAI, and 5 = THABD.
Forecast of salary costs
With FORMASY it is possible to compute the actual and future salary costs
for all grades. These computations are based on a salary scheme where for
every class there is a minimum salary (for grade seniority 1), a constant
yearly increase with grade seniority, and a maximum salary. Such a salary
scheme is applied in Dutch public services and in several large firms.
Of course, other methods for the computation of salary costs can also be
included as long as the salary depends on the characteristics considered in
the model.
An example of the forecasts of salary costs is given in table 4.7.
Data with respect to career schemes
One further point in the use of Markov models is that several calculations
can be made that are very helpful for the evaluation and design of career
Table 4.6. Age Distribution of Employees, Average Ages, and Stan-
dard Deviations in Each Grade
YEAR : 1982
AGE *
63-64 *3445
61-62 *233445
59-60 *34455
57-58 *3344455
55-56 *233444555
53 ..·54 *23344455
51'··52 *33444455
49-50 *233334444455
47-48 *23333344444455
45-46 *2333344444455
43-44 *223333333444445
41-42 *222333333333344444
39-40 *22222233333333333334444
37-38 *2222222233333333333333333444
35-36 *222222222222222233333333333333333333344
33-34 *1222222222222222222333333333333333
31-32 *1122222222222222222333
29-30 *111111111222222222222222223
27-28 *11111111111111111112222222
25-26 *11111111111112
23-24 *11
1 FIGURE = 4 EMPLOYEES
AVERAGE AGES OF MANPOWER
GRADE AVERAGE DEVIATION

AGE
---------:------------ -----------
TA 27.6 2.2
TAl 34.6 5.9
THA 40.5 7.2
THAI 48.2 7.4
THABD 53.2 5.8
---------:------------:-----------
THE AVERAGE AGE OF TOTAL MANPOWER IS 39.0 YEARS
THE DEVIATION IS 9.5 YEARS

Table 4.7. Actual and Forecasted Salary Costs for Each Grade (in Million
Guilders)
SALARY COSTS IN MILLION GUILDERS

+ • • + • • • • • • • • • ++ • • • • • + • • • +.+.+~ ••
GRADE
YEAR * TA TAl THA THAi THABD * TOTAL
:l977
* 7.46 22.7E1
*
19.5::'i
.t ...................................................
6"'"
10.94 4.92 6::'i.65
** 5.77 19.97 21.97 12.06 5.49 ** 65.15

* 6. 22.24
197E1 20.39 1.1.1:3
~j2 -=" r)c:
\.J + ..:..~}
o
\.J + 0::'"
,.J~
c· "")"')
1979 21.19 21.30 1.1 .49
5.9~; ...J + ':.. ':H
5.::'i9 *
1980 65.26
1981
1982 ** ::'i.67 17.42 23.19 13.44 5.91. ** 65.64
5.70 :1.8.66 22.53 1.2.6E1 6::'i.16
-----------------------------------------------------
policies. FORMASY can present a number of data with respect to the aver-
age career scheme for the employees. Such data include, for instance, the
average grade seniority at which promotion occurs in each class (under the
condition that promotion will take place) and the total fraction of employ-
ees who will be promoted in each class.
Define for some given class (see figure 4.8):
Tps == average time until promotion, under the condition that promotion
will take place, for employees with grade seniority;
Vs == direct wastage fraction for employees with grade seniority
s;
Ps == direct promotion fraction for employees with grade seniority s;
7rs == average fraction of employees with grade seniority s who will be
promoted after some time .
.- --) s s + 1 1-----'"
FIGURE 4.8. Transition scheme indicating the function of grade seniority in

the model for one class.
102 CHAPTER 4
It should be remarked that promotion is defined here as a transition to a dif-

ferent class. The influence of the employee's retirement is neglected.
Now we can give the following formula:
(4.2)
that is, the fraction of employees promoted directly plus the fraction of em-
ployees who will be promoted later.
Tps can be written as
T'ps
(4.3)
'lrs '
in which T' ps might be interpreted as the average time until promotion for
an employee with grade seniority, if we count the time that is not followed
by promotion as zero. T' ps can be computed by using the following rela-
tion:
T'ps = 'lrs + (l - Vs - Ps)T'p,s+ I' (4.4)
These formulas are easy to handle for computation. In the same way, the
average grade seniority at which turnover occurs can be calculated. The
average time an employee will spend in his class is computed as well.
Define:
Ts == average time yet to spend in his class for an employee with grade
seniority s.
Then:
(4.5)
Another important item concerning career policies is the average fraction
of employees who will be promoted within s years after entering the current
class.
Define:
Qs == average fraction of the entrants in a certain class who will be pro-

moted within s years.
Then:
Qs = ptllpf .IIt: 1(1 - Vk - Pk)l. (4.6)
The index Qs gives some information about the career prospect of new en-
trants in a certain class. However, high turnover fractions will decrease the
Table 4.8. Extra Information about Actual Career Policies

TRANSITION 3
MIN. GRADE SENIORITY - 1.0

MAX. GRADE SENIORITY -13.0
AVERAGE GRADE SENIORITY- 6.2
TRANSITION 4
MIN. GRADE SENIORITY - 1.0

MAX. GRADE SENIORITY =13.0
AVERAGE GRADE SENIORITY'" 6.3
PERCENTAGE OF ENTRANTS IN CLASS 2 WHO MAKE FOLLOWING TRANSITIONS

PROMOTION TURNOVER STAYERS
AFTER 13 YEARS: 71.1 ~~3.1 5.9
TOTAL 73.5 26.5 0.0
AVERAGE TIME SPENT IN CLASS 2: 6.2
value of Qs' This turnover may be caused by employees who do not perform
well in the class or who want to leave the class for better positions in other
organizations.
Table 4.8 shows an example of FORMASY results concerning average
career opportunities. In the table, transition 3 is the promotion from grade
TA 1 to grade THA; transition 4 is turnover from grade TA 1; class 2 is grade
TA 1. For this example we have the following results: Tpo = 6.2 (for transi-
tion 3); QI3 = 0.711; Qoo = 0.735; To = 6.2.
Simulation possibilities
As we mentioned before, a manpower planning system should offer some

possibilities for changing parameters. The forecasting of the consequences
of changed parameters is called simulation. Simulation helps the personnel
manager to design policies and may support the manager's decisions.
FORMASY offers the following simulation opportunities:
1. Forecasting the consequences of a change of transition fractions in

order to evaluate effects of changes in wastage patterns and career
policies,
2. Forecasting the consequences of a change of recruitment (a recruit-
ment-planning procedure),
3. Forecasting the consequences of a change of retirement age,
4. Forecasting by using FORMASY dynamically.
Forecasting the consequences of a change of transition fractions.

FORMASY contains three options for changing the transition fractions:
104 CHAPTER 4
1. A shift (lengthening or shortening) of the time until promotion from

a certain class. For example, if the historical promotion fractions for
all grade seniorities out of some class are:
Grade Seniority: 2 3 4 5 ~6
Historical Fractions: .1 .2 .3 .2 .1 .1
then, a lengthening of the time until promotion by one year would
result in the following fractions:
Grade Seniority: 1 2 3 4 5 ~6
New Fractions: o .1 .2 .3 .2 .1
This procedure is helpful in indicating the effects of changes in pro-
motion possibilities.
2. Changing one or more single fractions of a certain transition.
Transition here means a change of class. In the example above, this
results in changes in the fractions for grade seniority 2 and 3, for in-
stance, in:
New Fractions: .1 .3 .4 .2 .1 .1
3. Multiplying all fractions of a certain transition by the same constant.
In the example above, multiplying all fractions by 1.5 would lead to:
New Fractions: .15 .30 .46 .30 .15 .15

This procedure turns out to be very useful for evaluating the effects
of a general increase or decrease of turnover fractions.
Forecasting the consequences of a change of recruitment. FORMASY fa-

cilitates evaluation of the impact of alternative recruitment policies by
allowing the user to choose an arbitrary future recruitment. Application of
the forecasting procedure then gives the impact of the recruitment on the fu-
ture distribution of employees over the categories.
With another option of FORMASY one may compute the necessary
recruitment that leads to numbers of employees in all grades equal to or (if
necessary) above given lower bounds. This option is constructed in such a
way that the user may give a maximum allowed number of recruits. More-
over, in case the upper bound for the allowed recruitment is reached in some
year, the user may allow recruitment some years earlier, which leads to
overoccupation of the grade for some time.
Using these upper bounds, one can take care of stable recruitment. Also,
these upper bounds can be used in case the possibility for the organization
to recruit people from the labor market is restricted. The overflow feature
makes it possible to choose a compromise between the maximally allowed
recruitment and the minimally desired manpower distribution.
Notice that in this procedure the recruitment is considered on the level of
grades instead of classes. This means that the original transition matrix has
to be reformulated for grades. We will not discuss this in more detail here.
The recruitment planning procedure has been based on a dynamic pro-
gramming approach and has been described by Van Nunen and Wessels
(1978). Tables 4.9 and 4.10 are an example of an application of this proce-
dure. The procedure shows the impact of changes in the manpower require-
ment on the necessary recruitment. Table 4.9 gives the forecasted recruit-
ments when the allowed numbers of recruits in each year for grades TA and
TAl are 80 and 50, respectively. The desired lower bounds for these grades
are 200 and 500, respectively, each year, and no earlier recruitment is
allowed. It is assumed that the recruitment during 1977 is first counted in
the manpower distribution of 1978. The resulting manpower distribution
over the grades is given in table 4.10. Notice that in grade TA 1 it takes some
time to reach the lower bound.
Forecasting the consequences oj a change oj retirement age. The user may

change the retirement age for every planning year and is able in this way to
assess the impact of alternative retirement ages.
Forecasting by using FORMASY dynamically. One limitation of the Mar-

kov model is that the transition fractions are assumed to be constant.
Dynamic use of FORMASY removes this objection. If the user expects cer-
Table 4.9. Recruitment Planning Procedure

RECRUITMENT IN PLANNING PERIOD
•••••••••••••• + ••••••• + •••••••
GRADE
YEAR * TA TAl THA THA1 THABD * TOTAL
.-----------------------------------------------------
.................
1977
* 57 0 0
*
0 0
........................................
57
**
***
1978 80 0 0 0 0 80
1979 73 28 0 0 0 101
1980
* 69 34 0 0 0 103
1981
* 73 29 0 0 0
* 102
106 CHAPTER 4
Table 4.10. Forecasted Manpower Distribution Including Recruitment

DISTRIBUTION IN PLANNING PERIOD
.....................................................
M~NPOWER
GRADE
YEAR * TA TAl THA THAl THABD * TOTAL
* 238 547 :576 :L80 70 * 1411

.....................................................
1977
:l979 * 75 *
1978 200 531 :394 184 75 1384
198() * 79 *
198 504 412 190 :L379
1981 * 81 *
200 !7iOO 425 200 1404
:L982 * 1:16 *
200 !7jOO 4:56 210 :L427
* 200 450
~.)()O 222
* 1451:1
-----------------------------------------------------
tain changes in promotion policies or turnover fractions, he may compute
the impact of these changes with FORMASY. He applies, for instance, the
actual transition matrix for the planning horizon for which he expects this
matrix to be correct. The forecasted manpower distribution for this plan-
ning horizon can be stored. Then he restarts the processing of FORMASY
and applies the new transition matrix to the stored manpower distribution
over the categories.
All options described in this section are executed interactively by com-
mands of the user. The simulation possibilities may be considered as the
main contribution to the applicability of this planning system. We will dis-
cuss the advantages in some detail at the end of this chapter. In the appen-
dix, an example is given of the use of FORMASY for the case study of chap-
ter 6. This listing demonstrates how the options mentioned above are used.
ESTIMATION OF TRANSITION FRACTIONS
In chapter 3 we described how transition fractions for the Markov model

can be determined from historical data. If one wants to obtain some infor-
mation with respect to the promotion fractions, the analysis of historical
data is a straightforward first step. If one wants to start with career plan-
ning and no data from history are available, then one may try to deduce
promotion fractions from an intended policy or one may evaluate by simu-
lation which promotion fractions are sensible.
For the estimation of the turnover fractions one may use:
1. Historical data on the considered level of aggregation,

2. Historical data on a higher aggregation level within the organization,
3. Data from other organizations or from the national level.
We will discuss these possibilities by means of example A, where the three

levels of engineers in a Dutch public service were considered. If one wants to
determine the turnover fractions for the highest-level engineers, one may
compute these data according to formula (3.6) by using past data on this
same group of engineers. If the number of leavers cannot be obtained in
such detail, then one may be satisfied with, for instance, the number of
leavers for the entire group of engineers in the past or turnover for the total
organization. If these data are not available, then one may try to obtain
such information from other public services or from national turnover frac-
tions for people with similar training and qualifications. In this situation,
however, the reasons for turnover should be taken into account explicitly.
COMPUTATIONAL EFFICIENCY OF FORMASY
To assess the computational efficiency of FORMASY, we will consider the

required storage capacity and the computation time for the system's appli-
cation. The required storage capacity depends on the way the possible
transitions of employees are stored. If we consider the transition fractions
Pi} in formula (3.1), which are based on the transitions between classes in
the terminology of this chapter, then these fractions are equal to zero in
general if j < i. This means that no degradations may occur. By this prop-
erty, the matrix of transition fractions can be stored in a very efficient way
by storing only the elements that can be positive:
PIl P12 P13 ••••• P1M PIO
P22 P23 • • • • • P2M P20
P = P33 P34 • P3M P30
PMM PMO
1
Here, M is the number of classes and PiO denotes turnover from class i.
Of course, we consider categories in the FORMASY system, not classes.
This means that for all classes in the matrix above, a number of transition
108 CHAPTER 4
fractions have to be stored-namely, for every grade seniority-but in order

to avoid a very technical discussion we will not treat this aspect in detail.
Another saving of storage capacity is achieved by deleting the positions
that do not occur in practice. Recall that the total size of the system would
be (G x Q x E x S + 1). For the examples of FORMASY models given
earlier, this would lead to the following number of states:
Example A.1: 51 and 66, respectively;

Example A.2: 241;
Example B.1: 151;
Example B.2: 131;
Example C: 736;
Example D: 301 (or with age classes included: 901).
However, a considerable saving on the number of states is usually possi-

ble. This gives for the examples:
Example A.1: nothing can be gained, but 51 and 66 are not very large;
Example A.2: 1 + 13 x 12 = 157 states suffice;
Example B.1: since the subgrades only work for the lowest grade, it is
sufficient to consider 1 + 7 x 10 = 71;
Example B.2: 1 + 13 x 7 = 92 states suffice;
Example C: nothing can be gained;
Example D: 1 x 16 x 10 = 161 states suffice (with age classes included, a
reduction to 391 appeared to be possible); here a further reduction is
possible since it is not necessary for all states to consider ten levels for
grade seniority.
The second aspect, the computation time, will be discussed briefly here.
The computation time depends of course on the number of categories and
the number of possible transitions between the classes. The forecasting pro-
cedure that leads to relatively low computation times is very simple. For the
example considered in the appendix-which required several executions of
the forecasting procedure-the total processing time on a B7700 computer
was about forty seconds.
USEFULNESS OF FORMASY
The program system FORMASY has been designed to make it simple to

simulate the effects of various possible policies with respect to manpower.
As before, the term simulation does not mean Monte Carlo simulation.
Here we mean that for several policies, forecasts are computed for the fu-
ture behavior of the manpower system. In this sense the effects of policies
are simulated. Using this aspect, one can apply FORMASY as an instru-
ment to match manpower requirement and manpower availability (or to
estimate the flexibility of the organization for changes in manpower re-
quirements). It is especially suitable for medium- and long-term planning.
Short-term planning (less than a one-year period) has to be so detailed that
it is usually preferable to consider individuals rather than categories, as we
discussed in chapter 1.
The main advantages of FORMASY are the following features:
It is an interactive computer program;

It can be used by experts in the manpower planning field;
It is based on a simulation approach; that is, there are no built-in optimi-
zation criteria to assess the policies;
It contains many options for changing the policies easily;
It has many output facilities.
There are also some conditions for optimal usefulness of this computer
program system:
The user has to be familiar with manpower planning problems;

The user should be creative in designing alternative policies and should
have the knowledge required to assess the alternatives;
The required input data should be easily accessible;
The computer system should have facilities for 'interactive use.
The usefulness of FORMASY and the implications for the organization

will be discussed in greater detail in chapter 5.
5 THE USE OF FORMASY
In this chapter we consider the use of the manpower planning system FOR-
MASY. Three parts can be distinguished in this chapter. In the first part we
discuss in some detail the process of matching manpower availability and
manpower requirement. In particular, the use of a system such as FOR-
MASY for the matching process is described and some examples of the
application of FORMASY for designing career and recruitment policies are
given.
In the second part we discuss some aspects of the system that have to be
considered before FORMASY can be applied. We indicate how the level of
aggregation may be chosen-that is, how should one choose function
groups such that manpower planning is possible and useful-and we discuss
how the length of the planning horizon may be chosen.
In the third part some organizational questions with respect to the use of
FORMASY are treated. We discuss the requirements for personnel registra-
tion and the requirements for the employees of the personnel department.
Finally, the place of FORMASY within the organization is described.
110
THE USE OF FORMASY 111
THE MATCHING PROCESS AND FORMASY
In chapter 1 we discussed the manpower planning activities of career plan-

ning, recruitment planning, training and development planning, and alloca-
tion planning. The importance of these activities for matching manpower re-
quirement and availability depends on the type of manpower flow or transi-
tion that is considered and on the restrictions imposed on these transitions.
Flows can be purely vertical, a transition to a similar function on a higher
organizational level. Flows can also be purely horizontal, a transition to a
different type of function on the same organizational level. And flows can be
a mixture, a transition to a different type of function on a higher level within
the organization.
If career restrictions are very tight-for instance, because of a high degree
of commitment-then the vertical component of the flows is very important,
and thus also career planning. If many functions can be occupied only by em-
ployees with experience in different types of functions, then the horizontal
component of the flows is very important, and thus also allocation planning.
In many situations both types of planning have to be considered. But if there
is a high degree of flexibility with respect to horizontal or vertical move-
ment-for instance, because of the high mobility of employees-it may be
sufficient to consider for the medium and long term only the vertical or only
the horizontal transitions. This example of aggregation will be discussed in
some detail later in this chapter. An important variable in each type of plan-
ning is recruitment. In certain situations-for instance, if some types of em-
ployees can rarely be recruited-it may be necessary to pay special attention
to recruitment planning. In the same way, training and development may be
an important variable in each type of planning.
In the discussions that follow, we will describe the use of FORMASY for
the development of career policies, allocation policies, recruitment policies,
and training and development policies.
Career policies
The design of career policies is a very important activity in organizations to-

day. While machines have an economic lifetime of about two to twenty
years, employees may stay with the organization for forty years or even
longer. There is growing acceptance that employees have their own career
rights and influence on career policies. The degree of acceptance of this de-
velopment by the organization determines the level of commitment of the
organization to its personnel.
112 CHAPTER 5
Restrictions with respect to the vertical flows of employees may be ex-

pressed by:
1. The organization's restrictions,

2. Personnel restrictions,
3. The environment's restrictions,
4. Stability restrictions.
The organization's restrictions are caused by the availability of functions on

the various organizational levels and by the availability of money to pay the
promotions. Personnel restrictions are caused by the employees. These re-
strictions are determined by the interests and potential abilities of the em-
ployees. The environment's restrictions are caused by the career opportuni-
ties offered by other organizations. And stability restrictions are caused by
the wish to have stable career policies over time.
Career policies can be described by the following characteristics:
The position of an employee when entering the organization,

The different positions an employee occupies subsequently,
The number of years spent in each position.
The starting position of an employee in the organization has to agree to

some extent with the usual level both within the organization and elsewhere
in society (3). This is true not only for the salary level but also for the con-
tents of the tasks.
The different positions an employee occupies during his stay with the
organization have to agree not only with the individual expectations of the
employee and his potential ability (2) but also with the organization's re-
quirements (1). After some time he may be able to perform a function on a
higher level. Particularly for highly trained employees who start in a learn-
ing function and who will stay with the organization for a long time, career
planning is very important. Therefore, one has to forecast future organiza-
tion requirements and the evolution of manpower distribution in order to
notice possible discrepancies in time.
The number of years until the position of an employee changes has to
agree with his personal ability and with the organization's demands. It is
important to develop a career plan for all relevant categories of personnel,
including position when entering the organization; the minimum, average,
and maximum time between the transitions; and the average fraction of em-
ployees reaching the given positions. Such a career plan indicates clearly the
prospects for the recruits and the employees already present within the
organization. This may avoid disappointments and a possible feeling of
arbitrariness with respect to promotions (4). Of course, such a scheme is an

average scheme, which leaves the possibility open that the better employees
may reach top positions within a few years, while others may never reach
top positions. See figure 5.1 for an example of this feature.
Many organizations do have a career plan for certain groups of person-
nel (usually employees with high training or experience). But often this plan
is not formalized. Sometimes only a minimum time for promotion is for-
malized because of experience requirements.
With respect to career opportunities, FORMASY can be used in the fol-
lowing way. First, compute the impact of historical career opportunities
(promotion fractions) on the evolution of the manpower distribution. The
career plan may depend on the qualifications of the employees. A compari-
son of the resulting manpower availability forecast with the manpower re-
quirement forecast indicates possible discrepancies. Second, alternative ca-
reer policies can be developed. This means that the position when entering
the organization is changed, or the sequence of occupied fractions, or the
time spent in each of the functions. This process of changing career policies
has to be continued until the career policies agree with the organization's re-
quirements. Later in this chapter we will give some examples of the use of
FORMASY for the design of alternative career policies.
agel agel age f

60 60 60
level 3
48 48
44
40
level 2
32
30r-__________ ~ 30 1-_______- \ ~---.
28
level 1
20~ __________
(a)
- ~
20 '--_ _ _ _ _ _.....
no. of employees (b)

--+
10~
no. of employees
________
-
(c) no. of
FIGURE 5.1. Three alternative career policies: (a) the average career where all em-
~
employ~es
ployees will reach the top level; (b) the case of a selection bound at age 44, such that
only a part of the employees will achieve the third level; (c) another case of selection
bounds. Employees are promoted to level 2 at age 28 or 32. Promotion to level 3
takes place between ages 40 and 48. For convenience, we have expressed the careers
according to age, but one may also replace age by, for instance, length of service.
114 CHAPTER 5
Allocation policies
Allocation policies indicate the numbers of employees who will occupy dif-
ferent types of functions for the medium and long term. Allocation plan-
ning is important if the organization wants to maintain a given intensity of
horizontal flows. This will be the case, for instance, if for certain functions
one requires experience in other types of functions. As an example, consider
a bank where the requirement for managers of large affiliations is several
years of experience in staff functions and several years of experience as
manager of some smaller affiliations.
Restrictions with respect to the horizontal flows of employees can be
classified much as the restrictions mentioned above for career policies. The
organization's restrictions and stability restrictions on allocation policies
need no.explanation. The environment's restrictions are caused by, for in-
stance, opportunities offered by other organizations. Personnel restrictions
are closely related to the willingness of employees to participate in the allo-
cation patterns (mobility). In chapter 1 we considered an example with two
function levels and three types of functions (A, B, and C) on the lowest
level. The requirement for functions on the second level is two years of ex-
perience in A-type functions, three years in B-type functions, and five years
in C-type functions.
The following model structure may be used for the application of FOR-
MASY:
g== type of function (g = 1,2,3, or 4 for A-type, B-type, C-type, and

level-2, functions);
e == experience of the employee (e = 0,1,2, ... , 7);
s == function seniority (s = 1,2, ... ).
The experience of employees can be given as: e = 0 (no experience), e = 1

(at least two years in type-A functions), e = 2 (at least three years in type-B
functions), e = 3 (at least five years in type-C functions), e = 4 (required ex-
perience in functions A and B), e = 5 (required experience in A and C),
e = 6 (required experience in B and C), and e = 7 (fully qualified em-
ployees).
We assume implicitly that employees are transferred only if they have ob-
tained the required experience in their functions. If the transition fractions
between the types of functions are given, then forecasts for the number of
employees in each type of function can be made. Notice that the possible
transition fractions depend on the willingness of employees to participate in
the allocation scheme (mobility). Evaluation of these results may lead to
changes in allocation policies. This model could also be used for the estima-
tion of mobility required to guarantee the matching of required and avail-
able experience.
Recruitment policies
Recruitment policies indicate the numbers and types of people who have to
be recruited for the medium and long term. The restrictions with respect to
recruitment policies are again classified much as the restrictions on career
policies. The interpretation of the environment's restriction is the situation
on the labor market. The personnel restriction indicates the willingness of
people to join the organization. This willingness will be related to job se-
curity, working circumstances, and so on.
It should be remarked, once again, that recruitment policies have a great
impact on career policies. Changes in an existing career plan may be neces-
sary because of unstable recruitment over time.
Consider an example. One group of employees is embodied into a grade
system with two grades, and the desired number of employees in every grade
is fixed. If, in a certain period, extra recruitment was necessary in grade 1
because of arrincrease in the number of functions in grade 1, and this will
continue in the following years, then average career prospects deteriorate
since relatively fewer functions become available in grade 2. A possible ac-
tion here might be the internal or external recruitment of older person-
nel-for example, less-trained people with some experience. This leads to an
increase of wastage (retirement) such that career prospects for younger em-
ployees need not deteriorate. Later in this chapter we will see an example of
such recruitment policies.
Often little attention is given to the desirability of having a stable recruit-
ment policy. Many organizations recruit people in situations of an increase
in sales, since one does not want to lose part of the market. However one
has to realize that these employees may stay a long time with the organiza-
tion; thus their employment and career opportunities have to be ensured.
Medium- and long-term recruitment planning is therefore important. Re-
cruitment planning is also important if the situation on the labor market
changes such that a shortage of certain groups of personnel may occur.
Recruitments may be restricted also by training facilities. In fact, if an
organization has its own training institute, as do police organizations, then
the capacity of this institute puts restrictions on recruitment.
The first question to be answered if recruitment is necessary is whether
the vacancies can be filled from within the organization (internal transfer),
116 CHAPTER 5
possibly after a training course, or whether external recruitment is desired.

In this approach also, education and training requirements have to be con-
sidered. Later in this chapter we will consider an example of the use of
FORMASY for the development of recruitment policies.
Development and training policies
Development and training policies indicate the numbers and types of em-
ployees who should attend courses in order to be prepared for future posi-
tions. The restrictions are classified much as the restrictions on career poli-
cies. Particularly important is the personnel restriction, which indicates not
only the willingness of the employee to participate in development and
training programs, but also the ability of the employee to attend such
courses. The environment's restriction indicates the development and train-
ing opportunities offered by other organizations and, for instance, the
capacity of training institutes.
Recall that development and training policies may influence and will be
influenced by both horizontal flows and vertical flows.
FORMASY has been used for the design of training and development
policies for the Dutch police forces. This has been described in Van
Meeteren (1978).
Preliminary steps for the application of FORMASY
Before we describe two applications of FORMASY, we will make some gen-

eral remarks on the preliminary steps necessary for such applications.
First, one has to determine which questions have to be answered. One
possible question is whether the available numbers of employees who will be
qualified in the future will agree with the required numbers of employees. In
the next phase, the relevant characteristics of personnel demand and supply
must be determined-for instance, training or experience. Before one can
apply a system such as FORMASY, the level of aggregation and the plan-
ning horizon also have to be fixed. The subsequent steps are indicated in fig-
ure 5.2.
Application of FORMASY means that the following data have to be
given:
The actual manpower distribution,

The transition percentages and recruitment (preferably based on histori-
cal data).
manpower plan
ning problem
choice of
---. personnel
group - cnOl.ce or
characteristics --.
choice of the
level of
aggregation
L cnOl.ce OI 1:ne
planning I - application of
horizon FORMASY
FIGURE 5.2. Scheme of necessary steps when using FORMASY.
The forecasts of manpower availability based on these data have to be com-

pared with the manpower requirement forecasts. Then transition fractions
and recruitment can be changed in order to diminish the deviations of man-
power availability from manpower requirement. The resulting policies have
to be evaluated. This may lead to a change of manpower requirement,
which restarts the matching process.
APPLICATIONS OF FORMASY IN DESIGNING CAREER

AND RECRUITMENT POLICIES
Application in the Netherlands Ministry of Public Works
As we described in chapter 4, the engineers of the Public Works Office have

their own grade system. The model structure was described in figures 4.1
through 4.3. We will consider here two examples of the application of FOR-
MASY:
Avoiding an overstaffing in top grades,

Maintaining career prospects by integration of groups.
Example 1: Avoiding an overstaffing in top grades. For the top-level

engineers of this example, as for many other groups of employees, one
meets the following situation. In the recent past an increase in the size of the
group occurred because of a steady inflow of young members. At the mo-
ment, this increase has stagnated, and no further increase is expected within
the foreseeable future. The consequences for the distribution of engineers
over the grades will be investigated. Notice that we will consider in particu-
lar the vertical flows of employees, whereas the horizontal flows are aggre-
gated. Thus all engineers in various fields of training are brought into one
group. Application of the forecasting procedure (without recruitment)
118 CHAPTER 5
Table 5.1. Allowed, Current, and Forecasted Distribution of Top-Level

Engineers
Grade
1 2 3 4 5 Total
Allowed No. of
Employees in 1977 288 210 99 86 683
No. of Employees
in 1977 120 204 120 90 78 612
Forecasted No. of
Employees in 1989 a 5 106 168 123 402
aThe forecast for 1989 is based on the current situation without recruitment and reflects
the actual promotion policy of the last five years.
shows that the current bottom-heaviness of the hierarchical pyramid will be

transformed into a rather strong top-heaviness in a relatively short time (see
table 5.1.)
From this forecast and the forecast for the intermediate years, one can
easily compute how the total population may be kept at the present level of
612 by recruitment. In this application, the computation is simple since
practically there can be inflow only in grade 1. The forecasted recruitment
for the forthcoming years is given in table 5.2.
It is clear that this recruitment, which is much lower than recruitment in
the past, will take care of filling the lower grades. However, the overrepre-
sentation in the top grades does remain (see table 5.3).
This prospective overstaffing in the top grades is not allowed for formal
and financial reasons. But it is also highly undesirable for social reasons be-
cause of the lack of appropriate positions for high-level employees with sev-
eral years of experience. The only solution is to start some slowing down of
promotions in the lower grades as early as possible. In our view, it is better to
keep all employees a bit longer on the lower levels of their career than to
allow the generation of a great stock of engineers waiting for promotion.
Table 5.2. Forecasted Recruitment in Grade 1 Needed to Maintain Total Work

Force at Full Strength
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988
19 21 22 20 21 22 22 22 21 23 21 22
Table 5.3. Allowed, Current, and Forecasted Manpower Distribution with

Recruitment under Actual Promotion Policy
Grade 1 2 3 4 5 Total
Allowed 288 210 99 86 683
1977 120 204 120 90 78 612
1978 101 195 143 91 82 612

1979 113 153 168 93 85 612
1980 96 140 192 97 87 612
1981 80 131 207 104 90 612
1982 95 95 218 III 93 612
1983 112 70 214 121 95 612
1984 115 66 200 134 97 612
1985 116 58 190 148 100 612
1986 116 56 176 159 105 612
1987 118 64 154 167 109 612
1988 118 73 135 170 116 612
1989 118 76 126 169 123 612
In tables 5.4 and 5.5 we show forecasts for alternative promotion poli-
cies. Each alternative transition matrix is derived from the original one by
shifting all promotions in some grades by one or more years. The recruit-
ment policy of table 5.2 is always used. Notice the small changes in the total
number of employees because of different turnover fractions for the various
grades. Since the number of years until promotion has been changed, the
employees will stay in a grade for a longer time. This may also change the
forecasted turnover.
In one processing session, several of these shifts can be tried out to ob-
tain a good indication of the extent to which the number of years until pro-
motion has to be increased in order to obtain reasonable prospects for the
distribution of employees over the grades. In a later stage, refinements may
be brought in by considering more then pure shifts.
Example 2. Maintaining career prospects by integration of groups. In

example 1 the situation for the top-level engineers was investigated in some
detail. For the two other groups of engineers in the same organization the
situation is similar. So by planning the three groups separately, the result is
a considerable lengthening of time until promotion for all engineers in-
volved.
Table 5.4. Forecasted Manpower Distribution for a Shift of 1 Year in Grade 2
and 1 Year in Grade 3
Grade
Year 1 2 3 4 5 Total
1977 120 204 120 90 78 612

1978 101 219 122 88 82 612
1979 113 188 138 86 85 610
1980 96 180 161 87 87 611
1981 80 167 184 90 89 610
1982 95 128 200 94 91 608
1983 112 93 212 99 92 608
1984 115 83 215 102 94 609
1985 116 80 209 110 94 609
1986 116 73 203 120 96 608
1987 118 70 190 132 98 608
1988 118 79 168 142 101 608
1989 118 87 148 149 104 606
Table 5.5. Forecasted Manpower Distribution for a Shift of 3 Years in Grade 2

and 2 Years in Grade 3
Grade
Year 1 2 3 4 5 Total
1977 120 204 120 90 78 612

1978 101 230 113 85 82 611
1979 113 224 108 81 85 611
1980 96 239 108 80 87 610
1981 80 240 120 79 89 608
1982 95 201 137 81 90 604
1983 112 158 156 86 91 603
1984 115 135 173 88 91 602
1985 116 117 191 86 90 600
1986 116 109 201 83 90 599
1987 118 106 201 84 90 599
1988 118 99 201 89 90 597
1989 118 98 194 97 90 597
However, with regard to tasks as well as salaries, there is a considerable

overlap between the three groups. Figure 5.3 shows that the top grades of
the surveyors have the same salary level as the lower grades of the middle-
level engineers, and the same is true for the top grades of the middle-level
engineers and the lower grades of the top-level engineers.
The main problem for all three groups is the large number of engineers
allowed in the lower grades against the small numbers allowed in the top
grades. In a nongrowing system this must inevitably lead to rather poor ca-
reer prospects for those involved. From figure 5.3 one can see that the lower
grades of the middle-level engineers are not fully occupied, whereas the sur-
veyor grades on the same level are overoccupied. This gives the clue to a
solution: transfer some of the allowed numbers of lower-grade engineers at
the middle-level to the top grades of the lower-level engineers and also some
of the allowed numbers of top-level engineers in lower grades to the top
grades of the middle-level engineers. The result of such an operation is bet-
ter prospects for all groups, although the recruitment of top-level engineers
must certainly be diminished further.
SALARY-
i ~ROUP top.level engineers
:~,
112~
a 100 400
rade
112
o lOa 500 lana
YOO
~ no. of employees
FIGURE 5.3. Representation of allowed and actual numbers of employees and the
salary levels for the three types of employees.
::J = allowed number of employees in each grade.
=:] = actual number of employees in each grade_
'In this figure the actual numbers of top- and middle-level engineers in grades 1 and 2
are aggregated since in the organization the allowed numbers are also aggregated_
122 CHAPTER 5
By using FORMASY, it is simple to find out what might be obtained in

this way. Forecasts can be made by choosing promotion and recruitment
policies for all three groups, and it is easy to compute whether the fore-
casted numbers fit the allowed numbers after the transfer. Figures 5.4 and
5.5 give the result of such an exercise for the three groups of engineers. Fig-
ure 5.4 shows the number of employees at the salary level that lower- and
middle-level engineers have in common. The dotted lines in the figure show
the numbers of functions on this salary level: 370 for lower-level engineers
and 1,830 for middle-level engineers. Furthermore, the striped lines give the
forecasted number of lower-level engineers and the total forecasted number
of employees for the given promotion and recruitment policy. Figure 5.4
shows that transferring 413 middle-level functions to lower-level engineers
in the following twenty years would be sufficient to maintain the current
promotion and recruitment policy; figure 5.5 shows that transferring 91
top-level functions to middle-level engineers in the following twenty years
would be sufficient to maintain the current promotion and recruitment pol-
icy.
2000
no'!; of
engt neers
1500
middle .level eng i neers
1000
413
sao
._._._--_.__
lower.level engineers
._._._._._--_._._-_._-_._. 1
100
1977 1980 1985 1990 1995 1997

_ years
FIGURE 5.4. Integration of middle- and lower-level engineers.

900
no's of
tnglneers
top-level eng i neers
500
91
L
middle-level engi neers
lOa
°19~7:::-74---=19~80:-+----+---I1--~1±8::-:5'-+--+--+----+-::iI99':::O:-+---t----i-+--;;19±975-+-1;;;!997
years
FIGURE 5.5. A representation analogous to figure 5.4 for the salary level
that top- and middle-level engineers have in common.
Using the program system, one can tryout some promotion and recruit-
ment policies to find which combination of promotion shift and recruitment
policy fits best for the purpose. As figures 5.4 and 5.5 show, an integrated
treatment of various groups may help considerably in reconciling the condi-
tions imposed by workload and salary constraints on one hand and career
prospects on the other hand. In order to realize the developed policies, one
has to execute function analysis, potential appraisal, and so on. This appli-
cation has also been described in Van der Beek, Verhoeven, and Wessels
(1978).
Application in a large industrial firm
As we have seen, FORMASY can be applied to match future demand for

and supply of personnel by considering the impact of changes in historical
policies. Sometimes, however, the historical transition fractions are not
appropriate. Consider the following example.
124 CHAPTER 5
The research lab of a large industrial firm employs two types of en-
gineers: engineers with a university degree on the master's or doctor's level
and middle-level engineers. Top management decides to decrease the total
personnel for this research lab by 3070 a year. The question is to which de-
gree this leads to an aging organization (a catastrophic situation for a re-
search organization) and what kind of recruitment will be possible in the
future.
The model structure for the engineers is given in figure 5.6. The model
has three grades (five classes); each class has fifteen grade seniorities.
Classes 1 and 3 stand for grades 1 and 2 for engineers with a university de-
gree (high-level engineers). Classes 2 and 4 indicate grades 1 and 2 for mid-
dle-level and for high-level engineers. Class 5 stands for grade 3, both for
middle-level and for high-level engineers. The engineers are recruited in
classes 1 and 2. The retirement age is 60.
We will forecast the evolution of manpower distribution over the classes.
There have been no promotions in recent years, but this trend may not con-
tinue. Therefore, it is not possible to use historical promotion fractions. In
the first instance, some promotion fractions are chosen, based on manage-
ment's ideas of a reasonable career for the engineers. Consider as an exam-
ple the transitions from class 2 to class 4. According to management about
40% of the people in class 2 can make this promotion, and the promotion
has to occur after eight to ten years of service.
Based on these transition fractions, forecasts for the next ten years have
been made. In the case of no recruitment, the forecasted evolution of man-
power distribution over the grades is given in table 5.6. In table 5.6 the total
university degree
middle-level 2
grade _
FIGURE 5.6. Classes and transitions for engineers of the research lab of a
large industrial firm. There are 3 grades, and in each grade two classes: engi-
neers with a university degree and engineers with a middle-level degree. In
grade 3 the engineers have been taken together. Grade seniorities have been
deleted in this figure.
Table 5.6. Actual Manpower Distribution and Forecasts for 1990

Grade
Year Gl G2 G3 Total
1980 50 70 120 240
1981 49 67 118 234

1982 48 65 115 228
1983 46 63 112 221
1984 40 59 110 209
1985 38 56 109 203
1986 35 52 108 195
1987 33 48 107 188
1988 28 47 103 178
1989 25 48 98 171
1990 21 47 92 160
number of employees decreases by 33070 in this ten-year period. 01 indicates

grade 1, and so on. The numbers of employees in the classes are lumped in
order to obtain the numbers of employees in the grades. The corresponding
actual and future age distributions are shown in table 5.7. In table 5.7 each
line indicates the age class and number of employees with the correspond-
ing ages. The figure indicates the number of the grade of the employee
(1 = grade 1, 2 = grade 2,3 = grade 3). For instance, eight employees are in
the age class 52-53 in 1980; two are in grade 1, three in grade 2, and three in
grade 3.
Now we will investigate whether the forecasted evolution of manpower
distribution over the grades agrees with the desired future manpower dis-
tribution. Table 5.8 shows the desired manpower distribution in the plan-
ning period based on the yearly decrease of 3%. Compare these numbers
with those in table 5.6. As in table 5.6, the numbers of employees in the
classes are lumped in order to obtain the numbers of employees in the
grades.
Comparing tables 5.6 and 5.8, we notice a discrepancy between the ac-
tual manpower distribution and the desired manpower distribution. The ac-
tual percentage of employees in the classes is as follows: class 1,0%; class 2,
20.8%; class 3, 12.5%; class 4, 16.7%; and class 5,50%. The desired man-
power distribution for 1980 is: class 1, 24%; class 2, 20%; class 3, 12%;
class 4, 8%; and class 5, 36%.
Table 5.7. Actual Age Distribution and Forecasts for 1990
Age Actual Manpower jor 1980
58-59
56-57 1111233
54-55 133333
52-53 11222333
50-51 1112333333333
48-49 1111222222233333
46-47 11122333333333333
44-45 122222222333333333
42-43 2333333333333333
40-41 1111122222222333333333333333
38-39 111112222233333333333333
36-37 11111111112223333333333333
34-35 1111111222223333333333
32-33 111122222222222222233333333
30-31 1222222222
28-29 2
26-27 2
24-25
Age Forecast jor 1990
58-59 1112222223333
56-57 11223333333333
54-55 12222223333333
52-53 2333333333333
50-51 1122222222333333333333
48-49 1122222233333333333
46-47 111122222233333333333
44-45 11122222333333333
42-43 1122222233333333333
40-41 122333
38-39
36-37
34-35
32-33
30-31
28-29
26-27
24-25
Note: 1 figure = 1 employee.
Table 5.8. Desired Manpower Distribution Based on Yearly Decrease of 3UJo
Grade
Year G1 G2 G3 Total
1980 110 50 90 250
1981 107 49 87 243

1982 103 47 85 235
1983 HlO 46 82 228
1984 97 44 80 221
1985 94 43 77 214
1986 92 42 75 209
1987 89 40 73 202
1988 86 39 71 196
1989 84 38 68 190
1990 81 37 66 184
In fact, comparison of tables 5.6 and 5.8 also shows that only a few em-
ployees can be recruited in this ten-year period. This would lead to an aging
research lab. In order to prevent this development, management wants to
consider alternative policies. One alternative policy is:
Recruitment of ten employees each year in class 2 (as was the historical
policy) and also ten employees in class 1 from other divisions,
Transfer of highly experienced personnel from class 5 to management
functions in other divisions (in the program system FORMASY this
can be accounted for by changing the turnover percentages-for
example, for grade seniorities 5,6, ... ,15 in category 5 from 2% to
12070 ),
Changing the retirement age from 60 to 58 years.
The forecasted manpower distribution over the grades for this alternative
policy is given in table 5.9. This table shows clearly that the desired man-
power distribution (table 5.8) also is not achieved. However, the problem of
the aging organization has been solved (table 5.10). In fact, this example
demonstrates that career prospects that seem acceptable for management do
not always result in a manpower distribution that is acceptable for manage-
ment.
We will now consider a second alternative policy to change the evolution
of manpower distribution over the grades such that the forecasts come close
Table 5.9. Forecasted Manpower Distribution Based on Alternative Policy
Grade
Year G1 G2 G3 Total
1980 50 70 120 240
1981 69 67 108 244

1982 83 64 96 243
1983 101 62 85 248
1984 116 59 75 250
1985 128 58 69 255
1986 137 57 63 257
1987 147 59 57 263
1988 156 62 50 268
1989 165 62 44 271
1990 171 64 39 274
Table 5.10. Forecasted Age Distribution in 1990 Based on Alternative Policy

Age
58-59
56-57 1122333
54-55 1222222333
52-53 23333
50-51 11222222223333
48-49 112222223333
46-47 11112222223333
44-45 11112222233333
42-43 1111111112222223333333
40-41 11111111111111222333
38-39 1111111111111111112
36-37 1111111111111111111111223
34-35 111111111111111111111111122222
32-33 1111111111111111111112222222
30-31 1111111111111111122222
28-29 111111111111111122
26;...27 111111111111
24-25 1111
Note: 1 figure = 1 employee.
to the desired manpower distribution given in table 5.8. The new alternative
policy is:
Recruitment of one middle-level engineer in class 2 and nine high-level

engineers in class 1 each year,
Increasing the promotion fractions from class 3 and decreasing the pro-
motion fractions from class 4,
Transfer of highly experienced personnel from class 5 to management
functions in other divisions as in the first alternative,
Retirement age of 58 years.
Notice that the promotion prospect for high-level engineers will improve
and for middle-level engineers will deteriorate. The forecasted manpower
distribution over the grades is given in table 5.11.
A comparison of tables 5.8 and 5.11 shows that the last alternative gives
a forecasted manpower distribution that is close to the desired distribution.
Of course, this is only one possible alternative. We will not discuss here
other reasonable alternatives.
This case study shows how manpower planning can be executed even
though historical data concerning career possibilities are missing. It also
makes clear why manpower planning for such organizations with highly
specific personnel is relevant (see typology in chapter 1).
Table 5.11. Forecasted Manpower Distribution Based on Second Alternative
Policy
Grade
Year Gl 02 03 Total
1980 50 70 120 240
1981 59 67 108 234

1982 67 63 96 226
1983 75 61 85 221
1984 83 53 79 215
1985 82 52 77 211
1986 80 52 74 206
1987 82 55 65 202
1988 86 53 57 196
1989 84 48 57 189
1990 81 45 58 184
130 CHAPTERS
LEVEL OF AGGREGATION
In this section we consider the choice of function groups to use in the

matching process. On the one hand, to guarantee a certain reliability of the
forecasts, the groups should not be too small. On the other hand, the
groups should not be too large if the matching is to be meaningfuL
Examples of the two extremes are consideration of the entire organization
as one personnel group and consideration of personnel groups as containing
just one individual. In the first case, reliable information can probably be
obtained with respect to the evolution of manpower size, but the question is
whether this information is also valuable for subgroups of the organization.
For instance, if the total number of employees increases by 50/0, this does not
necessarily mean that both the number of engineers and the number of secre-
taries will increase by 5%. More detailed information is usually necessary in
the planning process. The other extreme case, however, is not executable in
practice since it is not possible to estimate the exact requirement for all func-
tions individually for the medium and long term. The minimum level of
aggregation required to maintain a certain reliability of the forecasts depends
on the degree of uncertainty and on the length of the planning horizon. The
maximum level of aggregation allowed to keep the matching meaningful de-
pends on the flexibility of manpower requirement and availability. We will
explain these factors in more detail below. For short-term planning, a de-
tailed approach is desired, whereas the consequences of decisions for the me-
dium and long term usually can be investigated on a higher aggregation level.
As we have seen in chapter 1, manpower planning has to be part of corporate
planning. In corporate planning, the future development of finance, ma-
terials, equipment, and personnel is considered. We already distinguished
three levels of planning: short term (less than one year ahead), medium term
(one to ten years), and long term (more than ten years ahead). The long-term
corporate plan provides goals and restrictions for the medium- and short-
term plan. Considering the short-term decisions, one has to accept the me-
dium- and long-term corporate plan as a restriction. This restriction may be
expressed in terms of total salary costs or in terms of numbers of allowed em-
ployees. Whether the restrictions are given in terms of money or in terms of
allowed personnel depends also on the decision structure of the organization.
This will be discussed later in the chapter.
Choice of the level of aggregation
Two important factors with respect to the level of aggregation are the length
of the planning horizon and the flexibility of manpower requirement and
availability.
The length of the planning horizon. Manpower requirement forecasts are

based on corporate planning. If the length of the planning horizon in-
creases, then the manpower requirement forecasts become less reliable for
the later years of the planning period because of the possibility of future
changes in corporate planning. Also, manpower availability forecasts are
less reliable for a longer planning horizon. Thus for a long planning horiz011l
a high level of aggregation is required, whereas for short planning horizons
more detailed computations are possible. The planning horizon will be con-
sidered in more detail later.
The flexibility of manpower requirement and availability. The degree of
flexibility of the manpower requirement is the degree to which the numbers
of functions and the types of functions can be changed if necessary. This
corresponds to the rigidity of manpower requirement as described in chap-
ter 1 (typology). If the degree of flexibility of manpower requirement is low,
then the level of aggregation has to be low, since otherwise a match on a
high level of aggregation gives no guarantee for a detailed match of require-
ment and availability.
The degree of flexibility of manpower availability is the degree to which
the organization can transfer employees to other functions or locations
(mobility). Dismissing employees of one type and recruiting personnel of
another type is also a form of indirect mobility. In an organization with de-
partments, a high level of aggregation is allowed if the mobility between de-
partments is so high that a matching problem in a subsystem will never arise
if the aggregated system has been matched. This is explained in greater de-
tail in the following example.
EXAMPLE. Consider a manpower system with two groups of functions
on the same organizational level (figure 5.7).
Define:
(XI == wastage fraction from category 1;
f31 == mobility fraction from category 1, indicating the maximal frac-
tion of employees in category 1 who can be transferred to the
other category each year;
ml(t) == the number of required employees in category 1 at time t;
n I (t) == the number of available employees i11l category 1 at time t;
rl(t) == recruitment in category 1 during the time interval (t - 1, t];
b l (t) == the numbers of employees in category 1 who are transferred to
the other category during the time interval (t - 1, t].
For the sake of simplicity we assume the wastage fractions for both cat-
egories to be equal and constant over time: (X = (XI = (Xz. Suppose that the
system can be controlled by recruitment and by transferring employees
132 CHAPTER 5
a
1
S2
/'?
category categor
1 2
Sl
FIGURE 5.7. A manpower system with two cat-

egories of personnel (functions) on the same organi-
zational level. O!j indicates the wastage fraction
from category 1, {3j denotes the mobility fraction
from category 1, and rj (f) indicates the recruitment
in category 1 during the time interval (f - l,f].
from category 1 to category 2, and vice versa. There is no restriction on re-

cruitment. The goal is to minimize the deviations between manpower re-
quirement and availability over a certain planning horizon of, say, Tyears.
We assume the costs of deviations between available and required numbers
of employees to be constant over time and equal for both categories.
We will investigate whether it is possible to solve this matching problem
by using an aggregated model for a long planning horizon and a disaggrega-
tion model for the first planning period. If this is not possible, then a de-
tailed model for the two-category system has to be used.
In the detailed model, the recruitment rl (t) and the numbers of transfers
b l (t) are determined simultaneously for both categories for all planning
years. The detailed model of the two-category system is:
T
Minimize E {Inl(t)
t= I
- ml(t)1 + In2(t) - mit)l} (5.1)
under the conditions:
nl(t) + n2(t) ~ (1 - a) (nl(t - 1) + n2(t - 1», for t = 1,2, ... ,r,

(5.2)
nt(t) ~ (1- a - (31)nt(t - 1), fort = 1,2, ... ,r, (5.3)
n2(t) ~ (1 - a - (32) n2(t - 1), fort = 1,2, ... ,T. (5.4)
Solving this problem gives the solution for nl (t) and n2(t), for all
t = 1,2, ... ,T. The values of rt(t) and bt(t) are determined by the values
of nt(t). For instance, if nt(t) < (1 - a) n](t - 1), then 'l(t) = 0;
r2(t) = nl (t) + n2(t) - (1 - a) [nl (t - 1) + n2(t - I)), and b l (t) =

(1 - a)nl (t - 1) - nl (t); b2(t) = o.
In the aggregated version of the model, total recruitment r(t) is deter-
mined for t = 1, ... , T. The aggregated version of the model is:
T
Minimize E
t= I
Is(t) - [ml(t) + m2(t)]l (5.5)

s(t) ~ (1 - a) s(t - 1), for t = 1,2, ... ,r, (5.6)
where s(t) is the total number of employees at time t. Solving this problem
gives the solution for s(t), for all t = 1,2, ... ,T. The values of r(t) are de-
termined by the values of s(t), namely:
r(t) = s(t) - (1 - a)s(t - 1). (5.7)
The total number of employees, as computed with the aggregated model,
has to be distributed over the categories. In order to determine n I (1) and
n2(1), one may use the following disaggregation model:
Minimize: Inl (1) - ml (1)1 + In2(1) - m2(1)1 (5.8)
nl(1) + n2(1) = s*(1), (5.9)
nl (1) ~ (1 - a - (31)nl (0), (5.10)
n2(1) ~ (1 - a - (32)n2(0), (5.11)
where s*(l) is the optimal value of s(1) as is found when using the aggre-
gated model.
Solving the aggregated model and then solving the dis aggregated model
gives values for nl (1) and n2(1). Values for nl (t) and n2(t) for t > 1 can be
obtained in the same way. Aggregation is allowed if these values for n I (t)
and n2(t) for t > 0 are optimal. This is the case if following conditions are
satisfied:
ml(t) ~ (1 - a - (3dml(t - 1), fort= 1,2, ... ,T, (5.12)
m2(t) ~ (1 - a - (32)m2(t - 1), for t = 1,2, ... ,r, (5.13)
ml(t) :;;; (1 - a)ml(t - 1) + (32m2(t - 1) + s*(t) - (1 - a)s*(t - 1),
fort= I,2, ... ,r, (5.14)
m2(t) :;;; (1 - a)m2(t - 1) + (3lml(t - 1) + s*(t) - (1 - a)s*(t - 1),

for t = 1,2, ... , r, (5.15)
where nl (0) = ml (0), n2(0) = m2(0), and s*( t) is the optimal solution for the
aggregated model.
134 CHAPTER 5
Conditions (5.12) and (5.13) indicate that manpower requirement may

not decrease yearly by more than the number of leavers and the number of
transferable employees for each category. Conditions (5.14) and (5.15) en-
sure that if both categories are sufficiently occupied at time t, then it is pos-
sible to obtain a sufficient number of employees in at least one of the cat-
egories at time t + 1.
For further research on the level of aggregation and its relationship to the
planning horizon in manpower planning, see Nuttle and Wijngaard (1979)
and Wijngaard (1980).
Consider now some numerical results for the example above. The param-
eter values are:
a = 0.1; {3, = 0.2; (32 = 0.1; T = 2;
n,(O) = m,(O) = 1,000; m,(1) = 1,200; m,(2) = 900;
n2(0) = m2(0) = 1,500; m2(1) = 1,300; m2(2) = 1,100.
An investigation of conditions (5.12) through (5.15) shows that aggrega-

tion is allowed. Solving the aggregated model (5.5) and (5.6) gives the fol-
lowing results:
s(O) = 2,500 ; s(1) = 2,500 ; s(2) = 2,250,
which makes a discrepancy in period 2 obvious. Executing the disaggrega-

tion step gives: n, (2) = 900 ; n2(2) = 1,350.
Solving the dis aggregated model (5.1) through (5.4) gives the following
solution:
n,(O) = 1,000; n,(1) = 1,200; n,(2) = 900;
n2(0) = 1,500; n2(1) = 1,300; n2(2) = 1,350.
This solution is indeed identical to the solution of the aggregated model.

The solution in terms of recruitment and transfers is:
b,(1) =0; b,(2) = 180;
b2 (1) = 50; b2 (2) = 0;
r,(1) = 250; r,(2) = 0;
r2(1) = 0; r2(2) = O.
Discussion
The choice of the level of aggregation is in general not unique. A function

can be classified roughly in a three-dimensional way: the organizational
level of the function, the place where the function is executed, and the
organizational sector where the function belongs. Aggregation based on
these dimensions is called vertical aggregation, geographic aggregation, and

sectoral aggregation, respectively.
The example above shows that aggregation is possible if mobility be-
tween the subsystems to aggregate is so high that a matching problem in a
subsystem will never arise if the aggregated system has been matched. The
match of requirement and availability in the aggregated plan guarantees the
match for each subsystem.
Even if the situation is not so ideal with respect to mobility between the
subsystems, the matching of the aggregated forecasts improves the possibil-
ity of obtaining a good future match in each of the subsystems. If there are
no more possibilities for aggregation (sectoral or geographic, for instance),
one can choose the dimension over which the mobility is highest. If sectoral
mobility is higher than geographic mobility, it is probably better to apply
sectoral aggregation.
The conditions fot: vertical aggregation are of the same type as the condi-
tions for horizontal aggregation. But the mobility is asymmetrical in this
case. It is impossible, in general, for people to move to a lower organiza-
tional level. Let, for instance, A and B be two different levels and let the
normal career scheme be that people start in B and move to A after a num-
ber of years. Assume a constant growth in the number of available func-
tions of 100/0 a year. If there are no restrictions on the numbers of recruits,
it is always possible to adapt the number of available people to this growth,
although one can get into difficulties with experience requirements. An-
other difficulty can be the existence of fixed career patterns. If one moves
from B to A after five years of experience, the number of people at level A is
not influenced by recruitment for the first five years. Thus each restriction
on the flows between B and A is a handicap for aggregation.
As mentioned already, there is also a relation between the level of aggre-
gation and the organizational structure. On the central level of an organiza-
tion, one usually considers a high level of aggregation since it is impossible
to solve possible discrepancies on a lower level of aggregation. Therefore,
budgets are expressed, for instance, in total salary costs. The number of al-
lowed employees in each category cannot be determined. This has to be
done on the lower level of the organization. In the Dutch police organiza-
tion, for instance, a total budget is given for each of the local forces. The
local organization has to fill in the concrete numbers and types of police
officers.
PLANNING HORIZON
As we have seen, there is a relation between the level of aggregation and the
length of the planning horizon. For organizations with a high degree of
136 CHAPTER 5
flexibility (a high degree of mobility), only a short planning horizon will be

required.
In this section we will discuss in particular the relation between planning
horizon and uncertainty. A simulation experiment has been executed in
order to illustrate this relation.
Rationale for a short planning horizon
There are two reasons why a short planning horizon may be sufficient:
1. Large flexibility of manpower requirement and availability,

2. Large uncertainties with respect to the forecasts.
Large flexibility of manpower requirement and availability. The first rea-

son corresponds to the requirement of flexibility for high levels of aggrega-
tion. In situations of a high degree of flexibility of manpower requirement
and availability, it is relatively easy to obtain a match of manpower demand
and supply. If wastage is high, for instance, or if the organization is gww-
ing, sufficient functions will be available for the employees to satisfy their
career expectations. In these situations a short planning horizon is suffi-
cient.
Large uncertainties with respect to the forecasts. In fact, two types of

uncertainties are important: the unpredictable variability in manpower
availability and the unpredictable variability in manpower requirement. In
chapter 3 we considered possible errors in manpower availability forecast-
ing. As has been shown (table 4.4), the standard deviations of the forecasted
manpower distribution over the grades are relatively small for the medium
term. The unpredictable variability in manpower requirement, however,
may be much more important. Many factors will influence this require-
ment: market share, technology, economic situation, and so on. In the case
of large uncertainties, a short planning horizon is sufficient.
Simulation experiment
We will use a simulation experiment in order to illustrate the relation be-

tween the length of the planning horizon and the variability (both predict-
able and unpredictable) of the requirement.
Suppose that we have one category of personnel (for instance one grade
or the entire personnel size) with recruitment and wastage. The wastage
fractionp is known and constant, say 0.05. The number of employees in the
category is assumed to be controlled by a rolling plan procedure. In order to
simulate such a controlled process, one has to indicate not only how the real
manpower requirements and availabilities originate, but also how the fore-
casts are obtained and how the matching is executed.
We distinguish the following phases of a rolling plan procedure:
1. Forecasting of manpower requirements,

2. Forecasting of manpower availabilities,
3. Matching of forecasted requirements and availabilities by means of
recruitment.
Define:
m(t, t + i) == estimated manpower requirement for time t + i made at

time t;
m(t) == manpower requirement at time t (realization);
n(t) == manpower availability at time t (realization);
r(t) == recruitment in the time interval (t - 1, t].
Forecasting of manpower requirement. We assume that the realization of

manpower requirement m(t) is obtained by the following process:
(5.16)
where, m\(t) is the unknown component of the manpower requirement;
m2(t) is the given, fluctuating component of the manpower requirement; c
is constant; and m\(t) is supposed to follow an autoregressive process of
order 1 (see formula [2.6]).
(5.17)
where 01 is the given process coefficient and the noise €t has the following
characteristics:
lE€t = 0 and IE (€t • €t - k) = 1~<T2 for k I- 0,

for k = 0, (5.18)
for t = 0,1,2, ....
In the simulation process, we will vary the values of A in order to investi-
gate the influence of a different degree of uncertainty on the length of the
planning horizon (0 ~ A ~ 1).
m2(t) is also supposed to follow an autoregressive process of order 1 but
is known for T years ahead according to:
(5.19)
138 CHAPTER 5
where the noise Ot has the following characteristics:

JO for k 1= 0,
lEot = 0 and IE (Ot • Ot - d = 1(1 _ A)(}"2 for k = 0,
for t = 0,1,2, . . .. (5.20)
It is assumed that the Ot are known T years in advance. In a simulation
procedure we obtain the manpower requirement forecasts at time t accord-
ing to the following formula:
fori= 1,2, ... ,T,
for t = 0,1,2, ... , (5.21)
where ml(t,t + i) == minimal mean square error forecast for the unknown
component of the manpower requirement at time t + i, made at time t, and
m2(t + i) == given component of manpower requirement at time t + i. The
forecasts, ml(t,t + i) for the unknown component can be obtained by:
(5.22)
Forecasting oj manpower availabilities. Wastage and recruitment bear in-

fluence on manpower availability. We assume the wastage fractionp to be
known and constant over time (deterministic situation). Thus:
n(t + 1) = n(t) . (1 - p) + r(t + 1), for t = 0,1,2, .... (5.23)
Matching by means oj recruitment. The matching of manpower require-

ment and availability is executed by means of recruitment. The recruitment
for time 0,1,2, ... , is obtained by minimizing
T
,E1 Im(t,t + i)
1=
- n(t + i)l· (5.24)
In order to evaluate the outcomes of the simulation process for different

planning horizons T, we must have a performance criterion. Let M be the
number of periods over which the rolling plan is simulated. The following
performance criterion is used to evaluate the outcomes:
1 M
_. E In(t) -
M t= 1
m(t)I. (5.25)
Results of the simulation for different values of A, (}"2, and T are given in
table 5.12. The simulations have been executed for M = 500 in five runs of
100 trials, each with a "warming-up" period of 50 trials. Because of the lin-
ear objective function (5.24), the value of the criterion (5.25) changes only
for all odd values of the planning horizon.
Table 5.12. Results of Simulation Experiment for Different Values of cl and A
Value of Performance Criterion for Planning Horizon

2
(J A 1 2 3 4 5 6 7 8 9 10 11 12
0.00 13.27 13.27 '10.03 10.03 9.26 9.26 8.90 8.90 8.85 8.79 8.79 8.75
.25 15.71 15.71 13.65 13.65 13.34 13.34 12.82 12.82 12.71 12.60
400 .50 17.09 17.09 15.71 15.71 15.43 15.43 15.29 15.29 15.17 15.16
.75 18.37 18.37 17.63 17.63 17.41 17.41 17.33 17.33 17.24
1.00 19.72 19.72 19.48 19.48 19.45
0.00 3.43 3.43 2.55 2.55 2.44 2.44 2.41

.25 5.78 5.78 5.16 5.16 5.09
100 .50 6.79 6.79 6.44 6.44 6.40
.75 7.63 7.63 7.46
1.00 8.59
0.00 0.50 0.50 0.42

.25 2.11 2.11 2.06
25 .50 2.85 2.85 2.82
.75 3.45
1.00 4.05
Note: The last number of each line indicates the minimum value of the criterion; an increase of Twould not improve the cri-
terion value. The parameters are: p = 0.05; IX = 0.8; c = 100.
140 CHAPTER 5
Table 5.12 clearly shows that in case of large but predictable fluctuations
of the manpower requirement (A - 0, (12 large), large planning horizons are
necessary, whereas in case of large but unpredictable fluctuations of the
manpower requirement (A - 1, (12 large), short planning horizons are suffi-
cient. Small fluctuations «(12 small) will also result in short planning hori-
zons.
Notice that many extensions can be investigated. Here we only consid-
ered one category of personnel (high level of aggregation). If the flexibility
of manpower availability is low-for instance, because of the existence of
fixed career patterns-then the planning horizon will lengthen. However,
we will not discuss such aspects in more detail. (See also Nuttle and
Wijngaard 1979.)
REQUIREMENTS FOR PERSONNEL REGISTRATION
In this section we will briefly discuss the relation between the introduction
of manpower planning-and more specifically of systems such as FOR-
MASY -and the personnel registration of an organization. As we have seen
throughout this book, manpower planning requires a lot of information,
both about the personnel and about the functions. First, information about
the numbers and types of functions, now and in the future, is necessary.
Therefore, a (automatized) function registration system can be helpful. In
such a system all functions are recorded together with information about
the organizational level and the necessary qualifications or training. This
function registration system agrees with the organization scheme of this mo-
ment. It may also contain information about the future development of the
functions-for example, the time at which a function will disappear or
when new functions will arise. An example of part of such a registration sys-
tem for a Dutch defense organization is shown in table 5.13.
The registration system shown in table 5.13 stems from a Dutch defense
organization. It fits the requirement of that organization. Other organiza-
tions may want to include different characteristics, but this system indicates
what such a registration system may look like.
Second, information is required on the number of employees. This may
be obtained by means of a (automatized) personnel registration system that
stores information about the actual employees. It may contain such data as
name of the employee, date of birth (age), sex, family status, date of entry
in the organization (length of service), grade at entrance, date of last pro-
motion (grade seniority), training, experience, expectations with respect to
functions, and actual functions. For an adequate use of FORMASY it is
helpful if such data are available. A personnel registration system gives in-
Table 5.13. Function Registration System for Dutch Defense Organization
Number oj Similar
Organization Function Functions Function Function Function
Element Code Serial Grade Function Function Authority Requirements
(Department No.) Number Peacetime Wartime (Rank) Name Code Level and Remarks
21.000 Manpower Planning Department
010 col. off. P 05027 63 Also controller
020 registr. 850105 63 Secr. also diary
registr. 850103 63 Secr. also diary
21.000 Planning Section
010 It. col. off. P 05026 62
030 maj. off. Klu 09900 63 To be replaced
by probationer
040 cap. off. Bez. 050043A 42
050 maj. off. P. 05025 42 Automation
knowledge
142 CHAPTER 5
formation about the actual numbers and types of employees and offers an
opportunity to compute historical transition fractions by comparing the
positions of employees, for instance, once a year. As we have seen in chap-
ter 4, these data are necessary in order to investigate the consequences of a
continuation of the historical policies on the evolution of the manpower dis-
tribution. Both the historical promotion and the turnover rates have to be
adjusted each year.
REQUIREMENTS FOR THE PERSONNEL DEPARTMENT
In our view, the personnel department should have the responsibility of

examining the evolution of manpower requirement and availability, of
noticing possible discrepancies, and of proposing or taking actions in order
to solve the matching problems. Thus this department should not only have
a passive role in the registration of functions and employees but also an ac-
tive role in the design of policies with respect to personnel.
The introduction of a manpower data base and a manpower planning
system such as FORMASYimplies certain tasks for the personnel depart-
ment. In the first place, the structure of the function and personnel registra-
tion system has to be (re)designed-in such a way that it can be used for
planning purposes-in cooperation with the computer; second, to detect
trends and systematic changes in both the organizational structure and the
personnel distribution, it is necessary that the personnel department update
the function and personnel registration system, make occasional surveys of
personnel flows and function changes, and investigate future consequences.
The introduction of a computer-aided manpower planning system such as
FORMASY leads to requirements with respect to the employees who have
to use this system and the staff members who have to interpret the results.
The introduction, therefore, has to be done in such a way that possible resis-
tance to computer-aided systems can be overcome, especially by showing
that all computations can be checked easily by hand and are only executed
by means of a computer program to save time.
The acceptance of these requirements implies that the management of a
personnel department must be qualified to understand the value of man-
power planning in general and of systems like FORMASY in particular.
Moreover, one has to be familiar with computerized function and personnel
registration systems.
CONCLUDING REMARKS
Many restrictions both from outside the organization, such as laws, labor
market, labor costs, and so on, and from within, such as personnel de-
mands, career expectations, commitment, and so on, enforce the necessity

of manpower planning. The planning of personnel has been a neglected part
of corporate planning for a long time.
In large organizations, quantitative methods are needed to forecast the
development of the numbers and types of functions as well as to forecast the
numbers and types of employees. The design of policies to avoid discrepan-
cies can be done best by considering the consequences of alternatives. These
consequences may be computed directly by using a manpower planning sys-
tem such as FORMASY.
By accepting such a manpower planning procedure, management can
take actions in time. This is important not only for the organization or man-
agement but also for the individual employees. Manpower planning systems
offer possibilities for dialogue between management and personnel (or
trade unions) in order to prevent sudden conflicts. In this way the organiza-
tion can function better and the employees can get rational information
about career possibilities.
6 A CASE STUDY
The manpower planning system FORMASY and its possible use have been
discussed in chapters 4 and 5. The use of FORMASY for practical man-
power planning will be illustrated in this chapter by means of a case study.
We will describe the way FORMASY is used in the manpower planning pro-
cess at the manpower planning department of the Royal Netherlands Air
Force.
All policies, models, and figures mentioned in this chapter are fictitious.
However, they have been chosen in such a way that a certain correspon-
dence to the current situation has been obtained.
MANPOWER PLANNING WITHIN THE ROYAL

NETHERLANDS AIR FORCE
The personnel plan of the Royal Netherlands Air Force is integrated into the
defense planning process, together with the finance and the materials plan.
This personnel plan indicates the numbers and types of employees in differ-
ent functions for the next ten years. The plan is adjusted every two years.
Information about corporate planning and the way personnel requirement
144
A CASE STUDY 145
forecasts are obtained has been given in chapter 2. Therefore, manpower re-
quirement forecasting will not be considered here.
Manpower policies within this organization aim at the following goals:
1. An adequate filling-both quantitatively and qualitatively-of the

peacetime organization as well as of the wartime organization. This
goal implies that a given number of employees in each type of func-
tion have to be reached in a peacetime situation and also that an
additional number of reservists (mobilization force) have to be main-
tained for wartime situations.
2. A planned development of the careers of the regular staff. Promo-
tions are often made after a fixed number of years in a certain rank,
but only if adequate higher-ranked positions are available. For this
reason, career prospects have to be controlled, especially since at the
moment an irregular age distribution exists.
3. Flexibility in the structure of the manpower composition in peace-
time. The possibility of recruitment of short-term volunteers and
conscripts to fill temporary staff shortages contributes to the main-
tenance of flexibility.
The manpower planning department controls the careers of the regular

staff and is involved in the determination of the numbers of recruits of sev-
eral categories of military personnel. The management of the air force is in-
terested in the evolution of the actual manpower distribution and in the
question of whether discrepancies will arise between staff supply and de-
mand.
Special problems are caused by the actual irregular age structure. The ir-
regular structure has been due to changes in the size of the organization and
to changes in recruitment, which occurs in only a few age classes. The
organization has known a period of growth, followed by several years of
about constant size, and has grown again in the last few years. This has re-
sulted in an age distribution with a high fraction of older employees and a
high fraction of younger employees, but relatively low numbers of em-
ployees in the intermediate groups (see table 6.2). Some consequences of
this age structure are the relatively high numbers of retirements in the next
few years, good career prospects for employees in the intermediate age
groups, and poor promotion possibilities for the younger personnel group.
Therefore, it is important to develop strategies that restore the age distribu-
tion-for example, by recruitment in different age groups, flexible retire-
ment ages, changes in career prospects, and so on.
Other problems of matching manpower supply and demand concern the
training level of the employees, their qualification level, and so on. Thus,
146 CHAPTER 6
training and development planning also is important. Allocation planning is

another important topic.
Rising salary costs and matching problems as discussed above have re-
sulted in a growing interest in manpower planning instruments within this
organization.
Three time horizons of manpower planning can be distinguished:
1. Long-term planning. This type of planning deals with the construc-
tion of equilibrium structures (manpower pyramids) for all groups.
This is executed by assuming stable manpower requirements and
then developing stable manpower populations. Long-term age dis-
tributions, career paths, and the numbers of recruits are thus deter-
mined.
2. Medium-term planning. Medium-term planning gives the forecasts
for the evolution of the number of employees in the various cat-
egories in the subsequent ten years and the determination of recruit-
ment of regular staff and short-term volunteers. The planning is also
used to adjust career paths and to provide forecasts for the develop-
ment of personnel expenses.
3. Short-term planning. This type of planning delivers the forecasts of
the evolution of manpower availability and of the requirements for
the following three years-in monthly figures-in all types of func-
tions and all categories of personnel. These forecasts are used for the
determination of the numbers of recruits for short-term volunteering
and of conscripts. These forecasts also provide the calculation basis
for the air force's financial estimates.
This case study presents some examples of these planning activities for the
peacetime strength of military staff.
Manpower planning is executed broadly for the different staff groups
(officers, noncommissioned officers, and so on) as well as in detail for the
different task groups (air operations, ground operations, electronics, and so
on). The next three sections cover the broad planning procedure, followed
by a description of the detailed planning.
MODEL STRUCTURE
We will examine the model structure for the group of officers. For the other
staff groups similar models have been constructed.
The group of officers is divided into five classes according to rank or
grade (some higher grades are lumped together because of small numbers):
lieutenant, captain, major, lieutenant-colonel, and finally colonel or higher
ranks. Two training levels can be distinguished. The officers for special ser-
A CASE STUDY 147
vices (O.S.S.) have two years of training. They are partly selected from the
lower ranks. The officers who have completed the Royal Military Academy
(R.M.A.) have four or five years of training.
The maximum grade for the O.S.S. officers is normally the rank of lieu-
tenant-colonel. However, some of the officers with this training who have
attended the course for field officers (about 150/0) can be promoted to the
training level of R.M.A. officers when they have the rank of captain. This
enlarges the chances fOl: fast promotion and reaching the highest rank of
colonel. In the grades of major and lieutenant-colonel, the R.M.A. officers
are divided into three qualification levels-namely, level 1 for those who do
not have the potential ability to reach the rank of colonel, level 2 for those
who have this potential ability, and level 3 for the officers with an air force
staff license or with a university degree.
A career scheme for the officers is given in figure 6.1. In each class, the
theoretical possibility of twenty-four grade seniorities is included. In the fig-
colonel or higher grades

(COL +)
~~::::~~:: coJ~9I G
major (MAJ)
o (6.7)
[boa)
5
captain (CAP)
lieutenant (LT)
T Promotion
O.s.s. R.M.A. 1 Recruitment
FIGURE 6.1. Career patterns for officers and average grade seniorities until pro-
motion.
148 CHAPTER 6
ure, average grade seniority until promotion is denoted in parentheses.

Transitions with ® are selection promotions (that is, not all officers make
these transitions). The ratio between O.S.S. recruits and R.M.A. re-
cruits = 5:3. This relationship is the result of the overall desired manpower
distribution and career paths. Classes 6 and 10 are designated for officers
without potential ability for colonel (qualification level 1), classes 7 and 11
for officers with potential ability for colonel (qualification level 2), and
classes 8 and 12 for officers with a university degree or staff license (qualifi-
cation level 3).
The average grade seniorities until promotion as well as the input data
needed for the use of the FORMASY system for this model have been ob-
tained directly from the personnel registration system.
LONG-TERM PLANNING
By means of a special option of the FORMASY system, one may compute

the steady-state distribution over the grades for some given recruitment (see
"Steady-State Distribution of Employees" in Chapter 4). The recruitment
has been chosen such that the ratio between O.S.S. recruits and R.M.A. re-
cruits is 5:3. The computed steady-state distribution is compared with the
desired distribution. Subsequently new recruitments are computed such that
the desired distribution is obtained, assuming that the given ratio will not
change.
The historical age distribution of the recruits is shown in figure 6.2. The
actual manpower distribution and the steady-state manpower distribution
.t 0.5.5. recruits
percentage 30
R.M.A. recruits
20
~
.t
10 .r .t
.f I
., .t .t .t-
o LI\ r rf t ot i¥.f "' ..... ~ age
"20 25 30 35 41'.l
FIGURE 6.2. Age distribution of recruited O.S.S. and R.M.A. officers in the last
five years.
A CASE STUDY 149
are presented in table 6.1. The steady-state distribution is based on recruit-

ment of forty-five O.S.S. and twenty-seven R.M.A. officers yearly and on
the assumption that historical promotion and turnover rates will remain
constant in the future.
The desired manpower distribution over the grades is assumed to be
equal to the actual distribution. Table 6.1 shows a forecasted overoccupa-
tion of lower grades and a shortage of personnel in the higher grades. The
reason for this result is the eXIsting number of years until promotion takes
place for each grade. A few years after World War II the Royal Netherlands
Air Force recruited many young officers, and these employees will retire iIll
the next few years at the age of 55 (see table 6.2). As a consequence, there
have been only a few vacancies in the higher ranks in recent years, and this
has resulted in a slowing down of promotions. If the existing career policies
were maintained, then the forecasted shortage of staff in the higher ranks in
the long-term future would be realized.
It is possible to decrease the average time until promotion without chang-
ing the fraction of officers who are promoted. This would speed up promo-
tions without affecting the selection criteria for promotion. By such a
speeding-up of promotions, one might obtain a steady-state distribution of
officers over the ranks that looks more like the actual distribution than the
steady-state distribution of the actual promotion policy (see table 6.1).
After some trial-and-error efforts, an alternative can be designed as indi-
cated in figure 6.3 (compare with figure 6.1). In the figure, the numbers in
parentheses denote the shortening of average grade seniorities until promo-
tion, compared with the original averages in figure 6.1. In this alternative
policy, the highly qualified officers make even faster promotion than in the
policy of figure 6.1.
Table 6.3 compares the actual distribution of officers over the ranks with
the steady-state distribution for the newly designed promotion policy of
figure 6.3. This alternative career path seems acceptable. The long-term age
distribution that will result is shown in table 6.4. Compare this age distribu-
tion with the current one exhibited in table 6.2.
Table 6.1. Actual and Steady-State Manpower Distribution of Officers
Grade
LT CAP MAJ LT COL COL + Total
Actual Distribution 318 466 467 220 85 1556

Steady-State Distribution 435 519 396 158 54 1562
Table 6.2. Actual Age Distribution of Air Force Officers
Age
54 33334444455555
53 22333333333333444444444445555555
52 122223333333333333334444444444444444455555555555555
51 12222333333333344444444444444455555
50 222223333333334444444444444444555555
49 122222333333333333333344444444444555
48 122222233333333333333333333444444444455
47 112222233333333333333344445
46 111112223333333333333333444444
45 112222333333333333344444
44 1122222333333333333333333444444
43 1122223333333333333333444
42 12222222223333333333333334444
41 112222222222333333333334
40 1122222222222233333333333
39 12222223333333333
38 112222222222333333333
37 111112222222222222222222233333333
36 111111222222222222222222223333
35 1111122222222222222223333
34 1111111111222222222222223
33 111111111111122222222222222222
32 11111111111112222222222222222
31 11111111111222222222222222
30 111111122222222222
29 11111111122222222
28 1111111111122222
27 111111111222
26 1111111111111111
25 11111111111
24 111111111
23 11111
22
21
Note: 1 figure = 2 employees; 1 = LT; 2 = CAP; 3 = MAJ; 4 = LT COL; 5 = COL+ .
LT COL
MAJ
CAP
LT
O.s.s. R.M.A.
FIGURE 6.3. Alternative career pattern for officers.
Table 6.3. Actual Distribution of Officers Compared with Steady-State Distri-

bution for Newly Designed Policy
Grade
LT CAP MAl LT COL COL + Total

152 CHAPTER 6
Table 6.4. Steady-State Age Distribution Based on Alternative Career Scheme

for Officers
Age
54 23333333333444444445555
53 2233333333334444444445555
52 2233333333334444444445555
51 223333333333444444445555
50 22233333333333444444445555
49 22233333333333444444445555
48 222333333333334444444555
47 12223333333333334444444555
46 122233333333333334444444555
45 1122233333333333334444444555
44 1122222333333333333444444455
43 11222222333333333344444455
42 112222222333333333344444455
41 1122222222223333333334444455
40 1122222222222333333333344445
39 112222222222223333333333444
38 12222222222222233333333344
37 1122222222222222233333333344
36 1122222222222222233333333334
35 1111222222222222223333333333
34 111111122222222222233333333
33 1111111122222222222222333333
32 111111111122222222222222333
31 1111111111122222222222222223
30 1111111111122222222222222
29 11111111111222222222222
28 1111111111111222222222
27 1111111111111222222
26 111111111111111222
25 111111111111111
24 11111111111
23 111111
22 111
21
Note: 1 figure = 2 employees; 1 = LT; 2 = CAP; 3 = MAJ; 4 = LT COL; 5 = COL + .
Another type of policy that can be considered in order to obtain an

appropriate steady-state distribution over the ranks is an increase of the re-
tirement age from 55 to, say, 58 years. The result of such a policy, when his-
torical career schemes and turnover fractions are maintained, is given in
A CASE STUDY 153
Table 6.5. Actual Distribution of Officers Compared with Steady-State Distri-

bution for Promotion Policy of Figure 6.1
Grade
LT CAP MAl LT COL COL + Total

table 6.5. The recruitment consists of forty O.S.S. officers and twenty-four
R.M.A. officers each year. With this increased retirement age, only small
changes in promotion policies would be necessary in order to obtain the de-
sired steady-state distribution.
MEDIUM-TERM PLANNING
Apart from recruitment of regular personnel, the air force has an oppor-
tunity to recruit volunteers on the officers' level in the grade of lieutenant
on a contract basis of four or six years. After this period they leave the or-
ganization. In the discussion above we calculated the recruitment numbers
for regular officers necessary to maintain total strength in the long run. If
this recruitment policy is really used, then the unfavorable age distribution
of this moment will disappear in the long run. However, in the short and
medium term this would lead to a shortage of officers (because of many re-
tirements). This shortage can be met by the recruitment of more volunteers.
The necessary extra recruitment of lieutenants for the next five years has
been calculated with FORMASY and is shown in figure 6.4 for the alterna-
Humber of extra
recruits
40
•
30
• •
20
10
o year of recruitment
1979 1980 1981 1982 1983
FIGURE 6.4. Recruitment of volunteers needed to keep the total strength con-
stant, while using the promotion policy of figure 6.3 and the original recruitment
policy of table 6.1.
154 CHAPTER 6
Table 6.6. Forecasted Distribution of Officers Including Volunteer Recruitment

Described in Figure 6.4
Grade
Year LT CAP MAJ LTCOL COL + Total
1979 318 466 467 220 85 1556
1980 168 546 516 260 88 1578

1981 191 547 480 263 82 1564
1982 271 516 458 250 61 1556
1983 313 502 451 234 55 1556
1984 374 459 459 213 51 1556
tive career scheme of figure 6.3. The resulting distribution of officers over
the ranks for the next five years is given in table 6.6.
Table 6.6 shows clearly that in the next two years the total number of
officers would be slightly more than is allowed with the alternative policy.
Moreover and worse, the distribution of officers over the grades differs
considerably from the actual distribution. This means that the alternative
career scheme that satisfies in the long term does not provide an appropriate
distribution of officers over the ranks in the medium term. One reason for
the irregular development of the distribution over the ranks is the actual dis-
tribution of grade seniorities. Because of fluctuating recruitments in the
Table 6.7. Actual Distribution over Grade Seniorities for Officers

Grade
Grade
Seniority LT CAP MAJ LTCOL COL +
1 8 55 48 26 5
2 35 39 52 28 8
3 48 43 47 28 9
4 49 54 71 25 10
5 89 73 0 28 6
6 31 53 55 29 2
7 57 48 48 20 10
8 0 32 39 6 4
9 0 19 40 8 7
10 0 14 29 9 5
11 0 24 6 3 4
~12 1 12 32 10 15
A CASE STUDY 155
actual scheme alternative scheme
FIGURE 6.5. Left: example of an actual promotion scheme for some grade; 500/0
of the officers with grade seniority 7 meet the promotion requirements and are pro-
moted. Right: example of an alternative for the left-hand scheme; promotions have
been spread over grade seniorities 6, 7, and 8. By promoting 10% of the officers
with grade seniority 6, 40% of the officers with grade seniority 7, and 8% of the of-
ficers with grade seniority 8, one would maintain the average time until promotion
and the total fraction of promoted officers.
past, this distribution is very unbalanced (see table 6.7). Promotion occurs
mostly at a fixed grade seniority, which results in heavily varying promotion
flows. A way to meet this problem in the future is by ceasing to promote
officers at a fixed grade seniority. By spreading the promotions from a
grade over a block of seniorities over three years (see figure 6.5), for in-
stance, the promotion flows would stabilize somewhat. The total percentage
of officers who are promoted can remain the same, but a distinction in abili-
ties can be made in this way.
SHORT-TERM PLANNING
In short-term planning one considers the number of conscripts and volun-

teers needed monthly to keep the air force continuously at full strength. Of
course, short-term planning is of interest for the group of officers only inso-
far as it concerns the recruitment of volunteers in the grade of lieutenant.
Further promotions cannot occur for this category of officers.
For short-term planning a special version of FORMASY has been devel-
oped in which promotions are not considered (the only possible flows are
turnover, retirement, and recruitment). This program deals with monthly
figures so that necessary recruitments of volunteers and conscripts can be
computed in detail. The same types of tables and histograms as shown for
medium-term and long-term planning can be obtained for short-term plan-
ning.
DETAILED PLANNING
Once average career schemes for the various staff groups are known,
detailed planning is necessary in order to determine whether manpower
156 CHAPTER 6
Table 6.8. Medium-Term Forecasts for Distribution of Ground Operations

Officers
Grade
Year LT CAP MAl LTCOL COL + Total
1979 105 128 135 53 8 429
1980 80 135 137 62 7 421

1981 67 139 123 66 9 404
1982 32 163 107 63 8 373
1983 16 167 106 64 7 360
1984 8 156 107 58 8 337
availability and requirement are matched for each task group. As an exam-
ple we consider the officers of the ground operations task group. Medium-
term forecasts, assuming no recruitment and using the historical promotion
scheme for this task group, are given in table 6.8.
The planning goal is to maintain, as much as possible, the actual distri-
bution over the ranks in the near future. In order to attain this goal, the fol-
lowing changes in the average promotion patterns are considered:
Delay all promotions in the MAJ grade by two extra years by shifting the
promotion fractions appropriately;
Accelerate all promotions in the CAP grade by one year by shifting the
promotion fractions appropriately;
Delay all promotions in the LT grade by one extra year.
The forecasted result of this operation corresponds to the desired man-

power distribution (apart from recruitment) as shown in table 6.9. The
Table 6.9. Forecasts for Officers of Ground Operations Task Group

Grade
Year LT CAP MAl LTCOL COL + Total
1979 105 128 135 53 8 429
1980 103 104 149 57 7 420

1981 79 120 140 56 9 404
1982 67 118 129 52 8 374
1983 31 136 127 57 7 358
1984 16 130 130 54 7 337
A CASE STUDY 157
needed recruitment can easily be computed now, as was shown in the discus-
sion on medium-term planning.
A comparison of this alternative career scheme for the officers of a task
group with the proposed career scheme for the officers (compare with sec-
tion "Medium-Term Planning") shows very clearly that an "ideal" career
pattern for the total staff group is not always good for each task group.
Especially if officers cannot be transferred to other task groups, many
problems will arise (compare with "Level of Aggregation" in chapter 5).
SUMMARY
In this case study a survey has been given of the manpower planning process
within the Royal Netherlands Air Force. The usefulness of the FORMASY
system has also been explained. Long-term planning is based on "ideal"
grade and age structures in order to determine necessary recruitment of
regular staff, avoiding irregular age distributions in the future. Medium-
term planning deals in particular with extra recruitment (that is, apart from
long-term recruitment) in the next ten years, needed partly because of fluc-
tuations in retirements. Furthermore, alternative career schemes are stud-
ied. Short-term planning concerns recruitment of volunteers on a contract
basis and of conscripts and is based on monthly figures. Furthermore, we
have described how detailed planning has been worked out for a certain task
group. Such excercises make it possible to compare career opportunities for
officers in different task groups and can help to reveal possible discrepan-
cies so that actions can be taken in time.
It has to be remarked that we have illustrated the manpower planning
process by the development of alternative policies in order to adjust the evo-
lution of the distribution of officers over the ranks to the given manpower
requirement estimations. Notice, however, that the matching of availability
and requirement may also lead to changes of the requirements.
APPENDIX
The appendix presents a summary of the system for the medium-term plan-
ning of the Royal Netherlands Air Force; see "Medium-Term Planning" in
chapter 6. Some results may differ from the results given in that section
since some transition fractions in the data base have been changed. The
notes at the end of this appendix give further explanations with respect to
the text. The symbol -> indicates the answers that have been typed in by the
user. Here we have used the English version of FORMASY.
159
160 APPENDIX
.. RUN FORMASytl)
'tRUNNING 0844
*******************************************************************
FORMASY IS AN INTEf~ACnVE COMPUTER PROGRAM FOR MANPOWER PLANNING,
DEVELOPED AT EINDHOVEN UNIVERSITY OF TECHNOLOGY
*******************************************************************
TYPE IN THE NAME OF THE DATA-BASE ?
:If:'i)
.. ell:'F I CERf,. (2)
CONMENTS : l3)
TEST DATA 'OFFICERS', ROYAL NETHERLANDS AIRFORCE, 1979
TYPE IN THE LENGTH OF THE FORECASTING PERIOD ?

+ 5
TYPE IN THE CODES, F~EASE ?

.. HELP(4)
THE POSSIBLE CODES ARE :

INPUTPRINT PRINTING OF INPUT DATA
EXTRAPRINT : PRINTING OF CAREER PROSECTS
RECRU : INPUT OF RECRUITMENT ,
SUBTOT : INPUT OF DESIRED SUBTOTALS IN TABLES~5)
AFTER n~E LAST CODE, TYPE IN ONE FIGURE ! (6)
.. INPUTPI,(INT
.. EXTI'~APRINT
.. 4
PRIN1ING OF INPUT DATA O)
NUMBER OF GRADES (FUNCTION GROUPS)

NUMBER OF QUALIFICATION GROUPS 4
NUMBER OF TRAINING GROUPS = 2
NUMBER OF GRADE SENIORITIES IN ALL CLASSES 12
NUMBER OF CLASSES = :14
NUMBER OF TRANSITIONS BETWEEN CLASSES = 26
************************************************************
NUMBERS OF THE CLASSES
NlJMBEf~ GRADE QUALIFICATION GROUP TRAINING GROUP

LT
2 LT 2 2
3 2 CAP
4 2 CAP ::~ :~~
5 3 MAJ
6 3 MA,j ~~~ ~~~
7 3 MAJ :3 2
APPENDIX 161
8 3 MA.J 4 2
9 4 lTCOl 1
10 4 LTCDL.. 2 2
11 4 lTCOl 3 2
12 4 LTCOl 4 2
L5 ~.i COL+ 4 2
14 0 0 0 (8)
************************************************************
TRANSITION PERCENTAGES FROM CLASS ••• TO CLASS •••
FOR GRADE SENIORITY •••
1. 3. 1 : 0.0 1, 14, 1: 1.0

1. :3, 2: 0.0 1, 14, 2: 1.0
1. 3, :3 : 0.0 1. , 14, 3: 1.0
1, :3,4: 0.0 1, 14. 4: :1..0
1. , :3,5: 0.0 1, 14, 5: 1.0
1, 3, 6: 0.0 1. , 14, 6: 1.0
1, :5, 7: 99.0 1, 14, 7: 1.0
1, , 8:
:~ 99.0 1, 14, 8: 1.0
1, 3, 9: 99.0 1, 14, 9: 1.0
1. 3dO: 9'1.0 b 14.10: 1.0
1, 3.11: 99.0 1, 14,11: 1.0
1, 3,12: 99.0 1, 14.12: 1.0 (9)
END OF INPUTPRINT
************************************************************
PRINTING OF CAREEf~ PI:WSPECTS UO )
TRANSITION
MIN. GRADE SENIORITY "' 7.0

MAX. GRADE SENIORITY 7.0
AVERAGE GRADE SENIORITY= 7.0
TRANSITION 2
MIN. GRADE SENIORITY 1.0


PROMOTION rURNOVER SrAYERS
AFTER 12 YEARS: 93.2 6.8 0.0
TOTAL 93.2 6.8 0.0
AVERAGE TIME SPENT IN CLASS 1 : 6.8
TRANSITION 3

Tf~ANSITION 4
MIN. GRADE SENIORI1Y t.o

162 APPENDIX
TRANSITION 25
AVERAGE GRAnE SENIORITY", 2.8
PERCENTAGE OF ENTRANTS IN CLASS 1.2 WHII MAKE FOLLOWING TRANSITIONS
PROMOTION TURNOVER STAYERS
AFTER 12 YEARS: 93.3 6.7 ().O
TOTAL 93.3 6.7 0.0
AVERAGE TIME SPENT IN CLASS 12: 4.5
TRANSITION 26
MIN. GRADE SENIORITY = 1.0
MAX. GRADE SENIORITY =12.0
AVERAGE GRADE SENIORITY=66.7
TURNOVER STAYERS
AFTER 12 YEARS: 16.6 83.4
TOTAL : 100.0 0.0
AVERAGE TIME SPENT IN CLASS 13 :66.7
DO YOU WANT TO CHANGE THE TRANSITION PERCENTAGES ?

... NO
TYPE IN THE ACTUAL RETIREMENT AGE AND THE RETIREMENT AGES

FOR THE YEARS 1980 - 1984 ?
DO YOU WANT THE MANPOWER DISTRIBUTION FORECAST? TYPE IN: 0

DO YOU WANT THE MANPOWER DISTRIBUTION/ AGE FORECAST? TYPE IN: 1
... 0
MANPOWER DISTRIBUTION IN PLANNING PERIOD
GRADE
YEAf? * I... T CAP MA,J LTCOL COl... + * TOTAL
:L979
, ....*.... . ... ...... .. .......
318 466 467 220
.. ........
19BO * 435
:~~ ~~2 489 90 * :I.!:!j()!::;
;;,~ ::~(t
19B1 * 182 4 I.)!::;
4~H fJ4 * 142()
2~'=;f.J
1982 * 92 412 4:3:1. 246 6:? * :l:5()3
:1983*
19B4 *
46 46() 426
* 1.
~'.:51. i:l 1;;'1::'
d.J ~'.1.
17 419 4:5~; 2:1.1 * 1 :1.:34 ~j2
************************************************************
WHICH OUfPUT DO YOU WANT (FIGURE, NO CODES) ?
... HELP
TYPE IN
() ND f'I.JlHI·IEI:? OUTPUT
1. 'IABLE OF NtJMBEf?S IN (1UAI....ITICAT [UN f)l~UI.lI'·'s
2 TABI.. E OF NUMBEF::S IN 'TRATNING GI'~OlJI"~)
:.~ TABLE OF NUMBEf?S FUf( THE GfMll[ SENJUI< II.[/-: S
4 FLUWS IN THE SYS1EM
APPENDIX 163
5 STEADY-STATE DISTRIBUTION FOR LAST CHOSEN RErRUITMENl

6 STANDARD DEVIATIONS (UPPER BOUND)
7 STANDARD DEVIATIONS (EXACT)
8 COMPUTATION OF SALARY COSTS
9 HISTOGRAM OF THE MANPOWER DIS1RIBUTIUN
~ 1
QUALIFICATION GROUPS
GRABE
YEAR * LT CAP MAJ LTCOL COLt * TOTAL
1979
,
* 224 266 194 o * 4
....................................................
688
1980
* 164 277 218 9 0
* 668
1981
** 133 282 209 12 0
* 636
1982 69 325 186 11 0
* 591
1983
* 38 323 177 18 0
* 55b
1984
* 17 307 175 21 0
* 520
GRABE
YEAR * LT CAP MAJ LTCOL COLt * TOTAL
1979 * 94 200 220 168 o * 682
:1.980
1981
** 68
49
158
149
218
196
197
200
o ** 641
594
o
1982 * 24 147 182 191 o * 544
1983 * 7 137 183 171 o * 498
1984 * o 127 177 153 o * 457
* *
GF~ADE
YEAF< LT CAP MAJ I.. rcol... COL';' TOTAL.
1979 * o o o o o * o
1.980
* 0 0 16 0 0
* 16
1981
* 0 0 21 0 0
* 21
1982
* 0 I) 2~) I) I)
* 2:::i
1983
* I) I) 31) I) I)
*
.·.·.M··.M .. .... ___. ___ ._._._ .._*
:31)
1984
* 0
------------------
0 34 I) I)
.._....
:54
* *
GF~ADE
YEAF~ L.T CAP MA,.! LTCOL COLt TOTAL
1979 * I) I) 413 8~; * HI6
19(1)
** I) I) :37 52 'tl)
* 1.79
**
1981 I) I) :39 46 FJ4 169
6'·)
1982
* () () :38 44 ~ 144
1983
19B4 *
*
I)
I)
I)
I)
36
3:5
4:3
3"7
55
52 ** 1:34
1")")
.:,.,,'-
164 APPENDIX
WHICH OUTPUT DO YOU WANT (FIGURE. NO CODES) ?

+3
FOR WHICH YEAR DO YOU WANT A TABLE (STOP BY 99) ?

+ 1984
YEAR : 1984
GR. SENIORITY LT CAP MA,) L. TCOl.. COL +
** 0 28 4~) 25 12
2 ** 0 41:!·
,j 44 24 7
3 ** 0 84 32 28 7
4 ** 0 45 36 2(1 7
c-
,J
** 0 76 90 49 f.!
6 ** 0 50 :34 14 0
7 ** 17 :32 29 l1. 0
8 ** 0 16 26 :l2 0
9 ** 0 2() 36. 6 ()
10 ** () 38 () 6. ()
l:l. ** () () :l::~ ~) ()
1 ')
** 0 0 :33 10 (11 )
FOR WHICH YEAR DO YOU WANT A TABLE (STOP BY 99) ?

+ 99
WHICH OUTPUT DO YOU WANT (FIGURE, NO CODES) ?

+()
HOW DO YOU WANT TO CONTINUE (CODES,NO FIGURES) 7

HELP
THE POSSIBLE CODES ARE :

INPUTPRINT PRINTING OF INPUT DATA
EXTRAPRINT PRINTING OF CAREER PROSPECTS
SURVEY SURVEY OF CHANGED TRANSITION PERCENTAGES
RECRU INPUT OF RECRUITMENT
CHANGE CHANGE OF TRANSITION PERCENTAGES
AGE FORECASTS OF THE GRADE/AGE DISTRIBUTIONS
SUBTOT INPUT OF DESIRED SUBTOTALS IN TABLES
PLANNING RECRUITMENT PLANNING
DATABASE TRANSFORMATION OF CHANGED DATA TO NEW DATA-BASE
STOP END OF PROGRAM
AFTER l~E LAST CODE. TYPE IN ONE FIGURE !
+ RECRU
+ CHANGE
-> :3
TYPE IN THE NUMBER OF CLASSES WHERE RECRUITMENT IS ALLOWED? (12)

-+2
TYPE IN F~R ALI... CLASSES THE NUMBER AND ON THE NEXT LINE
THE RECRUITMENT FOR THE 5 PLANNING YEARS ?
APPENDIX 165
.... 1.
+2
+ ~"~?" 2'l" :~~7, 27,27
************************************************************
IN HOW MANY TRANSITIONS DO YOU WANT TO CHANGE PERCENTAGES 1(13)
.... 8
TYPE IN ON SINGLE LINES THE NUMBERS OF THE TRANSITIONS 1
.... 1.
.... 3
-+ I>
.... 8
-+ 9
-+ to
-+ 'IB
HOW DO YOU WANT TO CHANGE IN TRANSITION t 1

-+ HELP
TYPE IN
o CHANGE OF SINGLE PERCENTAGES
1 SHIFT OF THE AVERAGE TIME UNTIL TRANSITION
2 MULTIPLYING THE TRANSITION PERCENTAGES WITH A FACIOR
-+1
HOW MANY YEARS DO YOU WANT TO SHIFT THE PERCENTAGES 1

+ = LENGTHENING OF TIME UNTIL TRANSITION, - = SHORTENING OF TIME UNTIl.
TRANSITION
+-2
WATCH THE STRUCTURE OF THE TRANSITION PERCENTAGES FOR THE GRADE SENIORITIES ! I
(14)
HOW DO YOU WANT TO CHANGE IN TRANSITION 3?

-+1

+ = LENGTHENING OF TIME UNTIL TRANSITION, - = SHORTENING OF lIME UNflL
TRANSITION
WATCH THE STRUCTURE OF THE TRANSITION PERCENTAGES FOR THE GRADE SENIORITIES I I
HOW DO YOU WANT TO CHANGE IN TRANSITION 5 1

-+1

+ = LENGTHENING OF TIME UNTIL TRANSITION, - = SHORTENING OF TIME UNTIL
TRANSITION
+-1
WATCH THE STRUCTURE OF THE TRANSITION PERCENTAGES FOR THE GRADE SENIOR!TIES I I
-------------------------------------------------------- (15)
HOW DO YOU WANT TO CHANGE IN TRANSITION 18 ?

-+1

+ = LENGTHENING OF TIME UNTIL TRANSITION , - = SHORTENING OF TIME UNTIL
TRANSITION
-+ -1
WATCH THE STRUCTURE OF THE TRANSITION PERCENTAGES FOR THE GRADE SENIORITIES I!
166 APPENDIX
DO YOU WANT TO CHANGE THESE lRANSITION PERCENTAGES AGAIN 7

+NO
DO YOU WANT THE MANPOWER DISTRIBUTION FORECAST? TYPE IN: 0

DO YOU WANT THE MANPOWER DISTRIBUTION/ AGE FORECAST? TYPE IN: 1
+0
TYPE IN THE AGE DISTRIBUTION OF RECRUITS (AGE, PERCENTAGE) ?

CLASS 1:
-+ ~~4, 6
-+ 25, 9
+26011
+27, 6
+28.13
-+ 29 II 6
+ 30, 14
-+31.11
-+ 32, 4
-+ 33, 2
-+ J~::i!l :~
-+ 36 ~1 "I.
-+ .j;' y :I.
-+ ,:m, 1.
+ Y), 3
+ 4(). 1.
-+ 4:1 l' ' I
-+ 421' . .~
+ 43, :l
-+ 44, 'I
F"F/IC/',:N TAUF 1:!3 : I. ()O
AVERAGE RECRUITMENT AGE IS 30.2
r:LAS~:; ~:~!
-+ 21. 'I :.~
-+ 2~.~,:!..1.
-+ ~.~.h24
-+ :.\4y:~)9
-+ ;!~:';!':I (,
-+ :'!.) l' 4
-+ ~~751 ~~)
-+ 2H y "
PERCENTAGE IS : 100
AVlRAGl RECRUI.IMENT AUE [8 lJ.6
MANP(.IWE'/'( D:r ~:; m [HtllION IN PI" ANNING PU:': I,em
r;F(AfJE
I",r CAF' M('U I,rcul. CDL.+ * lOlA!...
1.919
I '~IHC)
. *.. . >I<
,'I.B
167
. ~:}~~9
...... . . .., . ..*. . ..
467
~:'j)9 :~'62
B~';
90 )/( :L~:j/?
'I 9f:l.I. I (,'0 49:) 04
*
)/( ::~ '~':"~ :·.~bl, 1.~,'64
I. q8~~ )/( \ I ~'i(n 4.'>0 2::):5 62 )/( .I !:,17
.l.9f:U
"
9H4 *
>I<
:.~ ~'1;'
.30()
409
44H
4M)
46/
~);19
:.~ 'I. ~)
.......................... _... - ................... ".
:::; ~::;
:13 *
)/(
',!',()()
14B3
~.·.M
APPENDIX 167

-+- 2
THAINING GRClUPS
GRADE
YEAH * 1..r CAP MA,J I. TeOL CO/..+ *
. ..*. ...... ......... .· .. . ...·.
1.9"7'1 266 1.94 4 o *. ...
6BB
*
:l.9BO :1.1.3 35~.:j 2:'~~j 9 ()
* "714
1981.
* 129 :,6B 2:1.B :1.2 ()
* n"7
:1.9B2
* 151 3~:;5 2()~i 1.1. 0
* 7~~~)
1983
* 1n :32() :1.99 18
n
0
* /14
1984
* ~!;!:I. 2n 2:1.:3 ()
* n7
GHADE
YEAI', * I... T CAP MA.J LTCOL CDl..+ * TOTAL
... .*
:1.9/9
.· 94 2()O
.. . ... 90 . . . .
:1.980
* :7j2 :1 "74 294 2~3:3
**
B6:5
19B:I.
* 6:1. 164 2"74 2~):'5 B4 f.l36
1982
* 79 14/ 26:;' 242 6;'.
** 792
19f.l:,
* /9 169 261 221 /t!~i
!:.)~.:j
1984
* /9 1.76 2~)4 1. ~}:::j :;;:3
* /!57
WHICH OUTPUT DO YOU WANT (F HiURE, NO CODES) ','

-.. 4
FLOWS IN THE SYSTEM
* FLOW *:1.9/9 YEAH

1980 19f.ll 19f.l2 19f.l3
LT f(ECf'. IN n n n n
I.. T PFWM. IN () () 0 0 0
I... r
LT
TOT. IN
·
PROM.OUT
. n
2:19 ·.... . · .
n
47
n
29
n
4:1
n
2~::;
I..T
I... T
L.T
TlII~N. OUT
RETI .OUT
Tor.OUT
4
()
22 ..~ 49
2
()
31
:~
() "
46
()
29
3
()
-------------_._---------------
CAP I:'EGF'. IN () 0 () () 0
CAP PfWM. IN 219 47 29 43 2~'=;
C,,\P
CAP
.· .
TOT. IN
pl~(JM.OUT
21'1
149 · .. ..· .. .
4/
:~~5
~.~9
4~)
43
46
::.~ !:.:;
!:.~ I
CAP TUF'N.OUT /' ? / 6 "7
CAP F,ErI .DU r 0 2 6 4 3
CAP lDT.OlJT 1 ~5b 44 ~it! ~5b 66
'·__..H._· ___ ,_ ........H··_· .. _··_· •• H·•• _ ....... _- .--,-" .. ,.-,.-
MA,J RECf(. IN () () () () ()
MA,J V'I:mM. IN 149 :~~:; 4~5 46 ~:.:;7
MAJ TOT. IN 149 ::~I.:.) 4::,:; 41> ::)7

168 APPENDIX
MA,J PROM.OUT 70 ;'P ::lO 2~) 24

MA,J TURN.OUT 9 1.0 " 9 9
MAJ RETI.OUT B :31 1.9 1B
MA,J TOT.OUT 137 6'1 5:~ ~jl
LTCOL RECF~. IN o o 0 ()
LTCDL pf~nM. IN 7() 37 2!::j 24
I... TCDL TOT. IN 70 :.p :. ~!.) ;:>4
. . . . . '" . . . . . . . . . . . . . . . . t ....... f .... t .. t .......... t "" .. t +
I..TCOL PROM.OUT 1.6 9 9 B 1 :.~

LTGOL TURN.OUT 3 4 4 ~~ 3
I.. TCOL. RETI. OUT 13 21. 30 2B 31
!...TCOL TOT.OUT 2B :'~4 42 ~W 47
--------------------------------------------
GOU RECF~.IN () () 0 0 0
COLt PROM. IN 16 9 9 8 13
COLt TOT. IN 16 9 9 B 13
f ••••••••••••••••••••••••••••••••
COL-!- PROM. OUT 0 () 0 0 0

COLt TURN.OUT 1 1. 1 1 1
COLt RET! • OUT :1.0 14 30 L~ :L5
COU TOT.OUT U 1~
. J ~H 14 16

-+ :7j
STEADY-STATE MANPOWER DISTRIBUTION FOR LAST CHOSEN RECRUITMENT
YEAI'~ : 20:L:3
(17)
(JI~ADE
LT CAP MA,J LTGOL COL + TOTAL
TOTAL: 30() 4f.l6 4136 95 ** :l6()4
NECESSARY RECRUITMENT TO MAINTAIN THE TOTAL NUMBER OF EMPLOYEES

CLASS 1: 44
CLASS
THE MANPOWER DISTRIBUTION WILL BE THEN
GF~ADE
LT CAP MA,J LTCOI... COLt TOTAL
TOTAL: 4n 4n 92 ** 15~;7
STANDARD DEVIATIONS OF THE FORECASTS (UPPER BOUND)

APPENDIX 169
GRADE
YEAH * LT CAP MAJ LTCOL COL+
1980
* 4 fJ 5 ~.~
19fJ1
* 1.0 fJ 3
~
d
1982
* 2 6 :1.1 9 4
1983
* 2 9 12 10 4
*
c·
1984 2 9 1:5 :1. J. "
WHICH OU1~UT DO YOU WANT (FIGURE, NO CODES) 7

-+7
STANDAHD DEVIATIONS OF THE FORECASTS (EXACT)

. . . . ++ • • • • • ++ • • • • • t ••••••••• ++ • • • • • • • • • • • • • • •
GRADE
YEAR * LT Cj!\P MA,J LTCOL COLf
1980
* 1 4 8 5 2
1981
* 1. 5 10 6 2
1982
* 2 6 11
1 '14
7 3
1983
* 2 7 7 ~~
1984
* 2 8 13 8 4
WHICH OUTPUT DO YOU WANT (FIGURE, NO CODES) ?

-+8
TYPE IN FOR THE 13 CLASSES (IN 1000-GUILDERS)

MINIMUM SALARY, PERIODIC INCREASE, MAX. NUMBER OF PERIODIC INCREASES
-+ 38,:1.• 2, 7
-+ 38,,1.2, 7
.,. 48.1.4.10
-+ 48.1.4.10
-+ 65.1.6, H
.,. 65>1.6, 13
-+ 65,1.6, 8
-+ 65,1.6, 8
-+ 82,1.8, 6
-+ 8~', 1 .8, 6
-+ 82,1 .. 8, 6
-+ 82,1.13, 6
-+ 100. 2 • 0, 6
(18)
TYPE IN THE YEARLY TREND IN PERCENTS (STOP BY 99)
-+0
SALARY COSTS IN MILLION GUILDERS
GRADE
YEAR * LT CAP MAJ LTCOL COl.+ * TOTAL
-----------------------------------------------------
1979
* 13.45 25.11 33.47 19.29 9.35 * 100.67
•••••••••••••••••••••••••••• t .........................
1981 * 9.07 * 101.El7

1980 6.66 27.05 37.14 22.78 9.76 103.:38
* 7.49 27.54 34.69 :?3.08
*
1983 * 5.86 *
1982 9.10 26.36 33.05 6.56 97.:L1. ;!2.0~)
* 10.17 25.87 32.71 20 ..

* 8~:j 9~;.46
1984
* 12 23.813 .. 22 lB.85
* 93
l!:' t::"C'
.O~! 3~5 •. J ...J0 .~):·5
------------------------------------_._---------------
170 APPENDIX
DO YOU WANT THE STANDARD DEVIATIONS 1

-T NO
TYPE IN THE YEARLY TREND IN PEpn rl rs (STOP BY 99)
-T 99
WHICH OUTPUT [If) YOU WAN r (F WIJRE, NO rOIJES) '~

-T 9
FOR WHICH YEAR DO YOIJ WANT '. d.fSrOGRAtl (STOP BY 99) '?
-T 1984
ONE FIGURE STANDS ~w'
YEAR 1984:
----A----A----A- ,--A----A----A----A----A----A----A----A----A
LT 111111111111111111111111111111 300
LT 111111111111111111111111111111 300
CAP 222222:~22222222 ,\ :~~2 22 222222222222222222222222 2 448
CAP 22222~~222222222:):?22222222222222222~~2222:;~22222 448
MAJ 33:333333~~33~~~~333;~;3:B;3333;3333:n~~3333333:333333333 467
MAJ 33333333333333333333333333333333333333333333333 467
LlCOL 4444444444444444444444 215
LTCOL 4444444444444444444444 215
COLt 55555 53
COLt 55555 53
FOR WHICH YEAR DO YOU WANT A HISTOGRAM (STOP BY 99) 1

-T 99
WHICH OUTPUT DO YOU WANT (FIGURE, NO CODES) 1

-T 0
HOW DO YOU WANT TO CONTINUE (CODES,NO FIGURES) 1

-T SURVEY
-T AGE
-T DATABASE
+ PLANNING
-T 3
------------------------------- •••• _ _ • N~ , ___ • __" " •• H •• H •• _ ••_ •• ."._._...__..... " .. - ..-..

(9)
3 I::'
6
'"
---------------------------------------_._----
1 , 3, 1 : 0.0 4, 1 : 0.0
...- ............ _......
, 4, 1 : 0,0
_,."
3,
....... _. _...OM._ .M. _..... _. _ M"_
:I.: 0,0
-'"
'1 I::'
~~
~~ ,
~,
1, 3, 2: 0.0 2, 4, 2: 0.0 3, 4, ;! : 0,0 3, 2: 0.0

1, :3, 3: 0.0 '1
.. ' 4, ::~ : 9B.0 :~ , 4, :3 : 14.9 :3, ..,, C'
3: 0.0
, ~~ ,
,,:
1, ~~ 4: 0,0 2, 4, 4: 9B.0 :{, 4, 4l 0,0 ~5 , 4: 0,0
1, 3, J . 1::'.
99.0 2, 4, ~j : 9B.O :5, 4, 0.0 1::"
:3, , ~:; 5: 0.0
,
,,}.
1, 3, 6: 99.0 2, 4, 6: 913.0 :~ 4, f>! 0.0 :3, , ~:; 6: 0.0

1, 3, 7: 99.0 .. ,
,,:'1 4, 7l 9B.0 3, 4, 7l 0.0 ~~ , 5, 7! 0.0
1, 3, B: 99.0 2, 4, B: 'IB.O 3, 4, 8: 0.0 :3, -'"
0'
B: 0.0
1, :~ , 9: 99.0 -, ,4, 9: 98.0 :3, 4, 9: 0.0 3, 1;:'
,.1 , 9: 99.0
-, ,
~,
1, :~,],O: 99.0 ..... 4.10: 9El.0 :3, 4,10: 0.0 :5, ~), 10: 99.0
1, 3.t 1.: 9'1.0 ..-, , 4,11: 9B.O :3, 4.11: 0.0 :5, ~~;, 1 :I. : 99.0
-, ,
~.
1, 3, 1. :~~: 99.0 ,,~. 4,12! 98.0 :3, 4" 1~~: 0.0 :5, :~j, 12: 99.0
APPENDIX 171
....... ..-..----....
~-
-_.-.--------------- . ..... -- - _. -. -'- ", .,,, .......... _........
-------------------
8 9 10 HI
.... _----------------------------------- _ _-------------------

....
4, 6, 1: 0.0 4, 7, 1: 0.0 4, 8. :1.: 0.0 8, 12, 1: 0.0
4. 6, 2: 0.0 4, 7, 2: 0.0 4. B, 2: 0.0 H, 1.2, ::!: 0.0
4, 6, ;3 : 0.0 4, 7, ;'5 : 0.0 4, 8, 3: 0.0 flo 1 ::.~, 3: 0.0
4, 6, 4: 0.0 4, 7. 4: 0.0 4, 8, 4: 0.0 H, 1. ~.', 4: :39.2
4, 6, 1::'.
,I. 0.0 4, 7, .,1.t:' ..
0.0 4, 8, ~.:; : 24.5 8, 12, 5: 98.0
4. 6, I.d 71. 't 4, 7, 6: 26.1 4, 8, 6: 0.0 8, 12, 6: 98.0
4, 6. 7: 71.. 'f 4, 7, 7: 26.1 4. 8, 7! 0.0 H. 12, 7: 98.0
4, 6, B: 71.9 4, 7, Bt :':~6 .. 1 4, 8, 8: 0.0 8, :1.2, 8: 'fB.O
4, 6, 9: 71.9 4, 7, 9: 26.1 4. 8, 9: 0.0 8. 1;? , 9l 98.0
4, 6dO: 71.9 4, 7dOl 26.1 4, 8dO: 0.0 Eh 12.10: 98.0
4. 6,11.: 71.9 4, '7, :1.1 :
26.1 4. 8.11: 0.0 8, 1 ~~, 11. : 98.0
4, 6, :1.2: '71.9 4. 7,12: 26.1 4, 8,12: 0.0 8, 12,1::.~: 9B.O
• • • • • • • • • • • • • • • • • • • • • • • • • • • • • ++ • • • • ++ • • • • • • • • • • • • • • • t •••••••••••••••
TYPE IN THE DESIRED FIGUF~E : (20)

o YOU WANT THE AGE DISTRIBUTIONS FOR ALL GRADES INCLUDING RECRUITMENT
N : YOU WANT THE: AGE DISnnBUTIONS nm GRADE N
-> 0
TYPE IN THE MINIMUM AND MAXIMUM AGE FOR THE

AGE DISTRIBUTION YOU WANT TO SEE: ?
..... 21,~i4
--_ ...._---------------------------------
ACTUAL MANPOWU~ DISnnBUTION : 1979
GRADE
AGE LT CAP MA.,J LTCClI... COL+ TOTAL
·._ ........ _._.._M··_H_
--------------------------------
::;i4
** 0 () 8 El 10 26 **
5;:~
** () 4 24 21.
:5:5
14 ** 63
~52
** 2 8 29 ~?B
** 100
:::;1
** l '7 2() 30 IO ** 68
~)()
** () 9 113
:52
32 1.2
** 11
49 ** l 9 22 6 ** '70
41:1
** :I. 12 40 19 4 ** 76
47
** :3 'l 29 '7 1 ** 49
46 ** 'f 6 32 1:.' ()
** 59
41::·
•.1
** 4 8 :~~ t, lO ()
** 48
44 ** 4 :1.0 :5f.,
:5:<
:1.2 0 ** 62
4:3 ** :3 B 6 0 ** 49
4':>
4'1. **
**
2
:3
1./
20
:50
2::!
7
1
0'
0
**
**
~5b
46
4()
** 4 :;~:3 2:1 0 ()
** 4B
:39 ** :1. 12 l'i' 0 0 ** ::12
~~B
** 4 20 If.! () ()
** 42
:3'7 ** 9 :59 If., 0 0 ** 64
:16
** 1. ~.> 40 7 0 0 ** ~j9
3~:.:j
** :1.0 :11 7 0 0 ** 4B
:34
** ;'0 27 1 0 0
** 4B
3:5 ** 26 :34 0 () 0 ** 60
3::.> ** 21.) 32 () 0 0 ** ~~i8
:n ** 22 :1() () 0 0 ** ~52
30 ** :1.:3 :;~ l () () 0 ** 34
29 ** 1.'7 :I. ~::, () () ()
** 32
21:1
** 22 :1.0 () 0 0 ** 3?
27 ** :I.B 5 () 0 0
** 2~~
26 ** 3:;~ () () 0 ()
** ,3'""·M
172 APPENDIX
2~.:j
** 21 ()
** 21 () () ()
n ** ** 17
24 1.7 () () () ()
22 **
** 1() () () ()
** 1()1 ()
2:1. **
1. () () ()
** ()
() () ()
**
._--------_._-------------------
()() ()
~···M
TOTAL:
..M_··_··.. __···_···_
318 466 467 ** 1::=;56
--_._ .. _----,----.---------------------
~!20 B~j
AVERAGE : DEVIATION :
AGE
........... : .....M.: _.... ___._. __ .... _._._:
I. I :31 +!::j
CAP :37.4
MA,j 4~) +:3
LTCClI.. 49.2
COI. . +
.. -----.---:------------J-----------:
IHf ~V[RAGE AGE OF TOTAL. MANPOWER IS 41.() YEARS
I HI:' /.I[VfAlJ.ON [S 8.5 YEARS
. _----------------- ... _--_.. __.__ ..--..

YEAf~ : 1984
Gf<ADE
AGE LT CAl'-' MA,J LTCOL COl.+ TOlAL
• _ _ ·OM·· __ ' __ •• _ _ _ _ _ · · _ , , _
... _------------------------------------
:::j4
** 0 2 3() n 1.0 ** 65
5:3 ** () 2 34 28 6 ** 7()
4 -,C '
~52
** () 4 20 113 3 **
51 ** () 9 2() 19 5 ** ~j3
5()
** () 4 19 16 6 ** 4'::'
,J
4'7 ** () 6
-,c·
24 21 7 ** 5B
4<3
** () 1~; 22 3 ** 4 ,JC'
47 ** 2 /
-,
1.<3 113 13 ** 53
46
** :3 ~
2:3 12 2 ** 4~j
4'::'
,J
** 4 / 28 <3 1
** 413
44
** 4 :5 17 9 1 ** ~H
43
4':' **
** "4 13
22
22
31 6
8 1
() **
**
44
63
41 ** 4 22 :U 2 ()
** 59
4()
** :1 21 22 2 ()
** 4B
39 ** 4 2/ 16 1 ()
** 4<3
::18 ** 3 :37 19 () ()
** :7,9
:17
** :5 :56 :18 1. ()
** 58
::16
** 4 2:3 24 1 ()
** ::=;2
:·~~.5
** 9 1.4 17 0 ()
** 4()
34 ** 14 19 J.l. () ()
** 44
33 ** 16 24 6 () ()
** 46
3::~
** 21 19 2 () ()
**
**
4~)
:3J.
:3() **
**
22
22
:50
21
()
()
()
()
()
()
**
52
4:3
APPENDIX 173
29 ** ~'1 21. 0 () 0 ** 4;.~

20 ** 22 ::~ () () () ()
** 42
27 ** 20 14 () () 0 ** 34
26 ** 24 :1.:1. 0 () ()
** :5~)
2~j
** 21,:.:} (, () () ()
** :3:1.
n
24 ** 22 :L 0 0 ()
**
2:5 ** 1.:3 () () 0 0 ** 13
22 ** 6 () 0 () 0 ** (,
21 **
..M._.H.._._.M.·····H
1
......... ....
0
_ _.._.... -_ ....
() 0 0
. . . . . . . . . . .M • • _
**
. . . . . . . . . . . . . . . . _ _ _ .·M •
1
TO"! AI.
M ...." .." . . . . . . . . . . . . . . . . . . . , •• •••
3()() 440 467
•••••••••••• H •• _ _ _ _ •• _ ..... H
2l~) ~53 ** :1.4B3
GI~ADE AVEF~AGE : DEVIATION

AGE
.... I ....., ......._............ : H . . . . H . _ _ _ _ •••••••••• _ •••• " ...... :
I...T 30.2
CAP 36. :3
MA.J 44.4
I.. TCOI. . 4'1.1.
COLt ~5 () • 0
.....M.:
THE AVERAGE AGE OF TOTAl.. MANPOWER IS 40.0 YEARS

THE DEVIATION IS
TYPE IN THE DESIRED FIGURE :

0: YOU WANT NO FIGURE OF THE AGE DISTRIBUTIONS
N: YUU WANT THESE FIGURES WHERE N SUCCESSIVE AGES
ARE TAKEN TOGETHER
FIGLJF~E OF THE ACTIJAL MANPOWEF~ : 1. 979
AGE *
52,-54 *223333333333334444444444445555555555
49-5:1. *222223333333333333344444444444444444555555
46-4B *1112222233333333333333333333444444445
43-45 *1.1222223333333333333333333444444
40-42 *:1.122222222222233333333333333344
37-39 *:1.1.1.2222222222222233333333333
34-36 *111:1.111:1.22222222222222222222333
31-33 *111:1.111111111112222222222222222222
2B-30 *11:1.1:1.1111:1.222222222
25-27 .111:1.111111.11:1.12
22····24 U1:1.1ll
21-,,21. *
t FIGURE " 5 EMPLOYEES

174 APPENDIX
YEAR 1. 'hI 4
AGE *
52-54 *2333333333333333334444444444444455~5
49-51 *22223333333333333444444444445555
46-48 *12223333333333344444444444555
43-45 *11.22223333333333333444445
40-42 *1122222222222223333333333333333344
37-39 *11.2222222222222222222233333333333
34-36 *11111222222222223333333333
31-33 *11111111.111122222222222222233
28-30 *1111.111111111222222222222
25-27 *1111111.1111111222222
*
n··-~~4 *1.11lU.l.:I.
21·-·~~:1.
:1 FIGUf~E = ~; EMPI... OYEES
YOU CAN CHOOSE FROM TWO DATA-BASES ; TYPE IN:

o ACTUAL TRANSITION PERCENTAGES AND BASIC YEAR 1979
1.: ACTUAL TRANSITION PERCENTAGES AND YEAR m 1984
-+ 1.
TYPE IN THE NAME Fe)/'( THE NEW DATA .... BASE -p(21)
-+ TEST ~
FL.OW SCHEME FDf! 1:!ECfWITS(22)
Ff!ClM TO AF"J Ef~ ••• YEAI'~S

GF~ADF GRADE .l 234
,.~ .. -.- .. .. --.-
~-
* 98 96 0
()
'n
* 2 () () 94
* :3 () () () ()
**
4 () () () ()
,.,
c:'
0 () 0 ()
',/
2
* 2 99 '16 9~:;
2
* :~ () () () ()
**
2 4 () () () ()
2 I.:'
"
() () () ()
:.\ :3 <t ~:; 'If, 94 88
*
,.,
:, **
:3 4 () () () I::'
,.,
I .. '
() () () ()
4 * 4 99 97 '16 84
4
~5 *
*
~.'.i
<
,.! 98
()
••••••••••• _ • • • _ • • OH ••••• _ . _
97
()
96
() :I.()
94
nn YOll WANT TO CHANGE. n·11':: FLOW SCHEME FOR I~ECI:~U:rTS '!~

-+ N(I
I'~ 11111.1 11()NY Gf~ADES .1:8 I·,FCI:WITMENT ALLOWED '?

APPENDIX 175
TYPE IN THE NUMBERS OF THESE GRADES AND ON THE NEXT LINE

THE MAXIMAL ALLOWED RECRUITMENT FOR THE 5 PLANNING YEARS ?
... 1
TYPE IN THE LOWER BOUND GRADE SIZES FOR fHL 5 PLANNING YEARS ?
GRADE 1:
->- 170,200,250,300,350
TYPE IN HOW MANY YEARS BEFORE RECRUITMENT IS ALLOWED ?

->- 0
MANPOWER DISTRIBUTION IN PLANNING PERIOD INCL. RECRUITMENT
GRADE
YEAR * LT CAP MA,J LTeDL. COL+ * TOTAL
.....* ....... ........... ..... .. .. ..*........

1979 31B 466 467 220 B~'i
:1980 * llO ~:;::!9 ~:j29 262 * 90 1,~';8()

1981 * 2()() ~.:;32 4'12 ;!bb
* 1,574
f.>2 *
84
1,982
* 2~.:;() ~';()3 468 2!":;::~ 1 ::i:5<,
:1983 * :30() 4'12 460 2:39 * ~i~) 1~"i46
:1.984 *
.__.._ ..H.....·H·· __ ··_ ...
:3::iO 4~iB 4""")
\.)/ 2:1.~.:i
* ~j3
-------_._---------------------------
1~'i43
RECRUITMENT IN PL.ANNING PERIOD
*
Gr~ADE
YEAR LT CAP MAJ L Tcm.. COL.+ * TOTAL.
1.979 *
° ° () *
.....................................................
3 ()
19130 * 7 ()
° 0 0 * 7
19131 * 1.0
°
() 0 0 * 10
1982 * 27
°
() () 0 * ;.'7
19K~
* 14 () 0 () 0 * 14
DO YOU WANT RECRUITMENT PL.ANNING AGAIN 9

... NO
HOW DO YOU WANT TO CONTINUE (CODES,NO FIGURES) ?

->- STOP
************************************************************
176 APPENDIX
THIS IS THE END OF THE INTERACTIVE COMPUTER PROGRAM
+++++++++ ++++H+H+ ++++++++++ H H +H+H++H HH+H+++ ++

+ ++ + + + + + + + + + + + +
+ + + + + + ++ + + + + +
++++++ + + ·HH++++++ + + + + + ++++·H++++ +
+ + + + + + + ++++++++++ + ~
+ + + + + + + + + + ~
+ + + + + + + + + + ~
+ + + + + + + + ' + + + ~
++ +H++HH+ +++ H+ H+ H+ +++ +++ HH++++H +~
(23)
APPENDIX 177
NOTES
1. By this command the user starts the processing of the computer program, which is con-
tinuously stored on disk (background storage).
2. The data base "OFFICERS" contains the necessary input data for the group of officers.
Such data bases are also stored on disk (background storage).
3. If such comments have been indicated on the data base, then they are printed at this
moment.
4. If the user does not know the answer, then he may type in "HELP," whereafter he re-
ceives the possible answers.
5. Sometimes the user wants to lump together some grades or function levels. In that case he
may indicate which grades have to be taken together by the code "SUBTOT."
6. The figure serves as a delimiter for the codes.
7. This list follows from the code "INPUTPRINT." It gives some data recorded on the data
base.
8. The class "turnover" has to be indicated by (0,0,0).
9. The transition percentages for transitions 3-26 have been deleted here.
10. This list follows from the code "EXTRAPRINT." The data with respect to transitions
5-24 have been deleted here.
11. For the maximum grade 'seniority (here, 12), the number of employees with this grade
seniority indicates the number of employees with this grade seniority or mote.
12. Because of the code "RECRU" (recruitment).
13. Because of the code "CHANGE" (changes of career policies).
14. Back-shift of the promotion percentages means that the first percentages of the transition
are skipped.
15. The changes for the transitions 6, 8, 9, and 10 have been deleted here.
16. The flows indicate for each grade the numbers of recruits, promotions into the grade,
total inflow, promotions out of the grade, turnover, retirements, and total outflow, re-
spectively.
17. In this year the steady-state distribution is obtained.
18. Such a trend may indicate, for instance, the inflation percentage or the yearly extra salary
increase.
19. Because of the code "SURVEY" (transitions for which transition percentages have been
changed).
20. Because of the code "AGE" (forecasts of the age distribution).
21. Because of the code "DATA BASE" (recording of changed data on a new data base).
22. Because of the code "PLANNING" (recruitment planning). The flow scheme indicates
for all grades the percentage of employees who are still in that grade and who have been
promoted to higher grades for 1,2,3, ... years after entrance in that grade.
23. The "elapsed time" for this run has been about 28 minutes, which was mainly caused by
the large numbers of options used for this example and by the number of users of the
computer system. The "process time" has been about 40 seconds, which was mainly
caused by the computation of the exact standard deviations. The "input-output time"
was about 4 seconds for the transport of data.
REFERENCES
Ackoff, R. L. 1970. A Concept 0/ Corporate Planning. New York: John Wiley.

Barber, J. W., ed. 1968. Industrial Training Handbook. London: Iliffe Books.
Bartholomew, D. J. 1973. Stochastic Models for Social Processes. 2nd ed. London:
John Wiley.
___ . 1975. "Errors of Prediction for Markov Chain Models." Journal 0/ Royal
Statistical Society 3:444-56.
Bartholomew, D. J., R. F. A. Hopes, andA. R. Smith. 1976. "Manpower Planning
in the Face of Uncertainty." Personnel Review 5:5-17.
Bartholomew, D. J., and A. F. Forbes. 1979. Statistical Techniques/or Manpower
Planning. Chichester, Eng.: John Wiley.
Bartholomew, D. J., ed. 1976. Manpower Planning. Harmondsworth, Eng.: Pen-
guin Books.
Beth-Von der Warth, A. M., and V. Steinecke. 1980. "Personalaufwandsplanung
mit Hilfe eines Erweiterten Linearen Modells." Operation Research Spektrum 1:
183-98.
Blau, P. M., and W. R. Scott. 1970. Formal Organizations: A Comparative
Approach. London: Routledge & Kegan Paul.
Bosworth, D. L., G. J. Evans, and R. M. Lindley. 1974. "Mechanistic Manpower
Models." In Wabe 1974, pp. 61-84.
179
180 REFERENCES
Botter, C.H. 1975. Industry and Organization (in Dutch). Eindhoven:

Philips/Kluwer, NIVE.
Bowey, A. M. 1969. "Labor Stability Curves and a Labor Stability Index." British
Journal oj Industrial Relations 7:71-83.
___ . 1977. "Corporate Manpower Planning." Management Decision 15:
421-96.
Box, G. E. P., and G. M. Jenkins. 1970. Time Series Analysis Forecasting and Con-
trol. San Francisco: Holden-Day.
Brown, R. G. 1963. Smoothing, Forecasting and Prediction oj Discrete Time Series.
Englewood Cliffs, N.J.: Prentice-Hall.
Bryant, D. T., and R. J. Niehaus, eds. 1978. Manpower Planning and Organization
Design. Nato Conference Series II: Systems Science, Vol. 7. New York: Plenum
Press.
Cameron, M. H., and J. E. Nash. 1974. "On Forecasting the Manpower Require-
ments of an Organization with Homogeneous Workloads." Journal oj Royal
Statistical Society A 137:584-95.
Casson, R. J. 1978. "Re-evaluating Company Manpower Planning in the Light of
Some Practical Experiences." In Bryant and Niehaus 1978, pp. 545-59.
Charnes, A., W. W. Cooper, and R. J. Niehaus, eds. 1978. Management Science
Approaches to Manpower Planning and Organization Design. TIMS Studies in
the Management Sciences. New York: North-Holland.
Clough, D. J., C. C. Lewis, and A. L. Oliver, eds. 1977. Manpower Planning
Models. New York: North-Holland.
Crichton, A. 1968. Personnel Management in Context. London: Batsford.
Draper, N. R., and H. Smith. 1966. Applied Regression Analysis. New York: John
Wiley.
Drui, A. B. 1963. "The Use of Regression Equations to Predict Manpower Require-
ments." Management Science 8:669-77.
Dudding, R. C., and W. G. Piskor. 1978. "A Computer-Assisted Manpower Plan-
ning Model." In Bryant and Niehaus 1978, pp. 331-39.
Esser, F. L. G. 1975. "FORMASY: A System in Order to Forecast the Manpower
Distribution and Necessary Recruitment." Master's thesis, Department of
Mathematics, Eindhoven University of Technology.
Etzioni, A. 1968. A Comparative Analysis oj Complex Organizations. New York:
Free Press.
Fair, R. C. 1969. The Short-Run Demand jor Workers and Hours. Amsterdam:
North-Holland.
Gani, J. 1963. "Formulae for Projecting Enrollments and Degrees Awarded in Uni-
versities." Journal oj Royal Statistical Society A 126:400-09.
Geisler, E. B. 1967. "Manpower Planning: An Emerging Staff Function." Ameri-
can Management Association Bulletin 101:1-28.
Ghosh, A. 1960. "A Note on Leontief Models with Non-homogeneous Production
Functions." Metro Economica 10:14-20.
REFERENCES 181
Gohl, J., and H. J. Opelland. 1978. "A Methodological Review of the Treatment of
Individual Goals in Some Recent German System Approaches towards Man-
power Planning." In Bryant and Niehaus 1978, pp. 57-72.
Goldfeld, S. M., and R. E. Quandt. 1972. Nonlinear Methods in Econometrics.
Amsterdam: North-Holland.
Gordon, T. J., and O. Helmer. 1964. "Report on a Long-Range Forecasting
Study." Santa Monica, Calif.: Rand Corporation, Report No. P-2982.
Grinold, R. C. 1977. "Interactive Design of a Manpower System." In Bryant and
Niehaus 1978, pp. 307-18.
Grinold, R. C., and K. T. Marshall. 1977. Manpower Planning Models. New York:
North-Holland.
Halpern, J. 1977. "A Forecasting Technique with an Application to the Civil Ser-
vice." In Clough, Lewis, and Oliver 1977, pp. 219-25.
Harvey, E. 1968. "Technology and the Structure of Organizations." American
Sociological Review 33:247-59.
Huisjes, W. R., E. H. Van Tuyll van Serooskerken, and F. J. Wijbenga. 1977. "In-
struments for Manpower Planning" (in Dutch). InJormatie 19:734-41.
Ignizio, J. P. 1976. Goal Programming and Extensions. Lexington, Mass.: Lexing-
ton Books.
Jackson, J. H., and C. P. Morgan. 1978. Optimization Theory: A Macro Perspec-
tive Jor Management. Englewood Cliffs, N. J.: Prentice-Hall.
Kahalas, H., and D. A. Gray. 1976. "A Quantitative Model for Manpower Decision
Making." Omega 4:685-97.
Katz, D., and R. L. Kahn. 1967. The Social Psychology oj Organizations. New
York: John Wiley.
Keen, P. G. W., and M. S. Scott Morton. 1978. Decision Support Systems: An
Organizational Perspective. Reading, Mass.: Addison-Wesley.
Kessels, J. H. 1974. "The Dynamical Behaviour of a Manpower System." Master's
thesis, Department of Mathematics, Eindhoven University of Technology.
Klein, L. R. 1962. An Introduction to Econometrics. Englewood Cliffs, N.J.: Pren-
tice-Hall.
Koontz, H., and C. O'Donnell. 1974. Essentials oj Management. New York:
McGraw-Hill.
Kiiching, K.-F. 1973. Personalplanung and UnternehmungsJorschung. Baden-
Baden: Gehlen Verlag.
Lapp, P.A., and I. W. Thompson. 1977. "Supply and Demand for Engineering
Manpower Related to the University System in Ontario." In Clough, Lewis, and
Oliver 1977, pp. 293-310.
Lawrence, J. R., ed. 1975. Company Manpower Planning in Perspective. London:
Institute of Personnel Management.
Likert, R. 1961. New Patterns oj Management. New York: McGraw-Hill.
Margerison, C. J., and N. Hunter, eds. 1978. Management and Organization Devel-
opment: Bibliography. 2nd ed. Bradford, Eng.: MCB Publications.
182 REFERENCES
McCosh, A. M., and M. S. Scott Morton. 1978. Management Decision Support Sys-
tems. London: Macmillan Press.
Milkovitch, G. T., A. J. Annoni, and T. A. Mahony. 1972. "The Use of the Delphi
Procedures in Manpower Forecasting." Management Science 19:381-88.
Nash, J., and A. Goddard. 1978. "MANPLAN: A Flexible Computer ]Package
for Practical Manpower Planning." In Bryant and Niehaus 1978, pp.
103-18.
Needham, D. 1972. Economic Analysis and Industrial Structure. London: Holt,
Rinehart & Winston.
Nuttle, H. L. W., and J. Wijngaard. 1979. "Planning Horizons for Manpower
Planning in a Two-Level Hierarchical System." Manpower Planning Reports 19,
Eindhoven University of Technology.
Parsons, T. 1960. Structure and Process in Modern Societies. Glencoe, HI.: Free
Press.
"Project Group Organization Structure: The Strength of the Police Force" (in
Dutch). 1979. Report of Netherlands Ministry of the Interior, The Hague.
Pugh, D. A., D. J. Hickson, and C. R. Hinings. 1969. "An Empirical Taxonomy of
Structures of Work Organizations." Administrative Science Quarterly 14:
115-26.
Purkiss, C. J. 1974. "Manpower Planning: A Contribution of Concepts and Prac-
tice." Dissertation, University of Lancaster.
Rowntree, J. A., and P. A. Stewart. 1976. "Estimating Manpower Needs II: Statis-
tical Methods." In A. R. Smith, ed. Manpower Planning in the Civil Service.
London: H. M. S. 0., No.3, pp. 36-53.
Sayles, L. R., and G. Strauss. 1977. Managing Human Resources. Englewood
Cliffs, N. J.: Prentice-Hall.
Schmidt, H., H. Hagenbruck, and W. Samann, eds. 1975. Handbuch der Personal-
planung. Frankfurt: Herder & Herder.
Seal, H. L. 1945. "The Mathematics of a Population Composed of k Stationary
Strata Each with Recruitment from the Stratum Below and Supported at the
Lowest Level by a Uniform Number of Annual Entrants." Biometrika
33:226-30.
Sherlock, 1. P. 1978. "Manpower Planning as a Basis for Current Changes in Peri-
ods of Engagement in the Canadian Forces." In Bryant and Niehaus 1978, pp.
513-27.
Siebelt, H. Th. E. M., T. Taal, and H. Van Wijk. 1977. "Career Planning at the
Netherlands Post, Telephone and Telegraph Organization" (in Dutch). Inter-
mediair 13.
Si1cock, H. 1954. "The Phenomenon of Labour Turnover." Journal oj Royal Sta-
tistical Society A 117:429-40.
Smith, A. R., ed. 1970. Models oj Manpower Systems. London: English University
Press.
Taylor, B., and G. L. Lippitt, eds. 1975. Management Development and Training
Handbook. New York: McGraw-Hill.
REFERENCES 183
Tuggle, F. D. 1978. "An Analysis of Employee Turnover." Behavioral Science

23:32-37.
Uebe, G., M. J. Beckmann, J. Fischer, and E. Marnitz. 1978. "Two Linear Pro-
gramming Models of a Large German Federal Department." In Bryant and
Niehaus 1978, pp. 307-18.
Vajda, S. 1947. "The Stratified Semi-Stationary Population." Biometrika 34:
243-54.
_ _ . 1970. "A Historical Survey." In Smith 1970, pp. 7-10.
___ . 1978. Mathematics oj Manpower Planning. Chichester, Eng.: Wiley-Inter-
science.
Van der Beek, E. 1977a. "Manpower Planning for the Public Works Office Using
FORMASY" (in Dutch). Manpower Planning Reports 7, Eindhoven University
of Technology.
___ . 1977b. "Prediction Errors in the Application of Markov Models for Man-
power Planning" (in Dutch). Memorandum COSOR 77-13, Department of
Mathematics, Eindhoven University of Technology.
Van der Beek, E., C. J. Verhoeven, and K. de Waij. 1978. "A Manpower Planning
Method" (in Dutch). Personeelbeleid 14:152-59.
Van der Beek, E., C. J. Verhoeven, and J. Wessels. 1978. "Some Applications of
the Manpower Planning System FORMASY." Operations Research Verjahren
28:19-32.
Van Meeteren, F. H. 1978. "Manpower planning with the Use of FORMASY at the
Local Police Forces" (in Dutch). Master's thesis, Department of Industrial Engi-
neering, Eindhoven University of Technology.
_ _ . 1979. "Application of FORMASY for the Police Force of The Hague" (in
Dutch). Report of Netherlands Ministry of the Interior, The Hague.
Van Nunen, J. A. E. E., and J. Wessels. 1978. "Forecasting and Recruitment in
Graded Manpower Systems." In Bryant and Niehaus 1978, pp. 353-64.
Van Winkel, E. G. F. 1979. "ARIMA Processes: The Importance of Univariate
Methods of Econometrics" (in Dutch). Dissertation, Free University, Amster-
dam.
Verhoeven, C. J. 1977. "FORMASY 2: User's Manual for an Interactive Manpower
Planning System" (in Dutch). Manpower Planning Reports 5, Eindhoven Uni-
versity of Technology.
___ . 1979. "Eine Interaktive Personalplanungsmethode." Manpower Planning
Reports 20, Eindhoven University of Technology.
___ .1981. "Corporate Manpower Planning." European Journal oj Operations
Research 7:341-49.
Verhoeven, C. J., andJ. Wijngaard.1978. "Application of the Manpower Planning
System FORMASY at Philips Ltd." (in Dutch). Manpower Planning Reports 13,
Eindhoven University of Technology.
Verhoeven, C. J., J. Wessels, and J. Wijngaard. 1979. "Computer-Aided Design of
Manpower Policies." Manpower Planning Reports 16, Eindhoven University of
Technology.
184 REFERENCES
Wabe, J. S. 1974. Problems in Manpower Forecasting. Westmead, Eng.: Saxon

House.
Wachter, H. 1974. Grundlagen der Langfristigen Personalplanung. Berlin: Herne
Verlag.
Wessels, J., and J. A. E. E. Van Nunen. 1976. "FORMASY: Forecasting and Re-
cruitment in Manpower Systems." Statistica Neerlandica 30:173-93.
Wijngaard, J. 1978. "Markov Models for Manpower Planning: Some Remarks and
an Application on a Problem of Decreasing Growth." Manpower Planning Re-
ports 11, Eindhoven University of Technology.
___ .1980. "Aggregation in Manpower Planning." Manpower Planning Reports
22, Eindhoven University of Technology.
Woodward, J. 1958. Management and Technology. London: H.M.S.O.
Young, A., and G. Almond. 1961. "Predicting Distribution of Staff." Computer
Journal 3:246-50.
Young, A., and P. C. G. Vassiliou. 1974. "A Non-linear Model on the Promotion
of Staff." Journal of Royal Statistical Society. A 137:584-95.
INDEX
Age distribution, 6-11, 99-100, 125-26, Decision support system, 84

128, 145, 148-52 Definition: of categories, 77-82, 85-93;
Aggregation, level of, 19-21,66,110, of manpower planning, 1-5
116-17,130-35 Delphi method, 49-50
Allocation planning, 18,23,32-34, 111, Deterministic models, 40-41,63-64
114-15 Development and training planning, 17,
23,32-34,111,116
CAMP, 86 Dimension, 28-30
Career: planning, 15-17,23,32-34;
prospects, 6, 8,11,111-13,117, Economic models, 46-47
119-21 Economies of scale, 44, 57
Category, 86 Errors, prediction, 66-68, 76
Class, 86 Exponential model, 40
Classification, 21, 25-34 Extrapolation, 40-42
Cohorts, 68
Commitment, 3-4, 24, 28-34 Flexibility, 130-31, 136, 145
Complexity, 11 Forecasting: of manpower availability,
Computation time, 81-82, 5,35-36,60-85,137-38;of
107-08 manpower requirement, 5, 35-36,
Corporate planning, 13, 130 38-59, 137-38
Covariances, 66, 94 FORMASY, 84-157
Cross-sectional models, 68 Function seniority, 16
185
186 INDEX
Grade seniority, 16,57,78,79,81-82, Output: elasticities, 46; level, 43--50, 57,

86-155 59
Growth,6-8 Overstaffing, 118-20
Inflexibility, 23-34 Performance criterion, 138

Interactive, 84-143 Personnel. See Manpower distribution;
Manpower planning
Linear model, 40-41 Personnel department, 110, 142
Longitudinal models, 68-69 Personnel registration, 110, 140-42
Planning horizon, 116, 130-31, 135-40
Management opinion, 42, 49, 51-58, 59 Prediction errors, 66-68, 76
MANPLAN,86
Manpower distribution, 11, 15,60-157 Quality of forecasts, 66-68, 94-96
Manpower planning: activities in, 4, 13,
15-19,111-16; importance of, 5-12, Recruitment planning, 17,23,32-34,
56; interpretations of, 2-4; process of, 111,115-57
5,12,14,35,83-157; system of, Regression: criterion, 40; models, 45-46
83-157; terms of, 5-6,146 Renewal models, 60, 69-74, 79-83
Markov models, 60-68, 79-83
Matching, 15,35-38,53,60,74,137, Salary costs, 10-11,75,99,101
138 Simulation, 103, 108-09, 136-40
Methods: delphi, 49-50; statistical, Stationary distribution, 64, 73, 75,
39-48,59; subjective, 39, 48-51 96-97
Mobility, 35, 131 Steady-state distribution, 64, 73, 75,
Models: autoregressive, 41-42; 96-97
deterministic, 63-64; economic, Stochastic models, 63-64
46-47; exponential, 40; goal
programming, 76; linear, 40-41; Technology, 23-58
longitudinal, 68-69; Markov, 60-68, Training and development planning, 15,
79-83; multicategory, 19-20; 17-18,23,32-34
multivariate, 43-48; optimization, 60, Transition: fractions, 61-108,116-29;
74-77,80-83; pull, 69; push, 62; matrix, 62
regression, 45-46; renewal, 60, 69-74, Trends: deterministic, 40-41; stochastic,
79-83; stochastic, 63-64; univariate, 41,42
39-43,47 Typology, 21-34
Multicategory planning, 19-20
Multivariate models, 43-48 Uncertainty, 136
Univariate models, 39-42, 47
Objective function, 75-77, 80
Optimization models, 60, 74-77, 80-83 Variances, 66
Organizational planning, 13-14 Variation coefficient, 67
Organizations: defense, 16,52-53,56;
stable, 5-6; typologies of, 21-34; VVastage,6,60-79,131
unstable, 5-6 VVork-studymethod, 49, 50-51, 53

6471 - (International Series in Management Scien PDF

Uploaded by

Copyright:

Available Formats

6471 - (International Series in Management Scien PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

6471 - (International Series in Management Scien PDF

Uploaded by

Copyright:

Available Formats

Techniques in

Corporate Manpower Planning

Kluwer • Nijhoff Publishing

DISTRIBUTORS OUTSIDE NORTH AMERICA:

Library of Congress Cataloging in Publication Data

Copyright © 1982 by C. J. Verhoeven

1 Corporate Manpower Planning 1

2 Manpower Requirement Forecasting 38

3 Manpower Availability Forecasting 60

4 FORMASY: An Interactive Manpower Planning System 85

5 The Use of FORMASY 110

6 A Case Study 144

The title of this book is Techniques in Corporate Manpower Planning:

approach to manpower planning. In our view, this medium- and long-term

manpower requirement and availability forecasts must be assessed in order

The research was performed under the supervision of Prof. Dr. J.

Corporate manpower planning is the planning of personnel for large or-

stances are labor laws, the impossibility of involuntary dismissal or transfer

In this section we introduce some views on corporate manpower planning

Interpretations considering only organization interests

In the view of Vajda (1970) "manpower planning is concerned with arrang-

1. A regular monitoring activity, through which manpower and the re-

An interpretation considering the interests of employees also

Here we consider an interpretation of the tasks of manpower planning that

In our opinion, the realization of individual interests should be an essential

Our own view on manpower planning

In our view, manpower planning is a set of activities designed to keep avail-

development planning, and allocation planning) will be described later in

The forecasting of manpower requirement (both qualitatively and quan-

The distinction-which will be sustained throughout the book-follows

IMPORTANCE OF MANPOWER PLANNING

In this section we describe some circumstances that emphasize the impor-

be executed easily. In fact, the only manpower planning activities to be con-

Circumstances calling for manpower planning

Decreasing growth of organizations. Some organizations have known sev-

age age age

no. of no. of no. of

(a) (b) (c)

ically-may not be a desirable development (for instance, for a research lab

Another example of decreasing growth of an organization is encountered

Irregular age structure. In some organizations the actual age distribu-

1. The distribution of the qualifications of the employees is irregular;

tirement ages, training and development, together with changes in career

Increasing complexity of organizations. Organizations have become more

Length of the feedback cycle of manpower policies. Another reason for

The manpower decision-making process. Decisions with respect to man-

MANPOWER PLANNING AND ITS RELATION

In this section we consider the relations between manpower planning and

Manpower planning and organizational planning

The goal of organizational or corporate planning is the matching of re-

FIGURE 1.4. Information flows for the planning of

Factors influencing manpower requirement and availability

Particularly manpower requirement depends on the levels of other re-

FIGURE 1.5. Influences of the various variables on manpower availability and

Manpower planning activities

In principle, a match of manpower requirement and availability can be ob-

Career planning. Career planning is the development of career policies.

Recruitment planning. Recruitment planning is the development of re-

Development and training planning. Development and training planning