Content uploaded by Frank A.G den Butter
Author content
All content in this area was uploaded by Frank A.G den Butter
Content may be subject to copyright.
Research Report 0011/E
Zoetermeer, March 2001
Frank den Butter
Edwin van Gameren
Jan de Kok
The effects of transaction costs and human
capital on firm size: a simulation model ap-
proach
This paper has been written in the framework of EIM’s economic modelling
programme SCALES, which is financed by the Netherlands Ministry of Eco-
SCALES
SCientific AnaLysis of Entrepreneurship and SMEs
ISBN: 90-371-0815-6
Price: NLG 35.-
Order number: H0011
This report has been written in cooperation with ALERT, the Applied Labour Eco-
nomics Research Team from the Free University, Amsterdam
The underlying study forms part of the Programmaonderzoek MKB en Ondernemer-
schap (Programme Research SMEs & Entrepreneurship) that is financed by the Neth-
erlands Ministry of Economic Affairs.
EIM / Business & Policy Research employs 170 professionals. EIM provides pol-
icy- and practice-oriented socio-economic information on and for all sectors in
private enterprise and for policy-makers. EIM is established in Zoetermeer.
Besides on the Netherlands, EIM also focuses on the European economy and
on other continents. You may contact us for more information about EIM and
its services.
Address: Italiëlaan 33
Mailing address: P.O. Box 7001
2701 AA Zoetermeer
Telephone: + 31 79 341 36 34
Fax: + 31 79 341 50 24
Website: www.eim.nl
The responsibility for the contents of this report lies with EIM.
Quoting of numbers and/or texts as an explanation or support in papers, essays and
books is permitted only when the source is clearly mentioned.
No part of this publication may be copied and/or published in any form or by any
means, or stored in a retrieval system, without the prior written permission of EIM.
EIM does not accept responsibility for printing errors and/or other imperfections.
Contents
Summary.......................................................................................5
1 Introduction .................................................................................. 7
2 Theoretical background ............................................................... 9
3 Internal labour flows in a hierarchical model of the firm ....... 11
4 Firm size distributions ................................................................ 17
5 Labour productivity, specialization and firm size .................... 19
5.1 Labour productivity and firm size................................................................... 19
5.2 Specialization and firm size............................................................................. 21
6 Different approaches to explaining firm size distribution ......23
6.1 Simulating small and large enterprises........................................................... 23
6.2 Sensitivity analysis ............................................................................................ 26
7 Conclusions ................................................................................. 33
References .................................................................................. 35
Appendix
The calibrated simulation model..................................................................... 37
EIM Business & Policy Research 5
Summary
The effects of transaction costs and human capital on firm size …
Why do firms differ in size? Traditional microeconomic theory is based on the
concept of the representative firm, and can only provide a partial answer to
this question. Economies of scale, resulting from technical and allocational ef-
ficiency, can explain cross-industry differences in average firm size. It cannot
explain, however, the large variation in firm size within certain industries.
Other theories have been developed to explain this heterogeneity in firm size.
These include the transaction cost approach and the labour flow approach.
The transaction cost approach can be used to examine (transaction) costs of
the internal organization. The labour flow approach has pointed towards the
interaction between firm size on the one hand, and employee turnover and
internal labour flows on the other hand. In combination with the assumption
that employees can differ in their individual qualities, these approaches may
be used to explain differences in the size of firms within a certain sector and
country.
… a simulation model approach
To examine the relative importance of these approaches in explaining differ-
ences in firm size distribution, we use a simulation model. This model formal-
izes decisions of a profit-maximizing employer on the optimal number of em-
ployees. This decision depends on costs and benefits of recruiting, allocating
and supervising employees, and on the costs and benefits of cooperation be-
tween employees. Individual employees can have different qualities. The qual-
ity of new employees (and the exit rate of incumbent workers) is modelled as
a stochastic process. To account for these random processes, all simulation ex-
periments are replicated 100 times.
The parameters of the model are calibrated in such a way that the baseline
calibration generates a benchmark representative firm with several hierarchi-
cal levels. The flow characteristics of this firm (quits, fires, and internal and ex-
ternal worker flows) mimic the results found in scarcely available empirical
studies. The first simulation experiments examine to which extent the survival
rate and size of this benchmark firm depend on average labour productivity
and specialization. Next, this information is used to model two new bench-
mark firms: a small firm (25 employees on average) and a large firm (790 em-
ployees on average). The final set of simulation experiments consists of sensi-
tivity analyses performed on these two benchmark firms.
Conclusions
The model not only explains size differences between enterprises of different
sectors, but also why enterprises within the same sector may differ in size. The
6 EIM Business & Policy Research
model shows that there are two sources for such size differences of firms
within the same sector.
The first source is differences in transaction costs that may lead to size differ-
ences between otherwise comparable firms. This effect is already explained by
microeconomic theory, but the possibility of firm size effects has been ig-
nored. Our study demonstrates that such effects can exist: simulation experi-
ments show that small firms are less sensitive to changes in transaction costs
than large firms are.
The second source of size differences between firms of the same sector is het-
erogeneity of labour supply and the reaction of personnel management on
such heterogeneity. Even when transaction costs are the same for similar firms,
their sizes may differ due to the qualities and qualifications of incumbent
workers. The magnitude of this effect appears again to be different in small
businesses and large businesses. The relative influence that labour heteroge-
neity has on firm size depends crucially on the ratio between transaction costs
and wage costs. In our calibrated model, small businesses with a relatively flat
organization and few hierarchical levels face relatively few transaction costs,
but transaction costs gain importance when the number of hierarchical levels
rises and intra-firm bureaucracy increases. Therefore, the impact of labour
heterogeneity (and the scope for HRM) on business performance and firm size
is more severe in large enterprises than in small enterprises. Moreover, it is not
so much the quantity of (internal) labour flows but more so the quality of
these flows that matters for business performance.
EIM Business & Policy Research 7
1 Introduction
Traditional neo-classical microeconomic theory, based on the concept of the
representative firm, cannot explain why firms differ in size. Economies of
scale, resulting from technical and allocational efficiency, can explain cross-
industry differences in average firm size. It does, however, not explain the size
distribution of firms found in the real world. Other theories have been devel-
oped to explain the vast heterogeneity in firm size. Three alternative ap-
proaches to explain firm size may be distinguished (You, 1995). According to
the transaction cost approach (or institutional approach), firm size is deter-
mined by transaction cost efficiency1. Within the industrial organization ap-
proach, size distribution is explained by market power. Thirdly, the growth
approach focuses on the dynamics of the size distribution of firms. This ap-
proach includes life-cycle models and evolutionary models on firm growth. The
relevance of these approaches has been examined in a number of empirical
studies. For example, Davis and Henrekson (1999) have examined the role of
institutions in explaining national differences in firm size distribution, and Al-
mus and Nerlinger (1999) combined elements from the neo-classical and
growth approach to explain growth of new technology-based enterprises.
Recently, another possible determinant of firm size has received the attention
of economic research: labour flows. Analysis of large longitudinal datasets at
enterprise level has provided much insight into the specific characteristics of
labour flows between enterprises and establishments, and their connection
with labour market dynamics (Davis and Haltiwanger, 1990,1992; Davis et al.,
1996). A major finding is that the largest part of job turnover (job creation +
job destruction) takes place within the same regions and branches of industry.
It implies that job creation and job destruction are much more driven by idio-
syncratic, firm-specific shocks than by demand and supply shocks at macro
level. Another finding is that worker flows exceed job flows. For the Nether-
lands, Hamermesh et al. (1996) find that worker turnover is roughly three
times as large as job turnover.
Differences in firm size are likely to affect worker turnover and internal labour
flows, and vice versa. From that perspective, this study examines the relation
between firm size and internal labour flows. In doing so, we combine litera-
ture on labour flows with the standard neo-classical and transaction cost ap-
proaches to explain firm size. For this purpose, we construct an empirical simu-
lation model that incorporates various mechanisms that may be regarded as
underlying sources of firm heterogeneity. Although firm behaviour is de-
1 From a macroeconomic point of view, transaction costs include costs associated with existing
institutions. Transaction cost theory then effectively becomes an institutional theory of the
firm (You, 1995, page 448).
8 EIM Business & Policy Research
scribed by profit maximization, building and solving an analytical model that
incorporates all required sources of heterogeneity appeared impossible.
Therefore, we were forced to recourse to a simulation model that is calibrated
using the scarce empirical evidence on internal labour flows (Van Gameren,
2000). In this paper, we investigate in what manner this combination of theo-
ries explains firm size differences1.
The content of the paper is as follows. The next chapter reviews what the
literature on transaction costs, hierarchical models of the firm, equilibrium
search theory and internal labour markets can teach us about the underlying
sources of heterogeneity amongst enterprises. This gives us a hint on how to
make this heterogeneity operational in the model. Chapter 3 discusses how
these features are implemented in the simulation model. Next, we identify the
various characteristics of firm size distribution that we shall examine with this
model. The simulation outcomes are presented and discussed in chapters 5 and
6. Chapter 7 concludes.
1 This report is also published as a Free University Research Memorandum.
EIM Business & Policy Research 9
2 Theoretical background
Transaction costs
Adam Smith stated that firm size is determined by benefits and costs of spe-
cialization of labour, resulting in economies of scale. The degree of specializa-
tion would be limited mainly by the extent of the market1. Coase (1937) exam-
ined whether this specialization should take place within one single firm, or
between several enterprises. He introduced the idea of transaction costs, to
explain which transactions should take place in the market, and which transac-
tions are more efficient within the framework of an enterprise. Transaction
cost theory assumes that all transactions are costly due to bounded rationality
and opportunism. We adopt the concept of transaction costs to incorporate
the costs of an enterprise’s internal organization in our simulation model.
A transaction is any transfer of goods or services from one individual to an-
other. Transactions can take place either within or between firms; in this pa-
per we focus on transactions within firms, and, therefore, on internal transac-
tion costs. All transactions require coordination and cooperation from the in-
dividuals involved; hence, transaction costs can be classified into coordination
and motivation costs (Milgrom and Roberts, 1992). Internal coordination con-
sists of several steps (each resulting in internal coordination costs): obtaining
information needed to determine an efficient plan for a transaction; using the
knowledge available to determine the plan to be implemented; communicat-
ing the plan to those responsible for implementing it; and monitoring the
plan. Motivation costs may arise due to information incompleteness, informa-
tion asymmetries and imperfect commitment (resulting in hold-up problems).
Nooteboom (1993) and Garnsey (1998) argue that small businesses have a be-
havioural advantage over large enterprises, in that employees in smaller busi-
nesses are more motivated. This results in higher motivation costs (per em-
ployee) for larger enterprises.
These transaction costs refer to costs of vertical transactions: transactions that
involve different hierarchical levels within an enterprise. However, transac-
tions may also take place within a certain level. Becker and Murphy (1992) ar-
gue that both horizontal co-ordination and motivation costs per employee in-
crease with the number of employees.
Hierarchical models of the firm
Williamson (1967) used a hierarchical model of an individual enterprise to ex-
amine determinants of optimal firm size2. This model delineates a price-taking
1 Assuming that employers want to maximize their profits.
2 This model has later been augmented by Calvo and Wellisz (1978) and Keren and Levhari
(1979).
10 EIM Business & Policy Research
enterprise with m administrative levels. Each employee may supervise s subor-
dinates, i.e. the span of control equals s. Williamson also introduced a ‘compli-
ance’ or ‘loss of control’ parameter that indicates the effective contribution of
an employee to the objectives of his supervisor. The compliance parameter is
less than 1, reflecting that ‘only a fraction of the intentions of the superior is
effectively satisfied by a subordinate’ (Williamson, 1967). Without loss of con-
trol, the enterprise would have an infinite number of levels; in effect, its size
would be limited by the size of the market only.
The loss-of-control parameter in this model may be interpreted as a measure-
ment for vertical internal transaction costs. This interpretation becomes clear
when we explore Williamson’s argument to introduce the loss of control. The
intentions of the supervisor will never be fully satisfied because either the
communication between supervisor and subordinate is imperfect, or because
subordinates do not follow up on agreements made. The first explanation re-
flects (vertical) coordination costs, the second motivation costs. Williamson’s
(1967) conclusion that the compliance parameter must be less than 1 for en-
terprises to become finite is, therefore, similar to the conclusion by Coase
(1937) that firm size is finite due to positive (internal) transaction costs.
Becker and Murphy (1992) argue that optimal firm size is related to the de-
gree of specialization and coordination costs within hierarchical levels. In their
model, individual production increases with team size. This is because larger
teams allow for more specialization. The benefits of specialization are bal-
anced with the increasing costs of coordination between a larger number of
more specialized workers. They find that under some general conditions, team
size is limited by coordination costs instead of market size.
Equilibrium search and internal labour markets
An objective for our simulation model is that it should outline the dynamic
time path of the enterprise in response to various types of external shocks.
Therefore, the actual firm size should continuously be adapted to its optimal
size, taking account of adaptation costs (e.g. costs of hiring, firing and train-
ing). So as to outline this dynamic adjustment process, the model combines in-
sights of modern equilibrium search models and the flow approach of the la-
bour market (see Mortensen and Pissarides, 1998) with insights from internal
labour market models (Doeringer and Piore, 1971). These theories also explic-
itly take worker heterogeneity into account. This is another feature that our
model incorporates.
EIM Business & Policy Research 11
3 Internal labour flows in a hierarchical model of the firm
Our simulation model focuses on formalizing decisions of personnel managers
as regards the allocation of employees over the jobs available. The model
specifies hierarchical levels in line with Williamson (1967), and applies the the-
ory developed by Becker and Murphy (1992) to model benefits and costs of
cooperation within teams/levels. This allows us to endogenize the span of con-
trol, which is exogenous in Williamson (1967). In this chapter, we outline the
specification and calibration of the model1.
Independent decisions
The key assumption in our hierarchical model is that each employee decides
whether he or she spends time on the production of output or on supervising
subordinates (or on a combination of both) independently from others in the
enterprise. Under specific conditions, this yields identical results as when all
decisions are centralized. We consider this decision for each individual to be
taken by the management of the business rather than by employees them-
selves.
Benefits of specialization
Optimization starts with the highest-ranked person in the enterprise. He de-
termines the optimal number of subordinates for his circumstances by maximi-
zation of his contribution to the business’s profits, weighting the costs and
benefits of recruiting additional subordinates. The benefits consist of the pro-
duction generated by the subordinates. We specify individual production func-
tions, based on the quality of the employee. If more subordinates are re-
cruited, the tasks that must be performed and coordinated by the supervisor
may be divided over a larger group of subordinates. This results in specializa-
tion of the subordinates, which yields an increased individual productivity. A
subordinate’s contribution to firm production is, therefore, modelled as a
function of individual productivity and the number of subordinates within his
team.
Horizontal and vertical transaction costs
However, increasing the number of subordinates also increases horizontal
transaction costs (both coordination and motivation). We assume that the
horizontal transaction costs per subordinate rise with the number of subordi-
nates (Becker and Murphy, 1992). The combined effect of specialization and
horizontal coordination results in a parabolic relation between the profits of
1 The model is presented in the appendix. It is a variant of the model developed by Van
Gameren (2000). We refer to his study for further details.
12 EIM Business & Policy Research
the supervisor and the number of subordinates. Individual productivity may
benefit from cooperation with other subordinates (of the same supervisor) at
the same hierarchical level; however, hiring too many subordinates turns the
cooperation into a negative factor when specialization is extended too far.
The model also incorporates vertical transaction costs. These are the sum of
foregone production and loss of control. Foregone production measures the
time a supervisor has to spend on supervising and coordinating the subordi-
nate; while coordinating, he cannot contribute to production. The amount of
coordination required by a subordinate depends negatively on his qualities.
Loss of control measures the extent as to which, even after coordination, sub-
ordinates still will not be able/willing to produce the output required by their
supervisor. Besides horizontal and vertical transaction costs, the costs of sub-
ordinates depend on their wages.
Finite firm size
Coase (1937) concluded that enterprises have a finite size due to positive
transaction costs. This is also the case for our model: in the absence of any
transaction costs, firm size is limited by the size of the market only (which is
infinite, since we assume perfect competition). Transaction costs are necessary
to ensure both a finite number of hierarchical levels and a finite team size.
When the enterprise has reached its optimal size, marginal profits become
zero and additional subordinates are no longer beneficial. In other words, the
production technology is modelled in such a way that after a certain point, the
supply curve becomes upward sloping (due to decreasing returns to scale).
Adjustment costs
Until now, we discussed the costs and benefits of having subordinates. A ma-
jor feature of our model is that it contains costs for changing the number of
subordinates as well. Adjustment costs arise if the optimal number of subordi-
nates differs from the actual number. We identify three types of adjustment
costs. In the case of superfluous employees, the employees having the lowest
qualities are fired. The firm must pay firing costs for each fired employee. If
the enterprise has a staff shortage, it has to fill vacancies by searching for suit-
able employees. The enterprise searches, first, among the employees currently
employed in other jobs at the business. We assume the supervisor has insight
in the qualities of the employees in the next lower rank in the hierarchy (per-
fect information). Second, if the capacities required are not available within
the enterprise, the supervisor may decide to recruit a new employee. This ne-
cessitates an external search procedure that bears a higher cost level. External
applicants (i.e. their age and quality) are drawn from a random distribution;
the enterprise has no influence on the arrival of candidates. This mimics, in a
way, incomplete information. If a candidate fails to meet the minimum re-
quirements, a training procedure can be considered, at a certain cost. A possi-
EIM Business & Policy Research 13
ble outcome of the (external) application procedure is that the job remains va-
cant.
Operation of the simulation model
Optimization of the profits of the entrepreneur and his search for subordi-
nates – the mechanisms of which are both discussed above – are the first two
steps in the operation of the simulation model. The third step is that for each
filled job, i.e. for each subordinate, we repeat the optimization and search
procedure, by taking into account the central assumption: each employee
takes independent decisions on whether he works on the production of out-
put, on the supervision of subordinates, or a combination of both. The optimal
number of subordinates is independent from the decisions made at other
ranks and in other branches in the hierarchy: it is (modelled as) a purely indi-
vidual decision. As outlined above, we modelled the structure of the profit
function and the level of the transaction costs in such a way that they set a
limit on firm size. Under this condition, the number of repetitions of steps 1
and 2 is finite, and it is possible to delineate the enterprise by number of em-
ployees, organizational structure, generated output and number of unfilled
vacancies (step 4 in table 1). Notice that both the number of hierarchical ranks
and the number of subordinates per team are endogenous in the model.
After steps 1 to 4, we obtain the hierarchical set-up of the enterprise at the
onset of a period. All workers in the hierarchy remain at their jobs for (at
least) one period and produce output during this period. The passing of time
generates an increase in the experience of employees within the enterprise
(‘learning by doing’), which is implemented as an increase in their personal
measure of quality. The effect depends on the tenure in the current job and
has a random component. At the end of the period, a random number of em-
ployees decide to quit the company. Here, we may think of workers who find
jobs elsewhere, or workers who have other reasons to leave the labour force.
A fraction of the employees will retire; we impose a mandatory retirement
age. Furthermore, employees may get dismissed if their qualities do not meet
minimum requirements. This is possible only for employees who were re-
cruited at the beginning of the period, and needed a training course to en-
hance their qualities. If that training doesn’t lift their quality levels to mini-
mum requirements they will be dismissed. The result of quits, retirements and
fires is the opening of vacancies at the old positions. Instead of immediately
searching for candidates who may fill these vacancies (and the unfilled vacan-
cies remaining from the previous search process), we return to the optimiza-
tion process (step 1) to determine whether it is optimal to search for employ-
ees to fill the vacancies, or whether it is best to close the vacancies altogether.
The next steps in the modelling algorithm are conducted successively, as out-
lined above.
14 EIM Business & Policy Research
Table 1 Set-up of the simulation model
STEP 1 Determination of optimal number of subordinates (per supervisor, per time period)
STEP 2 In the case of vacancies: search for employees
• Promotion of insiders (causes vacancy chains)
• Recrui tment of outsiders (training might be necessary )
In the case of superfluous workers: dismiss the least qualified subordinates (the result of
this step might be that there remain unfilled vacancies)
STEP 3 Perform steps 1 and 2 for each subordinate until reaching the rank where the (optimal)
number of subordinates equals zero
STEP 4 Determine the number of employees, production (optimal, actual), hiring, firing, (flows,
costs) and organizational structure of the enterprise
STEP 5 • Random quits will occur
• There will be an increase in the experience of the employees who stay (‘learning by
doing’)
• Repeat steps 1 to 4 for the following period
Calibration of the model
The parameters of the model are calibrated upon the scarcely available em-
pirical evidence. Our baseline calibration generates a benchmark representa-
tive firm whose flow characteristics (quits, fires, and internal and external
worker flows) mimic the results found in a study by Hamermesh et al. (1996).
This study presents estimates of the annual worker flows in the Netherlands in
1990, drawn from a stratified sample of about 1,000 enterprises with 10 or
more employees1. The selection of the model parameters, to generate our
benchmark firm, is based on case studies on the internal economics of enter-
prises by Baker et al. (1994) and Van Gameren et al. (1999). Both case studies
utilize personnel records of a large enterprise, and specify how the internal
structure such as the span of control and the wage scales of the business are
organized. The calibration of search costs is based on linear approximation of
the quadratic adjustment cost function of Pfann and Verspagen (1989). Their
results suggest that in the case of small adjustments, recruitment costs seem to
be somewhat higher than the firing costs, while for more expansionary firms,
hiring costs increase exponentially. We assume that external search is more
expensive than internal search, which implies that the first option to fill vacan-
cies is through internal moves.
Simulation results with our calibrated model of a representative business are
shown in table 2. The flows are based on simulations over 50 periods (or
1 Allaart et al. (2000) use a more recent data set for the Netherlands (concerning 1996), and find
very similar worker flow estimates.
EIM Business & Policy Research 15
years)1, and replicated 100 times to account for the random processes incorpo-
rated in the model. Averages are taken over the last 25 years since during the
initial years, the business grows to its optimal size. The simulation results can
be compared with the results of Hamermesh et al. (1996) that are presented in
the column ‘target size’.
Table 2 Simulation results*
Simulation results
Type and direction of worker flows Mean Standard deviation Target size
Number of employees in the firm 123 1.90 136
Inflow
Hire to a newly created job (%) 2.0 1.70 3.2
Hire to an existing job (%) 7.8 3.39 8.7
Outflow
Quits/retirements (%) 7.6 2.97
Direct fires (%) 1.9 1.53 } 8.2
Outflow from a destroyed job (%) 0.35 0.66 1.9
Internal mobility
To a newly-created job (%) 0.04 0.24 1.2
To an existing job (by direct subordinate) (%) 0.2 0.51
To an existing job (from other team) (%) 2.2 1.82 } 2.2
* The target sizes are taken from Hamermesh et al. (1996), table 1, with the exception of the
number of employees in the firm, which is taken from Van Gameren (2000), table 6.3. All per-
centages denote percentages of employment. The sample standard deviations are calculated
over the last 25 periods.
Table 2 indicates that average firm size hardly fluctuates between the various
simulations: the standard deviation of the number of employees is small. A
closer inspection of the simulation results provides insights into the hierarchi-
cal structure. The employer hires four employees (say heads of units or plants).
Each of these employees wants to hire five subordinates (say heads of
branches), and is eventually able to keep these positions filled each period.
These subordinates (20 in total) also want to hire five employees each, but
they are not always able to keep these positions occupied (due to quits, re-
tirements or dismissals, and the absence of internal candidates). It is only at
this lowest level that the simulated enterprises show any variation in number
of employees. Apparently, the heterogeneity between enterprises (quality,
age and tenure of employees) is not large enough to change the optimal
number of subordinates and levels between enterprises.
The table shows that the random processes that hit the enterprise cause more
variation in some of the labour flows than in firm size. Our calibrated bench-
1 In the model, a period is defined as a year. This facilitates both the modelling of the ageing of
employees, and the calibration using estimates of annual flows.
16 EIM Business & Policy Research
mark model is able to reproduce the target values with respect to the inflow
and outflow of employees reasonably well. Its distribution of total outflow
over outflow from existing jobs (quits/retirements and direct fires1) and out-
flow from destroyed jobs seems less successful. Jobs are being destroyed (the
annual job destruction rate is 1.2%), but most employees whose jobs are de-
stroyed can find employment elsewhere in the enterprise. It is, however, im-
portant to notice that the target sizes taken from Hamermesh et al. (1996)
represent a growing business: total inflow (11.9%) exceeds total outflow
(10.1%). Our calibrated benchmark model represents a business in equilibrium,
with constant firm size, and inflow and outflow being equal. Hamermesh et
al. (1996) find that the dismissal rate (both direct fires and outflow from de-
stroyed jobs) is lowest for enterprises with constant employment level, which
suggests that the target size for outflow from a destroyed job is set too high.
Target values for internal flows are the most difficult to reproduce in the cali-
bration procedure. Internal mobility towards new jobs is very low: once a
business has stabilized, new jobs are almost exclusively created at the lowest
rank in the hierarchy (where vertical mobility is, by definition, not possible).
Vacancies that arise at higher levels are mostly filled by internal mobility; ex-
ternal inflow occurs almost exclusively at the lowest rank.
1 Direct fires occur when employees are fired because their qualities are insufficient. These em-
ployees directly leave the firm, whilst their jobs remain intact. Indirect fires occur when jobs
are destroyed; these employees can apply for vacancies elsewhere in the firm.
EIM Business & Policy Research 17
4 Firm size distributions
The size distribution of a population of firms may be described by various
characteristics. Our simulation model enables us to examine the following fea-
tures1:
1. The average size of firms that survive for a certain number of years: the
model allows for the possibility that enterprises do not survive after 50
years, either because the original owner cannot find a successor, or because
at a certain point all employees leave the business. Average firm size is
taken over all surviving enterprises.
2. The survival rate: the fraction of all simulated firms that survive after 50
years.
3. The average start-up length: the length of the start-up period is deter-
mined by the first year in which the business reaches a size of at least 95%
of the average size (for the scenario in table 2, the start-up length is 4
years).
4. The within-standard deviation of firm size: a measure of the average stan-
dard deviation within each firm, over all periods of time: it indicates how
the same firms differ in size over time.
5. The between-standard deviation of firm size: a measure for the difference
in average firm size between enterprises: it indicates how different firms
differ in size at the same time.
To examine these characteristics of firm size distribution, the simulation model
combines elements from various approaches to explaining firm size. The rele-
vance of the neo-classical (or microeconomic) and transaction cost approaches
have been examined before (You, 1995). Elements from these approaches that
are incorporated in the model are the relevance of labour productivity, wage
costs, costs and benefits of specialization and vertical transaction costs (loss of
control and costs of supervision). Equilibrium search theory is represented by
random quits of employees, search costs, and requirements for internal and
external candidates. Finally, to take account of the relevance of human capital
of individual employees, the simulation model allows for variation in the
qualities of external candidates, and (variation in the effects of) learning by
doing and firm-provided training.
The following two chapters examine the relevance of these approaches by as-
sessing their impact on the five size distribution characteristics. Chapter 5 fo-
cuses on the microeconomic and transaction cost approaches, by studying the
impact of changes in labour productivity and costs and benefits of specializa-
tion on firm size and labour flows. In chapter 6, the working of the neo-
1 With the exception of the survival rate, all characteristics are calculated over the last 25 peri-
ods.
18 EIM Business & Policy Research
classical mechanisms and the effect of transaction costs is compared with the
relevance of search theory and (heterogeneous) human capital of individual
employees for the size distribution of firms.
EIM Business & Policy Research 19
5 Labour productivity, specialization and firm size
5.1 Labour productivity and firm size
Our first simulation examines the impact of changes in average labour produc-
tivity (the annual production of a new employee with average quality). The
calibration discussed in the previous chapter resulted in a business with an av-
erage labour productivity of 150 units a year (with the price of a unit of pro-
duction normalized to 1). Figure 1 shows the relation between average labour
productivity and firm size. If average labour productivity is too low and does
not cover (transaction) costs, entrepreneurs don’t recruit any employees, and
enterprises do not survive after 50 years. At a certain threshold point1, labour
productivity becomes high enough to make it profitable to recruit employees,
and a level is added to the firm. The survival rate of enterprises now suddenly
shifts to 100%.
Figure 1 Relation between firm size and average labour productivity*
* The dotted lines represent average firm size +/- 2 x the standard deviation of firm size.
Firm size increases only if productivity becomes high enough to add a third
level to the firm, and later on a fourth level2. Changes in average labour pro-
ductivity have no effect on the size of the teams. Additional profits from in-
creased productivity are not large enough to justify the costs of increased
horizontal and vertical coordination that are associated with an expanding
team size.
This changes however, if average productivity is increased further. A small in-
crease at the next threshold point (from 155 to 156) now has two effects. An
additional level is added to the firm, which increases average firm size. More-
over, firms now differ also in the sizes of their teams. Not only at the fifth
1 At an average labour productivity of 136; this value depends on the other parameter values.
2 At an average labour productivity of 139 and 145, respectively.
0
200
400
600
800
1000
1200
130 135 140 145 150 155 160
Average labour productivity
No. of employees
20 EIM Business & Policy Research
level, but at all levels of the hierarchy. Variation in team size increases with hi-
erarchical level. This results in a large variation in firm size. Beyond this
threshold point, average firm size is determined by team size, and not so much
by the number of hierarchical levels (for example: increasing the average pro-
duction level from 155 to 160 raises average firm size, whilst the number of
hierarchical levels remains the same).
This intriguing change in the working of the model may be explained as fol-
lows. All stochastic elements in the simulation model are related to the human
capital of individual employees: qualities of new applicants, returns to train-
ing, effects of learning by doing and voluntary quits (voluntary quits result in
a loss of human capital, and open an opportunity to gain new human capital).
It is, therefore, the heterogeneity in the human capital available that causes
the variation in firm size. Two combined effects make this mechanism work in
large enterprises in particular. Below, we explain why.
Human capital influences marginal costs of production
A supervisor will hire employees as long as marginal benefits of the additional
production exceed marginal costs. The marginal benefits of an additional unit
of production are by definition equal to output price, which is normalized to
1. The marginal costs consist of marginal wage costs and marginal transaction
costs (supervision costs and adjustment costs). We assume marginal productiv-
ity wages: wages per output are independent of human capital. In contrast,
supervision costs per unit of output are negatively related to the amount of
human capital: more human capital implies both higher production and lower
supervision costs1. To conclude: human capital influences marginal costs of ad-
ditional production and, therefore, - in theory - the decision on how many
employees to hire.
This finding also explains why the variation in team size increases with the hi-
erarchical level. This is because with each additional level, transaction costs
(which depend on human capital available) increase relative to wage costs.
With each additional level, the costs of managing the hierarchical firm become
more important2.
1 The marginal adjustment costs of an additional unit of production are also negatively related
to human capital, but less strong than marginal supervision costs. This is because adjustment
costs are independent of human capital.
2 This is not a consequence of the structure of the model, but of the calibration process. Differ-
ent values for the parameters that determine transaction costs could result in different conclu-
sions (see appendix).
EIM Business & Policy Research 21
Human capital influences marginal costs and benefits of employees
Another way of analyzing the recruitment decision is to compare marginal
costs and benefits of recruiting an additional subordinate1. Due to the costs
and benefits of specialization, both marginal costs and marginal benefits de-
pend on the number of incumbents.
The human capital of (incumbent) workers influences both marginal costs and
marginal benefits of an additional subordinate. Whether this actually affects
the (discrete) recruitment decision, depends on how strong marginal costs and
benefits depend on the number of incumbent workers and on their human
capital. Our simulations show that average labour productivity must exceed a
certain threshold before human capital actually affects firm size.
5.2 Specialization and firm size
Firm size is determined by the number of workers within each team, and the
number of hierarchical levels. Williamson (1967) modelled the number of
workers within each team as an exogenous variable: the span of control. In his
model, increasing the span of control resulted in an increase in the number of
levels, so that the effect on firm size is twofold.
Instead, the span of control is endogenized in our model by introducing (the
costs and benefits of) specialization. The net contribution of an individual to
the total production of its team is the difference between the benefits of spe-
cialization and the costs of horizontal coordination. This combined effect is
modelled as a parabolic relation, along the lines of Becker and Murphy (1992).
Hence, we have an endogenous span of control determined by the efficiency-
maximizing team size (defined as the number of employees for whom the av-
erage net contribution per employee is maximal). This efficiency-maximizing
team size may be manipulated by simultaneously changing costs and benefits
of specialization.
Increasing the efficiency-maximizing team size from 1 to 7 employees in-
creases the average firm size from 4 to more than 700 employees (see figure
2). This is exactly according to the expectations of the traditional microeco-
nomic approach: economies of scale (or specialization) have a positive impact
on firm size. Contrary to Williamson (1967), we find that increases in the effi-
ciency-maximizing team size have no effect on the number of hierarchical lev-
els.
1 Both the number of subordinates within a team and the number of hierarchical levels are
determined by equating the marginal costs and benefits of recr uiting an additional
subordinate (the number of hierarchical levels may be found by deriving at which level it is
optimal to recruit zero subordinates). In the appendix, an equation is derived for this problem.
As in the model by Williamson (1967), this equation can only be solved numerically.
22 EIM Business & Policy Research
Figure 2 Relation between firm size and efficiency-maximizing team size
* The dotted lines represent the average firm size +/- 2 x the standard deviation of firm size.
The efficiency-maximizing team size is independent of wage and transaction
costs. As a result, the simulated (profit maximizing) team size is not equal to
the efficiency-maximizing team size. In fact, the simulated team size differs
between hierarchical levels (since transaction costs differ between levels). With
the exception of the highest level, the simulated team size is larger than the
efficiency-maximizing team size.
0
100
200
300
400
500
600
700
800
01234567
Efficiency-maximizing team size
No. of employees
EIM Business & Policy Research 23
6 Different approaches to explaining firm size distribu-
tion
This chapter examines the effects of changes in several model parameters, for
both a small (25 employees) and a large (600 employees) benchmark firm.
These parameters represent elements of the various approaches to explain
firm size distribution: the technical approach (wage costs, benefits of speciali-
zation), transaction cost approach (costs of specialization, loss of control and
supervision), equilibrium search theory (search costs, requirements for candi-
dates, quit rate) and human capital of individual employees (variation in quali-
ties of external candidates, effects of learning by doing and training).
6.1 Simulating small and large enterprises
The small firm is simulated by selecting the value for average productivity. This
yields an enterprise with three levels, with approximately 25 employees (see
table 3). The large enterprise is simulated by enhancing average productivity
so that profit maximization yields an enterprise with five levels. Adding a fifth
level results in higher standard deviations of firm size, both within and be-
tween enterprises (see table 4). Simulation experiments show that this effect
does not only occur when average labour productivity is enhanced: changes in
other parameters may also result in large enterprises with five levels and high
within-/between-standard deviations.
The average rates of in- and outflow are comparable for small and large en-
terprises: inflow is 9% of total employment for small and 11% for large enter-
prises. The difference is caused by the difference in average quit rates be-
tween small and large enterprises (because the quit rate differs between hier-
archical levels, large enterprises have ceteris paribus higher quit rates). The na-
ture of the flows differs, however. For small businesses, the majority of inflow
concerns existing jobs, whilst for large enterprises it is mostly inflow into
newly created jobs. There are fewer fires in large than in small businesses. The
outflow from destroyed jobs is very similar.
24 EIM Business & Policy Research
Table 3 Simulation results for a small benchmark firm
Type and direction of worker flows
Average
Within-standard
deviation
Between-standard
deviation
Survival rate 99%
Start-up length (years) 4
Number of levels 3 0 0
Number of employees in the firm 24.5 0.72 0.16
Inflow
Hire to a newly created job (%) 1.8 0.67 0.11
Hire to an existing job (%) 7.2 1.29 0.25
Outflow
Quits/retirements (%) 6.4 1.22 0.14
Direct fires (%) 2.3 0.74 0.16
Outflow from a destroyed job (%) 0.4 0.29 0.05
Internal mobility
to a newly-created job (%) 0 0.02 0.00
to an existing job (by direct subordinate) (%) 0.6 0.40 0.01
to an existing job (from other team) (%) 1.0 0.50 0.03
The internal mobility is clearly higher for large enterprises as they have more
opportunities for job movers than small businesses have. This is due to the lar-
ger pool of incumbent workers with sufficient qualifications. This result is in
accordance with the findings of Hamermesh et al. (1996) and Hassink (1996).
EIM Business & Policy Research 25
Table 4 Simulation results for a large enterprise
Type and direction of worker flows
Average
Within-standard
deviation
Between-standard
deviation
Survival rate 99%
Start-up length (years) 19
Number of levels 5
Number of employees in the firm 792 44.2 93.9
Inflow
Hire to a newly created job (%) 8.6 48.2 25.4
Hire to an existing job (%) 2.4 15.7 16.0
Outflow
Quits/retirements (%) 9.3 9.3 8.5
Direct fires (%) 0.9 5.0 1.0
Outflow from a destroyed job (%) 0.6 13.1 2.7
Internal mobility:
to a newly created job (%) 3.5 21.6 18.6
to an existing job (by direct subordinate) (%) 0.1 0.8 0.3
to an existing job (from other team) (%) 9.4 16.5 27.4
Figure 3 illustrates the variance in firm size for the large enterprise, both
within individual businesses over time (measured by the within-standard de-
viation) and between enterprises (measured by the between-standard devia-
tion). It shows the development of four different simulated firms: a firm with
an average within-deviation, the firm with the highest within-deviation, and
the firms with - on average - the most and least employees. It is clear from fig-
ure 3 that the number of employees often changes; this is also reflected in ta-
ble 4 by the fact that variation in inflow (and internal mobility) far exceeds the
variation in outflow rates.
26 EIM Business & Policy Research
Figure 3 Four different ‘large’ enterprises
Allaart et al. (2000) have calculated worker flows for different size classes. This
enables us to compare our simulation outcomes with some empirical informa-
tion. Allaart et al. (2000) find that in- and outflow of workers vary less with
size class than internal flows do. Particularly for enterprises with 20-49 em-
ployees they find that inflow equals 12.5% of the number of employees, out-
flow 10.5% and internal flow 2.3%. For enterprises with more than 500 em-
ployees, these volumes are 12.6%, 11.1% and 6.9%, respectively. These results
are comparable with Hassink (1996), who finds internal labour flows of 2.4%
for businesses with less than 100 employees, and 4.9% for enterprises with
more than 100 employees. Our model reproduces the (small) difference in out-
flow between small and large enterprises rather well, but it underestimates in-
ternal mobility for small businesses, and overestimates internal mobility for
large enterprises.
6.2 Sensitivity analysis
By way of sensitivity analysis, our final set of simulation experiments illustrates
the influence of parameter changes on size distribution of firms and on labour
flows. These parameter changes represent various options for changing the
performance of the business. They may be associated with the various theo-
retical approaches to explain heterogeneity amongst enterprises, which are
combined in the model. The aim of these simulations is to give some indica-
tion, both in the case of a small business and of a large enterprise, of the rela-
tive impact (on firm size) of various ways in which enterprises may adapt their
production process, internal organization and personnel management. Simu-
lations are conducted with the following parameter changes:
1. Wage costs: these costs are defined by two parameters, viz. wagc, repre-
senting the wage at the highest hierarchical level and wagl, representing
wage differences between the hierarchical levels. It should be noted that a
change in wagc, given wagl, implies a change of all wages in the company.
0
200
400
600
800
1000
1200
0 5 10 15 20 25 30 35 40 45 50
Age of firm (years)
No. of employees
average small large high (var)
EIM Business & Policy Research 27
Also, since product price is fixed and the model assumes completely elastic
product demand, a change of wage costs of the enterprise should be inter-
preted as a firm-specific change and may not be considered the conse-
quence of a general wage restraint or wage push.
2. Specialization: costs and benefits of specialization are represented by two
different parameters.
3. Vertical transaction costs: here, the model includes 3 parameters which rep-
resent various types of vertical transaction costs, viz. the loss-of-control pa-
rameter, indicating vertical motivation costs, and two parameters which de-
termine foregone production due to supervision (fgpc, representing the
costs at the highest hierarchical level, and fgpl, representing cost differ-
ences between the hierarchical levels).
4. Search costs: external and internal search costs have been altered
proportionally, as both costs have the same influence on the working of
the model.
5. Search requirements: here, changes in three parameters are considered, viz.
in the ‘baseline’ minimum requirements reqc (the minimum requirements
at the highest hierarchical level), the differences in requirements between
hierarchical levels reql, and the additional minimum requirement for an in-
ternal applicant, rqie. Given the average human capital of external appli-
cants, lowering the requirements enhances the probability of finding a
suitable applicant, but lowers the average quality of employees. Increasing
additional minimum requirements for internal applicants enhances the av-
erage quality of those who are promoted but decreases internal mobility.
6. Quit rate: the probability that employees decide to leave the enterprise, for
other reasons than retirement.
7. Average human capital: here, changes may occur due to a change in the
average quality of external applicants (pdfu) or a change in the effects of
learning by doing and training (grwe).
Tables 5 and 7 show the effect on the characteristics of the size distribution of
the firms and on labour flows1, when the parameter changes represent an in-
crease in the performance of the business; tables 6 and 8 show the effect of
opposite changes in these model parameters.
As the response of our model to various shocks and parameter changes is
highly non-linear, mainly as a consequence of ratchet effects (change of num-
ber of levels), it appears that in some cases, small businesses react less strongly
to changes than large enterprises do. Therefore, we have conducted our simu-
1 Since our model represents firms in equilibrium, outflow and inflow rates are virtually identi-
cal. We therefore only present the inflow rates in our tables.
28 EIM Business & Policy Research
lation experiments with larger parameter changes for the small business (ta-
bles 5 and 6) than for the large enterprise1 (tables 7 and 8).
Table 5 Changing parameter values to stimulate performance of small businesses
Wages Specialization Vertical transaction costs
bench-
wagc
wagl
benefits
costs
loss of
control
fgpc
gpl
mark -5% -2.5% +2.5% -5% -1% -10% -2.5%
Survival 99% 100% 99% 100% 99% 100% 100% 100%
Size (no. empl.) 24.5 24.5 122.9 123.8 151.0 122.7 122.9 122.8
Within st. dev. 0.7 0.7 1.8 2.3 4.8 1.7 1.5 1.5
Between st. dev. 0.2 0.1 0.4 0.7 6.1 0.3 0.3 0.3
Inflow 9.0% 9.0% 9.8% 9.7% 9.5% 9.8% 9.8% 9.9%
Internal mobility 1.6% 1.5% 2.4% 2.5% 4.5% 2.3% 2.0% 2.0%
Search Search requirements Quit Human capital
costs reqc reql rqie rate pdfu grwe
-10% -35% -25% -50% -2% pt +30% +75%
Survival 99% 100% 100% 100% 100% 100% 100% 100%
Size (no. empl.) 24.5 122.1 24.5 24.4 24.5 24.6 24.5 24.6
Within st. dev. 0.7 1.7 0.7 0.7 0.7 0.6 0.7 0.7
Between st. dev. 0.2 0.3 0.2 0.2 0.1 0.1 0.2 0.1
Inflow 9.0% 9.7% 7.1% 6.8% 9.2% 5.8% 7.7% 8.5%
Internal mobility 1.6% 2.6% 3.3% 3.7% 1.6% 1.6% 2.3% 1.8%
1 For example: for large firms, the parameter on the effects of learning by doing and training
(grwe) was changed with +/- 10% (tables 7 and 8). For small firms, this change had no effect.
Instead, tables 5 and 6 report the effects of changes of +/- 75%.
EIM Business & Policy Research 29
Table 6 Changing parameter values to hamper performance of small businesses
Wages Specialization Vertical transaction costs
bench-
wagc
wagl
benefits
costs
loss of
control
fgpc
fgpl
mark +5% +2.5% -2.5% +5% +1% +10% +2.5%
Survival 99% 100% 100% 100% 100% 100% 100% 99%
Size (no. empl.) 24.5 4.8 4.9 4.9 4.8 4.9 4.9 24.4
Within st. dev. 0.7 0.4 0.4 0.4 0.4 0.4 0.4 0.7
Between st. dev. 0.2 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Inflow 9.0% 9.5% 9.2% 8.6% 9.8% 9.8% 9.1% 9.2%
Internal mobility 1.6% 1.5% 1.5% 1.4% 1.5% 1.4% 1.4% 1.7%
Search Search requirements Quit Human capital
costs reqc reql rqie rate pdfu grwe
+10% +35% +25% +50% +3% pt -30% -75%
Survival 99% 100% 6% 49% 69% 99% 68% 57%
Size (no. empl.) 24.5 4.8 23.4 3.8 24.4 24.0 23.6 24.2
Within st. dev. 0.7 0.4 1.2 1.4 0.8 1.0 1.1 0.9
Between st. dev. 0.2 0.1 0.4 0.7 0.1 0.2 0.3 0.2
Inflow 9.0% 9.9% 13.0% 31.3% 9.3% 18.3% 15.6% 12.4%
Internal mobility 1.6% 1.5% 1.6% 3.7% 1.4% 2.4% 2.1% 5.0%
Table 7 Changing parameter values to stimulate performance of large enterprises
Wages Specialization Vertical transaction costs
bench-
wagc
wagl
benefits
costs
loss of
control
fgpc
fgpl
mark -2.5% -2.5% +2.5% -2.5% -0.5% -5% -2.5%
Size (no. empl.) 792 884.6 1020.8 1176.3 1171.6 849.7 837.9 848.6
Within st. dev. 44.2 44.7 47.3 30.9 29.1 44.9 44.5 40.8
Between st. dev. 93.9 112.5 115.3 1.9 40.5 59.9 99.1 97.2
Inflow 11.0% 10.7% 10.6 10.6 10.5% 10.9 10.9% 10.8%
Internal mobility 13.0% 14.2% 12.0% 7.9% 7.6% 14.7% 14.3% 14.5%
Search Search requirements Quit Human capital
costs reqc reql rqie rate pdfu grwe
-5% -10% -10% -10% -2% pt +10% +10%
Size (no. empl.) 792 849.6 801.0 817.7 796.5 910.8 809 845
Within st. dev. 44.2 42.9 44.0 46.5 43.4 43.0 44.5 46.0
Between st. dev. 93.9 107.0 97.7 108.6 86.2 34.3 72.0 42.1
Inflow 11.0% 10.8% 10.7% 10.0% 10.9% 4.2% 10.6% 11.0%
Internal mobility 13.0% 15.7% 14.1% 14.6% 13.7% 12.2% 14.4% 17.5%
30 EIM Business & Policy Research
Table 8 Changing parameter values to hamper performance of large enterprises
Wages Specialization Vertical transaction costs
bench-
wagc
wagl
benefits
costs
loss of
control
fgpc
fgpl
mark +2.5% +2.5% -2.5% +2.5% +0.5% +5% +2.5%
Size (no. empl.) 792 709.8 122.7 122.2 609.8 123.6 779.5 123.3
Within st. dev. 44.2 34.3 1.6 1.6 4.2 2.2 43.1 2.1
Between st. dev. 93.9 95.7 0.3 0.4 0.7 0.5 88.7 0.6
Inflow 11.0% 10.8% 9.9% 9.8% 10.5% 9.9% 10.9% 9.8%
Internal mobility 13.0% 10.3% 2.3% 2.5% 3.9% 2.5% 13.6% 2.7%
Search Search requirements Quit Human capital
costs reqc reql rqie rate pdfu grwe
+5% +10% +10% +10% +3% pt -10% -10%
Size (no. empl.) 792 123.8 800.3 736.4 806.5 622.7 758.9 666.1
Within st. dev. 44.2 2.1 38.8 32 43.6 15 37.9 26.6
Between st. dev. 93.9 0.5 82.4 100 79.8 19.1 101.9 83.1
Inflow 11.0% 9.8% 11.1% 13.2% 10.8% 24.6% 11.2% 10.7%
Internal mobility 13.0% 2.4% 13.7% 9.4% 13.8% 7.3% 11.7% 6.3%
These simulation exercises lead to the following conclusions with respect to
differences between small and large enterprises:
• Small businesses react less strongly to changes in their (internal and exter-
nal) environment than large enterprises do.
• Large enterprises always survive the 50-year period of our simulations
(therefore, the survival rate has not been reported in tables 7 and 8). For
small businesses, this is not the case.
• Variation in firm size is always due to variation in team size (at all levels),
never in number of levels. The number of hierarchical levels is determined
by model parameters representing the production and management of the
business, and not by the stochastics of the internal and external labour
markets.
• As long as enterprises have no more than four levels, firm size shows very
little variation over the simulations (4.9, 24.5 and 123 employees for 2, 3
and 4 levels, respectively). In contrast, there is much more variation in busi-
nesses with five levels (610 to 1,176 employees). This variation is ultimately
caused by the variation in human capital of individual employees. Appar-
ently, the factors that cause enterprises to become so large that their or-
ganization consists of five hierarchical levels, also enhance the relation be-
tween optimal firm size and human capital of their (incumbent) employees.
This suggests that with large (5-leveled) firms, personnel management (hir-
ing and selection of new employees and internal mobility of incumbent
employees) and organizational changes may influence firm size.
EIM Business & Policy Research 31
With respect to the technicalities of the production process (wage costs, costs
and benefits of specialization), the following conclusions emerge from the
simulation experiments:
• Changes in the respective parameter values have a clear effect on the num-
ber of employees. For small businesses, the number of employees changes
because a hierarchical level is added or removed. For large enterprises, lev-
els may be removed, but a sixth level is never added to the firm.
• The benchmark model of the large enterprise shows a substantial variation
of average size between firms, which is caused by the heterogeneity of the
labour force. Changes in the benefits and costs of specialization may, how-
ever, make the heterogeneity of employees become irrelevant again (as is
the case for the small businesses). Enhancing the benefits of horizontal co-
ordination by 2.5% results in a 48% increase of average firm size, while the
between-firm standard deviation reduces almost to nil (table 7). A 2.5% in-
crease of the costs of horizontal co-ordination lead to a 23% decrease of
average size (without removing a hierarchical level), and again the be-
tween-firm standard deviation becomes very small (table 8).
With respect to the features of equilibrium search theory (search costs, re-
quirements for candidates, quits) that are incorporated in the model, the
simulation experiments of this sensitivity analysis give rise to the following
conclusions:
• Search/adjustment costs have a substantial influence on the equilibrium size
of the firm: increasing costs have a negative effect on the number of em-
ployees, both for small and for large enterprises.
• Factors that determine the quality requirements for internal and external
candidates have a different effect on small and large enterprises: for small
businesses, they influence the survival rate (and the size of the inflow), and
for large enterprises, they influence the number of employees.
• The observed variation in firm size for the large enterprises depends
strongly on the quit rate: if employees do not leave the company (except
when retiring), the between-firm standard deviation is reduced from 93.9
to 19.1.
• An increase in the quit rate leads to a rise in direct dismissals. The underly-
ing mechanism is that an increase in quit rate results in a rise of external
recruitments. With a constant fraction of new employees being dismissed
after a year (because their qualities turned out to be insufficient), an in-
crease in external recruitments leads to a rise in outflow by direct fires.
Finally, with respect to human capital of individual employees (variation in
qualities of external candidates, effects of learning by doing and company-
provided training), the following conclusions are in order.
• Decreasing the available quality and/or lowering the effects of learning by
doing and the returns to training has a negative effect on the survival rates
of small businesses and on the average size of large enterprises.
32 EIM Business & Policy Research
• For small businesses, the characteristics of individual employees are nega-
tively related with inflow (and outflow): if the average quality is higher
and/or training becomes more effective, then in- and outflow rates decline.
For large enterprises, the characteristics of individual employees have no
effect on inflow rates.
• The relations with internal mobility are less clear for small businesses than
for large enterprises. For large enterprises, there is a positive relation with
internal mobility. With small businesses, both increases and decreases in the
relevant model parameters seem to have a positive effect on internal mo-
bility rates.
EIM Business & Policy Research 33
7 Conclusions
By using a calibrated simulation model, this paper provides a quantitative view
on the importance of various determinants of the size distribution of firms. Al-
though the model has a neo-classical background in the sense that optimal
firm size is determined by profit maximizing, it combines a number of other
approaches from economic literature which aim at explaining firm heteroge-
neity and variations in firm size. In this respect, our model pays ample atten-
tion to the various forms of transaction costs. Moreover, the model delineates
external and internal labour flows; and in doing so, it shows how the per-
formance of the firm and, therefore, its size is influenced by aspects of human
capital and personnel management, such as hiring costs, firing costs, search
costs, wage policy, training, job matching and setting requirements for worker
qualification. In fact, our modelling exercise fully appreciates the observation
by Conlisk (1996) that ‘a central insight is that the existence, size, structure and
workings of organizations are critically shaped by a need to economize on
various transaction costs’. Our model is capable of reproducing all these in-
sights, and the experiments with the model show the relative effectiveness of
such economizing.
The sophistication of the model does not only enable us to explain size differ-
ences between enterprises of different sectors (which had already been ex-
plained by microeconomic theory) but, also, to explain why enterprises within
the same sector may differ in size. The model shows that there are two sources
for such size differences of firms within the same sector. The first source is dif-
ferences in transaction costs that may lead, as theory predicts, to size differ-
ences between firms that operate otherwise in the same circumstances. Our
simulation experiments also show that the elasticity of transaction costs - i.e.
the difference in firm size evoked by a 1% difference in transaction costs - de-
pends on firm size itself.
The second source of size differences between firms of the same sector is het-
erogeneity of labour supply and the reaction of personnel management on
such heterogeneity. Even when transaction costs are the same for similar firms,
their sizes may differ due to the qualities and qualifications of incumbent
workers. The magnitude of this effect appears again to be different in small
businesses and large enterprises. The relative influence that labour heteroge-
neity has on firm size depends crucially on the ratio between transaction costs
and wage costs. In our calibrated model, small businesses with a relatively flat
organization and few hierarchical levels face relatively few transaction costs,
but transaction costs gain importance when the number of hierarchical levels
rises and intra-firm bureaucracy increases. Therefore, the impact of labour
heterogeneity (and the scope for HRM) on business performance and firm size
is more severe in large enterprises than in small businesses. This conclusion is
34 EIM Business & Policy Research
in line with Boone and van Witteloostuijn (1996), who find that the impact of
human capital is more pronounced in large than in small businesses. More-
over, it is not so much the quantity of (internal) labour flows but more so the
quality of these flows that matters for business performance.
Transaction costs may be categorized into coordination costs and costs of mo-
tivation. Coordination costs are indicators for the quality of management and
for how well structured the organization of the enterprise is. In smaller busi-
nesses, where the owner is both entrepreneur and manager, coordination
costs also relate to entrepreneurial qualities. The model simulations show that
the success and survival probabilities of new businesses depend heavily on
these entrepreneurial qualities.
The specification and calibration of our model needed a number of assump-
tions on both the shape and parameter values of the production process and
the transaction costs associated with company management. Although we
have exploited as much as possible existing empirical evidence for specifying
and calibrating the model, it is obvious that considerable part of the informa-
tion that is crucial for the working of the model, is still lacking. E.g. much
more empirical data are needed in order to come to a more robust specifica-
tion of the relationship between the span of control, vertical and horizontal
transaction costs and optimal team size. The sensitivity analysis of the previous
chapter indicates that these data, and data on human capital and costs associ-
ated with hiring, firing, quitting and training, are essential for a better under-
standing of the reasons why profit-maximizing businesses differ in size. Col-
lecting these data in individual case studies of enterprises seems an important
scope for future research. Our modelling exercise provides a framework for
the collection of these data.
EIM Business & Policy Research 35
References
Allaart, P., M. Kerkhofs and M. de Voogd-Hamelink (2000), ‘Simultaneous job
creation and job destruction on establishment level’, OSA Working Paper
WP2000-7.
Almus, M. and E.A. Nerlinger (1999), ‘Growth of new technology-based firms:
which factors matter?’, Small Business Economics, 13, 141-154.
Baker, G., M. Gibbs and B. Holmstrom (1994), ‘The internal economics of the
firm: Evidence from personnel data’, Quarterly Journal of Economics, 109, 881-
919.
Becker, G.S. and K.M. Murphy (1992), ‘The division of labour, coordination cost
and knowledge’, Quarterly Journal of Economics, 57 (4), 1137-1160.
Boone, C. and A. van Witteloostuijn (1996), ‘Industry competition and firm
human capital’, Small Business Economics, 8, 347-364.
Calvo, G.A. and S. Wellisz (1978), ‘Supervision, loss of control and the optimum
size of firm’, Journal of Political Economy, 86, 943-952.
Coase, R.H. (1937), ‘The nature of the firm’, Economica, 386-405.
Conlisk, J. (1996), ‘Why bounded rationality’, Journal of Economic Literature,
34, 669-700.
Davis, S.J. and J. Haltiwanger (1990), ‘Gross job creation and destruction: Mi-
croeconomic evidence and macroeconomic implications’, NBER Macroeconom-
ics Annual, 5, 123-168.
Davis, S.J. and J. Haltiwanger (1992), ‘Gross job creation, gross job destruction
and employment reallocation’, Quarterly Journal of Economics, 107, 819-863.
Davis, S.J., J. Haltiwanger and S. Schuh (1996), Job Creation and Destruction,
Cambridge: MIT Press.
Davis, S.J. and M. Henrekson (1999), ‘Explaining national differences in the size
and industry distribution of employment’, Small Business Economics, 12, 59-83.
Doeringer, P.B. and M.J. Piore (1971), Internal Labour Markets and Manpower
Analysis, Lexington, Mass.: Heath Lexington Books.
Gameren, E. van (2000), The internal economics of firms, Thela Thesis, the
Netherlands.
Gameren, E. van and M. Lindeboom (1999), Promotion probabilities vs. exits,
mimeo, Tinbergen Institute/Vrije Universiteit Amsterdam.
Hamermesh, D.S., W.H.J. Hassink and J.C. van Ours (1996), ‘Job turnover and
labour turnover: a taxonomy of employment dynamics’, Annales d’Economie
et de Statistique, 41/42, 21-40.
36 EIM Business & Policy Research
Hassink, W.H.J. (1996), ‘An Empirical Note on Job Turnover and Internal Mobil-
ity of Workers’, Economics Letters, 51(3), 339-344.
Keren, M. and D. Levhari (1979), ‘The optimum span of control in a pure hier-
archy’, Management Science, 25, 1162-1172.
Milgrom, P. and J. Roberts (1992), Economics, organization and management,
Prentice Hall.
Mortensen, D.T. and Ch.A. Pissarides (1998), ‘New developments in models of
search in the labor market’, mimeo, prepared for the Handbook of Labor Eco-
nomics, 3, Chapter 39.
Nooteboom, B. (1993), ‘Firm size effects on transaction costs’, Small Business
Economics, 5, 283-295.
Pfann, G.A. and B. Verspagen (1989), ‘The structure of adjustment costs for la-
bour in the Dutch manufacturing sector’, Economics Letters, 29, 365-371.
Williamson, O.E. (1967), ‘Hierarchical control and optimum firm size’, Journal
of Political Economy, 75, 123-138.
You, J.I. (1995), ‘Small firms in economic theory’, Cambridge Journal of Eco-
nomics, 19, 441-462.
EIM Business & Policy Research 37
Appendix: The calibrated simulation model
In this appendix, the simulation model is described. First, the specification of
the simulation model is provided, including the calibrated model parameters
of the baseline simulation. Next, the marginal costs and benefits of hiring ad-
ditional subordinates are derived. For the special case where all employees
have the same average quality, we examine under which conditions a positive
and finite team size (the profit-maximizing number of subordinates) is guar-
anteed.
Specification
The profit function in period t of each supervisor (occupied at hierarchical
level i) is specified as:
å
=
−+++ −−−=
t
n
j
tttijtijtijtttit nnACfpwqnapnprf
0
*
1,1,,1,,1, ),(})({)( ,
and is maximized with respect to the number of subordinates nt (with j=0 to nt
an index of the subordinates). Here, *
1−t
n is the currently available number of
subordinates.
The constituent parts are the following functions:
Production: qj,i,t = γq,1 (γq,2)i cj,i,t,
with γq,1 = 150 and γq,2 = 0.85
(loss-of-control parameter).
Small and large benchmark firms are simulated by assuming an average
production of 141 and 154, respectively.
Supervision costs: fpj,i,t = γfp,1 (γfp,2)i (1/cj,i,t),
with γfp,1 = 37.5 and γfp,2 = 1.
Cooperation: a(n) = -
α
n2 +
β
n + 1,
Costs:
α
= 0.05625,
Benefits:
β
= 0.45.
Wage wj,i,t = γw,1 (γw,2)i cj,i,t,
with γw,1 = 175 and γw,2 = 0.75.
In this specification, the loss of control (vertical transaction costs) is incorpo-
rated in the production function. The costs of cooperation (horizontal transac-
tion costs) and benefits of cooperation are combined into a parabolic func-
tion.
The function pt defines the price of the output as it will be received by the
firm. In our simulations, we assume a constant price: pt = 1.
38 EIM Business & Policy Research
The relative quality measure cjit is defined as:
cjit = qualjt / E(qAit) if the job is occupied by subordinate j, and
= 1 if the job is vacant,
where qualjt is the actual quality of subordinate j, and E(qAit) the expected
quality of an external applicant.
Dynamics are incorporated in the model through the adjustment costs,
AC(nt, *
1−t
n) = –frit (nt–*
1−t
n) I(nt–*
1−t
n<0) + siit min(nt–*
1−t
n, nint) +
+ seit max(nt–*
1−t
n–nint, 0) I(nt–*
1−t
n>0),
with nint the number of potential internal candidates for the job (defined as
the number of employees at the next lower level), and I(.) the indicator func-
tion. The three different search costs are:
External search costs seit = 50
Firing costs frit = 0.5 seit
Internal search costs siit = 0.66 seit .
Both quality and age of external candidates are drawn from a uniform prob-
ability density function with upper and lower bounds that vary per level:
Quality of applicant qAit ~ UNIF(qlbit, qubit),
qlbit = γqlb,1 (γqprob,2)i,
qubit = γqub,1 (γqprob,2)i,
with γqlb,1 = 3, γqub,1 = 11.6 and γqprob,2 = 0.87.
Hence, expected quality is equal to E(qAit) = γprob,1 (γprob,2)i, with γprob,1 = (γqlb,1 +
γqub,1)/2 = 7.3.
Age of applicant Agejt ~ UNIF(albit, aubit),
albit = γalb,1 (γaprob,2)i,
aubit = γaub,1 (γaprob ,2)i,
with γalb,1 = 40, γaub,1 = 80 and γapro b,2 = 0.80.
The ‘baseline’ minimum requirements are given by:
mqit = γmq,1 (γmq,2)i,
with γmq,1 = 10 and γmq,2 = 0.75.
For an internal applicant, the minimum requirements are:
mqi,t
int = (1+
γ
RQIE) mqi,t,
with
γ
RQIE = 0.10,
while for an external applicant, the minimum requirements equal:
mqi,t
ext = (1–
γ
RQTR) mqi,t,
with
γ
RQTR = 0.20.
EIM Business & Policy Research 39
Furthermore, we have a number of functions that specify the relations be-
tween successive periods. We have a random quit probability, which is actually
defined as the probability that an employee remains in the firm, (1–γQUIT)i, with
γQUIT = 0.02. There is a retirement age: an employee who reaches the age of γRETR
= 60 is retired. For each worker who does not quit, we introduce an accumula-
tion of quality, qualj,t+1 = qualjt (1+2(½)tenJjtU), where U is the random factor
drawn from a uniform distribution U ~ UNIF(0, γGRWE) with γGRWE = 0.20, and tenJjt
is the tenure in the current job.
Marginal costs and benefits
The optimal number of subordinates may be determined by comparing the
marginal costs (mc) and benefits (mb) of hiring an additional subordinate. The
marginal costs for a (potential) supervisor of hiring an (additional) subordinate
are the sum of additional coordination costs, supervision costs, wage costs and
adjustment costs. For a supervisor at hierarchical level (i-1) who currently em-
ploys N subordinates (N≥0), the marginal costs mci-1,N+1 of hiring an additional
subordinate are1:
The marginal benefits of hiring an additional subordinate are the sum of the
benefits from the additional employee, and the changes in the benefits of the
incumbent N employees. These changes are caused by changes in the benefits
of cooperation if an additional employee would be hired:
If an employee decides to become (or stay) a supervisor, then the profit-
maximizing number of subordinates N* is given by the conditions
mci-1,N* < mb i-1,N* and mci-1,N*+1 > mb i-1,N*+1. The unique solution N* may be found
by finding the solution to mci-1,N+1 = mb i-1,N+1 (and rounding off the solution)2:
1 The time index t is removed from all equations, since it has no relevance for the calculations
presented here.
2 In addition, the first derivative of the marginal costs with respect to the number of subordi-
nates must exceed the first derivative of the marginal benefits.
ACcc
cNcNN
ACwfpqNqNmc
iN
i
wwiN
i
fpfp
N
j
ij
i
qqiN
i
qq
iNiN
N
j
ij
N
j
ijNi
+++
++++=
+++−+=
++
=
+
++
=
+
=
+−
å
åå
,12,1,,1
2,1,
0
,2,1,,12,1,
2
,1,1
0
,
2
1
0
,
2
1,1
)()/1()(
)()()12()()2(
)1(
γγγγ
γαγγγααα
αα
å
å
=
+
=
++−
+++=
−++++=
N
j
ij
i
qqiN
i
qq
N
j
ijiNNi
ccN
qNNqNmb
0
,2,1,,12,1,
0
,,11,1
)()()1(
))1(()1)1((
γβγγγββ
βββ
40 EIM Business & Policy Research
Team size in firms with homogeneous employees
In the special case where all employees have the same, average quality (cj,i = 1
∀ j,i), the equations for marginal costs and benefits become less complicated.
The marginal costs equation may be simplified to:
with ϕk>0 for k=1,2.
The values of these parameters depend on the hierarchical level i (for nota-
tional convenience, the hierarchical level index i has been left out). Both the
first-order and second-order derivative with respect to N are positive, so the
marginal costs are a strict convex function of the number of incumbent subor-
dinates.
The marginal benefits may be rewritten as:
with θk>0 for k=1,2.
The marginal benefits are now increasing linearly with the number of incum-
bent subordinates (again, the hierarchical level index i has been left out).
A necessary condition for a finite team size is that mci-1,N / mb i-1,N>1 for N → ∞.
This condition is always met (given that all model parameters are strictly posi-
tive):
Whether or not an employee becomes a supervisor, depends on other criteria.
A sufficient condition is that mci-1,1 / mb i-1,1<1: the benefits of hiring the first
subordinate exceed the costs. For the calibrated model, this condition is met
for the first three levels of the firm. This implicates that the baseline firm
should consist of at least four levels. As discussed in the main text, this is the
actual number of levels for the baseline simulation. For the small business, this
condition is met for the first level only, and for the large enterprise for the
first three levels.
21
2
1
2,1,
2,1,
2,1,2,1,
2
2,1,
2,1,
2,1,
2,1,2,1,
2
1,1
)()()()(3)(3
)()()()12()()2(
ϕϕϕ
γγγγγαγγαγγαγ
γγγγγγαγγααα
++=
+++++=
+++++++=
+−
NN
ACNN
ACNNNNmc
i
ww
i
fpfp
i
qq
i
qq
i
qq
i
ww
i
fpfp
i
qq
i
qqNi
21
2,1,2,1,
2,1,2,1,1,1
)()12()(2
)()()1(
θθ
γγβγβγ
γγβγγββ
+=
++=
+++=
+−
N
N
NNmb
i
qq
i
qq
i
qq
i
qqNi
1
limlimlim
1
1
1
1
1
1
2
1
,1
,1 >+=
+
=
∞→∞→
−
−
∞→
θ
ϕ
θ
ϕ
θ
ϕϕ
N
N
NN
NN
Ni
Ni
Nmb
mc
ACcc
cNcNN
mbmc
iN
i
wwiN
i
fpfp
N
j
ij
i
qqiN
i
qq
NiNi
++=
+−−+++−−−−⇔
=
++
=
+
+−+−
å
,12,1,,1
2,1,
0
,2,1,,12,1,
2
1,11,1
)()/1()(
)()2()()1)2((
γγγγ
γγβααγγβαβαα
EIM Business & Policy Research 41
Both the small and the large enterprise have one level more than the mini-
mum implied by the condition mci-1,1 / mb i-1,1<1. Apparently, even if this condi-
tion doesn’t hold, it may still be profitable to hire several employees. This is
because marginal costs are a convex function of N, and marginal benefits a
linear function (see figure 4 for an example). Necessary (but not sufficient)
conditions for this solution are:
1. mci-1,1 / mb i-1,1>1
2. {∂ mci-1,N /∂N}/ {∂ mb i-1,N/∂N}< 1 for N=0.
This second inequality is equivalent with ϕ1 / θ1 < 1⇔ α/β < 2/3. This second
condition is met in our calibrated model.
Figure 4 Marginal costs and marginal benefits
If the conditions that guarantee a finite number of subordinates are met, then
team size is implicitly defined by the following equation:
Marginal benefits per unit of production only depend on the benefits of co-
operation and the number of incumbent workers, and are independent of the
hierarchical level i. In contrast, marginal costs per unit of production differ be-
tween hierarchical levels. In our calibrated version of the model, the relative
share of the supervision costs increases with the level, while the relative
weight of the adjustment and wage costs decreases.
mb
mc
Number of subordinates (N)
12
1
33
//
2,1,2,
2,
1,
1,
2,
2,
1,
1,
2
,1,1,1,1
1,11,1
++=
÷
÷
ø
ö
ç
ç
è
æ
+
÷
÷
ø
ö
ç
ç
è
æ
+
÷
÷
ø
ö
ç
ç
è
æ
+++⇔
=⇔
=
+−+−
+−+−
ββ
γγγ
γ
γ
γ
γ
γ
γ
γ
ααα
N
AC
NN
qmbqmc
mbmc
i
qq
i
q
w
q
w
i
q
fp
q
fp
ijNiijNi
NiNi
42 EIM Business & Policy Research
EIM Business & Policy Research 43
List of Research Reports
The research report series is the successor of both the research paper and the
'researchpublicatie' series. There is a consecutive report numbering followed
by /x. For /x there are five options:
/E: a report of the business unit Strategic Research, written in English;
/N: like /E, but written in Dutch;
/F: like /E, but written in French;
/A: a report of one of the other business units of EIM/Small Business Re-
search and Consultancy;
/I: a report of the business unit Strategic Research for internal purposes;
external availability on request.
9301/E The intertemporal stability of the concentration-margins relationship
in Dutch and U.S. manufacturing; Yvonne Prince and Roy Thurik
9302/E Persistence of profits and competitiveness in Dutch manufacturing;
Aad Kleijweg
9303/E Small-store presence in Japan; Martin A. Carree, Jeroen C.A. Potjes
and A. Roy Thurik
9304/I Multi-factorial risk analysis and the sensitivity concept; Erik M. Ver-
meulen, Jaap Spronk and Nico van der Wijst
9305/E Do small firms' price-cost margins follow those of large firms? First
empirical results; Yvonne Prince and Roy Thurik
9306/A Export success of SMEs: an empirical study; Cinzia Mancini and
Yvonne Prince
9307/N Het aandeel van het midden- en kleinbedrijf in de Nederlandse in-
dustrie; Kees Bakker en Roy Thurik
9308/E Multi-factorial risk analysis applied to firm evaluation; Erik M. Ver-
meulen, Jaap Spronk and Nico van der Wijst
9309/E Visualizing interfirm comparison; Erik M. Vermeulen, Jaap Spronk
and Nico van der Wijst
9310/E Industry dynamics and small-firm development in the European print-
ing industry (Case Studies of Britain, the Netherlands and Denmark);
Michael Kitson, Yvonne Prince and Mette Mönsted
9401/E Employment during the business cycle: evidence from Dutch manufactur-
ing; Marcel H.C. Lever and Wilbert H.M. van der Hoeven
9402/N De Nederlandse industrie in internationaal perspectief: arbeidspro-
duktiviteit, lonen en concurrentiepositie; Aad Kleijweg en Sjaak
Vollebregt
9403/E A micro-econometric analysis of interrelated factor demand; René
Huigen, Aad Kleijweg, George van Leeuwen and Kees Zeelenberg
9404/E Between economies of scale and entrepreneurship; Roy Thurik
44 EIM Business & Policy Research
9405/F L'évolution structurelle du commerce de gros français; Luuk Klomp et
Eugène Rebers
9406/I Basisinkomen: een inventarisatie van argumenten; Bob van Dijk
9407/E Interfirm performance evaluation under uncertainty, a multi-
dimensional frame-work; Jaap Spronk and Erik M. Vermeulen
9408/N Indicatoren voor de dynamiek van de Nederlandse economie: een
sectorale analyse; Garmt Dijksterhuis, Hendrik-Jan Heeres en Aad
Kleijweg
9409/E Entry and exit in Dutch manufacturing industries; Aad Kleijweg and
Marcel Lever
9410/I Labour productivity in Europe: differences in firm-size, countries and
industries; Garmt Dijksterhuis
9411/N Verslag van de derde mondiale workshop Small Business Economics;
Tinbergen Instituut, Rotterdam, 26-27 augustus 1994; M.A. Carree en
M.H.C. Lever
9412/E Internal and external forces in sectoral wage formation: evidence
from the Netherlands; Johan J. Graafland and Marcel H.C. Lever
9413/A Selectie van leveranciers: een kwestie van produkt, profijt en partner-
schap?; F. Pleijster
9414/I Grafische weergave van tabellen; Garmt Dijksterhuis
9501/N Over de toepassing van de financieringstheorie in het midden- en klein-
bedrijf; Erik M. Vermeulen
9502/E Insider power, market power, firm size and wages: evidence from Dutch
manufacturing industries; Marcel H.C. Lever and Jolanda M. van
Werkhooven
9503/E Export performance of SMEs; Yvonne M. Prince
9504/E Strategic Niches and Profitability: A First Report; David B. Audretsch,
Yvonne M. Prince and A. Roy Thurik
9505/A Meer over winkelopenstellingstijden; H.J. Gianotten en H.J. Heeres
9506/I Interstratos; een onderzoek naar de mogelijkheden van de Interstratos-
dataset; Jan de Kok
9507/E Union coverage and sectoral wages: evidence from the Netherlands; Mar-
cel H.C. Lever and Wessel A. Marquering
9508/N Ontwikkeling van de grootteklassenverdeling in de Nederlandse Indus-
trie; Sjaak Vollebregt
9509/E Firm size and employment determination in Dutch manufacturing indus-
tries; Marcel H.C. Lever
9510/N Entrepreneurship: visies en benaderingen; Bob van Dijk en Roy Thurik
9511/A De toegevoegde waarde van de detailhandel; enkele verklarende theo-
rieën tegen de achtergrond van ontwikkelingen in distributiekolom,
technologie en externe omgeving; J.T. Nienhuis en H.J. Gianotten
EIM Business & Policy Research 45
9512/N Haalbaarheidsonderzoek MANAGEMENT-model; onderzoek naar de
mogelijkheden voor een simulatiemodel van het bedrijfsleven, gebaseerd
op gedetailleerde branche- en bedrijfsgegevens; Aad Kleijweg, Sander
Wennekers, Ton Kwaak en Nico van der Wijst
9513/A Chippen in binnen- en buitenland; De elektronische portemonnee in
kaart gebracht; een verkenning van toepassingen, mogelijkheden en
consequenties van de chipcard als elektronische portemonnee in binnen-
en buitenland; drs. J. Roorda en drs. W.J.P. Vogelesang
9601/N Omzetprognoses voor de detailhandel; Pieter Fris, Aad Kleijweg en Jan
de Kok
9602/N Flexibiliteit in de Nederlandse Industrie; N.J. Reincke
9603/E The Decision between Internal and External R&D; David B. Audretsch,
Albert J. Menkveld and A. Roy Thurik
9604/E Job creation by size class: measurement and empirical investigation;
Aad Kleijweg and Henry Nieuwenhuijsen
9605/N Het effect van een beursnotering; drs. K.R. Jonkheer
9606/N Een Micro-werkgelegenheidsmodel voor de Detailhandel; drs. P. Fris
9607/E Demand for and wages of high- and low-skilled labour in the Nether-
lands; M.H.C. Lever and A.S.R. van der Linden
9701/N Arbeidsomstandigheden en bedrijfsgrootte. Een verkenning met de
LISREL-methode; drs. L.H.M. Bosch en drs. J.M.P. de Kok
9702/E The impact of competition on prices and wages in Dutch manufactur-
ing industries; Marcel H.C. Lever
9703/A FAMOS, een financieringsmodel naar grootteklassen; drs. W.H.J. Ver-
hoeven
9704/N Banencreatie door MKB en GB; Pieter Fris, Henry Nieuwenhuijsen en
Sjaak Vollebregt
9705/N Naar een bedrijfstypenmodel van het Nederlandse bedrijfsleven; drs.
W.H.M. van der Hoeven, drs. J.M.P. de Kok en drs. A. Kwaak
9801/E The Knowledge Society, Entrepreneurship and Unemployment; David
B. Audretsch and A. Roy Thurik
9802/A Firm Failure and Industrial Dynamics in the Netherlands; David B.
Audretsch, Patrick Houweling and A. Roy Thurik
9803/E The determinants of employment in Europe, the USA and Japan;
André van Stel
9804/E PRISMA'98: Policy Research Instrument for Size-aspects in Macro-
economic Analysis; Ton Kwaak
9805/N Banencreatie bij het Klein-, Midden- en Grootbedrijf; Henry Nieu-
wenhuijsen, Ben van der Eijken en Ron van Dijk
9806/A Milieumodel; drs. K.L. Bangma
9807/A Barriers for hiring personnel; Jacques Niehof
9808/A Methodiek kosten en baten Arbowetgeving; drs. K.M.P. Brouwers, dr.
B.I. van der Burg, drs. A.F.M. Nijsen en ir. H.C. Visee
46 EIM Business & Policy Research
9809/E Business Ownership and Economic Growth; An Empirical Investiga-
tion; Martin Carree, André van Stel, Roy Thurik and Sander Wen-
nekers
9810/E The Degree of Collusion in Construction; M.H.C. Lever, H.R. Nieuwen-
huijsen and A.J. van Stel
9811/E Self-employment in 23 OECD countries; Ralph E. Wildeman, Geert
Hofstede, Niels G. Noorderhaven, A. Roy Thurik, Wim H.J. Verhoeven
and Alexander R.M. Wennekers
9812/E SICLASS: Forecasting the European enterprise sector by industry and
size class; Niels Bosma and Ton Kwaak
9901/E Scanning the Future of Entrepreneurship; drs. N.S. Bosma, drs. A.R.M.
Wennekers and drs. W.S. Zwinkels
9902/E Are Small Firms Really Sub-optimal? Compensating Factor Differen-
tials in Small Dutch Manufacturing Firms; David B. Audretsch, George
van Leeuwen, Bert Menkveld and Roy Thurik
9903/E FAMOS; A size-class based financial analysis model; W.H.J. Verhoeven
and E.A. van Noort
9904/E Conduct and Performance in Dutch Manufacturing; An Application of
Appelbaum 1982 with a Plausibility-Check; Frank A. Hindriks, Henry
R. Nieuwenhuijsen and Adriaan J. van Stel
9905/E Non-competitive Rents in Dutch Manufacturing; Conduct and Per-
formance in the New Empirical Industrial Organization; Frank A. Hin-
driks
9906/E A human-resource-based theory of the small firm; Charlotte Koch
and Jan de Kok
9907/N Van werknemer naar ondernemer; Een hybride of directe start?; ir.
H.C. Visee en drs. W.S. Zwinkels
9908/E Modelling returns to R&D: an application on size effects; Peter Brou-
wer and Henry Nieuwenhuijsen
9909/E Turbulence and productivity in the Netherlands; Niels Bosma and
Henry Nieuwenhuijsen
9910/E Start-up capital: Differences between male and female entrepre-
neurs. ‘Does gender matter?’; Ingrid Verheul and Roy Thurik
9911/E Modelling Business Ownership in the Netherlands; Niels Bosma,
Sander Wennekers, Gerrit de Wit and Wim Zwinkels
9912/A Measuring innovative intensity: Scale construction; J.P.J. de Jong
9913/A Determinants of firm size; Y. Bernardt and R. Muller
0001/E Strategies, uncertainty and performance of small business start-ups;
Marco van Gelderen, Michael Frese and Roy Thurik
0002/E Determinants of Successful Entrepreneurship; Niels Bosma, Mirjam
van Praag and Gerrit de Wit
0003/E Comparative Advantages in Estimating Mark-ups; Frank A. Hindriks,
Henry R. Nieuwenhuijsen and Gerrit de Wit
EIM Business & Policy Research 47
0004/A The ARKO labour-cost model: Characteristics and application; G.Th.
Elsendoorn and A.H. Nieuwland
0005/E The impact of contestability on prices in manufacturing industries;
Frank Bosman
0006/A Estimating missing data within an accounting and aggregation
framework; Ton Kwaak and Niels Bosma
0007/N Kennis-spillovers en economische groei; Henry Nieuwenhuijsen en
André van Stel
0008/A KTO2000; Een sectormodel naar grootteklasse voor de analyse en
prognose van Korte Termijn Ontwikkelingen; drs. K.L. Bangma en
drs. W.H.J. Verhoeven
0009/E Making sense of the new economy; Joris Meijaard
0010/A Determinants of innovative ability; An empirical test of a causal
model; J.P.J. de Jong, R. Kemp and C. Snel
