Doi - 10.1016 - J.jebo.2009.04.011

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 42

Accepted Manuscript

Title: Profit Sharing and Firm Size: The Role of Team


Production

Authors: John S. Heywood, Uwe Jirjahn

PII: S0167-2681(09)00120-6
DOI: doi:10.1016/j.jebo.2009.04.011
Reference: JEBO 2379

To appear in: Journal of Economic Behavior & Organization

Received date: 20-3-2008


Revised date: 20-4-2009
Accepted date: 22-4-2009

Please cite this article as: Heywood, J.S., Jirjahn, U., Profit Sharing and Firm Size: The
Role of Team Production, Journal of Economic Behavior and Organization (2008),
doi:10.1016/j.jebo.2009.04.011

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
* Title Page (with Full Author Details)

Profit Sharing and Firm Size: The Role of Team Production

t
ip
John S. Heywood*

Uwe Jirjahn**

cr
us
* Professor of Economics and Director of the Graduate Program in Human Resources

an
and Labor Relations, University of Wisconsin-Milwaukee, Milwaukee, Wisconsin USA

** Research Scientist, Faculty of Economics, Department of Labor Economics, Leibniz


University of Hanover, Hanover Germany
M
d

Abstract
te

This paper presents a model showing that profit sharing is subject to the 1/N problem in
the case of independent worker productivity but not in the case of interdependent worker
p

productivity. This implies the role of firm size on the likelihood of profit sharing will
differ by the nature of the underlying technology. We test this implication using German
ce

establishment data and using a proxy for interdependent worker productivity. The results
conform to the theory showing that firm size is associated with reduced profit sharing use
when technology is independent but not when technology is interdependent.
Ac

JEL: J33, J41

Keywords: Profit Sharing, 1/N Problem, Team Production

Corresponding author: Prof. J. Heywood, Department of Economics, University of


Wisconsin-Milwaukee, P.O. Box 413, Milwaukee, WI 53201 USA.

Page 1 of 41
Blinded Manuscript (NO Author Details)

Profit Sharing and Firm Size: The Role of Team Production

t
ip
cr
us
an
Abstract
M
This paper presents a representative model showing that profit sharing is subject to the
1/N problem in the case of independent worker productivity but not in the case of
interdependent worker productivity. This implies the role of firm size on the likelihood of
d

profit sharing will differ by the nature of the underlying technology. We test this
implication using German establishment data and using a proxy for interdependent
te

worker productivity. The results conform to the theory showing that firm size is
associated with reduced profit sharing use when technology is independent but not when
technology is interdependent.
p
ce

JEL: J33, J41

Keywords: Profit Sharing, 1/N Problem, Team Production


Ac

Page 2 of 41
I. Introduction

While profit sharing helps align the interests of workers with those of shareholders, its

effectiveness in motivating workers may be limited by the 1/N problem. All else equal,

the proportion of any profit increase that a worker receives, and so her incentive to

t
ip
provide effort, decreases with the number of workers participating in the profit sharing

cr
scheme, N. Indeed, Prendergast (1999) reviews evidence from medical and legal

partnerships showing that profit sharing becomes increasingly irrelevant in motivating

us
workers as the size of the partnership increases. More generally, it has been anticipated

that larger firms would avoid using profit sharing because of the 1/N problem. Yet, most
an
studies find either no significant association (e.g., Wagar and Long 1995 for Canada,

Drago and Heywood 1995 for Australia, FitzRoy and Kraft 1987 and Heywood and
M
Jirjahn 2002 for Germany, Cheadle 1989 and Kruse 1996 for the US, Pendelton 1997 for
d

Britain) or even a positive link (e.g., Gregg and Machin 1988 for Britain, Jones and
te

Pliskin 1997 and Adams 2002 for Canada, FitzRoy and Kraft 1995 for Germany,

Amisano and Del Boca 2004 for Italy).1


p

While fixed costs in the adoption of profit sharing might explain these results,
ce

Adams (2002, 2006) suggests an alternative explanation. He argues that manufacturing

firms, unlike legal or medical partnerships, have a higher degree of interdependent


Ac

worker productivity and that the extent of this interdependency may grow with firm size.

Interdependent worker productivity, often called team production (Alchian and Demsetz

1972), implies that shirking by an individual worker decreases not only his own

1
One study using UK data (Estrin and Wilson 1989) and a second using Spanish data (Bayo-

Moriones and Herta-Arribas 2002) estimated significant negative associations.

Page 3 of 41
productivity but also the productivity of other workers. This increases the cost of shirking

as it implies a more drastic decline in total production and so in individual profit sharing

income. Thus, it is possible that an increase in firm size has two opposing incentive

effects. On the one hand, the 1/N problem gets more severe reducing each workers

t
incentive to exert effort. This would be anticipated because larger firms are likely to

ip
have profit sharing schemes covering more workers. On the other hand, larger firms may

cr
have greater productivity interdependencies increasing the cost of shirking. These

us
offsetting influences could make the role of firm size ambiguous and help explain the

usual failure to confirm that profit sharing is less common in larger firms.
an
If this explanation is correct, statistical identification of the 1/N problem requires

examining a sample of firms with low degrees of team production. Among these firms, as
M
size grows, profit sharing becomes increasingly ineffective in motivating workers. Hence,

larger firms should tend to avoid the use of profit sharing. Yet, for firms characterized by
d

a high degree of team production, the 1/N problem will be offset or even dominated by an
te

increased cost of shirking. In that case, there should be no or even a positive association
p

between firm size and the use of profit sharing.


ce

Using German establishment data, we are the first to test these implications. We

divide the establishments into those we think more likely to be characterized by team
Ac

production, interdependent worker productivity, and those we think less likely to be so

characterized. Within each subsample we estimate the relationship between employment

size and the probability of using profit sharing. The results are persuasive and broadly

supportive of the alternative explanation provided by Adams. We find that among those

establishments without team production, the prediction of the traditional model receives

Page 4 of 41
support. Larger firms are less likely to use profit sharing consistent with the expectation

that the large N makes the motivational effects of profit sharing irrelevant. On the other

hand, among those establishments with team production, we find no significant

association between establishment size and the use of profit sharing.

t
In the next section we provide a theoretical illustration that contrasts technologies

ip
with and without team production. We demonstrate that the role of profit sharing differs

cr
dramatically between the two technologies. While technology without team production

us
encourages workers to shirk under profit sharing, technology with team production

encourages effort provision under profit sharing. We develop the testable hypotheses

regarding firm size. an


In the third section, we defend our focus on teams as a proxy for

team production, discuss other determinants of profit sharing and the available control
M
variables. While "work teams," a notion from human resource management, is not the

same as team production from the economics literature, we argue there exists substantial
d

overlap. Section four presents our data and provides descriptive statistics. The fifth
te

section presents the key results and undertakes a series of robustness checks. The final
p

section concludes.
ce

II. Profit Sharing and Team Production


Ac

Making workers' income depend on profit would seem to provide an incentive advantage.

Yet, this incentive advantage might be illusory because of the well-known 1/N problem.

A worker who increases his productivity in response to a group reward receives only the

share 1/N (where N is the number of workers in profit sharing plan) of that increase back

as a reward with the remainder divided among the other members of the group. Using a

Page 5 of 41
simple analytical example we illustrate that the severity of the problem depends on the

production technology.

In what follows, we assume that the size of the firm is also the size of the profit

sharing plan. Thus, as N increases, the firm is larger and so more workers are covered by

t
the profit sharing plan. We recognize that this may not always be the case. First, even a

ip
large firm may confine a profit sharing scheme to only a minority of employees. Second,

cr
the relevant profit for a sharing scheme may not be the total for a firm but that generated

us
by a particular establishment or even a smaller subunit of production. While we will

account for the imperfect link between firm size and profit sharing scheme size in our
an
empirical work, we ignore the distinction in our theoretical consideration.
M
No Team Production

We consider a group of N homogenous workers. The group receives a share  of total


d

output Q . In the absence of teamwork (NT), each worker’s contribution to output is


te

independent of that of other workers:


p

N
Q NT   ei (1)
ce

i 1

where ei denotes both worker i’s effort and worker i’s individual output with ei  {0, 1} .
Ac

His utility depends on money income, yi , and effort: U i ( yi , ei )  yi  Ci ( ei ) where

Ci ( ei ) denotes the disutility of effort function with C ( ei )  cei and 0  c   . Let

consider an equal sharing scheme:


yi  Q (2)
N

Page 6 of 41
Workers’ total welfare is:

N
W  U j (3)
j 1

As c   , the social optimum requires that each worker exerts effort, ei  1 , yielding the

t
maximum total welfare, W  (  c) N  0 . This yields total output Q*  N and utility

ip
U *    c for each worker.

cr
Yet, an individually rational worker considers only his own utility. To analyze the

us
individually rational solution, let us define:

C c  /N (4)
an
Exerting effort under production technology (1) and pay scheme (2) results in a change of

worker i’s utility, U i   / N  c . If c  c and, hence, Ci (1)  C , exerting effort is the


M
dominant strategy as it continues to increase utility, U i  0 . Individual rationality and
d

social optimality coincide. Yet, if c  c and, hence, Ci (1)  C , exerting effort decreases
te

utility, U i  0 . Thus, ei  0 becomes the dominant strategy. The free-rider fully avoids
p

the disutilty of effort, while the income loss is small since the decrease in individual
ce

output is shared with coworkers. The resulting Nash equilibrium is a typical prisoner’s

dilemma. No one exerts effort, resulting in zero total output and zero utility for each
Ac

worker. Clearly, workers could be better off if everyone exerts effort. Yet, this is not

individually rational.

The threshold defined in equation (4) represents the maximal personal cost an

individually rational worker will bear to exert effort. This threshold decreases in the size

Page 7 of 41
of the group: C / N  c / N   / N 2 .2 As the size increases, the free rider problem

gets more severe as the willingness to bear a given effort decreases. Thus, as size varies

across firms, firms with a larger N face a decreased probability that their workers will

exert effort. Consequently, larger firms (bigger N) will be less likely to motivate workers

t
ip
by adopting profit sharing.

cr
Team Production

us
The preceding illustration confirms standard analyses showing that effort decreases in N

(e.g., Kandel and Lazear 1992). However, the incentive effects of profit sharing may
an
depend on the degree of interdependent worker productivity. Here we provide a simple

analysis illustrating that regardless of size socially optimal effort can be achieved under
M
profit sharing scheme (2) if technology is characterized by team work (T). Assume that

the production technology is given by Kremer’s (1993) ‘o-ring’ production function


d

N
Q T  N  ei
te

(5)
i 1
p

If all workers exert effort, group output is QT *  Q*  N . If only one single worker free
ce

rides, the team loses the entire output. Like production function (1), technology (5)

implies that exerting effort is socially desirable as long as c   . Unlike production


Ac

function (1), technology (5) involves an alternative to the free riding equilibrium. All

workers exerting effort is also a Nash equilibrium even when remuneration is

characterized by sharing scheme (2). Given that all of the colleagues exert effort, i’s own

2
While we focus on a dichotomous effort choice, the threshold c is linked continuously to N.

This can be seen as an equivalent to a continuous effort choice.

Page 8 of 41
effort, ei  1 , ensures that Q* is reached and is the best response of worker i. This implies

income yi   and utility U *    c . Most importantly, this does not depend on the size

of the team. Hence, the use of sharing schemes will not depend on firm size if production

technology is characterized by interdependent worker productivity.

t
ip
Team production has the same consequence as the target based scheme suggested

by Holmstrom (1982). Holmstrom shows that a socially optimal effort choice may be

cr
achieved if the profit sharing is discontinuously linked to performance. Workers

us
participate in output only when a certain goal is reached generating a more severe

punishment for shirking. Our analysis shows that team production plays a similar role
an
without the target-based scheme. Team production severely punishes free riding since a

single worker’s shirking entails a drastic decline of group output. This has practical
M
implications. While Holmstrom argues that an outside agent is required to enforce the

target-based scheme, team production can play a similar role.3 Profit sharing will be more
d

effective under team production since it helps overcome the 1/N problem. The empirical
te

implication is that while profit sharing should clearly become less common among larger
p

firms without team production, there exists no clear relationship between profit sharing
ce

and firm size among firms with team production.

Our model of profit sharing contrasts a negative relationship with size among
Ac

firms without team production (1) and the absence of a relationship among firms with

team production (5). These production functions should be recognized as two extremes

with many actual cases falling in between. The critical point remains that as the degree of

3
A hierarchical and centralized personnel management and the main banks in Japan or Germany

may take this role as an a external agent (Aoki 1993).

Page 9 of 41
interdependence increases the output loss associated with free riding grows reducing its

extent. Indeed, Adams (2006, 2002) makes this point by manipulating a constant

elasticity of substitution production function across individual worker efforts. His work

differs in that he shows that the extent of interdependent worker productivity can increase

t
to such a degree that the traditional prediction between firm size and profit sharing is

ip
reversed. In addition to this difference in hypotheses, the welfare implications also

cr
differ.4 In his model (Adams 2006), the individually rational incentives induced by profit

us
sharing fall below the socially desirable incentives regardless of the degree of

interdependent worker productivity. Moreover, the gap between these incentives


an
increases with firm size. Profit sharing as a response to team production can only be a

second best solution. In our theoretical illustration, profit sharing achieves a first best
M
solution under team production that is independent of firm size.
d

III. Determinants of Profit Sharing Use


te

The fundamental empirical exercise estimates the establishment level determinants of the
p

use of profit sharing. While this is well-traveled terrain (see Kruse 1993 for an early
ce

review of a dozen studies and Perotin and Robinson 2003 for a recent survey), our

approach is unique in its focus on the role of team production in determining the role of
Ac

establishment size. As mentioned in the introduction, existing studies have reached no

consensus with size playing a role as a positive determinant in some studies, a negative

determinant in other studies and often having no role at all. We hypothesize that this

results from the failure to properly control for team production.

4
Adams (2002) provides no welfare analysis.

Page 10 of 41
Teams and Team Production

Following earlier work largely unrelated to profit sharing (Heywood and Jirjahn

2004, Heywood et al. 2008), we use workplace teams as a proxy for underlying team

t
production. While management increasingly views workplace teams as a tool to increase

ip
productivity (Spreitzer et al. 1999, Devaro 2006, 2008), they succeed only in certain

cr
circumstances. In equilibrium, their use reflects underlying production technology.

us
Theoretical work illustrates the connection between teams and interdependent worker

productivity. Teams facilitate communication and information sharing among team


an
members. Aoki (1990) and Carter (1995) stress that this communication takes on value

when there are gains associated with coordinating workers' actions and allocating their
M
tasks. Such coordination and allocation is the essence of team production suggesting that

teams will be far more likely in circumstances characterized by interdependent worker


d

productivity.
te

This theoretical work has been complemented by empirical evidence. While exact
p

measures of team production are rare, indicators include assembly line manufacturing
ce

(Adams 2006), the need to cross-train workers, particularly high costs associated with

absence and the hiring of workers who have greater job attachment. Each of these related
Ac

indicators is also associated with the use of teams. Heywood et al. (2008) and Heywood

and Jirjahn (2004) show that firms using teams face higher costs of absence. Brown et al.

(2007) estimate the determinants of workplace teams showing that firms with higher job

attachment workers and in assembly line manufacturing are more likely to use teams. In

short, both theory and evidence indicate a close, albeit imperfect, relationship between

10

Page 11 of 41
team production and the use of teams.5 Clearly, this does not mean that interdependent

worker productivity is absent in firms without teams only that interdependencies are

likely to be far smaller in firms without teams.

Our empirical strategy uses the presence of workplace teams as a primary indicator

t
of whether or not an establishment has team production.6 The critical estimations

ip
examine the role of establishment size (number of employees in the establishment) on the

cr
probability of the establishment using profit sharing. As the theory makes clear, we

us
anticipate fundamentally different relationships among the firms characterized as having

team production and among the firms characterized as not having team production.
an
We note that to the extent our proxy mismeasures the presence of team production,

we face an errors-in-variables problem making it more difficult to isolate the differences


M
we predict. Thus, the establishments without teams may include some with true

production interdependencies, e.g. basic assembly lines. As a consequence, we


d

underestimate the negative link between size and profit sharing among these
te

establishments. Similarly, the establishment with teams may include some that do not
p

5
Teams may also be taken, in part, as an indicator of greater intensity of mutual monitoring but
ce

as the number of teams within a workplace increases with its size, the prediction of a negative

relationship between size and profit sharing would remain among the establishments that have
Ac

teams (Adams 2002). In essence, each team becomes subject to a greater 1/N problem as size

increases.
6
As noted in our theoretical discussion, team production can be thought of as a continuum

reflecting the degree of interdependency between worker productivities. Thus, our identification

of team production should be thought of as identification of those establishments with greater

degrees of interdependence.

11

Page 12 of 41
have true production interdependencies (teams created as a managerial fad). As a

consequence, we might estimate a false negative link between size and profit sharing.

The important point remains that such mismeasurement biases us away from finding the

theoretically predicted results. Despite this likely bias, we find the presence of teams

t
plays a robust role in influencing the relationship between size and profit sharing.

ip
cr
Other Determinants of Profit Sharing

us
In addition to our focus on the roles of teamwork and establishment size, we

include a wide range of other determinants that might be anticipated to influence the use
an
of profit sharing. We characterize these into four rough, if interrelated, categories: the

nature of technology, job security and job tenure, managerial strategy, and industrial
M
relations. We consider each in turn and discuss the available variables in our data source,

the Hanover Panel.


d

A series of indicators capture the variety in technology. When workers perform


te

multi-dimensional tasks, individual performance measures are often unavailable for all
p

tasks. An emphasis on individual performance as measured by one or a few indicators


ce

causes workers to cut back on productive behaviors for which they are not rewarded

(Holmstrom and Milgrom 1991, Baker 1992).7 In contrast to individual performance pay,
Ac

profit sharing provides incentives to exert effort in all activities relevant to the firm’s

profit (Jirjahn 2000, Baker 2002). Therefore, profit sharing will be more likely when

technology requires multi-tasking and, hence, complex skills. To capture multi-skilling

7
Thus, a piece rate encourages workers to increase quantity but may not reward increasing

quality, maintaining equipment or reducing chances of workplace injury.

12

Page 13 of 41
we control for the establishment’s per capita expenses on training and the share of

employees who have a university degree.8 Further, a variable for usual weekly hours is

taken into account to proxy the complexity of tasks. Complex tasks require extensive

training and hence entail substantial quasi-fixed labor costs inducing increased labor

t
utilization through longer average hours.9 In addition, we include the share of blue-collar

ip
workers and broad industry dummies to control for technology differences across

cr
industries.10

us
The second group of controls is related to job security and job tenure. An

important managerial decision is determining the extent of deferred compensation


an
(Lazear 1979, 1981). Unlike the United States, profit sharing in Germany is all most

entirely contemporaneous rather than deferred. Thus, it should be viewed as more likely
M
in those circumstances in which deferred compensation will be unsuccessful in eliciting

effort and in which call for contemporaneous variable pay. Thus, women with their
d

shorter expected tenure will be better motivated by contemporaneous variable pay


te

(Goldin 1986) and earlier German evidence confirms that female dominated workplaces
p

are more likely to use profit sharing (Heywood and Jirjahn 2002).11 Part time workers
ce

8
Independent of multi-skilling, we note the importance of controlling for training as Azfar and

Danninger (2001) confirm a close empirical link in the US between the incidence of profit
Ac

sharing and the extent of training.


9
Using employee data, Hart and Hubler (1990) confirm that longer average hours as a proxy for

task complexity are positively associated with profit sharing.


10
The dummies follow from dividing manufacturing in to four broad industrial groupings.
11
Similarly, Australian establishment evidence confirms that female dominated workplaces are

more likely to have profit sharing (Drago and Heywood 1995).

13

Page 14 of 41
and temporary workers are also likely to have lower expected tenure and should be

associated with contemporaneous payment schemes. While we don't have an indicator of

the age of workers in the establishment, we include a measure indicating whether the

manager views the average worker as "too old".12 In conclusion, we note a broad tension

t
between theory and evidence that shorter term workers are more likely to be paid

ip
contemporaneous profit sharing and the notion that a stable longer term workforce is

cr
needed for profit sharing to successfully change group norms (Kandel and Lazear 1992,

us
Che and Yoo 2001).

The third set of controls captures a variety of aspects of the managerial strategy.
an
Both teams and establishment size are likely to be associated with specific managerial

practices that may also influence worker effort and the establishment’s use of profit
M
sharing.13 If profit sharing is part of a complicated bundle of managerial practices

designed to change behavior and increase productivity in the face of particular


d

technologies (Robinson and Wilson 2006), controlling for those practices helps to avoid
te

omitted variable bias.


p

Profit sharing will induce effort only if workers trust that managers pursue firm
ce

strategies and investments designed to increase financial performance (Ross 1973). We

use a proxy for the presence of managerial profit sharing to capture these circumstances.
Ac

We also include a related variable measuring whether or not an owner is active in

12
There is no prohibition on age discrimination in Germany at the time of the survey.
13
For example, Pendleton (1997) argues that larger establishments tend to develop formal

controls and performance measures. The quality of monitoring in larger establishments might

help reducing the free rider problem associated with profit sharing.

14

Page 15 of 41
managing the establishment. Further, intense competition may encourage profit sharing

as a way to improve productivity or it may force firms away from deferred compensation

as the chance increases that they will have to renege on such commitments (Bertrand

2004). While product market competition can take many forms, competition over market

t
share usually plays an important role. Thus, we include whether or not increasing market

ip
share stands at the heart of managerial strategy.

cr
We also include several indicators of performance management to recognize that

us
practices other than team working may be used to mitigate the free-rider problem.

Specifically, improved monitoring and control can reduce free riding. We recognize that
an
the firm may break the establishment into smaller profit centers that bring workers closer

to their value added and in so doing reduce the free-rider problem. Freeman et al. (2008)
M
have shown that the extent of mutual monitoring that workers actually undertake is

reduced in larger workplaces. Breaking large establishments into smaller profit centers
d

might be expected to increase the extent of mutual monitoring making profit sharing
te

more likely.
p

Moreover, performance management may include direct monitoring of individual


ce

output. Pendleton (2006) argues that the use of individual performance pay should not be

thought of as a substitute for profit sharing but rather as a complement to profit sharing
Ac

that helps reduce free-riding. We include both the use of piece rates and premium pay as

controls. While piece rates reward the quantity of output, premium pay rewards other

dimensions of output such as reduced wastage or increased quality of workmanship.

Profit sharing can also be part of an efficiency wage strategy (Pendleton 1997).

Free riding may be reduced if the establishment pays high wages and threatens to dismiss

15

Page 16 of 41
workers who are caught shirking.14 We capture this strategy with two managerial

attitudes: whether or not managers believe that high wages and/or the threat of dismissal

are likely to motivate workers. While indirect, these attitudinal variables suggest the

relative importance that management places on wages and dismissals as incentives.

t
Further, we control for managers’ attitudes toward employee involvement in

ip
decision making. This variable will be positively associated with the use of profit sharing

cr
if managers try to mitigate the free-rider problem by providing opportunities for mutual

us
monitoring and peer pressure. Yet, it will be negatively associated with profit sharing if

managers aim at reducing free riding by hierarchical monitoring and limiting employees’
an
discretion. Indeed, Bryson and Freeman (2007) found that managerial monitoring is

positively associated with financial participation of the workforce.


M
The fourth group of controls is designed to capture elements of employment

relations within the establishments. German industrial relations are characterized by a


d

dual structure of employee representation (Hubler and Jirjahn 2003). While collective
te

bargaining agreements are usually negotiated at the sectoral level, works councils provide
p

a highly developed mechanism for establishment-level participation. Workers of each


ce

establishment may elect council members but the creation of the council depends on the

initiative of the establishment’s employees. Hence, councils are not present in all eligible
Ac

establishments. The rights of the works councils are given by the Works Constitution Act

(WCA). The WCA provides that the works council must agree to the implementation of

14
Incentive schemes may be either input- or output-related. While performance pay is typically

based on monitoring the workers’ outputs, efficiency wages are typically modeled as input-

related incentive devices based on (incompletely) monitoring workers’ efforts (Jirjahn 2006).

16

Page 17 of 41
any piece rate making them the result of consultation and negotiation between

management and workers. Works councils do not enjoy the same codetermination rights

with regard to profit sharing plans under the WCA and German trade unions have

historically seen profit sharing as undercutting their function. Thus Germany makes

t
relatively modest use of profit sharing but extensive use of piece rates and premium pay

ip
(Jirjahn 2002).

cr
Nonetheless the use of profit sharing is often thought to depend on cooperative

us
and trustful employer-employee relations.15 Not only must workers believe that managers

will pursue increased financial performance but they must trust the accounting of
an
profits.16 The works council plays an explicit role in promoting such relations and with its

rights to financial information it can help enforce profit sharing schemes making workers
M
more likely to accept them in the first place. Freeman and Lazear (1995) argue that works

councils are more likely to build trust and cooperation when distributional conflicts are
d

settled between unions and employers’ associations outside the firm. Thus, following
te

Heywood et al. (1998) we anticipate that works councils are positively associated with
p

profit sharing in establishments covered by collective bargaining agreements but not in


ce

uncovered establishments.
Ac

15
In this sense, the successful use of profit sharing may require that two different types of

prisoners’ dilemmas be tackled. The 1/N problem refers to a prisoners’ dilemma among

employees. Uncooperative employer-employee relations refer to a prisoners’ dilemma between

the employer and the workforce of the establishment (Cable 1984).


16
Evidence from the US movie industry suggests such trust can be unjustified (Cheatham et al.

1996).

17

Page 18 of 41
IV. The Hanover Panel, the Data and Methodology

The Hanover Panel provides a set of variables highly appropriate for testing the

determinants of profit sharing use. Funded by the Volkswagen Foundation, the first wave

t
was conducted in 1994 with a core set of questions asked each wave but with each wave

ip
also having unique questions. Individual interviews were held with top managers to elicit

cr
answers to around 90 questions on issues of employment, personnel and industrial

us
relations. We focus on the third wave from 1996 that contains accurate information on

the presence of teams. The survey population is all manufacturing establishments in the
an
German state of Lower Saxony with at least five employees. A total of 721

establishments participated in the third wave and Brand et al. (1996) describe the
M
statistical design of the survey and provide a complete copy of the questionnaire. Our

analysis uses all establishments from the third wave that provided data for each question
d

we use.
te

The critical dependent variable indicates whether or not the establishments


p

provide profit sharing for their non-managerial employees. Moreover, we have


ce

information on the share of workers covered by the scheme that we will use for

robustness checks. We emphasize that the unit of observation for both the incidence of
Ac

profit sharing and workplace size is the establishment rather than the firm. Table 1

provides the definitions and descriptive statistics for all variables used in our estimations.

A total of 14.2 percent of all establishments use profit sharing. This figure is smaller than

the 39.1 percent of establishments that organize their workforces into production teams.

The average size of the establishments is 148 workers.

18

Page 19 of 41
Table 2 splits the establishments by the use of profit sharing showing that those

establishments using profit sharing are substantially more likely to be organized into

teams. Moreover, they are slightly larger on average. Thus, simple means do not reveal

evidence of free riding as the establishments using profit sharing do not emerge smaller

t
as predicted. We now turn to multivariate estimation.

ip
cr
V. Multivariate Estimates

us
To account for the range of potential determinants described in the third section, we

estimate multivariate probit specifications through maximum likelihood. Each entry


an
includes the coefficient, the t-statistic based on the asymptotic standard error and the

marginal effect of a unit change in the independent variable upon the probability of the
M
establishment providing profit sharing. When using interaction variables, we will present

the marginal effect calculated by the Ai and Norton (2003) and Norton et al. (2004)
d

routine to account for cross-partial effects that emerge from the nonlinear estimator but
te

are not incorporated in the traditional STATA estimates.17 The first column presents a
p

17
However, note that interpreting marginal effects of interaction variables in nonlinear models
ce

remains highly controversial. While marginal effects provide a sense of practical importance,

calculating the cross derivative of the expected value of the dependent variable in nonlinear
Ac

models has been criticized as potentially resulting in artificial and atheoretical predictions (Frant

1991). The functional form of a nonlinear model implies that all explanatory variables have

nonlinear effects on the probability of interest. Hence, the method suggested by Ai and Norton

(2003) can produce interaction effects by (distributional) assumption rather than by conclusion.

For example, the cross derivative of the expected value may be nonzero even if the coefficient of

the interaction variable is zero. To avoid such spurious interpretations, Nagler (1991) suggests

19

Page 20 of 41
base estimation across all observations. It confirms that usual hours, training, the share

of university graduates, expansion plans, executive profit sharing and the share of part-

time workers are each associated with a higher probability of profit sharing. Premium pay

is also a positive determinant suggesting that multiple incentive devices may be bundled.

t
The interaction between collective bargaining and works councils takes a positive

ip
coefficient, significant at ten percent, hinting that variable pay schemes may be more

cr
easily arranged in Germany when works councils exist and distributional conflicts are

us
handled outside the workplace by collective bargaining. The presence of teams also takes

a positive coefficient indicating that those workplaces using teams are 5.3 percent more
an
likely to make use of profit sharing than workplaces without teams. There is no apparent

role for size. It has a small and statistically insignificant coefficient. Thus, matching
M
earlier results from Germany (and elsewhere) there remains no evidence of the 1/N

problem.
d

In our first attempt to more fully isolate the role of team production, we add an
te

interaction of our proxy, teams, with establishment size. The results are presented in
p

column 2. As might be anticipated, the bulk of the results from the base estimation
ce

remain. Importantly, the coefficient on the size variable emerges as six times larger than

in the base estimate and achieves statistical significance at the five percent level. While
Ac

interpreting interaction effects with respect to the cross derivative of the underlying latent

variable and, hence, to focus solely on the coefficient of the interaction variable. Our main focus

will be on the sign and the statistical significance of the estimated coefficients. However, to get

an impression of the magnitudes, we present marginal effects in Table 3. In the other tables we

will focus solely on the coefficients.

20

Page 21 of 41
this result provides evidence that the 1/N problem may be important, it holds only for the

establishments without teams. The interaction of teams with establishment size takes a

nearly identical coefficient of the opposite sign.18 Indeed, the sum of the coefficient on

size and that on the interaction of team with size is not significantly different from zero.

t
Thus, there is no evidence that size influences the use of profit sharing among

ip
establishments characterized by team production. This fits our theoretical development

cr
that team production helps eliminate the 1/N problem of free riding. Moreover, the

us
general tendency in the literature to not isolate the negative association with size

anticipated from the 1/N problem may flow from the failure to distinguish the role of

team production. an
Columns 3 and 4 of Table 3 divide the sample and estimate the determinants of
M
profit sharing separately for establishments with and without teams. While the sample

sizes of the separate estimates are smaller, implying reduced precision in the point
d

estimates, the separate estimates allow the full range of determinants to vary in their
te

influence. This proves important for the industrial relations variables and others. Among
p

the set of establishments without teams, works councils, collective bargaining and their
ce

interaction play no role as determinants. On the other hand, among the establishments

with teams, works councils appear to be associated with reduced use of profit sharing
Ac

while the interaction between works councils and collective bargaining coverage emerges

18
The Norton et al. (2004) marginal effect of .26 is nearly identical to the uncorrected Stata

marginal effect of .27. Yet, we note that while the traditional Stata marginal effects are calculated

at the mean characteristics, the Norton et al. (2004) measure is the average marginal effect across

the individual observations.

21

Page 22 of 41
with a large and statistically significant positive coefficient. Premium pay appears to be a

significant positive determinant of profit sharing only in the sector with teams. In

contrast, high wages, threat of dismissal and limiting employee involvement are positive

determinants only in the sector without teams. Yet, despite these and other variations

t
between the two sets of establishments, the role of size identified in the simple interaction

ip
specification remains largely similar. Among the establishments without teams, larger

cr
size is associated with a significantly reduced probability of profit sharing. An increase of

us
one hundred workers is associated with a reduction of almost three percentage points in

the probability of having profit sharing. This represents a change that is more than 20

percent of the average probability of about .14. an Size plays no role among the

establishments with teams.19


M
We recognize that the dummy variable of whether or not establishments use profit

sharing scheme could be misleading if establishments cover only a small share of their
d

workers with the scheme. To examine this we use a secondary dependent variable that
te

measures the share of workers covered by the profit sharing scheme. While there is a
p

handful of missing observations, 551 establishments do not using profit sharing, 83 do


ce

use profit sharing and of those 35 establishments report that all of their workers are
19
As the dependent variable indicates the use of profit sharing for non-managerial workers, it
Ac

remains possible that the share of production workers might differ from the share of blue-collar

workers and that it might be correlated with establishment size. If so, the apparent correlation

with firm size might simply reflect differences in the share of production workers. Yet, Berman et

al. (1998) show that production workers in manufacturing are typically in blue-collar occupations

suggesting that the share of blue-collar workers should be highly correlated with the share of

production workers.

22

Page 23 of 41
covered. The mean coverage rate among those using profit sharing is .538 or fifty four

percent. To use this coverage variable we estimate a double truncated tobit in order to

account for both the zero and one observations. At issue is whether allowing for the

variation in coverage rates (rather than simply an indicator of zero or one), indicates a

t
different role for firm size.

ip
The first two columns of Table 4 mimic the last two columns of Table 3

cr
estimating the tobit separately on the sample without and with teams.20 The results also

us
mimic the earlier table. Firm size has a significant negative influence on profit sharing

coverage rates but only among the sample without teams. Among the sample with teams
an
there is no indication of size playing a role. Similar results also emerge from a full

sample interaction framework in which the interaction with teams takes a significant
M
positive coefficient and offsets the significant negative influence of size.

A related concern is that the profit sharing indicators do not reveal the possibility
d

of multiple smaller sized schemes within a single establishment. Thus, a profit or value
te

added measure could be used for a particular assembly line, part of an operation or group
p

of workers. Even if all workers in the establishment are covered by profit sharing, the
ce

actual schemes could be decentralized and involve a small numbers of workers

potentially overcoming the 1/N problem. While we have no direct evidence on this, we
Ac

do know whether or not the establishment has decentralized by identifying separate profit

centers within the operation. The establishment of such centers would seem a necessary

step toward creating a profit sharing scheme that applies differently to different subsets of

20
Interval estimation has been used in STATA to generate robust standard errors for the tobit

model.

23

Page 24 of 41
workers within the establishment. We now split our sample four ways to examine the

influence of firm size among those establishments with and without teams with the

subsamples of those with and without designated profit centers. These estimations are

shown in columns 3 to 6 of Table 4 and while the sample sizes become small, the results

t
continue to tell a consistent story. Regardless of whether or not the establishments

ip
identify profit centers, the negative influence of firm size is confined to the sample

cr
without teams but there is no evidence of influence in the sample with teams. Thus, to the

us
extent we can tell, measurement errors associated with using firm size as an indicator for

profit sharing size do not seem to be driving our results.


an
We now differentiate between those establishments that use teams covering more

than half of the workers and those that use teams covering half or less of the workers. We
M
use this information to create two team variables indicating higher and lower coverage.

We reproduce the critical estimations using both of these dummies. The first column in
d

Table 5 shows that higher coverage seems to be associated with greater use of profit
te

sharing but returns the familiar result that size is not a statistically significant determinant
p

of profit sharing use. We then interact each of the team coverage dummies with the size
ce

variable and include them in the same regression over the full sample. The results

indicate a six-fold increase in the coefficient on the basic size variable meaning that
Ac

among those establishments without teams, larger employment is associated with reduced

use of profit sharing. Interestingly, both interactions emerge with nearly identical

coefficients that each offset the coefficient on original size variable. Thus, among both

those with establishments with low team coverage and those with high team coverage,

size is not a determinant of profit sharing use.

24

Page 25 of 41
This may suggest that the critical issue is the use of teams itself not how widely

they are spread within the organization. As a check, we estimate three separate

specifications according to whether the establishment has no teams (as before), low team

coverage or high team coverage. As anticipated, among establishments without teams

t
size is a significant negative determinant while among those establishments with either

ip
low coverage or high coverage, size plays no significant role as a determinant.

cr
Nonetheless, it is interesting to note the large size of the negative coefficient among those

us
establishments with high coverage, admittedly only 98 observations.

An additional exploration led us to refine our definition of teams in a second


an
dimension. We assume that meaningful teams, those that accurately reflect team

production, involve workers doing separate tasks that must each be completed for
M
production to move forward. In this context, an absent worker imposes great costs on the

firm if other workers do not know how to complete the absent worker's task causing
d

production to slow or even stop (Heywood and Jirjahn 2004). As a consequence,


te

establishments characterized by strong interdependent worker productivity have an


p

increased incentive to provide cross-training enabling other workers to perform the tasks
ce

of absent workers. Thus, we take as an indication of meaningful team production the

combination of teams in the establishment and employer provided training among


Ac

workers. Gerlach and Jirjahn (2001) have shown a strong, albeit not perfect, link between

teams and employer provided further training in Germany. This has been seen as

(indirect) supporting evidence that teams increase the need for cross training. Hence

25

Page 26 of 41
employer provided training in firms with teams is likely to include cross training.21

Certainly, in the complete absence of separate and inter-related tasks, cross-training

would not be sensible investment by the establishment and the probability of observing

employer provided training would be lower.

t
The Hanover Panel has for each establishment an indicator of whether or not

ip
workers receive employer provided training. We develop an interaction indicator of

cr
meaningful teams that requires both that the establishment have teams and that it engages

us
in training. Nearly one-third of all establishments that report teams, do not engage in

training. We enter two variables, teams without training and teams with training into the
an
earlier regression in place of the single team indicator. The first column of Table 6 shows

the resulting estimation. As before, size is not a significant determinant of profit sharing
M
use in the initial estimation even as those establishments with teams with training are

significantly more likely to use profit sharing. Those establishments with teams without
d

training are no more or less likely to use profit sharing than establishments without
te

teams.
p

The second column of Table 6 interacts each of the team variables with size.
ce

First, and confirming all previous estimates, this causes the basic size variable to emerge

as a significant negative determinant. Thus, among those establishments without teams,


Ac

large establishments continue to be less likely to use profit sharing. The interaction with

those establishments with teams and training takes the familiar positive and significant

coefficient that roughly offsets the original size coefficient. Size plays no role among

21
Indeed, Lynch and Black (1998) directly confirm that the link between teams and employer-

provided training is driven by a higher incidence of "teamwork training".

26

Page 27 of 41
these establishments. One cannot reject the hypothesis that the sum of coefficients equals

zero. The interaction with those establishments with teams without training takes a

negative but insignificant size. Indeed, if one excludes this interaction (placing the

establishments in with the base group), the negative coefficient on size grows in both size

t
and significance. Thus, it appears that while coverage may not be a key indicator of team

ip
production, the combination of teams and training may be critical.

cr
us
VI. Conclusions

The basic proxy of teams as an indicator of team production fits the theoretical model
an
presented earlier in the paper. We are able to routinely confirm across a variety of

specifications that among those establishments without teams, profit sharing becomes
M
less likely as the size of the establishment grows. This fits nicely with the well-known

1/N problem associated with profit sharing but not typically found in previous studies.
d

Team production introduces a cost to shirking that is otherwise absent. Shirking


te

reduces not only the output of the shirking worker but potentially creates nontrivial
p

reductions in the output of the entire establishment. This serves to internalize, at least to
ce

some extent, the externality of shirking. Thus, we have contended that the 1/N problem

should not be observed or be as severe in settings of team production. We took teams as


Ac

an indicator of team production and tried several alternative definitions to check for

robustness. The results consistently show that size is not a significant determinant of

profit sharing use among establishments with teams even as a variety of other important

determinants were identified. This finding argues that for a class of large manufacturing

workplaces in which substantial team production is likely, concern over free-riding may

27

Page 28 of 41
be misplaced. It also argues that the choice of production technology can influence the

effectiveness of profit sharing as a motivational scheme. In general, emphasizing team

production makes profit sharing more successful.

We recognize the need for continued research within this theme. As a single

t
cross-section of establishments, the data we use can only provide a coherent pattern of

ip
statistical regularities rather than confirm causation. Longitudinal data may control for

cr
establishment fixed effects but would do so with a dramatic reduction in the variation in

us
establishment size if size changes only slightly from year to year. Future work might try

to refine the indicators of team production. We were able to make use of the presence of
an
training to identify types of teams that we thought were more likely to indicate team

production. Alternative data sources may have greater detail on the functions and
M
purposes of teams. Those that are jointly responsible for the entire production of output

or a component may be more likely to indicate team production. Those that require
d

extensive communication and the close working together of team members might
te

similarly be more indicative of team production. Entirely different measures of team


p

production may also be possible by examining the way workers interact with machines or
ce

by understanding the detailed nature of the assembly process. The point remains that

there exist potentially fruitful ways to measure team production and that these might be
Ac

used for even stronger tests of the hypotheses presented in this paper.

28

Page 29 of 41
Table 1: Variable Definitions and Descriptive Statistics

Variable Description (mean, standard deviation), n = 640


Profit Share Dummy variable equal to 1 if the establishment provides profit sharing to
employees other than executives (.142, .350).
Percent Profit Share Percent of an establishment's employees participating in the establishment's
profit sharing plan (7.22, 24.4).
Size Total employees in the establishment divided by 100 (1.48, 3.45).

t
ip
Team Dummy variable equal to 1 if blue-collar workers are organized in production
teams (.391, .488).
Team (<50) Dummy variable equal to 1 if less than 50 percent of blue-collar workers are

cr
organized in production teams (.237, .425)
Team (>50) Dummy variable equal to 1 if at least 50 percent of blue-collar workers are
organized in production teams (.153, .360)

us
Team with Training Dummy variable equal to 1 if blue-collar workers are organized in production
teams and the establishment finances further training (.272, .445)
Team without Training Dummy variable equal to 1 if blue-collar workers are organized in production
teams and the establishment does not finance further training (.119, .324)
Women

Usual Hours
an
Women as a proportion of all employees (.286, .233).

Usual weekly hours for production workers excluding overtime hours (37.6,
1.83)
M
BlueCol Blue-collar workers as a proportion of total employees (.633, .185).

Temporary Temporary workers as a share of total employees (.025, .082)


d

Part time Part time workers as a share of total employees (.081, .124)
te

Works Council Dummy variable equal to 1 if the establishment has a works council (.588,
.493)
Collective Bargaining Dummy variable equal to 1 if the establishment is covered by a collective
p

bargaining agreement (.664, .473).


Expansion Dummy variable equal to 1 if management plans to increase the market share
ce

of the firm (.569, .496).


Active Owner Dummy variable equal to 1 if owners are active in establishment management
(.632, .482).
Executive PS Dummy variable equal to 1 if the executive managers have a profit sharing
Ac

plan (.483, .500).


Profit Centers Dummy variable equal to 1 if profit centers have been created within the
establishment (.203, .403)
TrainExp Expenditures on employer provided further training divided by total employees
(145.9, 309.2)
Uni University graduates as a proportion of total employees (.036, .057)

Piece Rates Dummy variable equal to 1 if the establishment uses piece rates for its
production workers (.178, .383)

Premium Pay Dummy variable equal to 1 if the establishment uses premium pay for its
production workers (.169, .375)

29

Page 30 of 41
Wages as Motivation Manager's attitude toward high wages as an instrument to motivate workers:
1=poor, 2=fair, 3=good or very good (2.56, .625)
Involvement as Manager's attitude toward employee involvement in decision making as an
Motivation instrument to motivate workers: 1=poor, 2=fair, 3=good or very good (2.78,
.479)
Dismissal as Manager's attitude toward the threat of dismissal as an instrument to motivate
Motivation workers: 1=poor, 2=fair, 3=good or very good (1.34, .585)
Old Workers Dummy variable equal to 1 if management regards the average age of

t
employees as too high (.134, .341)

ip
cr
us
an
M
d
p te
ce
Ac

30

Page 31 of 41
Table 2: Profit sharing, Size and Teams

Profit Share = 1 Profit Share = 0 Difference

t
Mean, standard deviation Mean, standard deviation

ip
Teams .560, .499 .362, .481 .198

cr
Size 1.72, 3.37 1.44, 3.47 .28

us
N 91 549

an
M
d
p te
ce
Ac

31

Page 32 of 41
Table 3: Estimating the Probability of Using Profit Sharing

(1) (2) (3) (4)


Team=0 Team=1
Team .3343 .0936
[.0533] [0.141]
(2.33)** (0.52)
Size -.0338 -.2071 -.3260 -.0114

t
[-.0051] [-.0307] [-.0291] [-.0024]

ip
(1.25) (2.53)** (2.98)** (0.67)
Team x Size .1836
[.0264]

cr
(2.23)**
BlueCol .1744 .1441 .5438 -.0894
[.0263] [.0214] [.0486] [-.0186]

us
(0.38) (0.31) (0.90) (0.13)
Usual Hours .1557 .1573 .1348 .1698
[.0235] [.0233] [0.120] [.0353]
(3.36)** (3.41)** (2.08)** (2.33)**
Women .1408
[.0213]
(0.37)
an
.1622
[.0240]
(0.43)
.3430
[0.306]
(0.61)
.2932
[.0609]
(0.55)
TrainExp .0007 .0007 .0010 .0005
M
[.0001] [.0001] [.0001] [.0001]
(2.87)** (2.92)** (3.01)** (1.90)*
Uni 2.592 2.609 3.394 2.728
[.3916] [.3868] [.3032] [.5669]
d

(2.02)** (2.04)** (1.56) (1.59)


Temporary -.6646 -.6897 -.0648 -.4722
te

[-.1004] [-.1022] [-.0058] [-.0981]


(0.72) (0.76) (0.05) (0.33)
Part time 1.687 1.691 1.331 1.902
p

[.2548] [.2507] [.1189] [.3953]


(2.46)** (2.44)** (1.50) (1.67)*
ce

Works Council -.4222 -.3923 .2537 -.9627


[-.0675] [-.0613] [.0219] [-.2212]
(1.50) (1.39) (0.64) (2.33)**
Collective Bargaining -.2448 -.2748 -.2438 -.3969
Ac

[-.0391] [-.0435] [-.0235] [-.0883]


(1.00) (1.12) (0.67) (1.09)
Works Council x .6670 .7469 .4629 1.274
Collective Bargaining [.1073] [.1182] [.0564] [.2312]
(1.87)* (2.11)** (0.96) (2.43)**
Expansion .2937 .2870 .1338 .4606
[.0433] [.0415] [.0118] [.0926]
(1.99)** (1.94)* (0.62) (2.15)**
Active Owner -.0517 -.0613 -.3931 .2838
[-.0079] [-.0092] [-.0394] [.0572]
(0.31) (0.36) (1.75)* (1.07)
Executive PS 1.151 1.175 1.291 1.177

32

Page 33 of 41
[.1868] [.1879] [.1474] [.2317]
(6.93)** (7.02)** (4.97)** (4.81)**
Profit Centers -.0552 -.0012 .2009 -.1649
[-.0082] [-.0002] [.0202] [-.0328]
(0.32) (0.01) (0.74) (0.60)
Piece Rates -.2787 -.2668 -.3436 -.3127
[-.0371] [-.0350] [-.0254] [-.0579]
(1.45) (1.36) (1.14) (1.05)

t
Premium Pay .4088 .4439 .1888 .6807

ip
[.0742] [.0805] [.0189] [.1749]
(2.36)** (2.53)** (0.69) (2.63)**
Wages as Motivation .2003 .1967 .3240 .0409

cr
[.0303] [.0292] [.0290] [.0085]
(1.85)* (1.81)* (1.90)* (0.27)
Involvement as -.0584 -.0518 -.3589 .3094

us
Motivation [-.0088] [-.0077] [-.0321] [.0643]
(0.38) (0.34) (1.87)* (1.13)
Dismissal as Motivation .1265 .1265 .3188 -.0001
[.0191] [.0188] [.0285] [-.00001]

Old Workers
(1.05)
.0468
[.0072]
an
(1.06)
.0067
[.0010]
(1.87)*
-.1638
[-.0132]
(0.00)
.1169
[.0255]
(0.21) (0.03) (0.46) (0.36)
M
Constant -9.188 -9.029 -8.392 -10.07
(4.55)** (4.49)** (3.18)** (3.08)**
Industry Dummies Yes Yes Yes Yes
Chi-squared 105.2** 110.0** 60.0** 69.5**
d

Pseudo R-squared .231 .239 .264 .270


N 640 640 390 250
te

Notes: The first entry is the coefficient, the entry in square brackets is the marginal effect and the
entry in parentheses is the asymptotic t-statistic. All estimates account for robust standard errors.
p

The marginal effects for interaction variables are calculated using the Norton et al. (2004) routine.
* significant at 10 percent; ** significant at 5 percent.
ce
Ac

33

Page 34 of 41
Table 4: Profit Sharing Coverage and Profit Centers

Percent Profit Share Use of Profit Sharing


(Tobit) (Probit)
Profit Center = 0 Profit Center = 1
Team=0 Team=1 Team=0 Team=1 Team=0 Team=1

t
ip
Size -52.11 -2.632 -.3176 .0354 -.7056 -.0127
(2.59)** (0.95) (1.91)* (0.60) (1.66)* (0.62)
Profit Center 32.59 -4.167

cr
(0.82) (0.14)
Industry Yes Yes Yes Yes Yes Yes
Dummies

us
Chi-squared 66.7** 65.20** 64.6** 51.9** 19.3 24.5
Pseudo R- .143 .118 .292 .327 .547 .263
squared
N 387 247 326 184 64 66
an
Notes: The first entry is the coefficient and the entry in parentheses is the asymptotic t-statistic.
All estimates account for robust standard errors. * significant at 10 percent; ** significant at 5
percent. The full set of variables listed in Table 3 are included in the estimations but not shown to
M
save space.
d
p te
ce
Ac

34

Page 35 of 41
Table 5: The Role of Team Coverage

(1) (2) (3) (4) (5)


Team=0 Team (<50) Team (>50)
Team (<50) .0812 -.2447
(0.48) (1.13)
Team (>50) .6539 .4284
(3.42)** (1.99)**

t
Size -.0318 -.2258 -.3260 .0359 -.0851

ip
(1.50) (2.67)** (2.98)** (0.96) (1.14)
Team(<50) x Size .2212
(2.52)**

cr
Team(>50) x Size .1913
(2.22)**
Industry Dummies Yes Yes Yes Yes Yes

us
Chi-squared 106.7** 112.33** 60.0** 49.6** 31.7
Pseudo R-squared .244 .254 .264 .4604 .295
N 640 640 390 152 98
an
Notes: The first entry is the coefficient and the entry in parentheses is the asymptotic t-statistic.
All estimates account for robust standard errors. * significant at 10 percent; ** significant at 5
percent. The full set of variables listed in Table 3 are included in the estimations but not shown to
save space.
M
d
p te
ce
Ac

35

Page 36 of 41
Table 6: Interaction of Training and Teams

(1) (2)

Team with Training .4231 0.169


(2.69)** (0.83)
Team without Training .0790 .0599

t
(0.30) (0.18)

ip
Size -.0362 -.1970
(1.25) (2.36)**
Size x Team with Training .1726

cr
(2.06)**
Size x Team without Training -.1135
(0.28)

us
Industry Dummies Yes Yes

Chi-squared 104.3** 109.8**

Pseudo R-squared

N
an .234

640
.241

640
M
Notes: The first entry is the coefficient and the entry in parentheses is the asymptotic t-statistic.
All estimates account for robust standard errors. * significant at 10 percent; ** significant at 5
percent. The full set of variables listed in Table 3 are included in the estimations but not shown to
d

save space.
p te
ce
Ac

36

Page 37 of 41
References

Alchian, A.A., Demsetz, H,. 1972. Production costs, information and economic organization.
American Economic Review 62, 777 – 95.

Adams, C.P., 2006. Optimal team incentives with CES production. Economics Letters 92, 143 –
48.

t
Adams, C.P., 2002. Does size really matter? Empirical evidence on group incentives. Working

ip
Paper No. 252, Federal Trade Commission, Washington DC.

Ai, C. Norton, E., 2003. Interaction terms in logit and probit models. Economic Letters 80, 123 –

cr
29.

Aoki, M., 1993. The informationally-participatory firm. In Bowles, S., Ginitis, H., Gustafsson, B.

us
(Eds.), Markets and Democracy: Participation, Accountability and Efficiency. Cambridge
University Press, Cambridge, 231 – 247.

Aoki, M., 1990. The participatory generation of information rents and the theory of the firm. In
an
Aoki, M.B. Gustafsson, B., Williamson, O.E. (Eds.), The Firm as a Nexus of Treatises.
Sage, Thousand Oaks, CA, 26 – 52.

Amisano, G., Del Boca, A., 2004. Profit related pay in Italy: A microeconometric analysis.
M
International Journal of Manpower 25, 463 – 478.

Azfar, O., Danninger, S., 2001. Profit sharing, employment stability and wage growth. Industrial
and Labor Relations Review 54, 619 – 30.
d

Baker, G., 2002. Distortion and risk in optimal incentive contracts. Journal of Human Resources
37, 728 – 751.
te

Baker, G., 1992. Incentive contracts and performance measurement. Journal of Political Economy
100, 1125 – 1156.
p

Bayo-Moriones, A., Emilio Herta-Arribas, E., 2002. Organisational incentive plans in Spain.
ce

Personnel Review 31, 128-42.

Berman, E., Bound, J., Machin, S.J., 1998. Implications of skill-biased technological change:
International evidence. Quarterly Journal of Economics 113, 1245 – 79.
Ac

Bertrand, M., 2004. From the invisible handshake to the invisible hand? How import competition
changes the employment relationship. Journal of Labor Economics 22, 723 – 65.

Brand, R., Carstensen, V., Gerlach, K., Klodt, T., 1996. The Hanover panel. Discussion Paper
No. 2, University of Hanover, Institute for Quantitative Economic Research.

Brown, M., Geddes, L.A., Heywood, J.S., 2007. The determinants of employee involvement
schemes: Private sector Australian evidence. Economic and Industrial Democracy 28, 259 –
91.

37

Page 38 of 41
Bryson, A., Freeman R.B., 2007. “Doing the right thing? Does fair share capitalism improve
workplace performance?” Employment Relations Research Series No. 81, London:
Department of Trade and Industry.

Cable, J., 1984. “Employee participation and firm performance: A prisoners’ dilemma
framework,. EUI Working Paper No. 84/126, University of Warwick.

Carter, M.J. 1995. Information and the division of labor: Implications for the firm’s choice of

t
organization. Economic Journal 105, 385 – 397.

ip
Cheadle, A. 1989. Explaining patterns of profit sharing activity. Industrial Relations 28, 387 –
400.

cr
Cheatham, C., Davis, D., Cheatham, L., 1996. Hollywood profits: Gone with the wind? CPA
Journal 66, 32 – 34.

us
Che, Y.K., Yoo, S.W., 2001. Optimal incentives for teams. American Economic Review 91, 525
– 41.

an
Devaro, J., 2008. The effects of self-managed and closely managed teams on labor productivity
and product quality: An empirical analysis of a cross section of establishments. Industrial
Relations 47, 659 – 97.
M
Devaro, J., 2006. Teams, autonomy, and the financial performance of firms. Industrial Relations
45, 217 – 69.

Drago, R., Heywood, J.S., 1995. "Choice of payment schemes: Australian establishment data.
Industrial Relations 34, 507 – 31.
d

Estrin, S., Wilson, N., 1989. Profit sharing, the marginal cost of labour, and employment
te

variability. Working Paper, London: London School of Economics, Department of


Economics.
p

FitzRoy, F., Kraft, K. 1995. On the choice of incentives in firms. Journal of Economic Behavior
and Organization 26,145 – 60.
ce

FitzRoy, F., Kraft, K., 1987. Cooperation, productivity and profit sharing. Quarterly Journal of
Economics 102, 23 – 36.
Ac

Frant, H., 1991. Specifying a model of state policy intervention. American Political Science
Review 85, 571 – 73.

Freeman, Richard, Douglas Kruse and Joseph Blasi. 2008. Worker Responses to Shirking Under
Shared Capitalism. National Bureau of Economic Research, Working Paper No. 14227,
Cambridge, MA.

Freeman, R.B., Lazear, E.P., 1995. An Economic Analysis of ‘Works Councils. In Rogers, J.,
Streeck, W. (Eds.), Works Councils – Consultation, Representation and Cooperation in
Industrial Relations, Chicago, University of Chicago Press, 27 – 52.

38

Page 39 of 41
Gerlach, K., Jirjahn, U., 2001. Employer provided further training: Evidence from German
establishment data. Journal of Applied Social Science Studies 121, 139 – 64.

Goldin, C., 1986. “Monitoring costs and occupational segregation by sex: A historical analysis.
Journal of Labor Economics 4, 1 – 27.

Gregg, P.A., Machin, S.J., 1988. Unions and the incidence of performance linked pay schemes in
Britain. International Journal of Industrial Organization 6, 91 –107.

t
ip
Hart, R.A., Hubler O., 1990. Wage, labour mobility and working time effects of profit sharing.
Empirica – Austrian Economic Papers 17, 115 – 130.

cr
Heywood, J.S., Hubler, O., Jirjahn, U., 1998. Variable payment schemes and industrial relations:
Evidence from Germany. Kyklos 51, 327 – 57.

us
Heywood, J.S., Jirjahn U., 2004. Teams, teamwork and absence. Scandinavian Journal of
Economics 106,765 – 82.

Heywood, J.S., Jirjahn, U., 2002. Payment schemes and gender in Germany. Industrial and Labor
Relations Review 56, 44 – 64. an
Heywood, J.S., Jirjahn, U., Wei, X., 2008. Teamwork, monitoring and absence. Journal of
Economic Behavior and Organization 68, 676 – 90.
M
Holmstrom, B., 1982. Moral hazard in teams. Bell Journal of Economics 13, 324 – 340.

Holmstrom, B., Milgrom, P., 1991. Multitask principal agent analyses: Incentive contracts, asset
ownership and job design. Journal of Law, Economics and Organization 7, 24 – 52.
d

Hubler, O., Jirjahn, U., 2003. Works councils and collective bargaining in Germany: The impact
te

on productivity and wages. Scottish Journal of Political Economy 50, 1 – 21.

Jirjahn, U., 2006. A note on efficiency wage theory and principal-agent theory. Bulletin of
p

Economic Research 58, 235 – 52.


ce

Jirjahn, U., 2002. The German experience with performance pay. In Heywood, J.S., Brown, M.
(Eds.), Paying for Performance: An International Comparison, New York, M.E. Sharpe,
148 – 78.
Ac

Jirjahn, U., 2000. Incentives for multitasking: Fixed wages or profit sharing?” Economic Analysis
3, 137 – 48.

Jones, D., Pliskin, J., 1997. Determinants of the incidence of group incentives: Evidence from
Canada. Canadian Journal of Economics 30, 1027 - 45.

Kandel, E., Lazear, E.P., 1992. Peer pressure and partnerships. Journal of Political Economy 100,
801 – 17.

Kremer, M., 1993. The ‘O-ring’ theory of economic development. Quarterly Journal of
Economics 108, 551 – 75.

39

Page 40 of 41
Kruse, D.L., 1996. Why do firms adopt profit sharing and employee ownership plans? British
Journal of Industrial Relations 34, 515 – 38.

Kruse, D.L. 1993. Profit Sharing: Does It Make a Difference? Kalamazoo Michigan, Upjohn
Institute.

Lazear, E.P., 1981. Agency, earnings profiles, Productivity and hours restrictions. American
Economics Review 71, 606 – 20.

t
ip
Lazear, E.P., 1979. Why is there mandatory retirement? Journal of Political Economy 87, 1261 –
84.

cr
Lynch, L.M., Black, S.E., 1998. Beyond the incidence of employer-provided training. Industrial
and Labor Relations Review 52, 64 – 81.

us
Nagler, J., 1991. The effect of registration laws and education on U.S. voter turnout. American
Political Science Review 85, 1393 – 405.

Norton, E., Wang, H., Ai, C., 2004. Computing interaction effects and standard errors in logit and
probit models. The Stata Journal 4, 154 – 67. an
Pendleton, A., 2006. Incentives, monitoring, and employee stock ownership plans: New evidence
and interpretations. Industrial Relations 45, 753 – 77.
M
Pendleton, A., 1997. Characteristics of workplaces with financial participation: Evidence from
the Workplace Industrial Relations Survey. Industrial Relations Journal 28, 103 – 19.

Perotin, V., Robinson, A., 2003. Employee Participation in Profit and Ownership: A Review of
d

the Issues and the Evidence. Luxembourg, European Parliament.


te

Prendergast, C., 1999. The provision of incentives in firms. Journal of Economic Literature 37, 7
– 63.
p

Robinson, A.M., Wilson, N., 2006. Employee financial participation and productivity: An
empirical reappraisal. British Journal of Industrial Relations 44, 31 – 50.
ce

Ross, S., 1973. The economic theory of agency: The principal’s problem. American Economic
Review 63, 34 – 39.
Ac

Spreitzer, G., Cohen, S., Ledford, G. 1999. Developing effective self-managing work teams in
service organizations. Group and Organization Management 24, 340 – 366.

Wagar, T.H., Long, R., 1995. Profit sharing in Canada: Incidence and predictors. Proceedings of
the Administrative Sciences Association of Canada (Human Resources Division) 16, 97 –
105.

40

Page 41 of 41

You might also like