Performance Pay, Union Bargaining and Within‐Firm Wage Inequality*
| Author | Bernt Bratsberg,Oddbjørn Raaum,Erling Barth,Torbjørn Hægeland |
| DOI | http://doi.org/10.1111/j.1468-0084.2011.00656.x |
| Published date | 01 June 2012 |
| Date | 01 June 2012 |
327
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford, 2011. Published by Blackwell Publishing Ltd,
9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 74, 3 (2012) 0305-9049
doi: 10.1111/j.1468-0084.2011.00656.x
Performance Pay, Union Bargaining and Within-Firm
Wage Inequality*
Erling Barth,†Bernt Bratsberg,‡Torbjørn Hægeland§ and
Oddbjørn Raaum‡
†Institute for Social Research, PO Box 3233 Elisenberg, N-0208 Oslo, Norway
(e-mail: erling.barth@socialresearch.no)
‡Ragnar Frisch Centre for Economic Research, Gaustadalleen 21, N-0349 Oslo, Norway
(e-mails: bernt.bratsberg@frisch.uio.no; oddbjorn.raaum@frisch.uio.no)
§Statistics Norway, PO Box 8131 Dep, N-0033 Oslo, Norway
(e-mail: torbjorn.haegeland@ssb. no)
Abstract
Theory predicts that performance pay boosts wage dispersion. Workers retain a share
of individual productivity shocks and high-efficiency workers receive compensation
for greater effort. Collective bargaining can mitigate the effect of performance pay on
wage inequality by easing monitoring of common effort standards and group-based pay
schemes. Analyses of longitudinal employer–employee data show that the introduction of
performance-related pay raises wage inequality in non-union firms, but not in firms with
high union density.Although performance-related pay appears to be on the rise, the overall
impact on wage dispersion is likely to be small, particularly in European countries with
influential unions.
I. Introduction
Performance-related pay and individualized wage setting appear to be on the rise, even
in economies with influential unions. Linking pay to firm performance has long been a
prevalent method of rewarding executive managers (Jensen and Meckling, 1976; Holm-
str¨om and Milgrom, 1987). But individualized pay and performance-related pay seem to
be spreading to lower ranks of employees within organizations, and throughout the labour
market as well, see Lemieux, MacLeod and Parent (2009) for recent evidence from the
US and Kersley et al. (2006) for the UK. In Europe, where union-negotiated, fixed wages
ÅWeare grateful to Carlo Dell’Aringa, Alex Bryson, Donna Ginther, Espen Moen and two anonymous referees for
comments on earlier versions of this paper. We also thank seminar participants at the Catholic University of Milan,
University of Oslo, workshop on the use and analysis of employer–employee data in Oslo and the lower conference in
Esbjerg for useful comments. This work has received financial support from the Norwegian Research Council (grants
# 156035/S20 and 173591/S20) the Norwegian Ministry of Labour, the EDWIN project of the European Commis-
sion IHP project (grant #HPSE-CT-2002-00108)and is part of the research activities of the centre of equality, social
organization, and performance (ESOP), University of Oslo.
JEL Classification numbers: J31, J33.
328 Bulletin
have been predominant, this development represents an important break with traditional
forms of remuneration. An important question is whether such developments have conse-
quences for wage inequality.This paper develops a theoretical framework for analysing the
effects of performance-related pay on within-firm wage dispersion and confronts theory
with evidence from a rich employer–employee data set.
Performance pay may affect wage dispersion differently depending on the institutional
environment in which it is implemented. In particular, the presence of unions and collective
wage bargaining will alter the conditions for and the design of performance-related pay
schemes. Understanding the interplay between pay systems and union bargaining is thus
crucial in order to understand the development of pay setting regimes, especially in Euro-
pean labour markets where unions are widespread, as well as to understand differences in
levels and changes in within-group wage dispersion across economies.
Three features of unions are important in the context of performance pay. First, unions
appropriate rent, see, e.g. Blanchflower and Bryson (2003) on the wage gaps between union
and non-union workers in the UK and the US. The bargaining power of unions needs to
be incorporated in models of wage formation even if contracts include performance based
components. Under bargaining, the wage policy of the firm is typically not constrained
by outside options (i.e. a participation constraint), but will be limited by union power.
Second, unions tend to compress wages. Wage dispersion is typically lower in union-
ized environments, see, e.g. Freeman and Medoff (1984) and Card, Lemieux and Ridell
(2004).1Since performance pay can be expected to induce worker heterogeneity in effort,
and therefore pay, a theoretical framework must specify the wage compressing motives
of the union. We consider a utilitarian union that represents the interests of risk-averse
employees. Risk aversion implies that more weight is placed on low pay than on high pay,
providing collective preferences that favour pay equality. Third, unions alter information
flows. Freeman and Medoff (1984) argue that unions provide workers with a ‘collective
voice’ that improves the information flow within the firm. In particular, unions may voice
grievances or preferences about workplace conditions that individual workers are reluc-
tant to put forward. In the agency literature, the important information problem is lack of
observability (or at least verifiability) of effort. We argue that the union may operate as a
monitoring device, as co-workers will be able to observe worker effort more easily and less
costly than management.2A higher level of trust among workers facilitates information
sharing across employees. Consequently, the union will be able to enforce a given job
standard without substantial monitoring costs. The problem for the firm, then, is how to
provide the union with incentives to uphold a certain job standard. A group bonus may
serve this purpose. The advantage of a bonus based on collective, rather than individual,
outcomes is that risk is shared among workers.
1A large literature has paired variation in unionism and wage inequality across time and space. DiNardo, Fortin
and Lemieux (1996), for example, conclude that declining unionism contributed to a considerable portion of the
increase in US wage inequality during the 1980s. Kahn (2000) argues that collective coverage explains a large part
of the observed differences in wage dispersion across countries.
2See Pencavel (1977) for an early discussion of such a mechanism. According to Metcalf (2003), the Donovan
Commission (1968) study of UK shop stewards emphasized the shop stewards’ role in communication, information
and discipline of workers. See also Vroman (1990), who uses a similar mechanism to explain the cyclical behaviour
of the union-nun-union wage differential in an efficiency wage model.
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford 2011
Performance pay and inequality 329
Our theoretical framework highlights performance pay as a remedy for agency prob-
lems: basically, performance pay is introduced in order to align the effort put forth by
workers with the interests of the firm. The model integrates individual worker effort deci-
sions in a collective wage bargaining framework and provides predictions about the impact
of performance pay on the wage distribution and the interaction between performance pay
and collective wage setting.3We analyse both individual and group-based performance-
pay schemes. In the collective bargaining case, where an effort standard is monitored by the
union, the trade-off between individual and group-based pay schemes is shown to depend
on the relative importance of risk sharing within the group vs. the efficiency loss associated
with enforcing a common job standard.
The theoretical model predicts that: (i) performance pay will raise within-firm wage dis-
persion; but (ii) the impact of introducing performance pay is smaller in the union case than
in the non-union case; (iii) within-firm wage dispersion under group-based performance-
pay systems is likely to be similar to that found under fixed-pay collective bargaining;
and that (iv) group-based performance-pay systems are likely to be more prevalent when
wages are set through union bargaining.
A careful assessment of the impact of performance-related pay on the firm wage struc-
ture requires linked employer–employee data as well as a strategy for identification. Cross-
sectional comparisons of firms or individuals may confound the heterogeneity of firms and
workers with the effects of performance pay.We use a representative panel of two employer
surveys from Norway, matched with the full set of individual employee records. The data
allow us to use a comprehensive set of controls, both at the worker and at the firm level.
Moreover, we obtain identification with establishment heterogeneity using the fixed effect
estimator. Bias due to differential trends is rejected by means of a placebo presample
test.
The literature on performance pay broadly relies on two types of mechanisms, both
related to informational problems: employee sorting and agency problems (see e.g.
Lazear, 2000a). In the theory section, we abstract from sorting issues by assuming that
the employer can tailor the performance-pay scheme to each worker’s outside option. In
the empirical analysis, we address any confounding impact of employee sorting by also
examining results in a subsample of workers who stay with the same employer for a 6-year
period. Because results based on the full and same-job samples are similar, we argue that
our empirical findings are unlikely to be driven by employee sorting.
II. Empirical studies of performance pay
The empirical literature analysing the effects of performance-related pay focuses pri-
marily on productivity and wage levels, and few studies explicitly address implications
for inequality. Early studies of incentive schemes provide evidence of wage premiums
for piece-rate workers (see, e.g. Pencavel, 1977). Seiler (1984) uses establishment data
for two US industries and finds that incentive workers receive higher earnings than
other workers, in part to compensate for their higher risk and partially due to their greater
3Recent evidence, such as that reported in MacLeod and Parent (1999), provides support for the idea that agency
problems motivate the practice of performance pay. For evidence showing effects on employee sorting, see, e.g.
Lazear (2000b).
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford 2011
To continue reading
Request your trial