The one constant: a causal effect of collective bargaining on employment growth? Evidence from German linked‐employer‐employee data
| DOI | http://doi.org/10.1111/sjpe.12180 |
| Date | 01 November 2018 |
| Published date | 01 November 2018 |
| Author | Tobias Brändle,Laszlo Goerke |
THE ONE CONSTANT: A CAUSAL
EFFECT OF COLLECTIVE BARGAINING
ON EMPLOYMENT GROWTH?
EVIDENCE FROM GERMAN LINKED-
EMPLOYER-EMPLOYEE DATA
Tobias Br€
andle* and Laszlo Goerke**,***,****
ABSTRACT
A large number of articles have analysed ‘the one constant’ in the economic
effects of trade unions, namely that collective bargaining reduces employment
growth by 2–4% points per year. Evidence is, however, mostly related to Anglo-
Saxon countries. We investigate whether a different institutional setting might
lead to a different outcome, making the constant a variable entity. Using linked-
employer-employee data for Germany, we find a negative correlation between
being covered by a sector-wide bargaining agreement or firm-level contract and
employment growth of about 1% point per annum. However, the correlation
between employment growth and collective bargaining is not robust to the use of
panel methods. We conclude that the results of the literature using cross-section
data might be driven by selection.
II
NTRODUCTION
What are the economic effects of trade unions on employment, that is, does
collective bargaining reduce employment? The neoclassical theory suggests
that if wages equal marginal productivity and trade unions raise wages, firms
will choose a lower quantity of labour, moving up along the (inverse) labour
demanded schedule (Hammermesh, 1993). However, if firms and trade unions
bargain not only over wages but also over employment, collective negotiations
might increase employment beyond the level chosen by the firm if it deter-
mined employment unilaterally at the level at which its marginal revenue pro-
duct equalled the wage (McDonald and Solow, 1981). Furthermore, trade
unions could increase labour productivity and boost employment in unionised
plants by raising the quality of job matches or reducing turnover, such that
the incentives to invest in firm-specific human capital may be larger.
*Institute for Applied Economic Research at the University of T€
ubingen
**IAAEU Trier, Trier University
***CESIfo
****IZA
Scottish Journal of Political Economy, DOI: 10.1111/sjpe.12180, Vol. 65, No. 5, November 2018
©2018 Scottish Economic Society.
445
Despite this theoretical ambiguity, empirical analyses at first sight provide a
clear-cut picture and have uncovered what Addison and Belfield (2004) refer
to as the one constant among the economic effects of trade unions: Unionism
reduces employment growth by 2–4% per annum. This interpretation is based
on data primarily from Anglo-Saxon countries which tend to be characterised
by a pluralistic system of industrial relations and low collective bargaining
coverage in the private sector.
1
Moreover, collective bargaining primarily
takes place at the plant level.
Therefore, in this paper we inquire whether the negative employment effects
observed for Anglo-Saxon countries are also present in a more cooperative and
corporatist industrial relations system such as Germany, the largest economy in
the European Union and fourth largest in the world. It represents an interesting
case for a number of reasons: First, while Germany has been called the sick man
of Europe only about a decade and a half ago, its economy has been remarkably
stable during the Great Recession. Nowadays, it is sometimes viewed as role
model, outperforming its European competitors, especially with regard to the
labour market (Dustmann et al., 2014). Second, collective bargaining coverage is
still relatively high in Germany: sector-wide bargaining covers more than half of
the workforce. If wages in a large part of the economy are resulting from
collective bargaining, it will be easier to shift forward cost increases into prices,
such that the employment effects of collective bargainingmay be less pronounced
than, for example, in Anglo-Saxon countries characterised by a lower level of
coverage. The strong export-orientationof many German firms will further affect
the employment consequences of collective bargaining. Exporting firms often
have higher productivity levels than non-exporters. Therefore, detrimental
employment effects may be less pronounced. However, if international competi-
tion is more intense than domestic competition, exporting firms cannot shift for-
ward higher wages into prices as easily as their non-exporting counterparts. This
suggests that the employment effects of collective bargaining are aggravated for
exporting plants. Third, there are also negotiations at the firm level for about
10% of employees. Accordingly, we can distinguish the employment effects of
more and less centralised collectivenegotiations. A fourth reason why we look at
Germany is that its labour market and institutional settings have undergone a
number of changes recently. Following unification, unemployment was rather
high at the dawn of the century and collective bargaining coverage has declined
by about 15% points over the last decades. Furthermore, a range of labour
market reforms (Hartz IV laws etc.) havebeen enacted in the last 15 years. Fifth,
we can utilise high quality linked-employer-employee panel data, namely the
widely used LIAB data set from the Institute for Employment Research (IAB).
Accordingly, we can look at various subgroups, analyse the interaction of
particular features of the German labour market, such as works councils, with
collective bargaining and scrutinise whether the relationship between collective
bargaining and employment growth can be interpretedcausally. Additionally, we
1
Australia is somewhat of an exception in the latter regard with a collective bargaining
coverage of about 60% (Visser, 2016).
446 TOBIAS BR
€
ANDLE AND LASZLO GOERKE
Scottish Journal of Political Economy
©2018 Scottish Economic Society
can observe plants annually for a period of almost 15 years, while previous
contributions either rely on cross-sectional data or on substantially fewer
observations per plant over time, which encompass larger time spans. Finally, if
the one constant were to exist for Germany, its overall impact would be substan-
tially larger than in economiesexhibiting lower coverage.
Correspondingly, we contribute to the literature, first, by providing evidence
for a major country, for which no such findings are available yet. Second, we
go beyond most contributions by looking at the question of whether observed
effects can be interpreted causally from a variety of vantage points. Finally,
we distinguish different degrees of bargaining centralisation.
Our baseline findings indicate that collective bargaining is associated with a
reduction in employment growth in German plants by 1% point per annum,
less than in those countries for which the one constant has yet been detected.
Importantly, we observe no large differences between negotiations at the plant
or at the sector level. When using the time dimension of the data, fixed-effects
and difference-in-differences estimates provide no indication that employment
growth is affected by the introduction or the abolition of collective bargaining.
More elaborate estimation techniques, which account for endogeneity, fail to
deliver valid results: Dynamic panel estimations and instrumental variables
techniques provide no basis for a causal interpretation due to the rejection of
the respective state-of-the-art test procedures. Therefore, our results suggest
that the employment effects associated with collective bargaining in Germany
are due to self-selection into bargaining regimes.
After having presented our motivation, we introduce the relevant literature
in Section II and use Section III to characterise the institutional setting. Sec-
tion IV provides an overview of the data, presents descriptive evidence and
outlines the empirical methods we employ. The basic results as well as findings
from various robustness checks are presented in Section V. Section VI con-
cludes and puts our findings into perspective.
II RELATED LITERATURE
The union employment literature has been established on the basis of cross-sec-
tional evidence, mainly for Great Britain or the United Kingdom. These stud-
ies generally employ data from Workplace Industrial Relations Surveys
(WIRS) or Workplace Employee Relations Survey (WERS). Since the early
2000s, contributions tend to rely on repeated cross-sectional or panel data.
Irrespective of the data utilised, a common and widely documented finding is
that employment in unionised plants grows between 2% and 4% less per
annum than in non-unionised plants.
2
A few analyses can establish such an
effect for selected plants only (Machin and Wadhwani, 1991), or cannot dis-
cern it at all (Blanchflower and Burgess, 1996, Bryson and Dale-Olsen, 2008).
The result that employment is lower in British and UK plants covered by
2
See, inter alia, Blanchflower et al. (1991), Blanchflower and Burgess (1996), Booth and
McCulloch (1999), Addison et al. (2000), Bryson (2004), Addison and Belfield (2004) and
Bryson and Nurmi (2011).
COLLECTIVE BARGAINING AND EMPLOYMENT 447
Scottish Journal of Political Economy
©2018 Scottish Economic Society
To continue reading
Request your trial