Comparative
Political Studies
http://cps.sagepub.com/
When Are Concerted Reforms Feasible? Explaining the Emergence
of Social Pacts in Western Europe
Sabina Avdagic
Comparative Political Studies 2010 43: 628 originally published online 6 January
2010
DOI: 10.1177/0010414009356178
The online version of this article can be found at:
http://cps.sagepub.com/content/43/5/628
Published by:
http://www.sagepublications.com
Additional services and information for Comparative Political Studies can be found at:
Email Alerts: http://cps.sagepub.com/cgi/alerts
Subscriptions: http://cps.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://cps.sagepub.com/content/43/5/628.refs.html
>> Version of Record - Mar 31, 2010
OnlineFirst Version of Record - Jan 6, 2010
What is This?
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
When Are Concerted
Reforms Feasible?
Explaining the
Emergence of Social
Pacts in Western Europe
Comparative Political Studies
43(5) 628­–657
© 2010 SAGE Publications
DOI: 10.1177/0010414009356178
http://cps.sagepub.com
Sabina Avdagic1
Abstract
Under what conditions do governments, employers, and unions enter formal
policy agreements on incomes, employment, and social security? Such agree­
ments, widely known as social pacts, became particularly prominent during the
1990s when European economies underwent major adjustment. This article
seeks to explain national variation in adjustment strategies and specifically
why concerted agreements were struck in some countries but not in others.
A fuzzy-set qualitative comparative analysis of 14 European countries is employed
to assess main arguments about the emergence of pacts. The analysis yields
two key findings. First, although prevailing arguments emphasize Economic and
Monetary Union–related pressures, or alternatively unemployment, these factors
were neither necessary nor in themselves sufficient for pacts to materialize.
Rather, a high economic “problem load” appears to be causally relevant only
when combined with particular political and institutional conditions, namely,
the prevalence of electorally weak governments and/or an intermediate
level of union centralization. Second, the analysis refines existing multicausal
explanations of pacts by demonstrating three distinct, theoretically and
empirically relevant causal pathways to concerted agreements.
1
University of Sussex, Brighton, UK
Corresponding Author:
Sabina Avdagic, University of Sussex, Department of Politics, Sussex European Institute, Falmer,
Brighton BN1 9RG, United Kingdom
Email: [email protected]
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
629
Avdagic
Keywords
policy concertation, social pact, economic adjustment,Western Europe, fuzzyset QCA
Much of the scholarly debate about economic adjustment and institutional
change has focused on the logic of negotiated reforms in advanced democracies. Particularly prominent in this debate have been endeavors to understand
the phenomenon of social pacts—formal, often multipolicy agreements
among the state, unions, and employers that spread across European countries during the 1990s. As the internationalization of the economy intensified
competitive pressures and sociodemographic changes increased the costs of
welfare provision, many governments turned to organized interests to negotiate reforms of wage-bargaining systems, social security, and employment
protection. This resurgence of concertation, however, differed in two crucial
aspects from classic neocorporatist bargains. First, unlike neocorporatism in
the 1970 when governments used welfare benefits and employment creation
to compensate unions for delivering wage moderation, concertation in the
1990s entailed a different type of exchange. Faced with stringent budgetary
constraints, governments could no longer play the same compensatory role. As
a result, new concerted deals relied more on the devolution of policy-making
functions to the unions than on redistributive policies and concessions (Regini,
1997). Second, somewhat unexpectedly, these new concerted agreements
emerged in countries lacking monopolistic and hierarchical interest organizations that the neocorporatist literature of the 1970s and 1980s identified as a
key precondition for a successful political exchange.
These distinct features of contemporary concertation generated a whole
new body of literature as scholars increasingly sought to explain why and
when such concerted solutions may be possible in the new economic environment. Although this literature generally agrees that concertation has the
potential to generate broader support for reforms and thus to promote reform
sustainability, there is no agreement on the conditions under which such concerted agreements are likely to materialize. As a consequence, we still lack a
consensus on one of the central questions addressed by this literature: Why
were such deals struck in some countries but not in others? Although some
accounts emphasize national variation in the extent of economic “problem
load,” particularly with respect to meeting the Maastricht convergence criteria for Economic and Monetary Union (EMU; Fajertag & Pochet, 2000;
Hancke & Rhodes, 2005), others highlight national differences in wagebargaining institutions (Hancke, 2002; Hassel, 2006) and, more recently, in
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
630
Comparative Political Studies 43(5)
government strength and specific patterns of electoral pressures (Baccaro &
Lim, 2007; Hamann & Kelly, 2007a). In most cases, however, these hypotheses have been derived and assessed on a small number of in-depth case
studies and attempts to evaluate them on a larger number of country cases
have been rather cursory.
The aim of this article is to more systematically assess the explanatory
power of the various arguments about national variation in the emergence of
social pacts. Relying on a fuzzy set qualitative comparative analysis (fs/
QCA) of 14 European countries during the 1990s, I find little support for the
dominant argument that sees pacts as a direct result of the extent of the underlying economic problem load, especially with regard to inflation and budget
deficit. This finding is in line with some recent accounts that argue that
policy concertation has been driven at least as much by governments’ political calculations as by their more general concerns about key macroeconomic
parameters (Baccaro & Lim, 2007; Hamann & Kelly, 2007a). But although
some of these accounts accept that EMU-related pressures may have been at
least necessary if not a sufficient condition for successful concertation during
the 1990s (Hamann & Kelly, 2007a, p. 974), my analysis shows that this condition by itself has been neither necessary nor sufficient. Instead, the analysis
reveals three possible paths to pacts, each combining at least two different conditions emphasized by the various strands of the literature. These three causal
configurations suggest that a high economic problem load generates pacts
only when combined with particular political and/or institutional conditions.
Specifically, weak governments and intermediate union centralization seem
to have been key ingredients for successful tripartite concertation. As far as
economic variables are concerned, it is evident that even during the run up to
EMU—when we would expect the explanatory power of the imbalances with
regard to the Maastricht criteria to be particularly strong—high unemployment
was almost as important for the emergence of pacts. As this article shows, the
three paths identified by the fuzzy set analysis are not only empirically relevant in that they explain all our cases well but also broaden our theoretical
understanding of the combinations of conditions that may be conducive to
successful policy concertation. In short, the findings of this analysis confirm that
the resurgence of concertation cannot be explained by a single factor and that
social pacts have multiple (albeit not entirely idiosyncratic) causes.
This article is organized as follows. Section I presents three main groups
of explanations of social pacts emphasizing economic, institutional, and
political causes, respectively, as well as several multicausal or conjunctural
explanations. Section II discusses the benefits of using fs/QCA to analyze
cross-country variation in the reliance on social pacts. Section III introduces
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
631
Avdagic
the main variables and data, whereas Sections IV and V discuss the main
steps of the analysis, model specifications, and findings. Section VI summarizes the findings and offers suggestions for future research.
Explaining the Reliance on Pacts:
Economics, Institutions, and Politics
Policy concertation is often depicted as an effective way to facilitate economic adjustment and sustainable reforms. Yet not all countries faced with
adjustment pressures of the 1990s saw the emergence of social pacts. Why
did governments pursue concerted agreements in some countries but not in
others? The literature emphasizes three sets of factors as the main reasons for
cross-national variation in adjustment strategies: economic problem load,
wage-setting institutions, and governments’ political calculations and electoral concerns.
Economic Problem Load
The predominant explanation of social pacts emphasizes the role of a crisis
or a high economic problem load. A general idea running through this literature is that an agreement on reforms is more likely when a country is stuck in
a deep crisis that threatens international competitiveness or when exogenous
shocks require adjustment across multiple policy areas. Such situations of
national emergency, so the argument goes, act as powerful incentives for concertation because a high problem load may help to create a shared understanding
that a cooperative game is the most appropriate course of action. Social pacts
are thus generally depicted as functional responses to various economic
problems.
Two sets of arguments figure prominently in this group of explanations.
The first one maintains that social pacts of the 1990s reflect the need of European economies to bring inflation and deficits down to the level imposed by
the EMU convergence criteria (Fajertag & Pochet, 2000; Hancke & Rhodes,
2005; Hassel, 2006). Hancke and Rhodes (2005), for example, argue that not
all countries saw the emergence of pacts because EMU-related pressures
were not distributed symmetrically. Only where inflation and budget deficit
were above the Maastricht limits (as defined in the appendix) did governments
rely on concertation “both to expedite the process of adjustment and reduce the
potential social costs of rapid disinflation” (Hancke & Rhodes, 2005, p. 201).
The second set of arguments emphasizes more general pressures associated with
internationalization and in particular the loss of competitiveness and (threat of)
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
632
Comparative Political Studies 43(5)
high unemployment as the key drivers of pacts (Culpepper, 2002, 2008;
Hyman, 1999; Rhodes, 2001). Rather than assuming a direct link between economic problems and concerted agreements, some of these accounts have
advanced sophisticated causal arguments that demonstrate how a crisis induces
cooperation by transforming actors’ understanding of feasible policy options
and ultimately their initial preferences (see Culpepper, 2008).
Taken together, this literature suggests two hypotheses about the emergence of pacts that stress either problems to meet the core EMU convergence
criteria or the loss of competitiveness as manifested in high unemployment.
Accordingly, a higher propensity to pacts should be evident in countries that
had encountered high inflation or government deficits (Hypothesis 1) or,
alternatively, high unemployment (Hypothesis 2).
Wage-Bargaining Institutions
The corporatist literature has long argued that centralized and monopolistic
interest organizations are the key institutional precondition for concertation.
Although this argument was strongly challenged during the 1990s when
concerted agreements emerged in a number of countries that lacked such
institutional preconditions, several contemporary accounts suggest that certain
institutional features are nonetheless important for they provide compelling
incentives for pacts. Unions’ organizational strength as expressed through
union density and the level of wage-bargaining centralization are the two
most emphasized factors. Whether or not the unions organize a sizeable proportion of the workforce is seen as relevant because governments are
unlikely to be interested in joint policy formulation when unions are organizationally too weak to be credible negotiating partners (Baccaro & Simoni,
2008). As far as wage bargaining centralization is concerned, there is no
consensus on what degree of centralization may be most conducive to pacts.
Some argue that high centralization is needed to convince the government
that unions would be able to hold up their side of the bargain (Harcourt &
Wood, 2003). Others, in contrast, maintain that an intermediate level of centralization creates stronger incentives for pacts (Ebbinghaus & Hassel, 2000;
Hassel, 2006). Drawing on the curvilinear hypothesis about the impact of
union centralization on wage demands (Calmfors & Driffil, 1988), these
authors argue that unions in highly centralized systems automatically internalize costs of excessive wage increases, which in turn would make a social
pact on wage moderation superfluous. The government should therefore be
more interested in gaining cooperation of moderately centralized unions,
which may have a substantial effect on nominal wages but are unlikely to
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
633
Avdagic
be equally concerned about employment costs of wage demands for their
members.
This group of explanations thus suggests three different hypotheses about
national variation in the reliance on pacts: Pacts should be more evident in
countries that have more encompassing unions (Hypothesis 3), higher levels
of union centralization (Hypothesis 4), or, alternatively, unions that are intermediately centralized (Hypothesis 5).
Governments’ Political Calculations
and Electoral Concerns
Governments’ willingness to negotiate reforms and sign social pacts has been
also linked to various political conditions. Three different conditions figure
most prominently in this group of explanations. First, the early literature on
the politics and effects of corporatism has identified Left party incumbency
as a key precondition for successful concertation (Alvarez, Garrett, & Lange,
1991; Hicks, 1999; Korpi, 1983). Left parties were thought to be particularly
well positioned to ensure union cooperation because of their traditionally strong
relationship with the unions. Unions were willing to accept wage restraint
because they could count on their long-standing partners to deliver on their
promises and ensure full employment and high investment. Although it is now
widely accepted that government partisanship was significantly more important in the context of a Keynesian economic policy than in the contemporary
monetarist context (Scharpf, 1991), a number of recent studies nonetheless
suggest that Left parties are still more likely to negotiate policies with unions
(Baccaro & Simoni, 2008; Hamann & Kelly, 2007b; Harcourt & Wood, 2003;
Hassel, 2006).
Second, several accounts argue that the emergence of pacts depends on
government strength and specifically on whether the governing party or coalition controls a majority of seats in the legislature (Avdagic, 2006; Baccaro &
Lim, 2007; Baccaro & Simoni, 2008). Although these accounts recognize
that multiparty governments may be less cohesive than single-party governments, their predominant focus is on the proportion of seats held by the
governing parties. The underlying logic here thus mirrors arguments on the
relationship between political institutions and the size of budget deficits,
which posit that it is “not negotiations within a government, as much as
negotiations in parliament that is the main obstacle” in undertaking reforms
(Edin & Ohlsson, 1991, p. 1602). Following this line of reasoning, Baccaro
and Lim (2007) argue most explicitly that minority and caretaker governments are more likely to craft policy agreements with unions and employers
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
634
Comparative Political Studies 43(5)
because their electoral weakness requires them to seek support for reforms
outside of parliaments.
Finally, it has been argued that social pacts are a consequence of broader
electoral pressures, manifested in fierce party competition, the emergence of
new parties, and declining partisan loyalties. Drawing on the literature on
party competition and changing voting behavior, Hamann and Kelly (2007a)
assume that governing parties are interested in winning votes and elections
and thus calculate the likely electoral gains of pacts versus unilateral imposition
of policies. Correspondingly, they argue that governments are more likely to
offer pacts when the governing parties are facing strong electoral pressures.
In sum, this literature suggests three plausible hypotheses: Pacts should be
evident in countries where governments have been on the whole electorally
weak (Hypothesis 6), dominated by Left parties (Hypothesis 7), or faced with
strong electoral pressures (Hypothesis 8).
Conjunctural and Multicausal Explanations
Although this distinction among the three groups of explanations is useful for
analytic purposes, it is important to recognize that many works referred to
above do not advance simple mono-causal explanations of pacts. Instead, they
often emphasize more than one causal condition and sometimes develop conjunctural explanations whereby the impact of a particular variable is conditioned
by the level of another variable. For example, Baccaro and Lim (2007) argue
that a combination of three conditions is needed for a pact to materialize: an
economic crisis, a minority government, and weakened but still sufficiently
powerful unions. Similarly, Hamann and Kelly (2007a) suggest that pacts
occur in situations where governments are facing not only economic problems
but also significant electoral pressures. Analyses of pacts that underline the
role of government partisanship draw on the literature on social democratic
corporatism (Alvarez et al., 1991; Hicks, 1999) and the power-resources perspective (Korpi, 1983), which suggest that negotiated compromises should
be more likely in countries that have both high Left cabinet incumbency and
organizationally strong unions.
Hassel (2006) also develops a multicausal argument that emphasizes not
only economic problems and wage-bargaining institutions but also the extent
to which monetary regimes are accommodating and the type of political systems. Following Iversen (1998), she argues that intermediate centralization
leads to inflationary pressures only when monetary policy is accommodating
because unions in restrictive monetary regimes will have a strong incentive to
restrain their wage demands for fear of employment losses. Simultaneously,
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
635
Avdagic
she draws on Lijphart’s (1999) argument that consensus democracies are
accustomed to negotiation because they have proportional electoral rules, multiparty systems, and a web of corporatist institutions guaranteeing representation
of all relevant interests in policy making. Accordingly, Hassel argues that
although unilateral reforms may be possible in majoritarian systems where
governments face fewer institutional constraints, negotiations with social partners (i.e., unions and employers) may be unavoidable in consensus democracies.
Her complex framework thus suggests that governments should be more likely
to pursue pacts in countries where economic problems are combined with
intermediate union centralization, insufficiently restrictive monetary regimes,
and a tradition of consensus democracy.
Finally, some earlier works also advance multicausal arguments that may
offer some insight about the rationale behind contemporary concertation.
Among these, Katzenstein’s (1985) argument about adjustment strategies
of small states appears to be particularly relevant because it focuses on the
link among competitive pressures, political institutions, and negotiated
adjustment. Although social pacts (and their absence) in the 1990s do not
seem to be strongly related to country size, a combination of two remaining
factors—a high degree of economic openness and consensus-based political
institutions—merits further consideration.
Evaluating the Hypotheses:
Why Use Fuzzy Sets?
The remainder of this article evaluates these hypotheses through a fuzzy set
analysis of 14 European countries during the 1990s. Fs/QCA is a technique
for systematic explorations of connections between hypothesized explanatory factors and outcomes (Ragin, 2000, 2008). Several reasons make this
technique particularly suitable for macro-comparative research in general
and for the analysis of social pacts in particular. First, explaining national
variation in adjustment strategies entails the well-known “few cases, many
variables” problem. Regression analysis is not appropriate in this case because
the relatively small N would render insufficiently reliable results and would
not permit inclusion of complex interaction terms to investigate multicausal
hypotheses. Fs/QCA, in contrast, was developed primarily for small- to mediumN analyses and for the exploration of causal configurations rather than net
effects of individual variables.
Second, the literature reveals different degrees of reliance on concerted
agreements. A distinction is made not only between “pact countries” and
“nonpact countries” but also within “pact countries” depending on the extent
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
636
Comparative Political Studies 43(5)
to which they relied on this mode of adjustment. In some of these countries
governments used social pacts extensively, linking adjustment in wage setting
with labor market and social security reforms. In others, pacts were either
primarily focused on nonwage issues, or alternatively governments avoided
tripartite negotiations, striving instead to induce central agreements between
unions and employers by threatening unilateral intervention. This variation
cannot be adequately captured by standard binary choice models or by the
older variant of QCA relying on Boolean sets, as these techniques require
dichotomization of the dependent variable. In contrast, fs/QCA permits cases
to have partial membership in both the outcome and causal conditions, as
captured by fuzzy membership scores in the interval between 0 (nonmembership) and 1 (full membership).Afuzzy set thus corresponds to a pseudo-continuous
scale on which cases are coded according to the degree to which they belong
to a specified category, for example, extensive concertation, encompassing
unions, high unemployment, and so on.
Third, many prominent arguments about social pacts have been framed in
terms of necessary and sufficient conditions for their emergence. Fs/QCA is
an obvious choice for assessing such arguments because this technique formalizes analysis of necessity and sufficiency by drawing on the logic of set
theory while simultaneously incorporating the assumption of probabilistic,
rather than veristic, causality. Finally, fs/QCA allows for equifinality, which
denotes a possibility of multiple pathways to an outcome. This idea is particularly relevant for our analysis because case studies suggest that different
combinations of conditions may have been responsible for the emergence of
pacts in different countries.
Data
Most causal conditions in this article are coded as six-value fuzzy sets with
membership scores 1, 0.8, 0.6, 0.4, 0.2, and 0. For reasons of empirical correspondence, the outcome is coded as a four-value fuzzy set with the matching
scores 1, 0.8, 0.2, and 0. Membership along the continuum is determined on the
basis of substantive and theoretical knowledge—a method considered preferable to simple mathematical rescaling (Ragin, 2008). Cases include all “old” EU
member states but Luxembourg, for which complete data were not available.
Outcome
The outcome of interest in this analysis is the reliance on social pacts as a
strategy of economic adjustment during the 1990s. Because the literature has
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
637
Avdagic
tried to primarily understand why pacts were signed in some countries but
not in others, the concern here is successful policy concertation, that is, concertation culminating in a social pact. For the purpose of this analysis, social
pacts are defined as publicly announced formal policy contracts between the
government and social partners over incomes, labor market, or welfare policies that explicitly identify policy issues and targets, means to achieve them,
and tasks and responsibilities of the signatories. This definition therefore
excludes purely declaratory agreements, general statements of intent, and
informal behind-the-scenes settlements that are sometimes seen as functional
equivalents of pacts. It also rules out pure bilateral agreements between unions
and employers, focusing attention instead on the role of the government. For
an agreement to be considered a pact, the government has to be a signatory,
or alternatively it has to publicly back up a union–employer agreement and
offer explicit policy incentives to the negotiating parties. I adopt such a
restrictive definition of pacts for two main reasons. First, considering all types
of agreements as pacts regardless of their formal status, substantive content,
and actors involved would artificially inflate the number of pacts. Put simply,
there would be hardly any cross-national variation left as some form of centralized agreement was evident in almost all EU countries during the 1990s.
Second, because several explanations of pacts emphasize political conditions
and governments’ electoral concerns, it would make little sense to test those
hypotheses on cases of union–employer agreements that were crafted without explicit government involvement.
The analysis focuses on the 1990s because most social pacts that brought
about far-reaching consensual reforms linking multiple policy areas were
crafted during the past decade. To be sure, some pacts were already signed in
the late 1980s, and pacting has continued in several countries beyond the
1990s. However, comprehensive tripartite pacts in the late 1980s were much
less frequent (the 1987 pact in Ireland being the most often cited example),
whereas pacts in the 2000s have been generally of a narrower scope (e.g., the
2003 wage agreement in the Netherlands, the 2001 agreements on training,
and health and safety in Portugal) and/or characterized by more serious implementation problems (e.g., the 2002 Pact for Italy). Focusing on the 1990s thus
ensures that the analysis considers pacts that are better comparable in terms of
their scope, content, and practice while simultaneously allowing a more precise assessment of countries’ fuzzy membership in specific categories than
would be possible in an analysis focusing on a longer period of time.
How widespread was the reliance on social pacts as a strategy of economic adjustment, and what form did such concerted deals take? Table 1
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
638
Comparative Political Studies 43(5)
Table 1. Social Pacts During the 1990s: Fuzzy-Set Membership in the Outcome
Countries
Characteristics
Italy, Ireland, Portugal, Finland
Spain
Netherlands, Belgium
Austria, Denmark, France,
Germany, Greece, Sweden,
United Kingdom
Broad tripartite pacts, linking
incomes policy with labor market
and welfare reforms
Narrower pacts, focus on labor
market and welfare policies
Union–employer pacts under threat
of government unilateralism
No major pacts
Fuzzy Score
1
0.8
0.2
0
summarizes the basic characteristics of pacts and classifies countries along a
4-point scale capturing their membership in the “extensive concertation” set.
The table indicates that during the 1990s governments in 7 out of 14 countries
used some form of concerted agreements. Out of the 7 countries where pacts
were evident, 5 are clearly positive cases, that is, cases where government was
directly involved in concertation and where the resulting agreements spanned
several policy areas. These cases are assigned fuzzy scores 1 or 0.8 depending
on the scope of agreements and the extent of government involvement.
Italy, Portugal, Ireland, and Finland are all given score 1, denoting their
full membership in the “extensive concertation” set. Social pacts in these
countries linked reforms across adjacent policy areas, and in all of them the
government played a central role. Italy saw the emergence of five such pacts
during the 1990s. Not only did these agreements help secure wage restraint,
but they also facilitated consensus over the abolishment of the wage indexation mechanism (scala mobile) and a reform of collective bargaining, a major
pension reform, and the creation of a range of new employment programs
(Baccaro & Lim, 2007; Regini & Colombo, 2006). Similarly, in Portugal,
there was a series of pacts covering pay-rise ceilings and levels of the minimum wage as well as reforms of labor legislation and working hours, social
security and pensions, the tax system, and employment policies and occupational training (Campos Lima & Naumann, 2005). Ireland adopted a practice
of 3-year tripartite agreements, started in 1987 with the Program for National
Recovery. Three successive pacts signed during the 1990s served to sustain
wage moderation in exchange for tax cuts while introducing a comprehensive
range of new welfare and labor market measures aimed to minimize social
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
639
Avdagic
inequalities and exclusion (Baccaro & Lim, 2007; O’Donnell, Adshead, &
Thomas, 2007). Like Ireland, Finland also relied on regular agreements linking incomes policy with institutional innovations in the areas of taxation,
social security, and employment policies. Although the government in Finland
was rarely a formal signatory of pacts, it not only endorsed these agreements
publicly but also participated in their negotiations, offering explicit incentives to the negotiating parties (e.g., tax relief, shortened working hours,
earnings-related unemployment benefits, and holiday bonuses) to ensure that
a consensus is reached (Kauppinen, 2000).
Spain is classified as having high albeit not full membership (0.8) in the
set of countries with extensive concertation. Its lower score reflects the narrower scope of its pacts and a more limited role of government in wage
determination. Here, the government tried to influence collective bargaining
only indirectly by offering social partners a seat at the bargaining table on
labor market and welfare reforms in exchange for responsible behavior in
wage setting. Consequently, Spain had no grand tripartite pacts linking multiple policy areas. Instead, concertation was pursued at separate tripartite
and bipartite tables focusing on narrower issues (Hancke & Rhodes, 2005).
This strategy resulted in five agreements between the government and social
partners over employment and social security (including pensions, part-time
contracts, and health and safety) and eight separate interconfederal agreements over wages and workplace industrial relations (Royo, 2006).
The Netherlands and Belgium also have nonzero membership in the outcome, but their score below the 0.5 threshold indicates that they cannot be
considered as fully positive cases. They are assigned fuzzy score of 0.2
because their “pacts” were in fact centralized union–employer agreements
crafted under explicit threats of government intervention. These agreements
are not “social pacts proper” because the government’s role in the negotiations was indirect and its commitment to concertation, and the willingness to
offer tangible concessions, clearly lower. Put simply, rather than carrots, these
governments used sticks to convince social partners to cooperate. If and when
the social partners refused to do so, the government did not try to buy their
cooperation but proceeded unilaterally by imposing wage ceilings through
legislation. This type of “concertation under the shadow of hierarchy” (Visser,
1999) was not new in these countries. The well-known Waasenar Agreement
(1982) in the Netherlands was also signed only after the government threatened to abolish wage indexation and freeze public and minimum wages and
main transfer programs (Ebbinghaus & Hassel, 2000). Subsequently, threats
of government intervention were crucial to ensure bipartite agreements on
wages and working-time flexibilization in 1994 and 1997. In Belgium, the
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
640
Comparative Political Studies 43(5)
failed negotiations of the Recovery Plan (1982) prompted the government
to freeze wages and suspend temporarily wage indexation, whereas the subsequent biannual interconfederal wage agreements were mostly dictated by
the government (Hassel, 2006). In the mid-1990s, the government tried to
convince the social partners to support a tripartite pact aimed to improve the
country’s declining competitiveness. When the unions refused to accept its
terms, the government again proceeded unilaterally, first by introducing a
wage freeze (1993) and then a new legal framework for wage bargaining
(1996), which limited wage increases to the levels of Belgium’s three main
trading partners (Baccaro & Lim, 2007). Consequently, the biannual agreements on wages and employment conditions concluded in the late 1990s
entailed a de facto threat of further government intervention.
The other seven countries are considered to be nonmembers in the set
“extensive tripartite concertation.” Some of them unsuccessfully tried to
reach comprehensive social pacts. The most notable example is the German
Alliance for Jobs (1996, 1998), a wide-ranging initiative that produced some
narrow agreements but ultimately failed to achieve a broad-based consensus
on labor market and welfare reforms for which its creators had hoped. Similarly, in Greece negotiations of the Confidence Pact (1997) failed to produce
substantial reforms, instead resulting in a rather general and largely declaratory agreement. The remaining countries did not try to reach explicit social
pacts. It is important to note though that they compose two qualitatively different groups. The first group includes countries with strong corporatist traditions:
Austria, Sweden, and Denmark. These countries have a range of long-established
participatory institutions, and social partners are often involved in day-to-day
governance of specific policy areas. In a sense, such developed corporatist
infrastructure may have made formal social pacts redundant. Indeed, governments and social partners in these countries were able to agree on some
wide-ranging reforms during the 1990s without “packaging” these instances
as formal social pacts (e.g., social security and welfare reforms in Austria in
1995 and 1996 and labor market reforms in Denmark in 1994, 1996, and
1998). The second group includes two noncorporatist countries where no
proper negotiations over major reforms took place: the United Kingdom and
France. In the former, the Conservative governments’ policies in the 1980s
and 1990s undermined union power and contributed to the decentralization
of bargaining structures. In the latter, weak and fractious unions staged large
protests in 1995 against the government’s proposal to reform the social security system but proved to lack the capacity to promote internal discussions
and convince their members to help reach a concerted agreement with the
government (see Culpepper, 2002).
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
641
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
0.8
1
0
0
1
0.2
1
0.2
1
0.8
1
1
1
1
MAAS
0
0.8
0.6
1
1
0.6
0.8
1
1
0.2
0.2
1
0.6
0.8
UNEM
Economics
0.6
0.6
0.8
0.8
0.2
0.4
0.4
0.6
0.4
0.4
0.4
0.2
1
0.4
DENS
1
0.8
0.6
0.6
0
0.8
0.6
0.8
0.4
0.8
0.4
0.4
0.8
0
CENT
Institutions
0
0
1
1
0
0
1
0
1
0
1
1
0
0
MEDC
Core Causal Conditions
0.8
0.8
0.8
0.4
0.8
0
0.8
0
0.4
0.6
0.6
0.8
1
0.2
LEFT
0
0
1
0
0.45
0
0
0.89
0.77
0
0.63
0.89
1
0
MINGa
Politics
0.6
1
1
1
0.6
0.6
0.2
0.6
1
1
0.4
0.6
0.8
0.2
EFFP
0.6
0.8
1
0.8
0
0.6
0.2
0.6
0.8
1
0.8
0.2
0.8
0
CONS
0
0.6
0.6
0.8
0.6
0
0.4
0.8
0.6
0.2
0.8
0.6
0.6
0.6
CBDP
0.8
1
0.8
0.6
0.4
0.6
0.4
1
0.4
1
0.6
0.4
0.6
0.4
OPEN
Additional Conditions
Note: See the appendix for variable definitions.
a. Because few countries had predominantly minority governments, this variable was recoded by calculating the square root of the original scores. This
operation is called dilation (Ragin, 2000, p. 178), and it is commonly used to set a lower standard for membership without affecting the score of full
members and nonmembers. In this case, the verbal qualifier very weak governments is replaced by predominantly weak governments.
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
United Kingdom
Country
Table 2. Fuzzy Membership in the Causal Conditions
642
Comparative Political Studies 43(5)
Causal Conditions
Taken together, the three main groups of explanations of pacts marshal eight
potentially relevant causal conditions, whereas the related political economy
literature suggests three additional variables that merit further exploration.
Data for 14 EU countries were compiled from various sources to evaluate
these explanations. Raw data for all causal conditions were recoded as sixvalue fuzzy sets apart from intermediate centralization, which was coded as
a dichotomous set. Table 2 reports fuzzy scores for all variables. The criteria
for calibration for all variables and data sources are described in more detail
in the appendix.
Measures of the economic problem load, emphasized by the first set of
explanations, include inflation, government deficit, and unemployment.
Three-year moving averages were used to determine the degrees of countries’
membership in the set of “high unemployment” (UNEM). Because average
figures for inflation and deficit would mask the extent of the problem load
that several countries were facing at some points, I created indices of EMUrelated problems that take into consideration both the number of years in
which deficit and inflation were above the Maastricht limits and the extent of
this imbalance in each year (see the appendix). The resulting sets were then
combined by using the “logical OR” function to create the macro variable
“Maastricht imbalance” (MAAS). Accordingly, a membership score in this
set indicates the degree to which a country was experiencing either considerable inflation or deficit problems during this period.
Measures used to evaluate explanations emphasizing institutional factors
include union density and the centralization index, which depicts union concentration and authority in wage bargaining at multiple levels. Because these
data did not vary greatly over the 1990s, decade means were used in the calibration process. Union density was used to create the set of “encompassing
unions” (DENS), whereas the centralization index was used to construct two
separate sets: “high union centralization” (CENT) and “intermediate union
centralization” (MEDC). To assess hypotheses that emphasize political conditions, I created sets “high Left cabinet incumbency” (LEFT), “electorally
weak governments” (MING), and “high political competition” (EFFP). The
degree of Left governance is measured by an index of cumulative Left cabinet incumbency, which corresponds to the sum of Left seats as a percentage
of parliamentary seats held by all government parties during the 1990s. As a
measure of electorally weak governments, I use the number of years in which
minority or caretaker governments were in power during this period. The
average effective number of parties serves as a proxy of competition and
electoral pressures.
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
643
Avdagic
Finally, to evaluate multicausal hypotheses from the related political economy
literature, I constructed three additional fuzzy sets: “insufficiently restrictive
monetary regime” (CBDP), “strong tradition of consensus democracy” (CONS),
and “high economic openness” (OPEN). Indicators used in the construction
of these sets include (a) an index of central bank independence as a proxy for
the degree of monetary accommodation, (b) a measure of a tradition of consensus democracy combining the index of proportionality of electoral systems with
the number of effective parties over a longer period of time, and (c) the share
of exports and imports in GDP as a proxy of economic openness.
Analysis
A standard fs/QCA protocol looks as follows. Like in regression analysis, the
first step is to specify the model, that is, to decide which hypothesized causal
conditions should be included in the analysis. Once the model is specified, fs/
QCA software generates the so-called truth table, which lists all logically
possible combinations of causal conditions and shows the correspondence of
cases to those configurations. At the same time, the truth table displays a
series of consistency scores which can range from 0 to 1 and which provide
information on the degree to which each causal configuration is consistently
associated with the outcome. The researcher then determines a particular consistency threshold, effectively instructing the program which configurations
should be considered as reasonable subsets of the outcome. Configurations
below this threshold are then discarded from the analysis. The second step
entails a decision on whether the analysis should include only empirically obse­
rvable configurations or also counterfactual configurations that lack empirical
instances. The former choice favors empirical complexity, whereas the latter
favors parsimony. Once these decisions are made, the program uses the set–
subset logic to simplify data patterns and identify configurations that can be
considered sufficient for the occurrence of the outcome. The ultimate product
of such analysis is a simplified solution set containing specific causal pathways
and several measures that indicate their “goodness of fit” (as explained below).
In this analysis, I used the software program fs/QCA 2.0 to run multiple,
predominantly four- or five-variable, models containing various combinations of causal conditions. The decision to explore multiple models, each
containing a modest number of conditions, was motivated by three reasons.
First, the inclusion of additional variables in a model would make each case
increasingly unique, yielding largely idiosyncratic solutions that are too complex and difficult to interpret. Second, adding additional variables exponentially
increases the number of logically possible causal combinations. Simultaneously
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
644
Comparative Political Studies 43(5)
considering the eight core causal conditions would yield 28 = 256 possible
configurations, out of which a maximum of 14 could be empirically observable. The resulting truth table would thus contain at least 242 rows displaying
configurations that lack empirical instances. The upshot of this is that a
parsimonious solution, which incorporates all these so-called logical remainders, would be based largely on “cases that could have been” rather than those
that are empirically observable. Although counterfactual analysis is a great
strength of QCA, recent experimental research has shown that a moderate
ratio of variables to cases is required to ensure the validity of results because
the risk of QCA finding solutions even on random data increases with the
addition of new conditions (Marx, 2006). The recommendation of this
research for analyses where N = 14 is to employ models with four explanatory variables, which should ensure highly valid results. Five-variable models
entail a somewhat higher risk of error, whereas models with six and more
variables should be avoided in this case (Marx, 2006, p. 19). Finally, specifying such multiple models allows us to compare the explanatory power of
different arguments, whereas adding and replacing variables in core models
serves to assess the robustness of solutions and minimize the risk of an omitted variable bias.
Correspondingly, this analysis entailed three steps to compare the broadest range of solutions while ensuring valid results. I first specified models
that approximate prominent conjunctural arguments, some of which include
variables beyond the core eight conditions emphasized by the literature on
social pacts. Next, the core variables were alternated systematically so as to
generate various theoretically plausible four-variable models. The final step
involved a test of several five-variables models. All models were run with and
without simplifying assumptions about configurations that lack empirical
instances so as to generate complex and intermediate solutions. The former
are based only on empirically observable configurations, whereas the latter
incorporate the so-called “easy counterfactuals” and are recommended as a
good bridge between complexity and parsimony (Ragin, 2008).
The assessment of causal sufficiency was based on a consistency threshold of 0.8—a measure indicating the extent to which membership in the
outcome is consistently greater than or equal to membership in a causal configuration. Effectively, this choice denotes that any combination of economic,
political, and institutional conditions that reaches or exceeds this threshold
can be considered “almost always” sufficient for extensive concerted agreements. As is customary in small-N analyses, a higher consistency benchmark
of 0.9 was used in the test of causal necessity. This test was performed prior
to specifying the different models and conducting sufficiency tests to
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
645
Avdagic
establish whether any of the individual causal conditions consistently appear
in the cases displaying the outcome.
Findings
The test of necessity revealed that none of the identified causal conditions
appears to be necessary for the outcome. High unemployment has the highest
consistency score of 0.84, but even this condition falls short of reaching the
required threshold for causal necessity (0.9). Interestingly, high inflation and
deficit—conditions often referred to as necessary for social pacts during the
1990s—display far lower consistency thresholds of 0.5 and 0.65, respectively. This result suggests that Maastricht-related problems alone may have
been in fact a less important driver of social pacts than it is often argued. The fact
that no individual condition turned out to be necessary suggests that the resu­
rgence of policy concertation is a phenomenon that is both complex and
profoundly influenced by context-specific conditions. In light of these results,
any search for a single cause or even a common impetus for pacts would
appear to be a futile exercise. The next step is therefore to search for different
combinations of conditions that may be linked to the outcome in terms of
causal sufficiency. For each model specification, the program identifies all
possible causal combinations and then uses set–subset logic to analyze combinations with consistency scores of at least 0.8. This process of “logical
minimization” yields simplified solution sets with the associated measures of
consistency and coverage, which are used to compare the explanatory power
of the models. Consistency indicates how well a given solution set explains
the outcome in question, whereas coverage indicates its empirical relevance,
that is, the proportion of membership in the outcome explained by the solution. As such, solution coverage bears some resemblance to the R2 (coefficient
of determination) in regression analysis. Together, these measures provide an
indication of the overall goodness of fit of solutions.
In this analysis, the test of sufficiency was first conducted on models that
approximate some prominent arguments about concertation and social pacts.
Five of these models turned out to be poor explanations of cross-national variation in the reliance on social pacts as none of their causal combinations
passed the required consistency threshold. These include (a) a model focusing on
the economic problem load, combining inflation, deficit, and/or unemployment
(Fajertag & Pochet, 2000; Hancke & Rhodes, 2005); (b) a model combining
high economic problem load with union centralization (Harcourt & Wood,
2003); (c) the power-resources model, combining high Left cabinet incumbency with encompassing and centralized unions and different measures of
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
646
Comparative Political Studies 43(5)
economic problems; (d) the classic corporatist model, focusing on institutional
conditions (union encompassingness and centralization); and (e) a model combining measures of economic openness and a tradition of consensus democracy
with various economic variables (Katzenstein, 1985). The remaining models
in this group perform better, displaying generally high consistency scores but
insufficiently high coverage. In simple terms, this means that they explain the
outcome well, but their explanation is relevant only in a small portion of
cases. This finding is not surprising because most of these arguments were
derived on the basis of in-depth qualitative analyses involving a few country
cases. These models include (a) a model combining Maastricht-related problems, intermediate union centralization, insufficiently independent central
bank, and a tradition of consensus democracy (Hassel, 2006); (b) a model
combining economic problem load with measures of government and union
strength (Avdagic, 2006; Baccaro & Lim, 2007); and (c) a model combining
Maastricht-related problems with various political variables including measures of government strength, political competition, and government’s
political orientation (Hamann & Kelly, 2007a, 2007b). All of these models
bring about solution sets with consistency scores exceeding 0.8, but their
coverage is rather low, ranging from 0.27 (the Hamann and Kelly model) to
0.46 (the Hassel model).
The next two rounds of sufficiency tests were conducted on a range of
four- and five-variable models including different combinations of the core
eight conditions. Table 3 presents the top three models in terms of their
explanatory potential. Both complex and intermediate solutions are reported.
As evident, all three models include Maastricht-related problems, electorally
weak governments and intermediate centralization. In addition, Model 1 incl­
udes high unemployment, Model 2 includes strong political competition, and
Model 3 includes all of these variables. Although all models perform well,
Model 1 is clearly the best specification in terms of the overall consistency
(0.98) and coverage (0.77). A further look reveals that this solution set contains three distinct paths or causal configurations that produce the outcome.
The last three columns in the table display information on consistency and
coverage for each of these paths. Raw coverage indicates the proportion of
all cases covered by a particular causal configuration. Unique coverage, on
the other hand, is in some ways similar to the partial regression coefficient,
and it denotes the proportion of cases explained exclusively by a given causal
configuration, that is, net of overlaps with other solutions. Such overlaps are
possible because fs/QCA permits cases to have different degrees of membership in given sets, which means that partially they could be explained by
more than one causal configuration. In this case, however, raw coverage and
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
647
Avdagic
Table 3. Best Models
Raw
Coverage
Unique
Coverage
Model 1
SOCP = f{MAAS, UNEM, MEDC, MING}
0.423077
0.423077
MAAS*MEDC*MING+
maas*UNEM*MEDC*ming+
0.192308
0.192308
maas*UNEM*medc*MING
0.153846
0.153846
Solution coverage: 0.769231
Solution consistency: 0.977995
Model 2
SOCP = f{MAAS, MEDC, MING, EFFP}
Complex solution
MAAS*MEDC*MING+
0.423077
0.423077
maas*medc*MING*EFFP+
0.115385
0.115385
maas*MEDC*ming*EFFP
0.192308
0.192308
Solution coverage: 0.730769
Solution consistency: 0.976864
Intermediate solution
ming*MEDC*EFFP+
0.328846
0.192308
MAAS*MEDC*MING+
0.423077
0.286538
maas*medc*MING*EFFP
0.115385
0.115385
Solution coverage: 0.730769
Solution consistency: 0.929095
Model 3
SOCP = f{MAAS, UNEM, MEDC, MING, EFFP}
Complex solution
maas*UNEM*medc*MING*EFFP+
0.115385
0.115385
MAAS*unem*MEDC*MING*effp+
0.115385
0.076923
maas*UNEM*MEDC*ming*EFFP+
0.192308
0.192308
MAAS*UNEM*MEDC*MING*EFFP
0.301923
0.263461
Solution coverage: 0.686538
Solution consistency: 1.000000
Intermediate solution
MAAS*MEDC*MING+
0.423077
0.319231
UNEM*MEDC*ming*EFFP+
0.296154
0.192308
maas*UNEM*medc*MING*EFFP
0.115385
0.115385
Solution coverage: 0.730769
Solution consistency: 0.929095
Consistency
0.960699
1.000000
1.000000
0.960699
1.000000
1.000000
0.895288
0.960699
1.000000
1.000000
1.000000
1.000000
1.000000
0.960699
0.885057
1.000000
Note: See the appendix for variable definitions. For Model 1, complex and intermediate
solutions are identical.
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
648
Comparative Political Studies 43(5)
unique coverage figures are the same, which suggests that there are no significant overlaps between the paths.
The results show that the first path explains the highest portion of the outcome, as indicated by its coverage of 0.42. This path combines high inflation
or deficit with a high proportion of minority governments and intermediate
union centralization. The other two paths, which jointly explain roughly 35%
of the outcome, reflect cases where unemployment, rather than inflation or
deficit, was the main economic problem. These two solutions suggest that
pacts are a viable strategy when high unemployment is combined with either
intermediate centralization or a high proportion of minority governments.
Theoretically, these three causal configurations provide plausible explanations of governments’ reliance on pacts during the 1990s. Although all
countries experiencing high inflation or deficits had to undertake substantial
reforms in the run up to EMU to meet the Maastricht criteria, those that had
predominantly minority governments were more likely to do so through
formal agreements with social partners. Because of their electoral weakness,
minority governments were compelled to seek extraparliamentary support
for reforms, and their incentives to do so would have been particularly strong
in cases where unions are neither strongly centralized nor decentralized. In
the former case, wage moderation was likely to be more or less automatic
because unions in such systems internalize the costs of their wage demands,
whereas in the latter systems unions would likely be too weak to affect aggregate wage developments. For similar reasons, it is understandable that minority
governments would have been inclined to pursue pacts in cases where unemployment was the main economic problem. Moreover, because unemployment
has a particularly strong impact on voters’ decisions and is always politically
costly, even nonminority governments could be expected to pursue concerted
solutions in cases of intermediate centralization. This is so because in such
systems unemployment may be perceived to be a consequence of intermediate centralization, which encourages wage hikes that may generate further
cuts in employment.
These three causal paths also seem to offer reasonable explanations for
our empirical cases. Path I, which combines high inflation or deficit with
weak governments and intermediate centralization, explains the cases of
Italy, Portugal, and Spain. During this period, all three countries struggled
with inflation and deficit problems and had intermediately centralized unions
and predominantly weak governments. This configuration captures perfectly
the case of Italy, which for the most part of the 1990s was governed by electorally weak and unstable governments burdened by several corruption scandals
(Baccaro & Lim, 2007; Regini & Colombo, 2006). These governments pursued
concertation because solving the inflation and deficit problems required
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
649
Avdagic
comprehensive reforms that they were too weak to design and impose unilaterally. Governments in Spain faced similar incentives and were unable to
solve Maastricht-related problems unilaterally. This was especially evident
after 1993 when the socialist PSOE government lost its majority and was
replaced in the next elections by the conservative Popular Party minority
government, which was also burdened by the need to prove its democratic
credentials (Hamann & Kelly, 2007a). The resurgence of concertation coincided with these developments as most pacts were signed in the mid-1990s.
A comparable scenario is evident in Portugal from 1995 when the PSD
majority government (1987–1995) was replaced by the PS minority government led by Prime Minister Guterres (1995–2001) who made concertation
an explicit part of government strategy of adjustment (Campos Lima &
Naumann, 2005). Path II, which combines high unemployment and predominantly minority governments, explains the case of Ireland. Although the first in
the series of regular pacts was signed already in 1987 under the minority Fianna
Fáil government, Irish social partnership was far from being solidly institutionalized prior to the late 1990s. Although the key Maastricht indicators had been
brought under control, unemployment remained stubbornly high (in the range
of 10% to 15%) for much of the 1990s. In this context, the continuation of
concertation was a rational strategy for internally divided multiparty coalition
governments, which apart from 1993 to 1996 had no majority in the Dáil.
Finally, Path III, combining high unemployment with intermediate centralization, corresponds to the case of Finland. The collapse of the Soviet
market drove the Finish economy into a deep recession, causing unemployment to increase rapidly from 3% in 1990 to around 17% in 1994, and it was
not before 2000 that unemployment finally fell below 10%. Simultaneously,
nonbinding centralized wage agreements became increasingly diluted,
whereas the growing tendency to determine wages at the industry and even
company levels produced a non-negligible wage drift, accounting for approximately 35% of wage growth (Uusitalo & Vartiainen, 2007). This context of
high unemployment and more frequent breakdowns in confederal bargaining
persuaded the Lipponen government to revive tripartite concertation despite
it being a stable and largely cohesive surplus coalition. Indeed, although this
so-called “rainbow coalition” comprised five parties representing different
sides of the political spectrum, the parties generally agreed on main policy
goals (albeit for different reasons; see Jungar, 2002). Moreover, the SDP and
the NCP—the two main parties that were also the principal advocates of a
comprehensive program of reforms—commanded a sufficient majority in
parliament and thus would unlikely have had problems securing legislative
approval for this program even without the other coalition partners. However, the government considered that a negotiated compromise with the
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
650
Comparative Political Studies 43(5)
unions was essential for successful implementation of the reforms and consequently emphasized a need for political unity and joint actions to fight
unemployment and restore the health of the economy (Jungar, 2002). Given
the government’s insistence on concertation and its resoluteness to obtain
union support, we could speculate whether the government may have in fact
felt vulnerable despite its electoral strength. Ultimately, however, this question can be answered only through a more in-depth case study analysis.
Overall, the presence of minority governments appears to have been a particularly important factor for successful concertation. Indeed, this was evident
in four out of our five cases of extensive pacting. A reconsideration of negative
cases also confirms the importance of this variable. The data show that where
economic problems were evident but governments were not electorally weak,
they either tried to correct imbalances unilaterally or alternatively threatened
unilateral intervention to convince social partners to craft appropriate bipartite
agreements. For example, the case of Greece, where Maastricht-related problems were particularly pronounced, suggests that an uninterrupted rule of
single-party majority governments may have acted as a deterrent to pursuing
tangible concerted agreements. The cases of Belgium and the Netherlands also
support the argument about the importance of government strength. Although
inflation was under control in both countries, they encountered non-negligible
problems with either deficit and debt or unemployment at various points
during the 1990s. Rather than investing heavily in tripartite concertation, governments in these countries tried to indirectly induce bipartite cooperation
between unions and employers. Occasional half-hearted attempts at broad tripartite agreements were supplanted by explicit threats of unilateral government
action if social partners failed to agree on an acceptable course of adjustment.
In both countries, the credibility of these threats was enhanced not only by the
history of government intervention but also by the fact that all governing coalitions during this period were able to secure a majority in the parliament. In the
Netherlands, the Lubbers government (1989–1994) controlled almost 70% of
seats in the parliament, and the subsequent Kok cabinets (1994–2002) held
60% to 65% of the seats. Similarly, the Dehaene government in Belgium was
a grand coalition of the Christian Democratic and Socialist parties that managed to secure an impressive 80% of parliamentary seats in the first part of the
1990s and a comfortable majority thereafter.
Conclusion
What explains national variation in the reliance on social pacts as a strategy
of economic adjustment in the 1990s? The findings of this analysis indicate
that dominant arguments, which emphasize the extent of economic problems
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
651
Avdagic
associated with the run up to EMU or more general economic pressures,
explain only one part of the story. High inflation and deficit or unemployment
turned out to be neither necessary nor in themselves sufficient for the extensive reliance on social pacts. Rather, a high economic problem load appears to
be causally relevant only when combined with particular political and institutional conditions, namely, the prevalence of electorally weak governments
and/or an intermediate level of union centralization. That the acuteness of
economic problems by itself does not guarantee pacts is understandable
because governments could be expected to pursue concertation only when
they expect the benefits of concerted agreements to outweigh their costs.
Indeed, if bargaining entailed no cost for the government, governments undertaking adjustment would always prefer pacts over unilateralism because such
agreements signal a broader support for reforms. Although the social pacts of
the 1990s included no major side payments reminiscent of the Keynesian-era
corporatist deals, they nonetheless entailed some non-negligible policy concessions (e.g., selective tax benefits, more gradual pension reforms, etc.) and
procedural gains for the unions. This analysis indicates that governments are
willing to pay such costs only when they believe that concerted agreements
would be politically beneficial and help them achieve their underlying goals.
Apart from specific economic pressures, government’s electoral strength and
expectations about the ability of wage-setting institutions to generate desirable behavior appear to be especially important in such calculations.
These findings are broadly consistent with the general thrust of some
recent accounts that question the sufficiency of economic pressures as the
underlying cause of social pacts (Avdagic, Rhodes, & Visser, 2005; Baccaro
& Lim, 2007; Hamann & Kelly, 2007a). The analysis presented here makes
two distinct contributions to this literature. First, it evaluates and compares
the explanatory power of different hypotheses on a larger number of country
cases while simultaneously generating more precise assessments of necessary and sufficient conditions for pacts. Given the consideration of a wide
range of variables and the multitude of model specifications, the resulting
causal configurations are less likely to reflect spurious relationships produced by omitted variables. Second, the fuzzy-set analysis presented here
extends and refines existing multicausal explanations of pacts by demonstrating that there is more than one causal pathway to concerted agreements. In
examining the conjunction of multiple causal factors, this analysis has identified three distinct, theoretically and empirically relevant combinations of
conditions that helped generate pacts during the 1990s. In countries where
EMU-related pressures were particularly pronounced, a combination of weak
governments and intermediately centralized unions was required for the
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
652
Comparative Political Studies 43(5)
adoption of pacts as a strategy of adjustment. Where unemployment, rather
than the Maastricht imbalance, was the main economic problem, either electorally weak governments or intermediate union centralization was sufficient
for concerted reforms. This study therefore bridges the gap between in-depth
single-case studies of social pacts that emphasize context specificity and the
uniqueness of cases and comparative accounts that search for general explanations applicable to all countries. Although the analysis highlights more
than one causal pathway in accounting for the resurgence of concertation, it
also reveals that these explanations are not idiosyncratic and that there is only
a limited number of relevant causal patterns.
To be sure, this analysis is limited. Although it considers most West European countries, it focuses on a relatively short period of time to control for
shifts in the broader economic context. Although such cross-national analysis
was necessary to assess some leading hypotheses about social pacts, countries may be too crude units of analysis even if we focus on a single decade
when they experienced broadly similar pressures, such as those associated
with preparations for EMU. A promising avenue for further research would
therefore be to move away from the macro level of analysis and to examine
conditions for the emergence of different types of pacts during a longer
period of time. Using individual pacts, rather than countries, as units of analysis would enable us to explore whether the logic of concerted agreements
has changed over time. The analysis presented here could serve as a basis for
such an undertaking. Because such an analysis would entail a considerably
larger N, fs/QCA could be fruitfully combined with regression analysis to
examine both configurational and tendential relationships.
Appendix
Variable Descriptions and Data Sources
Maastricht imbalance (MAAS): A macro variable joining sets “high
deficit” and “high inflation” by “logical OR.” The sets are based on
cumulative indices for 1990–1999, calculated as follows. For deficit, score 1 is assigned for each year when the deficit is above the
Maastricht limit of 3% of GDP but lower than 4%, score 2 when it is
between 4% and 5%, and score 3 when it exceeds 5%. For inflation,
score 1 is assigned when inflation exceeds the Maastricht limit (1.5
percentage points above three lowest inflation member states) by
no more than 2.5 percentage points, score 2 by up to 5 points, and
(continued)
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
653
Avdagic
Appendix (continued)
score 3 by more than 5 points. For both sets, at least 10 on the index
is required for full membership, at least 5 to be more in than out
of the set, and a maximum of 1 for full nonmembership. Source:
Organisation for Economic Co-operation and Development (OECD)
Economic Outlook (2007) and International Monetary Fund World
Economic Outlook data set (2007).
High unemployment (UNEM): Three-year moving averages. Doubledigit figures correspond to full membership in this set; full nonmembership is below 4%; a cross-over point is 7%. Source: OECD
Main Economic Indicators (2007).
Encompassing unions (DENS): Average net union membership as a
proportion of wage and salary earners in employment during the
1990s. At least 80% is required for full membership, less than 5%
indicates full nonmembership, 40% is a cross-over point. Source:
Jelle Visser, ICTWSS data set, University of Amsterdam (2008).
High union centralization (CENT): Iversen’s centralization index as
updated by Visser (2008), averaged over the 1990s. In theory, the
index ranges from 0 to 1, but only Austria reaches figures above 0.6.
France and the United Kingdom are at the bottom of the scale with
figures below 0.3. Accordingly, at least 0.6 is needed for full membership, figures below 0.3 indicate full nonmembership, and 0.4 is a
cross-over point. Source: Jelle Visser, ICTWSS data set, University
of Amsterdam, 2008.
Intermediate union centralization (MEDC): A dichotomous set based on
the same data as CENT. Figures between 0.35 and 0.45 reflect membership in this set (J. Visser, personal communication, April 9, 2008).
All other figures reflect nonmembership.
High Left cabinet incumbency (LEFT): Cumulative index of Left cabinet incumbency, 1990–1999. The maximum value in any given year
is 1. At least 8 on this index is required for full membership, 4 is a
cross-over point, and values below 1 reflect full nonmembership.
Source: Armingeon et al., Comparative Political Dataset 1960–
2005, Institute of Political Science, University of Berne (2008).
Electorally weak governments (MING): Number of years in which minority or caretaker governments were in power during the 1990s. At least
9 years of minority governments reflects full membership in this set,
5 is a cross-over point, and no minority governments reflects full nonmembership. Source: Armingeon et al., Comparative Political Dataset
1960–2005, Institute of Political Science, University of Berne (2008).
(continued)
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
654
Comparative Political Studies 43(5)
Appendix (continued)
High political competition (EFFP): The Laakso–Taagepera index
reflecting both the number of parties and their relative weight. At
least 4.5 is required for full membership, a maximum of 2 for full
nonmembership, and 3 is a cross-over point. Source: Armingeon
et al., Comparative Political Dataset 1960–2005, Institute of Political Science, University of Berne (2008).
Insufficiently restrictive monetary regime (CBDP): An index combining four well-known indices of central bank independence. It
ranges from 1 to 3, with higher values reflecting a lower degree
of independence. On this index, 3 reflects full membership, 1 full
nonmembership, and 2 is a cross-over point. Source: Armingeon
et al., Comparative Political Dataset 1960–2005, Institute of Political Science, University of Berne (2008).
Strong tradition of consensus democracy (CONS): A macro variable
joining sets “proportional electoral systems” and “multiparty systems” (1971–1996) by “logical AND.” The first set is based on the
Gallagher disproportionality index, where lower values indicate
more proportional systems. Values below 2 correspond to full membership, values above 10 to full nonmembership, and 6 is a crossover point. The second set is based on the Laakso–Taagepera index
as in EFFP above. Source: Armingeon et al., Comparative Political Dataset 1960–2005, Institute of Political Science, University of
Berne (2008).
High economic openness (OPEN): Sum of exports and imports as percentage of GDP averaged over the 1990s. At least 100 on this index
is needed for full membership, 50 is a cross-over point, and less than
10 reflects full nonmembership. Source: Penn World Table, Version
6.2, University of Pennsylvania (2006).
Acknowledgments
The author would like to thank Claudius Wagemann, Martin Rhodes, and three anonymous reviewers for helpful comments on earlier drafts of this article.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the authorship
and/or publication of this article.
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
655
Avdagic
Financial Disclosure/Funding
The author disclosed receipt of the following financial support for the research and/or
authorship of this article: Research for this article was undertaken in the framework
of the Integrated Project on New Modes of Governance, coordinated by the European
University Institute in Florence and supported by the European Commission’s 6th
Framework Program (Project CIT1-CT-2004-506392).
References
Alvarez, M., Garrett, G., & Lange, P. (1991). Government partisanship, labor organization, and macroeconomic performance. American Political Science Review,
85(2), 539-556.
Avdagic, S. (2006). The emergence of pacts: Framework evaluation (NEWGOV
project report). Florence, Italy: European University Institute.
Avdagic, S., Rhodes, M., & Visser, J. (2005). The emergence and evolution of social
pacts: A provisional framework for comparative analysis (European Governance
Papers N-05-01). Retrieved August 11, 2008, from www.connex-network.org
Baccaro, L., & Lim, S.-H. (2007). Social pacts as coalitions of the weak and moderate:
Ireland, Italy and South Korea in comparative perspective. European Journal of
Industrial Relations, 13(1), 27-46.
Baccaro, L., & Simoni, M. (2008). Policy concertation in Europe: Understanding
government choice. Comparative Political Studies, 41, 1323-1348.
Calmfors, L., & Driffil, J. (1988). Bargaining structure, corporatism and macroeconomic
performance. Economic Policy, 6, 14-47.
Campos Lima, M. D. P., & Naumann, R. (2005). The emergence and evolution of
social pacts in Portugal (NEWGOV project report). Florence, Italy: European
University Institute.
Culpepper, P. (2002). Powering, puzzling, and “pacting”: The informational logic of
negotiated reforms. Journal of European Public Policy, 9(5), 774-790.
Culpepper, P. (2008). The politics of common knowledge: Ideas and institutional
change in wage bargaining. International Organization, 62, 1-33.
Ebbinghaus, B., & Hassel, A. (2000). Striking deals: Concertation in the reform of continental European welfare states. Journal of European Public Policy, 7(1), 44-62.
Edin, P.-A., & Ohlsson, H. (1991). Political determinants of budget deficits: Coalition
effects versus minority effects. European Economic Review, 35(8), 1597-1603.
Fajertag, G., & Pochet, P. (2000). A new era for social pacts in Europe. In G. Fajertag
& P. Pochet (Eds.), Social pacts in Europe—New dynamics (pp. 9-25). Brussels,
Belgium: European Trade Union Institute.
Hamann, K., & Kelly, J. (2007a). Party politics and the reemergence of social pacts in
Western Europe. Comparative Political Studies, 40, 971-994.
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
656
Comparative Political Studies 43(5)
Hamann, K., & Kelly, J. (2007b, May). Social pacts in Western Europe: The roles
of European integration and electoral competition. Paper presented at the EUSA
meeting, Montreal, Canada.
Hancke, B. (2002). The political economy of wage-setting in the Eurozone. In P. Pochet
(Ed.), Wage Policy in the Eurozone. Brussels, Belgium: Peter Lang Publishing.
Hancke, B., & Rhodes, M. (2005). EMU and labor market institutions in Europe:
The rise and fall of national social pacts. Work and Occupations, 32(2), 196-228.
Harcourt, M., & Wood, G. (2003). Under what circumstances do social accords work?
Journal of Economic Issues, 37(3), 747-767.
Hassel, A. (2006). Wage setting, social pacts and the euro. Amsterdam: Amsterdam
University Press.
Hicks, A. (1999). Social democracy and welfare capitalism: A century of income
security politics. Ithaca, NY: Cornell University Press.
Hyman, R. (1999). National industrial relations systems and transnational challenges:
An essay in review. European Journal of Industrial Relations, 5(1), 89-110.
Iversen, T. (1998). Wage bargaining, central bank independence, and the real effects
of money. International Organization, 52, 469-504.
Jungar, A.-C. (2002). A case of a surplus majority government: The Finish rainbow
coalition. Scandinavian Political Studies, 25(1), 57-83.
Katzenstein, P. (1985). Small states in world markets: Industrial policy in Europe.
Ithaca, NY: Cornell University Press.
Kauppinen, T. (2000). Social pacts in Finland. In G. Fajertag & P. Pochet (Eds.),
Social pacts in Europe—New dynamics (pp. 161-185). Brussels, Belgium: European Trade Union Institute.
Korpi, W. (1983). The democratic class struggle. London: Routledge.
Laakso, M., & Taagepera, R. (1979). Effective number of parties: A measure with
application to West Europe. Comparative Political Studies, 12, 3-27.
Lijphart, A. (1999). Patterns of democracy: Government forms and performance in
thirty-six countries. New Haven, CT: Yale University Press.
Marx, A. (2006). Towards more robust model specification in QCA: Results from a
methodological experiment (Compasss working paper). Retrieved May 7, 2008,
from www.compasss.org
O’Donnell, R., Adshead, L., & Thomas, D. (2007). The emergence and evolution
of social pacts: Ireland. (NEWGOV project report). Florence, Italy: European
University Institute.
Ragin, C. (2000). Fuzzy-set social science. Chicago: University of Chicago Press.
Ragin, C. (2008). Redesigning social inquiry: Fuzzy sets and beyond. Chicago: University
of Chicago Press.
Regini, M. (1997). Still engaging in corporatism? Recent Italian experience in comparative perspective. European Journal of Industrial Relations, 3(3), 259-278.
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013
657
Avdagic
Regini, M., & Colombo, S. (2006). The emergence and evolution of social pacts in
Italy (NEWGOV project report). Florence, Italy: European University Institute.
Rhodes, M. (2001). The political economy of social pacts: Competitive corporatism
and European welfare reform. In P. Pierson (Ed.), The new politics of the welfare
state (pp. 165-194). Oxford, UK: Oxford University Press.
Royo, S. (2006). Beyond confrontation: The resurgence of social bargaining in Spain
in the 1990s. Comparative Political Studies, 40, 969-995.
Scharpf, F. (1991). Crisis and choice in European social democracy. Ithaca, NY:
Cornell University Press.
Uusitalo, R., & Vartiainen, J. (2007). Finland: Firm factors in wages and wage changes.
Cambridge, MA: National Bureau of Economic Research, Personnel Economics
Project on Comparison of Wage Structures.
Visser, J. (1999). Concertation and the art of making social pacts. In E. Gabaglio &
R. Hoffmann (Eds.), European Trade Union yearbook (pp. 217-232). Brussels,
Belgium: European Trade Union Institute.
Bio
Sabina Avdagic is UK Research Councils academic fellow in the Department of
Politics, and Sussex European Institute at the University of Sussex in Brighton, UK.
Downloaded from cps.sagepub.com at Universitaetsbibliothek on March 27, 2013