Government formation in Multi-Level Settings

Party Strategy and Institutional Constraints

I. Stefuriuc

• Éditeur: Palgrave Macmillan
• Année: 2013
• ISBN:
  9781137300744
• Sujet: Politics & International Relations
• Sous-sujet: European Politics

This book examines how parties negotiate coalition deals at the subnational level using the examples of Germany and Spain. In such multi-level settings, parties are present at various negotiation tables often having to make difficult choices about their role in the coalition and the relative merits of being in government over the opposition.

1

Towards a Theory of Coalition Formation in Multi-Level Settings

Electoral results seldom give parliamentary majorities to single parties. Very frequently, political parties need to negotiate among themselves in order to form a multi-party government. Likewise, if they go into a minority government, they need to strike deals with opposition parties so that they can ensure their survival in office and support for government-initiated policies.¹

How parties make these decisions, on what criteria they choose their partners and what type of governments are formed are questions that have been on the political science research agenda since the early 1950s. Existing research developed a solid theoretical apparatus for the analysis of government formation. This chapter shows that while this constitutes a useful anchoring point for exploring coalition formation in multi-level settings, several of the fundamental underlying assumptions of classical coalition theory need to be revised. A new theoretical framework building on these revised assumptions is necessary for studying regional government formation in multi-level countries.

This chapter is organized as follows: the first section gives a brief overview of the main insights of the classical theory of coalition formation. The discussion then moves on to evaluate the applicability of these insights to multi-level settings. The final section presents a new theoretical framework which takes into account the peculiarities of coalition formation in multi-level settings.

Coalition formation, the traditional explanations

The rational choice approach: office, policy and votes

The first coalition formation studies, published in the early 1950s, were situated in the rational choice paradigm. Starting from the premise that political parties playing the coalition game are rational actors that pursue utility maximization, scholars elaborated a range of theoretical models and empirical expectations, all addressing one fundamental question: what makes a particular combination of parliamentary parties more likely than others to successfully form a coalition government?

The main concern of coalition formation studies was to develop a theoretical model based on an explanation able to closely predict the party composition of the coalition which succeeds taking office. Predicting the coalition combination that is most likely to form is by no means an easy task, since in a typical five-party assembly a total of 31 combinations of parties are theoretically possible and only one of these actually gets the chance to govern.²

In finding the determinants of successful coalition formation, the rational choice literature relies on various assumptions about the fundamental goals political parties seek to accomplish by entering coalition governments. Thus, the early theories started from the assumption that parties are purely office-oriented organizations, seeking to maximize the spoils derived from holding executive office. The main expectations following from this assumption are derivations of the winning proposition, which states that only those governments that contain “over half of the membership or votes or weight in the decision making system” (Riker, 1962: 256) will form. This expectation follows logically from the assumption that parties are pure office-seekers and see no utility in supporting or tolerating a minority cabinet without getting any share of the government office spoils. Further refinements to this idea led to predictions such as the minimal winning proposition (Von Neumann and Morgenstern, 1953), which states that only those coalitions which do not contain any unnecessary members above the majority threshold will form – as parties are rational actors who seek to maximize the utility derived from holding office, sharing the benefits associated with it with as few partners as possible is the only logical expectation. Another variety of the winning proposition is the minimum seats (Riker, 1962) proposition, which states that if different minimal winning solutions exist, the alternative that controls the minimum total number of seats will form.³

But these early theories commonly known as the office school of coalition formation were criticized by scholars pointing out to the importance of other factors, most notably that of ideology. According to the policy school, parties are fundamentally guided by a desire to implement their preferred policy goals and choose their coalition partners according to their relative policy positions. The theoretical predictions are therefore focused on characteristics such as the ideological range covered by the possible party combinations (Leiserson, 1966) or the inclusion of the median legislator (Black, 1958; Laver and Schofield, 1990).

Both the classical office- and policy-seeking theories conceptualized coalition formation games as snapshot phenomena, analysing them in isolation from other coalition formation instances occurring in the past or in the future. Critics of this approach argued that previous coalition experience was part of the minimal information baggage with which actors entered the bargaining game: “the formation of a governing coalition should be viewed as part of a historical sequence of events in which past experience plays an important role” (Franklin and Mackie, 1983: 276). Because bargaining involves transaction costs and because familiarity with coalition partners and inertia help reduce these costs, Warwick (1996) hypothesized that incumbent coalitions have more chances to stay in office when a new formation opportunity arises. In a similar vein, starting from the assumption that parties are long-term vote-seekers, Strøm (1990: 69) hypothesized that parties are primarily concerned with the consequences of government participation for their future electoral fortunes, which may explain why “short-term portfolio gains may be sacrificed for longer-term electoral advantage”.

“The European perspective” versus the rational choice approach

Rational choice theories dominate the study of coalition formation. The so-called European political science approach (Browne and Dreijmanis, 1982; Bogdanor, 1983; Pridham, 1986) criticizes these theories for resting on overly unrealistic and restrictive assumptions – such as attributing unique and mutually exclusive goals to political parties, modelling coalition formation as snapshot rather than dynamic processes or viewing parties as monolithic organizations that follow unique strategies and goals in conditions of strict discipline (Pridham, 1986; Maor, 1998).

This approach also proposes a different method for studying coalition formation. Rather than starting from theoretical assumptions about party goals and behaviour, it suggests that coalition research should adopt an inductive approach based on thick description of the context of coalition formation and of the actors involved. It is only by having an exhaustive inventory of the variables that might explain coalition formation that we can start identifying the real determinants of coalition choices. Projects like those conducted by Laver and Budge (1992), Müller and Strøm (2000) or Blondel and Müller-Rommel (1988) collected a huge and highly valuable amount of systematic comparative data on West European cabinets and served as basis for a more exploratory approach to coalition making.

The constraining role of institutions

The early rational choice approaches to coalition formation were also criticized for focusing too narrowly on the actors involved and neglecting the constraining role of institutions. In the real world, neoinstitutionalists argued, institutions act as restrictions on “the set of feasible cabinet coalitions that [are] beyond the short-term control of the players” (Strøm et al., 1994: 308).

Strøm et al. (1994) and Bergman (1993) provide a comprehensive inventory of the institutional constraints on coalition formation. These constraints affect various aspects such as the likely composition of cabinets (e.g. consociational provisions like they exist in Belgium or India), their parliamentary status (e.g. some systems require governments to pass a positive vote of investiture, which makes it more difficult for minority governments to form, while some bypass this step and simply require that the parliament does not cast a no confidence vote, that it merely tolerates the government), or a coalition’s prospects for survival in office (e.g. the requirement of a constructive vote of no confidence which allows the parliament to overthrow an existing cabinet only in the presence of a viable alternative solution). Special rules might exist that favour the inclusion of specific parties in the coalition cabinet – such as the rules granting the formateur status to the largest parliamentary party or to the party approved by a majority of parliamentary party leaders.

In addition to these rules that might considerably narrow down the set of viable coalition alternatives, other systemic variables have been found to make more or less likely the formation of specific types of governments. Thus, Strøm (1990) finds that minority governments are generally more likely to occur in those systems in which the difference between the policy gains attained by stepping in government and the policy influence parties can exert even when in opposition is low.⁴

Lijphart (1984) and Sjölin (1993) suggest that the structure of the legislature will also impact on the type of coalition that forms. Thus, in bicameral legislatures in which the second chamber is an important arena for passing legislation, oversized governments are more likely to occur. This is because bicameral legislatures often produce divided majorities and in order to secure majority support in both chambers a government might need extra members.⁵

Which approach for studying coalition formation in multi-level settings?

The brief literature review presented in the previous section outlines the evolution of coalition formation theory. As theoretical knowledge progressed, the simple models based on the assumption of unique party goals were gradually replaced by more complex models including variables from more than one category. In later approaches, parties were conceptualized as actors seeking both office benefits and the maximization of their policy impact, while keeping an eye on the possible consequences of certain governing formulae for their future electoral fortunes and learning from past coalition experiences. Also, the role of institutions is by now fully acknowledged as restricting the set of feasible coalition combinations after each given election. Finally, authors have recently paid more attention to the specificity of the context and other less quantifiable attributes of the actors involved in negotiations. The result of this evolution is the development of complex theoretical models which are able to correctly predict the coalition that was successfully formed in reality in about 40% of the cases from among dozens or even hundreds of possible theoretical combinations (Martin and Stevenson, 2001).

Nevertheless, this “kitchen-sink approach” (De Winter et al., 2002) has been criticized on two counts. First, it has been argued that by focusing too much on increasing the rate of correct predictions of the governments that form, models of this type fail to elaborate on how the explanatory factors actually lead to the observed effect (Bäck and Dumont, 2007). In other words, a focus purely on the predictive power of statistical models, therefore on explaining the result of coalition negotiations, leaves us with only limited knowledge about the process of coalition formation.

Second, while authors using it have generally contributed to an increase in the amount and quality of empirical data on coalition formation across time and countries, this approach fares less well in terms of theoretical innovation. Laver (1989) observed that all the theories reviewed in the previous section were developed on the basis of knowledge about the formation of national governments, mainly in several western European countries. He argued that, by continuing to focus on national government formation, coalition theory condemns itself to going round in circles, as the observed data about national parties and governments generate hypotheses which are then further tested on the same data. Using new datasets is a solution for both adequately testing the empirical relevance of existing models and improving them by incorporating new, previously ignored variables (Dodd, 1976; Pridham, 1986; Laver, 1989).

These two critical arguments bring us closer to the focus of this book, which is coalition formation at the sub-national level. Local and regional governing coalitions provide us with a fresh and “fast-growing” set of observations for adequately testing old coalition theories and for developing novel explanations that take into account additional factors which previous literature left unacknowledged (Downs, 1998). By focusing on the process as much as on predicting the results of coalition formation, one can actually improve both the empirical and the theoretical understanding of this phenomenon.

The approach taken here for researching sub-national coalition formation responds to these two fundamental critiques of traditional coalition research. The aim of the book is to study coalition formation in multi-level settings through the prism of a new theoretical framework which makes it possible both to predict the successful combination across a large number of cases from a cross-country sample and to explain how the process of coalition formation in multi-level settings actually works in such settings.

In building such a framework, existing knowledge about coalition formation at the national level is obviously a useful starting point. Nevertheless, the argument here is that several of the fundamental assumptions made by most of the existing coalition formation theories need to be revised before their explanatory power is tested in multi-level contexts. This is because the layered institutional structure forces political parties into dynamics that are very different from those at work in unitary settings. Political parties need to coordinate their strategies across levels, and regional coalition formation may in fact respond to stimuli that are to be found at the national level. This revision is discussed in the following pages.⁶

Assumption no 1: parties are not unitary actors

The first way in which the multi-level nature of the institutional setting affects our understanding of coalition governments is connected to how we conceptualize the party units involved in the formation game. Classical coalition literature is largely based on the assumption that parties are unitary actors. The strongest argument in favour of the unitary-actor assumption was made by Laver and Shepsle, who argued on the basis of empirical evidence that “parties both enter and leave cabinet coalitions as unified blocks” (Laver and Shepsle, 1996: 25). But more recent research has found that internal party politics, and most notably factionalism, help give a better explanation of real-world coalition behaviour (Mitchell, 1999; Bäck, 2008).

If in centralized systems the unitary-actor assumption is questionable, in politically decentralized countries it is simply untenable, at least as regards statewide parties. A statewide party (SWP) is defined here as a political party that contests both regional and national elections in all or nearly all regions of the country, largely under the same electoral banner; if regional or national elections are fought in certain regions by a regional division of a party competing under a different banner – but this organizational division is not competing against any other organizational division of the same party and does not form a separate parliamentary group in those parliamentary contexts in which it coexists with the national division – then it will be counted as the same statewide party (Ştefuriuc, 2009a).

By contrast, a “non-statewide party” (NSWP) is one that contests either regional or national elections, or both, in a limited territory of the country (one or several, but never all regions), and which retains a separate parliamentary group organization in sub-national parliaments. If two or more non-statewide organizations run elections using a common list and, following elections, form one single parliamentary group, then, for the same reasons as above, they are considered a single non-statewide party (Ştefuriuc, 2009a).⁷

While non-statewide parties may be internally divided over coalition politics, for the purposes of this research they can in fact be treated as unitary actors. When it comes to statewide parties, however, the multi-level nature of their organization cannot be ignored: different party units are present at the negotiation tables at different levels, and the stakes and goals of their leaders are also likely to be different. Thus for regional party leaders the immediate stakes of coalition talks are most often related to their personal participation in the government to be formed, while for national party leaders observing or interfering in the process at the regional level, the stakes are often less personal and are related to the coordination of an overall party strategy.

Of course, the level of vertical organizational integration of parties in such systems often varies (Deschouwer, 2003), but all parties competing in multi-level systems are subject to substantial centrifugal forces. In some cases, the territorial arrangement of the state is mirrored by the organization of political parties. In others, their response to the centrifugal tendencies induced by territorial decentralization is to tighten the centralization of their organization. These are two different strategies that target the same problem: how to maintain the desired degree of vertical integration of the party organization? The first strategy proposes accommodation, by quasi-federalizing the party organization and devolving powers to the territorial party ...