1 Introduction

This article examines the influence of the interplay of party system factors on the formation of minority governments in Central and Eastern Europe. This research topic is of great interest for a number of reasons:

First, minority governments, defined as governments without a majority in parliament, are an empirically relevant phenomenon: They make up about one-third of all governments in Western as well as in Central and Eastern Europe (Woldendorp et al. 2000; Döring and Manow 2012).

Second, the formation of minority governments always puzzled both the public and political science. At first glance the formation of minority governments is a counterintuitive phenomenon. It seems to contradict a basic principle of representative democracy, which is the formation of a government representing the majority that won the election. Additionally, the formation of minority coalitions seems to contradict the office-seeking motivation of parties. Why should a party decide to stay in opposition but support a government that has no majority? Accordingly, minority governments were seen as “accidents” and “crises phenomena” (see, for example, von Beyme 1970; Taylor and Herman 1971). Only in the 1990 s did perception change. Researchers could show that the formation of minority governments is not accidental, but in most cases brought about by the voluntary and rational choices of actors (Strøm 1990; Bergman 1995). But these findings refer solely to the Western European countries (mainly to Scandinavia), not to Central and Eastern European countries.

Third, while government formation in the established Western European democracies is well studied, only recently has government formation in Central and Eastern Europe aroused the interest of coalition researchers. This is surprising, first because knowledge of government formation is crucial to understanding the functioning of representative democracies, and second because other topics – such as the development of the Central and Eastern European party systems, which is strongly related to the study of coalition formation – are well studied. With regard to the formation of minority governments, the situation is similar to that in Western Europe before the change of perception in the 1990 s: The occurrence of minority governments is mainly perceived as a crisis phenomenon (see, for example, Kropp et al. 2002a), but no substantive work on it exists.

Against this background, this article, which is based on a comprehensive study (Keudel-Kaiser 2014), focuses on the formation of minority governments in Central and Eastern Europe. The following eight Central and Eastern European EU member states are included: Bulgaria, the Czech Republic, Estonia, Latvia, Lithuania, Poland, Romania and Slovakia. The study comprises all government formation processes after parliamentary elections from the first regular elections in the early 1990 s up to 2010. The research question is: Which combination of party system factors leads to the formation of minority governments in Central and Eastern Europe?

I chose to focus on party system factors for the following reasons: First, party system factors are generally seen as central in coalition research, as parties and party elites are key players in the process of government formation (Laver and Budge 1992a; Müller et al. 2008; Savage 2016). Second, in the political systems of Central and Eastern Europe, parties play an extraordinarily central role because they were key players in the transformation processes of the young Central and Eastern European democracies (Segert 1994a, 1994b, von Beyme 1997; Bos and Segert 2008b). Many studies show a link between the development of the party systems and government stability (see, for example, Toole 2000; Bakke and Sitter 2005; Müller-Rommel 2005). Thus a large segment of coalition research on Central and Eastern Europe focuses on party system factors (see, for example Nikolenyi 2004; Grotz and Weber 2011, 2012; Savage 2013a, 2013b, 2014). Finally, to explain the formation of minority governments in Western Europe, especially in the Scandinavian countries, the coalition literature refers to central factors that do not apply to Central and Eastern Europe. In many cases the factors are simply not present, such as the central Scandinavian factor of “negative parliamentarism” (whereby a party or coalition that seeks entry into office does not need an explicit level of support from the parliament, Bergman 1995).

The method chosen is Qualitative Comparative Analysis (QCA) (Ragin 1987, 2000). QCA is an approach situated between qualitative and quantitative methods. It allows for generalisation, but at the same time focuses on every single case. The special strength of QCA lies in mapping out combinations of conditions that lead to the outcome of interest (Rihoux and Lobe 2009). Thus far, QCA has seen limited use in coalition research where purely quantitative approaches have prevailed. This is surprising because the advantages of the use of QCA in this context are obvious: First, it is evident that a number of interacting factors are involved in the complex process of government formation. Second, the use of QCA complies with the urgent appeal of coalition researchers to pay more attention to the contexts of the single cases that often fall by the wayside in the complex and abstract models, including large numbers of cases (see, for example, the criticsm by Pridham 1986; de Winter et al. 2002). Third, with the mixture of qualitative and quantitative elements, it responds to the call to combine different research approaches in coalition research (see, for example, Bäck and Dumont 2007).

The article is structured as follows: In the next section (part 2), the research question is embedded into the research context and the theoretical background. Part 3 introduces method, case selection and the operationalization of the conditions. The main section (part 4) presents and discusses the results of the Qualitative Comparative Analysis and gives insight into selected cases and conditions. The article concludes with a summary of the results (part 5).

2 Minority governments in theory and research

In the following, I give a short overview of coalition research on minority governments in Western Europe and then turn the focus to Central and Eastern Europe (for a detailed presentation of the wider research context, see Keudel-Kaiser 2014, Chap. 2.1Footnote 1). For a considerable time, coalition theory neglected minority governments. Scholars perceived the determining factors of coalition building to be the size of the coalition and the office-seeking motivation of political parties and their leaders, who were conceived as rational, unitary actors (among others, Gamson 1961; Riker 1962). At first glance, the formation of minority coalitions contradicts the office-seeking motivation of parties as well as the size principle: Why should a party decide to stay in opposition but support a government that has no majority? Minority governments are a “counter-intuitive phenomenon” (Strøm 1990, p. 8) and were therefore seen as “anomalies” and “accidents” despite the fact that about one-third of coalitions in Western Europe since 1945 have been minority governments (Woldendorp et al. 2000). Suggested explanations for the emergence of minority governments ranged from deep crises (von Beyme 1970; Taylor and Herman 1971) and a conflict-ridden political culture (Luebbert 1984) to the extreme instability of the political systems, high degree of fragmentation and deep cleavages (Dodd 1976).

A real change in perception took place with Strøm’s pioneering work on minority governments grounded in the rational choice tradition (Strøm 1990). Strøm demonstrates that minority government formation can be understood as the consequence of rational behaviour by the party leaders evoked by incentives provided by institutions (Strøm 1990, p. 23). He points to the fact that legislative and executive coalitions need not coincide: “Majority status is not necessarily the effective decision point in parliamentary legislature” (Strøm 1990, p. 38, emphasis included in the original). One of Strøm’s main arguments, the thesis that the greater the influence of the opposition, the more likely the formation of minority governments, reappears in assumptions about the influence of political culture on government formation: A consensual political culture is said to promote the formation of minority governments because all political forces are integrated regardless of whether or not they are part of the government (Luebbert 1984; Jahn 2002; Pehle 2002).

Bergman (1993, 1995) complements Strøm’s mainly rational-choice-based assumptions with further institutional factors. He analyses how specific constitutional arrangements concerning government formation influence the type of government that will form (Bergman 1995, p. 22). Bergman points out that negative parliamentarism – the fact that a new government must only be tolerated by parliament and need not win a vote of investiture – often leads to a high number of minority governments in countries with proportional systems (Bergman 1993, p. 60, 1995). Apart from such studies dealing with minority governments per se, minority governments feature in comprehensive studies that use statistical models to try to explain or predict which type of government (minority, minimal winning or surplus) will form and under which circumstances (for example, Martin and Stevenson 2001; Mitchell and Nyblade 2008). Institutional variables and party system features prove to be of particular importance for the formation of minority governments. In accordance with Bergman, Strøm et al. (1994) and Martin and Stevenson (2001) show that the absence of a formal investiture vote makes the formation of minority governments more likely, while other investiture rules, such as a constructive vote of non-confidence, make it unlikely. Martin and Stevenson (2001) prove that the existence of anti-system parties excluded from government formation has a significant impact on the formation of minority governments. A number of studies show that the existence of an ideologically central and at the same time numerically strong party that “dominates” the party system plays an important role in the formation of minority governments.Footnote 2

Other than for Western Europe, almost no literature exists on the formation of minority governments in Central and Eastern Europe despite the fact that, as in Western Europe, they make up a third of all governments. This corresponds to a general delay: Only in recent years have researchers begun to include the Central and Eastern European countries (see, for example, Kropp et al. 2002b; Kropp 2008). In contrast to coalition research on Western Europe that for a long time focused primarily on government formation, coalition research on Central and Eastern Europe was, given the (at least initially) very unstable party systems, concerned with government stability. Footnote 3 Recently, however, interest in the processes of government formation has increased. Some studies compare the factors that determine government participation in Western and Eastern Europe and for the most part reach the conclusion that “substantial differences in patterns of government formation” exist between East and West (Döring and Hellström 2013, p. 684, see also Bergman et al. 2015).

Other studies focus solely on Central and Eastern Europe. Savage (2014) examines the influence of the parties’ ideological positions on government membership in five Central and Eastern European countries. Contrary to previous research, he shows that “ideological variables are highly significant in determining which parties get into government” (Savage 2014, p. 558): Parties attempt to build ideologically homogeneous governments; parties close to the median position are more likely to be included in government coalitions. Some authors analyse the influence of (single) institutional factors and party system features on government formation, such as presidential influence (Protsyk 2005) or intra-party conflict (Ştefuriuc 2004). One particular party system factor is of special importance for government formation processes in Central and Eastern Europe: the regime divide, i. e. the divide between post-communist and post-oppositional parties. Grzymala-Busse (2001, p. 85) describes it as the “fundamental predictor of coalition formation” (see also Druckman and Roberts 2007). Membership in one of the two camps often outweighs the policy proximity that is generally considered an essential predictor for government formation processes: Parties will form coalitions with parties within their camp even if the policy positions of parties from the other camps would be much closer (Grzymala-Busse 2001, p. 88).

Some authors examine single government formation processes or specific aspects of government formation. A phenomenon that has attracted growing attention is the role of newly founded parties in government formation that often immediately manage to enter parliament or even government (Tavits 2007; Grotz and Weber 2013, 2016; Savage 2016). Grotz and Weber (2013) show, inter alia, that “genuinely new parties” are often included in oversized governments, but not in minority governments.

Very few studies look explicitly at the formation of different types of governments (minimal winning, minority, surplus). An exception is the study by Grotz and Weber (2011) on coalition governments in the ten Central and Eastern European EU member states. The authors emphasise the influence of party system features on the type of government that will form. They show that minority governments form most often in moderately fragmented and highly polarised party systems. They are ideologically more homogeneous than minimal winning coalitions, that is to say, party leaders seem to attach more importance to ideological coherence than to a parliamentary majority (Grotz and Weber 2011, pp. 203–204).

Thus far, few studies have looked explicitly at the formation of minority governments. The first in-depth study analysing the formation of minority governments in Central and Eastern Europe is the study upon which this article is based (Keudel-Kaiser 2014).Footnote 4 It is notable that in studies where minority governments are mentioned in passing, they are described as crisis phenomena and less-than-ideal solutions that can primarily be traced back to instable party systems (for example, Kropp et al. 2002a; Pridham 2002; Kropp 2008). But neither data nor profound reasoning are given to support such theses. Thus a parallel exists to early coalition research on the established democracies where minority governments were perceived as accidents, and no detailed examination was performed.

In sum, there is a multiple research gap as Central and Eastern Europe has generally been neglected in coalition research. Only recently has the formation of governments attracted more attention. Within the study of government formation, the formation of minority governments has so far received only cursory treatment despite the fact that minority governments constitute a third of all governments and are thus an important empirical phenomenon.

3 Method, case selection and operationalization

3.1 Methodological approach: Qualitative Comparative Analysis (QCA)

The aim of the study with its broad focus on government formation processes in eight of the Central and Eastern European EU member countries from the early 1990 s up to 2010 is, on the one hand, to describe “general features of government formation processes” (Laver and Budge 1992b, p. x). Accordingly, it can be assigned to the “comparative European politics tradition” (Laver and Budge 1992b, p. x). On the other hand, it is interested in the contexts of the single cases and in the processes that lead to the outcome and thus shares basic features with the “case study approach”.

This dual interest is reflected in the choice of method: The government formation processes are analysed by means of Qualitative Comparative Analysis (QCA). QCA, developed by Charles Ragin (1987, 2000), is a case-oriented method that sees itself as a “third way” between statistical-standardised and case study approaches (Schneider and Wagemann 2007). It can analyse a larger number of cases than a purely qualitative comparative study and can thus produce a certain level of generalisation (Schimmelfennig et al. 2006). At the same time, it provides “in-depth insight in the different cases” (Rihoux and Lobe 2009). The particularity of the QCA method lies in its “concept of multiple conjunctural causation” (Rihoux and Lobe 2009, p. 223, emphasis included in the original). Typically, researchers focus on the “net effects of causal conditions” (Ragin and Sonnett 2005, p. 180, emphasis included in the original), that is to say, they treat each condition as an independent cause of the outcome, regardless of the value of the other relevant conditions (Berg-Schlosser et al. 2009, p. 9). QCA looks at the combinations of conditions necessary or sufficient for the outcome of interest.

So far, QCA has hardly been used in coalition research where purely quantitative approaches have prevailed. This is surprising given that the advantages of QCA in this context are obvious. First, it is evident that a number of interacting factors are involved in the complex process of government formation. Accordingly, coalition researchers have long pledged to model “the interactions between the various types of explanatory variables used to understand coalition formation” (Mitchell and Nyblade 2008, p. 233) instead of looking at the influence of single factors. Second, the use of QCA complies with the urgent appeal by coalition researchers to pay greater attention to the contexts of the single cases that often fall by the wayside in the complex and abstract models, including large numbers of cases (see, for example, Pridham 1986; de Winter et al. 2002). Third, with the mixture of qualitative and quantitative elements, it responds to the call to combine different research approaches in coalition research (see, for example, Bäck and Dumont 2007).Footnote 5

A QCA procedure consists of three main steps (Rihoux and Lobe 2009): (1) an in-depth examination of the cases, comprising case selection and description (2) the “analytic moment”, based on a computer programme where the different combinations of conditions, described in a dichotomised manner, are compared pairwise and logically reduced by use of formal logic (Boolean algebra), and (3) the “downstream” interpretation of the results that goes back to the single cases. Here, the so-called crisp set version of the QCA is used: Based on theoretical assumptions combined with empirical knowledge, the researcher has to describe the conditions in a dichotomised manner. He/she has to set thresholds for their presence (=1) or absence (=0) (for details see Berg-Schlosser and De Meur 2009).

3.2 Operationalizing the outcome: formation of minority governments in Central and Eastern Europe

Minority governments, as compared to minimal winning or surplus coalitions, are defined as governments that do not possess a majority in parliament, that is to say, that do not comprise more than half of the members of parliament. Minority governments can be single-party governments as well as party coalitions (Woldendorp et al. 2000, p. 17).

Government formation: The standard definition in coalition research includes governments formed after elections as well as governments formed during the legislative period (Woldendorp et al. 2000; Müller-Rommel et al. 2008). Here, however, only those governments are counted as “new governments” that are formed after regular elections. This much narrower definition is used because there are different kinds of logic behind the formation of minority governments after elections, compared to those formed during the legislative period. While the former are built according to a regular process and under comparable conditions, the latter are precipitated by crisis of the precedent government.

Here, the term Central and Eastern Europe is used for the ten Central and Eastern European EU member states (accession 2004/2007). These countries share the same background: All are former communist states that began the transformation process after the regime changes of the early 1990 s. The basic setting in which government formation takes place is similar; the institutional structures of these states are homogeneous (Ismayr 2010; Müller-Rommel and Grotz 2011). All ten countries are unitary states. And following Steffani’s definition (Steffani 1979), all have parliamentary systems. The process through which a government comes into office differs only slightly among the countries. All ten have “positive parliamentarism” (Bergman 1995), that is to say, the government must be voted into office by parliament. The starting conditions are thus similar.

But not all ten countries are included in the analysis. In four of the ten countries, no minority government was formed during the period under investigation. Two of the countries, Hungary and Slovenia, were marked by special circumstances that made the formation of minority governments highly unlikely. In the Hungarian case, it is mainly the electoral system with its exceptionally strong majoritarian elements favouring strong parties that made the formation of minority governments quasi impossible.Footnote 6 In Slovenia, it was primarily the consensus-oriented politics rooted in the political culture of the former Yugoslavian federalism (Fink-Hafner 2006) that made the formation of minority governments highly unlikely.Footnote 7 Thus Hungary and Slovenia are not included in the study.

For methodological reasons, the two other countries (Estonia and Slovakia) where up to 2010 no minority government was formed are included in the QCA; in neither case do obvious reasons exist that could explain the absence of minority governments. To be able to show that a specific combination of conditions leads to the formation of minority governments, it must be shown that the same combination of conditions does not lead to the formation of majority governments in other cases (other countries) with the same starting conditions.

Concerning government formation in the remaining eight countries, further restrictions are applied. First, only those elections that took place after the adaptation of a constitution are included in the QCA. The choice of this starting point guarantees that the elections took place in a relatively stable and settled context. Second, some elections are excluded due to their exceptional context. This is true for the early Czech and Slovak elections that took place within the framework of the Czechoslovak federation. Third, only those governments are included in the analysis that formed after elections in which no party controlled an absolute majority of seats from the beginning, as the presence of a party controlling a majority makes the formation of a minority government almost impossible. Following these definitions, the total number of governments is 39, 20 of which can be called “minimal winning”, 13 “minority” and 6 “surplus” (see Table 1).

Table 1 Post-electoral governments in the eight Central and Eastern European countries, first regular elections up to 2010
Full size table

3.3 Operationalizing the conditions: five party system factors

Several party system factors discussed in coalition literature are supposed to influence the formation of (minority) governments. These factors refer to numerical aspects, such as the size of single parties or the numerical composition of the party system as a whole, and/or to ideological aspects, such as the ideological position of single parties or the overall ideology of the party system. Not all factors are transferable one-to-one to the Central and Eastern European context. Some do not apply because there is no variance between the cases.Footnote 8 Some factors specific to Central and Eastern Europe must be considered, such as the strong divides structuring party competition especially in the first decade following the regime change. The following factors turned out to be theoretically relevant (for a detailed discussion of all possible factors, see Keudel-Kaiser 2014, Chap. 4).

Seat share of the largest party (Nearmaj): The formation of minority governments is said to be more likely when the seat share of the largest party is large, i. e., nearly a majority (almost 51% of the seats). If this is the case, the party can easily form a minority government without being challenged by a powerful opposition (Herman and Pope 1973, p. 197; see also Taylor and Laver 1973). This basic assumption seems very plausible and even intuitive. However, it seems appropriate to extend the condition a bit further: A party, even if it comes very close to a majority, will not form a minority government if it does not have the (formal or informal) commitment from at least two of the opposition parties, namely that they will support the government (to enter into office as well as during the parliamentary term). The commitment of only one party would be too uncertain: The withdrawal of its support would destabilise or even bring down the minority government. In its extended form (party near to a majority with the support of at least two opposition parties), this factor is integrated in the analysis. The point of reference for the threshold setting for this condition is the fundamental study by Herman and Pope (1973). Based on theoretical considerations supported by the empirical evidence, they classify a percentage of 48% of seats as near to a majority (Herman and Pope 1973, p. 199). To check whether the additional requirement – the formal or informal commitment of at least two opposition parties to support the formateur party – is met, the study relies on the respective literature on elections and government formation processes.

Two-party dominance combined with lacking mutual coalition potential (Twodom): Not only the strength of a single party, but also the strength of several parties that dominate the party system is supposed to influence the formation of (minority) governments. If two large parties lack mutual coalition potential, the formation of a majority government becomes difficult. In research on party systems, the dominance of two parties is defined as follows: Two parties control more than a quarter of the seats each and together more than two-thirds of the seats. The third-largest party controls not more than half of the seats than the smaller of the two leading parties (Niedermayer 2010, p. 3). With regard to government formation, the crucial additional point is whether the two dominant parties are able to form a coalition with each other. The parties are classified as not mutually coalitionable if they are described as such in the respective literature (for details, see Keudel-Kaiser 2014, p. 71).

High percentage of “non-coalitionable” parties (Noncoal): A widespread but controversial hypothesis in coalition research is that the formation of minority governments is more likely if there are parties playing an outsider role in the party system, as such a role tends to limit the options of (majority) coalition formation. These parties are usually referred to as “extremist” or “anti-system” parties. These terms are, however, defined in various ways and used in different contexts, which leads to contradictory conclusions regarding their influence on government formation.Footnote 9 Assuming that “extremist parties” are automatically “non-coalitionable parties” is problematic: Especially in Central and Eastern Europe, parties that are “extremist” in the ideological sense are nevertheless often accepted as coalition partners by the so-called mainstream parties.

In this context, the “conceptual reassessment” of the term “anti-system parties” by Capoccia (2002) is helpful. He distinguishes between two types of anti-systemness: The concept of “‘ideological’ anti-systemness” refers to a party’s inherent ideological character. It “consists in the incompatibility of its ideological referents, and therefore its potential goals, with democracy” (Capoccia 2002, p. 24). The concept of “‘relational’ anti-systemness” refers to a party’s ideological distance from other parties in a particular system: “a party’s relational anti-systemness is not given per se by its ideological character, but rather by its ideological difference from the other parties in the system” (Capoccia 2002, p. 30, emphasis included in the original). Drawing on this second concept, parties here are classified as “non-coalitionable” if they are excluded from forming a government by the other parties (or the president) because of their “relational anti-systemness”. The necessary information can be deduced from the literature on the respective electoral campaigns and processes of forming a government. There is no agreement (and generally no specification) in coalition research about what percentage of seat share controlled by “non-coalitionable” parties is to be classified as “high”.Footnote 10 The data on “non-coalitionable” parties underlying this study exhibit no “natural” threshold: The percentage of “non-coalitionable” parties at the single elections varies between 8.5 and 24.3%. Setting a threshold below 8.5 would mean that the presence of “non-coalitionable” parties per se is automatically equal to a high percentage of “non-coalitionable” parties. This would take all discriminative power from this condition. The empirical evidence indicates that the presence of “non-coalitionable” parties carries weight in the process of coalition building only when these parties control at least 12% of the seats. In all three cases where the percentage was lower, the “non-coalitionable” parties did not play an influential role in the government formation process.Footnote 11 Therefore, a threshold of 12% is chosen. Threshold setting on an empirical basis contains the danger of circular reasoning, but as shown above, in this case it seems to be the only way out.

Lacking policy closeness (Lackpol): Assuming that parties stand for specific policy positions that they aim to enforce, the question of whether parties share basic policy positions with the formateur party is crucial for government formation. In contrast to the common assumption that parties in Central and Eastern Europe are only interested in office, recent research proved that ideological considerations do matter in coalition formation in Central and Eastern Europe, and are even “one of the most influential determinants of coalition membership in CEE countries” (Savage 2014, p. 548). A lack of potential coalition partners sharing at least roughly the same policy positions should thus play a role in the formation of minority governments. If no suitable coalition partner exists, a party might prefer to govern without a majority instead of governing in an ideologically inhomogeneous majority government that constantly forces the formateur to make policy compromises. Policy closeness is said to be lacking if there are basic differences in central policy fields between the formateur parties and the other parliamentary parties. The parties’ policy positions are deduced as described in Sect. 3.4.

Strong divide (Divide): Not only does the ideological standpoint of single parties play a role, but also the ideological structure of the whole party system. Coalition researchers point to the effect of “cleavage conflicts” (Dodd 1976) or “ideological polarisation” (Warwick 1998) on the formation of minority governments. The theoretical assumption emerging from this literature for the study at hand is the followingFootnote 12: When a party system is strongly divided into two camps, options for forming coalitions are from the beginning rather limited. This factor should therefore play a role in the formation of minority governments. Scholars point to the special importance of specific Central and Eastern European divides characteristic of the first years following regime change: The deep divide between post-communist and post-oppositional forces (regime divide) in most Central and Eastern European countries and the deep divide between pro-Russian and anti-Russian forces in the Baltic States often overshadowed policy closeness or distance between the parties.Footnote 13 A strong divide is said to be present when the entire party competition is structured along a specific and deep dividing line, when this is reflected in the electoral campaign, and when the political science literature classifies it at such (for details, see Keudel-Kaiser 2014, pp. 68–69).

3.4 Data sources: the challenge of party position determination

The QCA is based on detailed case descriptions, entailing descriptions and discussions of the above-introduced conditions and their operationalization for every single case (see Keudel-Kaiser 2014, Chap. 5). One aspect deserves special attention: To answer the question whether the party charged with government formation can identify potential coalition partners that share its main policy positions to build a majority government, information about the parties’ policy positions is needed. The measurement of policy positions and the use of the data in coalition research are highly contested issues. Two approaches prevail: analyses of party manifestos and expert surveys. The most frequently used data are the manifesto datasets, although their use in coalition research is very problematic. Most coalition researchers that employ the manifesto data admit that they chose them due to the absence of alternatives, despite their very limited appropriateness.Footnote 14

The first problem is that the data coding is based on the salience of issues. The results reveal only vague information about the parties’ spatial positions, although such information is essential for coalition formation.Footnote 15 Second, manifestos have the character of strategic documents: Ideological differences between parties important in the process of coalition formation can hardly be inferred from manifestos addressed to voters and not to potential coalition partners (Taylor and Laver 1973, p. 215). The expert surveys have a big advantage in that they position parties within policy spaces and draw a much more realistic picture of the issues dominating party competition. But their use in coalition research also has pitfalls. One of the major shortcomings is that they cover only single points in time. Second, there is the danger of circular reasoning. The country experts that classify the parties may have their coalition behaviour in mind (Müller 2009, p. 234, see also, Budge 2000). Third, although the experts base their judgement on deep case knowledge, the survey data have a certain degree of abstraction (Mair 2001). The scales give a general picture of dominant issues in party competition but do not always reflect the issues relevant for coalition formation.

For the case-oriented study at hand that aims to be as close to the single government formation processes as possible, the use of expert surveys is undoubtedly preferable. Amongst the expert surveys, the Benoit and Laver (2006) dataset succeeds best in minimising the described shortcomings, mainly with regard to the regional focus of the study at hand, because it takes into account the special Central and Eastern European context: It deploys a core set of scales for all countries, but adds a secondary set of dimensions such as “treatment of former communists” for the post-communist countries (Benoit and Laver 2006, p. 129). Yet use of the expert surveys is not the final answer. The problem remains that the dataset covers only a very narrow time period (2000–2004). Due to the large research gap in this field, only a few examples exist of how other researchers working on Central and Eastern Europe deal with this problem. Some use manifesto data (Bergman et al. 2015; Döring and Hellström 2013), and some compromise by using the problematic classification of parties into party families (Grotz and Weber 2010, 2011). Savage bases his analyses on a new expert survey of party policy positions, using a left-right space “defined in a way that is meaningful to CEE countries” (Savage 2014, p. 547). His is a promising approach, but the dataset covers only five of the Central and Eastern European countries and only up to the year 2006.

Against this background, I decided to take an alternative approach. Because party positions and the decisiveness of issues change frequently in Central and Eastern European post-communist countries, I define the topics that dominate party competition and the parties’ positions on them separately for every single election. The determination of the dominant issues and the parties’ positions are based on descriptions of the electoral campaigns and the parties’ current profiles, as provided by journals such as Electoral Studies or the European Journal of Political Research. Additionally, reports provided by the press and by political foundations are consulted. In short, this is nothing other than what Mair (2001, p. 13) describes as one way to assess parties’ policy preferences amongst others, the so-called “secondary reading”: “Scholars immersed themselves in as much of the available literature as possible on a given party system, and from this they derived their own estimates of relative party positions, as well as of changes in these positions over time.” Certainly my own approach may also be subject to the danger of circularity. However, the case study results show that I mostly managed to circumvent this pitfall: In a number of cases, parties classified as coherent in terms of policy do not form coalitions or vice versa.

Where possible, these assessments are carefully crosschecked with the existing survey data, mainly the Benoit and Laver (2006) dataset.Footnote 16 The long version of the study (Keudel-Kaiser 2014) includes detailed descriptions of the derivation of the policy position for every single party and every single election.

4 The analysis

4.1 Summary of the data: truth table

In the following, the interplay of the five conditions described above is examined by means of a Qualitative Comparative Analysis (QCA). One of the five conditions – the condition “one party coming close to a majority (‘Nearmaj’)” – is excluded from the start because it shows almost no variance and has therefore only limited explanatory value (see Schneider and Wagemann 2007, p. 98). There is only one case (Bulgaria 2009) where a single party comes close to a majority and is supported by at least two opposition parties.

The truth table (see Table 2) summarises the dichotomised data: It shows the different combinations of the remaining four conditions. Each row stands for one of the possible 16 combinations of conditions and the related outcome (minority government = 1; majority government = 0). The column “n” shows the number of cases covered by the specific configurations; the last column lists the corresponding case names.

Table 2 QCA: truth table
Full size table

Of the 16 possible configurations, 14 are covered by empirical cases. Only two configurations are without empirical equivalents, so-called logical remainders (row 2 and row 11). I chose a conservative strategy (Rihoux and Lobe 2009, p. 234) and included only those configurations that are represented by actual cases, as the main aim of the study is not to come up with the most parsimonious solution, but “to identify the configurations under which the outcome to be explained consistently occurs” (Schwellnus et al. 2009, p. 10).

There are four so-called contradictory configurations: The same combination of conditions leads in some cases to minority, in other cases to majority governments (rows 5, 8, 13 and 16). In one of the four configurations (row 5), a re-codification of the outlying case seems appropriate: Three of the four conditions are present. The expected outcome would thus be a minority government. Indeed, three of the four cases covered by this solution are minority governments (consistency: 0.75). The only exception is the 1995 Latvian government. This contradiction is easily resolved: Following the 1995 Latvian election, a minority government was built that asked the parliament for its investiture vote. But the government did not receive a majority in the parliament (for details, see Davies and Ozolins 1996; Tiemann and Jahn 2002) and further government formation attempts failed. In the end, as a last resort, a businessman without party affiliation was nominated as prime minister. The conditions under which government formation took place were no longer the same as at the beginning. It therefore seems appropriate to recode the case as minority government (Latvia 1995 = 1). As I chose a conservative consistency threshold (1.00), the other three contradictory configurations are not used to explain the outcome. There are, however, convincing explanations for the deviance of cases for at least two of the three contradictory configurations.Footnote 17 Thus, 11 minority governments are included in the analytic part of the QCA (for a detailed analysis of the excluded cases, see Keudel-Kaiser 2014, pp. 251–254).

4.2 Minimisation of the data: solution term

For the minimisation process, the fsQCA software (Ragin et al. 2006), Version 2.0 is used.Footnote 18 The different configurations are “logically simplified through a bottom-up process of paired comparison” (Ragin 2006). The following report of the results (see Table 3a) follows the convention in QCA: Capital letters are used to indicate the presence of a condition, small letters for its absence. The * sign stands for a logical AND, the + sign for a logical OR (Schneider and Wagemann 2010, p. 414). The result shows the solution term with the fewest possible conditions (Rihoux and Lobe 2009, p. 224). It shows the different paths leading to the same outcome (equifinality). Under each path, the cases covered by it are listed.

Table 3a QCA: solution terms. Complex solution
Full size table
Table 3b QCA: solution terms. Necessity of single conditions
Full size table

As I chose a conservative consistency threshold excluding all configurations not consistently showing the same outcome, the consistency value of the whole solution term is 1.00. Because I excluded three minority governments from the minimisation process as they were part of contradictory configurations (see above), the solution covers (coverage) 11 of the 14 minority governments (including the recoded 1995 Latvian minority government), that is to say, 79% of the positive outcomes. This is a fairly good result.Footnote 19

The solution term shows that there are three equifinal paths leading to the formation of minority governments. Minority governments are formed

  • when the party system is strongly divided and when there is a lack of potential coalition partners sharing the main policy positions with the formateur party (path 1) or

  • when there are two dominant parties that are mutually not coalitionable and a lack of potential coalition partners sharing the main policy positions with the formateur party (path 2) or

  • when the party system is strongly divided and there is a high number of “non-coalitionable” parties and an absence of two-party dominanceFootnote 20 (path 3).

The result shows that all four factors theoretically deduced as being relevant for the formation of minority governments do play a role.Footnote 21 The influence of party system features on the formation of minority governments is thus confirmed. None of the conditions is, however, a necessary condition, that is to say, none of the conditions must be present for a minority government to form: For all the conditions, the consistency value is clearly below 0.9 (see Table 3b), the (informal) threshold that prevails in the literature.Footnote 22 But each path contains either the condition “strong divide” or the condition “two party dominance”. Both conditions stand for a bifurcation of the party system. That is to say, the presence of a bifurcation of the party system is a necessary condition for the formation of a minority government. Indeed, it seems intuitive that the division of the party system into two blocs, either by a strong divide or by two dominant parties, most limits the majority government formation options. A high degree of “non-coalitionable” parties or a lack of parties sharing the main policy positions is only the final straw that, together with the bifurcation, hinders the formation of a majority government. One might object that the two conditions – a strong divide and the presence of two dominant parties lacking mutual coalition potential – are too similar. This can be rejected. The truth table shows that a number of cases exist where the condition “strong divide” is present but “two party dominance” is absent, or vice versa.Footnote 23

A surprising result is the fact that there is no relationship between a certain solution term and a certain country or country group, despite the different backgrounds. The Baltic States were the only countries entirely subordinated to the Soviet Union during the communist regime; Romania and Bulgaria were in many ways “delayed” compared to the other Central and Eastern European countries, which is reflected in their slower transformation processes and later accession to the European Union. There is also no difference between the first and second decades after the regime changes: There are neither different conditions nor is there a different interplay of the conditions. There are, however, differences in the details. This is especially true for the condition “strong divide”, which is discussed in greater detail in the next section.

4.3 “Downstream interpretation”

The strength of the QCA method lies in mapping out key combinations of conditions leading to the outcome of interest. But QCA gives no hints regarding “the ingredients and mechanisms producing (or not) an outcome of interest” (Rihoux and Lobe 2009, p. 235). Here, the case study approach as an essential complement to the QCA comes to the fore: The main strength of this approach lies within its ability to identify the intervening causal processes leading from the conditions to the outcome (Gerring 2004; George and Bennett 2005). For lack of space, I cannot carry out detailed case descriptions here. To nevertheless shed some light on the “black box” of the underlying processes, I have selected illustrative examples, namely two conditions that are of major importance and two cases that represent two extremes inside the QCA: the “prime example” of a minority government formation and an “outlier”.Footnote 24

4.3.1 A closer look at single conditions

Two conditions are of particular interest because they both appear in two of the solution paths and because their relevance for coalition formation is prominently discussed in the literature: a strong divide structuring party competition and the lack of a coalition partner sharing the same policy positions with the formateur.

Strong divide: The presence of a strong divide structuring party competition occurs in two of the three solution terms. There are nine cases (corresponding to six configurations) in which a strong divide is present and linked to the formation of a minority government. It is interesting to know the exact content of the divide (Table 4).

Table 4 Factors behind the condition “strong divide”
Full size table

The divides that were of importance in the first decade after the regime change are mainly related to topics characteristic for this period: the regime divide between post-communist and post-oppositional parties (Bulgaria 1991; Poland 1991; Romania 1992), and the divide between pro- and anti-Russian forces (Latvia 1993, 1995) respectively. This observation confirms the literature. In particular, the regime divide is said to strongly influence government formation, first and foremost in the first decade following the regime change (Grzymala-Busse 2001). New is the proven link between the regime divide and the formation of minority governments.

The conflicts linked to the early transformation process are, however, not the only driving force behind the strongly divided party systems. In the three Czech cases party competition was structured by a deep economic left-right divide. This divide was combined with strong antagonism between the leaders of the two dominant parties representing the left and right wings of the party system, the Civic Democratic Party (ODS) and the Social Democratic Party (ČSSD). Strong antagonism between party leaders that influences government formation processes reappears in other cases. One case, the 2005 Polish government formation process, deserves special attention. Here, the strong divide was driven primarily by personal animosity: The ideological divide between a “liberal and a social Poland” (Szczerbiak 2006, p. 16), represented by the two main contenders, the Law and Justice (PiS) and the Civic Platform (PO) parties, was aggravated by strong personal animosity between the party leaders, Jarosław Kaczyński (PiS) and Donald Tusk (PO). This divide grew even further during the aggressive presidential campaign (Jasiewicz and Jasiewicz-Betkiewicz 2006; Millard 2007) and became so strong that country experts spoke of a new divide having replaced the former regime divide (Szczerbiak 2006). It hindered the formation of a majority government between the PO and PiS, although the two parties had publicly stated before the election that they would form a government coalition.

It is worth considering whether it is possible to include the “personality factor” as a condition in the QCA. It is, however, very difficult to define and operationalise this very fuzzy and often informal factor. This is a problem that the whole of coalition research is confronted with and has not yet solved. De Winter states that he sees the only way out through “thick descriptions of government formations” (de Winter 2002, p. 205, emphasis included in the original). In this sense, the detailed and “thick” case descriptions included in the long version of this study (Keudel-Kaiser 2014) offer a first approximation to the inclusion of this factor.

The lack of a coalition partner sharing the main policy positions with the formateur (Lacking policy closeness): This condition plays an important role in the formation of minority governments. It has the highest (albeit still low) consistency threshold (0.79) and appears in two of the three equifinal paths. That is to say, in a number of cases, parties preferred forming a minority government (or remaining in opposition) to having to make policy compromises. This contradicts the assumption often referred to in the literature that in the young, unstable and often populist Central and Eastern European party systems, policy orientation does not play a major role (see, for example, Bos and Segert 2008a ). It affirms the results of recent research on Central and Eastern Europe, namely that ideological competition is central to any understanding of coalition formation and membership in these countries (Savage 2014).

One Central and Eastern European effect deserves special attention: In some cases, the regime divide overshadows policy closeness. The most impressive examples are two Polish cases, namely the governments formed after the 1993 and 1997 elections when the regime divide was still extraordinarily strong. In 1993, the post-communist formateur party, the Alliance of the Democratic Left (SLD), would have had enough close coalition partners in terms of policy: Two parties that originated from the post-oppositional camp, the Labour Union (UP) and the Democratic Union (UD), shared basic policy positions with the SLD, such as the position on church-state relations. The leader of the SLD pointed to these similarities and tried to convince the UD to join the coalition. But the UD and the UP were not willing to bridge the regime divide (Grzymala-Busse 2001, p. 95). Government formation after the 1997 election followed the same pattern: The formateur party, the post-oppositional Solidarity Electoral Action (AWS), would have had an ally relatively close in terms of policy in the Polish Peasants’ Party (PSL), but coalition formation was unthinkable because the PSL stemmed from the opposite camp. The final coalition partners, the AWS and the UW, both originated from the post-Solidarity camp but did not have much in common in terms of policy (Chan 1998). These examples point to two important aspects: the vital role of policy positions that should be treated more carefully in coalition research on Central and Eastern Europe, and the role of legacies from the communist past.

4.3.2 A closer look at single cases

From entire set of cases, I chose two opposite examples: the 1998 Czech government formation process where all four conditions are present (truth table row 1), and one of the outliers, the Latvian government that formed in 1998. The latter is the only minority government compared to four majority governments in a configuration theoretically linked to a majority government (truth table row 8).

Czech Republic 1998: The 1998 Czech government formation process can be seen as the “prime example”. It is the only case where all four conditions theoretically linked to the formation of a minority government are present. There was deep antagonism between the two dominating parties, the Civic Democratic Party (ODS), led by Václav Klaus, and the Social Democrats (ČSSD), led by Miloš Zeman, both controlling more than 30% of the seats (two-party dominance). This antagonism corresponded to an overwhelming left-right divide between the parties (strong divide). The unreformed communist party KSČM, which controlled 12% of the seats, was from the beginning excluded from government formation. Given its far leftist positions and its anti-EU attitude, the KSČM was ideologically distant from the “mainstream” parties, treated as a “pariah party” and was thus, in Cappocia’s (2002) terms, an “anti-system party” in the relational sense (presence of a high percentage of “non-coalitionable” parties). The ČSSD formateur party had no coalition partner on the left of the party system (lacking policy closeness) and in the end, the formation of a minority government was the only way forward. However, some intermediary steps came to light through the case study approach. The situation appeared to be deadlocked, but the small centre-right parties, the Freedom Union (US) and the Christian Democratic Union (KDU-ČSL), had some range of manoeuverability. They could have formed a coalition with the ODS, for example. But they delayed the government formation process, partly due to disagreements, partly for strategic reasons: After a number of negotiations failed, President Havel suggested that the formation of a government of experts might be the solution. This proved an attractive solution for the small parties and tempted them to delay the negotiation process further (Brokl and Mansfeldová 1999, p. 365). For the ODS and ČSSD, on the contrary, the formation of an expert government would have meant a loss of influence. It was likely primarily this “threat” that ultimately resulted in the surprising agreement between the ODS and ČSSD on the so-called “opposition pact”: The ODS tolerated a ČSSD minority government in exchange for a number of important ministerial positions. This example demonstrates that the conditions are generally well chosen and that the interplay of conditions can very well explain the formation of a minority government. At the same time, it shows that there are always additional, situational factors that can hardly be grasped by theory. This finding underlines the crucial importance of case studies in combination with QCA.

Latvia 1998: The 1998 Latvian government is the only minority government compared to four majority governments in a configuration theoretically linked to a majority government. Only one condition, strong divide, is present. Why did the government formation process end in a minority government? The parties of the conservative and liberal camp, the People’s Party (TP), Latvia’s Way (LC) and For Fatherland and Freedom/Latvian National Conservative Party (TB/LNNK), were the clear winners of the election and the formation of a majority government by these parties seemed the most obvious solution: All three parties were relatively close to each other in terms of policy and could be seen as “natural coalition partners” (Davies and Ozolins 2001, p. 137). What hindered the formation of a majority government? Strong personal antipathy existed between Andris Šķēle, leader of the People’s Party and Vilis Krištopāns, leader of Latvia’s Way. Krištopāns ruled out any coalition with the People’s Party. Against this background, President Guntis Ulmanis urged Krištopāns to attempt to form a government. Latvia’s Way was only the second strongest party, but in contrast to the People’s Party it was open to coalitions with parties to both its right and left. This led the president to believe that Latvia’s Way would be able to “command enough cross-party support to survive in the Saeima for the immediate future” (Davies and Ozolins 2001, p. 140), even if it was not able to build a majority government. Once again, this example demonstrates the importance of personal factors in coalition formation that sometimes overshadows policy closeness.

5 Conclusion

The analysis shows that the formation of minority governments in the eight Central and Eastern European countries (Bulgaria, the Czech Republic, Estonia, Latvia, Lithuania, Poland, Romania and Slovakia) can be coherently explained by the interplay of party system factors. Minority governments form when there is a strong bifurcation of the party systems, caused either by the dominance of two antagonistic parties or by a strong divide that structures party competition, combined with a high percentage of “non-coalitionable” parties or a lack of parties that share the main policy positions with the formateur party. This result does not necessarily mean that party system features are the only factors that play a role, but the analysis shows that it is possible to explain the formation of minority governments by focussing solely on party system features.

What does this mean? First, at a very general level it confirms the crucial role of parties in coalition formation. Second, it confirms the key role of parties in the political systems of the young Central and Eastern European democracies. In the two decades under investigation, the parties and party systems in Central and Eastern Europe underwent many changes. Although there is currently a trend toward stabilisation, the party systems are still very much in flux. Frequently, new parties appear, party alliances regroup or parties change their programmatic orientation. Therefore, government formation often takes place against an unpredictable background.

Does the result thus affirm the common hypothesis that the formation of minority governments in Central and Eastern Europe is first and foremost a less-than-ideal solution? To answer this question, the main characteristics of the party system conditions that proved to be decisive are reviewed once again. A bifurcation of the party system, either triggered by a strong divide or by two-party dominance, turned out to be a necessary condition for the formation of minority governments. This result confirms the important role of “a pattern of polarised competition between competing blocs of parties” in government formation in Central and Eastern Europe (Savage 2013b, p. 1046). The “blocking” of the party system by bifurcation, often aggravated by personal antagonism, at least partly confirms the widespread hypothesis that the formation of minority governments is an expression of problematic characteristics of the party systems. The influence of the third condition, the presence of “non-coalitionable parties”, corroborates this thesis. In five cases, the presence of non-coalitionable parties played a role in the formation of a minority government. In all five cases, these non-coalitionable parties were anti-system parties not only in the relational sense (that is to say, seen as distant by the “mainstream parties”), but also in the ideological sense (see Keudel-Kaiser 2014, Chap. 5): Their ideology was of “anti-democratic” character (see the above introduced concept by Capoccia (2002)). The fact that the seat share of these parties was so strong that it influenced government formation is alarming with regard to democracy.

The influence of the fourth condition, “lacking policy closeness”, highlights a second important aspect. In a number of cases, parties preferred to form a minority government (or to remain in opposition) so that they did not have to make policy compromises. This is an interesting result with regard to the on-going discussion in political science about whether “policy matters” in the competition between Central and Eastern European parties. The common opinion is that policy orientation does not play a major role in the young, unstable and often populist party systems of Central and Eastern European (see, for example, Bos and Segert 2008a) . Recent studies have shown that, compared to Western Europe, the ideological position of the parties plays only a minor role in government formation in Eastern Europe (see, for example, Döring and Hellström 2013, p. 684). However, other studies confirm the result of the analysis at hand that policy positions matter: With regard to the formation of minority governments, Grotz and Weber (2011) show that party leaders seem to attach more importance to ideological coherence than to a parliamentary majority (Grotz and Weber 2011, pp. 203–204). In his pioneering study on the influence of party positions on government formation in Central and Eastern Europe, Savage (2014) demonstrates that ideological competition is even one of the most influential factors determining coalition membership: “Previous research has indicated otherwise, largely due to a failure to take into account the specific ideological context of CEE countries” (Savage 2014, p. 548). I agree entirely: Clearly, “policy matters” in coalition formation in Central and Eastern Europe, but the existing datasets on policy positions are not suitable for the analysis of the party systems in these countries. I reacted to it with a different approach of determining party policy positions (see Sect. 3.4). Alternative measurements of parties’ policy positions must be advanced further.

How to proceed with future research?

Regarding method, Qualitative Comparative Analysis (QCA), which has seldom been used in coalition research thus far, proved an appropriate and compelling tool for the study of government formation that should be more frequently utilized. The configurative character of the QCA allowed for the elaboration of different paths leading to the outcome and acknowledged the interplay among factors in the process of coalition formation. The combination with case studies proved promising: It facilitated the opening of the “black boxes” of processes lying behind the different paths and revealed at least one additional factor – personality factors (mainly strong antipathy) – that plays an important role. This factor should be operationalised and integrated into an enhanced QCA analysis. The operationalisation of personality factors remains, however, a challenging task that coalition research as a whole should face.

Regarding content, it would be interesting to analyse the interplay of factors leading to the two other types of government, namely minimal winning and surplus coalitions. The formation of surplus coalitions that contain more coalition partners than numerically necessary is of great interest because like the formation of minority governments, these challenge basic assumptions of coalition theory. Examining other government types as well as different stages of the life cycle of governments will be important: The stability of governments in Central and Eastern Europe has attracted attention, but no studies focus specifically on the stability of minority governments. The empirical part of the detailed study upon which this article is based – the description of the 39 government formation processes (Keudel-Kaiser 2014, Chap. 5) – offers a broad pool for future case-based research.

The empirical developments show that research on minority governments is a topic that does not loose its relevance. To the contrary, given the on-going polarisation of party systems (and societies) in many European countries (see, for example, Pisano-Ferry 2015) often combined with the emergence of new (populist) parties (Hartleb 2011, p. 30), there is evidence that in the future the formation of minority governments will be the rule rather than the exception.