Coalition formation at the local level

Abstract

This article examines the determinants of coalition formation on the local level. In addition to standard office- and policy-seeking variables, we incorporate the local institutional setting and the constraints on local coalition politics emerging from patterns of party competition at the superior level of the political system. We test our expectations on the basis of a dataset providing information on the characteristics of potential and formed coalitions in 29 German cities. The results show that – even on the local level, which is often described as less politicised – not only office-seeking variables but also the ideological positioning of parties are good predictors for local coalition formation in German cities. Additionally, our findings suggest that local political actors take the party affiliation of the directly elected mayor into account when forming coalitions in local councils. The findings imply that political actors on all levels of political systems try to maximise their payoffs and form coalitions accordingly.

Keywords

coalition formation coalition theory institutional constraints local politics multi-level system

Introduction

While coalition theories have been frequently applied to the study of government formation on the national level (see, e.g. Martin and Stevenson, 2001, 2010; Strøm et al., 2010) and also to the analysis of government formation in the regional sphere (see, for example, Bäck et al., 2013; Downs, 1998; Ştefuriuc, 2009, 2013), research on party competition, coalition politics, and its impact on policy-making on the local level has been ‘terra incognita’ (Laver, 1989: 17). The application of coalition theories to the local level only recently gained momentum due to the notion that the local level is a promising research area to test coalition theories with new data, to increase the number of cases, and to hold the institutional context and ‘time’ constant (Camões and Mendes, 2009: 72; Laver et al., 1998: 352; Skjæveland et al., 2007: 726). All in all, the study of local coalition formation is one way to solve the problematic, ‘incestuous’ relationship in coalition research between theory development and data since there is only a limited number of national and regional coalitions (Laver, 1989: 16). In addition, turning the perspective to the local level, which is – at least in the country we study here, i.e. in Germany – often seen as a less politicised part of the political system and where decision-making is assumed to follow a more pragmatic, consensual way, we can analyse whether political actors even on this level of a political system adopt a utility-maximising strategy, as the standard models of coalition formation suggest (e.g. Axelrod, 1970; De Swaan, 1973; Laver and Shepsle, 1996; Riker, 1962; von Neumann and Morgenstern, 1944). If this is the case, then models on political decision-making that are based on rational choice theory receive empirical evidence from a level of the political system that is often described as ‘different’ from other spheres of a polity.

The aim of this paper is to identify the factors that drive the formation of coalitions on the local level. To restrict our analysis not only to office- and policy-seeking theories of coalition formation, we widen our theoretical and empirical perspective by incorporating incentives that emerge from the multi-level structure of a political system and further institutional constraints. We select the multi-level system of Germany and argue that – besides policy distances and office-seeking incentives – coalition formation on the local level is also influenced by institutional constraints. Particularly, the implementation of direct elections of the mayor and the abolishment of local councils’ rights to appoint the mayor strengthened the mayor’s political role. We argue that local political actors have an incentive to include the mayor’s party in a coalition to avoid situations of ‘cohabitation’, i.e. a situation where a coalition in the local council is opposed to the party of the directly elected mayor. Consequently, the party of the mayor should have an advantage in the local coalition formation game. Additionally, the relationship between the local and regional level sets specific incentives for the coalition formation process on the local level. Since local units in Germany depend on the support of the regional government to a significant degree, political actors should be more likely to include parties in a coalition on the local level that also hold key offices in the regional government.

Our findings, which are based on a completely new dataset that covers information on local party competition in 29 cities with more than 100,000 inhabitants in the German state of North Rhine-Westphalia, show that there is evidence that local political actors in Germany are office- and policy-seekers. Furthermore, there exists a ‘cohabitation effect’ with respect to the local coalition formation process. The directly elected mayor – or more precisely, the mayor’s party affiliation – is a key factor in the local coalition game. These findings remain stable after performing several robustness checks.

This finding sets incentives to refer to the local level when testing theories on political decision-making in specific institutional contexts like quasi-presidential or semi-presidential systems, where there are only few studies (see Clark et al., 2012: 506). The number of existing hybrid systems is not only small but also their institutional characteristics vary to a decisive degree, so that it is hard to isolate the key mechanisms that influence government formation and coalition politics. ‘Scaling down’ (Snyder, 2001) to the local level might help to solve this problem. Furthermore, our results add to our understanding of local coalition formation in Western European countries by analysing local coalition formation in Germany for the first time in a large-N quantitative design, and – in contrast to other studies (Bäck, 2003; Skjæveland et al., 2007) – by covering a 10-year time period.

The remainder of the paper is structured as follows. First, we give a brief overview of the local institutional setting, party competition, and coalition formation in Germany in general and in the selected 29 cities in particular. In the third section, we provide our main arguments regarding the analysis of local coalition formation. We discuss case selection, data, operationalization of the variables, and the applied statistical method in the fourth section. The fifth section presents the results and evaluates the hypotheses. The concluding section summarises the findings and discusses further research possibilities that emerge from focussing on the local level of political systems.

Institutional setting and patterns of political decision-making on the local level

The application of coalition theories to the coalition bargaining processes on the local level requires the recognition of office and policy payoffs for political actors, the identification of a coalitional arena, and the explanation of the ‘nature’ of local coalitions. Since our empirical analysis focuses on the local level in Germany – more precisely on coalition formation in 29 larger cities in the state of North Rhine-Westphalia – we briefly have to outline the institutional setting and patterns of party competition in which coalition formation on this level of the German political system takes place.

Despite the different local institutional settings in the German Länder, all local constitutions create incentives for political actors to compete over policy issues and local offices. German municipalities still have considerable rights to self-government and can shape the local policy. Especially during the economic and financial crisis, local political actors had to decide between budget consolidation and the preservation of public social spending. In addition, recent research shows that the ideological positions of parties are a key predictor on policy outcomes that deal with highly politicised issues such as the degree of registration fees for same-sex unions (Debus et al., 2013) or fees for Kindergarten places (Goerres and Tepe, 2013).

Concerning office payoffs, the German local political system does indeed hold some positions that can be seen as ‘functional equivalents’ to ministerial offices on the state and the national level. For example, heads of departments and deputy mayors are elected by the local council and therefore have to rely on the support of a parliamentary majority (Bogumil and Holtkamp, 2013: 40; Egner et al., 2013: 86; Mellors and Brearey, 1986: 288). Additionally, in most of the municipalities, the composition of committees depends on the allocation of seats in the local council. Hence, the largest party frequently can install one of its actors as committee chairperson. Since in most cases no single party is able to win a majority of seats in the local council, there is an incentive for parties to form coalitions on the local sphere in a way that they maximise their office benefits.

The German local government system has recently been called a ‘semi-presidential’ one due to the implementation of directly elected mayors (Bäck, 2005). In North Rhine-Westphalia, there exists a dual executive with the mayor (and the local administration) on the one hand and the local council on the other hand. This means that both the mayor and the local council are shaping the local policy agenda (Egner et al., 2013). The differences to parliamentary and presidential systems are obvious. In contrast to a parliamentary system, the local council and the mayor are elected separately (most of the time on different election days), and they both have their independent legitimation. Hence, the mayor does not rely on a local council’s majority, and the local council cannot deselect the mayor.¹ As distinct from pure presidential systems, the directly elected mayor (as functional equivalent to the president) has no possibility to choose his/her ‘ministers’, i.e. the deputy mayors and heads of departments. Consequently, this feature corresponds to Bäck’s (2005: 83) definition of a local semi-presidential system: “In semi-presidentialism […] the mayor would surround himself with a collective executive appointed by the council” (emphasis in original). All in all, the local council is the main sphere of activity for local parties and independent local lists. Turning to the process of local coalition formation, the local council is the coalitional arena in German local politics (see also Hogwood, 1999).

Nevertheless, local government coalitions are quite distinct from national and regional government coalitions in general (Laver, 1989). This is particularly true for the German case. Likewise to presidential and parliamentary systems, where the government (or parts of it) do not need the support of a parliamentary majority to get into or to remain in office (Clark et al., 2012; see also Bergman, 1993), the end of a coalition can result in the formation of a new coalition in the local councils. But there is no requirement to build a new coalition immediately. Instead, the local parties can also build ad hoc coalitions for specific policy issues.

This raises the question of why local political actors form coalitions in the first place, and why they are not exclusively dealing with ad hoc coalitions. Descriptive analyses of local coalition politics show that local political actors are forming coalitions and are simultaneously signing coalition agreements (which imply the prospect of a permanent cooperation) in order to reduce their bargaining and transaction costs (see, for example, Martinussen, 2002). Second, the signing of a coalition agreement sends a signal to local voters in a way that there will not be abrupt changes in the policy-making process. Thirdly, this comes with higher credibility and accountability (see Lupia and Strøm, 2010). To sum up, local coalitions with signed coalition agreements are not an equivalent to national or regional executive coalitions, but they have a more formal character than pure legislative ad hoc coalitions.

Theoretical considerations and hypotheses

Like in other European countries (see, e.g. Bäck, 2003; Skjæveland et al., 2007), parties play an important role in coalition formation on the local level in Germany. From a general perspective, political parties – the most important actors in modern democracies – are assumed to pursue three major goals: maximising the number of votes in an election, maximising their share of offices in a (coalition) government and maximising the number of enforced policy proposals (Strøm and Müller, 1999). Local politicians are office-, policy-, and vote-seeking actors as well (Skjæveland et al., 2007: 726). This perspective has straightforward implications for local coalition formation: parties should prefer coalitions that (1) have the support of a majority in parliament, (2) include a preferably small number of partisan actors, and (3) are ideologically cohesive. Therefore, coalitions on the local level should be more likely if they reflect the characteristics of key coalition theories that help to explain the coalition game outcome on other levels of the political system.

From an office-seeking point of view, coalitions should be more likely to form if they are minimal or minimum winning coalitions (Riker, 1962; von Neumann and Morgenstern, 1944). Coalitions are not expected to be oversized because otherwise the coalition parties’ share of cabinet offices decreases. Furthermore, parties should not only prefer coalitions with small majorities but also coalitions with the smallest number of parties in order to reduce transaction costs (Leiserson, 1966). In addition, non-cooperative game theorists like Austen-Smith and Banks (1988) stress the importance of the largest parliamentary party in the coalition game, which has the best chance to become a member of the government (see also Baron and Diermeier, 2001; Baron and Ferejohn, 1989). Policy-seeking theories, by contrast, focus on the policy preferences of political parties (as unitary actors) and predict coalitions that not only fulfil the criteria of office-seeking theories but also have the smallest programmatic distance on relevant policy dimensions (e.g. Axelrod, 1970; De Swaan, 1973). In line with the famous model by Downs (1957), coalition theorists argue that the party that controls the median legislator on the general left-right dimension has a pivotal position in the coalition formation process and is thus more likely to become a member of the next (coalition) government (Müller and Strøm, 2000; Saalfeld, 2000; Strøm, 1990; van Roozendaal, 1993). Furthermore, there is ample evidence that the status quo matters for the coalition formation game (see, e.g. Franklin and Mackie, 1983; Strøm et al., 1994; Warwick, 1996). In line with the current literature, we are expecting that incumbent coalitions should have an advantage in the local coalition formation game as well (Martin and Stevenson, 2010). This brings us to the following hypotheses derived from ‘classical’ theories of coalition formation: Coalitions on the local level should be more likely to form:

H1) if they are minimal or minimum winning coalitions;

H2) if they comprise the smallest number of parties (‘bargaining proposition’);

H3) if they include the largest party in the local parliament;

H4) the smaller the distance between the positions of local political actors is on key policy dimensions;

H5) if they include the party of the median legislator;

H6) if the respective party combination forms the current coalition in the local council.

The direct election of the mayors creates an additional institutional constraint for local political actors with regard to coalition formation. The mayors can shape the local policy agenda, in particular in their function as the head of the local administration. The relationship between the mayor and the local council is crucial for understanding the process of political decision-making, in particular since the mayors in North Rhine-Westphalia get directly elected in separate elections since 1999. The mayor does not have the right to appoint a local politician to form a coalition, but local political actors are likely to take into account the party affiliation of the mayor. It is not uncommon that directly elected mayors declare their coalition preferences in public. For example, the Christian Democratic mayors in the North Rhine-Westphalian cities Duisburg and Cologne stated that they are strongly against coalitions between the Social Democrats, the Greens, and The Left (see WAZ, 2009). To put it in other words, the mayors are against coalitions excluding their own party. Hence, the role of the directly elected mayor can be understood as being a ‘powerful player’ (cf. Strøm and Swindle, 2002). They can approve or disapprove a coalition proposal and influence the local coalition game, but their “consent is neither sufficient nor necessary” (Strøm and Swindle, 2002: 579) for the outcome of the coalition formation process.

Viewed from the perspective of local political parties, majority groups in the local council cannot be sure that they are going to bargain in everyday business with a mayor from their own party since voters might prefer – for whatever reason – a candidate from a smaller party in the mayoral election. Furthermore, note that in situations of opposing party memberships of local coalition members and of the directly elected mayor, local council majority groups often cut back responsibilities of directly elected mayors. In turn, mayors delay or even do not implement some council decisions (Bogumil and Holtkamp, 2013: 189). Therefore, such situations of ‘cohabitation’ could potentially increase bargaining and transaction costs, which will result in gridlock situations or at least in a slower process of political decision-making. The latter is likely to result in frustration among voters, who could punish the responsible parties – the ones forming the coalition and/or the parties supporting the mayor – in the upcoming elections. Consequently, parties should favour a political constellation that secures an efficient governing process since one key goal of political parties on all levels of the political system is to maximise their vote share (see, e.g. Strøm and Müller, 1999). Hence, we assume that local political actors take the mayor’s party affiliation into account and that they have an incentive to form coalitions including the mayor’s party in order to avoid a situation with opposite party orientations as well as to reduce transaction costs.² We formulate the following hypothesis:

H7) Coalitions on the local level should be less likely to form if they do not include the party of the mayor (‘cohabitation’).

The main theoretical approaches ignore more or less contextual features that also affect patterns of government formation, especially in ‘nested games’ (Tsebelis, 1990). However, the behaviour of parties that act in multi-level political systems is shaped by different kinds of incentives or pressure exerted by higher levels of the political system (see, e.g. Deschouwer, 2006; Downs, 1998; Müller, 2013; Swenden and Maddens, 2009; van Houten, 2009). Recent research argues that coalitions that are congruent with the partisan composition of government on the national sphere are significantly more likely to form in European multi-level systems (see, e.g. Bäck et al., 2013; Ştefuriuc, 2009, 2013). This is because congruent coalitions have some advantages for political actors in multi-level systems: they facilitate political decision-making in areas with shared competencies (Bolleyer, 2006; Hough and Jeffery, 2006; Ştefuriuc, 2009) and with multiple veto players like in federal systems (for the German case see, e.g. Fortunato et al., 2013).

However, local political actors are not represented in any kind of ‘second chamber’ at the regional level in Germany, where they could influence legislative decision-making. Therefore, there seems to be no need to form congruent coalitions to the regional government. Nevertheless, local units depend to a decisive degree on the (financial) support of regional governments. In order to increase their chances of receiving benefits from the regional government, incentives exist for local political actors to include those political parties in the coalition that are in charge of key positions in the regional government. In the case of North Rhine-Westphalia, the Ministry of the Interior is responsible for all local affairs. By considering ministerial discretion and by assuming that cabinet ministers can act as gatekeepers (Laver and Shepsle, 1996; for a recent analysis of portfolio allocation on the German state level see, e.g. Raabe and Linhart, 2013), a better connection to the state minister of the interior can be an advantage for local political actors by using intra-party channels for the purpose of getting important information, so that political actors from the respective local unit have some advantages in bargaining about financial grants (see, e.g. Ştefuriuc, 2009: 98). Therefore, we expect that:

H8) Coalitions on the local level are more likely to form if they include the party of the state minister of the interior.

Case selection, data, and methods

We test our expectations by studying coalition formation in North Rhine-Westphalian cities with at least 100,000 inhabitants. We define a local coalition as any cooperation between two (or more) parties (or independent local lists) that is based on a written and signed coalition policy agreement and that follows a local election.³ Many studies on local politics in Germany show that particularly the local councils in large cities can be seen as equivalents to national and regional parliaments with the ‘classical’ antagonism between majority and minority factions (see, e.g. Egner et al., 2013; Gunlicks, 1986). In addition, the impact of national and regional parties on local politics increases with the number of inhabitants (see Brearey, 1989; Fried, 1976).

The case selection of North Rhine-Westphalia for an application of coalition theories to local coalition formation in Germany has five advantages. First, compared to an analysis of more than one German local government system, we are able to hold the institutional setting constant in the sense that we are not coping with different types of local constitutions (see Laver et al., 1987: 503). Therefore, our research design has almost ‘laboratory-like conditions’ (Stecker, 2013). Second, local coalition formation does not take place in all German municipalities. Local politics in German municipalities is more and more politicised as the number of inhabitants is increasing, and the incentives to form coalitions are increasing as well. North Rhine-Westphalia is the most populated German state with almost 18 million inhabitants, and 29 out of 80 German cities with more than 100,000 inhabitants are located there. Third, the local government system of North Rhine-Westphalia can be seen as a ‘prototype’ of a competitive political system (Holtkamp, 2008). Local politics is characterised by a high politicisation of local parties and (to a lesser extent) by independent local lists. This results in conflicts between majority and opposition groups in the local council and leads to a coherent voting behaviour of parliamentary party groups. Fourth, concentrating on coalition formation on the local level in North Rhine-Westphalia allows for studying the potential impact of an additional institutional constraint in terms of the directly elected mayors and their party affiliations in the coalition bargaining process. Unlike the situation in Baden-Wuerttemberg or Bavaria, where mayors have always been elected directly, the implementation of direct mayoral elections in North Rhine-Westphalia in 1999 is relatively new. There should be a lesser extent of path dependence regarding the relationship between local council majorities and directly elected mayors, so that we are dealing with a least likely case to identify an impact of the partisan affiliation of the mayor on the outcome of the local coalition formation process. Fifth, local councils in North Rhine-Westphalia have a considerable amount of power because they are electing the deputy mayors and are defining their business circles without the interference of the mayor (Egner et al., 2013: 49).

We created a dataset that comprises information on local election results, on the allocations of seats in the local council, on local parties’ and independent local lists’ policy positions, and on the outcome of the coalition formation process in the local councils.⁴ Furthermore, we included information on the party affiliations of the directly elected mayors and of the state minister of the interior. The time period covered ranges from 1999 to 2009, so that our observation period starts in exactly the year when the first direct elections for the mayor positions were held in North Rhine-Westphalia.

For the application of office-seeking theories, we use data on the distribution of seats in the local councils in order to determine how many local coalitions could have been formed and which potential local coalitions are minimal or minimum winning coalitions and how many parties are parts of a potential coalition. Furthermore, we created an additional dummy variable indicating if a potential coalition includes the largest party in the local council, i.e. the party with the highest number of parliamentary seats. Further binary coded variables indicate if a potential coalition already exists, if it includes the party of the directly elected mayor⁵ and if it covers the party of the state minister of the interior.

Contrary to other studies on local coalition formation (see Denters, 1985; Laver et al., 1998; Steunenberg, 1992), we are not using national party positions as a proxy for local party positions. Following recent research on party positions on the German state level showing that there are decisive differences between the policy positions of the national parties compared to those of their regional branches (Bräuninger and Debus, 2012; Müller, 2009, 2013), we are expecting local party positions to differ as well since local parties address different electorates which are likely to have other policy preferences than the electorate on the regional or national level. Unfortunately, unlike for studies on local coalition formation in Sweden (Bäck, 2003) and Denmark (Skjæveland et al., 2007), there exists no expert survey information on the positions of local parties’ policy positions on key policy dimensions that cover our time period from 1999 to 2009.⁶

To determine policy positions of local parties, we examine party manifestos from the local level by using quantitative content analysis. On that basis, we derive positions of local parties and independent local lists on a general left-right axis and on an economic and a societal policy dimension – the two policy dimensions that structure party competition in Germany (e.g. Pappi, 2009) – by applying the Wordscores approach. Wordscores requires the identification of ‘reference texts’ and ‘reference scores’ (for a detailed description see Laver et al., 2003 and Lowe, 2008). We use the election manifestos of CDU, FDP, Greens, and SPD for the state elections in North Rhine-Westphalia in 1990, 1995, 2000, 2005, and 2010 as well as the election manifesto of The Left for the state election in 2010 as ‘reference texts’.⁷ The respective ‘reference scores’ for the ‘reference texts’ were assigned by using information on the policy positions of the North Rhine-Westphalian state parties with respect to the general left-right axis, the economic, and the societal dimension.⁸ The policy positions of German state parties are provided by a study on party competition in the German states (Bräuninger and Debus, 2012). We test the policy-seeking aspect by including, first, the policy-area specific veto player distance (Tsebelis, 2002) within each potential coalition and, second, by identifying the parliamentary party of the median legislator in the local council on the basis of the parties’ position on the general left-right dimension.⁹

We apply conditional logit (CL) models (McFadden, 1974) to analyse the determinants of local coalition formation. CL models have become the standard in the empirical analysis of government formation over the last decade (see, e.g. Martin and Stevenson, 2001, 2010) and have also been successfully applied to the analysis of local coalition formation (Bäck, 2003; Skjæveland et al., 2007). Each local coalition formation process is represented by a ‘choice set’ that comprises not only the actual formed coalition (which receives a value of one in the dependent variable) but also all other logically possible combinations of parties and independent local lists having gained seats in the local council (which are coded zero in the dependent variable). Therefore, the actual number of parties and independent local lists elected to local councils determines the size of the choice set.¹⁰

Results

Overall, 55 coalitions in North Rhine-Westphalian cities were formed in the time period under investigation (see Table 1). Since the implementation of direct mayor elections in 1999, 48 out of 55 local council coalitions have included the mayor’s party. There are only seven situations of ‘cohabitation’, where the mayor’s party is not a member of the coalition in the local council. This descriptive information already provides support for the hypothesis that parties in the legislature have an incentive to include the mayor’s party in the coalition and that local political actors try to avoid ‘cohabitation’. The party of the state minister of the interior is in 30 out of 55 cases part of the local coalition, giving support for our expectation that stressed the impact of ministerial discretion and intra-party communication on local coalition politics.

Table 1.

Characteristics of actually formed and potential coalitions in large North Rhine-Westphalian cities, 1999–2009.

	Formed coalitions		Potential coalitions
	Frequency	Share (percentage)	Frequency	Share (percentage)
Minimal winning	43	78.2	549	5.1
Minimum winning	13	23.6	300	2.8
Minimal winning with smallest no. of parties	44	80.0	1211	11.3
Including largest party	47	85.5	5248	48.9
Including median party (left-right dimension)	43	78.2	5784	53.9
Incumbent coalition	12	21.8^a	47	0.4^a
Including party of the mayor	48	87.3	5392	50.3
Including party of the state minister of the interior	30	54.6	5392	50.3
Total	55	100	10,729	100

^aThere had not been an incumbent coalition in 8 of 55 cases.

The overwhelming majority of local coalitions have been minimal winning coalitions. Minority and surplus majority coalitions are rarely formed (10 and two times, respectively). The average number of parties in local coalitions is 2.2. These descriptive findings provide evidence for our office-seeking hypotheses that local political actors do not only try to form minimal winning coalitions but rather minimal winning coalitions with the smallest number of parties included (see also Table A1 in the Appendix).

Figure 1 displays the local parties’ policy positions (and the policy positions of independent local lists, subsumed under the label ‘others’) in large North Rhine-Westphalian cities on the economic, the societal, and a general left-right dimension. Note that the policy position patterns on all three dimensions show a high resemblance to the positioning of German regional and national parties (see, e.g. Bräuninger and Debus, 2012; Linhart and Shikano, 2007, 2009; Müller, 2009, 2013; Pappi, 2009). The results also reveal that there is substantial diversity both between and within the programmatic orientation of local parties. It is therefore appropriate to rely on the policy positions of local parties in order to cover specific patterns of party competition in the respective cities instead of applying policy positions of parties from the regional or national level. Furthermore, the parties’ order from left to right based on our Wordscores estimation is identical to the self-assessments of local council members in German municipalities and cities (cf. Egner et al., 2013).

Figure 1.

Positions of local parties and independent local lists in North Rhine-Westphalian cities on three policy dimensions, 1999–2009.

Having shown that local political parties indeed adopt different programmatic positions in their election manifestos that vary within parties and across cities, the question is if these policy positions between parties have an impact on local coalition formation. Are local political actors policy-seeking and do they – from a general perspective – act in accordance with rational choice-based theories on maximizing the number of offices, despite the perception of local politics as less ideological and more pragmatic in terms of political decision-making? Or are local institutional, contextual, and multi-level features decisive for the outcome of coalition formation processes on the local level in the case under study here?

The results of the CL models are presented in Table 2.¹¹ Besides variables reflecting office-seeking theories of coalition formation as well as institutional and contextual constraints, the first three models include – as an indicator for the impact of policy-seeking approaches – the veto player distance between the two most extreme parties on the general left-right dimension and a variable identifying whether the median party on the left-right dimension is included in the coalition or not. Models 4 and 5 are based on a two-dimensional policy space and include the veto player distance between the most extreme parties in potential coalitions on the economic and societal policy dimension, respectively.

Table 2.

Conditional logit analyses of coalition formation in 29 North Rhine-Westphalian cities, 1999–2009.

	Model 1	Model 2	Model 3	Model 4	Model 5
Office-seeking factors
Minimal winning coalition	2.91***	2.70***	2.62***	2.85***	2.54***
	(0.57)	(0.57)	(0.57)	(0.58)	(0.60)
Minimum winning coalition	0.28	0.45	0.69	0.49	0.85⁺
	(0.45)	(0.47)	(0.51)	(0.46)	(0.51)
Bargaining proposition	1.99***	1.98***	2.14***	2.03***	2.24***
	(0.49)	(0.52)	(0.53)	(0.50)	(0.58)
Largest party included	0.89	0.55	0.71	1.04⁺	0.70
	(0.61)	(0.60)	(0.57)	(0.61)	(0.58)
Policy-seeking factors
Left-right policy distance	–0.27***	–0.28***	–0.28***
	(0.07)	(0.07)	(0.08)
Inclusion median party left-right dimension	0.38	0.24	0.18
	(0.45)	(0.47)	(0.47)
Economic policy distance				–0.09	–0.09
				(0.08)	(0.09)
Societal policy distance				–0.12*	–0.11*
				(0.05)	(0.05)
Institutional/contextual factors
Incumbent coalition	1.85***	1.84***	1.89***	1.89***	1.99***
	(0.52)	(0.53)	(0.53)	(0.57)	(0.56)
‘Cohabitation’		–1.21*	–0.96⁺		–1.04*
		(0.49)	(0.50)		(0.52)
Party of the state minister of the interior included			0.72⁺		0.63
Party of the state minister of the interior included			(0.40)		(0.39)
Number of potential coalitions	10,729	10,729	10,729	10,729	10,729
Formation opportunities	55	55	55	55	55
Pseudo R²	0.55	0.56	0.57	0.54	0.56
Log pseudolikelihood	120.90	117.51	115.66	124.27	119.14
AIC	255.80	251.01	249.32	262.53	256.28
IIA-test	0.83	0.87	0.88	0.92	0.91

Robust standard errors are shown in parentheses. ⁺ = p < 0.1; * = p < 0.05; ** = p < 0.01; *** = p < 0.001.

The results mostly support our hypotheses. The decision-making process is – even on the local level – about maximising the control over the number of offices. A potential coalition is more likely to form if it reflects the characteristics of a minimal winning coalition (von Neumann and Morgenstern, 1944) and especially of Leiserson’s (1966) bargaining proposition: local political actors prefer coalitions with as few parties as possible to reach a parliamentary majority. There is, however, no evidence that parties in the local parliament prefer minimum winning coalitions (Riker, 1962). These results corroborate findings for local coalition formations in Belgium (Geys et al., 2006) and the Netherlands (Steunenberg, 1992). The inclusion of the largest party into a potential coalition does not increase the chance that this potential coalition forms. The variable has the expected positive sign but does not reach conventional levels of statistical significance in all models presented in Table 2. Therefore, it seems not to be a common pattern in local politics to give the largest party in parliament the ‘right’ to make the first move to form a coalition. The findings remain robust when performing several robustness checks: even when deleting those potential coalitions that include one or more ‘anti-system’ or ‘pariah’ parties from the set of possible outcomes of the coalition formation process (see Ştefuriuc, 2013: 69, fn. 3) or when including office-seeking factors in the analysis only, the direction and statistical significance of the variables reflecting office-seeking coalition theories remain basically the same (see Tables A2 and A3 in the Appendix).¹²

The variables providing information on the veto player distance between the most extreme parties within a potential coalition have the expected negative direction. However, only the distances measured on the basis of the general left-right axis (models 1, 2, and 3) and the societal policy dimension (models 4 and 5) reach standard levels of statistical significance. Overall, potential coalitions are more likely to form if they comprise parties that are ideologically close to each other. When excluding potential coalitions that comprise parties with extremist policy positions from the analysis a priori, the statistically significant and negative effect of the variable reflecting the societal policy distance disappears, while the distance on the overall left-right dimension remains negative and significant according to models 2 and 9 in Table A3. The inclusion of the party that covers the median legislator on the general left-right axis does not significantly increase the chances that a theoretically possible coalition forms in reality.

We also find that throughout all the models incumbent local coalitions have a significantly higher chance of being the next coalition than other potential coalitions. This result is in line with the ones for local coalition formation processes in other European countries (see Bäck, 2003; Skjæveland et al., 2007).

Furthermore, the results provide empirical evidence for our expectation regarding the impact of the partisan affiliation of the directly elected mayor on the outcome of the local coalition formation processes: Directly elected mayors and their party affiliation play a key role for the outcome of the coalition formation process in the local government system under study here. The estimated effect of the variable that identifies potential coalitions that do not include the mayor’s party (‘cohabitation’) has the expected negative sign and is statistically significant at the standard level of statistical significance (see Table 2). These findings are corroborated by the models presented in Tables A2 and A3 in the Appendix: even when including only a few of the identified independent variables and/or when excluding all coalitions that include at least one ‘anti-system’ party from the analysis, potential coalitions excluding the party of the mayor have a smaller chance to form. Hence, studies on local politics and policy-making – at least in Germany and in similarly structured polities – should consider the role of the mayor as a decisive actor in the process of political decision-making.

Turning to multi-level factors, which are introduced in models 3 and 5 in Table 2, the estimated coefficients show mixed evidence for our expectations that local political actors in North Rhine-Westphalian cities include the party of the state minister of the interior, who is in charge of local affairs in the state government. The estimated effect is positive and statistically significant at the 90% level in model 2, but fails to reach the significance level mentioned last in model 5. Since the effect does not remain statistically significant from zero when looking at the models presented in Tables A2 and A3, we refrain from saying that this particular multi-level factor is decisive for coalition formation in the local council. In order to evaluate this hypothesis adequately, it is required to increase the number of observations under study.

Conclusion

The purpose of this study was to ‘scale down’ (Snyder, 2001) and to analyse the determinants of coalition formation on the local level by using a new data set on local election results and local coalition formation. Building on ‘classical’ theories of coalition formation, we argued that local political actors are additionally constrained in their search for a coalition by institutional and contextual factors. Our findings confirmed most of our expectations and showed that political actors from the local level follow by and large the same utility-maximising strategies as their fellow politicians acting on other, superior levels of a political system. Office-seeking motivations play a role in local coalition formation processes. We find evidence that local political actors rather prefer to form minimal winning coalitions with the smallest number of parties than to form minority or surplus majority coalitions. Moreover, coalitions are more likely to form if the ideological distance between parties decreases. In addition, incumbent coalitions have a greater chance of becoming the next coalition as well.

Second, our findings reveal that studies on coalition formation in hybrid settings like the local level in the selected cases of North Rhine-Westphalia should consider the role of the head of the executive – in our case, the directly elected mayor – as a decisive or ‘powerful player’ (Strøm and Swindle, 2002) in the coalition formation game. As the results and the robustness checks on the basis of different models and of a smaller database indicated, parties are less likely to form a coalition when the respective coalition does not include the mayor’s party. However, we do not know whether this is because of the mayor’s potential agenda-setting power, their indirect formateur role, or whether it is in the interest of the legislators to incorporate the MPs from the mayor’s party into a coalition. In order to understand the causal process leading to the inclusion of mayor’s parties into local coalitions, a small-N strategy that investigates typical and deviant cases on the basis of the predictions of the various models in this study seems worthwhile (see, e.g. Bäck and Dumont, 2007; Lieberman, 2005).

These findings give rise to a whole host of additional research questions that further studies could address. The empirical finding that political actors most often form coalitions with the smallest number of parties seems plausible because there are only a finite number of offices to allocate on the local level. Given that local political actors always have the possibility to form ad hoc coalitions in the local council, it seems worthwhile to analyse why local parties are forming coalitions at all. Therefore, one should further investigate the reasons or ‘conditions’ for the forming of coalitions on the basis of a written and signed coalition agreement.

Shifting the focus to the local sphere provides enough cases for performing empirical tests of theoretical arguments and models by concurrently holding the institutional setting constant. This provides incentives to study patterns of coalition politics and governance in presidential or semi-presidential systems – like the role of the head of state in the coalition bargaining process – in a quantitative manner by referring to the local level as a proxy. One could, for instance, study the determinants of government formation in hybrid or semi-presidential systems which differ in their institutional structure – the degree of power provided to the head of state by the constitution – to a significant degree, so that it is complicated to isolate effects emerging from independent variables that researchers are interested in, like for instance the influence of the head of a state on patterns of political decision-making.

Furthermore, the findings for large North Rhine-Westphalian cities presented here set incentives both for a broader analysis of the determinants of local coalition formation within the states of Germany, which still differ to some degree with regard to their local institutional setting, and for cross-country studies by including various European countries. Recent local government reforms in European countries show that there is a great variety of local competencies with regard to political decision-making (see Loughlin et al., 2011). Moreover, most of these local government reforms implemented direct elections of mayors. Therefore, we expect to find similar results regarding the impact of the mayor’s preferences in local decision-making processes in other European countries.

Footnotes

Acknowledgements

We wish to thank the participants of the panel ‘Coalition Formation, Stability and Policy Making’ at the 3rd EPSA annual conference, the participants of the panel ‘Local and Urban Politics in Multilevel Systems’ at the SPSA annual convention 2014, and the Party Politics anonymous reviewers for very helpful comments on previous versions of this article. In addition, we thank Anna Adendorf, Yannik Buhl, Hatice Kücük, Samuel Müller, Torben Schütz, Kenneth Stiller and Lea Maria Straβheim for assisting in the preparation of this study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental material

The online appendices are available at .

Notes

References

Austen-Smith

Banks

(1988) Elections, coalitions, and legislative outcomes. American Political Science Review 82(2): 405–422.

Axelrod

(1970) Conflict of Interest – A Theory of Divergent Goals with Applications to Politics. Chicago, IL: Markham Publishing Company.

Bäck

(2003) Explaining and predicting coalition outcomes: Conclusions from studying data on local coalitions. European Journal of Political Research 42(4): 441–472.

Bäck

(2005) The institutional setting of local political leadership and community involvement. In: Haus

Heinelt

Stewart

(eds) Urban Governance and Democracy: Leadership and Community Involvement. London and New York: Routledge, pp. 65–101.

Bäck

Dumont

(2007) Combining large-n and small-n strategies: The way forward in coalition research. West European Politics 30(3): 467–501.

Bäck

Debus

Müller

. (2013) Regional government formation in varying multilevel contexts: A comparison of eight European countries. Regional Studies 47(3): 368–387.

Baron

Diermeier

(2001) Elections, governments, and parliaments in proportional representation systems. Quarterly Journal of Economics 116(3): 933–967.

Baron

Ferejohn

(1989) Bargaining in legislatures. American Political Science Review 83(4): 1181–1206.

Benoit

Laver

(2006) Party Policy in Modern Democracies. London: Routledge.

10.

Bergman

(1993) Formation rules and minority governments. European Journal of Political Research 23(1): 55–66.

11.

Bogumil

Holtkamp

(2013) Kommunalpolitik und Kommunalverwaltung. Eine praxisorientierte Einführung. Bonn: Bundeszentrale für politische Bildung.

12.

Bogumil

Grohs

Holtkamp

(2010) Zersplitterte Kommunalparlamente oder Stärkung lokaler Demokratie? Warum die Abschaffung der kommunalen Fünfprozenthürde in Nordrhein-Westfalen ein Fehler war. Zeitschrift für Parlamentsfragen 41(4): 788–802.

13.

Bolleyer

(2006) Intergovernmental arrangements in Spain and Swiss federalism: The impact of power-concentrating and power-sharing executives on intergovernmental institutionalization. Regional and Federal Studies 16(4): 385–408.

14.

Bräuninger

Debus

(2012) Parteienwettbewerb in den deutschen Bundesländern. Wiesbaden: VS Verlag für Sozialwissenschaften.

15.

Brearey

(1989) City coalitions in West Germany: A case-study of Nordrhein-Westfalen. In: Mellors

Pijnenburg

(eds) Political Parties and Coalitions in European Local Government. London: Routledge, pp. 275–300.

16.

Camões

Mendes

(2009) Party politics and local government coalition formation in Portugal. In: Giannetti

Benoit

(eds) Intra-Party Politics and Coalition Governments in Parliamentary Democracies. London and New York: Routledge, pp. 69–85.

17.

Clark

Golder

(2012) Principles of Comparative Politics. Washington, DC: CQ Press.

18.

De Swaan

(1973) Coalition Theories and Cabinet Formations: A Study of Formal Theories of Coalition Formation Applied to Nine European Parliaments After 1918. Amsterdam: Elsevier.

19.

Debus

Knill

Tosun

(2013) Registration fees for same-sex unions, local party politics and societal demand. Local Government Studies 39(6): 756–776.

20.

Denters

(1985) Towards a conditional model of coalition behaviour. European Journal of Political Research 13(3): 295–309.

21.

Deschouwer

(2006) Political parties as multi-level organizations. In: Katz

Crotty

(eds) Handbook of Party Politics. London: Sage, pp. 291–300.

22.

Downs

(1957) An Economic Theory of Democracy. New York: Harper.

23.

Downs

(1998) Coalition Government, Subnational Style: Multiparty Politics in Europe’s Regional Parliaments. Columbus, OH: Ohio State University Press.

24.

Egner

Krapp

M-C

Heinelt

(2013) Das deutsche Gemeinderatsmitglied. Problemsichten – Einstellungen – Rollenverständnis. Wiesbaden: Springer VS.

25.

Fortunato

König

Proksch

S-O

(2013) Government agenda-setting and bicameral conflict resolution. Political Research Quarterly 66(4): 938–951.

26.

Franklin

Mackie

(1983) Familiarity and Inertia in the Formation of Governing Coalitions in Parliamentary Democracies. British Journal of Political Science 13(3): 275–298.

27.

Fried

(1976) Party and policy in west German cities. American Political Science Review 70(1): 11–24.

28.

Geys

Heyndels

Vermeir

(2006) Explaining the formation of minimal coalitions: Anti-system parties and anti-pact rules. European Journal of Political Research 45(6): 957–984.

29.

Glasgow

Golder

(2012) New empirical strategies for the study of parliamentary government formation. Political Analysis 20(2): 248–270.

30.

Goerres

Tepe

(2013) Für die Kleinen ist uns nichts zu teuer? Kindergartengebühren und ihre Determinanten in Deutschlands 95 bevölkerungsreichsten Städten zwischen 2007 und 2010. dms – der moderne Staat 6(1): 169–190.

31.

Gunlicks

(1986) Local Government in the German Federal System. Durham, NC: Duke University Press.

32.

Hogwood

(1999) Playing to win: Adapting concepts of rationality and utility for the German coalition context. In: Sturm

Kropp

(eds) Hinter den Kulissen von Regierungsbündnissen: Koalitionspolitik in Bund, Ländern und Gemeinden. Baden-Baden: Nomos, pp. 15–43.

33.

Holtkamp

(2008) Kommunale Konkordanz- und Konkurrenzdemokratie. Parteien und Bürgermeister in der repräsentativen Demokratie. Wiesbaden: VS Verlag für Sozialwissenschaften.

34.

Hough

Jeffery

(eds) (2006) Devolution and Electoral Politics. Manchester: Manchester University Press.

35.

Kang

S-G

(2009) The influence of presidential heads of state on government formation in European democracies: Empirical evidence. European Journal of Political Research 48(4): 543–572.

36.

Laver

(1989) Theories of coalition formation and local government coalitions. In: Mellors

Pijnenburg

(eds) Political Parties and Coalitions in European Local Government. London: Routledge, pp. 15–33.

37.

Laver

Shepsle

(1996) Making and Breaking Governments: Cabinets and Legislatures in Parliamentary Democracies. Cambridge and New York: Cambridge University Press.

38.

Laver

Benoit

Garry

(2003) Extracting policy positions from political texts using words as data. American Political Science Review 92(2): 311–331.

39.

Laver

Rallings

Thrasher

(1987) Coalition theory and local government: Coalition payoffs in Britain. British Journal of Political Science 17(4): 501–509.

40.

Laver

Rallings

Thrasher

(1998) Policy payoffs in local government. British Journal of Political Science 28(2): 333–353.

41.

Leiserson

(1966) Coalitions in Politics. PhD thesis, Yale University, USA.

42.

Lieberman

(2005) Nested analysis as a mixed-method strategy for comparative research. American Political Science Review 99(3): 435–452.

43.

Linhart

Shikano

(2007) Die Generierung von Parteipositionen aus vorverschlüsselten Wahlprogrammen für die Bundesrepublik Deutschland(1949–2002). Mannheim: MZES Working Paper No. 98.

44.

Linhart

Shikano

(2009) Ideological signals of German parties in a multidimensional space: An estimation of party preferences using the CMP data. German Politics 18(3): 301–322.

45.

Loughlin

Hendriks

Lidström

(eds) (2011) The Oxford Handbook of Local and Regional Democracy in Europe. Oxford: Oxford University Press.

46.

Lowe

(2008) Understanding wordscores. Political Analysis 16(4): 356–371.

47.

Lupia

Strøm

(2010) Bargaining, transaction costs, and coalition governance. In: Strøm

Müller

Bergman

(eds) Cabinets and Coalition Bargaining: The Democratic Life Cycle in Western Europe. Oxford: Oxford University Press, pp. 51–83.

48.

McFadden

(1974) Conditional logit analysis of qualitative choice behavior. In: Zarembka

(ed.) Frontiers in Econometrics. New York: Academic Press, pp. 105–142.

49.

Martin

Stevenson

( 2001) Government formation in parliamentary democracies. American Journal of Political Science 45(1): 33–50.

50.

Martin

Stevenson

(2010) The conditional impact of incumbency on government formation. American Political Science Review 104(3): 503–518.

51.

Martinussen

(2002) In search of the government in local government: Coalition agreements and office payoffs in Norway. Scandinavian Political Studies 25(2): 139–171.

52.

Mellors

Brearey

(1986) Multi-dimensional approaches to the study of local coalitions: Some cross-national comparisons. In: Pridham

(ed.) Coalitional Behaviour in Theory and Practice: An Inductive Model for Western Europe. Cambridge: Cambridge University Press, pp. 278–295.

53.

Müller

(2009) The impact of the socio-economic context on the Länder parties’ policy positions. German Politics 18(3): 365–384.

54.

Müller

(2013) On a short leash? Sub-national party positions between regional context and national party unity. Journal of Elections, Public Opinion and Parties 23(2): 177–199.

55.

Müller

Strøm

(eds) (2000) Coalition Government in Western Europe. Oxford: Oxford University Press.

56.

Pappi

(2009) Regierungsbildung im deutschen Fünf-Parteiensystem. Politische Vierteljahresschrift 50(2): 187–202.

57.

Raabe

Linhart

(2013) Does substance matter? A model of qualitative portfolio allocation and application to German state governments between 1990 and 2010. Party Politics. DOI: 10.1177/1354068813487107.

58.

Riker

(1962) The Theory of Political Coalitions. New Haven and London: Yale University Press.

59.

Saalfeld

(2000) Coalitions in Germany: Stable parties, chancellor democracy and the art of informal settlement. In: Müller

Strøm

(eds) Coalition Government in Western Europe. Oxford: Oxford University Press, pp. 32–85.

60.

Skjæveland

Serritzlew

Blom-Hansen

(2007) Theories of coalition formation: An empirical test using data from Denmark. European Journal of Political Research 46(5): 721–745.

61.

Snyder

(2001) Scaling down: The subnational comparative method. Studies in Comparative International Development 36(1): 93–110.

62.

Stecker

(2013) How effects on party unity vary across votes. Party Politics. DOI: 10.1177/1354068813509514.

63.

Ştefuriuc

(2009) Government formation in multi-level settings: Spanish regional coalitions and the quest for vertical congruence. Party Politics 15(1): 93–115.

64.

Ştefuriuc

(2013) Government Formation in Multi-Level Settings: Party Strategy and Institutional Constraints. Basingstoke: Palgrave Macmillan.

65.

Steunenberg

(1992) Coalition theories: Empirical evidence for Dutch municipalities. European Journal of Political Research 22(3): 245–278.

66.

Strøm

(1990) Minority Government and Majority Rule. Cambridge: Cambridge University Press.

67.

Strøm

Müller

(eds) (1999) Policy, Office, or Votes? How Political Parties in Western Europe Make Hard Decisions. Cambridge: Cambridge University Press.

68.

Strøm

Swindle

(2002) Strategic parliamentary dissolution. American Political Science Review 96(3): 575–591.

69.

Strøm

Budge

Laver

(1994) Constraints on cabinet formation in parliamentary democracies. American Journal of Political Science 38(2): 303–335.

70.

Strøm

Müller

Bergman

(eds) (2010) Cabinets and Coalition Bargaining: The Democratic Life Cycle in Western Europe. Oxford: Oxford University Press.

71.

Swenden

Maddens

(eds) (2009) Territorial Party Politics in Western Europe. London: Palgrave.

72.

Tsebelis

(1990) Nested Games: Rational Choice in Comparative Politics. Berkeley, CA: University of California Press.

73.

Tsebelis

(2002) Veto Players: How Political Institutions Work. Princeton, NJ: Princeton University Press.

74.

Van Houten

(2009) Multi-level relations in political parties: A delegation approach. Party Politics 15(2): 137–156.

75.

Van Roozendaal

(1993) Cabinets in the Netherlands (1918–1990): The importance of “dominant” and “central” parties. European Journal of Political Research 23(1): 35–54.

76.

Von Neumann

Morgenstern

(1944) Theory of Games and Economic Behaviour. Princeton, NJ: Princeton University Press.

77.

Warwick

(1996) Coalition government membership in west European parliamentary democracies. British Journal of Political Science 26(4): 471–499.

78.

WAZ (2009) Adolf Sauerland will Koalition der Vernunft. Available at: www.derwesten.de/staedte/duisburg/adolf-sauerland-will-koalition-der-vernunft-id158797.html (accessed 5 December 2013).