FROM THE PERIPHERY TO POWER Explanations for the government participation of niche parties in West European parliamentary democracies

Transcription

1 FROM THE PERIPHERY TO POWER Explanations for the government participation of niche parties in West European parliamentary democracies Sarah L. de Lange European University Institute * ASTRACT Niche parties have joined government coalitions in many West European parliamentary democracies in recent years. This paper investigates the rise to power of these parties using a new dataset that includes information about niche parties in fourteen countries and concludes that the factors that explain the recent government participation of niche parties are almost identical to those that explain the government participation of mainstream parties. Like mainstream parties, niche parties assume office when the formateur party are small, when they are located in close proximity of this party, and when they are politically experienced, both in parliament and in government. These factors also can account for the fact that it has taken niche parties some time to make the transition from opposition to government, that this transition has not (yet) occurred in all countries in which these parties have been represented, and that they remain less likely to govern than mainstream parties. KEY WORDS: niche parties; coalition governments; coalition membership theories. Paper prepared for presentation at the ECPR Joint Sessions, Lisbon, April 2009 * Jean Monet Fellow, Robert Schuman Centre for Advanced Studies, European University Institute, Via delle Fontanelle 19, San Domenico di Fiesole, Italy, sarah.delange@eui.eu. I would like to thank the participants of the Colloquium on Political ehaviour at the European University Institute for their valuable comments on an earlier draft of this paper. I am grateful to Catherine de Vries, Marcel van Egmond and Tom van der Meer for their statistical advice.

2 Introduction Niche parties have become powerful electoral contestants in post-war Western Europe. In the 1980s and 1990s green, radical right, and regionalist parties emerged as successful electoral forces in most West European countries, but for many years these parties have been unable to transform their electoral success into governmental power. Only in recent years a limited number of niche parties have been invited to join government coalitions, with the Freiheitliche Partei Österreichs (FPÖ) in Austria and die Grünen in Germany as the prime examples. Other successful niche parties, like the Vlaams elang (V) in elgium and GroenLinks (GL) in the Netherlands, remain part of the opposition. This observation raises many questions. First and most importantly, what explains the government participation of niche parties, and what explains the national and temporal situation of this phenomenon? This paper addresses this question on the basis of an analysis of the determinants of the coalition membership of niche parties. Second, how does the government participation compare to that of mainstream parties? To answer this question the paper compares niche parties to mainstream parties and establishes whether niche parties are less likely to participate in government coalitions than mainstream parties. It also examines whether the determinants of coalition membership of niche parties are different from those of mainstream parties. Finally, it addresses the more general question whether niche parties are less attractive coalition parties than mainstream parties. None of these questions has been answered yet on the basis of a systematic investigation of the government participation of niche parties. 2 This paper seeks to fill this void. It addresses the listed questions on the basis of a new dataset that includes information about niche and mainstream parties in fourteen West European parliamentary democracies. Focusing on parties that have been represented in West European parliamentary democracies between 1981 and 2008, the paper explores potential explanations for the (lack of) government participation of niche parties. Through studying the composition of coalition governments, and changes therein, the motives that guide parties behaviour in the coalition process can be uncovered. More specifically, it helps identify the circumstances under which mainstream parties are willing to change coalition practices and niche parties are willing to assume government responsibility. As such, the paper directly addresses the consequences of the emergence of niche parties in 2 Traditionally, the government participation of niche parties has been studied on the basis of (collections of) case studies, but see Dumont and äck (2006) on the government participation of green parties and De Lange (2008) on the government participation of radical right parties. 1

3 West European parliamentary democracies by focusing on their impact on patterns of coalition formation, and hence their broader impact on party systems. Moreover, the paper touches upon a series of questions that have important normative implications. How long does it take niche parties to make the transition from pariah to power? Are niche parties deliberately kept out of office by mainstream parties that have formed cartels? This paper consists of four parts. First, the recent government participation of niche parties is investigated in more detail. Second, potential explanations for the government participation of these parties are presented and hypotheses about the government participation of niche parties formulated. Third, the research design, data, and method used to test these hypotheses are presented and particular methodological issues that come up when studying niche parties are addressed. Fourth, the hypotheses are tested and the implications of the results of these tests explored. Niche parties in government Niche parties have benefited from increasing scholarly attention in recent years. Meguid (2008: 4-5) argues that niche parties have three defining characteristics. They reject the traditional class-based orientation of politics, raise issues that are novel and do not coincide with the existing left-right lines of political division, and differentiate themselves from mainstream parties by limiting their issue appeals to one issue only. According to Meguid (2008: 4), the group of niche parties, in which she includes green parties, radical right parties, and regionalist parties, consists of a set of political actors quite unlike those seen before. 3 Until recently, most studies that focus on niche parties have primarily been concerned with the behaviour of these parties in the electoral arena. Scholars have investigated whether niche parties react differently to changes in public opinion than mainstream parties (Adams et al. 2006), whether they benefit electorally from policy radicalization (Ezrow 2008), and how mainstream parties react to or attempt to counter the electoral success of niche parties (Greene 2008; Meguid 2005). These studies univocally conclude that niche parties are different from mainstream parties, in the sense that they face different political trade-offs and hence make different strategic decisions than mainstream parties. Scholars have not fully come to terms with the implications these findings have for the study of party behaviour outside of the 3 Other scholars take a 'big tent approach (ale and Dunphy forthcoming: 4) and also include communist and other radical left parties in the group of niche parties. Adams et al. (2006: 513) see niche parties as members of party families who present either an extreme ideology (such as Communist and extreme nationalist parties) or a noncentrist niche ideology (i.e., the Greens). Thus, for these scholars it is not essential that niche parties reject traditional conceptions of left and right, as long as they are sufficiently different from mainstream parties. 2

4 electoral arena. If niche parties behave differently from mainstream parties in the electoral arena, they are also likely to behave differently in the parliamentary and governmental arenas. After all, there are strong linkages between the three arenas (Narud 1996; Strom 1990). Moreover, the studies of Meguid show that mainstream parties interact differently with niche parties than with other mainstream parties, which has profound implications for the way in which these parties cooperate in parliament and government. Hence, it is imperative to also examine the interaction between mainstream and niche parties in other arenas than the electoral. This paper therefore investigates the government participation, or absence thereof, of niche parties. On the basis of a quick count Meguid (2008: 5) estimates that almost 12 per cent of niche parties have at some point in time participated in government coalitions, and half of these parties were crucial to the formation of a majority government. Table 1 provides a more systematic overview of the West European niche parties that have participated in government coalitions. It lists the niche parties that have been fortunate enough to join a government coalition, identifies the party families to which these parties belong, the periods during which they have been in government, and details whether these parties were fully fledged cabinet members, or served merely as government support parties. The overview demonstrates that sixteen niche parties became junior coalition members during the period under investigation and that most niche parties that crossed the threshold of government representation have been in office more than once. 4 Moreover, it also shows that in most West European parliamentary democracies niche parties have only joined government coalitions in recent years and in many countries more than one niche party has been able to penetrate the coalitional market. The Freiheitliche Partei Österreich (FPÖ), for example, rose to power in 2000 and despite the controversy the government participation of the radical right party created, it continued in office after the 2002 elections. In elgium the Flemish and Walloon green parties, Anders Gaan Leven (AGALEV) and ECOLO, and the Flemish regionalist parties, the Nieuw-Vlaamse Alliantie (N-VA) and Volksunie (VU), have been junior coalition members in 1999, 2007, and 1987 respectively. Danish niche parties have been unable to become junior coalition members, but the Dansk Folkeparti (DF) has supported several minority governments. In Finland the green Vihreä Liitto (VIHR) joined the government coalition that was formed after the 1995 elections and returned to government in Moreover, Finnish governments have always included the Svenska Folkepartiet i Finland (SFP), an 4 Niche parties have never managed to secure formateur status in coalition negotiations or succeeded in obtaining the prime ministership. 3

5 ethnonationalist party that promotes the interests of the Swedish minority that lives in Finland. In France Les Verts (VERTS), the French green party, participated in the government coalition formed by the Parti Socialiste (PS) in 1997, while in Germany the greens, or die Grünen (GRÜNE), joined a government coalition led by the Sozialdemokratische Partei Deutschlands (SPD) in The latter coalition was reformed in 2002 when the two parties managed to maintain their parliamentary majority. In Ireland the Green Party (GP) was invited to join Fianna Fáil (FF) in government after the most recent elections, which took place in Given the complexity and fluidity of the Italian party system it should come as no surprise that many niche parties have assumed office in this country. The Alleanza Nazionale (AN), Lega Nord (LN), Movimento per l Autonomia (MPA), and the Federazione dei Verdi (VERDI) each have participated in one or more coalition government. In the Netherlands the Lijst Pim Fortuyn (LPF) has been a junior coalition member in 2002, whereas the Norwegian Fremskrittspartiet (FRP), a radical right party that has been around for several decades, never made it into government but did support the minority government that assumed in Finally, in Sweden niche parties have never been cabinet members, but several of these parties (Miljöpartiet de Gröna, or MP) has acted as government support party as well. In Greece, Iceland, Luxembourg, Portugal, and Spain the coalitional market has been completely closed for niche parties. Parties like Íslandshreyfingin Lifandi Land (IH) and Vinstrihreyfingin Grænt framboð (VG) in Iceland, Déi Gréng (GRÉNG) in Luxembourg and the Partito Ecologista Os Verdes (PEV) in Portugal have never been in government and have not acted as government support parties either. <<Table 1 about here>> The overview presented in Table 1 suggests that there are two important patterns of variation in the government participation of niche parties, the first cross-national, the second cross-temporal. Table 2 contains details about the pattern of cross-national variation and demonstrates that niche parties are more likely to participate in government coalitions in some countries than in others. In Austria (15.8 per cent), Finland (71.4 per cent), France (16.7 per cent), Germany (28.6 per cent), and Italy (22.9 per cent) the government participation of niche parties has been above average, while in the other West European countries it has been below average. Especially the case of Finland stands out here, because the SFP, the party that represents the Swedish minority in Finland, has been part of nearly every single cabinet that has been formed since

6 <<Table 2 about here>> In addition to a clear pattern of cross-national variation one can also observe a clear pattern of temporal variation, of which the details are presented in Table 3. The government participation of niche parties clearly increases over time, most notably since the early 1990s. During the 1980s the percentage of niche parties that is a junior coalition member remains below 10.0 per cent, to rise to 11.9 per cent for the period and 22.2 per cent for the period From 2001 to 2005 this figure diminishes to 15.0 per cent, to climb to 18.9 per cent again for the period << Table 3 about here>> In sum, Tables 2 and 3 reveal that niche parties have participated in government coalitions in roughly half of the countries in which they have been represented, mostly during the 1990s and the 21 st century. The tables also show that the percentage of niche parties that has been in government is significantly lower than that of mainstream parties. Of the 173 junior coalition members that participated in West European government coalitions between 1981 and 2008, 143 were mainstream parties and 30 niche parties. Although the number of niche parties that has been represented in West European parliaments during this period is significantly lower than the number of mainstream parties, the difference is nevertheless noteworthy. Of the 518 mainstream parties 27.6 per cent has been a junior coalition member, while this has only been the case for 14.4 per cent of the 208 niche parties. These figures indicate that niche parties are less likely than mainstream parties to become junior coalition members. Of course, the cross-national and cross-temporal patterns of variation in the government participation of niche parties should be compared to those of the government participation of mainstream parties as well. This comparison of that especially in Austria (difference 38.7 per cent) and elgium (difference 43.5 per cent), and to a lesser extent in Denmark (24.6 per cent), Iceland (difference 28.6 per cent), Luxembourg (difference 29.4 per cent), and Norway (difference also 29.4 per cent), government coalitions are dominated by mainstream parties. A comparison of the patterns of cross-temporal variation demonstrates that the difference between the government participation of niche parties and mainstream parties varies over time, with small differences in the early 1980s and from 1996 onwards, 5

7 and larger differences between 1986 and Especially during the period the government participation of niche parties is almost equal to that of mainstream parties, but surprisingly differences have become larger again in recent years. Potential explanations for niche parties government participation In recent years scholars have become increasingly concerned with explaining why some parties join government coalitions, while others (are forced to) stay in opposition, and have developed coalition membership theories, which focus on the factors influencing the chances of individual parties joining the government (Mattila and Raunio 2002: 263, italics in original). According to coalition membership theories, the coalition formation process consists of two distinct phases: (1) the selection of the prime minister party; and (2) the selection of junior coalition members (e.g. Mattila and Raunio 2002; äck and Dumont 2006; Glasgow et al. 2007; Grofman et al. 1987; aron 1991). When the election results are in, parties do not randomly start to bargain about the composition of the government coalition. Rather, the coalition negotiations are lead by a formateur, which normally belongs to the party that is most likely to obtain the prime ministership. The formateur invites other parties to join the coalition negotiations and, if the coalition negotiations are successful, the government coalition. Given this set-up the party whose leader officially forms the government and assumes the leadership role clearly enjoys a more central role than is exercised by the other party or parties in the coalition (Warwick 1996: 473). This party selects the parties with which it wants to govern. In other words, in concentrated party systems junior parties are usually invited to join a coalition by the prime minister party (olleyer 2007: 131 emphasis added). As a result the coalition preferences of the prime minister party play an important role in the selection of junior coalition members. 5 This is clearly reflected by the factors that influence parties chances to become junior coalition members. Previous studies have found that junior coalition members are mainly selected on the basis of (1) their previous experience in government, including whether they were in the previous government and how long it has been since they were last in government; 5 Of course prospective junior coalition members are not hapless victims at the mercy of the coalition preferences of the prime minister party. They can refuse to take part in the coalition negotiations altogether or put conditions on their entry in the prospective government coalition. As Warwick (1996: ) notes the term selection should not be taken to imply that the formateur (or anyone else) acts as selector. The prime minister party nevertheless has the opportunity to bias government composition in line with its preference (äck and Dumont 2006: 1). It is not completely in control of the composition of the eventual government coalition, but its coalition preferences will have a substantial impact on the composition of the government coalition. 6

8 (2) their smallness, except where responsiveness to electoral success is called for; (3) their ideological distance from the centre of the left-right dimension as well as from the formateur; and (4) the need to curtail the number of coalition partners, particularly in fragmented parliaments (Warwick 1996: 497; see also äck 2003; Isaksson 2005; Mattila and Raunio 2002; Mattila and Raunio 2004; Warwick 1998). Since these findings are the point of departure for the analyses in this paper, it is useful to discuss them in more detail. Parties are more likely to become junior coalition members when they have experience in national executives. According to Warwick (1996: 488) government experience in general and, more particularly, membership in the outgoing government increase the odds of a place in government, while a lengthy absence from government lowers those odds. This shows that parties are neither amnesiac nor myopic (Strom 1990: 569) and previous coalition experiences inspire new ones for two reasons: familiarity and inertia. Familiarity refers to the fact that in any coalition forming situation a coalition will form containing that set of partners most familiar with working together (Franklin and Mackie 1983: 277). In other words, parties with government experience have an advantage in the coalition formation process and parties that have governed together are especially likely to govern together again. Familiarity makes that government outcomes can be explained precisely by the predisposition of parties to continue coalitions fund practicable before (Daalder 1983: 15). The reasons for the occurrence of the familiarity effect are that (1) parties that have governed together have created trust and channels to communicate and co-operate; (2) parties that have governed together are more certain about each others preferences and future actions; (3) government parties can get a reputation as good coalition partners; and (4) parties with government experience know the ins and outs of the government formation process (Franklin and Mackie 1983; Warwick 1996). Government coalitions that contain parties without prior government experience lack these advantages. The members of these government coalitions are faced with higher transaction costs (Lupia and Strom 2008); that is, higher levels of uncertainty about the future behaviour of their coalition partners and hence the durability of the government coalition. A special form of the familiarity effect is the incumbency effect, which entails that incumbent governments have a high propensity to govern again when they maintain their majority in the elections. In addition to the reasons that produce the familiarity effect, the incumbency effect is also caused by the fact that incumbent governments form the reversion point in the event of breakdown of the coalition negotiations (Strom et al. 1994; Martin and Stevenson 2001; De Winter et al. 2002). The incumbency effect creates high levels of inertia in the government formation process (Franklin and Mackie 1983). 7

9 The likelihood that a party becomes a junior coalition member also increases when the size of the party decreases, or inversely the larger the party, the less likely it is to take a seat at the cabinet table (Warwick 1996: 488). Small parties are attractive junior coalition members, because they are easy to dominate and can enable formateur parties to form bare majority government coalitions in which they reap most of the office benefits. It therefore also matters how large the formateur party is. A larger formateur party has a less need for junior coalition members and/or needs smaller junior coalition members than smaller one. The largest party in the system is especially unlikely to join a government coalition as junior coalition member, because it would threaten the position of the formateur party (Isaksson 2005: 341). Additionally, winning the elections that preceded the government formation process increases a party s odds to join the government coalition. Warwick asserts that this indicates that the government formation process contains an element of electoral responsiveness. Follow-up studies demonstrate, however, that there is no linear relationship between a party s electoral fortunes and the likelihood that it becomes a junior coalition member. Mattila and Raunio (2002: 276; 2004: 218) observe that winning elections does not help parties becoming a coalition government member and that if a party wants a place in the government it is more important to avoid losing seats in the elections than to be an actual winner (see also Isaksson 2005: 341). Moreover, they conclude that electoral success is not as important as a party s size or experience when explaining the selection of junior coalition members. A third cluster of factors that affects the odds that parties becomes a junior coalition members is policy-related. Parties that are centrally located are more likely to become junior coalition members than parties that are located towards the fringes of the political space. The reason for this is that centrally located parties have more coalition opportunities than parties that are located towards the extremes of the left-right dimension, that is, they can for coalition with parties on their left and right. In addition to the centrality of the position of the party, the distance of the party to the formateur party also matters. Formateurs generally have a preference for forming coalitions with their ideological neighbors, or other parties that take positions that are in close proximity of their own, because it is less costly to make policy compromises with these parties. With proximal parties formateurs can conclude policy agreements that are close to their ideal positions, which pleases their members and voters. Finally, the size of the government coalition also matters. The larger the government coalition, the more parties are needed to form it, especially in highly fragmented party systems. Hence, parties are more likely to become junior coalition members in countries 8

10 traditionally governed by majority coalitions than in countries where minority governments are the rule. Other party system features matter as well. High levels of party system polarization and large relative distances between parties make it less likely that parties participate in government coalitions as junior coalition members, because they increase the chances that minority governments are formed. For the same reason the size and centrality of the formateur play an additional role, because it can make it possible for formateur parties to survive in office without constructing a majority government coalition (Warwick 1998: ). Niche parties and coalition membership theories Most coalition membership theories suggest that niche parties have a disadvantageous position in coalition negotiations. Niche parties usually lack parliamentary and governmental experience and are located on the fringes of the political space, two factors that normally significantly decrease parties chances to join government coalitions. Niche parties have gradually entered West European parliaments since The parliamentary representation of green and radical right parties often dates back to the 1980s, but in some countries these parties have only emerged in the 1990s or even more recently. Table 3 already demonstrated that, despite having been represented for one, two, or even three decades, most niche parties have only started to participate in government coalitions in recent years. Consequentially, it is unlikely that they can benefit from any previous experiences in government. Niche parties also have the major disadvantage that they often can only make overtures towards the centre of the political space. According to Smith (1997: 36), most niche parties are situated at the outer extremes of the mainstream left and right. Even though their relative size may be significant, their coalition potential is limited since normally they can make overtures only to a single large party that has an affinity with them in some respects, and that party will usually not wish to estrange itself from its more moderate neighbours. This assertion has been confirmed by several large-n studies. Isaksson (2005: 336), for example, notes that a party on the outskirts of a traditional left-right scale reasonable has a less beneficial starting point in government negotiations. The probability of a right- or leftwing party being included in a coalition is, hence, smaller than that of a party in the ideological centre. Hence, left-wing niche parties are only attractive coalition partners to leftwing mainstream parties, while right-wing niche parties are only attractive coalition partners to right-wing mainstream parties. Consequentially, niche parties are more likely to become junior coalition members when formateur parties have a clear left- or right-wing signature. 9

11 When formateur parties take an unambiguous centre position, such as most Christian- Democratic parties, niche parties are less likely to become junior coalition members. Of course a lot of variation exists within the group of niche parties when it comes to party positions on the left-right dimension. Even within particular party families substantial differences in party positions exist, and niche parties sometimes change positions over time (Adams et al. 2006). This can make that some niche parties are more attractive coalition partners than others. Much depends on the positions of formateur parties as well, because those, together with the positions of niche parties, determine the policy distance between the two and hence the grounds for cooperation. The only thing niche parties seem to have going for themselves is that fact that they are usually small to comparative standards and electorally successful. The first factor would make that niche parties can easily complement formateur parties and make coalitions winning, the second factor would make that formateur parties are inclined to include niche parties in their coalitions because the election results suggest they deserve a position in government. Dumont and ack (2006) argue, however, that for parties that remain small by comparative standards [ ] the larger they become the greater their chance of getting into government. 6 The rationale behind their argument is the observation that small parties can simply be too small to contribute to any government coalition in a significant way, especially when the formateur party is not very big. It is therefore unlikely that small parties are invited by formateur parties to join government coalitions. The logic developed by Dumont and ack could explain why niche parties have been especially successful at entering government coalitions in recent years. During the 1980s they were probably simply too small to impose themselves during coalition negotiations and contribute in a significant way to the formation of majority coalitions. In short, coalition membership theories thus suggest by-and-large that niche parties are unattractive coalition partners, which could explain why they have participated less frequently in government coalitions than mainstream parties. Moreover, most theories also offer clues as to why niche parties become more attractive coalition partners in recent years. Niche parties have become more experienced at the parliamentary game and have experimented with executive power at the local and regional level. They have also realized important victories in elections and increased their representation in West European parliaments. Moreover, there is some evidence that niche parties have moderated their stances to prepare for government 6 Dumont and äck essentially suggest that there is a non-linear relationship between a party s size and the likelihood it becomes a junior coalition member. 10

12 participation. Some scholars even argue that they have deliberately pursued office-seeking strategies and attempted to make themselves more attractive on the coalitional market by moving in the direction of mainstream parties. At the same time there are indications that mainstream parties have moved closer to niche parties to counter the electoral success of these parties (Meguid 2005). These two dynamics could have created a fertile ground for executive cooperation between mainstream and niche parties. Hypotheses On the basis of these considerations the following eleven hypotheses are formulated regarding potential determinants of junior coalition membership. With regard to certain hypotheses it is expected that the predicted effects are either stronger or weaker for niche parties. The first four hypotheses pertain to the role party size plays in the coalition formation process. It is hypothesized that parties are more likely to become junior coalition members the larger their seat share, the larger the increases in their seat share, and the smaller the formateur party. It is also hypothesized that the effect of party size and changes therein have a greater effect on the likelihood that niche parties become junior coalition members than on the likelihood that mainstream parties take up this position. It is plausible that the government participation of niche parties is first and foremost related to the electoral and parliamentary pressure these parties exert. y governing with niche parties formateur parties might seek to incorporate these parties in existing structures and take away their electoral appeal as opposition parties (De Lange 2008: ). H1: Parties are more likely to become junior coalition members the larger their parliamentary seat share. This effect is expected to be stronger for niche parties. H2: Parties are less likely to become junior coalition members when they are the largest party in parliament. H3: Parties are more likely to become junior coalition members the larger their gains at the elections. This effect is expected to be stronger for niche parties. H4: Parties are more likely to participate in government coalitions the smaller the size of the formateur party. The effect for niche parties is expected to be the same as that for mainstream parties. The fifth and sixth hypotheses concern the relationship between party positions and distances between those positions, on the one, and the likelihood that a party becomes a junior coalition 11

13 member, on the other hand. For reasons of simplicity the hypotheses only cover positions and distances on the left-right dimension. Since niche parties compete primarily on issues that are not part of this dimension, it is expected that the distance to the formateur party on the leftright dimension is not as good a predictor of junior coalition membership for niche parties as it is for mainstream parties. H5: Parties are more likely to become junior coalition members the smaller the distance to the formateur party on the left-right dimension. This effect is expected to be weaker for niche parties. H6: Parties are more likely to become junior coalition members the closer they are located to the centre of the left-right dimension/the less extreme their position. The effect of the distance to the centre is expected to be weaker for niche parties. The seventh and eighth hypotheses predict that a party s experience in parliament and government will also influence the likelihood that it becomes a junior coalition member. It is hypothesized that parties are more likely to become junior coalition members the greater their parliamentary experience, and when they have incumbency status. Most mainstream parties have been represented in West European party system since Differences in parliamentary experience between mainstream parties are therefore negligible. Niche parties are generally far less experienced than mainstream parties and differences in experience between niche parties are more substantial. Hence, it is expected that parliamentary experience will have a stronger effect on the likelihood that niche parties become junior coalition members than the likelihood than niche parties become junior coalition members. Recent studies have demonstrated that niche parties tend to incur serious electoral losses when they have participated in government (uelens and Hino 2007). This could imply that they are less likely to benefit from an incumbency bonus in the coalition formation process than mainstream parties. It is therefore expected that incumbency status will have a weaker effect on the likelihood that niche parties become junior coalition members than the likelihood than niche parties become junior coalition members. H7: Parties are more likely to become junior coalition members the greater their parliamentary experience/the longer they have been represented in parliament. This effect is expected to be stronger for niche parties. H8: Parties are more likely to become junior coalition members when they have incumbency status. This effect is expected to be weaker for niche parties. 12

14 Party system characteristics are also hypothesized to influence the likelihood that parties become junior coalition members. It is hypothesized that the effective number of parties and the level of polarization will have an impact on parties chances to assume office and that parties are less likely to become junior coalition members in countries that are ruled by minority governments. These effects are expected not to be conditional on the type of party under investigation. H9: Parties are more likely to become junior coalition members the higher the effective number of parliamentary parties. H10: Parties are more likely to become junior coalition members the lower the level of polarization. H11: Niche parties are less likely to become junior coalition members in countries ruled by minority governments. Research design, data, and method To test these eleven hypotheses a new dataset has been constructed that contains information about the government coalitions that assumed office in West European parliamentary democracies between 1981 and 2008, and that documents which parties participated in these government coalitions. The dataset covers seventeen countries and 110 government coalitions, in which a total of 172 junior coalition members have participated. 7 The dataset is novel in several respects. First, it includes countries that are often excluded from comparative analyses, such as Greece, Iceland, Luxembourg, Portugal, and Spain. Second, the time span covered by the dataset is relatively large. More importantly, government coalitions that have been formed in recent years have also been included in the dataset. For the study of the government participation of niche parties this is essential, because these parties have only entered government coalitions in recent years. The dependent variable, Junior Coalition membership, is dichotomous and measures whether parties have been a junior coalition member in a particular government coalition, or not. To test the hypotheses twelve independent variables have been measured. The key independent variable in this study is a dummy, which measure whether a party is a niche party. Niche parties are identified on the basis of their party family affiliation reported in the 7 The dataset only includes information about government coalitions that have been formed after new elections have been held. Government coalitions that have formed mid-term are not party of the dataset. Moreover, it only contains entries for elections that produced a minority situation, that is, elections that did not leave a single party with a majority of seats in parliament, a situation which necessitates the formation of a government coalition. 13

15 Parliamentary Data Archive. Parties that are labelled as green parties, as extreme right-wing parties, or as regional, separatist, or ethno-nationalist are coded as niche parties, while all other parties have been coded as mainstream parties. The other eleven independent variables have been taken from previous coalition membership studies. The focus is first and foremost on those variables that have proven to have a significant effect on the likelihood that a party becomes a junior coalition member. These variables can be divided in four main groups. The first group of variables measure size characteristics of parties. The variable Size measures the proportion of seats controlled by a party at the time of the formation of the government coalition. The variable Largest party measures whether the party controlled the largest proportion of seats of the parties represented in parliament in the year the government coalition was formed. The variable Size formateur measures the proportion of seats controlled by the formateur in the same year. The variable Change size measures what proportion of votes the party has won or lost at the elections that preceded the formation of the government coalition. The second group of variables measures policy characteristics of parties. The variable Distance to formateur measures the policy distance between the party and the formateur party on the left-right dimension, while the variable Distance to centre measures how far this position is from the centre of this dimension. To avoid that the two variables are highly correlated, the latter variable does not reflect a party s distance to the weighted mean or median, but to absolute mid-point of the left-right dimension. The two variables are both included in the analysis, because the first tell us whether a policy rationale is likely to exist between prospective coalition partners, whereas the second informs us about the centrality of a party s position. oth are deemed relevant in the literature on junior coalition membership. The third group of variables measures how experienced parties in parliament and government. The variable Experience measures the number of years a party has been represented in parliament, counted from the first year a party has entered parliament. The variable Incumbency measures whether a party was part of the outgoing government coalition and thus has incumbency status. The fourth group of variables measures party system characteristics. The variable ENPP measures the effective number of parliamentary parties, using the Laakso and Taagepera index (Laakso and Taagepera 1979). The variable Polarization measures the level of polarization in the party system by looking at the range of positions taken by parties in this system. This measure is preferred over more sophisticated measures of polarization, because these are likely to be highly correlated with other independent variables and thus create 14

16 multicollinearity problems. The variable Minority government is a dummy variable that measures whether countries are predominantly ruled by minority governments (Denmark, Norway, and Sweden) or majority governments. The size and experience variables are measured on the basis of authorative sources that report election results, most notably The International Almanac of Electoral History (Rose and Mackie 1991), the Political Data Yearbook published by the European Journal of Political Research, and the website The combination of sources assures the completeness of information, especially for those countries in which the composition of parliamentary groups is rather fluid (e.g. France and Italy). On the basis of these sources, and the Parliamentary Data Archive, Müller and Strøm (2000) and Woldendorp et al. (2000) information was also collected about the composition of government coalitions and the identity of formateur parties. 8 The policy variables, including the level of polarization, are measured on the basis of party positional data collected through five expert surveys (Castles and Mair 1984; Huber and Inglehart 1995; Laver and Hunt 1992; Lubbers 2000; enoit and Laver 2006). 9 Extensive discussions of the most appropriate way to estimate parties positions are widely available (e.g. udge et al. 2001; Keman 2007; Laver 2001). The estimation of the party positions of niche parties is more difficult than that of other party families. Expert surveys provide the most reliable and valid estimates of niche parties positions (De Lange 2008) and are thus used in the analyses presented in this paper. The results of the analyses are cross-validated by running the same analyses using the Franzmann and Kaiser estimates of party positions (Franzmann and Kaiser 2006). 10 The hypotheses are tested by means of logistic regression analysis, because the dependent variable is dichotomous and the relationship between the independent and dependent variables consequentially non-linear (erry 1993). Logistic regression is essentially a form of OLS regression in which a logarithmic transformation is included to approximate the linearity assumed in OLS regression (erry and Feldman 1985). OLS regression and logistic regression are thus in many respects equivalent procedures developed for different kinds of dependent variables. To ensure that the results of the logistic regression analyses are not affected by multicollinearity, the variables Size formateur, Distance to formateur, Distance to centre, ENPP, and Polarization have been centred around their means. 8 Note that formateur parties are in most cases identical to the prime minister parties. Hence, prime minister status has been taken as a proxy for formateur status. 9 The left-right scores of parties reported in these surveys have been standardized (see Carter 2005) and have been attributed to five-year periods (e.g. the scores from Castles and Mair to the period ). 10 The Franzmann and Kaiser estimates have been calculated using the data collected by the Comparative Manifesto Project (udge et al. 2001; Klingemann et al. 2007). 15

17 The determinants of the government participation of niche parties The logistic regression analyses identify the determinants of the junior coalition membership of niche parties in three steps. In a first step the independent variables that have been found significant in previous studies of junior coalition membership are introduced to establish which have a significant effect on the likelihood that a party becomes a junior coalition member and which do not. In this study the most important determinants of junior coalition membership are 1) the size of the formateur party, 2) the distance to the formateur party, 3) experience in parliament, 4) incumbency status, and 5) the existence of a tradition of minority or majority government (see Model 1 in Table 4 and Table 5). 11 The effects are in the expected directions, as parties are more likely to become junior coalition members the smaller the formateur party, the smaller the policy distance to this party, and the greater their experience in parliament. They are also more likely to become junior coalition members when they have incumbency status and when majority coalitions are the norm. Interestingly, parties are not more likely to become junior coalition members the larger their share of seats 12, the larger their electoral gains, the closer they are located to the centre of the left-right dimension, the higher the effective number of parties, and the higher the level of polarization. Moreover, parties that are the largest in parliament are not significantly less likely to become junior coalition members. 13 The results of he logistic regression analysis thus provide support for the first part of the hypotheses 4, 5, 7, 8, and 11, while the hypotheses 1, 2, 3,6, 9, and 10 are not supported. <<Tables 4 and 5 about here>> In a second step the niche party dummy is added to the model to examine whether niche parties are less likely to become junior coalition members than mainstream parties, everything else being equal. The effect of this variable is in the expected direction, but not statistically significant (see Model 2 in Table 4 and Table 5). In other words, niche parties are not significantly less likely to become junior coalition members than mainstream parties, at 11 In Model 1 presented in Table 5 the variable Change Size also has a significant effect on the likelihood that a party becomes a junior coalition member. 12 To test for non-linearity of the relationship between a party s size and the likelihood that it becomes a junior coalition member Model 1 has also been run with a squared size variable, but this does not improve the fit of the model. 13 These determinants of junior coalition membership differ to some extent from those identified in previous studies of junior coalition membership. The differences are the result of the focus on a different set of countries and a different time period, as well as the inclusion of many smaller parties in the dataset. 16

18 least when one also takes the effect of other independent variables into consideration. The introduction of the niche variable leaves the previously established effects of the independent variables in tact. Direction, magnitude and significance of the effects of the independent variables in Model 2 are identical to those in Model 1. Hence, the fact that niche parties are less likely to participate in government coalitions is entirely attributable to their party characteristics, as well as to the characteristics of formateur parties and party systems. Niche parties are less likely to become junior coalition members than mainstream parties, because they are usually located further away from formateur parties than mainstream parties, because they usually have less parliamentary experience than mainstream parties, and because they benefit less frequently from incumbency status than mainstream parties. In a third step the independent variables are interacted with the niche party variable to investigate whether the effects of the independent variables are similar for niche and mainstream parties (see Models 3 to 12 in Table 4 and Table 5). Only three interaction terms are statistically significant: the interaction term between the size of the formateur party and niche party status, the interaction term between the level of polarization and niche party status, and the interaction term between the tradition of either minority or majority governments and niche party status. 14 Surprisingly, none of these interaction effects were hypothesized, while none of the hypothesized interaction effects is significant. A closer look at the b-coefficients of the conditional effects of the variables Size Formateur, Polarization, and Minority Government reveals how effects of the independent variables differ between niche and mainstream parties. The b-coefficients for Size Formateur are for mainstream parties and for niche parties, which indicates that, although both types of parties less likely to become junior coalition members the larger the formateur party, the effect is stronger for mainstream parties than for niche parties. The likelihood that mainstream parties assume office as junior coalition members is thus more strongly dependent on the size of the formateur party than the likelihood that niche parties achieve this. The level of polarization interacts in a different way with the Niche Party variable. The b-coefficients for the Polarization variable are.039 (not significant) and respectively. Thus, while the level of polarization does not affect the likelihood that mainstream parties become junior coalition members, it does have a strong negative effect on the likelihood that niche parties become junior coalition members. An increase in the level of 14 Table 5, estimated on the basis of the Franzmann and Kaiser data, does not include a model with an interaction term between Minority Government and Niche Party, because this term would predict the absence from government of niche parties perfectly. 17