Do Populist Parties Increase Turnout? Evidence from over 40 Years of Electoral History in 31 European Democracies

Transcription

1 Do Populist Parties Increase Turnout? Evidence from over 40 Years of Electoral History in 31 European Democracies Arndt Leininger Maurits Meijers September 7, 2017 Abstract Many established democracies have seen an increase in populist parties and candidates contesting elections in recent decades. While the policies and politics of populist parties are often deemed undesirable, the electoral presence of such parties may in fact have some desirable side-effects. These actors may, as they sometimes claim themselves, mobilize previous non-voters who are dissatisfied with current developments to participate in elections. Populist parties, in particular the radical right kind, may also motivate centrist non-voters to turn out to cast a vote against them. However, they may also impact negatively upon turnout because their negative rhetoric may further people s distrust towards politics, politicians and representative democracy. If populist parties do indeed affect turnout such an effect is mostly like to manifest itself in proportional representation systems which pose lower barriers to the entry of new parties. Examining both the effects of radical left and radical right populist parties, we test for the turnout effect of populist parties on a comprehensive aggregate dataset on elections in 20 West European and 11 Eastern European countries since as early as Draft version prepared for EPOP 2017, University of Nottingham. University of Mainz, leininger@politik.uni-mainz.de Radboud University, Nijmegen, m.meijers@fm.ru.nl 1

2 1 Introduction From the late 1990 s onward, the number of politically relevant political parties in Europe considered as populist has increased. This phenomenon was accompanied by a surge of studies in comparative politics on the issue of populism (Betz, 1994; Meny and Surel, 2002a; Mudde, 2007; Kessel, 2015). Particular emphasis was placed on the growing electoral clout of the populist radical right (Mudde, 2007, etc.). Yet, also the populist radical left is a significant political player in many European democracies - particularly in Western Europe (March and Mudde, 2005; March, 2011; March and Keith, 2016). Following Cas Mudde, we define populism as a thin-centred ideology that considers society to be ultimately separated into two homogeneous and antagonistic groups, the pure people versus the corrupt elite, and which argues that politics should be an expression of the volont gnrale (general will) of the people (Mudde, 2004, p. 543). As populist parties have become increasingly relevant in many European democracies, the normative debate about whether is essentially a good or bad thing for democracy also took hold. As Mudde and Rovira Kaltwasser, (2017, p. 79) note, the conventional potion is that populism constitutes an intrinsic danger to democracy. Yet, under some circumstances, the authors argue, populism can be a corrective for democracy rather than a threat. In line with many other scholars of populism, Mudde and Rovira Kaltwasser argue that populism can be healthy for a democracy as it puts previously ignored issues on the agenda and give voice to hitherto unrepresented voters. As such, populist parties can increase voter turnout in national elections. The literature on turnout suggests that aggregate-level turnout may increase when the competitiveness of elections is higher (Blais, 2006; Geys, 2006). Arguably, the presence of populist parties challenging the dominant political parties adds a competitive dimension to party politics and, thereby, they can increase voter turnout. On the other hand, there is also room for scepticism concerning the influence populist parties have on turnout. The presence of populist parties may increase political disenchantment of citizens, which in turn might negatively affect their willingness to partake in the electoral process (Meny and Surel, 2002b). In addition, an increase in political parties due to the 2

3 presence of populist parties must not necessarily increase turnout (Blais and Dobrzynska, 1998; Blais, 2006) as this increases the complexity of the electoral process. Therefore, this paper sets out to answer the question: do populist parties affect the aggregate-level turnout in national parliamentary elections? On the basis of a dataset of all national parliamentary elections in European democracies since 1970 based on the textitparlgov database (Döring and Manow, 2016) accompanied by election specific data from the Comparative Political Data Set (CPDS) (Armingeon et al., 2016) and further sources. While we predominantly focus on the presence of populist parties as our central independent variable, we also examine other operationalizations of our populist party independent variable. These include the number of populist parties, the total populist vote share and seats share, the vote share of the largest populist party and whether a new populist party entered parliament. We estimate panel regression models with country fixed effects. Importantly, in this paper we focus on the aggregate-level turnout in national elections, with the election-year as our unit of analysis. While micro-level analysis of voting turnout can yield important information about individual-level characteristics that can pave or block the way toward the polling booth (Smets and Ham, 2013; Immerzeel and Pickup, 2015), studying the dynamics of turnout at macro-level in cross-national fashion is important in order to gain insight in the net effects of populist parties on turnout levels. As Matsusaka and Palda, (1999, p. 442) point out, the study of aggregated voting behavior is important as indiviudal indiosyncracies cancel each other [out] and allow the estimation of models with greater explanatory power. We find that the presence of populist parties does not affect the aggregatelevel of turnout in European democracies once we include the necessary controls. In addition, we find that the disproportionality of an electoral system does not condition the effect of populist party presence on turnout. These conclusions also hold for Central and Eastern European party systems. 2 Theorizing the Effect of Populism on Turnout More often than not, populist parties have been characterized as a danger to the functioning of democracy by normative scholars and political com- 3

4 mentators (Downs, 2012). Yet, scholars have also argued that populism may have a positive role as it can increase political participation (Mudde and Rovira Kaltwasser, 2017, p.83). In this vein, Meny and Surel (Meny and Surel, (2002b) argue that populism should also be studied as a form of political mobilisation. While the concept of populism has been debated, scholar tend to agree that populism is a set of ideas that underline a fundamental opposition between the people and the establishment or the elites (Mudde, 2004; Hawkins, 2010; Rooduijn et al., 2014). Since populist parties espouse an anti-elite sentiment, they attempt to capitalize on disenchantment with the current political elites (Meny and Surel, 2002b). Hence, it has been argued that populism can be a response to the growing distrust and disinterest in politics (Dalton and Wattenberg, 2000; Betz, 1994). Related to his argument, scholars have argued that populism can be a response to excessive depoliticized decision-making produced by European integration and the collusion of mainstream parties (Mair, 2007) without the input of citizens and the policy output of which is not sufficiently responsive to citizens demands (Mouffe, 2005; Mair, 2006). Indeed, [f]eelings of powerlessness, of not being able to voice dissatisfaction effectively, of not being able to make oneself heard, are all fertile ground for populist parties. (Meny and Surel, 2002b, p. 11) Issue competition theory suggests that there can be clear electoral benefits for parties that emphasize publicly salient issues that were previously ignored by other competitors (Schattschneider, 1960; Carmines and Stimson, 1986). With their anti-establishment approach, populist parties often capitalize on sensitive political questions that dominant parties have chosen to ignore. Vries and Hobolt, (2012) have found that parties which advanced an outspoken on a previously depoliticized issue, the issue of European integration, tend to benefit electorally. This suggests that such parties have a strong mobilising potential. The supposition, then, is that if populist give voice to a silent majority then this should attract former non-voters to the polling stations. In addition, apart from increasing voter turnout by attracting new populist voters, radical right and radical left populist parties can also spur political participation by drawing anti-populist non-voters to the voting booth. In their individual-level turnout study of radical right populism Immerzeel and Pickup, (2015) show that, in Western Europe, radical right 4

5 populist parties tend to mobilize highly educated and politically interested voters voters likely to be opponents of (radical right) populist parties. Also the literature on voter turnout brings arguments to the table why we could expect an increase in aggregate-level turnout when populist parties participate in the election in question. A widely shared assumption is that voters are more likely to be drawn to elections when these are meaningful. In other words, citizens turn out to vote when they have the impression that there is something at stake. In this vein, extant research has shown that the competitiveness of elections matters for voter turnout (Blais, 2006; Geys, 2006). The extant literature suggests that political fragmentation can affect turnout. Specifically, a higher number of political parties broaden a voter s choice palette, which might motivate her to go the polls (Blais, 2006). In other words, when the political parties that compete in an election strongly differ from one another, the clarity of choice increases. This may motivate voters to go and vote. Moreover, Geys, (2006) finds in his meta-analysis of aggregate-level voter turnout studies that the closeness of an election (i.e. vote gap between the [main] competitors) is important. One possible reason for this is that when elections are closer, the individual s probability to be able to affect the outcome increases, hence making voting more relevant. Another possible reason is that political parties are more keen on mobilizing voters, i.e. by campaigning, when the race is close. This relates to the notion that the amount of expenditures for election campaigns is positively correlated with turnout, which is a rather robust finding in the literature (Geys, 2006). Applying these findings to the case of populist parties, it can be postulated that the presence of populist parties increases the urgency of elections both for potential populist voters and for their opponents. Indeed, populist parties may increase the clarity of choice in an election. In particular, when the populist party in question emphasizes an issue that was previously ignored or adopts an unorthodox, extreme position on a given issue. Populist parties, right and left, are known to do precisely that. The populist radical right championed the immigration issue (Mudde, 2007) and the populist radical left stands out with its anti-neoliberal critique (March, 2011). In addition, both the populist radical right and left put pressure on mainstream parties on the issue of European integration (Meijers, 2017). Also, the arguments about the closeness of elections and the amount of campaign 5

6 expenditures can be applied to the case of populist presence : elections are likely to be more competitive when anti-establishment, populist parties participate. These considerations allow us to formulate the following hypothesis: Mobilization Hypothesis: When populist parties are present in a given election, aggregate-level voter turnout will be higher. Yet, although the arguments for a particpatory effect of populism are compelling, both the literature on populism and the literature on voter turnout suggest it is also possible that there is no participatory effect of populism on aggregate-level turnout. First of all, the presence of populist parties might increase the political disenchantment that has spurred their nascence in the first place. One possible argument is that populist parties capitalize on anti-elite sentiments, but are not able to effectuate the political change they envisage. Hence, rather than motivating people to participate in the proper functioning of representative democracies, populist parties may convince that there is not much utility in voting when the same elites stay in power (Meny and Surel, 2002b). In a similar vein, as argued by Immerzeel and Pickup, (2015), populist parties may in fact demobilize voters because they increase the amount of negative campaigning in an election. In particular, by engaging in it themselves. Negative campaigning can affect the willingness of voters to become politically engaged as the negativity surrounding a political campaign might scare potential voters off (Kahn and Kenney, 1999). This, of course, particularly applies to the radical right and less so to the radical left. Therefore, it is important to estimate the effects of populist party presence for left and right separately. In addition, Immerzeel and Pickup, (2015) find in their individual-level analysis on the basis of six waves of the European Social Survey that the core-electorate of the populist radical right the male, low-educated, unemployed voters with anti-immigrant attitudes (Lubbers et al., 2002) were not mobilized to vote by right-wing populist parties Furthermore, while a higher of number of political parties may increase voter turnout because there is more political produce on offer, Blais and Dobrzynska, (1998) find that a higher number of political parties does not increase turnout (see also Blais, 2006). Rather, they argue that a higher 6

7 number of political parties increases the complexity of the election, which can hamper voters willingness to vote. Also, a higher number of (relevant) parties may blur one s clarity of choice for a government. When the number effective parties is low, it is usually clear from the outset which (bloc of) parties are likely coalition partners. When the number of parties is higher, however, it is less obvious with which competitor the party one voted for will attempt to form a coalition government with after the elections. As a result, the opaqueness of the political process is also increased, which might negatively affect turnout. Given these qualifications on the effect of populist parties on turnout, we should also formulate the following null-hypothesis to contrast our initial hypothesis: Null Hypothesis: When populist parties are present in a given election, aggregate-level voter turnout will not be higher. Yet, it is possible that the effect of populist parties on turnout is mediated by the degree of disproportionality in a given electoral system. One of the most consistent findings in the literature on turnout is that turnout seems to be higher under proportional representation (PR). PR systems generally are less disproportional than plurality systems (Blais and Carty, 1990; Blais, 2006; Geys, 2006). Many authors suggest that highly disproportional systems may lead to the belief that one s vote is less consequential this is particularly the case when voting for a smaller party or voting for a party that is unlikely to win the plurality in a given district (Jackman, 1987; Geys, 2006). A good example of this is the discrepancy between the vote share of the UK Independence Party (UKIP) and the Scottish National Party (SNP) in the UK General Election of In the UK, which has a single-member plurality system, UKIP garnered 12.7 per cent of the national vote in 2015, but due to the lack of geographical concentration of its votes it only obtained a single seat in the House of Commons. The SNP, on the other hand, won 4.7 per cent of the vote while winning a whopping 56 seats. Such a situation of high disproportionality might of course affect the mobilizing potential of populist parties. In addition, Blais and Carty, (1990) argue that in PR systems virtually all votes are competitive. In plurality systems, on the other hand, it is not 7

8 unusual that parties have certain stronghold districts rendering political competition and campaigning futile. Hence, PR systems may foster greater incentives for political campaigning (Blais and Carty, 1990, p.167). This is in line with evidence that an increase in campaign spending is also related to higher aggregate-level turnout (Geys, 2006, p.648). Therefore, it seems very likely that the effect of the presence of populist parties is conditioned by the degree of disproportionality in the systems with less disproportional electoral system, such as proportional representation, showing a positive effect for the presence of populist parties, whereas this effect is not present in (highly) disproportional electoral systems. This allows us to formulate the following conditional hypothesis: Disproportionality Hypothesis: The effect of the presence of populist parties in a given election on aggregate-level voter turnout decreases in the disproportionality of the electoral system. Finally, it is possible that the relationship between the presence of populism and turnout is different in Central and Eastern (CEE) and Western Europe. First, it is important to note that the party systems in CEE are relatively young at least in comparison with West European party systems. Many authors have noted the lack of stability of CEE party systems (Offe, 1996; Lewis, 2000). The presence of populist parties, which try to delegitimize the (newly) established political order with anti-establishment politics, might therefore not have same effect as in Western Europe. Rather than convincing voters that politics has become (more) meaningful again, populist parties in CEE countries might spur doubts about the utility of competitive democracies. The findings of Immerzeel and Pickup s (2015) individual-level study on the effect of radical right populist parties on voter turnout in CEE countries also point in this direction. The authors find that particularly young voters and those opposed to radical right populist parties are less inclined to turn out when radical right parties are successful. To sum up, it is possible that populist parties in CEE party systems have a demobilizing effect as their presence highlights the fragility of the nascent party systems in these countries: CEE Hypothesis: When populist parties are present in a given election in a party system in Central and Eastern Europe, aggregate-level voter turnout 8

9 will be lower. 3 Research Design To be able to test our hypotheses, we constructed a dataset of all national parliamentary elections (to the lower house) in European democracies since See Table 1 for an overview which countries and national elections are included. The data on election results come from the ParlGov database (Döring and Manow, 2016) which provides information on which parties ran in national parliamentary and European parliament elections and how they fared. For now, we focus on national parliamentary elections only. We coded the presence of populist parties variable based on the literature on both radical right and radical left populism. We primarily relied on the classifications provided by Kessel, (2015) as well as from Mudde, (2007) and March, (2011). In addition to the classifications proposed by these authors, we consider, on the radical left, the Workers Party of Belgium (PVDA-PTB) (Pauwels, 2014) and the Spanish party Podemos (Ramiro and Gomez, 2016) to be populist parties. On the right we consider, the German radical right Alternative for Germany to be a populist radical right party (Lewandowsky et al., 2016). We aggregate to the election-level to obtain aggregate measures, such as whether at least one populist party competed, how many populist parties competed in total or cumulative vote share of populist parties among others. Our key independent variable captures whether at least one populist party obtained a seat in parliament, Populist Presence. We focus on this variable because it indicates the existence of at least one viable populist party in an elections. Other candidate variables that we considered are Populist Party Presence (whether at least one populist party ran), Populist Parties (how many populist parties ran), Populist Vote Share (what percentage of the vote all populist parties combined received), Populist Max Voteshare (result of largest populist party), Populist Seatshare (what fraction of seats all populist parties combined hold after the election), and New Populist Party (whether there is at least one new populist party). We report models employing these independent variables in the appendix. 9

10 Country Elections Austria 1970, 1971, 1975, 1979, 1983, 1986, 1990, 1994, 1995, 1999, 2002, 2006, 2008, 2013 Belgium 1971, 1974, 1977, 1978, 1981, 1985, 1987, 1991, 1995, 1999, 2003, 2007, 2010, 2014 Bulgaria 1991, 1994, 1997, 2001, 2005, 2009, 2013, 2014 Croatia 2000, 2003, 2007, 2011, 2015, 2016 Cyprus 1976, 1981, 1985, 1991, 1996, 2001, 2006, 2011, 2016 Czech Republic 1990, 1992, 1996, 1998, 2002, 2006, 2010, 2013 Denmark 1971, 1973, 1975, 1977, 1979, 1981, 1984, 1987, 1988, 1990, 1994, 1998, 2001, 2005, 2007, 2011, 2015 Estonia 1992, 1995, 1999, 2003, 2007, 2011, 2015 Finland 1970, 1972, 1975, 1979, 1983, 1987, 1991, 1995, 1999, 2003, 2007, 2011, 2015 France 1973, 1978, 1981, 1986, 1988, 1993, 1997, 2002, 2007, 2012 Germany 1972, 1976, 1980, 1983, 1987, 1990, 1994, 1998, 2002, 2005, 2009, 2013 Greece 1974, 1977, 1981, 1985, 1989, 1989, 1990, 1993, 1996, 2000, 2004, 2007, 2009, 2012, 2012, 2015, 2015 Hungary 1990, 1994, 1998, 2002, 2006, 2010, 2014 Iceland 1971, 1974, 1978, 1979, 1983, 1987, 1991, 1995, 1999, 2003, 2007, 2009, 2013, 2016 Ireland 1973, 1977, 1981, 1982, 1982, 1987, 1989, 1992, 1997, 2002, 2007, 2011, 2016 Italy 1972, 1976, 1979, 1983, 1987, 1992, 1994, 1996, 2001, 2006, 2008, 2013 Latvia 1990, 1993, 1995, 1998, 2002, 2006, 2010, 2011, 2014 Lithuania 1992, 1996, 2000, 2004, 2008, 2012, 2016 Luxembourg 1974, 1979, 1984, 1989, 1994, 1999, 2004, 2009, 2013 Malta 1971, 1976, 1981, 1987, 1992, 1996, 1998, 2003, 2008, 2013 Netherlands 1971, 1972, 1977, 1981, 1982, 1986, 1989, 1994, 1998, 2002, 2003, 2006, 2010, 2012 Norway 1973, 1977, 1981, 1985, 1989, 1993, 1997, 2001, 2005, 2009, 2013 Poland 1991, 1993, 1997, 2001, 2005, 2007, 2011, 2015 Portugal 1975, 1976, 1979, 1980, 1983, 1985, 1987, 1991, 1995, 1999, 2002, 2005, 2009, 2011, 2015 Romania 1990, 1992, 1996, 2000, 2004, 2008, 2012 Slovakia 1990, 1992, 1994, 1998, 2002, 2006, 2010, 2012, 2016 Slovenia 1990, 1992, 1996, 2000, 2004, 2008, 2011, 2014 Spain 1977, 1979, 1982, 1986, 1989, 1993, 1996, 2000, 2004, 2008, 2011, 2015, 2016 Sweden 1970, 1973, 1976, 1979, 1982, 1985, 1988, 1991, 1994, 1998, 2002, 2006, 2010, 2014 Switzerland 1971, 1975, 1979, 1983, 1987, 1991, 1995, 1999, 2003, 2007, 2011, 2015 United Kingdom 1970, 1974, 1974, 1979, 1983, 1987, 1992, 1997, 2001, 2005, 2010, 2015 Table 1: Countries and national elections included in our sample. 10

11 Our dependent variable, Turnout, also comes from ParlGov. 1 Geys, (2006) cautions that defining turnout as a ratio limits the range of this variable to the 0-100% interval. However, this is not the case for our models as can be seen in Table A.2. Hence, we use panel regression models on the originally scaled variable rather than using a logistic transformation. Also from ParlGov we add Gallagher s (1991) Disproportionality index and the effective number of parties (ENP) calculated on the seat shares of the parties represented in parliament. Additional election specific variables come from the Comparative Political Data Set (CPDS) (Armingeon et al., 2016). These are the Economic openness (measured as total trade, sum of import and export, as a percentage of GDP in current prices), and the national Unemployment rate (percentage of civilian labour force). We choose these variables because we have rather clear expectations about their correlation with both the dependent and our key independent variable. In order to test the Disproportionality Hypothesis we rely on Gallagher s Disproportionality index. This index effectively captures differences between plurality rule (SMD) and proportional representation (PR) electoral systems as well as the variance within PR and mixed systems. If a party system has a low barrier to entry this will be reflected in a low disproportionality score. Because disproportionality is negatively correlated with turnout (Jackman and Miller, 1995; Blais and Carty, 1990; Jackman, 1987) omitting Disproportionality would induce a positive bias leading us to overstate a possible effect of turnout. Therefore, we include it as a control in all models. The effective number of parties counts the number of parties that make up a party system taking into account their relative sizes. We use the ENP calculated on seat shares because we are interested in the number of viable parties. Greater choice of parties is associated with higher turnout. At the same time a greater number of available parties should also make it harder for newcomers, including populist parties, to make a successful dbut. If we were to omit ENP (parties) we would expect a negative bias. Research has shown that Economic Openness, a measure of globalisation, is negatively associated with turnout (Steiner, 2010). Under globalisation governments have less room to manoeuvre, diminishing perceived efficacy of government, difference between parties (Hellwig, 2008; Hellwig, 2001) which should affect both party choice and the decision whether to 1 We fill in a missing values, Switzerland 2015, with data from International IDEA. 11

12 vote at all. Globalisation and its alleged effects have been an important campaign issue for populist parties both right, think FN or UKIP, and left, think Podemos or Movimento 5 Stelle. Therefore, if we fail to account for a country s exposure to global markets we would induce a negative bias in our estimate for populist party presence. Our other economic variable unemployment also has a clear connection to turnout. Unemployment is a strong predictor of turnout both on the individual and the aggregate level in a cross-section. However, based on prior research we expect a different for changes in unemployment over time: Increases in the Unemployment rate are associated with an increase in turnout (Burden and Wichowsky, 2014). Populist parties tend to target unemployed and citizens in risk of unemployment specifically. Hence, they are an important electorate for them (Immerzeel and Pickup, 2015). Consequently, rises in unemployment should strengthen them. Thus, if we were to omit unemployment from the specification this would potentially lead us to overestimate the effect populist parties have on turnout, if there is any. Finally, we include a control for the size of a population. A robust finding in the extant research on turnout has found that turnout is usually highest in smaller countries (Blais, 2006). We estimate panel models with country fixed effects. We control for time (Year) since there is a clear time trend in both our dependent and key independent variable: a negative trend in turnout and a positive trend in presence and success of populist parties. If we were to ignore these trends this would leave us with a negative bias. Therefore, we include a linear time trend in our models. Using fixed effects effectively cancels out any time-constant confounders, whether observable or unobservable. Some between-country differences in turnout may indeed be due to populist parties. Nevertheless, we take a more conservative approach by using fixed effects to minimize the possibility of bias through unobserved confounders as far as possible. We estimate country fixed effects via the within-transformation. Therefore, the pseudo-constant routinely calculated by Stata is omitted from all tables. As a first step we run a simple model including only Turnout and Year as explanatory variables. Only in a next step do we estimate a full model with covariates. Of course, there is some risk that measuring both variables at the same time point despite our use of a fixed-effects panel model introduces endogeneity into our estimates. As one solution we replace our key 12

13 independent variable with its first lag. In a second set of models we interact our populism variable with Gallagher s disproportionality index. Besides having a direct effect on turnout and the presence and success of populist parties, Disproportionality should condition the effect of populist parties as well. As noted in the theory section, populist party s impact should be greater in a system with low proportionality. Lower barriers to entry mean that a vote for a populist party is more likely to result in parliamentary representation. This gives sympathizers of a populist party a greater incentive to cast a vote. At the same time, because even small parties are more likely to gain representation, established will have to campaign harder to keep the party out. Furthermore, media attention is distributed more evenly across parties. This makes it easier for the populist party to mobilize its supporters. Yet, the greater presence of negative campaign message may also demobilize citizens. Furthermore, we distinguish between populist left and populist right parties. We believe that there are good reasons to suspect differences in mobilization potential. Populist left parties often resemble social movements and are less demonized by elites than populist (radial) right parties. The latter do have a strong potential to mobilize former non-voters. They did have recent successes in using their anti-immigrant rhetoric for mobilizing anti-globalist grievances. However, such negative campaigning may also alienate some citizens, as was noted in the theory section Finally, we distinguish post-communist democracies from other democracies in our sample. We do so by estimating models separately on these two samples, since the fixed effects models do not allow us to estimate variables that do not vary over time. 4 Results We begin our empirical analysis with some rather simple models. The dependent variable, as in all models presented here, is turnout in a national parliamentary election. Table 2 shows four different models. We consistently see a negative and highly significant time trend in turnout. This result in conjunction with a strong positive time trend in populism, reported in Figure A.1 and Table A.1 in the appendix, highlights the importance of controlling for time to avert bias in our estimates. Consequently, we include this vari- 13

14 able in all model specifications. All models also include country fixed effects estimated via the within-transformation. Populist Presence (1.28) (1.42) (1) (2) (3) (4) Populist Presence t (1.02) (1.22) Year (0.06) (0.08) (0.06) (0.07) Disproportionality (0.16) (0.19) ENP (Seats) (0.60) (0.67) Unemployment (0.14) (0.12) Economic openness (0.04) (0.03) Population size (0.14) (0.16) R N p < 0.05, p < 0.01 Table 2: Fixed-effects regression models of turnout on Populist Presence, models (1) and (2), and its lag, (3) and (4). All models include country fixed-effects. Another control variable, disproportionality is also negatively and significantly related to turnout. This is in line with research finding that PR systems yield higher levels of turnout (Blais, 2006; Geys, 2006).However, this variable and other additional covariates do not add much explanatory power to our models: R 2 changes little between models (1) and (2) and (3) and (4) respectively. Results presented in Table 2 provide weak support for the Mobilization Hypothesis and no support for the Null Hypothesis. There is a positive correlation between Populist Presence and turnout. On average, the presence of at least one populist party is associated with an increase in turnout between one and two percentage points. This correlation is stronger when we lag our key independent variable. Both the contemporaneous and lagged 14

15 correlation are slightly weaker when controls are included. Yet, only the contemporaneous Populist Party Presence variable is statistically significant at the 0.05 level in the model without economic controls (Model 1) Concerning the theoretical ideas which we test, these results are at best suggestive evidence for a positive effect. Populist Presence (1.74) (2.00) (1) (2) (3) (4) Populist Presence t (1.48) (1.50) Year (0.06) (0.08) (0.05) (0.06) Disproportionality Pop. Presence Disproportionality (0.25) (0.25) (0.24) (0.22) (0.22) (0.29) Pop. Presence t-1 Disproportionality (0.24) (0.24) Unemployment (0.15) (0.13) Economic openness (0.04) (0.03) R N p < 0.05, p < 0.01 Table 3: Fixed-effects regression models of turnout on Populist Party Presence, models (1) and (2), or its lag respectively, (3) and (4), interacted with Gallagher s Disproportionality index. All models include country fixed-effects. Figure 1 shows a plot of the average marginal effects of Populist Party Presence for different levels of disproportionality of party systems. The positive correlation between populist presence and turnout is strongest when disproportionality is at its lowest (see Table 3), which is in line with our expectations in the Disproportionality Hypothesis. However, as the 95 per cent confidence intervals overlap with 0, the estimate is not statistically 15

16 Figure 1: Marginal effects plot of Populist Party Presence for different levels of disproportionality of party systems. significant. Hence, the Disproportionality Hypothesis cannot be confirmed. In this case, in contrast to the models presented in Table 2, the correlation between populist presence and turnout is weaker when populist presence is lagged. Again correlations are slightly weaker and fail to attain statistical significance when controls are included. Next, we consider the difference between left and right populist parties. We estimate separate sets of models which replace our Populist Party Presence variable with a dummy that indicates the presence of at least one left or right populist party respectively. The models in Table 4 mirror the specifications of models (2) and (4) in Table 2 and Table 3. We see a stronger contemporaneous correlation for right populist parties, but a stronger correlation between turnout and the lag of populist presence for the populist left. None of these estimates, however, are significantly different from zero. Finally, we considered whether the potential impact of populist parties on turnout differs between Central and Eastern European party systems and the other European democracies. We find no difference between CEE countries and non-cee countries in the contemporaneous correlation between populist presence and turnout. In contrast, the estimates for the lag of populist presence differ strongly even though both are not significant. The 16

17 Populist Presence (Left) 0.25 (1.95) Populist Presence (Left) t (2.62) (1) (2) (3) (4) Populist Presence (Right) 2.73 (1.93) Populist Presence (Right) t (1.96) Year (0.07) (0.07) (0.07) (0.06) Disproportionality (0.14) (0.18) (0.14) (0.18) Unemployment (0.14) (0.15) (0.14) (0.12) Economic openness (0.04) (0.03) (0.04) (0.03) R N p < 0.05, p < 0.01 Table 4: Fixed-effects regression models of turnout on the presence of at least one populist left party in the current, model (1), or preceding election, (2), and the presence of at least one populist right party in the current, (3), or previous election, (4). All models include country fixed-effects. 17

18 (1) (2) (3) (4) Populist Presence (4.67) (1.23) Populist Presence t (1.94) (1.05) Year (0.28) (0.35) (0.07) (0.06) Disproportionality (0.34) (0.33) (0.15) (0.21) Unemployment (0.22) (0.22) (0.13) (0.13) Economic openness (0.10) (0.11) (0.02) (0.01) R N p < 0.05, p < 0.01 Table 5: Fixed-effects regression models of turnout on Populist Presence or its lag on a sample of CEE countries, models (1) and (2), and a sample of non-cee European countries, (3) and (4). All models include country fixed-effects. 18

19 two models estimated on a sample of post-communist countries are clearly hampered by the small size of that sample. Both populist presence and disproportionality do not attain statistical significance. The only significant coefficient is that of year which indicates a much steeper decline in turnout in post-communist countries than in Western democracies. 2 Throughout the different models, we saw a consistently positive relationship between populist presence and voter turnout, although these estimates were only seldomly distinguishable from zero. In most cases, correlations where stronger for the lag of our independent variable. In the appendix, in Tables A.3 through A.26, we present replications of the models using alternative operationalisations of our key independent variable. Results are generally weaker when other operationalisations are used. Results for Populist Party or Populist Parties point in the same direction but are weaker. We also see the pattern that correlations are weaker when controls are included. Furthermore, when looking at the results in the appendix in conjunction with the results here another pattern emerges: correlations are stronger in post-communist states than in Western established democracies. The evidence we are able to provide thus far is maybe best described as suggestive but not conclusive. However, these initial results clearly merit further analyses. 5 Conclusion Starting from the observation that many established democracies have seen an increase in populist parties and candidates contesting elections in recent decades, we tested for a possible effect of populist parties on turnout based on a comprehensive sample of national parliamentary elections. The main theoretical ideas which link populist parties to turnout can be summarized as follows. Populists may, as they sometimes claim, mobilize previous non-voters who are dissatisfied with current developments to participate in elections. Populist parties, in particular the radical right kind, may also motivate centrist non-voters to turn out to cast a vote against them. However, they may also impact negatively upon turnout if their negative rhetoric furthers people s distrust towards politics, politicians and representative democracy. 2 This may indicate that we need a more flexible approach to account for the time trend. 19

20 We test these countervailing expectations on an aggregate dataset of elections in 20 Western and 11 Central and Eastern European countries since as early as 1970 with fixed effects panel regression models. Concretely, we test whether turnout increases when populist parties compete in elections. Additionally, we included a time trend as there is a clear downward trend in turnout and an upward trend in populism. In short, we find no support for the Mobilization Hypothesis. We only find a weak indication of a positive effect of the presence of populist parties on turnout. The Disproportionality Hypothesis, which suspected that the positive effect of populist party presence is particularly outspoken in countries with electoral systems that exhibit low degrees of disproportionality could also not be corroborated. Finally, we did not find significant differences between the presence of radical left and radical right populist parties on national aggregate turnout levels. In addition, no statistically significant differences between Central and Eastern European party systems and the remaining European party systems could be discerned. Although these results are at best suggestive and by no means conclusive and merit further analyses, they suggest that populist parties are not able to spur turnout. Hence, if populist parties can indeed be a corrective for democracy, as Mudde and Rovira Kaltwasser, (2017) suggest, this does not apply to electoral participation. It is possible, of course, that populist parties spur other forms of political participation, such as portest, party membership or social movements. The recent Pulse of Europe demonstrations against nationalism and Euroscepticism across Europe are certainly a case of counter-mobilization. Mobilization against populist right protest movements such as the German Pegida. They contrast with Immerzeel and Pickup s (2015) result that East Europeans are demobilized by populist radical right parties. However, differences between countries and changes within countries over time can be and often are very different. Therefore, there is a contrast but not necessarily a contradiction. There is still much to be learned about the populism-turnout nexus. Further research should extend it to other countries, consider candidate elections, the subnational level or both. 20

21 References Armingeon, Klaus et al. (2016). Comparative Political Data Set Betz, Hans-Georg (1994). Radical Right-Wing Populism in Western Europe. Houndmills, London: Macmillan. Blais, André (2006). What Affects Voter Turnout? In: Annual Review of Political Science 9.1, pp Blais, André and R. K. Carty (1990). Does proportional representation foster voter turnout? In: European Journal of Political Research 18.2, pp Blais, André and Agnieszka Dobrzynska (1998). Turnout in Electoral Democracies. In: European journal of political research , pp Burden, Barry C and Amber Wichowsky (2014). Economic Discontent as a Mobilizer: Unemployment and Voter Turnout. In: The Journal of Politics 76.4, pp Carmines, Edward G. and James A. Stimson (1986). On the Structure and Sequence of Issue Evolution. In: The American Political Science Review 80.3, pp Dalton, Russell J. and Martin P. Wattenberg (2000). Unthinkable Democracy: Political Change in Advanced Industrial Democracies. In: Parties without Partisans. Political Change in Advanced Industrial Democracies. Ed. by Russell J. Dalton and Martin P. Wattenberg. Oxford: Oxford University Press, pp Döring, Holger and Philip Manow (2016). Parliaments and governments database (ParlGov): Information on parties, elections and cabinets in modern democracies. Development version. Downs, William M. (2012). Political Extremism in Democracies. Combating Intolerance. Basingstoke: Palgrave Macmillan. Gallagher, Michael (1991). Proportionality, disproportionality and electoral systems. In: Electoral Studies 10.1, pp Geys, Benny (2006). Explaining voter turnout: A review of aggregate-level research. In: Electoral Studies 25.4, pp Hawkins, Kirk A (2010). Venezuela s Chavismo and Populism in Comparative Perspective. Cambridge: Cambridge University Press. Hellwig, Timothy (2008). Globalization, Policy Constraints, and Vote Choice. In: Journal of Politics 70.4, pp

22 Hellwig, Timothy T. (2001). Interdependence, Government Constraints, and Economic Voting. In: Journal of Politics 63.4, pp Immerzeel, Tim and Mark Pickup (2015). Populist radical right parties mobilizing the people? The role of populist radical right success in voter turnout. In: Electoral Studies 40, pp Jackman, R. W. and R. a. Miller (1995). Voter Turnout in the Industrial Democracies during the 1980s. In: Comparative Political Studies 27, pp arxiv: Jackman, Robert W (1987). Political Insitutions and Voter Turnout in the Industrial Democracies. In: American Political Science Review 81.2, pp Kahn, Kim Fridkin and Patrick J. Kenney (1999). Do Negative Campaigns Mobilize or Suppress Turnout? Clarifying the Relationship between Negativity and Participation. In: American Political Science Review 93.4, pp Kessel, Stijn van (2015). Populist Parties in Europe. Agents of Discontent? Basingstoke: Palgrave Macmillan. Lewandowsky, Marcel, Heiko Giebler, and Aiko Wagner (2016). Rechtspopulismus in Deutschland. Eine empirische Einordnung der Parteien zur Bundestagswahl 2013 unter besonderer Berücksichtigung der AfD. In: Politische Vierteljahresschrift 57.2, pp Lewis, Paul G. (2000). Political Parties in Post-communist Eastern Europe. London and New York: Routledge. Lubbers, Marcel, Mérove Gijsberts, and Peer Scheepers (2002). Extreme right-wing voting in Western Europe. In: European Journal of Political Research 41.3, pp Mair, Peter (2006). Ruling the void? The Hollowing of Western Democracy. In: New Left Review 42, pp (2007). Political Opposition and the European Union. In: Government and Opposition 42.1, pp March, Luke (2011). Radical left parties in Europe. London: Routledge. March, Luke and Daniel Keith, eds. (2016). Europe s radical left : from marginality to the mainstream? London and New York: Rowman & Littlefield. 22

23 March, Luke and Cas Mudde (2005). What s Left of the Radical Left? The European Radical Left After 1989: Decline and Mutation. In: Comparative European Politics 3.1, pp Matsusaka, John G. and Filip Palda (1999). Voter turnout: How much can we explain? In: Public Choice , pp Meijers, Maurits J. (2017). Contagious Euroscepticism? The Impact of Eurosceptic Support on Mainstream Party Positions on European Integration. In: Party Politics 23.4, pp Meny, Yves and Yves Surel (2002a). Democracies and the Populist Challenge. Basingstoke: Palgrave Macmillan. (2002b). The Constitutive Ambiguity of Populism. In: Democracies and the Populist Challenge. Ed. by Yves Meny and Yves Surel. Basingstoke: Palgrave Macmillan, pp Mouffe, Chantal (2005). The End of Politics and the Challenge of Rightwing Populism. In: Populism and the Mirror of Democracy. Ed. by F. Panizza. London: Verso, pp Mudde, Cas (2004). The Populist Zeitgeist. In: Government and Opposition 39.3, pp (2007). Populist radical right parties in Europe. Cambridge: Cambridge University Press. Mudde, Cas and Cristobal Rovira Kaltwasser (2017). Populism: A Very Short introduction. Oxford: Oxford University Press. Offe, Claus (1996). Varieties of Transition. In: Cambridge: Polity Press. Pauwels, Teun (2014). Populism in Western Europe: Comparing Belgium, Germany and The Netherlands. London: Routledge. Ramiro, L. and Raul Gomez (2016). Radical-Left Populism during the Great Recession: Podemos and Its Competition with the Established Radical Left. In: Political Studies, pp Rooduijn, Matthijs, Sarah L. de Lange, and Wouter Van Der Brug (2014). A populist Zeitgeist? Programmatic contagion by populist parties in Western Europe. In: Party Politics 20.4, pp Schattschneider, E. E. (1960). The semisovereign people : a realist s view of democracy in America. New York: Holt, Rinehart and Winston. Smets, Kaat and Carolien van Ham (2013). The embarrassment of riches? A meta-analysis of individual-level research on voter turnout. In: Electoral Studies 32.2, pp

24 Steiner, Nils D. (2010). Economic globalization and voter turnout in established democracies. In: Electoral Studies 29.3, pp Vries, Catherine E. de and Sara B. Hobolt (2012). When dimensions collide: The electoral success of issue entrepreneurs. In: European Union Politics 13.2, pp

25 Appendix to The Limited Impact of Populist Parties on Turnout 25

26 Turnout Figure A.1: Turnout in national parliamentary elections, as early as

27 (1) year (0.0529) R N 337 Standardized beta coefficients; p < 0.05, p < 0.01 Table A.1: Fixed-effects regression model regressing turnout on time including country fixed-effects.. Range of predictions Model Range of predictions Table 2, Model (1) [52.38, 88.55] Table 2, Model (2) [57.91, 86.23] Table 2, Model (3) [66.49, 85.58] Table 2, Model (4) [55.39, 86.61] Table 3, Model (1) [61.93, 87.26] Table 3, Model (2) [53.33,87.38] Table 3, Model (3) [61.18, 86.63] Table 3, Model (4) [54.14, 86.24] Table A.2: Range of predictions generated from models presented in Tables 2 and

28 Populist Party (1) (2) (3) (4) Populist Party (1.38) (1.54) Populist Party t (1.22) (1.08) Year (0.06) (0.08) (0.06) (0.07) Disproportionality (0.15) (0.18) Unemployment (0.14) (0.13) Economic openness (0.04) (0.03) R N p < 0.05, p < 0.01 Table A.3: Fixed-effects regression models of turnout on Populist Party, models (1) and (2), and its lag, (3) and (4). All models include country fixed-effects. 28

29 (1) (2) (3) (4) Populist Party (1.46) (1.43) Populist Party t (1.33) (1.34) Year (0.06) (0.08) (0.06) (0.06) Disproportionality (0.23) (0.26) (0.18) (0.24) Pop. Party Disproportionality (0.21) (0.22) Pop. Party t-1 Disproportionality (0.17) (0.17) Unemployment (0.14) (0.13) Economic openness (0.04) (0.03) R N p < 0.05, p < 0.01 Table A.4: Fixed-effects regression models of turnout on Populist Party, models (1) and (2), or its lag respectively, (3) and (4), interacted with Gallagher s Disproportionality index. All models include country fixed-effects. 29