Voting Angry: An Exploratory Analysis of the Transformation of the German Party System∗ Peter Haschke† Paulina A. Marek‡ December 2012 Abstract Two arguments are considered to explain the increasing fragmentation of the German party system following 30 years of party system stability. According to the first argument, disillusioned citizens register protest votes that are out of line with Downsian expectations. The increase in party system fragmentation is thus seen as a consequence of voter dissatisfaction with the political mainstream. An alternative holds that the increasing fragmentation is attributable to increasing heterogeneity of policy preferences among labor. Being ignored by traditional social democratic parties, some segments of labor as a consequence are driven towards new and extreme parties. Relying on an analysis of recently published survey data, this paper finds strong support for the former but not the latter argument. ∗ We thank Dick Niemi, Curt Signorino, Kevin Clarke, Bonnie Meguid, Miguel R. Rueda, and Jonathan P. Olmsted for helpful comments, suggestions, and discussion. All errors remain our own. † PhD. Candidate, Department of Political Science, University of Rochester, Harkness Hall 333, Rochester, NY 14627, email: [email protected] ‡ Ph.D Candidate, Department of Political Science, University of Rochester, Harkness Hall 333, Rochester, NY 14627, email: [email protected] Following a period of stability which lasted from about the late 1950s through the 1970s, the German party system underwent significant changes during the last 25 years. Beginning with the emergence of the Green party in the 1980s and culminating with the consolidation of the socialist Die Linke during the first decade of the 21st century, the dominance of the center right Christian Democrats (CDU) and the center left Social Democrats (SPD), has become increasingly tenuous. Whereas the party system of the first thirty years of post-war democracy has frequently been described as a “two-and-a-half party system” (see for example Lijphart, 1999), true multi-partism has arrived, with vote shares of five parties exceeding 10 percent for the first time in Germany’s national elections in September of 2009. Although scholars have devoted considerable effort to document and explain the rise of Germany’s Green party, specifically (Kitschelt, 1989; Frankland, 1995), and the electoral fortunes of niche parties, more generally (see for example Inglehart, 1990; Kitschelt, 1994, 1995; Meguid, 2008), the electoral success of the new socialists in Germany has been a largely ignored development. Thus whereas, niche and single-issue parties, as well as nationalist and other parties of the far right, have received considerable scholarly and media attention, developments on the far left of the political spectrum have attracted only little notice among political scientists.1 Motivated by the arguably rather dramatic changes of the German party system over the last 10 years, this paper attempts to shed some light on these recent developments. In particular, this paper evaluates two possible explanations for the rise of the socialist Die Linke.2 Voters – according to the first explanation which I refer to as the protest vote argument – are said to be increasingly disaffected with the way democracy operates. Dissatisfied, especially, with the performance of the mainstream parties these individuals then as a consequence vote for parties outside of the political mainstream as a sign of protest. The recent electoral success of new political parties must thus be seen as an indicator of voter 1 The attention given to leftist and socialist parties in the literature focusing on post-communist Eastern Europe, constitutes an exception. 2 The name Die Linke translates to The Left. In order to avoid confusion, especially when considering party or voter positions on the ideological spectrum, I refer to The Left as Die Linke, throughout this paper. 1 discontent. According to the second explanation – which borrowing from Rueda (2005, 2006) I call the insider-outsider argument – the rise of new political parties can be understood as the outcome of certain segments of labor responding to new strategies adopted by traditional social democratic parties. Instead of being discontent with the performance of mainstream political parties generally, it is the policies advocated by social democratic parties specifically, that are to “blame” for the improving electoral fortunes of new parties. As citizen preferences and opinions are key to both the protest vote argument and the insider-outsider argument, this paper relies on recently updated survey data from 2011 to show that only the protest vote argument appears to be supported by the data. The paper is structured as follows. Before analyzing the data and estimating a set of multinomial votechoice models, a brief overview of the German party system is given to motivate the analysis. The subsequent two sections describe the data and results of the empirical analysis. A final section concludes. Motivation and Background In August of 1949, following the ratification of a provisional constitution (Grundgesetz or Basic Law) in May, Germany held its first democratic national elections since the collapse of the Weimar Republic. Of the ten parties winning seats to the Bundestag (the lower house), the newly formed center right Christian Democratic Union (CDU), and its Bavarian sister party, the Christian Social Union (CSU) combined, controlled about 34.6 percent of the seats in the lower house, while the center left Social Democratic Party (SPD) held about 32.6 percent. As the third largest vote getter, the Free Democratic Party (FDP) was able to secure another 12.9 percent of the seats. No other party was able to win more than 5 percent of the seats to the first Bundestag, and the three parties – the CDU/CSU (Union), the SPD, and the FDP – came to dominate German politics for the next 30 years.3 Given 3 The CDU and CSU have an electoral arrangement according to which the CDU agrees not to campaign and compete electorally in the state of Bavaria, whereas the CSU agrees to campaign and compete in Bavaria 2 that in the history of the Second Republic no single party has been able to win a majority of the seats to the legislature, all governments until 1998 thus consisted of at least two of these three parties. After the first election, and with the adoption of a nationally binding 5 percent electoral threshold in 1953, the field of parties winning seats to the lower house declined (see: Figure 1). By 1961 the Union, the SPD, and the FDP remained as the only parties winning seats to the legislature – both the Union and the SPD vying to become the largest party in the legislature, and the FDP adopting the role of “king-maker” in coalition formation bargaining.4 As evidenced by Figures 1 and 2, the emergence of the Green Party, which contested nationally for the first time in the election of 1980, broke the monopoly of the Union, the SPD, and the FDP on Bundestag seats beginning in 1983. According to Kitschelt (1994) the emergence and consolidation of the Green Party can be attributed to a failure of the SPD to adopt an appropriate and consistent electoral strategy. Kitschelt argues that the Social Democrats, particularly under the SPD chancellorship of Helmut Schmidt in the 1970s, had pursued a centripetal strategy of vote and office maximization (164). Having ignored its vulnerability to electoral challenges arising from new competitors on the ideological left, the Social Democratic party was unable to respond to growing “left-libertarian, feminist and ecologist demands” (164). After the social democratic Schmidt government had fallen and Helmut Kohl assumed the chancellorship, leading a coalition of the Union and the FDP, the SPD over the next decade vacillated between electoral strategies. Initially, the party moved sharply to the left to reclaim voters it had lost to the Greens. This oligopolistic strategy to eliminate the Green threat, however, was abandoned quickly as the SPD adopted a votemaximizing centrist strategy for the 1987 campaign. Following the defeat at the ballot box, only. As is convention, I will treat the two parties as one, referring to both of them combined as Union. 4 Of the minor parties winning seats to the first Bundestag most dissolved, joined the Union, or failed to cross the electoral threshold. Only the Communist Party (KPD) ran competitively but in 1956 the party was declared unconstitutional and it was subsequently banned from electoral competition. 3 Figure 1: Parties in the Bundestag: 1949-2009 Number of Parties 11.0 Total 6.0 5.0 Effective 4.0 3.5 3.0 2.5 2.0 0.0 1949 1953 1957 1961 1965 1969 1972 1976 1980 1983 1987 1990 1994 1998 2002 2005 2009 Election Year Note: The height of a bar represents the number of parties winning seats in the Bundestag each national election. Three independent members of the 1949 Bundestag are treated as one party. The solid line shows the corresponding effective number of legislative parties for that election. Own computation based on data taken from the website of the Federal Returning Officer (2009). however, the party yet again shifted strategies attempting to compete with the governing Union-FDP coalition by advocating more moderate redistributive policies, while at the same time engaging the Greens directly through libertarian appeals. Although the the SPD was able to regain some 600,000 votes from the Greens in the 1990 election, the Union-FDP government won the election on a wave of popular support for the Kohl government which had brought about unification with the East (Kitschelt, 1994, 166-167). The 1990 election saw yet another party win seats in the Bundestag (see again Figure 1). The successor party of the communist East German Socialist Unity Party (SED), the Die Linke, won seats due to relaxed threshold requirements.5 Despite failing to cross the 5 5 Initially the SED successor party ran under the name Party of Democratic Socialism (PDS). It later 4 Figure 2: Vote Shares by Party 50 ● ● 45 ● ● ● ● ● ● ● ● ● ● Percent of PR−Vote 40 ● 35 30 ● ● ● ● 25 Right of Center Parties Union 20 15 ● FDP ● ● 10 ● ● ● ● ● ● ● ● 5 ● ● ● ● ● 1994 1998 ● 0 1949 1953 1957 1961 1965 1969 1972 1976 1980 1983 1987 1990 2002 2005 2009 50 ● 45 ● Percent of PR−Vote 40 ● ● ● ● 35 30 ● ● ● ● ● ● ● Left of Center Parties ● ● ● 25 ● 20 SPD Greens 15 ● ● 10 ● ● 5 1949 1953 1957 1961 1965 1969 1972 1976 ● ● ● ● 1994 1998 ● ● 0 ● 1980 1983 1987 1990 ● Die Linke ● ● ● 2002 2005 2009 Election Year Note: Displayed are the vote shares of all right of center parties (top panel) and all left of center parties (bottom panel) which have won seats to the lower house following the first three consolidating elections. The vote shares reflect the proportional vote (Zweitstimme) only. As such the data points only approximate the seat shares of the parties in the legislature. Note, also, the solid black line indicating the 5% electoral threshold. Vote shares are displayed for all elections contested by the respective party. Data source: Federal Returning Officer (2009). percent threshold, which again applied in 1994, the Die Linke, due to highly concentrated support in the formerly East German states, was able to take advantage of Germany’s mixed member proportional electoral system. Winning a plurality in several districts in Berlin, the Die Linke was thus able to retain a presence in the Bundestag until the party finally crossed the electoral threshold in 1998. Surprisingly, despite two leftist parties challenging the Social Democrats on the left, the 1998 election resulted in the ouster of the center right Union-FDP government. As a whole, parties traditionally considered left of center won more than 50% of the popular vote, and having moved sharply to the right under the leadership of Gerhart Schröder, the SPD, in coalition with the Green party, returned to the chancellory. competed as Linkspartei/PDS (Left Party/PDS) before finally adopting the name Die Linke. 5 Importantly, however, the German party system, despite the entering of the Green Party in 1983 and the Die Linke in 1990, remained surprisingly stable until 2005. As evidenced in Figure 1, the effective number of legislative parties increased only slightly following the addition of these two parties.6 With the 2005 election, however, the effective number of parties increases rapidly to levels not seen since 1949, with the FDP, the Greens and Die Linke achieving double digit vote totals.7 It is the purpose of this paper to evaluate two competing hypotheses that could explain the decline of the Union and the SPD and the increasing fragmentation of the German party system, and in particular the rise of Die Linke. Two Explanations The proliferation of new parties in the electoral systems of the world since the 1960s has spawned a vast literature attempting to understand and explain the success of parties outside of the political mainstream. Attempts to explain the explosion of new parties (often referred to as single issue or niche parties) can be grouped roughly into two broad categories – a category of institutional explanations and a category of sociological explanations (see for example: Meguid, 2008). Proponents of the institutional approach have long argued that the institutional context in which political actors operate, plays an integral part in influencing the behavior of both voters and political elites. Electoral rules, state structure and governmental type have thus been the primary focus of scholars in this vein. Building on work by Duverger (1954); Lijphart (1994); Cox (1997), Jackman and Volpert (1996), for example, stress the importance of party-system factors, arguing that these factors can explain the success of new parties on 6 The effective number of parties is calculated by dividing 1 by the sum of the ith party’s squared seat share. It can be understood as a measure of the fractionalization of the party system, which discounts extremely small parties. 7 The standard deviation of the of the effective number of parties over the history of the Second Republic (since the adoption of the 5% national threshold in 1953) is 0.45. The 2005 and 2009 elections produces an increase of 0.64 and 0.53 effective parties respectively. 6 the extreme right. They find, for instance, that the proportionality of election rules is positively related to the likelihood of formation and electoral success of extremist parties. Electoral disproportionality in electoral systems such as those with single-member districts or those employing high electoral thresholds, undermine the success of these new parties by raising the cost of entry (517). The determinants of a parties electoral fortunes, according to the sociological approach, are to be found in the salience of the issues advanced by the relevant parties. Proponents of this approach thus point to the socio-economic and cultural particularities of the electorate to explain the electoral success of new parties. Changing voter attitudes and values are of critical importance to explanations of the success of Green Parties and new political movements (Inglehart, 1990; Kitschelt, 1994). Research also suggests that the success of radical right-wing parties is a direct consequence of socioeconomic conditions within a country (Betz, 1993; Kitschelt, 1995). A common line of argument, for example, holds that economic misfortune, in particular, high levels of unemployment, push the marginalized to cast their votes for anti-system and radical right-wing parties (Betz, 1990; Swank and Betz, 2003; Golder, 2003; Givens, 2005). It should be clear that the institutional approach should be of limited utility in explaining the developments of the German party system in recent years. As argued above, institutional factors have remained more or less constant throughout the history of the Second Republic. Neither the state-structure nor governmental type have changed. The developments of the party system since 2005 can thus not be attributed to increasing decentralization, or to a shift from a presidential to a parliamentary system, as these factors remained constant. The electoral rules, also have not changed significantly since the adoption of the the national 5 percent threshold for the 1953 election.8 A focus on the sociological approach is therefore warranted. 8 In the 1949 election a 5 percent threshold only was used only at the state level. The election of 1990 can perhaps be seen as an exception, as two thresholds were in place for this election. A party would gain representation in the Bundestag if it crossed the 5 percent threshold in either the states of former West Germany or in the former East German states. 7 The Protest Vote Argument The first explanation evaluated in this paper, I call the protest vote argument. According to this argument, the electoral success of new parties in Germany’s recent elections can be explained as consequence of voter dissatisfaction with the performance of the political establishment on the one hand, and disillusionment with the operation of democracy on the other. This argument suggests that supporters of parties of the political mainstream have become increasingly dissatisfied with the parties performance in addressing the problems and concerns important to voters. Importantly, the disenchanted, do not have particularly extreme policy demands. As a matter of fact, if the protest argument is correct, much of the support for new parties – parties which as a group tend to advocate more extreme policy positions – comes from voters who are ideologically moderate yet support these parties merely as a sign of political protest. The protest vote argument yields the following testable hypotheses. Hypothesis A1: If the protest voter argument is correct, all mainstream parties should be more or less equally affected by desertion of dissatisfied voters. If only one of the mainstream parties is abandoned in favor of non-mainstream alternatives, a party specific explanation is preferable. Hypothesis A2: All non-mainstream parties should benefit from protest voting. Again if only one of the non-mainstream party attracts protest voters a party specific explanation might be preferable. Hypothesis A3: Lastly, and most importantly, protest voters are moderates. Rather than supporting non-mainstream parties for their relatively more extreme platforms, protest voters support these parties despite of their relatively extreme platforms. In a Downsian sense, protest voters vote incorrectly (i.e. they do not vote for the party closest to them). 8 The Insider-Outsider Argument The second argument evaluated in this paper, borrows heavily from Rueda (2005, 2006). Rueda’s work challenges the assumption that social democratic parties are monolithic defenders of the working class. He claims that although most of the comparative economy literature treats social democratic parties as champions of labor (see among others: Hibbs (1977); Alesina, Roubini and Cohen (1999); also Bartels (2008)), this assumption is no longer warranted. Specifically, the conventional division of voters into upscale groups, those disproportionately affected by inflation, and labor, those disproportionatly affected by unemployment is inappropriate. Instead, labor itself must be disaggregated into two groups – insiders and outsiders. Following a period of economic growth and stability of the late 1960s combined with social unrest and union activism in the early 1970s, “highly restrictive tenure and severance pay arrangements” were accepted by firms in industrialized countries (Rueda, 2005, 63). As a consequence, large segments of labor became relatively insulated from market fluctuations and unemployment. Those segments that benefitted from these developments, and which are thus more or less immune to labor demand shocks, Rueda calls “insiders”. At the same time, however, many OECD countries experienced labor supply shocks following the Oil Shock crises (63). Women entering the labor force, and “a general intensification of international competition” thus lead to an increase of part-time work and temporary employment contracts (63). Individuals that are more precariously employed and more affected by economic fluctuations thus make up a second segment of labor, which Rueda calls “outsiders”. Importantly therefore, insiders and outsiders, differ in terms of their vulnerability to unemployment and labor cannot be conceptualized as a homogenous political actor. It follows that social democratic parties face important choices, insofar as a broad focus on reducing unemployment will only benefit part of the parties’ traditional constituency. Whereas outsiders are disproportionately interested in reducing unemployment, insiders are seen to be more concerned with maintaining their job security, while securing other work related 9 benefits. Unfortunately, the policy demands of these two groups are not necessarily consistent and might actually be conflicting. Demands for more job security and employment protection legislation, for example, could actually hinder hiring, as firms would become less flexible in their ability to respond the market fluctuations which might require them to shed some of their workforce. In his research Rueda finds that social democratic parties are more likely to cater to insiders than to outsiders because among other reasons, insiders tend to be more active politically, and thus a more reliable source of votes (62). The rise of insider-outsider politics in Rueda’s work is therefore seen as an important determinant of social democratic policy (2006, 387). Interestingly, Rueda speculates that if outsiders are ignored by social democratic governments specifically, or social democratic parties more generally, “powerful anti-system parties (especially extreme right ones)” could emerge (2005, 72). Although outsiders may have reason to abandon mainstream parties in favor of anti-system parties of the far right, the availability of a party on the far left of the political spectrum advancing stark redistributive policies could be even more appealing. As such, the increasing fragmentation of Germany’s left of center parties and particularly the rise of the decidedly socialist Die Linke, could be explained employing Rueda’s insider-outsider framework. His framework suggests the following hypotheses. Hypothesis B1: Outsiders are more likely to support anti-system parties, particularly those of the far right end of the political spectrum. Hypothesis B2: Outsiders are more likely to support the socialist Die Linke given the lack of a viable anti-system party on the right. Hypothesis B3: The mainstream social democratic parties (in this case the SPD) should bear the brunt of desertion by outsiders, whereas parties who’s core constituency are upscale 10 segments of society should be unaffected by insider-outsider politics. Methodological Approach and Data To assess the explanatory power of the protest vote and the insider-outsider argument, I rely on survey data taken from the European Social Survey 2008 (Norwegian Social Science Data Services, 2009).9 Before estimating a set of regressions this section provides an initial descriptive overview of the data to determine whether the two explanations indeed merit further analysis. Additionally, this section serves as an outline of the operationalization of the key variables used to test the sets of hypotheses above. Given the nature of the two explanations, survey responses relating to partisanship and vote choice are of critical importance to this analysis. All survey respondents were asked whether they considered themselves close to a particular party and if so were asked for the party. Furthermore, all respondents were also asked how they voted in the last national election, and where they would place themselves on a 11-point scale, ranging from extreme left to extreme right. Figure 3 provides an overview of voter self-placement and voting behavior among survey participants. As can be seen, the voting behavior of survey participants approximates the election outcome of the 2005 election relatively well. About 50% of respondents report having voted for the center-right Union and the center-left SPD. The minor parties (FDP, Greens, and Die Linke) were supported by about 30% of respondents (only 9 respondents claimed to have voted for parties of the extreme right).10 Importantly, just over 20% of respondents did not vote. The right panel of Figure 3 shows the distribution of survey respondents on the Left-Right scale. As can be seen most respondents tend to be moderates. The mean self-placement is 9 The 4th wave of the survey used for this paper was fielded in over 30 countries during the years 2008 and 2009. The German part of the survey was conducted in 2008, one year prior to the 2009 national elections. 10 Multiple extreme right parties compete in Germany’s national elections. The three parties that were mentioned by survey respondents were the German Peoples Union (DVU), the National Democratic Party (NPD), and the Republikaner (REP). I treat these three parties as one in this paper. 11 Figure 3: Vote Choice and Ideology 25 0.30 20 0.25 Vote Choice 0.20 Abstained 15 Party Extreme Right FDP Greens Density Percent Die Linke 0.15 SPD 10 Union ● Die Linke ● Extreme Right ● FDP ● Greens ● SPD ● Union 0.10 5 0.05 ● 0.00 0 0 Vote 1 2 3 ●● 4 ●● 5 6 7 ● 8 9 10 Left−Right Scale Note: The left panel shows the voting behavior of all respondents who were able to recall whether and for whom they voted in the 2005 election. The right panel shows the distribution of those individuals on the Left-Right scale. The vertical lines in the right panel represent the means of the Left-Right scales among respective partisans. about 5.3, the median is 6. The means among partisans of the various parties conform to expectations. The mean among partisans of the socialist Die Linke, for example is relatively far on the left (the mean ≈ 3.0). The partisans of the social democrats and the Green Party are not discernable in terms of their self-placement, yet they place themselves significantly further to the center of the scale (the means ≈ 4.5 and 4.3, respectively). Much the same can be said about partisans of the Union and the FDP which are located closely together on the center right of the Left-Right scale (the means are about 6.7 and 6.5, respectively). Partisans of extreme right parties locate themselves far on the right (mean ≈ 8.5). Throughout the remainder of the analysis I take the mean positions of partisans of the respective parties as their parties’ position on the left-right scale. According to the protest vote argument, disenchanted and dissatisfied individuals aban12 Figure 4: Voter Satisfaction Die Linke Die Linke Die Linke Extreme Right Extreme Right Extreme Right FDP FDP FDP Greens Greens Greens SPD SPD SPD 1 3 5 7 9 11 Satisfaction with Government Union Union Union Satisfaction with Democracy Abstained Abstained Abstained 1 3 5 7 9 11 1 3 5 7 9 11 Satisfaction with the Economy Note: Each column shows the distribution of voters of the various parties on a 10point scale representing their satisfaction with democracy (left column), satisfaction with government (center column), and satisfaction with the economy (right column). The vertical lines represent the median level of satisfaction of all respondents for the respective issue. don mainstream parties in favor of more extreme non-mainstream alternatives. Critically, Hypothesis A3 holds that these protest voters are ideologically moderate, voting for more extreme parties only as a sign of disillusionment. If this argument is correct, one would expect that voters of non-mainstream parties (especially voters of Die Linke, the Greens, and the extreme right), are more moderate in terms of their self-placement on the Left-Right scale than are partisans of those parties. When comparing the means of self-placement among partisans and voters, the evidence appears consistent with Hypothesis A3. Voters of the Green Party and Die Linke (with means of about 4.5 and 3.4 respectively) tend to be slightly more moderate than partisans of these parties (with means of 4.3 and 3.0 respectively). The same 13 is true for voters of the extreme right parties who tend to be more moderate than partisans. The partisan mean is 8.5, the mean among voters 7.7.11 Of course this evidence is only suggestive, as the difference between the means depends on how many voters of the respective parties are also partisans of those parties. Also this simple comparison ignores strategic voting calculations. SPD voters for example have strong incentives place their vote for the Greens to ensure that the Green party crosses the 5% threshold and is available as a potential coalition partner. Further descriptive evidence for the protest vote argument, however, is found when looking voter satisfaction. The survey contained three items asking respondents to express their degree of satisfaction with the operation of democracy, satisfaction with government, and satisfaction with the state of the economy. Voters were asked to state their degree of satisfaction on a 10-point scale ranging from 1 if “extremely dissatisfied” to 11 “extremely satisfied”. Figure 4 shows the distribution of voters of the various parties on these scales. What can be seen is that voters of two nonmainstream parties (Die Linke and the parties of the extreme right) are significantly less satisfied across the three issues. Interestingly, voters of the Greens and the FDP are more satisfied than voters of the other two minor parties. To provide an initial evaluation of the insider-outsider argument, I follow Rueda (2005, 2006) and divide survey respondents into three groups – insiders, outsiders, and upscales. Insiders are defined as those individuals who are employed full-time with permanent job, or as those with part-time employment but no desire to work full-time. Individuals in this group have permanent or unlimited work contracts. Outsiders on the other hand, are defined as those who are unemployed or as those who are employed full-time in temporary or fixedterm jobs.12 Finally the last group, upscales, are defined individuals who are self-employed (including professionals, owners of shops and small businesses, but also employed managers). 11 Using a simple one-tailed t-test, the difference in means is significant at the 90% level for the Greens and Die Linke. The difference is not statistically different for voters of the extreme right. 12 According to Rueda, students must be included in the outsider category as well, as “they have no certainty about their future employment” (Rueda, 2005, 63). However, all individuals below the voting age of 18, are dropped from the analysis leaving very few students in the sample. 14 Figure 5: Policy Demands by Type Type Insiders Outsiders Upscales 4.00 4.25 4.50 4.75 5.00 Taxes and Social Spending (10=Raise by a lot) Type Insiders Outsiders Upscales 5.00 5.25 5.50 5.75 6.00 6.25 6.50 6.75 7.00 Guaranteed Job (10=Gov't Responsibility) Type Insiders Outsiders Upscales 6.50 6.75 7.00 7.25 7.50 Paid leave (10=Gov't Responsibility) Note: Displayed are the mean responses according to questions regarding labor market policy and employment protection and benefits for insiders, outsiders, and upscales. According to the insider-outsider argument, members of these three groups differ in terms of their preferences and policy demands, especially those regarding labor market policy and employment protection. Figure 5 reports information about these preferences by type. The top panel of Figure 5 shows the mean responses of insiders, outsiders, and upscales to the question of whether government should raise or lower taxes and social spending.13 According to the insider-outsider argument, both upscales and insiders should be less supportive of increased social spending and higher taxes, because both groups by virtue of their higher job-security are less interested in “dedicating more resources to labor market policies” (2005, 64). As can be seen in the top panel of the figure, this expectation finds some support.14 13 Respondents were offered 11 choices ranging from 0, if they believed that “government should decrease taxes a lot and spend much less of social benefits and services” to 10 if they believed taxes should be raised a lot. 14 A one-tailed t-test shows that the mean for outsiders is different at the 90% level. 15 A similar finding is revealed when looking at the middle panel of Figure 5. Respondents were asked to state their view as to whether it is or is not the governments responsibility to guarantee jobs for all those who need one. Again, outsiders are most likely to believe that government is responsible for providing guaranteed employment, with insiders falling somewhere in the middle, and upscales least likely to hold the government responsible. Finally, in the bottom panel of Figure 5, insiders are seen to be most likely to hold views that it is governments responsibility to ensure that in cases of emergencies or pregnancy, employees are guaranteed paid leave. This conforms with the insider-outsider argument, according to which insiders’ demands regarding employment security and demands should differ from the policy demands made by outsiders. Outsiders should be more concerned with active labor market policies such as job training, and unemployment insurance, whereas insiders’ preferences put higher emphasis on maintaining job security and achieving additional employment benefits such as paid leave in cases of emergencies. The initial evidence provided in Figure 5 thus suggests that it is indeed reasonable to disaggregate labor into insiders and outsiders, and that this framework can serve as an explanation for the developments of the German political landscape in recent years. The Models The descriptive overview of the data presented above, of course, only offers cursory evidence as to the reasonableness of the protest vote and insider-outsider arguments. It is the purpose of this part of the paper to outline a set of regression models to assess the factors that are associated with citizens voting choices. To evaluate the explanatory power of the two arguments, three multinomial logit models are estimated. I first estimate a base model, regressing vote choice on a set of control variables. Then I estimate two additional models, one containing a set of regressors for the protest vote argument, and one containing a set of regressors for the insider-outsider argument. Each of the two additional models contain the variables of the base model as controls. The operationalization of the variables included in 16 the three models can be found in Table 1 below. Variable Name DV: Vote Choice Table 1: Models and Variables Operationalization a set of indicator variables, one for each of the following voting choices: SPD, Union, FDP, Greens, Die Linke, Extreme Right, and abstention; Base Model: Age respondent age in years; Gender coded 1 if the respondent is male, zero otherwise; Education respondent educational attainment in years of schooling; Income respondent income percentile; Rural dummy variable coded 1 if the respondent lives in “country village”, a “farm or a home in the countryside”, zero otherwise; Region dummy variable coded 1 if the respondent lives in an east Germany zero otherwise; Ideology respondents self-placement of the Left-Right Scale; Strategic Vote Protest Vote Model Satisfaction-Dem coded 1 if the respondent voted for different parties with her PR and SMD vote; zero otherwise (plus Base Model variables) Respondent’s level of satisfaction with democracy; Satisfaction-Gov Respondent’s level of satisfaction with government; Satisfaction-Econ Insider-Outsider Model Insider Respondent’s level of satisfaction with the economy; (plus Base Model variables) dummy variable coded 1 if the respondent is an insider, zero otherwise; Outsider dummy variable coded 1 if the respondent is an outsider, zero otherwise; Across all models the choice, Union, is treated as the reference category. 17 Results and Discussion The Base Model Before evaluating the explanatory merit of the protest vote and the insider-outsider arguments, a multinomial logit regression was estimated for a set of variables typical to vote choice model. The dependent variable contained seven choices, one for each party (Union, SPD, FDP, Greens, Die Linke, and Extreme Right) as well as the choice to not vote at all (Abstention). The choice to stay at home on election day was included because this option is directly relevant to the protest vote and the insider-outsider argument. The ability not to vote, for example, can be seen as a form of protest. Additionally it can also be seen as an option available to outsiders who do not feel adequately represented by social democratic policy making. As independent variables, a respondent’s age, gender, educational attainment, and income are common and generally uncontroversial in research of electoral behavior and public opinion. Two explanatory variables in the base model are, however, of particular importance for the analysis of German voting behavior. The first such variable is Region. This variable is coded 1, if a respondent lives in the formerly communist east German states. The inclusion of this indicator is necessary because election outcomes, even 20 years after the fall of the Berlin Wall, differ significantly between the East and West. The most important difference relates to much larger support for Die Linke. As argued above, Die Linke founded as the successor to the pre 1990 Socialist Unity Party, considers the east German states its regional stronghold. To evaluate the arguments of interest, it is necessary to control for this regional variation. A second variable of importance is the variable Strategic Vote. This variable is coded 1 if respondents split their two votes between two parties and zero otherwise. The ability to assess the protest vote argument, which holds that mainstream parties will be abandoned by disillusioned and dissatisfied individuals, is thus complicated by the fact that many support18 Figure 6: Predicted Probabilities: Base Model 0.8 ● 0.4 ● ● ● 0.7 ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.6 ● ● 0.3 ● ● ● ● ● ● ● ● Probability Probability ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 ● ● ● ● ● ● 1 ● ● ● ● ● ● 2 ● ● ● ● ● ● ● ● ● ● 3 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 ● ● ● ● ● ● ● 0.1 ● 0.1 ● ● ● ● ● 0.2 ● ● ● ● ● ● ● ● ● ● 0.4 0.2 ● ● ● ● 0.3 ● ● ● ● ● ● ● ● ● ● ● ● 0.5 ● ● ● 4 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 5 ● ● 6 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 7 ● ● ● ● ● ● 8 ● ● ● ● ● ● 9 ● ● ● ● 0.0 10 1 ● ● 2 ● ● 3 ● 4 Left−Right Scale ● ● ● 5 ● 6 7 8 9 10 Income Percentile 0.7 0.6 Vote Choice 0.5 Union SPD Greens Probability 0.4 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Abstention ● ● ● ● ● ● ● ● ● ● ● ● 5 ● ● ● 0.0 ● ● ● ● ● ● ● ● ● ● ● Extreme Right ● ● ● ● ● ● 0.1 ● ● ● ● 0.2 ● ● Die Linke FDP ● ● 0.3 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 10 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 15 ● ● 20 25 Years of Education All three panels display the predicted probabilities of voting for a particular party for typical survey participant (A profile holding all regressors at the respective sample mean or median). The typical respondent is a 43 year old West German male, with 14.6 years of education, in the 50th income percentile, living outside of rural areas, who is ideologically moderate (mean = 5.3), and who does not vote strategically. The top left panel varies the respondent’s self-placement on the 11-point left-right scale. The panel on the top right, varies Income, and the panel on the bottom, educational attainment. 19 ers of the mainstream parties abandon their party strategically to ensure that their preferred choice, will be able to find a suitable coalition partner. As such it is quite common for satisfied voters to vote against their first preference with their proportional vote, in order to ensure that their preferred coalition partner crosses the 5 percent electoral threshold while only voting for their preferred party with their district or SMD vote. To account for this strategic behavior the inclusion of the variable Strategic Vote is warranted. The estimated coefficients for the base model can be found in Table A1 in the appendix. Because a multinomial logit regression with seven choices and eight regressors, results in the estimation of 54 coefficients, the results are difficult to interpret directly. In order to interpret the effects of regressors, I thus rely primarily on plots of predicted probabilities. Figure 6, provides a set of predicted probability plots for the base model. Across all three panels the predicted probabilities for voting for a particular party (or not voting at all) are computed for a profile representing a typical survey respondent. The predicted probabilities reported in Figure 6 are quite intuitive. In terms of respondent ideology, more leftist respondents tend to vote for more leftist parties, whereas more conservative voters vote increasingly for parties on the right of the political spectrum. Interestingly, among the most leftist voters the probability for voting for the center-left social democratic party is still higher than for the socialist Die Linke. Similarly, among the most conservative respondents the probability of voting for the center-right Union, is estimated to be almost 80% with other center right parties receiving less than 10%. The effect of the typical respondent’s income is less powerful predictor of vote choice. As a matter of fact the predicted probabilities for the minor parties are relatively similar across income percentiles. Income has the most significant effect on the probabilities of voting for the two large parties. As the income percentile falls from the 100th to the 10th the probability of voting for the Union and the SPD declines by over 10% respectively. Other parties, however, do not benefit from this decline as most of those who abandon these two centrist parties simply stay at home on election day, to the point where among the poorest 20 but otherwise typical respondents the probability of abstention is estimated at just under 35%. The bottom panel of Figure 6 reveals the impact of varying levels of educational attainment among typical respondents. As can be seen, among the highly educated, the Greens have the largest estimated probability with around 35%, whereas among the least educated, the predicted probability of abstention is highest with about 70%. The Protest Vote Model Having provided an overview of some of the more important regressors of the base model, I now turn to an evaluation of the protest vote argument. As outlined in Table 1, the model estimated to evaluate the protest vote argument includes all regressors of the base model plus three variables measuring respondents degree of satisfaction with democracy, government, and the economy. The estimated coefficients for the protest vote model can be found in the second column of Table A1 in the appendix. Again a figure with predicted probabilities is provided to visualize the effects of the regressors of interest (see Figure 7). I begin with Hypothesis A1 and A2, according to which all mainstream parties should be equally affected by loosing support among discontent individuals, whereas all non-mainstream parties should gain equally from discontent. The top panel of Figure 7 is designed to test these hypotheses. The predicted probabilities in that top panel are computed as follows. Each line in the plot is computed for a different profile of respondents (as such the predicted probabilities of the six lines do not sum to one). Specifically, the profile used for a given line represents a typical partisan of that party. For example, the profile used to compute the black line, defines a typical Union partisan. All regressors for this line are held at the respective means or medians of Union partisans, whereas the profile used to compute, say, the red line defines the typical SPD partisan. Having, thusly specified six unique partisan profiles, I then computed the probability of voting for that same party. In other words the black line shows the probability of voting for Union for respondents that look like the typical 21 Figure 7: Evaluating the Protest Vote Argument Partisanship and Vote Choice Agreement ● ● 0.6 ● ● ● ● 0.5 ● ● ● ● ● ● ● ● 0.4 Typical Partisan Probability ● ● Union ● SPD Green ● 0.3 Die Linke ● ● 0.2 FDP Extreme Right ● ● ● ● ● 0.1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 2 ● ● ● ● ● ● ● ● ● 1 ● ● 3 4 ● ● ● ● ● ● ● ● ● 5 6 7 8 9 10 11 Overall Satisfaction Vote Choice among Typical SPD Constituents Vote Choice among Typical Union Constituents ● ● 0.6 0.6 ● ● ● 0.5 ● ● ● 0.5 ● ● Vote Choice ● Union ● ● 0.4 SPD ● 0.4 Green ● ● 0.3 ● ● ● ● Probability Probability ● Die Linke ● FDP 0.3 Extreme Right ● ● ● ● 0.2 ● ● 0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 2 3 ● ● ● Abstention ● ● 0.1 ● ● ● 0.0 1 ● ● ● ● ● ● ● ● ● ● ● 0.1 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 4 5 6 7 8 9 10 11 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 5 6 7 8 9 10 11 0.0 1 Overall Satisfaction 2 3 4 Overall Satisfaction The top panel of the figure shows predicted probabilities of voting for the respective parties for typical partisans of these parties. Each predicted probability is computed for a profile of the respective typical partisan identifier. The bottom panels display the predicted probabilities of voting for a given party or abstaining for typical SPD identifiers (bottom left) and typical Union identifiers (bottom right). 22 Union partisan. What can be seen in the top panel of Figure 7 then are the probabilities with which individuals that look like partisans of the respective parties actually vote accordingly as their level of overall satisfaction varies.15 It is clear, that individuals who look like partisans of the two mainstream center-right and center-left parties (Union and SPD), are increasingly less likely to vote according to expectations as their level of satisfaction decreases. Moreover, the decline of support among their their core constituents as satisfaction declines is dramatic. Whereas the probability of voting for the Union and SPD are 65% and 50% respectively among highly satisfied typical partisan identifiers, these probabilities drop to 20% among the least satisfied. Although the relationship is much weaker among partisans of the Green Party and the FDP, their support also declines with decreasing overall satisfaction. Considering that the Union and SPD and their coalition partners the FDP and the Greens have held a monopoly on all German governments, they can be considered mainstream. Given that they all lose support among individuals approximating their respective core constituents as they become dissatisfied, whereas the two other parties gain support as their constituents become less satisfied, this must be seen as strong evidence in favor of Hypothesis A1 as well as Hypothesis A2. To probe this finding deeper and to evaluate Hypothesis A3, according to which most protest voters are moderate, I include two additional plots in Figure 7. The two bottom panels of the figure, show the predicted probability of voting for the respective parties or abstaining, among ideologically moderate voters who otherwise look like partisans of the SPD (bottom left panel) and partisans of the Union (bottom right panel). Note that in these two plots the probabilities again must sum to one. The results are astonishingly consistent with the protest vote argument. It can be seen in the left panel, that as moderate respondents who otherwise approximate SPD partisans become increasingly dissatisfied, the probability of voting for the non-mainstream Die Linke increases. Among the least satisfied SPD constituents the probability of voting for the Die Linke in fact mirrors the probability 15 A value of Overall Satisfaction of 1 represents extreme dissatisfaction (on all three satisfaction measures), whereas a value of 11 represents extreme satisfaction (on all three satisfaction measures). 23 of voting for the Linke among their own partisans (this can be seen when comparing the line of Die Linke in the top panel to the line of Die Linke in the bottom left panel). The fact that dissatisfied yet moderate SPD constituents vote for Die Linke to the same degree as the more extreme partisans of Die Linke is consistent and strong support for protest vote Hypthesis A3. Consistent also with this hypothesis, moderate supporters of the CDU, as they become dissatisfied, are predicted to vote for the Extreme Right. The predicted probability of voting for the extreme right among typical Union partisans in fact exceeds the probability of voting for Union among the most dissatisfied. The Insider-Outsider Model Having found strong support for the three hypotheses associated with the protest vote argument, I next turn to those associated with the insider-outsider argument. As speculated by Rueda (2005), Hypothesis B1 holds that outsiders are more likely to support anti-system parties, particularly those on the far right, whereas Hypothesis B2 states that outsiders should turn to the socialist Die Linke if no viable right wing anti-system party exists. To evaluate these claim, I estimated a final regression containing two dummy variables according to Rueda’s insider-outsider framework. The estimated coefficients can be found in Table A1 in the appendix. As can be seen in Figure 8, neither Hypothesis B1 or B2 seem to be supported. The figure plots the predicted probabilities of individuals of a certain type (insiders, outsiders, and upscales) with otherwise mean or median characteristics of voting for a particular party. The predicted support for extreme right parties or Die Linke remain relatively unchanged, when controlling for other factors yet varying respondent type. Hypothesis B3 which holds that the social democratic SPD should bear the brunt of defection from outsiders is equally unsupported. Although support for the SPD is predicted to be lowest among outsiders, the same is true for Union or any other party for that matter. Most importantly, outsiders are not more likely to vote for anti-system parties or for parties 24 Figure 8: Evaluating the Insider-Outsider Argument Upscales 0.4 Vote Choice Probabiltity 0.3 Abstained Die Linke Extreme Right 0.2 FDP Greens 0.1 SPD Union 0.0 Abstained Die Linke Extreme Right FDP Greens SPD Union Vote Choice Insiders 0.4 Vote Choice Probabiltity 0.3 Abstained Die Linke Extreme Right 0.2 FDP Greens 0.1 SPD Union 0.0 Abstained Die Linke Extreme Right FDP Greens SPD Union Vote Choice Outsiders 0.4 Vote Choice Probabiltity 0.3 Abstained Die Linke Extreme Right 0.2 FDP Greens 0.1 SPD Union 0.0 Abstained Die Linke Extreme Right FDP Greens SPD Union Vote Choice Plotted are the predicted probabilities of voting for a particular party among upscales, insiders, and outsider. The profiles used to compute the predicted values holds all variable constant at their respective means and medians. of the extreme left or right, but to not vote at all. As can be seen in the bottom panel of Figure 8 a plurality of outsider is predicted to abstain. Although Rueda’s framework appears reasonable and appealing, its only merit for the purpose of this paper is separating voters from non-voters. 25 Conclusion Employing survey data from 2008, this paper attempted to shed some light the recent developments of the German party system. In light of the increasing fragmentation of the German party system, this paper considered two arguments to explain why voters have increasingly turned away from traditional mainstream parties. According to the first argument, voters – if disillusioned and dissatisfied with the political process – have strong incentives to register votes that conflict with typical Downsian expectations. Instead of voting for their preferred alternative, these voters choose to register protest votes by voting for non-mainstream and even extreme parties despite their own moderate political preferences. The second argument, building on work by Rueda (2005, 2006) suggested that the increasing fragmentation of the German party system can be explained by the increasing heterogeneity of preferences within labor. Rueda argued that Labor must be disaggregated into insiders and outsiders because these two groups have inconsistent and often conflicting preferences regarding labor market policies. Because social democratic parties can only cater to one of these groups, one must necessarily be ignored seeking redress from another political party. Three multinomial vote-choice models were estimated to assess the explanatory power of these two arguments. Substantively, strong empirical evidence was found in support of the protest vote argument. Disillusioned voters do indeed “vote wrong”, that is they turn away from the party that seems to most closely represent their interests. Instead these voters vote for parties outside the political mainstream, despite or rather because these parties are extreme and non-mainstream. The insider-outsider argument on the other hand seemed to be little supported by the data. The social democratic party does not appear to be abandoned by outsiders in favor of extreme anti-system parties or parties on the far left. Rather outsiders tend to not vote at all. The increasing fragmentation of the German party systems thus appears to be not a consequence of increasing conflict within labor but rather a consequence of moderate but dissatisfied voters choosing spatially distant parties in an act of electoral protest. 26 Appendix Figure 9: Distribution on the Left-Right Scale by Type 0.30 0.25 0.20 Density Type Insiders 0.15 Outsiders Upscales 0.10 0.05 0.00 0 1 2 3 4 5 6 7 8 9 10 Left−Right Scale Plotted are the distributions of insiders, outsiders and upscales on the LeftRight Scale 27 Variable Choice: Abstention Constant Age Gender Education Income Rural Region Ideology Strategic Vote Satisfaction-Dem Satisfaction-Gov Satisfaction-Econ Insider Outsider Choice: Die Linke Constant Age Gender Education Income Rural Region Ideology Strategic Vote Satisfaction-Dem Satisfaction-Gov Satisfaction-Econ Insider Outsider Choice: Extreme Right Constant Age Gender Education Income Rural Region Ideology Strategic Vote Satisfaction-Dem Satisfaction-Gov Satisfaction-Econ Insider Outsider Choice: FDP Constant Age Gender Education Income Rural Region Ideology Strategic Vote Satisfaction-Dem Satisfaction-Gov Satisfaction-Econ Insider Outsider Choice: Greens Constant Age Gender Education Income Rural Region Ideology Strategic Vote Satisfaction-Dem Satisfaction-Gov Satisfaction-Econ Insider Outsider Choice: SPD Constant Age Gender Education Income Rural Region Ideology Strategic Vote Satisfaction-Dem Satisfaction-Gov Satisfaction-Econ Insider Outsider N Log-Likelihood BIC Table A1: Results Base Model Protest Vote Model 8.36 -0.05 0.13 -0.25 -0.21 -0.20 0.32 -0.40 -13.70 (0.90) (0.00) (0.22) (0.00) (0.00) (0.22) (0.22) (0.10) (93.1) 10.12 -0.05 0.19 -0.23 -0.18 -0.37 0.03 -0.40 -13.78 -0.23 -0.06 -0.06 (0.98) (0.00) (0.22) (0.00) (0.00) (0.22) (0.24) (0.10) (97.2) (0.00) (0.00) (0.00) Insider-Outsider Model 7.70 (1.03) -0.04 (0.00) 0.10 (0.22) -0.25 (0.00) -0.15 (0.00) -0.16 (0.22) 0.29 (0.22) -0.42 (0.10) -13.66(99.7) -0.01 (0.39) 1.12 (0.44) 5.57 -0.01 0.42 -0.08 -0.15 -0.35 1.06 -1.05 1.09 (1.07) (0.00) (0.28) (0.00) (0.00) (0.30) (0.30) (0.10) (0.33) 7.12 -0.01 0.37 -0.06 -0.12 -0.46 0.92 -1.03 1.05 -0.11 -0.32 0.08 (1.15) (0.00) (0.30) (0.00) (0.00) (0.32) (0.32) (0.10) (0.35) (0.10) (0.10) (0.10) 6.26 -0.01 0.39 -0.09 -0.14 -0.32 1.06 -1.07 1.11 (1.25) (0.00) (0.30) (0.00) (0.00) (0.32) (0.30) (0.10) (0.35) -0.48 (0.45) -0.12 (0.54) -12.92 -0.00 12.13 -0.26 -0.15 0.28 -0.06 0.20 2.22 (96.8) (0.00) (96.7) (0.17) (0.17) (0.75) (0.82) (0.24) (0.79) -10.61 0.01 12.47 -0.21 0.07 0.10 -0.65 0.22 2.27 -0.56 -0.17 -0.33 (85.9) (0.00) (85.8) (0.17) (0.17) (0.82) (0.89) (0.26) (0.91) (0.26) (0.26) (0.22) -14.25 0.00 12.28 -0.26 -0.11 0.25 -0.05 0.25 2.31 (96.3) (0.00) (96.2) (0.17) (0.17) (0.75) (0.81) (0.26) (0.82) 0.53 (1.21) 0.99 (1.38) -0.67 -0.02 0.51 0.02 0.01 -0.50 -0.03 -0.17 2.74 (1.11) (0.00) (0.28) (0.00) (0.00) (0.28) (0.32) (0.10) (0.30) -0.37 -0.02 0.48 0.03 0.01 -0.55 -0.05 -0.18 2.73 0.02 -0.11 0.03 (1.18) (0.00) (0.28) (0.00) (0.00) (0.30) (0.32) (0.10) (0.30) (0.10) (0.10) (0.10) 1.12 -0.03 0.42 0.00 0.01 -0.45 -0.05 -0.22 2.79 (1.27) (0.00) (0.28) (0.00) (0.00) (0.30) (0.32) (0.10) (0.30) -1.19 (0.35) -0.54 (0.47) 2.69 -0.02 0.01 0.12 -0.08 -0.74 -0.36 -0.80 2.26 (0.91) (0.00) (0.24) (0.00) (0.00) (0.26) (0.26) (0.10) (0.26) 2.93 -0.02 -0.04 0.13 -0.07 -0.81 -0.33 -0.82 2.27 0.07 -0.20 0.06 (1.00) (0.00) (0.24) (0.00) (0.00) (0.28) (0.28) (0.10) (0.28) (0.00) (0.10) (0.00) 3.67 -0.02 -0.04 0.11 -0.06 -0.71 -0.37 -0.83 2.29 (1.07) (0.00) (0.24) (0.00) (0.00) (0.26) (0.26) (0.10) (0.26) -0.76 (0.33) -0.17 (0.44) 4.72 -0.01 0.02 -0.02 -0.01 -0.16 -0.08 -0.71 0.67 (0.75) (0.00) (0.20) (0.00) (0.00) (0.20) (0.20) (0.00) (0.24) 4.90 -0.01 -0.01 -0.02 -0.01 -0.21 -0.07 -0.73 0.68 0.03 -0.10 0.04 1078 1444 3266 (0.81) (0.00) (0.20) (0.00) (0.00) (0.20) (0.22) (0.00) (0.24) (0.00) (0.00) (0.00) 1078 1400 3303 28 4.84 -0.01 0.02 -0.02 -0.01 -0.15 -0.08 -0.73 0.68 (0.88) (0.00) (0.20) (0.00) (0.00) (0.20) (0.20) (0.00) (0.24) 0.01 (0.28) 0.10 (0.37) 1078 1423 3308 References Alesina, Alberto, Nouriel Roubini and Gerald D. Cohen. 1999. Political Cycles and the Macroeconomy. Cambridge, MA: MIT Press. Bartels, Larry M. 2008. Unequal Democracy: The Political Economy of the New Guilded Age. Princeton, NJ: Princeton University Press. Betz, Hans-Georg. 1990. “Politics of Resentment: Right-Wing Radicalism in West Germany.” Comparative Politics 23(1):45–60. Betz, Hans-Georg. 1993. “The New Politics of Resentment: Radical Right-Wing Populist Parties in Western Europe.” Comparative Politics 25(4):413–427. Cox, Gary W. 1997. Making Votes Count: Strategic Coordination in the World’s Electoral Systems. New York, NY: Cambridge University Press. Duverger, Maurice. 1954. Political Parties: Their Organization and Activity in the Modern State. London: Methuen. Federal Returning Officer. 2009. “Elections to the German Bundestag.”. URL: http://www.bundeswahlleiter.de/en/bundestagswahlen (Retrieved: December 2011) Frankland, E. Gene. 1995. Germany: The Rise, Fall, and Recovery of Die Grünen. In The Green Challenge: The Development of Green Parties in Europe, ed. Dick Richardson and Chris Rootes. New York, NY: Routledge chapter 2, pp. 23–44. Givens, Terri E. 2005. Voting Radical Right in Western Europe. New York, NY: Cambridge University Press. Golder, Matt. 2003. “Explaining Variation in the Success of Extreme Right Parties in Western Europe.” Comparative Political Studies 36(4):432–466. 29 Hibbs, Douglas A. 1977. “Political Parties and Macroeconomic Policy.” American Political Science Review 71(4):1467–1487. Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton, NJ: Princeton University Press. Jackman, Robert W. and Karin Volpert. 1996. “Conditions Favouring Parties of the Extreme Right in Western Europe.” Britisch Journal of Political Science 26(4):501–521. Kitschelt, Herbert. 1994. The Transformation of European Social Democracy. New York, NY: Cambridge University Press. Kitschelt, Herbert. 1995. The Radical Right in Western Europe: A Comparative Analysis. Ann Arbor, MI: The University of Michigan Press. Kitschelt, Herbert P. 1989. The Logics of Party Formation: Ecological Politics in Belgium and West Germany. Ithica, NY: Cornell University Press. Lijphart, Arend. 1994. Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945-1990. New York, NY: Oxford University Press. Lijphart, Arend. 1999. Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries. New Haven, CT: Yale Univeristy Press. Meguid, Bonnie M. 2008. Party Competition between Unequals: Strategies and Electoral Fortunes in Western Europe. New York, NY: Cambridge University Press. Norwegian Social Science Data Services. 2009. “ESS Round 4: European Social Survey Round 4 Data (Data file edition 2.0).”. URL: http://ess.nsd.uib.no/ess/round4 (Retrieved: Dec. 2011) R Development Core Team. 2011. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. URL: http://www.R-project.org 30 Rueda, David. 2005. “InsiderOutsider Politics in Industrialized Democracies: The Challenge to Social Democratic Parties.” American Political Science Review 99(1):61–74. Rueda, David. 2006. “Social Democracy and Active Labour-Market Policies: Insiders, Outsiders and the Politics of Employment Promotion.” Britisch Journal of Political Science 36(3):385–406. Swank, Duane and Hans-Georg Betz. 2003. “Globalization, the Welfare State and RightWing Populism in Western Europe.” Socio-Economic Review 1(2):215–245. Wickham, Hadley. 2009. ggplot2: Elegant Graphics for Data Analysis. New York, NY: Springer. URL: http://had.co.nz/ggplot2/book 31
© Copyright 2026 Paperzz