No Elections for Big Parties Elias Dinas 1 Pedro Riera 2 1 University of Nottingham elias.dinas@nottingham.ac.uk 2 University of Strathclyde pedro.riera@strath.ac.uk EUDO Dissemination Conference Florence, November 29, 2013 EU Incumbency e ects EUDO: Campaign E ects Panel 1
Question Francesca s case Italy 1984... Francesca is a PhD student from the North of Italy that was born on the fall of 1984. Being only 16, she was not able to vote in the 2001 Italian general election and had to wait until the 2004 Eureopan elections to become eligible to vote. In that election she voted for an essentially small party such as I Radicali (the so-called, Emma Bonino List). Two years later, in the first general election in which she could vote, she did it for another small party, that is, I Verdi or Green Party. Francesca has not voted in any single election since then. EU Incumbency e ects EUDO: Campaign E ects Panel 2
Question EU vs National Elections: # of Parties 12 10 8 6 4 2 0 Effective Number of Electoral Parties: National vs. European Elections Austria Belgium Bulgaria Cyprus Czech Denmark Estonia Finland France Germany Greece Hungary Ireland Italy Latvia Lithuania Luxembour Malta Netherlands Poland Portugal Romania Slovakia Slovenia Spain Sweden UK National European 2009 EU Election 2 3 4 5 6 7 8 9 10 Number of Parties 2 3 4 5 6 7 8 9 10 National Elections EU Incumbency e ects EUDO: Campaign E ects Panel 3
Question EU vs National Elections: Incumbent s % 60 40 20 0 Vote Shares of PM's party: National vs. European Elections Austria Belgium Bulgaria Cyprus Czech R. Denmark Estonia Finland France Germany Greece Hungary Ireland Italy Latvia Lithuania Luxembour Malta Netherland Poland Portugal Romania Slovakia Slovenia Spain Sweden UK 2009 EU Election 10 20 30 40 50 Vote share of PM's party National European 10 20 30 40 50 National Elections EU Incumbency e ects EUDO: Campaign E ects Panel 4
Question EU vs National Elections Government Parties % 80 60 40 20 0 Vote Shares Government Parties: National vs. European Elections Austria Belgium Bulgaria Cyprus Czech R. Denmark Estonia Finland France Germany Greece Hungary Ireland Italy Latvia Lithuania Luxembour Malta Netherland Poland Portugal Romania Slovakia Slovenia Spain Sweden UK National European 2009 EU Election 10 20 30 40 50 60 70 Vote Share of Government Parties 10 20 30 40 50 60 70 National Elections EU Incumbency e ects EUDO: Campaign E ects Panel 5
Electoral Volatility Question Pedersen's Index Electoral Volatility Pedersen's Index Electoral Volatility Pedersen's Index Electoral Volatility 5 10 15 20 5 10 15 20 5 10 15 20 Belgium Germany Luxembourg Pedersen's Index Electoral Volatility 0 20 Pedersen's Index Electoral Volatility 0 5 10 15 Pedersen's Index Electoral Volatility 5 10 15 20 25 30 Denmark Ireland Netherlands Pedersen's Index Electoral Volatility 5 10 15 20 25 Pedersen's Index Electoral Volatility 0 10 20 30 40 Pedersen's Index Electoral Volatility 0 5 10 15 France Italy United Kingdom Source: Authors own calculations EU Incumbency e ects EUDO: Campaign E ects Panel 6
Number of Parties Question Belgium Effective Number Electoral Parties 2 4 6 8 10 Effective Number Electoral Parties 4 5 6 7 Effective Number Electoral Parties 4 5 6 7 Germany Denmark Effective Number Electoral Parties 3 3.5 4 4.5 5 5.5 Effective Number Electoral Parties 2.5 3 3.5 4 Effective Number Electoral Parties 3 4 5 6 7 8 Luxembourg Ireland Effective Number Electoral Parties 3 3.5 4 4.5 5 Effective Number Electoral Parties 4 5 6 7 Effective Number Electoral Parties 2 2.5 3 3.5 4 Netherlands France Italy United Kingdom Source: Authors own calculations EU Incumbency e ects EUDO: Campaign E ects Panel 7
Question Negative Incumbency Advantage Chang&and&Golden&2004&& Source: Strom 1990; Narud & Valen 2008 EU Incumbency e ects EUDO: Campaign E ects Panel 8
Theory Elections for the EU Parliament: Direct E ects The Dominant View As second order elections, EU elections reflect the dynamics of the national electoral cycle (Weber 2011), serving to either to punish current governing parties or to accommodate their policy demands through electoral balancing (Erikson and Filippov 2001; Rodden and Wibbels 2010). The Second Order Thesis Lower turnout Protest voting Small parties Anti-incumbent Anti-system behaviour tends to wear o as national election campaigns bring structure into the political system. No immediate spillover e ects from EU to national elections. EU Incumbency e ects EUDO: Campaign E ects Panel 9
Theory Elections for the EU Parliament: Socialization E ects Our Argument Early voting experiences leave an imprint on voters partisan outlooks. EU elections characterized by di erent voting patterns. Di usion of voting patterns from EU to national elections. Political socialization and political persistence: sticking with your vote (Mullainathan and Washington 2009; Meredith 2009) The mirror image: De-mobilisation e ects of EU elections on national electoral turnout (Franklin and Hobolt 2012). From turnout to vote choice: Learning to vote for small parties Socialised against the Incumbent. EU Incumbency e ects EUDO: Campaign E ects Panel 10
Observational Implications Theory Hypotheses Becoming of age during an EU election will increase the likelihood of voting small parties in future national elections Becoming of age during an EU election will decrease the likelihood of voting for the incumbent party or parties during the period of the first eligible election. EU Incumbency e ects EUDO: Campaign E ects Panel 11
Identification Strategy Research Design The Problem People choose if they want to vote and in which elections to vote; Impossible to disentangle these early voting experiences from the selection process that resulted in voting for the first time in national or EU elections. The Design Use eligibility to vote for the first time. Distinguish between two groups: those becoming of age before an EU election and those becoming of age before a national election. Treatment is not turnout but type of first-eligible election. Assuming year of birth not age is randomly assigned; E.g. A German citizen being born in 1985 becomes of age in 2003, thus encountering first an EU election. A German citizen born either in 1983 or in 1986 becomes first eligible to vote in a Federal election. EU Incumbency e ects EUDO: Campaign E ects Panel 12
Research Design Measuring Treatment E ects H1: EU! Small Party Support E[Small Party Eligible EU ]=b 0 + b 1 Eligible EU Small Party: Vote in 2009 for any of the non-government parties. Eligible EU :1 If first eligible to vote in an EU election; 0 otherwise H2: EU! Vote against the socialization Incumbent E[Anti Incumbent Eligible EU,Incumbent when Eligible]= b 0 + b 1 Eligible EU + b 2 Party Eligible + b 3 Party Eligible Eligible EU Anti-Incumbent: 1 Voting in 2009 against the national incumbent when first eligible to vote; 0 otherwise Eligible EU : 1: If first eligible to vote in an EU election; 0 otherwise We exclude all respondents becoming of age after 2004, because the treatment might di er for them (Vote vs Not vote rather than type of vote) EU Incumbency e ects EUDO: Campaign E ects Panel 13
Research Design Measuring Treatment E ects H1: EU! Small Party Support E[Small Party Eligible EU ]=b 0 + b 1 Eligible EU Small Party: Vote in 2009 for any of the non-government parties. Eligible EU :1 If first eligible to vote in an EU election; 0 otherwise H2: EU! Vote against the socialization Incumbent E[Anti Incumbent Eligible EU,Incumbent when Eligible]= b 0 + b 1 Eligible EU + b 2 Party Eligible + b 3 Party Eligible Eligible EU Anti-Incumbent: 1 Voting in 2009 against the national incumbent when first eligible to vote; 0 otherwise Eligible EU : 1: If first eligible to vote in an EU election; 0 otherwise We exclude all respondents becoming of age after 2004, because the treatment might di er for them (Vote vs Not vote rather than type of vote) EU Incumbency e ects EUDO: Campaign E ects Panel 13
Research Design Measuring Treatment E ects H1: EU! Small Party Support E[Small Party Eligible EU ]=b 0 + b 1 Eligible EU Small Party: Vote in 2009 for any of the non-government parties. Eligible EU :1 If first eligible to vote in an EU election; 0 otherwise H2: EU! Vote against the socialization Incumbent E[Anti Incumbent Eligible EU,Incumbent when Eligible]= b 0 + b 1 Eligible EU + b 2 Party Eligible + b 3 Party Eligible Eligible EU Anti-Incumbent: 1 Voting in 2009 against the national incumbent when first eligible to vote; 0 otherwise Eligible EU : 1: If first eligible to vote in an EU election; 0 otherwise We exclude all respondents becoming of age after 2004, because the treatment might di er for them (Vote vs Not vote rather than type of vote) EU Incumbency e ects EUDO: Campaign E ects Panel 13
AGraphicalIllustration Research Design Dendity 0.02.04.06.08.1 EU eligible National eligible 25 30 35 40 45 50 Age EU Incumbency e ects EUDO: Campaign E ects Panel 14
AGraphicalIllustration Research Design Dendity 0.02.04.06.08.1 EU eligible National eligible 25 30 35 40 45 50 Age EU Incumbency e ects EUDO: Campaign E ects Panel 15
Research Design Balance in the Distributions Quantile-Quantile Plot Age [First election EU] 25 30 35 40 45 50 25 30 35 40 45 50 Age [First election national] EU Incumbency e ects EUDO: Campaign E ects Panel 16
Research Design Measuring Treatment E ects H1: EU! Small Party Support E[Small Party Eligible EU ]=b 0 + b 1 Eligible EU Small Party: Vote in 2009 for any of the non-government parties. Eligible EU :1 If first eligible to vote in an EU election; 0 otherwise H2: EU! Vote against the socialization Incumbent E[Anti Incumbent Eligible EU,Incumbent when Eligible]= b 0 + b 1 Eligible EU + b 2 Party Eligible + b 3 Party Eligible Eligible EU Anti-Incumbent: 1 Voting in 2009 against the national incumbent when first eligible to vote; 0 otherwise Eligible EU : 1: If first eligible to vote in an EU election; 0 otherwise We exclude all respondents becoming of age after 2004, because the treatment might di er for them (Vote vs Not vote rather than type of election) EU Incumbency e ects EUDO: Campaign E ects Panel 17
Findings Results: Small Parties H1 Becoming of age during an EU election increases the probabity of voting a small party in the national election preceding the 2009 election Pr(Small =1 Eligible EU ) LPM Logit-Model Only EU 10 Eligible EU.024 (.016).106 (.069).023 (.014) Country fixed-e ects Yes Yes Yes n 3471 3471 1699 EU Incumbency e ects EUDO: Campaign E ects Panel 18
Findings Result: Anti-Incumbent Vote H2 Becoming of age during an EU election decreases the probabity of voting for the incumbent party during the period of the first eligible election. Pr(Incumbent =1 Eligible EU,Incumbent Eligible ) OLS Coe s b 1 + b 3 Eligible EU.028 (.027) Incumbent Eligible.058 (.020) -.085 (.038) Eligible EU Incumbent Eligible -.113 (.046) Country fixed-e ects Yes Yes n 2529 2529 EU Incumbency e ects EUDO: Campaign E ects Panel 19
APlaceboTest Findings The Logic Treatment is an EU election that never took place in the country in question because it was not yet an EU member. Pr(Incumbent =1 Eligible EU,Incumbent Eligible ) Small Party Anti-Incumbent Eligible EU.010 (.032) -.245 (.113) Incumbent Eligible.058 (.020) -.062 (.203) Eligible EU Incumbent Eligible -.113 (.046) -.148 (.289) Country fixed-e ects Yes Yes n 2529 2529 EU Incumbency e ects EUDO: Campaign E ects Panel 20
Implications Extensions: Heterogeneity Unpacking the Mechanism Not a path into anti-system socialization Rather, a voter learning process: creating cohorts of small party support. Anti-incumbent e ects conditioning on EU attitudes: 2009 b 1 + b 3 EU apple 5 EU 5 Average Treatment E ect -.037 (.068) -.095 (.039) Country fixed-e ects Yes Yes n 741 1230 EU Incumbency e ects EUDO: Campaign E ects Panel 21
Implications Extensions: Heterogeneity Unpacking the Mechanism Not a path into anti-system (anti-eu) socialization Rather, a voter learning process: creating cohorts of small party support. Anti-incumbent e ects conditioning on EU attitudes: 2004 b 1 + b 3 EU apple 5 EU 5 Average Treatment E ect -.041 (.090) -.113 (.074) Country fixed-e ects Yes Yes n 664 1265 EU Incumbency e ects EUDO: Campaign E ects Panel 22
Implications & Extensions Implications Concluding Remarks The long shadow of the voting booth: early voting experiences leave an imprint on future behaviour. New elections might then have spillover e ects A similar argument might apply for: The national e ects of subnational elections The anti-incumbent spillover e ects of referendums (e.g. the EU constitution, the Italian 4Si referendum. An argument that holds in normal times is probably stronger in di cult times. Imagine a 2012 EU election... EU Incumbency e ects EUDO: Campaign E ects Panel 23
Implications 2014: No election for big parties EU Incumbency e ects EUDO: Campaign E ects Panel 24