An institutional safety net? How electoral institutions mediate the fortunes of parties under threat

ABSTRACT

How do electoral institutions condition the electoral fortunes of parties under threat? In this article, we examine how Germany’s mixed-member proportional (MMP) system has influenced the vote share of the Sozialdemokratische Partei Deutschlands (SPD), a party under threat which has consistently lost votes over the past two decades. Using the 2013 and 2017 waves of the German Longitudinal Election Study (GLES), we find that voters are likely to engage in “sincere” vote switching on both the district and party ballots and that they are less likely to engage in “strategic” vote switching when they cast their district votes than when they cast their party votes, which protects SPD district candidates. Moreover, voters who stay with the SPD when casting their district vote are also less likely to switch their party votes, which protects the SPD’s overall vote share. We thus conclude that Germany’s MMP electoral system serves as an institutional safety net for the SPD as a party under threat. Our findings have important implications for understanding the ways in which electoral rules shape voting behavior and how different rules can mediate the decline of parties under threat.

Acknowledgement

We thank Herbert Kitschelt, Ruth Dassonneville, and the three JEPOP reviewers for their helpful comments and suggestions. All errors remain our own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Such “contamination” effects in mixed electoral systems have been considered in a number of studies (see Cox and Schoppa 2002; Ferrara, Herron, and Nishikawa 2005; Hainmueller and Kern 2008; Herron and Nishikawa 2001; Karp 2009; Rheault et al. 2020).

2 https://www.politico.eu/europe-poll-of-polls/france/

3 For details see, Klingemann and Wessels (2003).

4 PR does not preclude all strategic voting, of course, since some threshold necessarily exists that a party has to clear to gain representation in parliament (see e.g. Gschwend 2007a; Bargsted and Kedar 2009).

5 Another source of vote choice in SMD and mixed electoral system may be a preference for a particular constituency candidate, i.e., a personal vote (Pappi and Thurner 2002); this does not play a central role in shaping vote choice in Germany, however (Moser and Scheiner 2005).

6 The top polling agencies in Germany do not conduct district level polls and even “media about state-level polls is rare in Germany” (Meffert and Gschwend 2011, 11). And while some new tools have recently been made available that gauge expected electoral success at the district level (see zweitstimme.org), it is questionable whether large numbers of voters use these, or other polls (Blais et al. 2018), to make voting decisions.

7 Among the reasons why most voters do not engage in split-ticket voting is that it is cognitively easier to make one rather than two independent voting decisions (Gschwend 2004). But split-ticket voting has increased over time (Jesse 1988); according to Gschwend and Pappi (2004), this is because it is facilitated by the clarity of Germany's ideological blocs and the coalition alternatives (although this is less clear today than it was at their time of writing).

8 We use the cross-sectional waves of the GLES and not the panel data, relying instead on the self-reported vote choice of respondents in the current and prior election. All question wording is available in Table A2 in the Appendix.

9 Tables A1a and A1b provide information on which parties the SPD voters switched from and switched to. Examining these patterns will be important for future work.

10 Importantly, we are interested in the patterns of vote switching from one election to the next on the same ballot and not split ticket voting where voters support two different parties in the same election. Unpacking this latter behavior is an important avenue for future work.

11 While we use the left-right position as a proxy for a party and voter's overall ideological position, it has the potential to lead to less precise measures of ideological position as it can conflate two or more separate political dimensions (see Caughey, O’Grady, and Warshaw 2019).

12 We rely on the party’s position from the Manifesto Project for both the party ballot and the district ballot. While voters are voting for individual candidates on the district ballot who have their own personal positions, some research suggests that MPs elected under the district ballots are not necessarily more likely to defect from the party (Becher and Sieberer 2008), and if they do defect at higher rates, it is not driven by local mandates (Sieberer 2010).

13 Importantly, while partisanship is also a useful proxy for holding a leftist social identity, partisanship does not capture the ties to a broader leftist social identity but instead captures specific ties to the SPD.

14 Male = 1; Education is recoded to a 5-category variable.

15 We also run models in which we replace partisan attachment to the SPD with whether a voter split her ballot and voted for the SPD on one ballot and not on the other in the previous election. This captures the voter’s loyalty to the SPD as it relates to her voting behavior. We find that this control provides substantively similar results to our PID control, with those who are weak SPD voters, or those who split their ticket, more likely to switch their vote in subsequent elections. The variables are highly correlated and therefore cannot both be included in our models.

16 We explored other measures of SPD strength, including the party’s vote share in the last federal election in the given state and a variable which measured SPD state and federal election results over time in a composite SPD stronghold variable. None of these variables had a statistically significant influence on vote switching on either ballot.

17 This result merits further exploration. Union strength arguably should also serve as a proxy for left-wing identity, especially for a union member, and decrease the probability of switching. This, however, is not what we find. This result may be due to the way the variable is operationalized—the percentage of respondents who are union members in a given state. Respondents may not be representative of union members in the population. Unfortunately, we have not been able to find actual union density data at the state level. As a proxy, we have a created a measure which approximates union density at the state level using 2014 state-level data from the Cologne Institute for Economic Research (Lesch 2016) multiplied by the percent change from the national level numbers from 2013 to 2014 and 2014–2017 using OECD (2020) data. When we use this variable, union strength is not significant in either model. The correlation between this variable and the data from the GLES is 0.594. Thus, our results should be understood in the context of these caveats.

18 In 2021, the CDU’s poor performance was largely the result of 1.5 million CDU/CSU voters switching to the SPD and not due to the party losing single-member districts to right-wing competitors (https://www.fes.de/en/german-election). Furthermore, there may be a socialization process at work whereby union members, who have a commitment to leftist values, are socialized into voting for the SPD. This same socialization process may not be at work between religious organizations and voting for the CDU/CSU. These mechanisms should be explored in future research.

19 While Ferrara, Herron, and Nishikawa (2005) maintain that SMD voting affects the PR component in mixed systems, Rheault et al. (2020) argue that the relationship is reversed.

20 Rheault et al. (2020) maintain that there should be stronger contamination from the party vote to the district vote, because the former determines a party’s number of legislative seats and is therefore more consequential. This, however, does not take into account that the potential negative consequences of vote switching at the district level – the selection of a political adversary – are more visible and tangible for voters than the diffuse benefit of contributing to the party’s overall vote share.