Navigating complexity: exploring the changing dynamics of coalition avoidance in Germany, 1946–2023

ABSTRACTState elections in Germany’s multi-level polity have important implications for politics at both the regional and the federal level: state governments’ composition makes a difference for both policy outputs and the federal government’s political leeway. We provide novel insights into long-term trends in state elections, including aspects of voting behavior, party systems, and government formations, covering elections between 1946 and 2023. We argue that the commonly held belief that state parties try to avoid coalitions bridging the government-opposition divide at the federal level is dependent on the level of legislative party system fragmentation. Furthermore, we argue that fragmentation influences the formation likelihood of ‘grand coalitions’ between CDU/CSU and SPD. Using conditional logit models, we find empirical support for our theoretical arguments. We contribute to a deeper understanding of the complexities of Germany’s multi-level polity and emphasize the need to consider evolving voting behavior and party competition patterns in coalition analyses.KEYWORDS: Regional electionsgovernment formationGermanycongruence Disclosure StatementNo potential conflict of interest was reported by the author(s).Notes1 Scholars used the German case to study party systems and their changes (e.g. Wagner Citation2023), voting behavior (e.g. Müller Citation2018), parties’ policy positions and issue emphasis (e.g. Bräuninger et al. Citation2020; Gross, Krauss, and Praprotnik Citation2023), government formation and coalition politics (e.g. Debus Citation2008; Gross and Krauss Citation2021), and the policy-effects of different political constellations (e.g. Jeffery et al. Citation2014).2 Federal- and state-level electoral institutions are quite similar across states regarding their mixed electoral systems, electoral thresholds, and electoral terms. Since these aspects are covered extensively by Müller (Citation2018) and because none of the respective factors have been changed in the last years, we do not explain them in great detail here.3 The smoothing line represents a curve generated using a local regression technique known as loess, implemented in ggplot’s geom_smooth-function in R.4 The NPD narrowly did not make it into the Bundestag in 1969, was voted out of the state parliaments and it only had a short-lived revival 35 years later.5 Besides the SED, the ‘block parties’ consisted of the East German CDU, the Liberal Democratic Party (LDPD), the National Democratic Party of Germany (NDPD) and the Democratic Farmers’ Party of Germany (DBD).6 Except from the CSU the PDS has been regarded as the most successful regionalist party between 1990 and 2005 (Hepburn and Hough Citation2012).7 This period is also marked by the lowest electoral volatility scores (see Figure 1).8 Yet, a pure policy-seeking perspective finds little surprise in this government as all three parties have been ideologically closer to each other compared to all other (majority) coalition options (Bräuninger et al. Citation2020).9 Prime Minister Reiner Haseloff (CDU) has been afraid that some of the CDU members of parliament (MPs) would rather vote with AfD MPs and not with the coalition partner.10 If we refer to the raw number of parties instead of the effective number of parties, we find a similar effect while the other associations essentially remain the same in the empirical models.11 Not all parties represented in the state parliaments are included in the original data. We thus use other parties’ positions as approximate values: the position of the CDU is used to approximate the position of the CSU, the CVP, the Free Voters and the Statt Party; the position of the SPD is used as a proxy for the Arbeit für Bremen und Bremerhaven. A similar approach was adopted for Alliance '90 (positions of the Greens), the BDV (positions of the FDP) and the DDU (positions of the KPD). For several right-wing parties (Bürger in Wut, DVU, NPD, the Republicans, the Schill party, and the Socialist Reich Party) we use the position that was determined for the AfD.12 We operationalized ‘pariah party’ as taking on the value of ‘1’ if a potential coalition includes the AfD, the Bürger in Wut, the DVU, the NPD, the Republicans, the Schill party or the Socialist Reich Party.13 Since the first federal government has been formed in 1949, the variable cross-cutting governments receives the value 0 for pre-1949 state governments. Dropping pre-1949 state governments does not change the substantial results of model 1.14 Note that polarization mainly influences the duration of government formations, both at the national and at the sub-national level (Bäck, Debus, and Imre Citation2022; Ecker and Meyer Citation2020; Golder Citation2010). Analysing whether the degree of polarization in the (parliamentary) party systems on the state level has a mediating effect on the factors that influence patterns of government formation in the German states requires a different theoretical set-up and is therefore a task for future research (see, e.g., Indridason Citation2011).Additional informationFundingThis work was supported by LMUexcellent Postdoc Support Fund.