LSE - Small Logo
LSE - Small Logo

Blog Team

January 28th, 2014

Evidence from Germany suggests that parties are influenced by voters’ preferences in their attempts to form coalitions


Estimated reading time: 5 minutes

Blog Team

January 28th, 2014

Evidence from Germany suggests that parties are influenced by voters’ preferences in their attempts to form coalitions


Estimated reading time: 5 minutes

How important are the preferences of voters in coalition negotiations? Marc Debus and Jochen Müller assess the potential governments that could have been formed at the state level in Germany from 1990-2009, and relate these to the coalition preferences of voters. They find that there is a reciprocal relationship between parties and the electorate, with voters more likely to support coalitions based on pre-electoral alliances, and parties more likely to support coalitions favoured by their supporters.

In many European countries coalition governments are the rule rather than exceptions. Their regular presence is not only relevant for legislation and policy-making. Discussions and speculation regarding the partisan composition of the next government and the likelihood of potential coalitions become a standard element of election campaigns and influence voting behaviour. The 2013 election campaign for the German Bundestag, for instance, was characterised by extensive discussions about possible, likely and preferable outcomes of the government formation game. Most voters preferred a government between the two large parties, Christian Democrats (CDU/CSU) and Social Democrats (SPD), which ultimately formed a coalition in December 2013.

German Vice-Chancellor Sigmar Gabriel and Angela Merkel, Credit: Martin Rulsch (CC-BY-SA-3.0)

Against this backdrop, it is surprising that plenty of advanced theoretical models on government formation focus only on size- and policy-driven factors that should influence the partisan composition of a government. While there is still research under way that emphasises the role of institutional and contextual factors on the outcome of the government formation process, coalition theories and their empirical evaluation still exclude the potential effect of the coalition preferences of voters on the outcome of the government formation process.

We have aimed to incorporate ‘vote seeking’ approaches and patterns of electoral competition into standard models of government formation. In doing so, we tested whether parties form coalitions according to the preferences of voters even when controlling for the ‘usual suspects’ to explain government formation, such as variables from the office and policy-centred approaches in coalition theory, and institutional constraints deriving from the political system of a country.

We selected the German state level and the processes of government formation there to test our expectations. The major reason for selecting the German states is not only the institutional constraints on government formation that arise from the German multi-level system on the one side, and the bicameral structure of the German political system on the other side. In contrast to the application of a cross-national research design, there is not much variation in the institutional structure between the German states, and hence we could isolate the effects of all independent variables without the interference of institutional factors.

We built our analysis on a dataset that covers the characteristics of all potential governments that could have been formed at the state level in Germany from January 1990 onwards. This dataset includes information not only on office-based features of each possible coalition, but also on the parties’ positions on the relevant policy dimensions and their pre-electoral commitments. Furthermore, it provides information on the question whether a potential coalition is in line with the patterns of party competition at the federal level and on the date an election was held.

In order to analyse the impact of the electorate’s coalition preferences on government formation, we had to identify a source that provides information on voters’ opinions on possible coalitions. Fortunately, a large number of surveys exist that were conducted in the run-up to 77 of the 79 state elections in the period under study, almost all of which include questions that we require to construct our variables. Using these surveys, we created a dataset that covers information on the coalition preferences of voters and party identifiers.

Table 1 below presents the potential coalitions that are favoured by voters in the time period under consideration. As shown, five different coalitions are preferred by the voters in at least one case. A coalition between Christian Democrats (CSU/CSU) and Free Democrats (FDP) is preferred in 24 of 77 government formation situations. In 23 out of 77 cases, the voters prefer coalitions between the SPD and the Greens or the Christian Democrats and the SPD, respectively. Except for the ‘grand coalition’ between CDU/CSU and SPD, these coalitions include only parties of the same ‘political camp’ and can thus be regarded as the natural choice of the parties involved in the coalition bargaining process. In the remaining cases, the electorate preferred a social-liberal coalition – in the state of Rhineland-Palatinate after it had been in charge for ten years – or a coalition between the Social Democrats and the socialist PDS, which renamed itself as Die Linke (The Left) in 2007.

Table 1: Coalition preferences of the electorate in German states (1990-2009)

Note: Authors’ calculations based on German state election surveys from 1990-2009

The description of voters’ coalition preferences leads to questions concerning their origins and implications. We argue that parties should take the voters’ attitudes toward possible coalitions into account when it comes to government formation. At the same time, it is the voters who use parties’ pre-electoral statements to estimate ‘realistic’ outcomes of the government formation process.

Table 2 presents some information on the overlap between the coalition preferences of voters and parties, i.e. pre-electoral alliances. Even though the depicted data does not allow for an analysis of causal mechanisms, it seems to be the case that the support of voters for specific coalitions is heavily influenced by the existence of pre-electoral alliances.

Voters who identify themselves with Christian Democrats or Liberals, for instance, do in general prefer a coalition between both parties. If there is a pre-electoral alliance between the CDU and FDP, the support for such a coalition among party identifiers clearly increases. While almost 49 per cent of CDU/CSU voters and 56 per cent of FDP voters preferred a coalition of these parties in state elections where they did not both state an intention to form a coalition government after the election; the share of supporters for a Christian-liberal coalition among the voters of the respective parties increased to 72 per cent (CDU/CSU) and 73 per cent (FDP) if a pre-electoral alliance between both parties existed.

A similar picture arises when looking at the support for a coalition between the SPD and FDP among their voters. Only 13.7 per cent of SPD voters and 17.2 per cent of respondents intending to vote for the Liberals preferred a coalition between both parties in situations where the Social Democrats and the FDP did not agree on a pre-electoral alliance. When the parties did send signals during state election campaigns that they were willing to form a coalition government, 48 per cent and 53 per cent of SPD and FDP voters wanted this coalition to emerge from the election. These findings from a descriptive analysis of the impact of parties’ coalition signals on the coalition preferences of voters show that it is necessary to control for the impact of pre-electoral pacts on the coalition preferences of voters in multivariate regression models.

Table 2: Pre-electoral alliances and coalition preferences by vote intention (percentage)

Note: Authors’ calculations based on German state election surveys from 1990-2009. Entries for parties belonging to a pre-electoral alliance are marked in bold.

Do the coalition preferences of voters have an effect on the outcome of the government formation game? Our results suggest that potential coalitions are more likely to form if they are favoured by a plurality of the electorate. This indicates that political parties take the coalition preferences of the electorate into account. If a coalition is preferred by a plurality among voters, then it is more likely that this combination of parties will form the next cabinet in a German state – even when controlling for potential endogeneity of voter coalition preferences and other factors that are usually considered to be decisive for coalition formation.

In addition, including information on voters’ coalition preferences increases the goodness of fit of the second regression model. In some government formation processes, it seems that the inclusion of voters’ coalition preferences helped to form or to renew a coalition government.

For a longer discussion of this topic see the authors’ papers ‘Do Voters’ Coalition Preferences Affect Government Formation? and ‘Expected utility or learned familiarity? The formation of voters’ coalition preferences’.

Please read our comments policy before commenting.

Note: This article gives the views of the author, and not the position of EUROPP – European Politics and Policy, nor of the London School of Economics.

Shortened URL for this post:


About the authors

Marc Debus University of Mannheim
Marc Debus is Professor of Comparative Government at the University of Mannheim, Germany. His research interests include political institutions and their effects on the political process, party competition and coalition politics, and political decision-making in multi-level systems.

Jochen Müller – University of Essex
Jochen Müller is Lecturer in the Department of Government at the University of Essex.

About the author

Blog Team

Posted In: Governance and policy evaluation | Jochen Müller | Marc Debus

Leave a Reply