Progressive Futures

Policy conflict and coalition formation in the upcoming Bundestag elections

19 September 2017

Angela Merkel will most likely secure her fourth term in office as German Chancellor following the Bundestag elections on 24 September. Although it may appear like a groundhog day experience, Germany’s political landscape has undergone a significant transformation since Merkel took office 12 years ago.

The increased fragmentation of Germany’s party system, paired with the advent of the right-wing populist Alliance for Germany (AfD), will make the formation of the future federal government a daunting task. A majority for the Christian Democrats (CDU/CSU) and Liberal Democrats (FDP) seems unlikely at this stage – the latest predictions by the forecasting team from zweitstimme.org suggest that this will happen with a probability of just a little more than 30%. At the same time, the SPD currently appears unwilling to join yet another grand coalition as junior partner. So, at the end of the day, an alliance of CDU/CSU, FDP and Greens – nick-named a ‘Jamaica coalition’ because of the three parties’ colours – turns out to be the most likely scenario.

To assess the ideological heterogeneity of different coalitions, we used data from the German voting advice application Wahl-O-Mat. The website allows users to answer 38 policy questions (with yes, no or neutral) and then provides a measure of personal spatial proximity to each party. In a similar vein, we take the issue-specific positions of parties to map out on which questions potential coalitions agree or differ.

Unsurprisingly, the traditional ‘bloc coalitions’ show the highest degree of ideological similarity. CDU/CSU and FDP agree on 20 out of 38 policy questions, whereas SPD and Greens agree on 24 issues. In light of the most recent polls, it is, however, unlikely that the CDU/CSU and FDP will gain a majority of seats in the next Bundestag, and for the SPD and Greens is it next to impossible.

Coalition-formation after the general election will therefore likely be an intricate matter. A renewed grand coalition is the only two-party alliance which almost certainly will gain a majority of seats in the Bundestag. Although not preferred by either partner, such an alliance agrees on 20 out of 38 issues after all.

Any of the three-party coalitions would face significantly more conflicts. The ‘Jamaica coalition’ only agrees on 9 topics, with explicit disagreement in 21 questions. Regarding environmental policies, for example, there is virtually no agreement between CDU/CSU, FDP and Greens. Life would not be much easier for a so-called ‘traffic-light-coalition’ comprising the SPD, FDP and Greens. These parties agree on 11 topics, yet differ on 20. While an alliance of the SPD, Greens and the socialist Left party would agree on almost double the number of issues, reaching a majority of seats is unlikely (zweitstimme.org suggests a probability of only 6% for this outcome).

Figure 1: Agreement and disagreement in different coalitions

Source: Own analysis of party positions in the Wahl-O-Mat for the general election 2017, published by Bundeszentrale für politische Bildung

The dualism of government versus opposition is not without alternative

We can of course only speculate about the exact composition of the future federal government in Berlin. Yet, we can predict with almost certainty that it is going to be a majority coalition. As many Britons think that only single-party majority governments are a viable solution, Germans are obsessed with majority coalitions. In a typical majority coalition, all parties in government commit to a unified position on virtually all political issues – in the best case facilitating joint government action, but in the worst case sticking to a sub-optimal status quo. Potential issue-specific majorities crossing the government-opposition divide are ignored. However, increasing fragmentation of Germany’s party system has made it much harder to form majority coalitions in the first place, and a majority setup frequently limits the scope of political action. Novel government constellations, as experienced in some German states (e.g. Greens and CDU in Baden-Württemberg, or CDU, SPD and Greens in Saxony-Anhalt), illustrate the problems when parties need to bridge significant disagreement over policy questions. And, at the same time, at least on some topics opposition parties would be available to strike a better deal. Yet, German parties cling to the idea of majority coalitions. While there is a lively discussion of policy issues, a debate regarding how parties could cooperate more effectively in a fragmented political environment is non-existent. In the recent decade, this preoccupation with majority coalitions mostly worked to the disadvantage of Social Democrats. Being relatively weak in terms of vote share, the SPD was caught as junior partner in a grand (majority) coalition. Being located close to the centre of the policy space, Social Democrats could have struck better deals if they had been willing to cooperate, on an issue-by-issue basis, with different partners to the left (Left Party) and right (FDP, CDU).

Other countries have long proven that such flexible forms of majority formation can be a desirable substitute for majority coalitions in the context of a fragmented party system. Danish minority governments, for example, build shifting majorities on different issues. Minority governments in New Zealand form confidence & supply agreements with different parties as stable framework for cooperation across the aisle of government and opposition. There are convincing arguments that such arrangements could work for the benefit of democracy also in Germany, Britain and beyond. Our analysis reveals, for example, that a flexible approach to majority formation would also maximize the scope for political action in a newly elected Bundestag. On the basis of current seat share estimates, a legislative majority would be attainable for 21 topics, even if the right-wing populist AfD is excluded from the equation. If the established parties opted for an issue specific-cooperation with the AfD, majorities are available for as many as 29 issues. This perspective also strikingly reveals how the SPD’s lost her strategic advantage in a Bundestag with hypothetical issue-specific majorities over the years. In 2005, when the SPD first entered a grand coalition with Angela Merkel’s CDU/CSU but could also have formed a majority government with Left Party and Greens it was part of 21 issue specific majorities (CDU/CSU: 11). In the newly elected Bundestag (based on seat share forecasts) it could even join 23 majorities, yet CDU/CSU has clearly taken the lead with 26 potential issue specific majorities.

Figure 2: Predicted share of conflicts (based on left-right positions)

Simulation of decision-making based on the left-right-dimension

The analysis presented so far treated all issues as equally important: disagreement over a peripheral issue entered the analysis in the same way as a question that lies at the heart of party competition. In the first case, however, parties should find it much easier to reach a compromise.

To take into account variation in political relevance of issues, we also make use of a more complex statistical model (a Bayesian implementation of the Nominal Response Model suggested by Bock, 1972). Based on all the observed issue-specific positions, this tool allows us to derive not only the more fundamental left-right positions of parties, but also how strongly each issue contributes to the main axis of political competition. Building on these inferences, we can then simulate the level of conflict for different coalitions. As an additional advantage, this procedure also quantifies the uncertainty in our conclusions. The graph shows the share of issues with explicit disagreement for the coalitions of main interest. In addition to the mean (shown as a point), the line illustrates the interval that covers nine out of ten simulated values. For comparison, the cross represents the result of simple counting as discussed above.

Grand coalition more, three-party coalitions less complicated

Looking at the two-party coalitions, we observe the biggest difference in results between the two methods for the Grand coalition. CDU and SPD need to resolve somewhat more issues than previously thought. For the other two-party coalitions, the changes are minor. Thus, the level of conflict for a Grand coalition falls somewhere between the classic “bloc coalitions” and an alliance of CDU and Greens.

There is also some good news for the negotiators and all those who are worried about the stability of the future government. Both the “Jamaica” and the “Traffic-light coalition” face conflict in less than 50% of the issues. Unsurprisingly, the three-party coalitions on the left and the right of the spectrum are more homogenous. It is rather unexpected, though, that there is a significant drop in disagreement within an alliance of CDU, FDP und AfD, when judged on the basis of the simulation. The mean share of disputed issues amounts to less than 30%, about the same as for a red-red-green alliance. The comparatively low level of issue-specific conflict should, however, not be equated with a high probability that such a coalition will actually be formed. As we know, government formation is also influenced by other factors, such as common values, trust and parties’ strategic considerations.


Däubler, Thomas und Debus, Marc (2009). Government formation and policy formulation in the German states. Regional & Federal Studies 19 (1):73-95.

Stecker, Christian und Ganghof, Steffen (im Erscheinen). Die Institutionalisierung wechselnder Mehrheiten: Minderheitsregierungen im internationalen Vergleich. In: A. Blätte & A. Steinfort (eds.), Regieren ohne eigene Mehrheit – Minderheitsregierungen in der Analyse. Wiesbaden: VS Verlag für Sozialwissenschaften.

Ganghof, Steffen, Stecker, Christian, Eppner, Sebastian und Heeß, Katja (2012). Flexible und inklusive Mehrheiten? Eine Analyse der Gesetzgebung der Minderheitsregierung in NRW. Zeitschrift für Parlamentsfragen 43(4): 887-900.

Bock, R. Darrell (1972). Estimating item parameters and latent ability when responses are scored in two or more nominal categories. Psychometrika 37(1):29–51.


This piece was first published in German by MZES (mannheimer zentrum für europäische sozialforschung)

The English edition of this report was launched at an event at Kings College London, hosted in partnership with the International Association for the Study of German Politics.


Image credit: Shutterstock.com / 360b