Portal de Periódicos da CAPES

Institutions and coalition formation: The German election of 2005

2006; Taylor & Francis; Volume: 29; Issue: 3 Linguagem: Inglês

10.1080/01402380600619884

1743-9655

Sven‐Oliver Proksch, Jonathan Slapin,

Social Policy and Reform Studies

Abstract The German election of 2005 creates three puzzles for the literature on coalition formation. First, the election led to a rare event in German politics and in parliamentary systems more generally, a 'grand coalition' between the two largest parties. Second, a minority government, something which has never occurred in postwar Germany (except briefly as the result of the breakdown of a government coalition), was in fact one of the two most likely governments to form. Third, the parties of the left retained a comfortable majority in the Bundestag; however they did not form a coalition. The election of 2005 appears unique in German politics, but we argue that its outcome is easily understood using existing institutional theories of coalition formation. We examine party positions in two dimensions (economic and social) using computer-based word scoring of party manifestos. We demonstrate that the conditions for a SPD minority government were present in Germany due to its central location in the policy space. While the configuration of policy positions would thus have allowed the SPD to form a minority government, the role of the Federal President as a veto player could have prevented it from forming, and the presence of an opposition-controlled upper house would have decreased its effectiveness. The mere possibility of forming a minority government gave the SPD a bargaining advantage in the coalition negotiations with the CDU/CSU. We show that in the final portfolio allocation, the SPD received ministries which control approximately two-thirds of the federal budget. Acknowledgements We would like to thank Thomas König, George Tsebelis, and two anonymous referees for their helpful comments. In addition, we appreciate the additional insights from the participants at the UCLA Political Science Speakers' Series, in particular from Kathy Bawn and Ron Rogowski. Notes 1. In his speech before the Bundestag, Schröder said: 'Ever since we decided on the Agenda 2010 [the reforms], the SPD has lost votes in state elections and in the election for the European Parliament . … This was a high price for pushing through the reforms . … Some SPD members threatened to join a backward, populist left party … The question, then, was whether the capacity to act still existed, given the slim majority [for the coalition government] in the Bundestag.' Gerhard Schröder, 185th Plenary Session of the 15th Bundestag, 1 July 2005, http://dip.bundestag.de/btp/15/15185.pdf. 2. In Germany, there are two types of confidence votes. The first is a constructive vote of no confidence initiated by the Bundestag. This requires that the Bundestag elects a new chancellor to replace the incumbent chancellor. The Federal President must comply and appoint the newly elected chancellor (Art. 67, German Constitution). We refer to the second type of confidence vote, which is called by the incumbent chancellor, not the Bundestag (Art. 68). 3. Following the election, Schröder declared that, 'for the next four years, Germany will have a stable government under my leadership. … No one else except me is able to form a stable government. … All those who wanted a change in the chancellor's office have failed spectacularly.' Der Spiegel, 18 September 2005. 4. Official statistics of the Bundesrat. 'Die Arbeit des Bundesrates im Spiegel der Zahlen, 1949–2004', http://www.bundesrat.de 5. For other applications of this approach see Laver and Benoit (2002 Michael, Laver and Benoit, Kenneth. 2002. Locating TDs in Policy Spaces: The Computational Text Analysis of Dail Speeches. Irish Political Studies, 17(1): 59–73. [Taylor & Francis Online] , [Google Scholar]), Giannetti and Laver (2005 Giannetti, Daniela and Laver, Michael. 2005. Policy Positions and Jobs in the Government. European Journal of Political Research, 44(1): 91–120. [Crossref], [Web of Science ®] , [Google Scholar]), Hug and Schultz (2005 Hug, Simon and Schulz, Tobias. Referendums in the EU's Constitution Building. Paper presented at DOSEI Conference in Brussels. June19–222005. [Google Scholar]), and Bräuninger and Debus (2005 Bräuninger, Thomas and Debus, Marc. Intraparty Factions and Coalition Bargaining in Germany. Paper presented at ECPR joint session. April2005, Granada. [Google Scholar]). 6. Benoit et al. have written a computer program, wordscores, to compute these scores, freely downloadable at http://www.wordscores.com. See Laver et al. (2003 Laver, Michael, Benoit, Kenneth and Garry, John. 2003. Extracting Policy Positions from Political Texts Using Words as Data. American Political Science Review, 97(2): 311–331. [Crossref], [Web of Science ®] , [Google Scholar]) for a more complete example of how the program works. 7. We do not claim that these two dimensions are always the most important ones. For example, in 2002, foreign policy (Iraq war) was certainly a divisive issue dimension as well. 8. The reference texts, party manifestos and replication information are available at http://jslapin.bol.ucla.edu/ 9. We thank Thomas König for pointing out these two possibilities. 10. We thank one anonymous referee for pointing this possibility out to us. 11. For the social dimension, we calculated the conditional probability that the Left Party's position was greater than some arbitrary point c given that the SPD's position was located at c for all possible values of c. Because the distributions of the positions are normal, this becomes . We integrate through Monte Carlo simulation. We do the same for the economic dimension (Greens and SPD). 12. In total, there were seven minimum winning coalitions: CDU/CSU–SPD, CDU/CSU–FDP–Greens, SPD–FDP–Greens, SPD–Left–Greens (all of which we discuss), as well as CDU/CSU–FDP–Left, CDU/CSU–Left–Greens, SPD–FDP–Left (which were not considered by any of the parties involved, and which we therefore exclude from our discussion). 13. For simplicity, we ignore the confidence intervals in this part of the analysis. The confidence ellipses would simply increase the size of the pareto sets, but our substantial findings would continue to hold. 14. 'Die Jamaika Koalition ist nichts fuer die Grünen', Tages-Anzeiger, 24 September 2005, p. 5. 15. 'Nicht geliebt, aber umworben. Union geht auf Grüne zu', Süddeutsche Zeitung, 20 September 2005, p. 2. 16. 'Westerwelle sagt Nein zur Ampel und "anderen Ampeleien"', General-Anzeiger, 19 September 2005, p. 5. The party chief repeated this statement several times. 17. 'Den Bundesrat als Gegner. Eine Ampelkoalition hätte wenig gemein – und träfe auf eine starke Länderkammer', Süddeutsche Zeitung, 20 September 2005, p. 5. 18. 'Müntefering schliesst Rot-Grüne Minderheitsregierung aus', Spiegel Online, 21 September 2005; 'Minderheitsregierung is besser als Große Koalition', Spiegel Online, 23 September 2005; 'Minderheitsregierung oder Neuwahl', Berliner Zeitung, 20 September 2005, p. 2. 19. Technically, because states vote in blocks in the Bundesrat, a state, not a party, controls the median. However, in this case it is quite clear that a state with a CDU government controls the median (see Appendix B). 20. Because a minority government must win a plurality, a CDU or CDU/FDP minority government was less likely than an SPD minority government. Some members of the Left Party may have supported the SPD if faced with the prospect of a right-wing minority government. Angel Merkel, CDU leader, was aware of this possibility, and therefore excluded a CDU/CSU minority government ('Merkel gegen Minderheitsregierung', Frankfurter Rundschau, 23 September 2005, p. 1). 21. 'SPD-Parteitag billigt Koalitionsvertrag', Frankfurter Allgemeine Zeitung Online, 14 November 2005.