The Push and Pull of Ministerial Resignations in Germany, 1969–2005

Abstract

When and why are cabinet ministers forced out of office? We argue that ministerial resignations cannot be understood as mechanistic consequences of serious personal or departmental errors as the classical responsibility hypothesis implies. Rather, they follow a systematic political logic. Cabinet ministers have to resign whenever the prime minister perceives the political costs of a minister staying in office to be higher than the benefits of keeping the status quo. We test this argument with resignation events in Germany in the period 1969 to 2005. Based on detailed data collection, we find 111 resignation events, i.e. serious public discussions about a cabinet minister's future, 14 of which ended in resignation. These data are analysed employing statistical as well as Qualitative Comparative Analysis based on Boolean algebra to detect patterns of ministerial resignations.

Acknowledgements

An earlier draft was presented at the ECPR Joint Sessions of Workshops, Granada, 14–19 April 2005 (Workshop ‘The Selection and De-Selection of Ministers’). We gratefully acknowledge constructive comments by the workshop participants and an anonymous referee. Thanks also to Simon Franzmann, Alexandra Patin and Natalie Ruppert for research assistance.

Notes

1. Home Secretary Rudolf Seiters, as quoted in Süddeutsche Zeitung, ‘Ich trete zurück ohne Bitterkeit’, 5 July 1993, p. 1.

2. Minister of Justice Sabine Leutheusser-Schnarrenberger, as quoted in Frankfurter Allgemeine Zeitung, ‘Justizministerin denkt nicht an Rücktritt’, 12 July 1993, p. 2.

3. The Home Office is in charge of the Federal Criminal Police Office, the Ministry of Justice of the Federal Prosecuting Attorney General.

4. This includes male and female office holders, although not surprisingly in Germany a large majority is still male. We do not distinguish between male and female cabinet ministers in the text for reasons of legibility.

5. Paul Simon, ‘Still Crazy After all these Years’, 1975.

6. Coding decisions are made according to the following rules: a push resignation is only coded as such if it was preceded by a resignation event (see below for the resignation event criteria). As a general rule, we prefer to code cases according to official statements rather than media speculation. Nevertheless, once we detect conspicuous discrepancies we report them in a footnote. However, the number of cases with discrepancies between the official communiqué and media reports is fairly small.

7. For a similar delineation see Sutherland (1991: 101).

8. We borrow these terms from migration research where push and pull motives of migration are distinguished, see Lee (1972).

9. This is clearly the case with 3 out of 11 ministers who were shuffled out during a term: Lauritz Lauritzen (Minister of Transport, 1974), Marie Schlei (Minister for Economic Cooperation and Development, 1978), and Rupert Scholz (Minister of Defence, 1989).

10. It is probably no coincidence that cabinet members exposed to public or internal party criticism may suddenly find themselves posted to Sarajevo or as an official in Nairobi. This happened to former Federal Minister for Special Tasks and Head of the Federal Chancellor's Office Bodo Hombach in 1999, who ended up as Special Coordinator of the Stability Pact for South-East Europe. He was unpopular with the left wing of the SPD who blamed him for poor coordination and communication within the, at the time, crisis-prone government. Furthermore, allegations about financial irregularities regarding the construction of his private home came up. Nevertheless, when he stepped down the worst intra-party criticisms were already three months away and accusations regarding his house were too vague to start a serious resignation debate. The former Minister of Regional Planning, Building and Urban Development Klaus Töpfer became Executive-Director of the United Nations Environment Programme in 1998. His intra-party standing was alright and he was far from being involved in any scandal. However, journalists argue that he had become too powerful for some party leaders and that Nairobi was considered to be the right place for him to spend the next years. Out of the 13 pull resignations covered from 1969 to 2005, Hombach is indeed the only one where the official reasons given for his resignation are doubtful and the connection between leaving the cabinet and his intra-party standing as well as personal enrichment allegations is rather obvious. With regard to neutral resignations, two out of a total of six can be argued to have an unequivocally ‘pushy’ character: Hans Leussink (Minister of Education, 1972) and Gerda Hasselfeldt (Health Minister, 1992).

11. In a comparative perspective readers might miss the sex scandal. However, the German media and public do not pay too much attention to politicians' private conduct, as long as it does not affect their administrative capabilities.

12. Our compilation, for the period 1969 to 1998 based on Kempf and Merz (2001).

13. From 1969 to 1992 the collection is based on a microfilm search; from 1993 to September 2005 we used an online Lexis-Nexis search with the following two search combinations: ‘!rücktr! + !minister! + !bundes!’ and ‘entlass! + !forder! + !minister! + !bundes!’. Due to an unexplained omission the Frankfurter Allgemeine Zeitung is not included in the Lexis-Nexis database for 2 June 1999 to 6 July 1999. Here we had recourse to the traditional microfilm collection. We tested for external validity by comparing the units of observation found in the Frankfurter Allgemeine Zeitung with the Süddeutsche Zeitung.

14. Dowding and Kang's (1998: 425) analysis of the United Kingdom follows a similar procedure with The Times.

15. This is called a ‘Bauernopfer’ (scapegoat), a term originating in chess, in the German language.

16. A = party leader, leader of parliamentary party, general secretary of party, candidate for federal chancellorship. B = e.g. deputy leader of parliamentary party, party spokesman on policy areas, manager of parliamentary party, leader of state party group in parliament, ‘the Opposition’, leader of state party, Minister of state government, Prime Minister of state government, chairman of major interest group. C = e.g. MP, MEP, local politician, spokesman of party-affiliated organisation, spokesman at demonstration.

17. This idea is similar to hypotheses about government popularity and voting behaviour in mid-term and second-order elections (see Miller and Mackie 1973; Reif and Schmitt 1980).

18. According to the Basic Law, the German constitution, a parliamentary petition for dismissal does not push the minister out of office automatically. Technically, it is a call to the Federal Chancellor to ask the Federal President to dismiss a minister.

19. However, not every resignation issue is suitable for the establishment of such a committee.

20. The minister's position is coded ‘against’ when he is unwilling to resign and ‘in favour’ when he offers his resignation. In a few cases the position is coded as ‘unclear’, for example when there are rumours in the news that a minister has privately offered to resign.

21. In all cases where the Federal Chancellor publicly supports the minister concerned we code the position as ‘against resignation’.

22. For the coding of the positions of media and public see below.

23. One can find mixed positions either with collective actors or when an actor changes his opinion during a resignation debate.

24. A prominent pull resignation was the one of Martin Bangemann (Free Democrats – FDP, Minister for Economic Affairs). In 1988 he took the post of EU Commissioner for Industrial Affairs, Information Technology and Telecommunications. Surprisingly, after a short while he was pulled out of this office as well, this time by the Spanish telecommunications company ‘Telefónica’. The most recent pull resignation was in 1999, when Franz Müntefering, Minister of Transport, Building and Housing, took the post of secretary general of the Social Democratic Party (SPD).

25. The most spectacular protest resignation in the last 30 years was surely Oskar Lafontaine's (SPD, Finance Minister) in 1999, who, after resigning, kept absolute silence for days, before he complained about the intra-governmental lack of team spirit. Interestingly, the Ministry of Finance seems to be particularly susceptible to protest resignations. Alex Möller and Karl Schiller did so in 1971 and 1972 respectively.

26. ‘Health problems’ were the official reasons both for the resignation of Health Minister Gerda Hasselfeldt (Christian Social Union – CSU, 1992) and Minister of Education and Science Rainer Ortleb (FDP, 1994).

27. Long-time Foreign Minister Hans-Dietrich Genscher (FDP) resigned in 1992 at the age of 65.

28. This finding of a highly variable resignation ratio among governments differs considerably from Dowding and Kang's (1998: 416–17) results for Great Britain. The overall average in Germany (12.6%) is significantly lower than in Britain (35%), but this is due to the fact that the ratio depends very much on what is counted as an event. The higher the threshold for identifying resignation events, the higher the ratio because the value in the denominator decreases.

29. The Christian Democratic Union and the Bavarian Christian Social Union form a parliamentary party together in the Bundestag, the federal parliament.

30. In terms of parties' ratios Dowding and Kang (1998: 417) reach a similar finding comparing the Conservatives and Labour in Britain.

31. See note 16 for the coding of the authors that are subsumed under each category.

32. This was Werner Maihofer (FDP).

33. The statistical association is far below a significant level here.

34. This refers to the resignation of former Minister of Transport, Building and Housing Reinhard Klimmt (SPD) in 1999.

35. Among those is one case concerning a minister of the smaller coalition party. However, this was due to a very peculiar constellation. In the course of a dramatic intra-government conflict between the SPD and the FDP concerning social, economic, and financial affairs in 1982, the publicly pro-resignation minded Federal Chancellor, Helmut Schmidt (SPD), did not make use of his constitutional powers against the Minister for Economic Affairs, Otto Graf Lambsdorff (FDP). The four FDP ministers, however, resigned en bloc during the running resignation issue.

36. One of the few examples where this happened is the interesting case of the discussion about Foreign Secretary Joschka Fischer's militant past as a street-fighter which ended in a non-resignation.

37. It would be possible to develop a regression model capturing multiple conjunctural causality (Braumoeller 2003). A regression model, however, requires specifying deductively configurations of conditions. QCA, in contrast, derives these configurations inductively from the data. We consider the inductive approach more appropriate. The deductive specification of a regression model might be feasible when one has a strong theory at hand, as for example in research on deterrence (Braumoeller 2003). Research on ministerial resignation is in its infancy and comparatively weak. Therefore, the inductive approach is more suitable for our purposes.

38. Interval-scaled data can be handled with fuzzy-set analysis (Ragin 2000; Ragin and Pennings 2005). Apart from the fact that the data can be easily dichotomised, we consider it not convincing to conceive of conditions as ‘nearly sufficient’ or ‘hardly necessary’ (cf. Symposium 2005).

The significant categories of the judicial and positional variables could be entered into QCA. Each category could be conceived of as a dichotomous condition. In this instance 15 conditions would be covered by the model. The ratio of conditions to cases would be too unfavourable, even for QCA that was designed for small-n and medium-n settings. The problem would lie in the simplifying assumptions that are made in the minimisation of the truth table (see below). Ceteris paribus, the smaller the ratio of cases to conditions, the fewer logically possible configurations are observed empirically. This boosts the simplifying assumptions tremendously. Since the less simplifying assumptions are the better (Ragin 1987), it is advisable to reduce the number of conditions on theoretical grounds. By recoding the categorical variables in this manner, the number of conditions decreases to nine.

40. NO_CRIM is a significant category of the judicial variable.

41. All positional conditions are significant categories of the respective positional variable.

42. BEF_AFT = date of offence, NO_CRIM = criminal relevance, POS_MINIS = minister's position, POS_CHANC = Federal Chancellor's position, POS_OWNP = position of the minister's party, POS_COAL = position of the coalition partner, POS_OPP = position of the parliamentary opposition, POS_MEDPUB = position of the media and public, POL_KIND = political or extra-political kind of resignation issue, REL_OFFICE = relationship of offence to office.

43. By leaving this variable aside, we also mitigate the problem of simplifying assumptions to some degree.

44. Note that our analysis is based on a decision-theoretic framework. We model resignation events from the perspective of the Federal Chancellor as the actor who has the formal power to dismiss a minister.

45. The ministers, the Chancellor, the minister's party, the coalition, and the media are not in favour of resignation. This speaks against a demission. The opposition, in contrast, is for a demission. This variable thus also suggests that the minister does not resign. In addition, no judicial steps are initiated against the minister. The other two factors do not clearly speak for or against a demission.

46. Charles Ragin considers the manipulation of data a viable solution toward the removal of contradictions if the most preferred means are not applicable (Ragin 1987: 118).

47. Positive outcomes are minimised and configurations that exist logically, but not in the truth table (so-called remainders), are treated as ‘don't cares’ in order to derive the minimal solution. The solution results automatically on the basis of the minimisation done by QCA 3.0 (http://www.u.arizona.edu/∼cragin/fsQCA/software.shtml) so that there is only one minimal solution.

48. The only case with this configuration is the resignation of Horst Ehmke. He resigned as Minister of Research, Technology and Postal Affairs, taking responsibility for an issue which he had been concerned with as Chancellery Minister in Willy Brandt's first cabinet – the ‘Guillaume Affair’. Brandt's top personal assistant turned out to be an Eastern German spy. The Chancellor resigned and thus the whole cabinet left office; however, Ehmke made his decision to leave his post public before Brandt announced his step. Generalising on this, one could argue that if an issue with a political dimension with roots in the past comes up, and therefore is apparently considered to be grave enough to be subject of public discussion, then the minister is in serious trouble.

49. Due to the German unification in 1990, four Eastern German politicians participated in Kohl's cabinet as ministers without portfolio. Just two and a half months later one of them, GDR's last prime minister, elected after the fall of the Berlin Wall in April 1990, Lothar de Maizière, fell victim to a media campaign concerning his alleged activity for the Ministry for the Safety of State in the GDR, which coordinated information gathering about suspected regime opponents. De Maizière, denying the allegations fiercely, was asked to give evidence of his innocence – an extremely difficult task that he was unable to accomplish. Being a side-entrant in politics he soon became fed up with the conduct in the Bonn Republic and resigned, although the opposition's calls for resignation were noticeably absent. This is due to the fact that the opposition was aware of the particularity of this matter in German–German history and did not participate in convicting on unproven accusations a person widely considered as behaving with integrity. This explains why this resignation is clearly an exception and unlikely to be repeated.

50. Negative outcomes are minimised and remainders are set as ‘don't cares’. For negative outcomes, several minimal solutions are obtainable. The above solution is derived by selecting by hand the prime implicants pos_chanc∗pos_ownp∗POS_OPP∗pol_kind and no_crim∗pos_chanc∗pos_ownp. The minimisation leading to this solution does not produce any simplifying assumptions contradictory to those underlying the minimisation of positive outcomes (cf. Vanderborght and Yamasaki 2003).