Meeting the Challenges of Representation and Accountability in Multi-party Governments

Abstract

In systems of proportional parliamentarism political parties play a double role. On the one hand they make delegation and accountability work; on the other they add complexity to the delegation regime, as minority situations require inter-party cooperation. Because coalition government usually involves policy compromises, the question arises how the coalition parties can ensure that the ministers stick to the coalition deal. Employing the principal–agent framework, this paper shows that coalitions can use several control mechanisms to pursue this goal. The authors consider ex ante mechanisms such as policy agreements that set the agenda for future policy decisions and coalition screening of ministerial candidates. Next they discuss the effects of ex post mechanisms such as strong committee systems and institutional checks like ‘watchdog’ junior ministers. Employing a simple spatial model, they illustrate how these instruments work. Using control mechanisms is not costless, however, and actors may want to avoid these costs. The article specifies conditions that make the use of control mechanisms likely to occur.

Notes

1. Following what now seems convention, we assume female principals and male agents throughout the article.

2. We can further distinguish sabotage as a sub-category of shirking (see Brehm and Gates 1997; Müller 2000a). However, the distinction between leisure-shirking and dissent-shirking is precise enough for the present purpose.

3. Other parties require the summoning of a party congress for making the decision to enter coalitions. These events typically take place only late in the game but nevertheless contain elements of risk that will be anticipated in the negotiations (see country chapters in Müller and Strøm 2000a).

4. Note that coalition parties might benefit from the appointment of ministers by their partners who are too incompetent to make full use of all the policy concessions these parties had been able to achieve. Yet failure is unlikely to be limited to issues contested between the coalition partners and indeed may undermine the entire cabinet's credibility and electoral prospects.

5. Such mechanisms were, for instance, developed during the first ‘grand coalition’ period in Austria (until 1966). At that time, the coalition parties normally did not negotiate policy issues ex ante but relied on rules how to resolve problems once they occur (see Müller 2000b). However, only few cabinets use procedural rules to a similar extent (Müller and Strøm 2000b: 575–8).

6. That is not to say that these rules are irrelevant. In Germany, for example, the agreement of the CDU/CSU–FDP coalition in 1961 prescribed the resignation of Chancellor Adenauer after two years. In return, the FDP promised to vote for Adenauer in the investiture vote (Saalfeld 2000).

7. If parties A and B agree to implement a policy within their winset (i.e. the cross-section of the circles surrounding their ideal positions), party C can offer another policy which is preferred by at least one of the parties A or B. This ‘cycling’ process continues if rarely existing conditions are not met. Laver and Shepsle's lattice points concept is a theoretical explanation for why cycling does not occur and produces point-predictions. Our more realistic approach of negotiated coalition policies implies that the cycling problem is back on the table. The control mechanisms discussed in this article are designed to keep it at bay.

8. For simplicity, we assume a symmetric interval (i.e. the margin of interpretation is the same to the left and the right).

9. Logically, the opposite can also occur: Respect of coalition discipline can prevent the coalition programme from being enacted. This may happen when a minister is not acting upon the coalition's policy programme while the parliamentary opposition would be ready to vote the policy into law on the basis of a parliamentary proposal. For all the other reasons we discuss here, and on the basis of our own non-systematic observations, we consider this alternative empirically less important.

10. As for the coalition agreement, we assume symmetry for the interval surrounding the proposal P. However, asymmetric pictures are also possible.

11. While Laver and Budge (1992) base their analyses on both party manifestos and government declarations, later publications of the manifesto project satisfy themselves with calculating estimates for government positions from party manifesto data and seat shares (see e.g. Budge et al. 2001). Whatever the merits of these estimates, they do not pay tribute to bargaining and its complexities.