Parliament without Government: The Belgian Parliament and the Government Formation Processes of 2007–2011

Abstract

Parliamentary systems are characterised by strong links between the executive and the legislature. While the importance of executive–legislative relationships is well-known, the extent to which executive dominance affects parliamentary behaviour is hard to grasp. This study uses the recent institutional crises in Belgium to study parliamentary behaviour in the absence of a government with full powers. Cabinet formation in Belgium has proved to be protracted in recent years, leading to long periods of government formation in both 2007–2008 and 2010–2011. Such circumstances provide a unique comparison between normal situations of parliament in the presence of government, and exceptional situations of prolonged periods of caretaker government. In particular the article looks at three aspects of parliamentary behaviour that are usually linked to executive–legislative relations: legislative initiatives, voting behaviour and party unity. The general hypothesis is that prolonged periods of government formation gave parliamentarians more opportunities to influence the legislative process and more (ideological) freedom. The results show a nuanced picture: parliament became more pro-active, the salience of the government–opposition divide declined, while party unity remained as strong as ever. It is concluded that government formation processes did not lead to drastic changes in the legislative–executive relationship, but rather permitted a modest correction to the extremely weak position of parliament.

Notes

Notes: Each dot represents one MP. Round dots are for Dutch-speaking MPs, square dots for French-speaking MPs. The grey dotted line indicates left–right positions, the solid circle contains the government parties (for the formation period this is the outgoing government). For the 2007–2008 period, the formation phase is defined as the period between June and the end of December, when an interim government was formed. The normal period runs from December 2007 to May 2010.

1. In the US context, scholars have partly overcome this problem by focussing on so-called lame-duck sessions in Congress. These exceptional sessions after the election but before the start of the new term allow the individual behaviour of Congress members that were not re-elected to be studied (see Jenkins and Nokken 2008; Rothenberg and Sanders 2000).

2. In the French-speaking part of Belgium the government is often labelled as being ‘démissionnaire’, a term that is also commonly used in the Netherlands.

3. Furthermore, in 2010–2011 the coalition had still a majority on the national level, but only 30 seats left (out of 88) on the Flemish side, which limited its legitimacy to take new initiatives. Also the Leterme government had no majority on the Flemish side, but it lacked only a few seats (41 out 88).

4. Besides Flanders and Wallonia, the two largest regions in terms of population (60 per cent and 30 per cent respectively), Belgium also consists of a third region: the bilingual capital Brussels (10 per cent).

5. These models were run using the ideal function in the pscl package in R. We ran 50,000 iterations, with a 200 thinning interval.

6. This method is called property fitting. The lines are estimated by regressing MPs left-right positions on their positions on the two IDEAL dimensions. The slope of the line plotted in the figure equals the ratio of the unstandardised regression coefficients (see Louwerse 2011: 88).

7. The Agreement Index (AI) is calculated as follows: where Y_i is the number of ‘yea’ votes, N_i is the number of ‘nay’ votes and A_i is the number of abstentions (Hix et al. 2005: 215).

8. Over the whole 1995–2011 period, 94 per cent of the governments-initiated bills and only 9 per cent of private member bills were passed by parliament. This success rate was hardly changed during the exceptional periods.

9. We have also estimated models for the 49th, 50th and 51st parliaments (not displayed for reasons of space). The patterns of parliamentary voting behaviour in these periods are similar to those of the ‘normal’ periods in the 52nd and 53rd parliament.

10. As we study a population of votes (and legislators) during a parliamentary period, it is strictly speaking not necessary or meaningful to calculate the statistical significance of these indicators. One might, however, argue that the actual votes form a (partly) random subset of all potential votes that could have been taken. Therefore, we do report difference-of-means significance tests for our indicators of government–opposition voting, language voting and left–right voting.