Portal de Periódicos da CAPES

The Swiss Upper House: ‘Chambre de Réflexion’ or Conservative Renegades?

2010; Taylor & Francis; Volume: 16; Issue: 2 Linguagem: Inglês

10.1080/13572331003740099

1743-9337

Sarah Bütikofer, Simon Hug,

European history and politics

Abstract Theoretical analyses suggest that bicameral systems have policy consequences if the preferences of the two chambers differ. This paper offers an analysis of the ideological positions of the MPs in the two chambers of the Swiss parliament. Contrary to conventional wisdom the analyses relying on MP surveys and roll call analyses suggest that the MPs of the same party hardly differ with respect to their ideological positions. While the MP survey suggests that the Swiss upper house is more conservative given the underrepresentation of leftist parties in the latter chamber, similar differences fail to appear in the roll call analyses in a systematic way. Hence, the Swiss upper house is hardly a conservative bastion. Keywords: bicameralismpolitical preferencesideological positionrepresentationSwiss upper houseroll call votes Acknowledgements Thanks are due to the Swiss Parlamentsdienste and especially Ernst Frischknecht and Andreas Sidler, who provided the main bulk of the empirical data used here, and helped us in using it. Franziska Spörri provided much appreciated research assistance, the Swiss National Science Foundation (Grant No 100012-111909) generous funding, and three anonymous reviewers greatly appreciated comments. Notes This obviously runs counter to Abbé Siyès' famous quip that ‘[s]i une seconde chambre n'est pas d'accord avec la première, elle est malveillante; si elle est d'accord, elle est superflue.’ (‘If the second chamber agrees with the first, it is useless, if it disagrees it is dangerous’ cited in Tsebelis and Money (Citation1997, p. 1). Levmore Citation(1992) offers a more general discussion along these lines. This becomes obvious in Lijphart's Citation(1999) classification in strong and weak bicameral systems in part according to differences in preferences. Similarly, the informational advantages of a bicameral setting obviously also depend on the presence or absence of differences in preferences between the two chambers. In an earlier version we used a related method proposed by Poole (Citation2000, Citation2005), which required us to estimate the ideological positions in two dimensions. To avoid this requirement, which as Hug and Schulz Citation(2007) demonstrate makes sense, we employ a technique based on Bayesian inference without affecting the main substantive results. Groseclose et al. Citation(1999) present an alternative way to connect preferences of actors belonging to different institutions. Riker Citation(1992) advances the same explanation for differences in policy stability between the bicameral US and the unicameral Great Britain, while McCarty and Cutrone Citation(2006) defend the view that bicameralism has no effect. Six MPs, representing only small electoral districts (cantons) are elected in plurality elections. In only one canton, namely the canton of Jura, the voters chose their MPs of the upper house in a PR system. The six half-cantons only elect one MP to the upper house. König Citation(1996) in part offers such an analysis based on power indices in the Swiss bicameral setting. As with all power index studies, however, preferences of the political actors are bracketed. These two surveys are the only comprehensive ones among Swiss MPs. Kerr's Citation(1981) survey of 1974 was based on personal interviews with a response rate of 95 per cent. In one question MPs were asked to place themselves on an ideological scale from 1 (extreme left) to 21 (extreme right). The second survey was carried out in winter 2006/07, first on the internet, followed by a postal survey and finally a few selected personal interviews. One hundred and sixty out of 246 participated in this survey, where one question asked the MPs to place themselves on a left–right scale going from 0 (extreme left) to 10 (extreme right). Two additional surveys were carried out by Riklin and Ochsner Citation(1984) and Wiesli and Linder Citation(2000), both of which focused, however, on very specific topics, and the data of which is no longer available. While both houses of parliament allow for roll call votes, those need to be requested by a certain number of MPs. The lower house, however, introduced an electronic voting system in 1994 that records all votes. Of these, only the requested roll calls and the final and global votes as well as decisions on emergency measures are published. Hence, starting with 1994 much more information is available on the individual decisions by MPs. In this paper we rely on data based on all votes recorded by the electronic voting system, which was made available to us by the Parlamentsdienste. When presenting the results we will return to some subtleties in this data. Both Bailey and Chang Citation(2001) and Bailey (2005) suggest in part a different way by integrating additional information in their Bayesian framework. A fifth roll call vote took place in the upper house on 15 December 2005, incidentally on whether this chamber should also use an electronic voting system. We refrain from using this additional vote, since it occurred in a different legislative period than the previous votes and thus would have required much stronger assumptions for our analysis. In between the two chambers voted again on this bill, on 20 March 2003 and 18 June 2003, respectively, such that the bill had shuttled three times between the two houses. The additional vote on this bill was, however, not recorded nominally in the upper house. We obtained this information from Ernst Frischknecht of the Parlamentsdienste, to whom we are much obliged. These MPs are Michel Béguelin (SP), Eugen David (CVP), Toni Dettling (FDP), and Simon Epiney (CVP). We could add additional MPs who had made the switch from the lower to the upper house earlier though this would require stronger assumptions about the stability of preferences over time, which explains why we refrained from doing so. Both of these possibilities of linking different chambers through roll call votes are discussed in detail by Poole (Citation2005). Poole Citation(2005) provides a detailed overview of these methods and their advantages and shortcomings. Hug and Schulz Citation(2007) discuss and compare these methods for the Swiss context. As mentioned above, in an earlier version of this paper we employed the so-called ‘optimal-classification’ method proposed by Poole Citation2000. This method relies on fewer largely untestable assumptions that other methods must adopt. One drawback of this method is that under the assumption that legislators distinguish themselves only on one dimension, only a rank-order of MPs can be derived. If more dimensions are assumed, the location can be estimated with greater precision. This is an additional advantage of IRT models over ‘optimal classification’ as the latter does not provide uncertainty estimates directly (see Lewis and Poole Citation2004). The same result is obtained if all parties represented in any of the two chambers are considered (see Appendix Table 2). Hug and Schulz's (2007) estimation of ideological positions while allowing dynamic changes from one legislature to the next provides some support to this assumption.