Government Alternation and Legislative Party Unity: The Case of Italy, 1988–2008

Abstract

The large literature on legislative party unity identifies the confidence relationship, i.e. the threat of being voted out of office and losing agenda setting powers, as well as cabinet membership, as two crucial institutional sources of party discipline. However, by focusing on the dramatic change in the Italian political system following the 1994 election, the article shows that the impact of these factors on party unity (and the direction of this impact) hinge crucially on the possibility of government alternation rather than mere cabinet turnover. This is illustrated by an index of party unity that explicitly focuses on the behaviour of individual MPs derived from a roll-call analysis of the Italian Chamber of Deputies during the period 1988–2008.

Acknowledgements

An earlier version of this article was presented at the annual meeting of APSA in Washington DC, 2010. We acknowledge helpful comments from the seminar's participants, from Willy Jou and also comments from two referees. Our work was supported by the Italian Ministry for Research and Higher Education, Prin 2009, prot. 2009TPW4NL_002.

Notes

 1. The fact that there was no expectation of cabinet alternation (defined below) in our case study (Italy) during a long period of its political history did not prevent it from being classified as a free and democratic country according to common indices (such as Polity or Freedom House).

 2. In other words, only the experience (or the expectation) of a major rather than minor governmental change (see Feng 1997 on this distinction) can be reasonably associated with government alternation.

 3. Some scholars argue that a new era of alternation began in Italy only after the 1996 general election, when for the first time a centre-left coalition dominated by the heirs of the Communist Party won the elections (see Newell 2000). Our results are not substantially affected by this alternative dating of when the new Italian political system was born.

 4. It is worth noting the very high correlation between our index of party unity and the Rice Index (around 0.90). This is reassuring for the reliability of our measure.

 5. More specifically, we recorded absence as a negative vote only when at least 90 per cent of MPs belonging to a parliamentary group were absent and the remaining MPs voted ‘nay’. For a similar approach see Ferrara (2004) and Curini and Zucchini (2010).

 6. To fit the uni-dimensional model, we constrained the ideal points to have a mean of 0 and a standard deviation of 1 across legislators. This is the ‘normalise’ option in the implementation of IDEAL using R statistical software (Clinton and Jackman 2009). Our analysis in IDEAL is based on 200,000 iterations with a thinning interval of 100.

 7. Replicating our analysis by using the ideal points estimated in two dimensions does not alter any of our findings.

 8. Under plausible assumptions, this fact can lead to underestimation of the ‘true’ level of intra-party policy conflict as we record it from the IDEAL analysis, due to the possibility of selection bias in roll-call votes (see Carrubba et al.2006). Note that this problem is present in any kind of party unity measure, including the classical indices.

 9. See Desposato (2006) for a similar measure of party unity based on a roll-call spatial analysis.

10. The DIVERGENCE variable reflects the overall unity of a party, where ‘unity’ is used to capture the observable degree to which members of a group act in unison (see Sieberer 2006). Whether unity is mainly due to cohesion (i.e. the existence of shared preferences among the MPs within the same party) or discipline (i.e. the by-product of effective sanctions and/or positive incentives that make members vote together even though their preferences differ) lies outside the scope of this article.

11. The dependent variable selected by most studies that investigate party unity at the level of individual MPs is the number of times each MP votes against the party line (see Stratmann 2006; Tavits 2009; for the Italian case see Curini et al.2011). In addition to being based on a clear behavioural model of individual choice (Poole 2005), the advantage of employing a spatial measure such as the one we employ here is that the spatial position of a MP compared to their own (median) party position depends on how they voted with respect to the cutline for each roll-call vote (that is, the midpoint of the Yea and Nay outcomes: see Poole 2005). In this sense, two MPs with an identical record of dissent can have a very different spatial position if the first MP voted against the party majority in a lopsided roll-call vote while the second MP did the same in a very close roll-call vote. Note that we have replicated our statistical analysis by employing as our dependent variable the raw number of dissenting votes, and the results point to the same qualitative conclusions of our analysis with respect to the conditional relationship linking ALTERNATION and CABINET. Data available upon request.

12. Conferring a value of 1 also to parties that provided external support to a government without taking up cabinet positions does not alter any of our conclusions.

13. Replicating our analysis by employing a model with fixed party effects does not affect our conclusions. We have also estimated a more complex multilevel model which assumes that observations for the same MP during different electoral periods are not independent; that is, a three-level model in which each observation for a MP is nested in the same MP across legislatures, in turn nested in a given party. Our findings, however, remain intact. Data and statistical code necessary to replicate all the analyses in this manuscript are available at http://www.socpol.unimi.it/docenti/curini/Papers2.htm

14. The presence or absence of a credible expectation of cabinet alternation can also affect other significant political processes. For example, Warwick (1994) discusses this in relation to government survival, while Chang (2005) links it to the frequency of corruption scandals.