Individual Submission Summary

Direct link:

Electoral Systems and Partisan Coalitions: A Regression Discontinuity Analysis

Sun, September 2, 8:00 to 9:30am, Hynes, 301


An influential literature in CPE suggests that the dominance of parties of the left or right across countries-- and hence different patterns of redistribution---are driven by electoral systems (Iversen and Soskice). According to this argument, PR systems are conducive to coalitions between the lower and middle classes against the wealthy; in these systems left parties are more likely to gain power. SMD systems, in contrast, are viewed as promoting coalitions between the middle and upper-classes against the working-class, thereby promoting the dominance of center-right political formations. Although intriguing, extant empirical evidence in favor of this proposition remains largely correlational: scholars have shown that parties of the left get higher vote shares in PR systems; likewise, center-right parties tend to do better in SMD systems. Moreover, recent qualitative studies have created substantial debate about the endogeneity of electoral system choice to partisan strength (Ahmed), raising further doubts about the linkages between political institutions and partisan dominance.
This paper intervenes into debates about the relationship between electoral systems, partisan coalitions and redistributive politics by offering a new micro-level research design through which to test posited mechanisms underlying the theory. It draws on an original dataset on rollcall voting from the French Senate, and leverages the fact that, depending on the size of the district, French senators are elected under distinctive electoral rules. We use the cutoff for assignment to SMD vs PR as a strategy for identifying the effect of electoral rules on voting patterns, investigating whether left and center-right Senators from PR districts are more likely to vote together than their counterparts in SMD districts (on either side of the thresshold), as suggested by the Iversen et al framework. If, however, there are no differences in voting patterns across the two types of districts then there is less evidence for the electoral systems claim.