Effect of the proportionality degree of electoral systems on corruption

摘要

This work makes two major innovations within the empirical literature on the effect of electoral systems on corruption. First, in order to distinguish electoral systems, we propose to measure their degree of proportionality, instead of using dummy variables. A measure of the proportionality degree seems the appropriate way to identify electoral systems in an empirical setting because it allows all their variants to be considered. Second, alongside a parametric panel data approach, we provide a new (in this field) semi-parametric technique to estimate a non-linear relationship between the proportionality degree of electoral systems and corruption. Our results show that, when we increase plurality elements into proportional, both voters' and opponents' monitoring power induce politicians to avoid corrupt behaviour. But it is beneficial only up to certain proportionality degrees, after which this effect stops. Moreover, data show that the 'best' range of proportionality degrees is most likely found under mixed electoral rules. For governors the choice of the proportionality degree therefore becomes fundamental.