Runoff Elections in the Laboratory

We study experimentally the properties of the majority runoff system and compare them to the ones of plurality rule, in the setup of a divided majority. Our focus is on Duverger's famous predictions that the plurality rule leads to a higher coordination of votes on a limited number of candidates than the majority runoff rule. Our experiments show that, in contradiction with Duverger's predictions, coordination forces are strong in majority runoff elections. We indeed observe similar levels of coordination under both rules, even when sincere voting is an equilibrium only under majority runoff. Our results suggest that the apparent desire to coordinate, and not vote sincerely, under the majority runoff rule is to some extent not rational. Finally, we find insignificant differences between runoff and plurality systems in terms of both electoral outcomes and welfare. This is so exactly because coordination forces are strong under both rules. But, this does not mean that the two rules are equally socially desirable. Majority runoff rule entails an additional cost: second rounds that take place frequently.