Stress-testing the runoff rule in the laboratory

Abstract

When a majority of voters has common values, but private information, then the runoff rule always admits an equilibrium that aggregates information strictly better than any equilibrium of the plurality rule. But there are cases in which the plurality rule supports equilibria that aggregate information strictly better compared to certain undominated equilibria of the runoff rule. Is there any risk with applying the runoff rule in these situations? We conduct a laboratory experiment and we show that the runoff rule consistently delivers better outcomes than the plurality rule even in such unfavorable scenarios. This establishes that the superiority of the runoff rule over the plurality rule in empirical settings outperforms its theoretical advantages.