Self-Equivalent Voting Rules

Abstract

In this paper, I introduce a novel stability axiom for stochastic voting rules, called self-equivalence, by which a society considering whether to replace its voting rule using itself will choose not to do so. I then show that the unique voting rule satisfying the democratic principles of anonymity, efficiency, monotonicity, and neutrality as well as the stability principle of self-equivalence must assign to every voter equal probability of being a dictator (i.e., uniform random dictatorship). Thus, this paper suggests one reason why most democratic societies normally choose whether to replace their voting rule using a voting rule other than itself, or in binary elections against just one other voting rule.