Optimal Voting Mechanisms on Generalized Single-Peaked Domains

This paper studies the design of voting mechanisms. There are multiple alternatives, and the voters have generalized single-peaked preferences derived from median spaces as introduced in Nehring and Puppe (2007b). This class of preferences covers a wide range of economically relevant domains, and it is much larger than the well-known single-peaked preferences on a line. I characterize the welfare-maximizing voting rules among all social choice functions satisfying strategy-proofness, anonymity, and surjectivity. The optimal mechanisms are composed of binary votes on subsets of alternatives involving flexible majority requirements. (JEL D71, D72, D82)