Strategy-Proofness and Efficiency of Probabilistic Mechanisms for Excludable Public Good

We study strategy-proof probabilistic mechanisms in a binary excludable public good model. We construct a new class of probabilistic mechanisms satisfying strategy-proofness, called α-mechanisms.We first show that the α-mechanisms are second-best efficient. Next, we identify the optimal α-mechanism with respect to the supremal welfare loss, and show that it improves the inefficiency of the equal cost sharing with maximal participation mechanism [Moulin (1994)] and the anonymous augmented serial mechanisms [Ohseto (2005)]