Bayesian implementation, efficiency, and independence classes

Abstract

The theory of Bayesian mechanism design is of interest to economists and computer scientists alike. It has focused on two extreme assumptions on the beliefs of the agents, full-freeness (or independence) and no-freeness (or Beliefs Determine Preferences). We discuss more general conditions that cover intermediate cases between these two extremes and characterize the corresponding set of implementable mechanisms. We also discuss applications of these results to economics and to computer science.

Reference

Jacques Crémer, and Claude D'Aspremont, “Bayesian implementation, efficiency, and independence classes”, TSE Working Paper, n. 24-1593, November 2024.