Efficiency and Mechanisms with no Regret
探讨在非对称信息经济中，能否设计一种机制使其所有均衡结果都满足临时个体理性效率或临时无嫉妒效率。研究发现，即使采用无遗憾机制（如廉价对话或事后视角），信息泄露也无法恢复效率，存在不可能性结果。
We address the following question:given a domain of asymmetric information economies, is it always possible to design a mechanism so that all of its equilibrium outcomes are either interim individually rational-efficient or interim envy-free-efficient in the domain?We show that if the solution concept were that of Bayesian equilibrium, the answer is no.It is known that in complete information economies such games can always be constructed.Can this gap be filled by "mechanisms with no regret" : games that leak information endogenously without enlarging the equilibrium set.We adopt two approaches.First, we introduce a theory of "cheap play".This is a natural model of mechanisms with no regret.Regret-free cheap play is shown to be characterized by posterior implement ability (Green and Laffont (1987)).The second is an extreme approach characterized by an ex post viewpoint (as in Cremer and McLean (1985)).Both approaches yield an impossibility result.Information leakage does not recover efficiency.