Neutral Optima in Informed Principal Problems with Common Values

In a class of informed principal problems with common values often used in applications we de\u85ne a particular mechanism which we call the assured allocation. It is always undominated, i.e. e ¢ cient among the di¤erent types of the principal. We show it is a perfect Bayesian equilibrium allocation of the three-stage game studied in Maskin and Tirole (1992) that coincides with the Rothschild-Stiglitz-Wilson allocation when the latter is undominated. Under familiar conditions on hazard rates we show that the assured allocation is a neutral optimum in the sense of Myerson (1983)