Selective revelation of public information and self-confirming equilibrium

Why is public information revealed? We endogeneize public information by modeling aggregate information release, in games of large populations with anonymous matching, as a rational, utility-maximizing choice. We assume that there is a “planner”, who knows and selectively reveals aggregate information, in order to maximize his objective function. we show that public information can affect individuals’ behavior, even if it is revealed after a self-confirming equilibrium (SCE) has been reached. By selectively revealing information, the planner may upset a given SCE, in order to achieve a better outcome for him. Hence, some SCE are “unstable” to public information release. It is shown that only equilibria supported by heterogeneous beliefs can be information-unstable. Hence, unitary SCE, including Nash, are robust to public information release. Two policy applications of the concept of revelation instability are considered, illustrating how concealing information (self-censorship) and inducing experimentation with untested actions (affirmative action) can be socially beneficial. We present several real examples of endogenous public information revelation, where our theory could be useful.