Coordination and Continuous Stochastic Choice
研究了在投资回报随他人投资比例和状态递增的情况下，玩家通过支付小成本获取信息时的随机选择规则，发现当不连续规则成本无限高时，唯一均衡中行动是对均匀信念的最优反应，推广了全局博弈的选择结果。
Abstract Players receive a return to investment that is increasing in the proportion of others who invest and the state and incur a small cost for acquiring information about the state. Their information is reflected in a stochastic choice rule, specifying the probability of a signal leading to investment. If discontinuous stochastic choice rules are infinitely costly, there is a unique equilibrium as costs become small, in which actions are a best response to a uniform (Laplacian) belief over the proportion of others investing. Infeasibility of discontinuous stochastic choice rules captures the idea that it is impossible to perfectly distinguish states that are arbitrarily close together and is both empirically documented and satisfied by many natural micro-founded cost functionals on information. Our results generalize global game selection results and establish that they do not depend on the specific additive noise information structure.