Monotone Perfection

报告题目：Monotone Perfection

报告人：徐韩萍（新加坡国立大学）

摘要：This paper proposes a new equilibrium concept called ``perfect monotone equilibrium'' in Bayesian games, which strengthens the standard notion of monotone equilibrium by accounting for the possibility that players might choose unintended strategies due to a trembling hand. A perfect monotone equilibrium satisfies the important property of admissibility in Bayesian games with finitely many actions, and the property of limit undominatedness in settings with infinitely many actions. In a general class of Bayesian games where each player's action set is a sublattice of multi-dimensional Euclidean space and players' types are also multi-dimensional, a perfect monotone equilibrium is shown to exist under the supermodularity and increasing differences conditions. These conditions model the scenarios in which, informally, players' payoffs exhibit complementarity in their own actions and monotone incremental returns in their own types. To illustrate practical relevance, we extend the analysis to Bayesian games with discontinuous payoffs, including application such as first-price auctions, all-pay auctions, and Bertrand competitions. Our result further generalizes to the setting with more general action spaces and type spaces.

时间：3月27日（星期四），15:30-17:00

地点：明德楼B201-1