Continuous Revision Games (joint with Akitada Kasahara)

東京大学大学院経済学研究科 学術交流棟(小島ホール)
1階 セミナー室

飯島良太 氏
Harvard University

We embed a one-shot strategic interaction in continuous-time models with a preparation phase. During the preparation phase before the final payoffs realize, players can make flexible and frequent adjustments. In the first model, imperfect monitoring model, players observe public signals about opponents' adjustment. In the second model, public noise model, players are subject to commonly observable noises that disturb game payoffs. We show that introducing such noise and adjustment cost, even if arbitrarily small, leads to a unique equilibrium. The proof is based on backward stochastic differential equations. Partial differential equations that describe players' equilibrium strategy are derived. They are generally applicable to a variety of economic settings and yield analytical solutions in special cases. Furthermore, we use our model to analyze equilibrium selection problem in potential games, and prove that the potential maximizer is selected as frictions vanish.