Seminar by Jeevant Rampal (IIM Ahmedabad)

In this paper we study a setting where players of a sequential-move game may have heterogeneous skill. Skill is captured by payoff responsiveness in quantal response models. Mckelvey and Palfrey (1998) provide the Quantal Response Equilibrium for extensive-form games (AQRE) where all players are assumed to have homogeneous skill. In this paper we extend the AQRE by modeling heterogeneous skill and uncertainty and belief-updating (BU) about opponents’ skills. First, we provide an equilibrium model incorporating skill-heterogeneity and uncertainty but not BU—this is called Heterogeneous AQRE (or HAQRE). Next, we incorporate naive disequilibrium belief-updating (BU) to define the HAQRE-BU. We show that these concepts exist, and in the context of finite perfect information games, they are unique, and they yield simple data applicability without fixed-point calculations. We use experimental data from a sequential-move game where players with different experience-levels interacted (Rampal (2020)) to show that modeling heterogeneity and belief updating about skills can each yield better data-fit in such settings.