Empirical models of discrete games

This paper develops econometric models for discrete games. Specifically, we model the payoffs of games where a researcher observes qualitative or censored information about agents' decisions and payoffs. These models extend single-person qualitative choice models introduced by McFadden (1974) and others to multiple-person choice problems. The equations describing players' equilibrium strategies depend on the game's structure and the equilibrium solution concept. We show that one can describe the equilibria of a simultaneous-move Nash game with a linear system of dummy endogenous variables. We also show that sequential-move and cooperative models have different, but related, econometric structures. A series of applied examples address identification and estimation issues. These examples include models of market entry, technology adoption, tax auditing, and cooperative family labor supply.