Estimation of Finite Sequential Games

Abstract

I study the estimation of finite sequential games with perfect information. The major challenge in estimation is computation of high-dimensional truncated integration whose domain is complicated by strategic interaction. I show that this complication resolves when unobserved off-the-equilibrium-path strategies are controlled for. Separately evaluating the likelihood contribution of each subgame perfect strategy profile that rationalizes the observed outcome allows the use of the GHK simulator, the most widely used importance-sampling probit simulator. Monte Carlo experiments demonstrate the performance and robustness of the proposed method, and confirm that misspecification of the decision order leads to underestimation of strategic effect.

Keywords: Inference In Discrete Games, Sequential Games, Monte Carlo Integration, GHK Simulator, Subgame Perfection, Perfect Information

JEL Classification: C35, C63, C72