Estimating Discrete Games of Complete Information: Bringing Logit Back in the Game

TL;DR

This paper introduces efficient estimation methods for discrete games of complete information, leveraging logit assumptions to simplify computation and improve speed while maintaining informative results.

Contribution

It develops convex conditional moment inequalities for unordered and ordered-action games, enabling faster and more tractable estimation without enumerating equilibria.

Findings

Generated informative identified sets

Achieved several orders of magnitude speedup

Validated methods with empirical examples

Abstract

Estimating discrete games of complete information is often computationally difficult due to partial identification and the absence of closed-form moment characterizations. This paper proposes computationally tractable approaches to estimation and inference that remove the computational burden associated with equilibria enumeration, numerical simulation, and grid search. Separately for unordered and ordered-actions games, I construct an identified set characterized by a finite set of generalized likelihood-based conditional moment inequalities that are convex in (a subvector of) structural model parameters under the standard logit assumption on unobservables. I use simulation and empirical examples to show that the proposed approaches generate informative identified sets and can be several orders of magnitude faster than existing estimation methods.