Tuesday, 1 December, 2026 | 14:15 | Room 6 | Micro Theory Research Seminar

Lea Nagel (Harvard University) "Algorithm-Robust Market Design"

Lea Nagel, Ph.D.

Harvard University, United States


Authors: Lea Nagel, Martino Banchio and Roberto Saitto

Abstract: We propose a model of batched algorithmic learning for Bayesian games, such as online ad auctions. We show that empirical play converges to the set of strategic-form correlated equilibria. Moreover,  every such equilibrium can be reached. Our main application shows that first-price auctions have a unique strategic-form correlated equilibrium, and hence robustly extract optimal revenues from algorithmic bidders. We also apply our framework to analyze regulatory interventions for third-party pricing algorithms. Preventing information sharing ensures prices no higher than in the worst Bayesian equilibrium.