Description:

Lorenzo Magnolfi (UW Madison): “Market Counterfactuals with Nonparametric Supply: An ML/AI Approach,” joint with Harold Chiang, Jack Collison, and Chris Sullivan

Abstract: This paper develops a flexible approach to perform market counterfactuals using
machine learning methods and nonparametric structure from economics. While stan-
dard structural methods rely on restrictive assumptions about firm conduct and cost,
we propose a data-driven framework that relaxes these constraints when rich market
data are available. Building on the identification results of Berry and Haile (2014), we
develop a nonparametric model of supply that nests traditional conduct specifications
while allowing for more complex competitive interactions. We adapt the Variational
Method of Moments (VMM) (Bennett and Kallus, 2023) to estimate this flexible model,
addressing the endogeneity of market shares and the high dimensionality of the prob-
lem. Our approach enables a wide range of counterfactual exercises including tax
policy analysis, product regulation, and merger simulation. Monte Carlo simulations
demonstrate that our method substantially outperforms standard approaches; applied
to the American Airlines-US Airways merger, our method produces more accurate
post-merger price predictions.