Deep Learning for Solving Economic Models

This paper is based on a series of lectures delivered at Johns Hopkins University as a Distinguished Visiting Scholar. I am deeply grateful to the Department of Economics at Johns Hopkins for the invitation and invaluable feedback. Previous versions were presented at the Bank of Spain, Cambridge, ECB, Harvard, Northwestern, Oxford, Penn, Princeton, UC Santa Barbara, and Stanford. I thank all who offered comments on those occasions, the editor, David Romer, and the five referees. I also wish to thank my coauthors on the many papers I have written on machine learning in economics: Lorenzo Bretscher, Sara Casella, David Childers, Mahdi Ebrahimi Kahou, Kenneth Gillingham, Sebastián Gómez-Cardona, Stephen Hansen, Samuel Hurtado, Yilian Li, Joël Marbet, Galo Nuño, Jesse Perla, Omar Rachedi, Jan Rosa, Simon Scheidegger, Minchul Shin, Arnav Sood, George Sorg-Langhans, Maximilian Vogle, Le Xu, and Francesco Zanetti. Their collaboration has been instrumental in shaping my understanding of this topic. Naturally, all remaining errors are my own. The companion webpage www.sas.upenn.edu/~jesusfv/ deeplearning.html will provide regular updates to the main text and code, as well as slide decks for teaching the material covered in this survey. The views expressed herein are those of the author and do not necessarily reflect the views of the National Bureau of Economic Research.

• September 11, 2025