Estimating Parameters of Structural Models Using Neural Networks

51 Pages Posted: 9 Dec 2019 Last revised: 5 Oct 2022

See all articles by Yanhao 'Max' Wei

Yanhao 'Max' Wei

University of Southern California - Marshall School of Business

Zhenling Jiang

University of Pennsylvania - The Wharton School

Date Written: October 1, 2022

Abstract

Machine learning algorithms are increasingly used in marketing to extract information from unstructured data or model flexible relations between variables in data. We explore an alternative use of machine learning algorithms. We train a neural net to provide parameter estimate for a given (structural) econometric model, e.g., discrete choice, consumer search. The training examples consist of datasets generated by the econometric model under a range of parameter values. The neural net takes the moments of a dataset as input and tries to recognize the parameter value underlying that dataset. The neural net can also be trained to provide the statistical accuracy of its parameter estimate. We establish that this neural net estimator (NNE) converges to limited-information Bayesian posterior when the number of training datasets is sufficiently large. We find that NNE is robust to redundant moments. We apply NNE in a consumer sequential search application. It gives accurate and robust estimates at a light computational cost, compared to the prevailing estimation approach. We discuss, more broadly, which applications are suitable (and unsuitable) for NNE.

Keywords: neural networks, machine learning, structural estimation, redundant moments, simulation burden, sequential search

Suggested Citation

Wei, Yanhao and Jiang, Zhenling, Estimating Parameters of Structural Models Using Neural Networks (October 1, 2022). USC Marshall School of Business Research Paper, Available at SSRN: https://ssrn.com/abstract=3496098 or http://dx.doi.org/10.2139/ssrn.3496098

Yanhao Wei (Contact Author)

University of Southern California - Marshall School of Business ( email )

701 Exposition Blvd
Los Angeles, CA California 90089
United States

Zhenling Jiang

University of Pennsylvania - The Wharton School ( email )

3641 Locust Walk
Philadelphia, PA 19104-6365
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
551
Abstract Views
2,134
rank
74,689
PlumX Metrics