On LASSO Inference for High Dimensional Predictive Regression

Prof. Zhentao SHI
Associate Professor,
Chinese University of Hong Kong

Date: 15 October 2024 (Tuesday)
Time: 10:30 – 12:00
Venue: E22-G008
Host: Prof. Yi DING, Assistant Professor in Business Economics

Abstract

While conducting regression analysis, the LASSO method introduces shrinkage bias, which can adversely affect the desirable asymptotic normality of the estimator. The desparsified LASSO has emerged as a well-known remedy for this issue. In the context of high-dimensional predictive regression, though, the desparsified LASSO faces an additional challenge: the Stambaugh bias arising from nonstationary regressors. To restore standard inferential procedure based on the t-statistic, we propose a novel estimator called IVX-desparsified LASSO (XDlasso). XDlasso eliminates both the shrinkage bias and the Stambaugh bias simultaneously, even when we lack prior knowledge about the identities of nonstationary and stationary regressors. We establish the asymptotic properties of XDlasso for hypothesis testing, and our theoretical findings are supported by Monte Carlo simulations. Applying our method to real-world data from the FRED-MD database — which includes a rich set of control variables — we investigate two important questions: (i) the predictability of U.S. inflation using the unemployment rate, and (ii) the predictability of U.S. stock returns based on the earnings-price ratio.

Speaker

Prof. Zhentao SHI is Associate Professor from Department of Economics, the Chinese University of Hong Kong. He obtained Ph.D. degree from Yale University, and was Associate Professor at the Georgia Institute of Technology. His research focuses on estimation and inference of machine learning methods for economic and financial big data. He is a recipient the National Science Fund for Distinguished Young Scholars, and he serves as an Associate Editor of the Journal of Business & Economic Statistics.

 

All are welcome!