A0555
Title: Robust estimation and inference for high-dimensional panel data models
Authors: Jiti Gao - Monash University (Australia)
Fei Liu - Nankai University (China)
Bin Peng - Monash University (Australia)
Yayi Yan - Shanghai University of Finance and Economics (China) [presenting]
Abstract: The relevant literature is provided with a complete toolkit for conducting robust estimation and inference about the parameters of interest involved in a high-dimensional panel data framework. Specifically, (1) non-Gaussian, serially and cross-sectionally correlated and heteroskedastic error processes are allowed for, (2) an estimation method is developed for a high-dimensional long-run covariance matrix using a thresholded estimator, and (3) the number of regressors is also allowed to grow faster than the sample size. Methodologically and technically, two Nagaev types of concentration inequalities are developed: One for a partial sum and the other for a quadratic form, subject to a set of easily verifiable conditions. Leveraging these two inequalities, a non-asymptotic bound is derived for the LASSO estimator, achieving asymptotic normality via the node-wise LASSO regression, and establishing a sharp convergence rate for the thresholded heteroskedasticity and autocorrelation consistent (HAC) estimator. The practical relevance of these theoretical results is demonstrated by investigating a high-dimensional panel data model with interactive effects. Moreover, extensive numerical studies are conducted using simulated and real data examples.
