Oracle Inequalities for High-Dimensional Panel Data Models

Abstract

This paper is concerned with high-dimensional panel data models where the number of regressors can be much larger than the sample size. Under the assumption that the true parameter vector is sparse we propose a panel-Lasso estimator and establish finite sample upper bounds on its estimation error under two different sets of conditions on the covariates as well as the error terms. In particular, we allow for heteroscedastic and non-gaussian error terms which are weakly dependent over time. Upper bounds on the estimation error of the unobserved heterogeneity are also provided under the assumption of sparsity. Next, we show that our upper bounds are essentially optimal in the sense that they can only be improved by multiplicative constants. These results are then used to show that the Lasso can be consistent in even very large models where the number of regressors increases at an exponential rate in the sample size. Conditions under which the Lasso does not discard any relevant variables asymptotically are also provided. In the second part of the paper we give lower bounds on the probability with which the adaptive Lasso selects the correct sparsity pattern in finite samples. These results are then used to give conditions under which the adaptive Lasso can detect the correct sparsity pattern asymptotically. We illustrate our finite sample results by simulations and apply the methods to search for covariates explaining growth in the G8 countries.