Abstract We propose a new inference method in high-dimensional regression models and high-dimensional IV regression models. The method is shown to be valid without requiring the exact sparsity or Lp sparsity conditions. Simulation studies demonstrate superior performance of this proposed method over those based on LASSO or random forest, especially under less sparse models. We illustrate an application to production analysis with a panel of Chilean firms.

Abstract We propose improved standard errors and an asymptotic theory for two-way clustered panels. Our theory allow for arbitrary serial dependence in the common time effects, which is excluded by existing two-way methods. Our asymptotic distribution theory is the first which allows for this level of inter-dependence. Under weak conditions, we demonstrate that OLS is asymptotically normal, our proposed variance estimator is consistent, and t-ratios are asymptotically standard normal. The results extend to two-way fixed-effect models; we argue that two-way clustering is still necessary even if two-way fixed effects are included. Simulation and empirical illustration are provided.