A1043
Title: Inference in high-dimensional two-way panel data models
Authors: Alexandra Soberon - Universidad de Cantabria (Spain)
Lindes Dominguez Diaz - Universidad de Cantabria (Spain)
Juan Manuel Rodriguez-Poo - Universidad de Cantabria (Spain) [presenting]
Abstract: The aim is to develop a consistent and asymptotically normal estimator for a triangular simultaneous two-way high-dimensional panel data model. The estimator addresses endogeneity arising from both individual and time effects, as well as the dependence between covariates and error terms. A two-stage procedure is proposed: First, removing fixed effects; second, applying instrumental variable estimation using regularization methods (lasso, cluster-lasso, and post-lasso) suitable for high-dimensional settings. Monte Carlo simulations demonstrate the estimator's favorable properties in terms of bias and RMSE, particularly when the regularization parameter is selected via cross-validation. Theoretical properties and practical implementation details are discussed, and the performance of the estimator is evaluated through extensive simulation studies.
