A1219
Title: Functional coefficient panel data models with endogenous selection: A flexible estimation approach
Authors: Alexandra Soberon - Universidad de Cantabria (Spain) [presenting]
Daniel Henderson - University of Alabama (United States)
Juan Manuel Rodriguez-Poo - Universidad de Cantabria (Spain)
Taining Wang - Capital University of Economics and Business (China)
Abstract: The purpose is to introduce an innovative estimation approach for functional coefficient panel data models that simultaneously addresses sample selection and fixed effects challenges. Nearest neighbor differencing of smoothing variables is exploited to eliminate fixed effects without restrictive identification assumptions. The proposed two-stage technique seamlessly integrates nonparametric selection modeling with flexible functional coefficient estimation. The first stage captures complex selection mechanisms using advanced nonparametric techniques, while the second stage employs a sophisticated generalized local weighting scheme that estimates primary relationships while purging selection bias. The approach operates under weak regularity conditions while achieving superior computational efficiency compared to existing methods. Asymptotic properties are established, and it is demonstrated through Monte Carlo simulations that the estimators consistently outperform alternatives, delivering substantial improvements in bias reduction and precision.
