A1662
Title: Empirical Likelihood based inference for varying coefficient panel data models with cross-sectional dependence
Authors: Luis Antonio Arteaga Molina - Universidad de Cantabria (Spain) [presenting]
Abstract: Local empirical likelihood based inference for semiparametric varying coefficient panel data models with cross-sectional dependence is investigated. A common factor structure is used to characterize the cross-sectional dependence. First, it is shown that the naive empirical likelihood ratio for the individual-specific and pooled varying coefficients are asymptotically standard chi-squared distributed when undersmoothing is employed. To avoid undersmoothing, bias-corrected empirical likelihood ratios that also have standard chi-squared limit distributions are presented. As a by-product, the maximum empirical likelihood estimators for the individual-specific and pooled varying coefficients are derived and their asymptotic distributions are obtained. The main advantage of empirical likelihood is that, unlike normal approximation, the plug-in estimator of the variance matrix is not needed. The feasibility of the technique and its small sample properties are demonstrated through a Monte Carlo simulation exercise and an empirical application.
