PARTIALLY LINEAR VARYING-COEFFICIENT PANEL DATA MODELS WITH FIXED EFFECTS

Wu Qifeng (P. R. China), Luo Xianhua (P. R. China) and Li Yuan (P. R. China)

Received January 17, 2008

Abstract

There is a rich literature on nonparametric and semiparametric estimations of panel or longitudinal data. However, few studies focus on the nonparametric and semiparametric panel models with fixed effects. In this paper, we consider semiparametric partially linear varying-coefficient models for panel data with fixed effects. We propose consistent estimators for both the parametric and nonparametric components in the models based on profile likelihood and local linear smoothing method. Simulation shows that our estimators behave quite well.

Keywords and phrases: partially linear varying-coefficients models, panel data, profile likelihood, local linear regression, fixed effects.