Fixed Effects Likelihood Approach for Large Panels

Abstract

This paper reviews and explains the techniques used in Hahn and Newey (2004) and Fernandez-Val and Weidner (2016) to derive the limit distribution of the fixed effects estimator of semiparametric panels when the time dimension tends to infinity jointly with the cross-section dimension. The techniques of these two papers are representative and understanding their working mechanism would be a good start point. Under a unified framework, this paper explicitly points out the difficulties in extending from models with fixed dimensional parameter space to panels with individual effects and from panel with individual effects to panel with both individual and time effects, and how Hahn and Newey (2004) and Fernandez-Val and Weidner (2016) solve them.

Keywords: panel data, nonlinear model, dynamic model, fixed effects, incidental parameters.

JEL Classification: C13, C23