Title: A semiparametric panel data model with common factors and spatial dependence
Authors: Alexandra Soberon - Universidad de Cantabria (Spain) [presenting]
Juan Manuel Rodriguez-Poo - Universidad de Cantabria (Spain)
Antonio Musolesi - University of Ferrara (Italy)
Abstract: New semiparametric heterogeneous panel data models are proposed which handle complex and relevant empirical problems, simultaneously: (i) functional misspecification by modelling stochastic observed common factors with a nonparametric function instead of assuming the usual parametric form; (ii) cross-sectional dependence originated simultaneously from common factors and spatial dependence, from the latter neither imposing a specific parametric spatial diffusion process nor requiring the specification of a given interaction matrix, but being directly derived from the data; iii) heterogeneous relations. We first propose a new estimation that extends the common correlated effect (CCE) approach to such a semiparametric spatially augmented framework. Then, Generalized Least Squares (GLS)-type estimators improving efficiency are proposed by taking into account the dependence structure. Asymptotic normal distributions are derived when the time dimension is large while the cross-sectional dimension need not be. Small sample properties of the estimators are investigated by Monte Carlo experiments, and an empirical application on the knowledge capital production function is conducted.