A0283
Title: Estimation and inference of panel data models with a generalized factor structure
Authors: Juan Manuel Rodriguez-Poo - Universidad de Cantabria (Spain) [presenting]
Alexandra Soberon - Universidad de Cantabria (Spain)
Stefan Sperlich - University of Geneva (Switzerland)
Abstract: A new class of panel data models are introduced where unobserved factors and factor loadings are introduced in a nonlinear fashion. To estimate the parameters of interest in this class of models, a consistent and asymptotically normal root-NT estimator is proposed. The approach shares the same philosophy as the common correlated effects estimation technique: it removes the unknown factors by transforming the model, but our proposal covers a wider set of models. Since a conditional independence assumption is the vault key of the estimation procedure, a consistent specification test is also proposed to check whether this assumption is fulfilled or not. The test relies on combining the methodology of conditional moments tests and nonparametric estimation techniques. Using degenerate and nondegenerate theories of U-statistics, the test is asymptotically distributed under the null while it diverges under the alternative at a rate that is arbitrarily close to the square root of NT. The finite sample performance is evaluated by the new estimators and test statistics with simulated data examples, and a real empirical problem is provided about the effect of the EU ETS on the economic development of the EU countries.
