Asymptotic Analysis of Panel Regression Models with Unobserved Interactive Individual Effects

University Of Southern California, Los Angeles CA

Investigators

Abstract

This project studies a panel regression model where the individual fixed effects interact with common time specific effects. This model is sometimes known as a factor model. The main objective is to establish the general asymptotic theory of the (Gaussian) quasi maximum likelihood estimator and the three classical tests (Wald, likelihood ratio, and Lagrangian multiplier tests) of the panel regression model with interactive fixed effects when both the cross sectional dimension and the time dimension are large. This model will then be extended to the case where the dimension of the factors is not known. The project also investigates the dynamic panel regression model with factors when the dynamic coefficient is close to one. Broader Impact The factor model studied in the project has a wide range of economic applications. These include arbitrage asset pricing models in finance, wage equations in labor economics, and global economy models in international economics. While the proposed research does not directly address these empirical applications, we believe that the proposed research provides new and innovative econometric techniques that can be used by empirical researchers in these areas. The results will be disseminated through publications in high quality journals that are widely read in the profession. The investigator also plans to make these estimation and inference procedures available on his internet web-pages through well documented MATLAB programs. The programs can then be used by other researchers who want to apply these methods.