Linear dynamic panel data models

Abstract

This chapter reviews the econometric literature on the estimation of linear dynamic panel data models. Frequently used in applied economics research, the estimation of these models is typically by generalized method of moments estimators which face several challenges particular to this context, including weak instruments and many moments. This chapter outlines possible solutions to these challenges and then discusses bias correction methods for fixed effects estimation that yield estimates that work well in moderately long panels. Lastly, it presents the motivation behind the use of dynamic panel data models, with particular emphasis on the importance of distinguishing between unobserved heterogeneity and state dependence. The chapter concludes with an empirical example of a relationship between economic environment and democracy estimated using Stata.