Bias-corrected estimation in dynamic panel data models

This study develops a new bias-corrected estimator for the fixed-effects dynamic panel data model and derives its limiting distribution for fixed T and N large. The bias-corrected estimator is derived as a bias correction of the least-squares dummy variable (within) estimator. It does not share some of the drawbacks of recently developed IV and GMM estimators and is relatively easy to compute. Monte Carlo experiments provide evidence for the bias-corrected estimator to perform well even in small samples. The paper contains an application to a model of unemployment dynamics at the U.S. state level for the 1991-2000 period.