The within or least squares dummy variable estimator is severely biased in homogeneous dynamic panel models with moderate T. We present a bias correction for this estimator based on an iterative bootstrap procedure. Monte Carlo simulations show that this procedure is a good alternative for the...

The within-groups estimator is inconsistent in dynamic panels with fixed T since the sample mean used to eliminate the individual effects from the lagged dependent variable is correlated with the error term. This paper suggests to eliminate individual effects from an AR(1) panel using backward...