Indirect Inference Estimation of a First-Order Dynamic Panel Data Model

A new estimator is proposed in this paper that inverts a binding function that is based on the asymptotic bias of the within-groups (WG) estimator in a first-order dynamic panel model with fixed effects under the asymptotic regime of large N and finite T. The new estimator is in the spirit of indirect inference (II) and corrects the inconsistency of the WG estimator. It is simulation-free, does not rely on any distributional assumption on the idiosyncratic error term, and is asymptotically normally distributed. Monte Carlo results indicate its good finite-sample performance in comparison with other consistent estimators.