Models with Fixed Effects

This paper analyzes the second order bias of instrumental variables estimators for a dynamic panel model with fixed effects. Three different methods of second order bias correction are considered. Simulation experiments show that these methods perform well if the model does not have a root near unity but break down near the unit circle. To remedy the problem near the unit root a weak instrument approximation is used. We show that an estimator based on long differencing the model is approximately achieving the minimal bias in a certain class of instrumental variables (IV) estimators. Simulation experiments document the performance of the proposed procedure in finite samples.