Half-panel jackknife estimation for dynamic panel models
This paper extends the half-panel jackknife (HPJ) estimator to GMM models with fixed effects. The Monte Carlo results show that the HPJ significantly reduces finite-sample bias for both the difference and system GMM estimators of the dynamic panel model.