Bias-corrected method of moments estimators for dynamic panel data models

A computationally simple bias correction for linear dynamic panel data models is proposed and its asymptotic properties are studied when the number of time periods fixed or tends to infinity with units. The approach can accommodate both fixed-effects random-effects assumptions, heteroskedastic errors, as well higher-order autoregressive models. Panel-corrected standard errors that allow robust inference in cross-sectionally correlated errors. Monte Carlo experiments suggest under assumption strictly exogenous regressors bias-corrected method moment estimator outperforms popular GMM estimators terms efficiency correctly sized tests.