Improved GMM estimation of panel VAR models

Improved IV/GMM estimators for panel vector autoregressive models (VAR) are proposed. It is shown that the proposed IV estimator has the same asymptotic distribution as the bias corrected fixed effects estimator in the VAR(1) case when both the sample sizes of cross section and time series are large. Since the proposed estimator is simply to change the form of instruments, it is very easy to implement in practice. As applications of the proposed estimators, a panel Granger causality test and panel impulse response analysis in which the asymptotic distribution of generalized impulse response functions is newly derived are considered. Monte Carlo simulation results show that the proposed estimators have comparable or better finite sample properties than the conventional IV/GMM estimators using instruments in levels for moderate or long time periods. (C) 2015 Elsevier B.V. All rights reserved.