A negative binomial model and moment conditions for count panel data

This paper proposes some moment conditions associated with an appropriate specification of negative binomial model for count panel data, which is proposed by Hausman et al. (1984). The newly proposed moment conditions enable researchers to conduct the consistent estimation of the model under much weaker assumptions than those configured by Hausman et al. (1984). In some Monte Carlo experiments, it is shown that the GMM estimators using the new moment conditions perform well in the DGP configurations conforming to the specification above.