作者: Douglas Rivers , Quang H. Vuong
DOI: 10.1016/0304-4076(88)90063-2
关键词:
摘要: A two-step maximum likelihood procedure is proposed for estimating simultaneous probit models and compared to alternative limited information estimators. Conditions under which each estimator attains the Cramer–Rao lower bound are obtained. Simple tests exogeneity based on new shown be asymptotically equivalent one another have same local asymptotic power as classical estimator. Finite sample comparisons between estimators presented some Monte Carlo evidence. The performance of also assessed.