Semiparametric estimation of duration models when the parameters are subject to inequality constraints and the error distribution is unknown
2017-06-08T01:13:37Z (GMT) by
The parameters in duration models are usually estimated by a Quasi Maximum Likelihood Estimator [QMLE]. This estimator is efficient if the errors are iid and exponentially distributed. Otherwise, it may not be the most efficient. Motivated by this, a class of estimators has been introduced by Drost and Werker (2004). Their estimator is asymptotically most efficient when the error distribution is unknown. However, the practical relevance of their method remains to be evaluated. Further, although some parameters in several common duration models are known to be nonnegative, this estimator may turn out to be negative. This paper addresses these two issues. We propose a new semiparametric estimator when there are inequality constraints on parameters, and a simulation study evaluates the two semiparametric estimators. The results lead us to conclude the following when the error distribution is unknown: (i) If there are no inequality constraints on parameters then the Drost-Werker estimator is better than the QMLE, and (ii) if there are inequality constraints on parameters then the estimator proposed in this paper is better than the Drost-Werker estimator and the QMLE. In conclusion, this paper recommends estimators that are better than the often used QMLE for estimating duration models.