Some Applications of Semi-Nonparametric Maximum Likelihood Estimation
Bas van der Klaauw Ruud H. Koning
October 1996
Abstract
An alternative to estimation of microeconometric models under the assumption
of normality of the distribution of the disturbances is semi-nonparametric
maximum likelihood estimation. In a particular class of this kind of models,
the density function of the disturbances is approximated by a Hermite series.
In this paper we will discuss this approach in the context of a popular
microeconometric model (the sample selection model) and we apply the model
to a truncated switching regression model with endogenous regimes. A new
choice of base functions of the Hermite series is presented and the semi-nonparametric
approach is used to examine sensitivity to the assumption of normality
of estimation results of a model for rent assistance and housing demand
in Koning (1995).