Taste variation in discrete choice models

摘要: This paper develops an extension of the classical multinomial logit model which approximates a class models obtained when there is uncontrolled taste variation across agents and choices in addition to stochastic noise inherent model. Unlike semiparametric parametric alternatives, extended easy estimate even are many potential choices. it does not require specification distribution varying tastes. The can give quick indication impact on estimates generates covariances shifters. It be used as exploratory device en route construction incorporating particular form random determine whether such effort required at all. When amount excessive approximate adequate itself. nests conventional leads misspecification diagnostic. A method for estimating using software proposed, asymptotic properties estimators derived application presented.