Semiparametric Estimation of a Heteroskedastic Sample Selection Model

This paper considers estimation of a sample selection model subject to conditionalheteroskedasticity in both the selection and outcome equations. The formof heteroskedasticity allowed for in each equation is multiplicative, and each ofthe two scale functions is left unspecified. A three-step estimator for the parametersof interest in the outcome equation is proposed. The first two stages involvenonparametrice stimation of the "propensitys core" and the conditional interquartilerange of the outcome equation, respectively. The third stage reweights thedata so that the conditional expectation of the reweighted dependent variable isof a partially linear form, and the parameters of interest are estimated by an approachanalogous to that adopted in Ahn and Powell (1993, Journal of Econometrics58, 3-29). Under standard regularity conditions the proposed estimator isshown to be V/--consistent and asymptotically normal, and the form of its limitingcovariance matrix is derived