Parametric and semiparametric multivariate sample selection models estimators’ accuracy: Comparative analysis on simulated data

Abstract

This article is devoted to the comparative analysis of parametric and semiparametric sample selection models with two selection equations. Comparison has been conducted on simulated data under different random errors distributional assumptions: student, beta and mixture of normal. The results suggest that for student and beta distributions parametric models’ estimates are more or equally accurate as semiparametric. However, former methods provide more accurate estimates under mixture distribution case. Therefore, parametric sample selection model estimators seem to be robust to violations of normality assumption in terms of tails thickness and asymmetry but fail to account for bimodality as good as their semiparametric counterparts