Ensayos sobre modelos econométricos sujetos a selección muestral y endogeneidad en el campo de la economía de la salud y del trabajo

Abstract

Fields like that of Labor Economics and Health economics have seen frequently concerned by econometric problems associated these types of models for his peculiarities. The method of ordinary least squares allows to obtain estimations of the effects as well as an intuitive interpretation and the consistent statistical inference. Nevertheless, this method is based on very rigid assumptions that not always are supported. Discrete choice models and with censured dependent variable have been widely studied in the last years trying to provide them more flexibility related with real cases, usually subject to problems of selection or endogeneity. So that, in this thesis we are concerned with the specification and estimation of this type of models, which if they are not correctly treated provide us inconsistent estimations. Concretely, it is usual that this type of models suffer from selection bias if the population of interest is not chosen in a random way. In this direction, a lot of studies have focused on extending the popular work of estimation in two stages of Heckman (1979). The above mentioned approach consists in obtaining consistent estimators of the equation of selection in the first stage and thus in a second phase, to estimate the parameters of the structural equation including a function of correction with the first stage estimators. From this popular approach, based on a context totally parametric, diverse authors have tried to adapt the necessary assumptions which are considered too restrictive. The problems of selection and endogeneity have special interest in this thesis, in which some methods are proposed to deal with the above mentioned problems. These problems have special relevancy in empirical problems of labor offer of certain special groups as persons with disability, married women or civil servants. In this context, the motivation of this thesis is twofold. On the one hand, it is about developing new estimation techniques consistent with the presence of sample selection and endogeneity. On the other hand, it is based on the application of some proposals able of handling with problems previously highlighted and appliying it to fields of interest in the Economy of our country.