Detección de instrumentos débiles al corregir endogeneidad en modelos logit binarios

Abstract

An instrumental variable is weak when it is not sufficiently correlated with the endogenous variable, precluding the correction of endogeneity. This problem has been studied only for linear models. In this paper we use Monte Carlo simulations to study the detection of weak instruments in binary Logit models, and to analyze their impact in practice. We conclude that, like in linear models, the use weak instruments precludes the correction of endogeneity in this context, and that weakness could be detected contrasting an F-test with critical values that are very similar to those that have been suggested for linear models.