Skewness and fat tails in discrete choice models

In discrete choice models, the probability that the dependent variable will assume 0 or 1 depends on a set of explanatory variables through a function F. In this paper we propose the class of skew Student t distribution as a function F in order to have a more flexible model that can account simultaneously for asimmetry and thick tails and such that probit and logit models can be considered as special cases. Two examples illustrates the performance of the proposed model.