Endog?n?it? dans les mod?les qualitatifs sur donn?es longitudinales

Statistical methods taking into account endogenous variables are now available with longitudinal data if the explanatory variables are continuous. But few papers deal with qualitative data. The paper proposes a method to estimate a model in which the female decision of activity depends on a qualitative endogenous explanatory variable (the number of children). Rather than a full parametric framework, the paper proposes a three-step method with less specification restrictions and more convenient to implement with panel data. Conditional on usual explanatory variables, the endogeneity of the number of children is rejected, as in cross-section. Moreover, panel data allow the identification of the effects of time-varying variables, as the partner's income.