Estimation of a semiparametric transformation model in the presence of endogeneity

Résumé

We consider a semiparametric transformation model, in which the regression func- tion has an additive nonparametric structure and the transformation of the response is assumed to belong to some parametric family. We suppose that endogeneity is present in the explanatory variables. Using a control function approach, we show that the pro- posed model is identified under suitable assumptions, and propose a profile estimation method for the transformation. The proposed estimator is shown to be asymptotically normal under certain regularity conditions. A simulation study shows that the esti- mator behaves well in practice. Finally, we give an empirical example using the U.K. Family Expenditure Survey.