Semiparametric Varying Coefficient Models with Endogenous Covariates

Though parametric methods are popular in applied settings, practitioners often require nonparametric alternatives. However, fully nonparametric methods are known to suffer from the curse of dimensionality, which limits their practical application. Semiparametric methods occupy a middle ground, have the desirable feature that they are both exible, and provide an attractive alternative to fully nonparametric methods, while attenuating the curse-of-dimensionality. Traditional semiparametric methods, such as the popular `varying coefficient' specification, do not account for endogenous covariates, which restricts their application. In this paper we consider the estimation of semiparametric varying coeffcient models when the functional coeffcients may contain (continuous) endogenous covariates thereby extending the reach of this exible and powerful class of models.