A simple estimator for partial linear regression with endogenous nonparametric variables

We propose a simple kernel estimator for semiparametric partial linear models with endogeneity in the nonparametric function. Compared to the existing backfitting estimator, our estimator is notationally simpler and relatively easier to implement. We also discuss data-driven bandwidth selection to implement this estimator in practice. Monte Carlo exercises show that the finite sample performance of these two estimators is similar.