A SPECIFICATION TEST FOR NONPARAMETRIC INSTRUMENTAL VARIABLE REGRESSION

We consider testing for correct specification of a nonparametric instrumental variable regression. In this ill-posed inverse problem setting, the test statistic is based on the empirical minimum distance criterion corresponding to the conditional moment restriction evaluated with a Tikhonov Regularized estimator of the functional parameter. Its asymptotic distribution is normal under the null hypothesis, and a consistent bootstrap is available to get simulation based critical values. We explore the finite sample behavior with Monte Carlo experiments. Finally, we provide an empirical application for an estimated Engel curve.