An equality test across nonparametric regressions

A procedure for testing equality across nonparametric regressions is proposed. The procedure allows for any dimension of the explanatory variables and for any number of subsamples. We consider the case of random explanatory variables and allow the designs of the regressors and the number of observations to dier across subsamples. The division into subsamples is defined through a variable C which can be either fixed or random. In the case of a random C, our procedure is a general test of significance for qualitative variables in a nonparametric regression. In the case of a fixed C, our procedure provides a 'nonparametric analysis of covariance'. In both case, the test is a one-sided normal test and is consistent against all alternatives. We study its small sample behavior through Monte-Carlo simulations.