- A NONPARAMETRIC TEST FOR SERIAL INDEPENDENCE OF REGRESSION ERRORS.
A test for serial independence of regression errors is proposed that is consistent in the direction ofserial dependence alternatives of first order. The test statistic is a function of aHoeffding-Blum-Kiefer-Rosenblatt type of empirical process, based on residuals. The resultantstatistic converges, surprisingly, to the same limiting distribution as the corresponding statisticbased on true errors.