- June 2016
Digital Object Identifier (DOI)
International Standard Serial Number (ISSN)
Electronic International Standard Serial Number (EISSN)
- This article proposes testing the hypothesis of a uniformly non-positive nonparametric regression function using a test statistic with tabulated critical values. The null hypothesis is characterized in terms of the significance of a parameter, which measures a distance from the double-integrated regression function to the class of concave functions. The test statistic is a suitably scaled parameter estimate, which does not require smooth estimation of the underlying regression and/or the conditional variance functions. The finite sample performance of the proposed test is studied by means of two Monte Carlo experiments, showing that the proposed method compares favorably to existing procedures. (C) 2015 Elsevier B.V. All rights reserved.
- nonparametric testing; conditional inequalities; least concave majorant; nonparametric regression; order statistics; iinference; variance; dfference; models