Halfline tests for multivariate one-sided alternatives

Abstract

Halfline tests studied in this paper are t type tests for testing inequality constraints under the alternative hypothesis. An appealing example of such tests in the literature is to find a halfline in the restricted parameter space such that the resultant test is most stringent in terms of the minimization of the maximum shortcoming. However, there appears to be no generally applicable procedure available for implementing this test. This paper is to fill this gap. We also propose a halfline test which has a computational advantage. Simulation studies are conducted to compare the finite sample performance of halfline tests against some existing tests. The results of our simulation studies suggest that halfline tests can have a better finite sample power property and are more robust against the normality assumption compared to likelihood ratio-based tests.