Testing Linear Regression Models in Non Regular Case

Abstract

We propose a new statistic for testing linear hypotheses in the non parametric regression model in the case of a homoscedastic error structure and fixed design. In contrast to most models suggested in the literature, our procedure is applicable in the non parametric model case without regularity condition, and also under either the null or the alternative hypotheses. We show the asymptotic normality of the test statistic under the null hypothesis and the alternative one. A simulation study is conducted to investigate the finite sample properties of the test with application to regime switching.

Keywords:

• Linear hypothesis
• Non linear regression
• Non parametric regression
• Regime switching.

Mathematics Subject Classification:

• 62G05, 62G10, 62G20, 62J02