Testing for a unit root with covariates against nonlinear alternatives

Abstract

This paper proposes a new procedure for testing the unit root null against stationary but nonlinear alternatives. This test can be viewed as a generalization of the one developed by Kapetanios et al. (2003) (the KSS test) by incorporating stationary covariates. The asymptotic distribution of the test is derived and the asymptotic critical values are tabulated. A set of Monte Carlo simulations show that our test generally achieves large power improvements over the KSS test. An illustrated empirical application indicates that our proposed test is able to unveil more evidence than the KSS test in favor of no unit root of real exchange rates in 15 Asian countries.