Title: Powerful self-normalizing tests for stationarity
Authors: Uwe Hassler - Goethe University Frankfurt (Germany) [presenting]
Mehdi Hosseinkouchack - Goethe University Frankfurt (Germany)
Abstract: A family of tests for stationarity against a unit root is proposed. It builds on the Karhunen-Loeve expansions of the limiting CUSUM processes under the null hypothesis and local alternatives. The test statistic becomes a ratio of quadratic forms of q weighted sums such that the nuisance long-run variance cancels asymptotically without having to be estimated. Critical values can be calculated by standard numerical means. Monte Carlo experiments show that q may not be too large in finite samples to obtain a test with correct size under the null. At the same time our test is more powerful than classical competitors that are not self-normalizing.