Covariate selection for testing purchasing power parity

This article employs Hansen's (1995) Covariate Augmented Dickey-Fuller (CADF) test to reexamine the issue of Purchasing Power Parity (PPP) using post-Bretton Woods exchange rate data for 20 industrialized countries. Instead of just using a single covariate as in the literature, we implement the test by using Bai and Ng's (2002) method to choose suitable stationary covariates. Our empirical results show that the real exchange rates generally display long-run mean reversion and support PPP, as contrasted with those obtained in Amara and Papell (2006) that less rejections of unit root hypothesis are made with the same test. The panel unit root test suggested by Chang (2004) is also performed to justify our results.