Tests for cointegration allowing for an unknown number of breaks

This paper introduces cointegration tests allowing for an unknown number of breaks. The introduced tests assume that the unspecified number of breaks is smaller than or equal to the maximum number of breaks set a priori. Monte Carlo simulations provide two main results. First, the proposed tests perform as well as the tests of Gregory and Hansen (1996a) and Hatemi-j (2008), which assume one or two breaks a priori, when the cointegration relationship has one or two breaks. Second, the proposed tests perform better than the tests of Gregory and Hansen (1996a) and Hatemi-j (2008) when the cointegration relationship has more than three breaks or persistent Markov switching shifts. We also provide empirical applications for the money demand of the U.S. The empirical results show that the proposed tests reject the null hypothesis of no cointegration as compared to other tests. (C) 2012 Elsevier B.V. All rights reserved.