Potential Pitfalls in Determining Multiple Structural Changes with an Application to Purchasing Power Parity

Tests for structural change play an important role in macroeconomics and international finance. We investigate the empirical performance of the Bai and Perron (1998) multiple structural change tests and show that the use of their critical values may cause severe size distortions in persistent series. To correct these size distortions, we implement the Bai (1999) structural change test, where we calculate bootstrap critical values. While the size is correct, his sequential method lacks power on data that includes breaks of opposite sign. We extend this test in two directions. First, we propose a new procedure to choose the number of breaks. Second, we develop a restricted version that specifically models breaks that imply mean or trend reversion. The simulation results show good power and satisfactory size of the new procedures. Using long-run real exchange rates, we illustrate the practical importance of these results. We investigate contradictory evidence of purchasing power parity from the application of unit root and structural change tests. We argue that the Bai and Perron test is the culprit and instead use the Bai test along with our newly proposed methodologies. Consequently, we reconcile the results of unit root tests with those of tests for structural change when testing for purchasing power parity