UNIT ROOTS AND MULTIPLE STRUCTURAL BREAKS IN REAL OUPUT: HOW LONG DOES AN ECONOMY REMAIN STATIONARY?

In recent literature on multiple structural change, the number of breaks is determined through a sequential test of parameter constancy. This paper explores the possibility of determining the number of breaks in a time series by relating structural breaks to the behavior of unit roots. Thus, rather than using the rule: stop adding breaks when the parameter variation form of nonstationarity is rejected, we examine the rule: stop adding breaks when the unit root form of nonstationarity is rejected. We use Monte Carlo studies to compare the performance of these two rules and find that the unit-root rule performs better for lower values of the autoregressive parameter. We illustrate the techniques in determining the number of breaks in Mexican real and real per-capita GDP series utilizing resampling methods.