Unit roots and multiple structural breaks in real output: How long does an economy remain stationary?

Abstract

Utilizing resampling methods, we present evidence on the rejection probabilities for difference-stationary and trend-stationary models for Mexico's real and real per capita annual gross domestic product. The trend stationary alternative allows for stationary fluctuations around a long-run trend function with endogenously determined multiple structural breaks, via global and sequential search methods. The number of breaks is determined using a unit-root rejection stopping rule and a parameter-constancy stopping rule.