Restricted structural change and the unit root hypothesis

In a classic paper, Nelson and Plosser could not reject the unit root hypothesis in favor of trend stationarity for 13 out of 14 long-term annual macro series. Subsequent studies, allowing for one or two structural changes, have found more rejections with a broken trend stationary alternative. Since these changes are defined to be permanent, the rejections do not provide evidence of trend stationarity. We propose new tests for a unit root in the presence of restricted structural change. Allowing for two offsetting structural changes, we reject the unit root null in favor of restricted trend stationarity for 6 out of 13 series.