Asymmetric unit root tests in the presence of structural breaks under the null

Using Monte Carlo methods, the behaviour of the momentum threshold autoregressive (MTAR) unit root test of Enders and Granger (1998) is examined in the presence of structural breaks under the null. It is found that for level breaks the MTAR test exhibits similar behaviour to that derived by Leybourne et al. (1998) for the Dickey-Fuller (1979) test, with size distortion apparent for early breaks only. In contrast, the results for breaks in drift show the MTAR test to experience severe size distortion when breaks occur both early and late in the sample period. The divergence in results for the MTAR and DF tests is further examined, showing that in the presence of late breaks the MTAR test can lead a practitioner to draw false inferences of both stationarity and asymmetry.