Testing the unit root hypothesis using generalized range statistics

The application of rescaled range statistics in econometrics is restricted to long memory detection in economic and financial time series. However, in this paper it is shown how such statistics can be generalized and used to test the unit root hypothesis. The proposed generalizations lead to testing procedures that are found to be consistent against I (0 , I (2 as well as against fractional integration and misspecification of the deterministic trend. A Monte Carlo simulation reveals that range-based tests can even outperform standard approaches to nonstationarity tests.