Bootstrapping the HEGY seasonal unit root tests

This paper proposes bootstrap versions of the seasonal unit root tests of, inter alia, Hylleberg, Engle, Granger and Yoo (J. Econometrics 44 (1990) 215–238) (HEGY). We report a simulation study of the properties of both the conventional and bootstrapped seasonal unit root tests when applied to series having higher-order serial correlation and/or periodic heteroscedasticity, both of which are known to severely distort the significance level of the conventional tests. Our results demonstrate that the bootstrap provides good approximations to the statistics’ null distributions. Moreover, the bootstrap corrects the adverse effects of data-dependent lag selection seen in the conventional augmented HEGY tests. The bootstrapped tests have comparable power to (infeasible) exactly significance-level-corrected lag-augmented HEGY tests, and their use is recommended.