The Performance of Variance Ratio Unit Root Tests Under Nonlinear Stationary TAR and STAR Processes: Evidence from Monte Carlo Simulations and Applications

摘要

This paper investigates the performance of variance ratio unit root tests under nonlinear stationary three-regime threshold autoregressive (TAR) and smooth transition autoregressive (STAR) processes that are significant for some economic theories and variables. Variance ratio unit root tests are effective tools in empirical analysis because they can theoretically consider broad classes of nonlinear stationary processes under the null or alternative hypothesis. Nevertheless, our Monte Carlo simulations demonstrate that these tests perform poorly (with severe size distortions or low power) under stationary TAR and STAR processes. To verify our Monte Carlo results, we apply these tests to yield spreads such as the TAR and STAR processes.