Non-parametric Test For Change In Persistence

This paper studies how to detect the change in persistence on the basis of ranks of a sequence.We derive a statistic to test,and investigate asymptotic distribution under the null hypothesis and show the consistence of the test under the alternative hypothesis.The Monte-Carlo simulations demonstrate that the non-parametric test has more correct sizes at the lost of less powers in finite samples,comparing with the test proposed by Kim[7],that means the test has much lower rejection rate when the series has no change in persistence.To illustrate this problem,We apply our test to the series of the monthly CPI rate and ISM non-manufacturing index of America.Then a new non-paxametric test method based on the ranks is proposed on the basis of the statistic of moving residual ratio test.And the asymptotic distribution is obtained under the null hypothesis,and its consistency is proved under the alternative hypothesis.The simulation results show that the size is smaller by using the non-parametric method for heavy-tailed distribution.Finally,the method is applied to the University of Michigan consumer confidence index,and the results show that the method is effective.