A0589
Title: Rank tests for randomness against time-varying MA alternative
Authors: Junichi Hirukawa - Niigata University (Japan) [presenting]
Abstract: The idea of the problem of testing randomness against ARMA alternatives to a class of locally stationary processes introduced by Dahlhaus is extended. The linear serial rank statistics are used, and the notion of the contiguity by LeCam for the testing problem is applied. Under the null hypothesis, the joint asymptotic normality of the proposed rank test statistics and the log-likelihood ratio is established using the local asymptotic normal property. Then, applying LeCam's third lemma, the asymptotic normality of the test statistic under the alternative is automatically derived.
