CFE 2019: Start Registration
View Submission - CMStatistics
Title: Nonparametric test for heteroscedasticity in time series Authors:  Herold Dehling - Ruhr-University Bochum (Germany) [presenting]
Roland Fried - TU Dortmund University (Germany)
Sara Kristin Schmidt - Ruhr-University Bochum (Germany)
Max Wornowizki - TU Dortmund University (Germany)
Abstract: A new test is presented for heteroscedasticity of a time series that is based on recent ideas. The test statistic is given by Gini's mean difference of logarithmic local sample variances. We investigate the asymptotic distribution of our test statistic under the null hypothesis of constant variance, and establish consistency against a large class of alternatives. Our results hold under mild conditions concerning the dependence structure of the underlying time series.