R/S-bootstrapping test for fractional integration

Abstract

The R/S-type statistic has been designed by Lo and used based on their asymptotic distributions by Giraitis et al. for the detection of fractionally integrated processes. The authors propose the use of a parametric bootstrap procedure based on the R/S-type statistics for testing for fractional integration. Simulation results show that the bootstrap approach shows the best performance in terms of producing good power results while maintaining control over type I error.

Keywords:

• R/S-type statistic
• Parametric bootstrap
• Long memory
• Fractional integration