Local Whittle estimation of the memory parameter in presence of deterministic components

Abstract

We discuss the estimation of the order of integration of a fractional process that may be contaminated by a time-varying deterministic trend or by a break in the mean. We show that in some cases the estimate may still be consistent and asymptotically normally distributed even when the order of magnitude of the spectral density of the fractional process does not dominate the one of the periodogram of the contaminating term. If trimming is introduced, stronger deterministic components may be neglected. The performance of the estimate in small samples is studied in a Monte Carlo experiment. © 2009 Blackwell Publishing Ltd.