Tackling and Understanding the Small Sample Bias in ARFIMA Models
Posted: 14 Sep 1999
Date Written: August 1994
It is shown that the approximate Whittle estimator suffers from an inherent small sample estimation bias which is not present in other frequency domain maximum likelihood methods. A bias corrected approximate Whittle estimator is derived and compared with existing alternatives. Furthermore, the behavior of bias in ARFIMA (p, d, q) models is addressed. The results of a systematic Monte Carlo study suggest that the size of the estimation bias and of the mean squared error is related to the asymptotic parameter correlation evaluated at the true parameters. Among other things, this implies the observed independence of bias and mean squared error from the memory parameter.
JEL Classification: G00
Suggested Citation: Suggested Citation