Generally the estimation techniques are not guaranteed to deliver results
for a d of, say, 1.7, whereas estimating a d of 0.7 is more
straightforward. Most published work therefore applies the estimators to
e.g. the inflation rate rather than the price level.
Kit Baum
Boston College
--On Fri, Jan 8, 1999 14:04 -0500 Suzanna De Boef
<[log in to unmask]> wrote:
> Can someone explain to me why we need to estimate the ARFIMA models on
> first differenced data rather than levels data? I am trying to compare
> GAUSS results purportedly using Sowell's exact MLE with the OX results
> and whether to difference before estimation makes a big difference in
> the comparison.
>
> Thank you very much.
> Suzanna De Boef
>
> Visiting Assistant Professor
> Harvard University
> Department of Government
> Littauer Center, North Yard
> Cambridge, MA 02138
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|