[Gretl-users] long memory parameter

[Riccardo (Jack) Lucchetti]

On Mon, December 11, 2006 23:48, Javier García wrote:
> Yes, in my opinion it would be more useful if the optional parameter was the
> exponent. Nevertheless, thanks for the fixes. They´ve helped me a lot.

Well, I'm open to be convinced of the contrary, but I'd rather keep m:
remember that all asymptotic results mandate that m should be *proportional*
to a certain power of T, not *equal*. As a consequence, in practical
situations one often wants to experiment what happens with m = T^a \pm k,
(with k "small"). Obviously, it's very easy to compute m as a function of T
and play with it; on the other hand, in order to have m vary as wanted, one
would have to fiddle with exponents in a much less straightforward way.

Riccardo (Jack) Lucchetti
Dipartimento di Economia
Facoltà di Economia "G. Fuà"
Ancona