On Mon, 11 Dec 2006, javier garcia enriquez wrote:
When Gretl gives us the periodogram, also appears the long
memory parameter estimation using GPH and Whittle estimators. In
Whittle Gretl tells us the size of "m" and we can obtain easily
the "a" (since m = T^a), but with GPH doesn´t. ¿What "a" does
the program use in GPH? ¿Is it possible change the value of "a"
using our own criterion?
I've made a few quick fixes here.
* Improved the documentation for the pergm command.
* Added an optional second parameter (with a corresponding GUI
dialog box) to set the bandwidth.
* Adjusted the default bandwidth for the GPH test.
Up till now, the GPH test used T^0.5, while the Whittle test
used the lesser of T/2 and T^0.6. I've changed it so that the two
tests have the same default, namely the second one mentioned.
pergm y # gets the default order, T^0.6
pergm y 16 # sets a bandwidth of 16
This change is in CVS, and there's a new Windows snapshot at
http://ricardo.ecn.wfu.edu/pub/gretl/gretl_install.exe
--
Allin Cottrell
Department of Economics
Wake Forest University, NC