[Gretl-users] AR , Movavg commands

Allin Cottrell cottrell at wfu.edu
Thu Dec 4 15:56:51 EST 2008


On Sun, 30 Nov 2008, [ISO-8859-1] "Matías F. Rebolledo" wrote:

> 1) I'd like to know if, there is any way to solve a ols equation
> with a MA second order (q=2), in a similar way as to solve with
> AR p order (that's running AR command rather than OLS).

If I understand you right, you should be able to do that using the
arma command in gretl (though of course estimation will not be via
OLS).  For example:

open denmark.gdt
arma 0 0 2 ; 1 2 3 4 0

successfully estimates a money demand equation with 2 MA lags, no
AR.

> Trying to use ARMA command, I've received a Collineality message.

Maybe the model is mis-specified?

> 2) Just checking the last mail lists, I've figured that running
> a ADF test over Eviews more than once, I've got the same
> different results of p-values. But then I noticed that the lags
> (calculated automaticly by eviews) were different in each cases,
> so the p-values were also different.

There are various ways of automating the selection of the lag
order in ADF (t or z test, F-test, Akaike; and if a test, what
p-value is used?).  The algorithm used by gretl is set out quite
explicitly in the command help.  I guess what Eviews does must be
a bit different.

Allin Cottrell



More information about the Gretl-users mailing list