Am 17.04.2019 um 23:39 schrieb Allin Cottrell:
On Wed, 17 Apr 2019, Olasehinde Timmy wrote:
> Thanks for your reply sir. I will now to explain my situation.
> I read it that we can save the $alpha (speed of adjustment or loading)
> and
> the $beta (long run) matrix after estimating VECM in gretl, and these two
> matrices can be used in the SVEC model estimation. However, I found it
> difficult to save them after estimating VECM in gretl as I keep getting
> error message that "the statistics you requested was not available"
The "accessors" you need to use after estimating a VECM via the built-in
command are $jalpha and $jbeta (where 'j' indicates "Johansen"), not
$alpha and $beta. Please see the "Function Reference" under the gretl
Help menu for specification of these and other accessors: a complete
list is provided.
OK, so we're talking about the prior cointegration/VECM estimate before
moving to the SVEC.
Tryint it out I'm noticing that it is not really possible in gretl's GUI
to save the objects that you get via $jalpha and $jbeta in a script.
From the VECM model window (menu Save) one can only save the EC term as
series. I wasn't actively aware of this limitation.
Now, I think it is possible to estimate the VECM in the GUI, and then
write '$jbeta' into the SVEC_GUI interface. But still it seems as if it
would be useful to have direct access to these estimates already at the
VECM step.
thanks
sven