Re: [Gretl-users] VECM, Impulse response: gretl crashes
Allin Cottrell schrieb:
On Wed, 3 Mar 2010, Artur T. wrote:
> Allin Cottrell schrieb:
>
>> Which impulse response plot gave the crash? (That is, which was
>> the response variable and which the innovation?) Thanks.
>>
> It seems to happen for all the impulse-response patterns I've just
> tried. Here the output of the terminal (by the way, I use the current
> cvs on ubuntu 9.10):
>
Thanks, I see what the basic problem is. I'll have to investigate
a bit more to find the best fix. Should be fixed by tomorrow.
Allin Cottrell
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
Hey Allin,
thanks for fixing it. There are no problems anymore!
Regarding Sven's question:
"Slightly OT: How many replications does gretl use for this bootstrap?
Shouldn't it be configurable by the user?"
What's the standard number of replications gretl uses, Allin?
@Riccardo:
Sorry, my question was not exact enough, I think. I had a look in your
SVAR package which you sent around a while ago and I know that you
implemented impulse response functions for SVAR/SVECM modeling what is a
great thing!
But is it also possible to do impulse response functions for a VECM
which has imposed long-run restrictions on beta? In Microfit it is
possible to setup a VECM, impose long-run restrictions, do some
diagnostic tests, and then run some Impulse response function
(orthogonal and generalized) based on this restricted model (please
don't understand me wrong, I don't ask for all these features).
If I find some time during the next couple of days I could do some
testing trying to replicate SVAR estimations which I have done with
JMulti before. At the moment I have a version of your package where a
pdf file (SVARdoc.pdf) dated with 1st August 2008 is attached.
Best,
Artur