Re: [Gretl-users] $irf(): excess collinearity

[artur tarassow]

Ok, that's weird. If I run your script no error at all occurs here using a
Windows-version from 2013-05-02. Despite the fact that both series are
almost perfectly correlated.

If I use the catch command in my actual dataset gretl does not produce a
matrix with three columns holding the point estimates and the CIs.
Actually, should gretl generate a three-column matrix if I use the "catch"
command in case of excess collinearity ignoring the error, or should it
just prevent termination by jumping to the next task?

Artur


2013/5/12 Allin Cottrell <cottrell(a)wfu.edu>

> On Sun, 12 May 2013, artur tarassow wrote:
>
> > Maybe flexibility could be increased by allowing for an option like "set
> > varsimultol int"?
>
> Maybe. Not sure it's worthwhile though.
>
> > "catch" does not seem to work here: gretl still stops and
> > prints the error.
>
> Strange. I'm trying this:
>
> <hansl>
> open data9-7
> series X = UNEMP + 0.1*normal()
> var 4 UNEMP X
> catch matrix m = irf(1, 2, 0.2)
> if $error
>    print "error: got here"
> else
>    print "no error"
> endif
> </hansl>
>
> Here the VAR prints OK; the irf() call prints "Excessive
> collinearity in resampled datasets"; and the $error clause
> prints "error: got here".
>
> Allin
>
> _______________________________________________
> Gretl-users mailing list
> Gretl-users(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-users
>

Ok, that's weird. If I run your script no error at all occurs here using a Windows-version from 2013-05-02. Despite the fact that both series are almost perfectly correlated.

If I use the catch command in my actual dataset gretl does not produce a matrix with three columns holding the point estimates and the CIs. Actually, should gretl generate a three-column matrix if I use the "catch" command in case of excess collinearity ignoring the error, or should it just prevent termination by jumping to the next task?

Artur

2013/5/12 Allin Cottrell <cottr...@wfu.edu>

On Sun, 12 May 2013, artur tarassow wrote:

> Maybe flexibility could be increased by allowing for an option like "set
> varsimultol int"?

Maybe. Not sure it's worthwhile though.

> "catch" does not seem to work here: gretl still stops and
> prints the error.

Strange. I'm trying this:

<hansl>
open data9-7
series X = UNEMP + 0.1*normal()
var 4 UNEMP X
catch matrix m = irf(1, 2, 0.2)
if $error
   print "error: got here"
else
   print "no error"
endif
</hansl>

Here the VAR prints OK; the irf() call prints "Excessive
collinearity in resampled datasets"; and the $error clause
prints "error: got here".

Allin

_______________________________________________
Gretl-users mailing list
gretl-us...@lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users