8:50 a.m.
Hi,
I cannot find the answer in the docs whether restricted exogenous terms
in the 'vecm' or 'coint2' commands are included with a lag or as-is.
This concerns the "rxlist" or "zlist" arguments to the commands.
I remember some discussion about it years ago on one of the lists (users
or devel), but I cannot find it in a web search. (And because I do not
agree to google's non-privacy policy, I cannot directly search in the
mailing list search interface which is powered by google.)
Obviously, for the standard and automatically included deterministic
terms like a constant or a trend it doesn't matter, but for others it does.
It would be helpful if that info could be added to the docs.
Thanks and happy Easter,
Sven
9:01 a.m.
Here is a list of the two options as they are implemented currently.
<hansl>
set verbose off
open denmark.gdt -q
list ylist = LRM LRY
list rlist = IBO
# exogenous only in the coint. space but not in the VAR
vecm 2 1 ylist ; ; rlist
# exogenous only in the VAR as-is w.o. any lags
vecm 2 1 ylist ; rlist
</hansl>
Of course this should be documented if it's not yet.
Artur
Am 15.04.2017 um 14:50 schrieb Sven Schreiber:
Hi,
I cannot find the answer in the docs whether restricted exogenous terms
in the 'vecm' or 'coint2' commands are included with a lag or as-is.
This concerns the "rxlist" or "zlist" arguments to the commands.
I remember some discussion about it years ago on one of the lists (users
or devel), but I cannot find it in a web search. (And because I do not
agree to google's non-privacy policy, I cannot directly search in the
mailing list search interface which is powered by google.)
Obviously, for the standard and automatically included deterministic
terms like a constant or a trend it doesn't matter, but for others it does.
It would be helpful if that info could be added to the docs.
Thanks and happy Easter,
Sven
1:24 p.m.
Am 15.04.2017 um 15:01 schrieb Artur Tarassow:
Here is a list of the two options as they are implemented currently.
list ylist = LRM LRY
list rlist = IBO
# exogenous only in the coint. space but not in the VAR
vecm 2 1 ylist ; ; rlist
Yeah, but that does not answer the question of whether the
error-correction term as written on the right-hand side of the
equation(s) is composed of LRM(-1)/LRY(-1)/IBO or of
LRM(-1)/LRY(-1)/IBO(-1). That's my question.
thanks,
sven
2:10 p.m.
Good and important question. Never thought about it as I assumed it has
to be the ordering LRM(-1)/LRY(-1)/IBO(-1).
Artur
Am 15.04.2017 um 19:24 schrieb Sven Schreiber:
Am 15.04.2017 um 15:01 schrieb Artur Tarassow:
> Here is a list of the two options as they are implemented currently.
> list ylist = LRM LRY
> list rlist = IBO
>
> # exogenous only in the coint. space but not in the VAR
> vecm 2 1 ylist ; ; rlist
Yeah, but that does not answer the question of whether the
error-correction term as written on the right-hand side of the
equation(s) is composed of LRM(-1)/LRY(-1)/IBO or of
LRM(-1)/LRY(-1)/IBO(-1). That's my question.
thanks,
sven
5:17 p.m.
On Sat, 15 Apr 2017, Sven Schreiber wrote:
Hi,
I cannot find the answer in the docs whether restricted exogenous terms in
the 'vecm' or 'coint2' commands are included with a lag or as-is. This
concerns the "rxlist" or "zlist" arguments to the commands.
OK, that should be in the docs, but one nice thing about open-source
software is that one can always quiz the source, either directly or
via the web:
https://sourceforge.net/p/gretl/git/ci/master/tree/lib/src/var.c
There we find the following comment, under the function
VECM_fill_Y():
"There's room for debate here (?) over whether we should
use the current value or the first lag of restricted
exogenous variables. But using the current value agrees
with Ox. See also VAR_set_sample()."
And the following code indicates fairly clearly that we use the
current value.
Allin
2737
days inactive
2737
days old
4 comments
3 participants
participants (3)
-
Allin Cottrell -
Artur Tarassow -
Sven Schreiber