10:43 a.m.
Hi all,
when performing fixed-effects regressions, we report the
classic Breusch-Pagan poolability test. However, there is an alternative
test by Pesaran,
https://link.springer.com/article/10.1007/s00181-020-01875-7
(section 4) which is getting increasingly popular, because it reportedly
has better small-sample properties. In fact, the working paper version has
been around for a long time, but now that it's been published, it may make
sense to have this in.
The test seems to be relatively easy to implement, so we could have
somebody write a function package for it. Any takers?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
12:21 p.m.
Am 26.06.2020 um 12:43 schrieb Riccardo (Jack) Lucchetti:
when performing fixed-effects regressions, we report the classic
Breusch-Pagan poolability test. However, there is an alternative test by
Pesaran,
https://link.springer.com/article/10.1007/s00181-020-01875-7
(section 4) which is getting increasingly popular, because it reportedly
has better small-sample properties. In fact, the working paper version
has been around for a long time, but now that it's been published, it
may make sense to have this in.
The test seems to be relatively easy to implement, so we could have
somebody write a function package for it. Any takers?
Hi Jack,
please check out Jörg Breitung's and my package CSDpanel if that's one
of the tests that are already in there. (My hunch would be yes.)
If so, we might update the package just to include the reference to this
publication.
cheers
sven
1:07 p.m.
Am 26.06.2020 um 14:21 schrieb Sven Schreiber:
Am 26.06.2020 um 12:43 schrieb Riccardo (Jack) Lucchetti:
> when performing fixed-effects regressions, we report the classic
> Breusch-Pagan poolability test. However, there is an alternative test by
> Pesaran,
>
> https://link.springer.com/article/10.1007/s00181-020-01875-7
>
> (section 4) which is getting increasingly popular, because it reportedly
> has better small-sample properties. In fact, the working paper version
> has been around for a long time, but now that it's been published, it
> may make sense to have this in.
>
> The test seems to be relatively easy to implement, so we could have
> somebody write a function package for it. Any takers?
Hi Jack,
please check out Jörg Breitung's and my package CSDpanel if that's one
of the tests that are already in there. (My hunch would be yes.)
If so, we might update the package just to include the reference to this
publication.
BTW, re-visiting the (closed) feature request 97 "cross-sectional
dependence test" I stumbled over my own remark that from 2016d on there
exists "modtest ... --xdepend" natively.
However, the user guide contains neither "cross dependence" nor
"cross-sectional dependence".
On the brighter side, a web search for "gretl cross-dependence" turns
out links to the feature request mentioned above as well as to the
command reference (modtest), and also to our CSDpanel package.
cheers
sven
6:10 a.m.
On Fri, 26 Jun 2020, Sven Schreiber wrote:
Am 26.06.2020 um 14:21 schrieb Sven Schreiber:
> Am 26.06.2020 um 12:43 schrieb Riccardo (Jack) Lucchetti:
>
>> when performing fixed-effects regressions, we report the classic
>> Breusch-Pagan poolability test. However, there is an alternative test by
>> Pesaran,
>>
>> https://link.springer.com/article/10.1007/s00181-020-01875-7
>>
>> (section 4) which is getting increasingly popular, because it reportedly
>> has better small-sample properties. In fact, the working paper version
>> has been around for a long time, but now that it's been published, it
>> may make sense to have this in.
>>
>> The test seems to be relatively easy to implement, so we could have
>> somebody write a function package for it. Any takers?
>
> Hi Jack,
> please check out Jörg Breitung's and my package CSDpanel if that's one
> of the tests that are already in there. (My hunch would be yes.)
Yes, that seems to be the one, thanks.
> If so, we might update the package just to include the reference
to this
> publication.
BTW, re-visiting the (closed) feature request 97 "cross-sectional
dependence test" I stumbled over my own remark that from 2016d on there
exists "modtest ... --xdepend" natively.
However, the user guide contains neither "cross dependence" nor
"cross-sectional dependence".
On the brighter side, a web search for "gretl cross-dependence" turns
out links to the feature request mentioned above as well as to the
command reference (modtest), and also to our CSDpanel package.
Possible enhancement: the variant for unbalanced panels seems to be
unimplemented yet.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
6:35 a.m.
Am 29.06.2020 um 08:10 schrieb Riccardo (Jack) Lucchetti:
On Fri, 26 Jun 2020, Sven Schreiber wrote:
> Am 26.06.2020 um 14:21 schrieb Sven Schreiber:
>> Am 26.06.2020 um 12:43 schrieb Riccardo (Jack) Lucchetti:
>>>
https://link.springer.com/article/10.1007/s00181-020-01875-7
>>> (section 4)
>> Hi Jack,
>> please check out Jörg Breitung's and my package CSDpanel if that's one
>> of the tests that are already in there. (My hunch would be yes.)
Yes, that seems to be the one, thanks.
OK, good; so adding that reference is one item on our to-do list for the
package.
Next I see various places where the documentation could use more updates
(notes to self):
- pkg install instead of just install
- add a remark about the native 'modtest ... --xdepend' (actually I
don't remember whether it's the same test, why the paper reference given
in the command doc is different [Pesaran, M. H. (2004) “General
diagnostic tests for cross section dependence in panels”. Cambridge
Working Papers in Economics (CWPE 0435)]
Possible enhancement: the variant for unbalanced panels seems to be
unimplemented yet.
Yes; do you need that? Does the native modtest cover that?
thanks
sven
6:39 a.m.
On Mon, 29 Jun 2020, Sven Schreiber wrote:
- add a remark about the native 'modtest ... --xdepend'
(actually I
don't remember whether it's the same test, why the paper reference given
in the command doc is different [Pesaran, M. H. (2004) “General
diagnostic tests for cross section dependence in panels”. Cambridge
Working Papers in Economics (CWPE 0435)]
It is!
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
1950
days inactive
1953
days old
5 comments
2 participants
participants (2)
-
Riccardo (Jack) Lucchetti -
Sven Schreiber