12:39 p.m.
For my panel data (all continuous variables; 1 DV, 4 IVs; 8 time points for
each of the 5 companies), the Breusch-Pagan test is nonsignificant
(Chi-square(4) = 1.77291, p= 0.777434) implying that both Random effects
model and fixed effect model are not appropriate. I used the OLS
regression with heteroscedasticy problem. Any suggestions on how to remedy
this?
Appreciate your help.
Lema.
1:49 p.m.
Your main problem here is a lack of data. A "Breusch-Pagan test is
nonsignificant (Chi-square(4) = 1.77291, p= 0.777434)" does not reject the
null of heteroskedasticity. With your short series, however, the test
has very low power. Check what you are doing here.
Without some knowledge of the economic model that you are trying to
estimate and your data, it is not possible to advise you. Perhaps you
should check your application of the Breusch Pagan test to your pooled
regression if that is what you have done.
John C Frain
3 Aranleigh Park
Rathfarnham
Dublin 14
Ireland
www.tcd.ie/Economics/staff/frainj/home.html
https://jcfrain.wordpress.com/
https://jcfraincv19.wordpress.com/
mailto:frainj@tcd.ie
mailto:frainj@gmail.com
On Fri, 1 Apr 2022 at 13:39, 3J LEMA <3jlema(a)gmail.com> wrote:
For my panel data (all continuous variables; 1 DV, 4 IVs; 8 time
points
for each of the 5 companies), the Breusch-Pagan test is nonsignificant
(Chi-square(4) = 1.77291, p= 0.777434) implying that both Random effects
model and fixed effect model are not appropriate. I used the OLS
regression with heteroscedasticy problem. Any suggestions on how to remedy
this?
Appreciate your help.
Lema.
_______________________________________________
Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
Website:
https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
5:11 p.m.
Am 01.04.2022 um 15:49 schrieb John C Frain:
Your main problem here is a lack of data. A "Breusch-Pagan test
is
nonsignificant (Chi-square(4) = 1.77291, p= 0.777434)" does not reject
the null of heteroskedasticity. With your short series, however, the
test has very low power. Check what you are doing here.
I believe the Breusch-Pagan test that was referred to is the one you get
with "panspec", so a test of the pooled model against panel RE. It would
be less confusing if those guys hadn't invented various tests!
I'm not saying that anything you wrote is wrong.
cheers
sven
9:15 p.m.
On Fri, 1 Apr 2022 at 18:11, Sven Schreiber <svetosch(a)gmx.net> wrote:
Am 01.04.2022 um 15:49 schrieb John C Frain:
> Your main problem here is a lack of data. A "Breusch-Pagan test is
> nonsignificant (Chi-square(4) = 1.77291, p= 0.777434)" does not reject
> the null of heteroskedasticity. With your short series, however, the
> test has very low power. Check what you are doing here.
I believe the Breusch-Pagan test that was referred to is the one you get
with "panspec", so a test of the pooled model against panel RE. It would
be less confusing if those guys hadn't invented various tests!
I'm not saying that anything you wrote is wrong.
You are correct. The mention of heteroskedasticity confused me. I still
believe that the problem is one of lack of data and the resulting low power
of the test.
Take Care
John
cheers
sven
_______________________________________________
Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
Website:
https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
754
days inactive
756
days old
3 comments
3 participants
participants (3)
-
3J LEMA -
John C Frain -
Sven Schreiber