Page 1 of 1

Dynamic Panel Method/Arelleno-Bond

Posted: Wed Apr 21, 2010 1:53 pm
by kumara
Hi all,

I am using Eviews-6 to implement Arellano-Bond (AB) method to estimate a dynamic panel model using dynamic panel wizard. AB estimates are based on the assumption that there should not be second-order serial correlation in the residuals of the first-difference equation. Arellano and Bond (1991) develop a test statistics to test for the serial correlation (m_2 statistics). The dynamic panel wizard in Eviews-6 does not have this test. How do I test for serial correlation in the residuals of the first-difference equation?

Cheers

A.

Re: Dynamic Panel Method/Arelleno-Bond

Posted: Sat Aug 07, 2010 9:36 am
by kelaynak
Hello,

I also need those Arellano-bond test for 1 st order autocorrelation (M1)
and second order autocorrelation (M2)

Is there M1 and M2 test for dynamic panel estimation in GMM in Eviews 7.1?
I could not find that option in Eviews

Or if there is no that choice automatic, is there any way to do it manually?
Is it different than standart autocorrelation test?

Please give a manual methodology to do find 1st order and second order aoutocorrelation in dynamic panel?

Thanks in advance
Rgds

Kelaynak

Re: Dynamic Panel Method/Arelleno-Bond

Posted: Sat Aug 07, 2010 11:42 am
by kelaynak
Hello
In eviews manual (panel estimation examples),
there is manually calculation method for first order auto correlation for panel.
- first saving residuals
then regress resid on resid(-1) and checking coeefcient whether it is less -0.5 or not. If higher, no first order autocorrelation. There is also formal test walt exapmple in there.

1) Is this method also valid for dynamic panel data?

2) What about second order autocorrelation? What is its manual method for it? Please explain?
what is its critical parameters?

Thanks

Re: Dynamic Panel Method/Arelleno-Bond

Posted: Wed Aug 11, 2010 10:40 am
by EViews Glenn
No. There's a different statistic which unfortunately, we haven't had time to implement.