st: Query on Dynamic Panel Technique / Granger Causality / Sargan Test

[Pejman Abedifar <pabedifar@gmail.com>]

Dear Statalist,

I intend to estimate dynamic panel data to test for Granger causality
test (ie I'll put one up to 3 lags of dependent variable (Y) and the
explanatory variable (X) which I think may cause Y. Then I need to
test whether the lags value of X are jointly significant)
I have a unbalanced panel data consisting of 8900 individuals for 18
periods (quarters).
I use Arellano-Bover/Blundell-Bond linear dynamic panel-data
estimation technique in stata (xtdpdsys).
Arellano-Bond (1991) test (estat abond test) for zero autocorrelation
in first-differenced error term do not reject the null hypethesis.
However, the Sargan test rejects the validity of the over-identifying
restrictions.

I don't know how I can solve the over-identifying restrictions
problem. When I try to define some of the control variables as
endogenous or predetermined, the chi-squared is increased, possibly
due to increase in the number of instrument.

Could you please let me know how I should deal with this issue?
Thanks in advance!

--
Kind regards,
Pejman Abedifar
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/