You may have found the solution in the meantime, otherwise my belated reply is: the Sargan is significant, thus your instruments may be not valid. But no suggestion on how to move on (I prefer other methods like -xtlsdvc- (from SSC) or the inverse probability of treatment weights (IPTW) (see Epidemiology, 11: 550-560 and see Jamie Robins and Miguel Hernan both Harvard School of Public Health, they have a website with all the relevant papers)
Nicola
P.S. I'll NOT receive/read any email but the Digest.
At 02.33 19/11/2008 -0500, "Eleonora Bartoloni06" wrote:
>** High Priority **
>
>Hallo,
>
>I have performed the xtabond procedure in stata.10 and I have obtained the following results.
>
>xtabond ros q_immmat leverage vafat offat innset add coslva d1998 d2003, inst(north_we n
>> orth_ea centre) lags(1) endog(inn) vce(robust) artests(2)
>
>Arellano-Bond dynamic panel-data estimation Number of obs = 15546
>Group variable: codice_i Number of groups = 2591
>Time variable: time
> Obs per group: min = 6
> avg = 6
> max = 6
>
>Number of instruments = 44 Wald chi2(11) = 2917.92
> Prob > chi2 = 0.0000
>One-step results
>- ------------------------------------------------------------------------------
> | Robust
> ros | Coef. Std. Err. z P>|z| [95% Conf. Interval]
>- -------------+----------------------------------------------------------------
> ros |
> L1. | .3233811 .0198598 16.28 0.000 .2844566 .3623056
> inn | .2425445 .1280361 1.89 0.058 -.0084016 .4934906
> q_immmat | .0023804 .0013162 1.81 0.071 -.0001994 .0049602
> leverage | .0000875 .0001947 0.45 0.653 -.000294 .000469
> vafat | .6048328 .0210364 28.75 0.000 .5636022 .6460633
> offat | -.6008628 .0672536 -8.93 0.000 -.7326773 -.4690482
> innset | .0411136 .0078649 5.23 0.000 .0256986 .0565286
> add | -.0017756 .0007399 -2.40 0.016 -.0032257 -.0003254
> coslva | -.7031861 .3948626 -1.78 0.075 -1.477103 .0707304
> d1998 | .6747555 .0723119 9.33 0.000 .5330267 .8164843
> d2003 | -1.008683 .0886336 -11.38 0.000 -1.182402 -.8349642
> _cons | -12.10147 .8268329 -14.64 0.000 -13.72203 -10.4809
>- ------------------------------------------------------------------------------
>Instruments for differenced equation
> GMM-type: L(2/.).ros L(2/.).inn
> Standard: D.q_immmat D.leverage D.vafat D.offat D.innset D.add D.coslva D.d1998
> D.d2003 north_we north_ea centre
>Instruments for level equation
> Standard: _cons
>
>.. estat sargan
>Sargan test of overidentifying restrictions
> H0: overidentifying restrictions are valid
>
> chi2(32) = 109.316
> Prob > chi2 = 0.0000
>
>. estat abond
>
>Arellano-Bond test for zero autocorrelation in first-differenced errors
> +-----------------------+
> |Order | z Prob > z|
> |------+----------------|
> | 1 |-16.228 0.0000 |
> | 2 |-.66241 0.5077 |
> +-----------------------+
> H0: no autocorrelation -
>
>
>
>Should I conclude that the Sargan test clearly does not reject the hypothesis that the sets of additional instruments used in the GMM specification are invalid?
>Is there someone who could help me in the correct interpretation and,if my interpretation is correct, in finding a better specification of the model?
>thank you
>Eleonora Bartoloni
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
