 |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: xtabond2
Dear Statalisters,
I also have some questions regarding the proper implementation of
-xtabond2-.
The guidance provided by David Roodman is doubtless wonderful peace of
work - it is quite straightforward and makes a lot of things clear.
However, there are several questions I could not answer by myself. Thats
why I am asking for your help.
I am trying to estimate dynamic panel model like that:
y[i,t] = y[i,t-1] + w[i,t] + v[i] + e[i,t]
where in the notation of the xtabond2-help y = dependent, w =
covariate(s) (apropos, there are some theoretical arguments to suspect
'w' endogenous), v = fixed effect, e = residuals.
I am trying to estimate the model in differences (not levels).
xtabond2 ln_y l1.ln_y ln_w, robust noleveleq gmm(l1.ln_y) gmm(ln_w).
I am not very sure that I understood the xtabond2-syntaxes correctly, so
the first question is whether this is the right way to estimated the model.
Second, Arelano-Bond test for serial correlation suggests first order
autocorrelation, rejects however second order AC.
Hence, I am asking myself whether the first lags are proper instruments
or not and whether I should start with deeper lags. For example:
xtabond2 ln_y l1.ln_y ln_w, robust noleveleq gmm(l1.ln_y) gmm(ln_w,
lag(2 .)).
Thanks for any help.
Viktor
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/