st: suggestions for adjusting the variance-covariance matrix of the endogenous regressor + 3sls over 2sls

["Chirantan Chatterjee" <chirantan@cmu.edu>]

Hi All,

I am working on a panel dataset estimating a specification y=f(x1,x2-xn)
where x1 is endogenous for which i have z1-z3 instruments.

What i am doing thus is to use the reg3 command and the ,2sls option to
get my results thus:

reg3 (y=x1 x2-xn) (x1= z1-z3), 2sls.

I have 3 set of questions here:

a. Are the errors for x1 getting adjusted for it being estimated in the
first stage using the z regressors and the estimated x1hat being used in
the 2nd stage. If not, can you folks suggest how to insert some additional
options in the command above to ensure i adjust the errors for my
endogenous term x1.

b. Also, i was trying to run 3sls on this - can somebody suggest if that
is a way to go, i just have 1000 odd observations in my data.

c. Finally, if i use ivregress command instead of reg3, can somebody
suggest what extra option to use in the above case, to ensure that i am
correcting the errors for the endogenous term x1. That is since i am using
x1hat in the 2nd stage, how can one correct the variance-covariance matrix
of the endogenous variable.


Thanks much, Chirantan.
PhD Student, Carnegie Mellon.


-- 


*
*   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/