st: Estimated regressors and GMM

["Tsatsaronis, Konstantinos" <Konstantinos.Tsatsaronis@bis.org>]

I am having trouble in estimating by stacked GMM a model with generated regressors. I checked through past postings and FAQs but I was not too successful. I was wondering if anyone could offer a helpful tip.

Here is the issue. I am estimating the following regression:

y = c0 + c1*W + c2*u_hat + e

where u_hat is the residual from a previous regression

x = b0 + b1*Z + u

Using u_hat in the regression messes up the inference of the c's so following McFadden and Newey's
chapter in the Handbook of Econometrics I tried stacking the two sets of moments conditions and estimating a GMM.
I have only used "canned" versions of GMM before in other software, but this is the first time
I have to be more explicit about it. I admit that I am not starting this with the greatest degree of confidence in my skill...

Here is one of my attempts to set it up:

gmm (eq1: x - {b0} - {xb:Z1 Z2}) (eq2: y - {c0} - {c2}*(x - {b0} - {yb:Z1 Z2}) - {cw:W1 W2 W3}), inst(Z1 Z2)


STATA complains variously about the weighting matrix not being positive definite, about the instrument list etc.

I realise that I am doing something fundamentally wrong but I cannot seem to get the grip of the GMM command.

Any ideas of what I am doing wrong, or how I can deal with this problem in a different way,
would be greatly appreciated.


Kostas Tsatsaronis


Disclaimer

This e-mail message shall not be construed as legally binding on the Bank for International Settlements (BIS).

As internet communications are not secure, the BIS does not accept responsibility for the content of this message.

This message is intended only for the recipient(s) named above. Any unauthorised disclosure, use or dissemination, either in whole or in part, of this message is prohibited. If you have received this message in error, please inform the sender immediately by return e-mail and delete this message and any attachments thereto from your system. 

Thank you for your cooperation.


*
*   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/