I'm not too familiar with 2SLS in panels...but if I'm reading your code correctly your first-stage regressions do not include the exogenous variables from the second-stage regression (fdstss2 fdstot2 fedss2 fedoth2 lbal). They need to be there as do all of the instruments. Note that your standard errors will not match regardless. The variance of the IV estimator is not equivalent to the variance of the standard least squares estimator.
This also is an opportunity to lobby for an option in the IV commands to show the output from the instrumenting equations.
Tracy
>>> "Stephanie Zobay" <[email protected]> 06/15/02 04:21PM >>>
Dear Listers,
I am not understanding the difference between manually doing a 2SLS and
letting Stata do it with xtivreg.
I estimated each of 2 endogenous variables separately, saving the predicted
values:
xtreg srev2 unemp pctmdcaid pctprsner gsp2 popgrow
. predict srevhat
(option xb assumed; fitted values)
. xtreg lrev2 pcy1 tel pctschkid pctincpcrime taxbur
. predict lrevhat
(option xb assumed; fitted values)
Then I ran my regression using the predicted values for the endogenous
variables:
. xtreg staid2 srevhat lrevhat fdstss2 fdstot2 fedss2 fedoth2 lbal
The result I get is completely different than if I use:
xtivreg staid2 fdstss2 fdstot2 fedss2 fedoth2 lbal (srev2 lrev2 = unemp
pctmdcaid pctprsner gsp2 popgrow pcy1 tel pctschkid pctincpcrime taxbur
fdstss2 fdstot2 fedss2 fedoth2 lbal)
What is the difference and how do I know which result is correct?
Stephanie
*
* 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/
*
* 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/
