Re: st: iterating a regression

["Hans J. Baumgartner" <hbaumgartner@diw.de>]

Hi,

I am not sure what you are trying to estimate but I guess the potential 
correlation between your two estimations will cause a problem.

Are you willing to restrict cj to a certain value? If yes, you could 
estimate

y=a+b1(x1+c1w1)+b2(x2+c2w2).

That is, inserting the second equation and multiplying out. In the new 
equation, only bj is estimated; cj is set, i.e. restricted.

Hans


tk2130@columbia.edu schrieb:


>    Dear All,
>    How can I iterate a regression by feeding back back some
> parameter estimates at each step back into one of the regressors in
> the regression? Should I use foreach forval commands?
>
> To be specific: I have an equation y=a+b1*x1+b2*x2+c1*z+error
>
> but z=b1*w1+b2*w2 (for some variables w1, w2)
>
> I start with some initial z and estimate the above equation then
> compute the new z with the b1 b2 estimates from the regression
> output and run the regression again with this new z. I want to
> iterate this until convergence is obtained. Does anyone know how to
> carry out this procedure? Thank you in advance for any comments and
> help.
>
>    Tumer Kapan
>    PhD Student
>    Economics Department
>    Columbia University
>
> *
> *   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/
--
Hans J. Baumgartner DIW Berlin
German Institute for Economic Research
Dept. Public Economics
K�nigin-Luise-Str. 5; 14195 Berlin; Germany
Tel.: +49/30/89789-307
Fax.: +49/30/89789-114
http://www.diw.de
http://www.hansbaumgartner.de


*
*   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/