st:Advice on Instruments

[Ayman Farahat <ayman.farahat@yahoo.com>]

Hello; 
I am working with an endogenous variable. Each observation has the form 

subject_id , variable_1 ...  variable_ind     => where variable_ind is my endogenous variable. 


Each subject belongs to multiple groups , for example the subject can belong to its gender , age, occupation , .. etc. 

I wanted to instrument for variable_end :
I was thinking of two options for instruments
1) Use the group indicators as my instruments. In that vein  in first stage of 2sls i will regress variable_ind on i.age i.gender .. 

2) Use the average or other statistics of the endogenous variable in the group as regressor, so my instruments would be the average in age, gender.. 



Any advice which set of instruments make more sense and why?
Thanks
Ayman

________________________________

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/