st: re: IV regression

[Kit Baum <baum@bc.edu>]

<>	
Carlos wonders

Second, if the relationship between the proposed instrument and the
endogenous regressor has to be causal (or likely causal).
 would the IV only be valid if this relationship = "IV (likely) causes
the endogenous regressor" or would the IV still be valid if the
relationship is = "endogenous regressor likely causes IV"?


From a statistical point of view, if the endogenous regressor  
'causes' the instrument, it is not an instrument. The whole notion is  
that the endogenous regressor is correlated with the error, which you  
are trying to purge by means of the instruments. If the instrument is  
'caused by' the endogenous regressor, it too will be correlated with  
the error. A test of overidentifying restrictions (in the case where  
you have overID) evaluates the null that the instruments, taken  
together, are suitably orthogonal to the error process.

Kit Baum   |   Boston College Economics and DIW Berlin   |   http://ideas.repec.org/e/pba1.html
An Introduction to Stata Programming   |   http://www.stata-press.com/books/isp.html
An Introduction to Modern Econometrics Using Stata   |   http://www.stata-press.com/books/imeus.html



*
*   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/