Kelejian (Journal of the American Statistical Association 66(334): 373-374) says that for a RHS endogenous regressor, IV is fine whether or not the variable is truncated, binary, whatever. You can get efficiency gains over IV using system estimation methods that take account of the form of the regressor, but IV is still consistent. Not 100% sure it applies to panels as well.
Nicola
P.S. I'll NOT receive/read any email but the Digest.
At 02.33 30/04/2009 -0400, Jeffrey Roberts wrote:
>Hi,
>
>Can XTIVREG be used in the case of discrete endogenous variables in some way?
>
>I have a problem such as:
>
>Y = f(a,b,c)
>a = f(z,b,c)
>
>where 'a' is a dichotomy. For 'z' to be an instrument for 'a' then
>there is a co-linearity problem.
>
>Thanks
>Jeff
>*
*
* 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/
