Nirina--
I would make Kit's point more forcefully: if you can find good
instruments, the rest of the analysis is simple, and you can publish
it in a top journal (so don't expect a _lot_ of free advice on good
instruments), but it's a good bet that what you think are good
instruments will not be thought so by others, or will turn out not to
be on further examination. Even a lot of the most famous clever IV
specifications were later called into question, e.g.
Hoxby and Rothstein, both later published in AER:
http://www.nber.org/papers/w4979
http://papers.nber.org/papers/w11215
or:
Angrist and Krueger. 1991. "Does Compulsory School Attendance Affect
Schooling and Earnings?" The Quarterly Journal of Economics, 106(4):
979-1014. [see also http://www.nber.org/papers/w3572]
Bound, Jaeger, and Baker. 1995. "Problems with Instrumental Variables
Estimation when the Correlation Between the Instruments and the
Endogenous Explanatory Variables is Weak," Journal of the American
Statistical Association, 90(430): 443-450. [see also
http://www.nber.org/papers/w5835 and http://www.nber.org/papers/t0137]
Also, please bear in mind that IV estimates are still biased (though
they are at least consistent when all the assumptions are satisfied--a
tall order, as I have said above).
On 9/21/07, Kit Baum <[email protected]> wrote:
> The issue of where to find valid instruments depends on the context
> of your model.
*
* 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/
