st: RE: Cragg Donald weak instruments statistics

["Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>]

Justina,

My short answer to your question below - do the Stock-Yogo-Cragg-Donald
critical values apply to a heteroskedastic- or cluster-robust F
statistic - is that I don't know of any study that has looked at this.
I believe Wooldridge's 2002 textbook remarks in passing that a robust
F-stat should be used if homoskedasticity is doubted, and he doesn't say
anything about any adjustments that should be made.  Given that, FWIW, I
would use Staiger-Stock's rule of thumb (the F stat should be >10) and
perhaps refer to the SYCD critical values just to see if they differ
from the rule of thumb in your specific case.

Hope this helps.

Cheers,
Mark

	-----Original Message-----
	From: Justina Fischer [mailto:justina.fischer@unisg.ch]
	Sent: Tuesday, December 12, 2006 1:46 PM
	To: stillman@motu.org.nz
	Subject: Cragg Donald weak istruments statistics
	
	

	Hi Steve,
	
	I just worked myself through another (new) version of ivreg2
command (it keeps constanly expanding...) and although you provide now
plenty of information, I fear that one question has not been answered in
full (or precisely enough).
	
	Here is my situation: I have a cross-section with one endogenous
regressor. Because of clustering I work with
heteroscedasticity-consistent standard errors.
	
	Based on the information given in the help file, the
Cragg-Donald criterion is only applicable in case of cond.
homoscedasticity. The help file then continues with " In the special
case of a single endogenous regressor, a robust test
	statistic for weak instruments is available with the first or
ffirst options;" , which then refers to the Cragg-Donald statistics
again.. and not, because it claims that this statistics then coincides
with the F-test on the exlucded instrumnts (which I always used as
criterion).
	
	So my question is wether in case of a single endogenous variable
and robust S.E. I can apply the Cragg-Donald criticial values or not.
The way the ivreg2 help file is designed in the moment, I find the
answer to my question quite opaque.
	
	You may put the Stata-list in the CC-line in your answer.
	
	Thanks a lot
	
	Justina
	
	
	
	Dr. Justina A.V. Fischer, M.A.
	University of St. GallenCH-9000 St. Gallen


*
*   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/