Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Cragg-Donald F statistic - Weak Indentification


From   Erasmo Giambona <e.giambona@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Cragg-Donald F statistic - Weak Indentification
Date   Mon, 28 Sep 2009 18:20:07 +0200

Thanks All. I had looked at Stock and Yogo, but I cannot really get
the intuition of the Cragg-Donald test.

Austin, hope I can ask for a clarification. If the third endogenous
variable is not pushed around by Z, then wouldn't the simple F-test
(excluded instruments) be very small? It seems this would tell me
already that I have a weak identification problem. What I cannot see
is what is the additional piece of information that I can gather from
the Cragg-Donald test.

Erasmo

On Mon, Sep 28, 2009 at 5:52 PM, Austin Nichols <austinnichols@gmail.com> wrote:
> Erasmo Giambona <e.giambona@gmail.com>:
> This is discussed by Stock and Yogo (reference given in the help file
> for -ivreg2- on SSC), among others.  Basically, the matrix version of
> the first-stage F-test gives the smallest effect of the various
> directions in which the endog X vars are pushed around by excluded
> instruments Z.  If two out of three endog X variables are strongly
> pushed around by Z, but one is hardly affected at all, you still have
> a weak instruments problem--and all of this is conditional on whatever
> other covariates are in the model, of course.
>
> On Mon, Sep 28, 2009 at 11:12 AM, Erasmo Giambona <e.giambona@gmail.com> wrote:
>> Dear Statalist,
>>
>> I am estimating a model with multiple endogenous variables using panel
>> data. I understand that one cannot use the F-test of excluded
>> instruments in the case of multiple endogenous variables to test for
>> weak identification. I understand that one should use instead the
>> Cragg-Donals F-statistic. What is the intuition of this test? Suppose
>> we have two endogenous variable (X and Y). Is the Cragg-Donals  a
>> joint test of the null hypothesis that we have at least one good
>> instrument for X and one good instrument for Y?
>>
>> Any comments and recommendation for readings would be appreciated?
>
> *
> *   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/
>

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index