Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: RE: ivreg2 questions


From   Robert Davidson <rhd773@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: ivreg2 questions
Date   Mon, 19 Mar 2012 19:25:28 -0400

Thank you for the response.

I was looking at the Cragg Wald F-statistic, testing if the equation
is weakly identified, and I did check the correlation with of the
endogenous regressor and the instrument.  I had always read that the
correlation generally needs to be high between the endogenous
regressor and the instrument in order for it to be a good instrument.
Is it just that the partial correlation between the two should be
high?

Thank you,
Rob

On Mon, Mar 19, 2012 at 6:53 PM, Schaffer, Mark E <M.E.Schaffer@hw.ac.uk> wrote:
> Rob,
>
>> -----Original Message-----
>> From: owner-statalist@hsphsun2.harvard.edu
>> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of
>> Robert Davidson
>> Sent: 19 March 2012 21:38
>> To: statalist@hsphsun2.harvard.edu
>> Subject: st: ivreg2 questions
>>
>> Dear Statalist,
>>
>> I am using ivreg2 (3.1.03) authors cfb & mes and have a few questions.
>>
>> My model is X = A + B + C (1)
>> The endogenous regressor, C, is binary, and I am using 1
>> continuous instrument (D) for C.
>>
>> Q1) does this version of ivreg2 not produce the Stock-Yogo
>> relative bias values?  My output only include the size values.
>
> It does.  However, the maximal IV relative bias test requires that the estimation be sufficiently overidentified (the reason is that the IV estimator has moments only up to the degree of overidentification).  You've got an exactly-identified model, so the relative bias test isn't well defined.
>
>> Q2) I have had a difficult time finding valid instruments for
>> C.  I have tried several that had reasonably high and
>> significant correlations with C and no significant
>> correlation with the error term from equation (1), but my
>> f-stats were often around 7 or 8 (reasonably weak
>> instruments).  Then, I tried another variable that was not
>> highly correlated with C and the f-stat was about 25 and the
>> results were similar to the main estimation.  Is it common
>> that a variable that does not seem strongly associated with
>> the endogenous regressor can serve as such a strong
>> instrument or am I doing something wrong?
>
> This might just be a misunderstanding.  The F stat is a kind of measure of partial correlation, i.e., with the exogenous covariates partialled out.  Perhaps when you were checking the correlations of the IVs with C, you were looking at correlations instead of partial correlations?
>
> HTH,
> Mark
>
>> I know I cannot
>> test for exogeneity, but this new variable does not seem
>> correlated with the error term from equation (1).
>>
>> I apologize that this may have been sent twice.
>>
>> Thank you,
>> Rob
>>
>> *
>> *   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/
>>
>
>
> --
> Heriot-Watt University is the Sunday Times
> Scottish University of the Year 2011-2012
>
> We invite research leaders and ambitious early career
> researchers to join us in leading and driving research
> in key inter-disciplinary themes. Please see
>
>    http://www.hw.ac.uk/researchleaders
>
> for further information and how to apply.
>
> Heriot-Watt University is a Scottish charity
> registered under charity number SC000278.
>
>
> *
> *   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   |   Site index