# Re: st: IV and Hausman Test

 From Mark Schaffer <[email protected]> To [email protected], Dongshu Ou <[email protected]> Subject Re: st: IV and Hausman Test Date Fri, 24 Dec 2004 14:53:46 +0000 (GMT)

```Dongshu,

Quoting Dongshu Ou <[email protected]>:

>   Hi,
>
>   I have some questions on IV regression.
>
>   I tried different instruments sets, can based on the first stage F
> value and the significance of the instruments to determine which is
> weak instrument? Normally how do you decide which is the best
> instrument for the model?

You might want to have a look at the paper Kit Baum, Steve Stillman and I
did on IV regression in Stata, and download the IV program we wrote for
Stata, -ivreg2-.  The paper was published in the Stata Journal, 3:1 2003,
but you can get the working paper version at
saying -findit ivreg2- in Stata and then following the links.

With respect to weak instruments, if you have a single endogenous regressor,
it's not hard to tell which are weak just by looking at the first-stage
regression.  The first-stage F is a test of the joint strength of the
excluded instruments, but the individual t's will tell which which are the
strongest.

>
>   Different IV sets also make my sub-samples have differnt size, can I
> conclude that the dependent variable are due to the sample size but
> not the choice of IV? (I am not sure whether my question is clear
> or not)

I'm not sure what you mean, but you need to think about sample selection
biases here.  If the missing values are purely random, though, then you

>   The last question is that, I run IV regression, then I run OLS based
> on the selected sample (reg Y X if e(sample) ) and then use Hausman test.
>
>   however, the Chi2 i got under every IV regressions are very small.
> and the prob>chi2 are all greater than .8. what shall I do? Is it
> meaningful to interpretate the 2sls results then?

This just means that there isn't much difference between your IV and OLS
coefficients, either because the coeffs are very similar and/or because the
standard errors are large.  Whether or not your IV results are meaningful
depends on the context.  If, for example, the standard errrors in your IV
equation are very large and none of the coefficients is significant, you
might conclude that the IV results aren't very meaningful.  If the SEs are
small and the IV equation passes the other relevant specification tests,
then you might conclude that the IV results are meaningful but that the OLS
results are preferable on efficiency grounds.

Hope this helps.

--Mark

>
>   thank you so much,
>
>   Dongshu
> *
> *   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/
>

Prof. Mark Schaffer
Director, CERT
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3294
email: [email protected]
web: http://www.sml.hw.ac.uk/ecomes
________________________________________________________________

DISCLAIMER:

This e-mail and any files transmitted with it are confidential
and intended solely for the use of the individual or entity to
whom it is addressed.  If you are not the intended recipient
you are prohibited from using any of the information contained
in this e-mail.  In such a case, please destroy all copies in
Watt University does not accept liability or responsibility
for changes made to this e-mail after it was sent, or for
viruses transmitted through this e-mail.  Opinions, comments,
conclusions and other information in this e-mail that do not
relate to the official business of Heriot Watt University are
not endorsed by it.
________________________________________________________________
*
*   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/
```