Dear statalister, Mark, I have read Baum, Schaffer, Stillman (2003) paper,
but I still don't understand why I should include the same set of
instruments in both the "consistent" and the "efficient" model when I test
for the endogeneity of a subset of regressors. Does someone has an easy
explanation? In any case, when I use -xtabond- how can I retrieve the set
of instruments to plug in in the "efficient" model? Thanks a lot.
Mario
----- Original Message -----
From: Mark Schaffer
To: [email protected] ; M Quagliariello
Sent: Thursday, June 17, 2004 9:59 PM
Subject: Re: st: Testing for endogeneity with xtabond
Mario,
Quoting M Quagliariello <[email protected]>:
Hallo!
I hope someone can help me.
Suppose I want to estimate a dynamic panel with xtabond having
three
sets of variables:
A) surely endogenous
B) surely exogenous
C) maybe endogenous
I was thinking to test for the endogenity of variables C) in this
way:
1) estimate a model in which variables C are considered as
exogenous.
The estimated coefficients should be consistent and efficient if
the
variables are actually exogenous, but inconsistent if the variables
C)
are endogenous (model1).
2) re-estimate the model considering the variables C) as endogenous
and
instrumenting them with their lagged levels (as for the lagged
dependent
variable and the other enedogenous regressors). The estimated
coefficients should be always consistent (model2).
3) test for endogeneity using -hausman model2 model1-
Is it reasonable?
You don't say so explicitly, but it looks like your category C variables
are regressors. In that case, what you want is a test of a subset of
orthogonality conditions. Because the suspect instruments are regressors,
the usual GMM "difference-in-Sargan" or "C test" for a subset of
orthogonality conditions is numerically equivalent to a Hausman test. See
Hayashi's (2000) textbook, pp. 233-34, or the 2003 Stata Journal paper I
did with Kit Baum and Steve Stillman.
You can proceed as you describe, or, since you're using xtabond2, even
more easily, just calculate the difference of the Sargan-Hansen statistics
of the two estimations. This will be distributed as chi-sq with
dof=number of suspect instruments.
IMPORTANT - you must make sure that every instrument that appears in model
2 is also an instrument in model 1. You want to use lagged C variables as
instruments when estimating model 2, so that means that you have to use
exactly these lagged variables as instruments when estimating model 1 as
well (even though the C variables are being treated as exogenous).
--Mark
Thanks a lot,
Mario
*
* 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 +44-131-451-3008
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
your possession and notify the sender by reply e-mail. Heriot
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/
