Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE : Heteroskedasticity and fixed effects (was: st: RE: Re: Weakinstruments)


From   GaulÚ Patrick <patrick.gaule@epfl.ch>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   RE : Heteroskedasticity and fixed effects (was: st: RE: Re: Weakinstruments)
Date   Thu, 17 Jul 2008 13:50:12 +0200

Dear Alice,

I would not worry about testing for heteroskedasticity. In practice, it just makes more sense to always use robust standard errors.

Stock and Watson puts it best in their introductory econometrics textbook:

"Economic theory rarely gives any reason to believe that the errors are homoskedastic. It therefore is prudent to assume that the errors might be heteroskedastic unless you have compelling reasons to believe otherwise. (...) If the homeskedasticity-only and heteroskedasticity-robust standard errors are the same, nothing is lost by using the heteroskedasticity-robust standard errors; if they differ, however, then you should use the more reliable ones that allow for heteroskedasticity. The simplest thing, then, is always to use the heteroskedasticity-robust standard errors."

Best regards,

Patrick GaulÚ
Research associate, EPFL


________________________________________
De : owner-statalist@hsphsun2.harvard.edu [owner-statalist@hsphsun2.harvard.edu] de la part de aapdm [aapdm_999@yahoo.co.uk]
Date d'envoi : jeudi 17 juillet 2008 13:12
└ : statalist@hsphsun2.harvard.edu
Objet : Re: Heteroskedasticity and fixed effects (was: st: RE: Re: Weak instruments)

Dear Mark,

Many thanks for this.

It seems to me that Stata 10 now adjusts standard
errors according to the paper by Stock and Watson.
Actually the standard errors reported after xtreg,fe
are different between Stata 9 and Stata 10.

However to know whether I should adjust standard
errors for heteroskedasticity (for instance using
cluster, as you suggest) I would first of all test
whether there is evidence of heteroskedasticity or
not.

I was wondering which test or command I should be
using. I found the xttest3 that can be used after
xtreg, or whitetst after reg.

I was wondering whether those tests would be reliable,
assuming that Stata 10 does adjust standard errors
correctly.

Any thought?

Many thanks.

Alice.

--- "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk> wrote:

> Alice,
>
> > -----Original Message-----
> > From: owner-statalist@hsphsun2.harvard.edu
> > [mailto:owner-statalist@hsphsun2.harvard.edu] On
> Behalf Of aapdm
> > Sent: 16 July 2008 23:34
> > To: statalist@hsphsun2.harvard.edu
> > Subject: Re: st: RE: Re: Weak instruments
> >
> > Hi
> >
> > My apologies if this email gets there twice but I
> think I
> > lost the one I wrote before.
> >
> > I am using xtreg, fe and would like to test for
> > heteroskedasticity after the estimation.
> >
> > Can someone tell me how to do that? Also, if I add
> the robust
> > command after xtreg, I would expect not to be able
> to reject
> > the null of homoskedasticity (for instance when
> using reg, fe
> > and including the dummies) but this is not the
> case: if I use
> > xttest3, I can't reject the null whether I add
> robust or not
> > after the estimation.
>
> A few thoughts here...
>
> - You should put a new title in the subject field
> when starting a new
> thread.  It misleads readers of the list and messes
> up threading if you
> don't.
>
> - You probably don't want to use -robust- with
> -xtreg,fe-.  Stock and
> Watson 2007 (see their NBER paper -
> http://www.nber.org/papers/t0323)
> show that the standard robust var-cov matrix is
> inconsistent.  In other
> words, the SEs are wrong.  They recommend either a
> modified vcv (which
> isn't implemented by -xtreg- the last I checked) or
> using the
> cluster-robust vcv.  The latter is easy to do and
> will get you SEs that
> are robust to clustering as well as
> heteroskedasticity.
>
> - The standard
> White/Koenker/Breusch-Pagan/Godfrey/Cook-Weisberg
> tests
> for heteroskedasticity (which is what you probably
> had in mind for
> testing) will probably also be inconsistent after
> fixed effects
> estimation, for the same reason that the robust vcv
> is inconsistent.  At
> their heart, they are contrasting the robust vcv
> with the classical
> (assumed homoskedasticity) vcv.  Under the null of
> no
> heteroskedasticity, there shouldn't be much
> difference.  But if the
> robust vcv is inconsistent, it's easy to see that
> you could reject the
> null even if you have homoskedasticity.
>
> Bottom line: the easiest thing to do is to switch to
> using -cluster-
> after -xtreg,fe- (or after -xtivreg2,fe-, if you'll
> pardon the
> self-plug).
>
> Hope this helps.
>
> Cheers,
> Mark
>
> Prof. Mark Schaffer
> Director, CERT
> Department of Economics
> School of Management & Languages
> Heriot-Watt University
> Edinburgh EH14 4AS
> tel +44-131-451-3494 / fax +44-131-451-3296
> http://ideas.repec.org/e/psc51.html
>
> > Can someone help me?
> >
> > Many thanks. Alice.
> >
> >
> >
>
__________________________________________________________
> > Not happy with your email address?.
> > Get the one you really want - millions of new
> email addresses
> > available now at Yahoo!
> http://uk.docs.yahoo.com/ymail/new.html
> >
> > *
> > *   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 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/
>



      __________________________________________________________
Not happy with your email address?.
Get the one you really want - millions of new email addresses available now at Yahoo! http://uk.docs.yahoo.com/ymail/new.html

*
*   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