Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

Re: st: Check heteroskedasticity after ivregress gmm


From   Jorge Eduardo Pérez Pérez <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: Check heteroskedasticity after ivregress gmm
Date   Mon, 24 Feb 2014 15:44:30 -0500

Why do you want to test for heteroskedasticity after using a robust
variance estimator?

If you implement a estimator that isn't robust to heteroskedasticity,
then it is a good idea to test for heteroskedasticity afterwards to
check whether your absence of heteroskedasticity assumption is
supported by the data.

But since you implemented a robust estimator, you are acknowledging
that there is heteroskedasticity. No point in performing a test of the
null of homoskedasticity against the alternative of heteroskedasticity
if you assumed heteroskedasticity in the beginning.

In other words, your workflow should be

* Estimate under homoskedasticity
ivreg2 y (endo = iv) X
* Test for heteroskedasticity
ivhettest

* If evidence of heteroskedasticity, then use robust estimator, with either
ivreg2 y (endo = iv) X, gmm robust
* Or
ivregress gmm y (endo = iv) X ,vce(hac nw opt)

Also notice that if you use the gmm option in -ivregress- the
resulting standard errors are robust by default.

--------------------------------------------
Jorge Eduardo Pérez Pérez
Graduate Student
Department of Economics
Brown University


On Fri, Feb 21, 2014 at 11:56 PM, Aksorn Lueanyod <[email protected]> wrote:
> Dear all,
>
> I am doing the estimation by using GMM with the robust Newey-West HAC variance
>
> Does anyone know the command to check heteroskedasticity after the command (STATA 12)
> ivregress gmm y (endo = iv) X ,vce(hac nw opt) ....(1)
>
> I know that the command "ivhettest" is work after ivreg2
> So, I try to replicate (1) by using
> ivreg2 y (endo = iv) X ,gmm robust bw(#) ... (2)
>
> "ivhettest" is work but I have to put # as a number that received from automatic lag choosing by Bartlett kernel  from (1)
> (I tried bw(#) as bw(auto), it didn't work)
>
> Please suggest the command to check heteroskedasticity for (1) or the command that (2) can choose automatic lag.
> Thank you in advance
>
> Best regards,
> Aksorn
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index