st: re: heteroskedasticity questions

[Kit Baum <baum@bc.edu>]

Richard said

-estat hettest- tests for a specific type of heteroskedasticity, i.e.
it tests whether the residual variances go up as yhat goes
up.
...
Finally, Goldfeldt-Quant and -estat hettest- seem to be testing
similar hypotheses.  Goldfeldt-Quant is a little clunky to estimate,
though, so is there any reason I would prefer it over -hettest-?  (I
actually asked this question 3 years ago, but I don't think I got a
definitive answer, or if I did I can't find it anymore).


(1) G-Q is strictly dominated by Breusch-Pagan and should never be  
used. It is essentially a two-sample test of the residual variances  
and is quite crude in comparison to B-P(-Cook-Weisberg).

(2) B-P (estat hettest) need not look only at yhat. You can specify a  
varlist and include the Xs (or any other candidate variables,  
including those with a parabolic shape) in the varlist and have a  
good test of the notion that the dist of the squared errors is indep  
of all of those. The addition of more variables (a la White test)  
reduces the power of the test, but may be more illuminating.

(3) If there is any potential endogeneity in the relationship the B-P  
and similar tests will not be useful. You should look at Mark  
Schaffer's ivhettest (which can reproduce pure B-P, but can also cope  
with heteroskedasticity in an IV setting). findit ivhettest

Kit

Kit Baum, Boston College Economics
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


*
*   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/