Michael S. Hanson <[email protected]> wrote that -estat hettest- after
-regress- does not produce the same calculation as the augmented regression
suggested by Wooldridge (2006).
Scott Merryman noted that -estat hettest- is computing the original version
of the Breusch-Pagan test which differs from the version discussed by
Wooldridge (2006). Scott also noted that the original version of the
Breusch-Pagan test requires a normality assumption that the version
discussed by Wooldridge (2006) does not.
To be fair, Wooldridge (2006, page 280) notes that LM test he presents is
due to Koenker (1981) even though it is known as the "Breusch-Pagan test for
heteroskedasticity". Moreover, Wooldridge is clear that the Koenker (1981)
version differs from the original Breusch-Pagan test.
Those who like to follow the literature closely will note that Godfrey
(1988, pages 128-129) derives the "Breusch-Pagan" test as the one computed
by -estat hettest- and also attributes the one discussed by Wooldridge to
Koenker (1981).
--David
[email protected]
References
---------
Godfrey, L.G. 1988. "Misspecification test in econometrics". Cambridge
University Press: Cambridge.
Koenker, R. 1981 "A note on studentizing a test for heteroskedasticity".
Journal of Econometrics 17:107-112.
Wooldridge, J.M. 2006 Introductory Econometrics: A modern approach, 3rd.
Thomson: Mason.
*
* 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/