Statalist The Stata Listserver


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

Re: st: Basic Autocorrelation Question


From   "Austin Nichols" <[email protected]>
To   [email protected]
Subject   Re: st: Basic Autocorrelation Question
Date   Fri, 8 Jun 2007 15:01:48 -0400

Kelley D Strawn --
One standard approach is to use the cluster(subject) option to
estimate cluster-robust SEs following Arellano and K�zdi or Stock and
Watson. This approach does not change your point estimates, but
produces SEs that allow errors to be arbitrarily correlated within
panel (subject, in your parlance). The approach works even if your
"panel" structure is not individuals over time (the N and T dimensions
are arbitrary) but does require that you have "enough" clusters--10 is
likely too few, 50 is probably fine, 31 may or may not be. Other
approaches are available from within -ivreg2- (-ssc install ivreg2,
replace- and -ssc install xtivreg2, replace- and read the excellent
help files), and note that -ivreg2- and -xtivreg2- do not require you
to specify any endogenous variables, or instruments, i.e. they run OLS
etc.  There are a variety of other options in -help xt- and more to
come in Stata 10.

Arellano. 1987. "Computing robust standard errors for within-groups
estimators." Oxford Bulletin of Economics and Statistics. Vol. 49. No.
4. p. 431-434

G�bor K�zdi. 2004. "Robust Standard Error Estimation in Fixed-Effects
Panel Models." Hungarian Statistical Review Special(9): 96-116.
http://portal.ksh.hu/pls/portal/docs/PAGE/STATSZEMLE/STATSZEMLE_ARCHIVUM/2004_ARCHIVUM/2004_SPECIAL_S9_ARCHIVUM/KEZDI.PDF

James H. Stock and Mark W. Watson. 2006. "Heteroskedasticity-Robust
Standard Errors for Fixed Effects Panel Data Regression." NBER
Technical Working Paper No. 323. [http://www.nber.org/papers/T0323]

On 6/8/07, Kelley D Strawn <[email protected]> wrote:
Hi,

I'm hoping someone can give me a starting point as I try to teach myself
how to make corrections for temporal autocorrelation in OLS models using
Stata. Tests tell me its there, but I am new at how to "remove" it so that
my estimators are more reliable. Is there a standard command or procedure
that I can familiarize myself with? The data structure is fairly simple:
31 subjects observed once a month for 24 months.

Thanks-
*
*   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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index