of logarithms
>
>I'm guessing you want -ineqdeco- but try -findit inequality- to see
>more options.
>
>>On 2/22/06, $B@>B<!!9,K~(B <[email protected]> wrote:
>> Statalist members,
>>
>> I am looking for a stata module to compute variance of logarithms on
>> personal income.and to compute this inequality decompositions by
subgroups.
>> Or please show me how to do.
>>
>> Yukimitsu Nishimura
Thanks for the plug, Austin. I am fairly confident in recommending
-indecdeco- to Yukimitsu Nishimura (and others) for inequality index
calculations with decompositions by population subgroup -- that is what
it is designed to do. Moreover, note that the variance of logarithms is
usually deprecated as an inequality measure because it does not always
satisfy the 'principle of transfers' axiom. (Contrast this with the
generalised entropy class of measures, which both satisfies the axioms
and is additively decomposable by population subgroup. See the
references in the help file to -ineqdeco-. NB -ineqdeco- assumes all
values are positive. If you wish to avoid this and allow zero values to
be included, use -ineqdec0-. Both programs available via -ssc-.)
I note that many labour economists persist in using the variance of
logarithms measure, despite what the inequality measurement literature
recommends! (Perhaps because labour economists typically estimate
log(wage) equations.)
>Date: Wed, 22 Feb 2006 09:41:29 -0300 (ART)
>From: <[email protected]>
>Subject: Re: st: RE: Re: Help with Ineqdeco
>I`m new and I don`t know what command ineqdeco do. But if someone are
>looking a program for the gini I`ve one
As others have recommended, use Stata tools to discover what particular
programs do, and to find programs that do particular tasks. -findit
<keyword> - is your friend! So too is -search-
For inequality measurement, there are a number of programs available,
including -ineqdeco-, -ineqdec0-, and some nice programs by Philippe Van
Kerm. For poverty measurement, with decompositions, see e.g. -povdeco-.
Philippe Van Kerm and I have also written -glcurve- for plotting
(generalised) Lorenz curves and related graphs (e.g. TIP curves,
concentration curves, etc.)
If you wish to compute standard errors for the estimates ("variance
estimation), you might take a bootstrap approach (some Stata programs
are available). But if you have weighted data, there are some
potentially unresolved issues I think.
Alternatively one can use a linearisation approach. -svygei- and
-svyatk- (-ssc desc svygei_svyatk-) can be used for generalised entropy
and Atkinson indices. (With the subpop() option, you can get estimates
for subgroups too.) For the Gini coefficient and for the Lorenz curve,
use -svylorenz- (also on SSC). Each of the programs cited allows for
complex survey designs. [Another different approach to variance
estimation for the Gini coefficient is taken by Roger Newson's -somersd-
program, and explained in its accompanying documentation. IMHO
-svylorenz- gets there more directly.]
Stephen
-------------------------------------------------------------
Professor Stephen P. Jenkins <[email protected]>
Institute for Social and Economic Research
University of Essex, Colchester CO4 3SQ, U.K.
Tel: +44 1206 873374. Fax: +44 1206 873151.
http://www.iser.essex.ac.uk
Survival Analysis using Stata:
http://www.iser.essex.ac.uk/teaching/degree/stephenj/ec968/
*
* 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/
