Johannes Schoder:
Stephen points out that you do not have micro data, which presents
problems. I am curious about what kind of inequality you hope to
capture by using the population dying each year as the population and
their age at death as the outcome of interest (analogous to income),
which conflates cohorts and ages. A more natural approach would be to
construct estimates of cohort-specific survival and to compute
inequalities across cohorts, or within cohort across country. Can you
clarify what you hope to accomplish?
Also, note that survival curves encompass a form of inequality
measurement, since given an equal distribution of age at death, the
survival curve would drop from one to zero at a single age; an unequal
distribution of age at death results in a number of steps in the
survival curve--and the Gini coef seems a poor approach to modeling
that inequality. At first glance, anyway--perhaps you have a good
theoretical justification.
On Sat, Apr 26, 2008 at 6:32 AM, Stephen P. Jenkins
<[email protected]> wrote:
> > Date: Thu, 24 Apr 2008 11:05:58 +0200
> > From: Johannes Schoder <[email protected]>
> > Subject: st: Gini coefficient of survival tables
> >
> > Dear Stata list users,
> > I am calculating the Gini coefficient of the survival tables
> > for 30 OECD countries for the time period of 1960-2004 using the Human
> Mortality
> > Database. I take years lived (from year 0 to 110) as income and death
> > numbers as population.
> > Thereby I use the Stata command "ginidesc death numbers, by(year)".
> > Unfortunately I receive totally different results when I
> > calculate the Ginicoefficient "by feet" with Excel.
> > So I am not sure if I used the command properly. Thanks a lot in
> advance!
> >
> > Johannes
> >
> > P.S.:
> >
> > The data is having the following structure:
> >
> >
> > Year Years lived Death Numbers
> > 1960 1 3400
> > 1960 2 478
> > 1960 3 488
> > .
> > .
> > .
> >
> >
> > 1961 1 3200
> > 1961 2 470
> > 1961 3 472
>
> --------------------------------------
>
> It seems that you have grouped data, not unit-record ("micro") data. All
> the Stata programs for the calculation of Gini coefficient that I am
> aware of assume that you have unit-record data. So, you will
> underestimate inequality if you use those programs. On the general
> issues, see FA Cowell and F Mehta (1982) "The estimation and
> interpolation of inequality measures", Review of Economic Studies, 49(2)
> 273-290, and references cited therein.
>
> By the way, -ginidesc- is, in many ways, simply a wrapper for -ineqdeco-
> (on SSC).
*
* 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/