Re: st: Index Construction in Stata

[Stas Kolenikov <skolenik@gmail.com>]

May be something like this:

webuse nlswork
gen wage = exp(ln_wage)
bysort year (id): cumul wage [aw=1], generate(cdf)
bysort id (year): gen cdf_diff = abs( cdf-cdf[_n-1] )
mean cdf_diff [pw=1], over(year)

You would need to substitute your own sampling weight wherever necessary.

Just out of curiosity: the above calculations give the income mobility
values of about 0.25-0.3. Is that considered low or high? I know that
a Gini index of 0.2 is a socialism (aka Nordic countries), and 0.5 is
one of US, Russia, Brazil or South Africa. This index appears to be
very similar in spirit, except that the difference is computed not
with the random person from the distribution, but with oneself a year
ago. What are the typical values of the income mobilities?

On Sat, Apr 14, 2012 at 11:20 AM, Krichevskiy, Dmitriy
<krichevskiyd@etown.edu> wrote:
> Dear members of the Stata community,
>
> I am trying to figure out how to construct in index in Stata. I want to implement Mobility index developed by Richard Dickens (see, Caught in a Trap? Wage Mobility in Great Britain: 1975-1994, Economica Vol 67. No 268 [Nov 2000] p492)
>
> The index construction is a fraction with the numerator -  double the sum of all individual absolute values derived from differencing income CDFs from one year to the other. Denominator - number of individuals.
>
>  I have a panel dataset of individuals with income reported with some periods missing for number of individuals. I am currently still using 11th version of Stata.

-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/