Estimation of the Concentration Index based on Kakwani and Wagstaff formulae.
concindexg [varlist] [if] [in] [weight] [ , welfarevar(varname) drop format(format) splitvars(varlist) se clean }
where :
welfarevar is the welfare variable - not optional - ,
drop is to clean the welfare indicator from missings,
format is the format type,
splitvars is to generate dummies from categorical variables,
se is to compute also standard errors of the concentration index,
clean is to erase matrices and scalars created by the program.
Description
concindexg computes the concentration index on grouped data. See work by Wagstaff and Kakwani.
Formulas
The concentration index, in the case of no standard deviations is computed as:
C = 2 cov(yi,Ri) / µ
yi is the individual i's indicator variable (e.g. health status),
µ its mean
Ri his fractional rank in the socioeconomic distribution
Standard errors of C are obtained by:
1 1 2 2 Var(C) = - [ - sum (a ) - (1 + C) ] n n
where :
yi a = - ( 2 Ri - 1 - C) + 2 - qi-1 - qi µ
and :
1 qi = - sum(yi) µn
is the ordinate of the concentration curve L(p), and q0 = 0.
Examples
. concindexg has_water has_elec, welf(quint)
. concindexg has_water has_elec, welf(quint) dr se cl)
. concindexg has_water [aw=weight], we(quint) se by(urbrur)
. concindexg var1 var2 [aw=weight], we(quint) f(%9.2f)
Saved Results
Concentration indices are saved in matrice r(CIG).
Author
Amadou Bassirou DIALLO, AFTPM, The World Bank -and- CERDI, Univ. of Auvergne (France).
Email: adiallo5@worldbank.org, adiallo5@gmail.com
Note
This program has been initially developped to use the Demographic and Health Surveys (DHS).
Also see
Online: help for concindexi, ineqdeco, inequal, povdeco, ineqerr if installed.
-------------------------------------------------------------------------------