Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • RE: Decomposition of the corrected version of the concentration index

    Dear Statalisters,
    I have been trying to decompose the corrected version of the concentration index suggested by Erreygers (2009) based on the following formula, CI=8*Cov(h,R). where CI=concentration index, h=the health variable of interest (in this case, dummy variable=1 if individual had 4 or more antenatal care visits during pregnancy and 0 otherwise), R=the fractional rank in the socioeconomic status distribution, and Cov(h, R) is the covariance between the health outcome and the rank. I am not too sure if i am doing this right and would greatly appreciate any help.

    I have been trying to follow the procedures laid out in O'Donnell, O., van Doorslaer, E., Wagstaff, A., & Lindelow, M. (2008) and modifying to suit the CI formula as given above but not too sure if i am doing this right. Below, is a copy of what i am doing and would greatly appreciate if anyone can peer review this code to find any mistakes i might be making. Thank you for your time and help.
    ***********************************

    NB: pnc_4vismore is the outcome variable; a binary variable =1 if 4 or more antenatal care visits and 0 otherwise. Analysis is based on Demographic and health survey data for Zimbabwe (2005-2006)

    ***code start here

    use "$inpath/zwe_inequalities_mhc", clear

    ***CONCENTRATION INDEX - CORRECTED VERSION
    egen raw_rank=rank(wealth_score), unique
    sort raw_rank
    quietly summ sweight
    gen wi=sweight/r(sum)
    gen cumsum=sum(wi)
    gen wj=cumsum[_n-1]
    replace wj=0 if wj==.
    gen rank=wj+0.5*wi
    drop raw_rank wi cumsum wj

    qui sum pnc_4vismore [aw=sweight]
    scalar mean=r(mean)
    cor pnc_4vismore rank [aw=sweight], covariance
    sca CI=8*r(cov_12)
    display "concentration index by convenient covariance method", CI

    ***Computing the semi-elasticities, CI and contributions of each factor
    clear matrix
    global X "wealthq1 wealthq2 wealthq3 wealthq4 age20to24 age25to29 age30to34 age35to39 age40to44 age45to49 educ_primary educ_secondary marstat_marr marstat_sep woman_working religion_christian fnews fradio family_plan preg_prev_terminated dum_urban region1 region2 region3 region4 region5 region6 region7 region8 region9"

    regress pnc_4vismore $X [pw=sweight], robust cluster(psu)

    *******decomposition starts here
    foreach x of global X {
    qui {
    scal b_`x'=_b[`x']
    corr rank `x' [aw=sweight], c
    sca cov_`x'=r(cov_12)
    sum `x' [aw=sweight]
    sca elas_`x'=(b_`x'*r(mean))
    sca CI_`x'=8*cov_`x'
    sca con_`x'=4*(elas_`x'*CI_`x')
    sca prcnt_`x'=(con_`x'/CI)*100
    }
    di "`x' elasticity:", elas_`x'
    di "`x' concentration index:", CI_`x'
    di "`x' contribution:", con_`x'
    di "`x' percentage contribution:", prcnt_`x'
    matrix Aaa = nullmat(Aaa) \ ///
    (elas_`x', CI_`x', con_`x', prcnt_`x')
    }
    matrix rownames Aaa= $X
    matrix colnames Aaa = "Elasticity""CI""Absolute""%"
    matrix list Aaa, format(%8.4f)
    *********

    Thank you for your help.

  • #2
    Welcome to Statalist. Please read the FAQ from top to bottom (hit the black bar at the top of the page), and note the advice about providing full references to all programs and literature that you cite. Also, please post Stata input and output using "CODE" delimiters for legibility's sake.

    For calculating the Erreygers (2009) CI and related variants, look at conindex on SSC (by O'Donnell et al.). But note that there are fundamental questions about whether the standard CI should be applied to binary outcome data, as you have. See: "On correcting the concentration index for binary variables" by Gustav Kjellssona and Ulf-G. Gerdtham, Journal of Health Economics, 32(2013), 659-670.

    Comment


    • #3
      Conindex is very useful for computing the different versions of the concentration index. But it doesn't do decompositions of these different versions of the concentration indicex - (as far as I'm aware).

      Comment


      • #4
        Originally posted by Marshall Makate View Post
        Dear Statalisters,
        I have been trying to decompose the corrected version of the concentration index suggested by Erreygers (2009) based on the following formula, CI=8*Cov(h,R). where CI=concentration index, h=the health variable of interest (in this case, dummy variable=1 if individual had 4 or more antenatal care visits during pregnancy and 0 otherwise), R=the fractional rank in the socioeconomic status distribution, and Cov(h, R) is the covariance between the health outcome and the rank. I am not too sure if i am doing this right and would greatly appreciate any help.

        I have been trying to follow the procedures laid out in O'Donnell, O., van Doorslaer, E., Wagstaff, A., & Lindelow, M. (2008) and modifying to suit the CI formula as given above but not too sure if i am doing this right. Below, is a copy of what i am doing and would greatly appreciate if anyone can peer review this code to find any mistakes i might be making. Thank you for your time and help.
        ***********************************

        NB: pnc_4vismore is the outcome variable; a binary variable =1 if 4 or more antenatal care visits and 0 otherwise. Analysis is based on Demographic and health survey data for Zimbabwe (2005-2006)

        ***code start here

        use "$inpath/zwe_inequalities_mhc", clear

        ***CONCENTRATION INDEX - CORRECTED VERSION
        egen raw_rank=rank(wealth_score), unique
        sort raw_rank
        quietly summ sweight
        gen wi=sweight/r(sum)
        gen cumsum=sum(wi)
        gen wj=cumsum[_n-1]
        replace wj=0 if wj==.
        gen rank=wj+0.5*wi
        drop raw_rank wi cumsum wj

        qui sum pnc_4vismore [aw=sweight]
        scalar mean=r(mean)
        cor pnc_4vismore rank [aw=sweight], covariance
        sca CI=8*r(cov_12)
        display "concentration index by convenient covariance method", CI

        ***Computing the semi-elasticities, CI and contributions of each factor
        clear matrix
        global X "wealthq1 wealthq2 wealthq3 wealthq4 age20to24 age25to29 age30to34 age35to39 age40to44 age45to49 educ_primary educ_secondary marstat_marr marstat_sep woman_working religion_christian fnews fradio family_plan preg_prev_terminated dum_urban region1 region2 region3 region4 region5 region6 region7 region8 region9"

        regress pnc_4vismore $X [pw=sweight], robust cluster(psu)

        *******decomposition starts here
        foreach x of global X {
        qui {
        scal b_`x'=_b[`x']
        corr rank `x' [aw=sweight], c
        sca cov_`x'=r(cov_12)
        sum `x' [aw=sweight]
        sca elas_`x'=(b_`x'*r(mean))
        sca CI_`x'=8*cov_`x'
        sca con_`x'=4*(elas_`x'*CI_`x')
        sca prcnt_`x'=(con_`x'/CI)*100
        }
        di "`x' elasticity:", elas_`x'
        di "`x' concentration index:", CI_`x'
        di "`x' contribution:", con_`x'
        di "`x' percentage contribution:", prcnt_`x'
        matrix Aaa = nullmat(Aaa) \ ///
        (elas_`x', CI_`x', con_`x', prcnt_`x')
        }
        matrix rownames Aaa= $X
        matrix colnames Aaa = "Elasticity""CI""Absolute""%"
        matrix list Aaa, format(%8.4f)
        *********

        Thank you for your help.
        Hi, I am using the same comment in DHS data. But I getting problem that "varlist not allowed". Please let me know how I will solve this problem. I look forward to your reply.

        Comment


        • #5
          can anyone help in decomposing the CI for binary variable. I would like to apply Wagstaff decomposition method. I have understood Wagstaff normalized concentration index and learnt Coindex command as well. However, I would like to decompose the same further across various socio-demographics. looking forward . Thanks

          Comment


          • #6
            Hi Aashima Singhal , can you help me on calculation of concentration index. I am working on inequality in 'use' and 'access' of credit and debit cards in India. And even if you got answer for decompose method for various socio-demographics.

            Comment


            • #7
              Hi Aashima Singhal , can you help me on calculation of concentration index. I am working on inequality in 'use' and 'access' of credit and debit cards in India. And even if you got answer for decompose method for various socio-demographics.

              Comment

              Working...
              X