Announcement

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

  • Calculating share of middle class (equivalising households)

    Dear Statalist community,

    I am an inexperienced coder new to Stata, who is trying to replicate and extend an existing paper. The first step is to calculate the share of the middle class for a number of countries and time periods using the LIS database. The definition of the middle class that I am using is households with an income between 75 percent and 125 percent of the median household income. In order to make households comparable, I'm using the OECD household equivalence scale (the oxford scale): 1.0 to the first adult; 0.7 to the second and each subsequent person aged 14 and over and 0.5 to each child aged under 14. However, so far I haven't managed to get the same numbers as the authors of the other paper and I'm not sure if I'm taking the right steps.

    use dhi nhhmem13 nhhmem hpopwgt using $ca91h
    sum dhi [w=hpopwgt*nhhmem], de
    gen edhi = dhi/(1+(0.7*nhhmem-1-nhhmem13)+0.5*nhhmem13)
    sum edhi [w=hpopwgt*nhhmem], de
    gen weight = hpopwgt*nhhmem
    sum weight if edhi >= .75*r(p50) & edhi <= 1.25*r(p50)
    display r(sum)
    Above is the code, where:
    dhi = disposable household income
    hpopwgt = household weight
    nhhmem = number of household members
    nhhmem13 = number of household members 13 or younger
    $ca91h = indicates that it is Canada 1994

    So how do I retrieve the middle class share from this information, I thought I could perhaps simply divide the sum of the weights fulfilling the if-criteria with the sum of weights given when summarizing the equivalized household income (without if-criteria)? Is that how you would go about this or am I missing some essential step?

  • #2
    Veronika:
    welcome to this forum.
    The only advice that I can provide is to search for Stephen Jenkins ' posts on this forum.
    Let's hope that he will chime in.
    Kind regards,
    Carlo
    (Stata 19.0)

    Comment


    • #3
      What is the unit of analysis? Are you intending to attribute each /individual/ with the equivalized income of the household to which s/he belongs? (Or is this a distribution of household income among households?) My guess it's the former, and that the weights gross up to a population of /individuals/, not households.
      I don't understand your use of the weights to define middle class. Does the following produce the sort of thing you're after?

      Code:
      gen wgt = hpopwgt*nhhmem
      sum edhi [w=wgt], de
      ge middleclass = 0 if !missing(edhi)
      replace middleclass = 1 if edhi  >= .75*r(p50) & edhi <= 1.25*r(p50)
      label def middleclass 0 "not middleclass" 1 "middleclass"
      lab val middleclass middleclass
      tab middleclass [w = wgt]

      Comment

      Working...
      X