Announcement

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

  • How can I do Das Gupta decomposition with Stata rdecompose?

    Hi all,

    I am learning the rdecompose syntax. I am trying to understand it by replicating the examples provided by Gutpa (1994). I can successfully replicate the one-factor example, but have problems in doing the 3-factors example.

    I am using the three-factor example by Gutpa (1994) (I am attaching the data and full paper below). In this example, we want to understand the changes in headship rate between 1970 and 1980, the crude changes in rate is: 4.39. The factors are sex structure, marital status structure, and age structure. Sex have two groups: 1 and 2, marital status have 5: 1-5, and age have 7: 1-7. Each row represents each combination of the three factors. The code I am using is:

    rdecompose size rate, group(population) sum(sex mar age) trans(size) function (size*rate)

    I got the results:
    Click image for larger version

Name:	Picture 1.png
Views:	1
Size:	18.4 KB
ID:	1766450



    It is not the same with reports in Gutpa (1994), like below. In brief, the rate effect is the same. However, I did not decompose the size effects into sex, age, and marital status. How do I correct my codes?
    Click image for larger version

Name:	gupta.png
Views:	1
Size:	141.3 KB
ID:	1766451



    Appreciate all your help!


    Data-example (full dataset please see attachment)
    sex marital status age size rate year
    1 1 1 3239 92.5 1970
    1 1 2 9710 98.47 1970
    1 1 3 9661 99.31 1970
    1 1 4 9501 99.32 1970
    1 1 5 7225 99.39 1970
    1 1 6 3979 99.04 1970
    1 1 7 1742 97.7 1970
    1 2 1 206 18.45 1970
    1 2 2 351 30.48 1970
    ... ... ... ... ... ...
    2 5 5 504 60.91 1980
    2 5 6 480 66.25 1980
    2 5 7 344 74.13 1980
    Attached Files
Working...
X