Regression based Decomposition

[Neeraj Kumar]

Hi
I am using the following command for the regression based inequality decomposition by factors. In this command coefficient of variation is computed and decomposed and if we use second command then half the squared coefficient of variation is computed and decomposed. I want to ask is there any command to measure other inequality measures like theil coefficient, mean log deviation etc. for decomposition of inequality by factors. My education, social group and religion is categorical variable and rest of the variables are continuous.

ineqrbd income age i.education i.religion i.social group employment income pension labor income

ineqrbd income age i.education i.religion i.social group employment income pension labor income, i2

[Stephen Jenkins]

You are referring to ineqrbd, a user-written program available on SSC. (Please read the Forum FAQ; you are enjoined to tell us the provenance of non-built-in commands.)

is there any command to measure other inequality measures like theil coefficient, mean log deviation etc. for decomposition of inequality by factors.

Please read the literature cited in the References shown in the help file. If you did, you wouldn't be asking your question. The decomposition rule, based on Shorrocks' original ideas, does not refer to a specific inequality index.

[Neeraj Kumar]

Thank you for your suggestion and I follow that and keep it in mind. It means a lot.

[Milan Das]

Hi
Stephen Jenkins,
When we will use the regression-based decomposition?


Milan

[Stephen Jenkins]

Milan: I suggest you look at some of the literature references references in the help-file. Also be aware that other methods have been developed since that work, including e.g. 'distributional regression'. On that, please also look at relevant literature. (There have also been presentations at Stata Conferences on the topic such as those by Philippe Van Kerm.)