Theil Index

[Farjana Misu]

Hello Everyone,

I will be grateful if anyone could advance my understanding of STATA in relation to the question below.

How do I write a command to calculate the Theil index to measure the inequality by region for a health indicator variable.


Thanks
Farjana

[Nick Cox]

See if ineqdeco from SSC by Stephen Jenkins does what you want.

[Farjana Misu]

Thank you for your support.

[Abhishek Kumar]

I did not understand that. ineqdeco command gives multiple results including gii coefficients. which one is theil index ? please help. Nick Cox

[Maarten Buis]

That is documented in the helpfile of ineqdeco. If something is unclear, then you type in Stata help program_name, where you replace program_name with the name of the program you have difficulty with. So in this case you type help ineqdeco . In that help-file you can read the following sentence: "GE(1) is the Theil index".

[Nick Cox]

https://en.wikipedia.org/wiki/Theil_index gives me the impression is that this index comes in different flavours. You may need to match the definition you want to the equations in the help and its references and/or the code. As already flagged, Stephen Jenkins is the author here.

[Abhishek Kumar]

that's helpful, thanks for the help Nick Cox Maarten Buis

[Stephen Jenkins]

As Nick points out, and is apparent from the help-file and substantive literature, -ineqdeco-'s Theil index is GE(1). GE(0), the Mean Logarithmic Deviation, is sometimes referred to as Theil's second measure. Confusing. So I don't use that term. To be clear, both belong to the Generalized Entropy family of inequality indices, GE(a), where "a" is an income-senstivity parameter. When a = 0, the index is "middle-sensitive". GE(-1) is markedly bottom-sensitive, GE(2) is markedly top-sensitive ... and so Theil = GE(1) is in between in sensitivity terms. (Read more in the literature.)