I calculated the Gini coefficient through a statistical program (MATLAB), and it came out much larger than I expected. I think I got the Gini coefficient formula right, but I don't know where I went wrong. Here's my calculation process. (I used file 1 with income, household size, and quarters table (1990/1Q-2022/4Q) and file 2 with CPI (2020=100)).

Calculating real income: dividing each the income by the CPI for corresponding quarter and multiply by 100.

Equalizing income: dividing the value obtained above by the square root of the number of household members.

Adding the number of people in the household for all households with the same income.

Calculating the Gini coefficient according to the following formula


$$ \frac{\sum_{i=1}^n f(Y_i) \sum_{j=1}^n f(Y_j)\vert Y_i-Y_j\vert}{2\bar Y} $$

where $Y_i$is the income of ith household, $f(Y_i)$ is the frequency of the people of income, and $\bar{Y}=\sum_{i=1}^{n}{f\left(Y_i\right)Y_i}$

I'm having a hard time keeping up with the reviews. Thank you for your help. I will share the MATLAB code if needed.


1 Answer 1


The steps you describe make sense except for the last one where you presumably forgot $t$ subscript. If you did not forgot it but it was deliberate choice it is very strange one since it does not make sense to look at inequality in pooled cross-section in an economy that presumably experienced growth.

  • $\begingroup$ You're right, I forgot. I distinguished income by period t and found the Gini coefficient for each t separately. The part I'm most unsure about is step1. Is it correct to run step1, which finds the real income at all t first, or after step2, which equalizes the income? $\endgroup$
    – user_A
    Commented Aug 7, 2023 at 6:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.