GINI calculation using inequal7

[Nerea Gomez]

Good morning,

I am interested in calculating the potential differences in the gini index for networth before and after an increase in inflation.

For that, I have my original networth values and then I create a new variable considering an increase in inflation of +10% (I do this by dividing my original variable by 1.10). I get the new variables in which I can clearly see that those who had a negative networth now have a higher value (they improve), but those who what a positive networth get now lower values (worse situation). Thus, I expect to calculate the Gini and obtain that it has decreased since all the values get closer after the increase in inflation. However, when I compute inequal7 for the original variable and the new one, I get the same Gini Index (and the other measures too) for both variables! Stata is supposed to use negative values (it states...has 511 values < 0 *used* in calculations) so I don't understand the issue.

If anyone has experience with this problem or knows the answer please I would really appreciate your help.

Thanks a lot,

Nerea.

[Ivica Rubil]

Dear Nerea,
there is actually no problem at all. You should know that the Gini coefficient is a measure of relative, as opposed to absolute, inequality. As such, it does not change when you multiply all values of your variable by the same factor. And you have done exactly that - multiplied all values of net worth by 1/1.1. The relative differences, namely the ratios, between different values do not change and thus the Gini does not change. Of course, the absolute differences are smaller, but that does not affect the Gini as it is a measure of relative inequality. If you wish to measure absolute inequality, you should use an appropriate measure, such as the variance or the Gini multiplied by the mean.
Best,
Ivica

[Nerea Gomez]

Dear Nerea,
there is actually no problem at all. You should know that the Gini coefficient is a measure of relative, as opposed to absolute, inequality. As such, it does not change when you multiply all values of your variable by the same factor. And you have done exactly that - multiplied all values of net worth by 1/1.1. The relative differences, namely the ratios, between different values do not change and thus the Gini does not change. Of course, the absolute differences are smaller, but that does not affect the Gini as it is a measure of relative inequality. If you wish to measure absolute inequality, you should use an appropriate measure, such as the variance or the Gini multiplied by the mean.
Best,
Ivica

Thanks a lot for your answer Ivica.

Do you know which commands should I use in Stata to measure absolute inequality? I am really interested in studying the impact of inflation so I would really like to find a way to compute it.

Thanks again,

Nerea.

[Ivica Rubil]

Dear Nerea,

it depends on which measure of absolute inequality you want to use. One of them is the variance. Another is the Absolute Gini coeff. If you opt for the latter, then take a look at Philippe Van Kerm's -sgini- command (ssc install sgini), which has the option to calculate the abosulute instead of the relative Gini coeff. In addition, you may choose the degree of inequality aversion, as the -sgini- command computes the so called S-Gini class of inequality measures (with "S" standing for "single-parameter").

Sorry for replying only now, I haven't been checking Statalist for quite a while.

Ivica