Knowing the average income per capita or the median income is a rather poor economical indicator of a country, even when given in PPP\$. Knowing the (dimension-less) Gini index is somehow better, but it's too abstract a quantity and doesn't - as a number - tell too much. The Lorenz curve from which the Gini index is derived contains a bit more information but it is still hard to grasp: "What exactly do these curves tell me?"

enter image description here

The most objective and intuitive way of displaying income statistics possibly is just plotting the number of people over income (blue) and the accumulated income per income group (silver):

enter image description here

Unfortunately I find such plots only rarely. What I am looking for is a database where I can find the appropriate data (e.g. relative sizes of 10 or 20 income groups) for ideally all countries of the world.

The World Bank seems to publish only income shares held by the kth 20% which is better than nothing and does allow to draw Lorenz curves. But it would not allow to draw more detailed income distributions.

Where do I find such data: number of capita per income (with appropriate income bins: < 10%, < 20%, .... > 90% of the average (or median) income).

  • $\begingroup$ From your question it sounds like you have Lorenz curve for each country. If you have that and total population, you can back out the income histogram for a country. $\endgroup$ – Art Oct 29 '19 at 5:33
  • $\begingroup$ @Art: Can you give me a hint how to do this, e.g. given the income share of the five quintils by 2.4%, 4.8%, 8.2%, 16.5%, 68.2% as for South Africa in 2014 (according to the World Bank - resulting in a Gini index of 63.0. How would I calculate or estimate from these numbers the percentage of people/households with 20%, 50%, 100%, 200%, 500%, 1,000%, 2,000% and 5,000% of the average (or median) income? $\endgroup$ – Hans-Peter Stricker Oct 29 '19 at 14:00
  • $\begingroup$ Suppose $f(x)$ is the PDF of your hisogram. The lorenz curve is just the CDF of the function, and you can take derivative to back out the PDF. This requires you to have a reasonably fine-binned Lorenz curve though, like percentile, instead of quintiles. $\endgroup$ – Art Oct 30 '19 at 7:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.