Income Mountains — v3 – Documentation

This describes how we made our rough calculations of the number of people on different income levels in all countries between 1800 and 2040.

(This data is following our Data Crunching Principles.)

Legal: The reproduction and communication of this data is authorized in all forms by any means under the Creative Common License CC BY 4.0 requiring that the following citation is clearly mentioned: “GAPMINDER: Income mountains v3 – based on World Bank’s Povcal, Maddison, IMF and Zanden see: www.gapminder.org/sources/income-mountains”

They were generated by using a method similar to the method described by Hellebrandt and Mauro in “The Future of Worldwide Income Distribution” April 2015.

For the forecast beyond 2013, we rely on the growth rates from our GDP per capita forecasts, where we have simply taken the IMF’s forecasts to 2022 and extended them to 2040. But in this forecast we apply the GDP per capita growth forecasts on the PovCal average household per capita incomes, which are much lower than GDP per capita numbers, as discussed by Hellebrandt and Mauro 2015. We use the same ginis from PovCal all the way up to 2040.

The log normal distributions were calculated with the method kindly provided by Bas van Leeuwen, as used in the paper he co-authored with Zanden & Foldvari “World Income Inequality 1820-2000“. But when comparing the resulting number of people on different levels against the PovCal survey based numbers, the differences were large. This is because the log-normal assumption is too simplistic. Most countries have a much more narrow central peak of people and far less people on extremely high and extremely low income levels, compared to what a log normal shape looks like. To adjust for this problem, we have adjusted all country distributions to make sure large parts of the world population are aligned with the PovCal numbers.

(to be continued)