Inequality Measures : Surveys and data : The Global Income Inequality Project : University of Sussex

This page is organised by world region and country. The bibliographic listing includes our estimates of various inequality measures compiled from the data we have extracted from the surveys. The list of countries for which we have estimated measures of inequality will be updated from time to time.

A note on the inequality measures we estimate

Estimating inequality measures from surveys with individual data is straightforwad, but many historical studies report tables of data on household income and expenditures where the households are grouped by income, or sometimes total expenditure. These grouped data allow us to make estimates of the degree of income/expenditure inequality among the sample of households. With such grouped data we can calculate total income of all households and also the total income shares of each income group. To estimate inequality we then need to use these shares to estimate how income shares would rise across all households if we had data on every household. There are two ways to do this: first, a curve could be fitted to the group shares, or we could simple join the group shares by drawing a line between them. To fit a curve we need a formula for the curve. Here we offer two possibilities: Lognormal (LN) and Beta-Lorenz (BL), described in detail in the technical discussion. We also offer the simplest join-the-dots method (Naive).

A fuller explanation of our methods can be found here:

Estimating inequality among households from grouped data reported in historical surveys. [PDF 503.29KB]

Once we have these curves and the join-the dots estimates of the sample income/expenditure distribution, we can use them to estimate inequality. We offer a Gini index, which is on a scale between 0 (perfect equality) and 1 (the richest person has all the income), as well as some popular percentile ratios. An example of a percentile ratio is the 90/10 ratio, which is the ratio of the income of the household for which 90% of households have lower income to the income of the household for which only 10% of households have lower income. The 50/10 ratio and 90/50 ratio are similarly defined.

By the standards of inequality measures estimated from modern (post-1950) household expenditure surveys, many of those presented here are low. This is primarily the outcome of the survey target group being drawn from a relatively narrow social or economic class, rather than being a sample of the entire population (or nearly so) as is generally the case with modern surveys. It is advisable to examine these inequality estimates in conjunction with the survey description given in the bibliography, as care is needed when interpreting trends derived from these estimates.