What the Gini index can and cannot tell you

The Gini index is the most quoted single number on inequality, and one of the easiest to misread. This site holds it as the World Bank publishes it on its Poverty and Inequality Platform: one file per reference year from 2002 to 2023, each value carried to the hundredth of a point. This piece reads those files for what the index says, which countries sit at each end, how the middle has moved and what the number cannot show. The figures are as they stood on 14 November 2026; the statistic page is the one kept current.

What the number is

The index runs from 0, where every person has exactly the same income or consumption, to 100, where one person has all of it. It is built by comparing every person in a survey with every other: the larger the typical gap between two people, measured against the country's mean, the higher the reading. It is a summary of how a whole distribution is shaped, not a level of income. A country where nearly everyone is poor can read low, and a rich country can read high. Two countries can also share a reading while their gaps sit in different places, one stretched at the top and the other hollowed out in the middle.

Income or consumption

Every figure comes from a household survey, and surveys measure one of two things. Some record what a household received: its disposable income, from work and other sources including transfers, less taxes. Others record what it consumed. The World Bank's methodology handbook explains why the choice matters: incomes are more likely than consumption to be very low, even negative, or very high in a given year, while households smooth their consumption against shocks, so "inequality often is higher when incomes are used rather than consumption". The Bank's data team said the same in January 2025: consumption surveys typically show less inequality than income surveys. For some country-years the platform publishes both kinds of reading. This site does not choose between them; it stores whichever one the Bank uses for its own headline figure.

Which kind a country uses follows its statistical system more than anything else. In the 2023 file, 46 of the 60 readings are income-based and 14 are consumption-based, and 10 of those 14 are in Asia. That is why setting the two groups against each other teaches nothing about the surveys themselves. The consumption group's middle country reads 34.13 and the income group's 33.72, because the income group is made of European countries at one end and countries of the Americas at the other. The difference a survey type makes shows inside a country, not across a table.

The lowest reading, and why it is low

India holds the lowest reading in the 2023 file, alone, at 23.00, ahead of Slovakia at 23.82 and Slovenia at 24.68. India's figure is consumption-based; the other two are income-based. The readings nearest it that share its method are Kiribati's 24.74 and Moldova's 26.76, both consumption figures. India's own series has been consumption-based throughout: 27.68 in 2004, 27.83 in 2009, 28.78 in 2011, 25.51 in 2022 and 23.00 in 2023. Its survey year is printed as 2023.58, which in the Bank's convention means the fieldwork ran across two calendar years and the reading is filed under the first. The figure is the one the Bank published; set beside Slovakia's, it is two different measurements side by side.

The highest readings depend on which year you read

In the 2023 file the eight highest readings are all in the Americas and all income-based, led by Colombia at 54.58, Brazil at 51.49 and Panama at 48.88. But that file holds only two African countries, so it says almost nothing about Africa. Read instead each country's newest survey from the ten years to 2023 and 156 countries are measured. The highest is then Namibia, at 59.07 in 2015, followed by Botswana at 54.91, also in 2015; Eswatini (2016) and Colombia (2023) share 54.58, and South Africa reads 54.05 in 2022. The southern African figures are consumption-based, the kind that tends to read lower, and they are still the highest of the 156.

By continent, the middle of those newest readings is 43.24 in the Americas, 38.09 in Africa, 32.91 in Asia, 31.52 in Oceania and 30.64 in Europe.

Why the years do not line up

A country appears in a year's file only if it ran a usable survey for that year. Over 2002–2023, 169 countries are measured at least once, but no single year holds more than 94, and the 2023 file holds 60: 28 in Europe, 15 in the Americas, 14 in Asia, 2 in Africa and 1 in Oceania. A country missing from a year has not become equal; it has not been surveyed. Nor do the years mean quite the same thing. Where fieldwork spans two calendar years, the Bank files the survey under the first; for the high-income surveys it draws from the European Union's income survey and the Luxembourg Income Study, the income asked about is the previous year's, so the reading is filed a year before the fieldwork. Comparing one year's file with another compares different groups of countries, measured in different ways.

How the middle has moved

The middle country of the 2002 file read 37.14 and the middle of the 2023 file 33.72, but those are different sets of countries. Following the same ones is safer. Of the 85 countries with a reading from 2002–2005 and another from 2020–2023 on the same kind of survey, 67 read lower at the end and 18 higher, and their middle moved from 37.73 to 33.89. The clearest falls are in South America, where surveys run every year: Bolivia from 59.27 in 2002 to 42.05 in 2023, Paraguay from 57.27 to 44.18 and Peru from 53.56 to 40.66 over the same years. Among the rises, Indonesia went from 30.16 in 2002 to 36.06 in 2023, Denmark from 25.63 in 2003 to 29.92 in 2023, and Germany from 29.88 in 2002 to 33.66 in 2022.

Some of the largest changes come from surveys run a decade or more apart: Zambia from 42.06 in 2002 to 51.48 in 2022, and Comoros from 55.92 in 2004 to 34.18 in 2020. The World Bank's handbook warns that a change in questionnaire design makes a country's estimates incomparable over time, and it records such breaks in a comparability database that this site does not hold. Over a long gap between two surveys, a change in the number can be a change in the instrument.

What it cannot tell you

The index is only as complete as the survey behind it, and surveys reach the top of a distribution least well. The World Bank's data team notes that "inequality estimates from household surveys often underrepresent the richest individuals", so the true spread at the top may be wider than a survey-based Gini shows. The index also says nothing about how poor a country is, only how unevenly what it has is shared: the share of income going to the poorest fifth, the Palma ratio and the poverty headcount each answer a question the Gini does not. And it describes one country at a time. Two countries with the same low reading are equally even within their borders, not equally well off.