Skip to content
Back to search results

Gini Coefficient

Introduced by
Corrado Gini
Year
1912
Type
Scalar measure of statistical dispersion / inequality
Family
Inequality measurement
Font size

What does it do?

The Gini coefficient is the most widely used single-number summary of inequality in a distribution such as income or wealth.

Introduced by the Italian statistician Corrado Gini in 1912, it equals twice the area between the Lorenz curve and the line of perfect equality, ranging from 0 when everyone has the same amount to a maximum approaching 1 when one unit holds everything.

The intuition behind the method

Suppose you pick two people at random and note the absolute difference in their incomes; average that difference over all pairs and scale it by the mean income. That is the Gini coefficient — a measure of how far apart, on average, any two members of the population are. Equivalently, draw the Lorenz curve of cumulative income against cumulative population: the more it sags below the 45-degree line of equality, the larger the gap, and the Gini is exactly twice the area of that gap. One number captures the whole shape of the inequality.

History

Corrado Gini introduced the index in his 1912 work Variabilità e Mutabilità, where he proposed several formulations of a concentration measure; Lidia Ceriani and Paolo Verme's 2012 centenary article translates and documents these origins. The index built on Max Lorenz's 1905 curve and on Gini's own mean-difference statistic. Through the twentieth century it became the standard inequality measure of economics and sociology, adopted by national statistical offices, the World Bank, and the OECD, and the focus of an extensive axiomatic and decomposition literature.

When to use it

Use the Gini coefficient for a compact, scale-invariant summary of inequality in any nonnegative distribution — income, wealth, land, firm size — when a single comparable number is wanted and the Lorenz dominance is not strictly required. It is the lingua franca of inequality comparison across countries and time. It is less suitable when you need sensitivity to a particular part of the distribution (the Gini is most sensitive to the middle and can be insensitive to extreme tails), when distributions' Lorenz curves cross (the single number can mislead), or when an explicit social-welfare interpretation with a tunable inequality-aversion parameter is desired (use the Atkinson index). It should not be applied to distributions with negative values without adjustment.

Limitations and considerations

  • Most sensitive to changes around the middle of the distribution and relatively insensitive to the extreme tails.
  • Two very different distributions can share the same Gini; it is uninformative when Lorenz curves cross.
  • Not cleanly decomposable into between- and within-group components (unlike entropy measures) because of an overlap term.
  • Carries no explicit social-welfare or inequality-aversion parameter, unlike the Atkinson index.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Single, scale- and population-size-invariant number enabling broad comparison across countries and over time.
  • Intuitive geometric and pairwise-difference interpretations.
  • Bounded on [0,1) with a clear meaning at both ends.
  • Decomposable by income source via the covariance form, and widely tabulated for almost every country.

Applications

  • Comparing income or wealth inequality across countries and over time.
  • Tracking the redistributive effect of taxes and transfers by comparing pre- and post-tax Ginis.
  • Measuring inequality in non-income domains such as land holdings, firm sizes, or health.
  • Decomposing inequality by income source (labor, capital, transfers).
  • Summarizing concentration in fields from ecology to information science.

How it works

How it works

1

The most fundamental definition is half the relative mean absolute difference: average the absolute income gap over all pairs of individuals and divide by twice the mean. This makes the Gini a direct measure of average pairwise disparity, requiring no reference to a Lorenz curve.

Formula
G = \frac{1}{2n^2\bar{x}} \sum_{i=1}^{n}\sum_{j=1}^{n} |x_i - x_j|
2

Geometrically, the Gini is one minus twice the area under the Lorenz curve L(p), or equivalently the ratio A/(A+B) where A is the area between the equality line and the Lorenz curve and B is the area beneath the curve. A perfectly equal distribution gives a Lorenz curve on the diagonal and a Gini of 0.

Formula
G = 1 - 2\int_0^1 L(p)\,dp = \frac{A}{A+B}
3

A convenient computational identity expresses the Gini as twice the covariance between income and its own rank (the cumulative distribution function), divided by the mean. This form underlies regression-based decompositions of the Gini by income source or population subgroup.

Formula
G = \frac{2\,\operatorname{cov}\!\bigl(x, F(x)\bigr)}{\bar{x}}
4

The coefficient lies in [0, 1): zero indicates perfect equality, and it approaches one as a single unit captures the entire total. In practice national income Ginis range from roughly 0.25 in the most equal countries to above 0.6 in the most unequal, and the bounded scale makes cross-population comparison straightforward.

Formula
G \in [0, 1):\ 0 = \text{perfect equality},\ \to 1 = \text{maximal inequality}

Common pitfalls

  • Comparing Ginis computed on different income concepts (market vs. disposable, individual vs. household) as if they were equivalent.
  • Treating equal Ginis as equal distributions when their Lorenz curves in fact cross.
  • Attempting a between/within subgroup decomposition without accounting for the residual overlap term.
  • Applying the standard formula to data containing negative incomes or wealth, where the coefficient can exceed one and lose its interpretation.

Frequently asked questions

What does a Gini of 0.4 mean?

It is a scale-free index of how unequal a distribution is, with 0 meaning everyone has the same amount and values near 1 meaning near-total concentration. A Gini of 0.4 indicates moderate-to-high inequality; for context, disposable-income Ginis cluster around 0.25–0.35 in the most egalitarian welfare states and reach 0.5 or higher in the most unequal countries. The number is comparable only when the same income concept and unit are used.

Why can two countries with the same Gini have different inequality?

The Gini summarizes the entire Lorenz curve in a single number, so distributions whose Lorenz curves cross can yield the same value while differing in where the inequality lies — one concentrated at the top, another at the bottom. When Lorenz curves cross there is no unambiguous inequality ranking, and a single index like the Gini may obscure the difference. Reporting tail-sensitive measures (Atkinson, Palma) alongside it helps.

Is the Gini decomposable across subgroups?

Not cleanly. Unlike the generalized entropy (Theil) family, the Gini does not split exactly into between-group and within-group parts: a decomposition leaves a residual 'overlap' or interaction term that arises when subgroup distributions overlap. For exact additive subgroup decomposition, entropy-based measures are preferred; the Gini does, however, decompose neatly by income source.

Related methods

Sources

  1. Ceriani, L., & Verme, P. (2012). The origins of the Gini index: extracts from Variabilità e Mutabilità (1912) by Corrado Gini. The Journal of Economic Inequality, 10(3), 421–443.

    Open DOI reference
    Copied.
  2. Lorenz, M. O. (1905). Methods of measuring the concentration of wealth. Publications of the American Statistical Association, 9(70), 209–219.

    Open DOI reference
    Copied.