Atkinson Index

The Atkinson Index is an inequality measure introduced by British economist Anthony B. Atkinson in his 1970 paper "On the Measurement of Inequality" in the Journal of Economic Theory. Unlike the Gini coefficient, which is purely descriptive, the Atkinson Index is explicitly normative: it builds a value judgment about inequality directly into the formula through an inequality aversion parameter, conventionally denoted ε (epsilon).

The index is derived from the concept of the equally distributed equivalent income — the per-capita income that, if distributed equally, would yield the same level of social welfare as the actual unequal distribution. The Atkinson Index equals 1 minus the ratio of that equivalent income to the mean income. A value of 0.30, for instance, implies that society could achieve the same welfare with only 70% of current income if it were distributed equally.

Key features:

• Parameter ε ≥ 0 controls how much weight is placed on transfers at the bottom of the distribution. When ε = 0, the index treats all incomes symmetrically and equals zero regardless of distribution. As ε rises (commonly 0.5, 1, or 2), the index gives progressively more weight to the poor.
• The index is subgroup decomposable, allowing researchers to separate within-group and between-group inequality — useful for analyzing regional, ethnic, or gender disparities.
• It satisfies the Pigou-Dalton transfer principle: any income transfer from a richer to a poorer person (that does not reverse their ranking) reduces measured inequality.

The Atkinson Index is widely used by the UNDP, OECD, World Bank, and Eurostat alongside the Gini coefficient and Theil index. It is particularly valued in welfare economics and policy analysis because it makes the analyst's ethical assumptions transparent rather than hidden. Critics note that choosing ε is itself a political choice, and results can shift substantially depending on the value selected.