Skip to content
Back to search results

Inequality-adjusted HDI

Introduced by
Sabina Alkire & James Foster; UNDP Human Development Report Office
Year
2010
Type
Distribution-sensitive composite development index
Family
Composite development indices
Font size

What does it do?

The Inequality-adjusted Human Development Index (IHDI) extends the Human Development Index by accounting for how achievements in health, education, and income are distributed across a population, not just their averages.

Designed by Sabina Alkire and James Foster for the UNDP and introduced in the 2010 Human Development Report, it discounts each HDI dimension by the inequality observed within it, using an Atkinson-class inequality measure. When there is no inequality the IHDI equals the HDI; as inequality rises the IHDI falls below it, and the percentage gap — the 'loss' — measures how much human development is eroded by being unequally shared.

The intuition behind the method

The HDI tells you a country's average health, schooling, and income, but an average can hide a sharply unequal reality: a society where a few live long, educated, prosperous lives and many do not can post the same HDI as a society where outcomes are evenly spread. The IHDI asks what the HDI would be if you cared about the typical person rather than the average. For each dimension it replaces the arithmetic mean with an inequality-sensitive 'equally distributed equivalent' — formally the geometric mean of individual achievements — which is always at or below the arithmetic mean and falls further the more unequal the distribution. Discounting each dimension by its inequality and recombining gives a development score that shrinks when achievement is concentrated, and the size of the shrinkage is itself a headline number.

History

The IHDI was proposed by Sabina Alkire and James Foster in a 2010 OPHI/HDRO working paper and introduced in the UNDP's 20th-anniversary Human Development Report of 2010, alongside the Gender Inequality Index and the Multidimensional Poverty Index, as part of a major methodological overhaul of the human development measures. It built on a long line of distribution-sensitive welfare measurement — Atkinson's 1970 inequality index and the concept of the equally distributed equivalent — and answered a standing critique that the HDI's reliance on averages made it blind to inequality. It has since been reported annually for most countries with adequate distributional data.

When to use it

Use the IHDI when the average picture given by the HDI is not enough and you want a distribution-sensitive measure that rewards equitably shared human development and penalizes concentration. It is the right companion to the HDI for comparing how well countries spread health, education, and income, for tracking whether progress is inclusive, and for highlighting the inequality 'loss' as an advocacy and monitoring figure. It requires household-level distributional data for each dimension, which is less universally available than the aggregate indicators behind the HDI, and like the HDI it remains a national summary rather than a measure of poverty, gender gaps, or sustainability.

Limitations and considerations

  • Requires household-level distributional data for all three dimensions, which is unavailable or low-quality in many countries, forcing imputation and limiting comparability.
  • It is association-insensitive in its standard form: it captures inequality within each dimension separately but not the overlap (whether the same people are deprived across dimensions).
  • The choice of inequality-aversion parameter (set to one) is a normative judgment that shapes how strongly inequality is penalized.
  • Like the HDI it remains a national average construct and says nothing about poverty incidence, gender gaps, or environmental sustainability.

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

Strengths

  • Makes the HDI distribution-sensitive, so a country is rewarded for spreading health, education, and income equitably rather than concentrating them.
  • The IHDI equals the HDI under perfect equality and the gap between them yields an interpretable 'loss due to inequality' headline figure.
  • Uses an Atkinson-class measure that is subgroup-decomposable and that handles inequality in each dimension separately, so the source of the loss can be identified.
  • Built directly on the familiar HDI structure, so it is easy to communicate as 'the HDI a country would have if achievements were equally distributed'.

Applications

  • Annual reporting of the inequality loss alongside the HDI in the UNDP Human Development Report to show how inclusive each country's development is.
  • Cross-country comparison of how equitably health, education, and income gains are shared rather than just their averages.
  • Monitoring whether economic and social progress is inclusive over time within a country.
  • Identifying which dimension — health, education, or income — contributes most to a country's inequality loss to guide policy.
  • Benchmarking sub-national regions or population groups by their inequality-adjusted human development.

How it works

How it works

1

For each dimension, inequality is measured by an Atkinson index with inequality-aversion parameter equal to one: A_x = 1 minus the ratio of the geometric mean of individual achievements to their arithmetic mean. Because the geometric mean never exceeds the arithmetic mean, A_x lies between 0 (perfect equality) and 1; it rises as the distribution becomes more unequal and is computed separately for the distributions of expected length of life, years of schooling, and income (or consumption).

Formula
A_x = 1 - \frac{\sqrt[n]{x_1 x_2 \cdots x_n}}{\bar{x}}
2

Each of the three HDI dimension indices is multiplied by (1 - A_x), its own inequality factor. This converts the average-based dimension index into an inequality-adjusted index that equals the average only when the dimension is perfectly equally distributed and is pushed down in proportion to the inequality measured in that dimension. The three dimensions are discounted independently because inequality in health, education, and income generally differs.

Formula
I_x^{*} = (1 - A_x)\, I_x, \qquad x \in \{\text{Health},\,\text{Education},\,\text{Income}\}
3

The three inequality-adjusted dimension indices are combined by the same unweighted geometric mean used for the HDI. Algebraically this equals the HDI multiplied by the cube root of the product of the three (1 - A_x) factors, so the IHDI inherits the HDI's structure but is scaled down by an aggregate inequality discount. The IHDI equals the HDI when all three dimensions are perfectly equal and is otherwise strictly lower.

Formula
\text{IHDI} = \left(I_{\text{Health}}^{*} \cdot I_{\text{Education}}^{*} \cdot I_{\text{Income}}^{*}\right)^{1/3} = \sqrt[3]{(1-A_{\text{H}})(1-A_{\text{E}})(1-A_{\text{I}})}\;\text{HDI}
4

The overall loss in human development due to inequality is the percentage difference between the HDI and the IHDI: L = (HDI - IHDI)/HDI, equivalently one minus the cube root of the product of the (1 - A_x) factors. This loss is the headline interpretive quantity — a country with HDI 0.80 and IHDI 0.64 suffers a 20 percent loss to inequality — and it can be decomposed to show which dimension contributes most to the shortfall.

Formula
L = \frac{\text{HDI} - \text{IHDI}}{\text{HDI}} = 1 - \sqrt[3]{(1-A_{\text{H}})(1-A_{\text{E}})(1-A_{\text{I}})}

Common pitfalls

  • Comparing the IHDI of one country with the HDI of another; the IHDI should be compared with IHDI, and the HDI-to-IHDI loss compared across countries.
  • Interpreting the loss as inequality in a single distribution when it is an aggregate over three independently measured dimensions.
  • Ignoring that the index does not capture cross-dimensional overlap, so it can understate disadvantage that is concentrated in the same people.
  • Treating IHDI changes over short periods as real when they may reflect changes in survey coverage or data quality of the underlying distributions.

Frequently asked questions

What does the inequality 'loss' in the IHDI mean?

The loss is the percentage by which the IHDI falls short of the HDI: L = (HDI - IHDI)/HDI. It quantifies how much potential human development is lost because achievements in health, education, and income are not equally distributed. A loss of 0 percent means perfectly equal distribution and IHDI equal to HDI; a loss of, say, 25 percent means a quarter of the country's average human development is, in effect, forfeited to inequality. Comparing losses across countries shows where development is most unequally shared.

Why does the IHDI use the Atkinson index rather than the Gini coefficient?

The Atkinson measure used in the IHDI is based on the ratio of the geometric mean to the arithmetic mean of individual achievements, which makes it subgroup-decomposable and gives it a clean welfare interpretation through the equally distributed equivalent. This lets each HDI dimension be discounted by (1 - A_x) and recombined with the same geometric-mean machinery as the HDI. The Gini coefficient, while widely used, is not subgroup-decomposable in the same convenient way and does not slot as cleanly into the multiplicative structure of the index.

Why is the IHDI called 'association-insensitive'?

The standard IHDI measures inequality in each dimension separately and then discounts each dimension, so it captures how unequal health is, how unequal education is, and how unequal income is, but not whether the same individuals are simultaneously disadvantaged across all three. Two societies with identical within-dimension inequalities receive the same IHDI even if in one the deprivations pile on the same people and in the other they are spread around. Capturing that overlap requires a joint, individual-level measure such as the Multidimensional Poverty Index.

Related methods

Sources

  1. Alkire, S., & Foster, J. (2010). Designing the Inequality-Adjusted Human Development Index (IHDI). OPHI Working Paper 37 / Human Development Research Paper 2010/28. UNDP Human Development Report Office, New York.

    Open publication reference
    Copied.
  2. UNDP (2022). Human Development Report 2021-22, Technical Notes. United Nations Development Programme, Human Development Report Office, New York.

    Open publication reference
    Copied.