Development Strategy and Policy Analysis Unit w Development Policy and Analysis Division

Department of Economic and Social Affairs

Inequality Measurement
Development Issues No. 2
21 October 2015

Comprehending the impact of policy changes on the distribu-

tion of income first requires a good portrayal of that distribution. Summary
There are various ways to accomplish this, including graphical
and mathematical approaches that range from simplistic to more There are many measures of inequality that, when
intricate methods. All of these can be used to provide a complete combined, provide nuance and depth to our understanding
picture of the concentration of income, to compare and rank of how income is distributed. Choosing which measure to
different income distributions, and to examine the implications use requires understanding the strengths and weaknesses
of alternative policy options. of each, and how they can complement each other to
An inequality measure is often a function that ascribes a value provide a complete picture.
to a specific distribution of income in a way that allows direct and
objective comparisons across different distributions. To do this,
inequality measures should have certain properties and behave in
a certain way given certain events. For example, moving $1 from the ratio of the area between the two curves (Lorenz curve and
a richer person to a poorer person should lead to a lower level of 45-degree line) to the area beneath the 45-degree line. In the
inequality. No single measure can satisfy all properties though, so figure above, it is equal to A/(A+B). A higher Gini coefficient
the choice of one measure over others involves trade-offs. The fol- represents a more unequal distribution. According to World
lowing measures differ with regards to the properties they satisfy Bank data, between 1981 and 2013, the Gini index ranged
and information they present. None can be considered superior, between 0.3 and 0.6 worldwide. The coefficient allows direct
as all are useful given certain contexts. A well-balanced inequal- comparison of two populations’ income distribution, regardless
ity analysis should look at several of these measures. of their sizes. The Gini’s main limitation is that it is not easily
decomposable or additive. Also, it does not respond in the same
Graphical representation of inequality way to income transfers between people in opposite tails of the
income distribution as it does to transfers in the middle of the
Lorenz curve distribution. Furthermore, very different income distributions
It is one of the simplest representations of inequality. On the can present the same Gini coefficient.
horizontal axis is the cumulative number of income recipients
ranked from the poorest to the richest individual or household. Atkinson’s inequality measure (or Atkinson’s index)
The vertical axis displays the cumulative percentage of total
income. The Lorenz curve reveals the percentage of income This is the most popular welfare-based measure of inequality.
owned by x per cent of the population. It is usually shown in It presents the percentage of total income that a given society
relation to a 45-degree line that represents perfect equality where would have to forego in order to have more equal shares of
each x percentile of the population receives the same x percentile income between its citizens. This measure depends on the degree
of income. Thus the farther the Lorenz curve is in relation to of society aversion to inequality (a theoretical parameter decided
the 45-degree line, the more unequal the distribution of income. by the researcher), where a higher value entails greater social
utility or willingness by individuals to accept smaller incomes
Indices in exchange for a more equal distribution. An important feature
of the Atkinson index is that it can be decomposed into within-
Gini index and between-group inequality. Moreover, unlike other indices,
It is the most widely cited measure of inequality; it measures it can provide welfare implications of alternative policies and
the extent to which the distribution within an economy deviates allows the researcher to include some normative content to the
from a perfectly equal distribution. The index is computed as analysis (Bellù, 2006).

Development Issues are intended to clarify concepts used in the analytical work of the Division, provide references to current development issues and offer a common
background for development policy discussions. This note was prepared by Helena Afonso, Marcelo LaFleur and Diana Alarcón in the Development Strategy and Policy
Analysis Unit in the Development Policy and Analysis Division of UN/DESA. For more information, contact: alarcond@un.org. The full archive is available at:
www.un.org/en/development/desa/policy/wess/ 1
 100%
Ratios
Ratios constitute the most basic inequality measures available.
They are simple, direct, easy to understand, and they offer
few data and computation challenges. Accordingly, they do
A not provide as much information as the complex measures

Cumulative percentage of income

described above.

Robin Hood Decile dispersion ratio (or inter-decile ratio)

rve
Index
It is the ratio of the average income of the richest x per cent of
ine

Cu
°) L

enz
the population to the average income of the poorest x per cent. It
(45

Lor
ion

expresses the income (or income share) of the rich as a multiple

ut
rib
ist

of that of the poor. However, it is vulnerable to extreme values

tD
fec

B and outliers. Common decile ratios include: D9/D1: ratio of the

r
Pe

income of the 10 per cent richest to that of the 10 per cent poor-
est; D9/D5: ratio of the income of the 10 per cent richest to the
Cumulative percentage of households 100% income of those at the median of the earnings distribution; D5/
Lorenz Curve and Robin Hood Index
D1: ratio of the income of those at the median of the earnings
distribution to the 10 per cent poorest. The Palma ratio and the
20/20 ratio are other examples of decile dispersion ratios.
Hoover index (also known as the Robin Hood index,
Schutz index or Pietra ratio) Palma ratio
It shows the proportion of all income which would have to It is the ratio of national income shares of the top 10 per cent of
be redistributed to achieve a state of perfect equality. In other households to the bottom 40 per cent. It is based on economist
words, the value of the index approximates the share of total José Gabriel Palma’s empirical observation that difference in
income that has to be transferred from households above the the income distribution of different countries (or over time) is
mean to those below the mean to achieve equality in the distri- largely the result of changes in the ‘tails’ of the distribution (the
bution of incomes. Higher values indicate more inequality and poorest and the richest) as there tends to be relative stability in
that more redistribution is needed to achieve income equality. It the share of income that goes to the ‘middle’ (Cobham, 2015).
can be graphically represented as the maximum vertical distance
between the Lorenz curve and the 45-degree line that represents
20/20 ratio
perfect equality of incomes.
It compares the ratio of the average income of the richest 20
Theil index and General Entropy (GE) measures per cent of the population to the average income of the poor-
est 20 per cent of the population. Used by the United Nations
The values of the GE class of measures vary between zero (perfect Development Programme Human Development Report (called
equality) and infinity (or one, if normalized). A key feature of “income quintile ratio”).
these measures is that they are fully decomposable, i.e. inequal-
ity may be broken down by population groups or income sources
or using other dimensions, which can prove useful to policy
References
Bellù, L. G., and Liberati, P. (2006), ‘Policy Impacts on Inequality: Welfare
makers. Another key feature is that researchers can choose a
Based Measures of Inequality – The Atkinson Index’, Food and
parameter α that assigns a weight to distances between incomes Agriculture Organization of the United Nations.
in different parts of the income distribution. For lower values Bellù, L. G., and Liberati, P. (2006), ‘Describing Income Inequality: Theil
of α, the measure is more sensitive to changes in the lower tail Index and Entropy Class Indexes’, Food and Agriculture Organization
of the distribution and, for higher values, it is more sensitive to of the United Nations.
changes that affect the upper tail (Atkinson and Bourguignon, Cobham, A., Schlogl, L., and Sumner, A. (2015), ‘Inequality and the Tails:
2015). The most common values for α are 0, 1, and 2. When The Palma Proposition and Ratio Revisited’, Department of Economic
and Social Affairs Working Paper No. 143 (ST/ESA/2015/DWP/143).
α=0, the index is called “Theil’s L” or the “mean log deviation”
‘Handbook of Income Distribution Volume 2’ First Edition (2015), edited by
measure. When α=1, the index is called “Theil’s T” index or, Anthony Atkinson and François Bourguignon, North-Holland (Elsevier).
more commonly, “Theil index”. When α=2, the index is called United Nations Development Programme, ‘2014 Human Development
“coefficient of variation”. Similarly to the Gini coefficient, when Trends by Indicator’, available at http://hdr.undp.org/en/data.
income redistribution happens, change in the indices depends World Bank, World Development Indicators. Accessed 13 October 2015.
on the level of individual incomes involved in the redistribution
and the population size (Bellù, 2006).

2 October 2015