Document Library

Key Text Why Horizontal Inequalities Matter: Some Implications for Measurement

Author: F Stewart and G Brown
Date: 2005
Size: 30 pages (210 KB)

Access full text: available online

Summary

Why do inequalities between groups matter as well as inequalities between individuals? What is the best way to measure such horizontal inequalities? This paper for the Centre of Research on Inequality, Human Security, and Ethnicity (CRISE) argues that horizontal inequalities (HI) matter for the well being of individuals within groups, and for their impact on wider growth and conflict.

Most discussion of inequality concerns Vertical Inequality (VI) between individuals, and is generally confined to a few economic variables such as income or consumption. Horizontal inequalities (HI) are inequalities between groups, and have been largely ignored by policy makers.

Group inequality is important because it can affect happiness, efficiency and political stability. It is difficult to assess HI because group identities are fluid, multiple, and may be endogenous. However, felt differences are important and clear enough in many societies to measure HIs if the contingent nature of group definitions is taken into account. Three alternative HI measures were reviewed and compared using data over time for Indonesia, South Africa and the USA; the coefficient of variation among groups (GCOV), the Group Gini (GGINI), and Group Theil (GTHEIL) indices. Findings were:

Each measure of HI is correlated with every other one, but less strongly than correlation between VI measures
Both HI and VI are persistent over time, but HI is more changeable. Durable HI is likely to cause political strife
Different HI measures can generate different trends. In particular, population unweighted GCOV moves in a different direction to weighted GCOV and GGINI.

Policy makers should pay more attention to HIs. In addition to the instrumental and well being implications mentioned, it is often not possible to improve the position of individuals without tackling the position of the group. For example, extending education to girls may be prevented by parental attitudes rather than numbers of schools and teachers. Addressing HI as well as VI can improve targeting of policies and interventions. The levels and durability of HI are of great significance in fragile/post-conflict states. Where there is political as well as economic inequality, HI can be a source of violent conflict. The value of a measure of HI depends on the purpose for which it is needed. If the concern is:

With the position of one group, or the relation of that group to one other group, group HI measures can conceal the variable of interest by including other groups. It is clearer to use ratio of performance to the mean, or ratio of performance of the two groups

Identifying a variable to test whether particular inequalities are correlated to other events such as conflict or criminality, group weighting by population is desirable (GGINI, GCOV)
With making a general statement about whether society is getting fairer or less fair over time from a group perspective, group weighting by population is also desirable
A stronger focus on fairness issues, the Esteban/Ray polarisation index or some other measure incorporating an evaluative element, may be most useful. However, unless the valuation is widely understood and shared, it is better to present the data using GGINI or GCOV.

Access full text: available online

Source: Stewart, F., Brown, G., and Mancini, L., 2005, ‘Why Horizontal Inequalities Matter: Some Implications for Measurement’, CRISE, Queen Elizabeth House, Oxford
Author: Centre for Research on Inequality, Human Security and Ethnicity (CRISE), http://www.crise.ox.ac.uk/index.shtml