The new inequality-adjusted HDI

Since it's creation in 1990, the HDI has been often critiqued for its inability to capture dimensions of economic and social inequality. The 2010 report, marking the 20th anniversary of the Human Development Report and published last week, has now responded to the demands by including an inequality-adjusted HDI

tabulation. As a quick briefing, I'd like to 1) discuss how inequality enters the index and 2) sketch some stark deviations caused by the adjustment.

In brief, inequality is accounted for by comparing the (arithmetic) mean of a distribution (e.g. life expectancy) to an ideal distribution of equality in utility (equally distributed equivalent). The differential between actual and ideal distribution is then used to generate an adjustment factor. While the idea sounds rather intuitive and objective, the definition of ideal equality adds another range of normative assumptions to the index: The UNDP uses a specific case of Atkinson's inequality index and implicitly assumes a welfare function of u(x)=ln(x) to arrive at the adjustment formula of (1-geometric mean/arithmetic mean). Just like arbitrarily assuming a weight of 1/3 for the dimensions of income, education and health, there is no particular reason why this should be so, apart from analytical elegance.

Nonetheless, the inclusion of inequality will certainly impact the discourse on development: Adjusting the HDI using Atkinson's index induces dramatic changes. For example, the HDI for Namibia, previously at 0.6 is almost halved when adjusting for inequality. This produces large changes in the overall ranking: The US, as another example, is ranked as the country with the fourth highest HDI but drops by 9 ranks once adjusting for inequality. When eyeballing the three dimensions separately, there appears to be a huge within variation: While the variation in life expectancy in Germany, for example, is almost equally distributed (only 4.4% downward adjustment), the distribution of income is much more unequal (16.7% downward adjustment). Azerbaijan, as a contrasting case, has an highly unequal distribution of life expectancy (23.8 downward adjustment), but income is much more equally distributed (only 4.4% downward adjustment).

To sum up: Despite its increasingly normative nature, the new IHDI will not only affect discourse but also provide an exciting range of new statistics: In particular, moving away from only income inequality towards inequality in health and education will contribute to a much deeper understanding of the distributional relationships between socio-economic indicators. Keep your eyes open for upcoming in-depth studies of the IHDI and keep us all posted.