A Newer Human Development Index

Abstract

The Human Development Index has experienced substantial modifications in the 2010 edition of the Human Development Report (changes in some of the variables, a different aggregation procedure, and the introduction of distributive considerations, among others). Those changes respond to some well-known shortcomings of the traditional design of this index and entail substantial improvements. There are still some inconsistencies in the new construction that have to be addressed (in particular, the use of a composite variable to approach educational achievements, the use of logs for the income variable and the type of normalization adopted). We discuss in this paper those inconsistencies and suggest some relatively minor changes that would suffice to avoid them.

Keywords:

• Human Development
• Inequality
• Measurement
• Capability Approach
• Education

JEL classification:

• 015
• 131

Acknowledgements

Thanks are due to two anonymous referees and the Associate Editor for their helpful comments and suggestions. Financial support to the following research projects, as well as the FEDER Funds, are gratefully acknowledged: Projects SEJ 2007-62656 (first author), ECO2008-03674 and ECO2011-029355 (second author) and ECO2010-21706 (third author) from the Ministry of Science and Innovation; Projects SEJ5980, SEJ4941 (second author) and SEJ-6882/ECON (third author) from the Junta de Andalucía.

Notes

See the discussion in the UNDP Research Papers' Series. See the criticisms in Anand and Sen (1994 a, b); Hicks (1997); Sagar and Najam (1999); Osberg and Sharpe (2002); Phillipson and Soares (2001); Pinilla and Goerlich (2005); Foster et al. (2005); Becker et al. (2005); Stiglitz et al. (2009); Herrero et al. (2010a, b).

Besides those methodological changes there are also some other relevant novelties, such as the reconstruction of the evolution of those variables since 1980 in a time consistent series and the introduction of a brand new multidimensional poverty measure.

This paper is built over a former contribution (Herrero et al, 2010b), written before the HDI 2010 appeared. Since some of our suggestions there have already been incorporated to the New HDI, we shall not insist on them.

Herrero et al. (2010b) propose to change life expectancy at birth by life potential, in order to get a better estimation of development capabilities, in the case of highly developed countries. This is so because life expectancy at birth is a variable that is independent on the demographic structure. As a consequence, it tends to over-weight the health component of those countries with a higher share of old people. That aspect is most arguable in the context of evaluating development capabilities, for it ignores the size of the present and future working age population. Be as it may, we leave aside here this question, bearing in mind that we aim at proposing some changes in the index without recurring to new data. Moreover, data on life potential are only available for OECD countries.

One is to add up all students that reach some reference level, as this implies that those with higher levels of studies are computed several times (for example, a person that gets a college degree weights three times more than a person that left studies after primary school); this is the approach followed by the American Human Development Index. (see Burd-Sharps, 2008). An alternative way is computing the average number of years of school attendance or the percentage of working age population with non-compulsory studies (as in Herrero et al. 2010c). A sensible option for highly developed countries is to take the ‘expected years of education between 15 and 29’ used by the OECD when assessing education and the labour market (Herrero et al. 2010b).

A natural way of interpreting the capabilities in the health and education variables is by means of the expectations of newcomers in the society. This is indeed the philosophy behind the selection of life expectancy at birth. Note that choosing mean years of education of adults would be more in line with the selection of life potential to measuring actual achievements.

Needless to say the geometric mean is just a special case of the generalized means, which have been already characterized (e.g. Foster et al., 2005; Seth, 2009). Our contribution here is that of presenting a very simple and self-contained characterization result. An alternative characterization appears in Herrero et al. (2010a).

A case in point is that in which material wellbeing is approached in terms of the egalitarian equivalent income (see below).

The min values chosen are: 20 years for life expectancy at birth, 0 for both education variables, and $163 (the lowest value attained for any country in recorded history, in Zimbabwe, 2008). As for the max values, we find: 83.2 years for life expectancy, 13.2 for mean years of schooling, 20.6 for expected years of schooling, and $108,211 for the per capita GNI.

To assess the empirical extent of this aspect we have calculated the HDI 2010 by normalizing the raw variables as shares to the same max values used in the report and keeping everything else the same. The result is that, in 2010, some 30% of the countries change their ranking by five or more positions and six countries change 10 positions or more.

See, for instance, Cowell (1995), Sen and Foster (1997), Goerlich and Villar (2009).

We consider here the family of relative inequality measures, which is the one taken as reference in the report.

Under the implicit assumption that inequality remains constant.

Just to give a hint: the 2010 figures show that the coefficient of variation of the per capita GNI is more than seven times that of life expectancy variable and more than four times that of the education index. Concerning the extreme values, the country with the highest life expectancy value was about twice that of the lowest one; the country with the highest education index was about ten times the lowest one; and the country with the highest per capita GNP was some 450 times the lowest one!

Additional information

Notes on contributors

Carmen Herrero

Carmen Herrero is Professor of Economics at the University of Alicante and Researcher of the IVIE, Valencia, Spain

Ricardo Martínez

Ricardo Martínez is Researcher at the University of Málaga, Málaga, Spain

Antonio Villar

Antonio Villar is Professor of Economics at the Pablo de Olavide University and Researcher of the IVIE, Valencia, Spain