one-dimensional focus on income.10
As a single-variable, summary measure of economic well-being the HDI has considerable political and popular appeal.11 However, complex socio-economic phenomena are seldom if ever accurately describe by such a measure. There is, in particular, no rationale for the equal weights assigned to the three components of the HDI – or, for that matter for the 2-to-1 weights assigned to the two components of the knowledge/education index. Why, for example, would we think that a one percent increase in the log of income should be equal in value to a one percent increase in the average life span – and why should this trade-off be the same at all levels?12 These formal problems in creating an index of economic well-being are not simply technical issues; they reflect real problems – perhaps the impossibility – of describing and
10 It is somewhat ironic, however, that a departure from the one-dimensional focus on income has been used as a basis for arguing the distribution of economic well-being in the world has been becoming more equal; see Becker et al (2005), who argue: “The effects of health are sufficient to revert the results regarding the evolution of cross-country inequality up to the 1990s. Once health is accounted for, there is a significant reduction in inequality throughout the world up to 1990 and, even with the AIDS epidemic, a much more significant reduction in inequality between 1960 and 2000 than can be perceived from income alone.” Given the nature of income, health and education/knowledge, it is of course the case that the inequality represented in the HDI will be less than the inequality in the distribution of income. Also, it is by no means surprising that the movement in the distribution of the latter two variables over time has been rather different than the movement of income distribution. But the meaning of these phenomena for appraising the distribution of economic well-being in the world is by no means an obvious or non-controversial issue.
11 In a 1999 address, the Deputy Chairman of the Indian Planning Commission, Shri K.C. Pant, defended the HDI in the following terms: “Over the years there has been some criticism of the human development index (HDI) on the ground that it is too simplistic and limited a measure of the extremely wide range of factors that determine a person’s quality of life. While such a criticism may be valid in terms of the academic appropriateness of the index, it completely misses the true utility of the concept. The average human mind, despite its potential to handle a wide range of concepts, feels most comfortable with a simple and intuitive idea even while recognising its limitations in representing complexity. It is perhaps for this reason that a summary measure like the per capita income has captured the thought processes not only of the average person but even of the academic world. Any effort to supplant this measure in a manner which would bring in the social aspects of human development would necessarily have to devise a measure which is equally simple and easy to understand. The HDI in my opinion has served this purpose admirably and has been instrumental in bringing about a significant shift in development perspectives.” Pant (1999).
12 This criticism of the HDI – that the weights have no rationale – is somewhat vitiated by another criticism of the index, namely that that the three basic components are so highly correlated with one another as to make the index redundant. This problem of redundancy was raised early on in the HDI’s evolution by McGillivray (1991). More recently, Cahill (2005) has demonstrated again the high degree of correlation among the three components, and he has pointed out that because of this correlation the index itself is not very sensitive to the weights that are used. A further problem, but one that may also make little practical difference in light of the redundancy, is that the upper and lower limits used to establish the separate indexes that make up the HDI are in some cases arbitrary. Why, in particular, is 0 taken as the lower limit on literacy and enrollment, respectively, when no country is near this level, while the lower limits on income and life expectancy are set at $100 and 25 years, respectively? See the appendix. Nonetheless, whatever quantitative problems exist with the HDI, its conceptual value – its focus on the multidimensional nature of human well-being – is undiminished.