In the last column (1st October), I mentioned HDI (human development index), the core measure used by UNDP’s Human Development Report (HDR) since 1990. Since 1990, HDI has gone through some modifications. Measuring human development and its opposite, human deprivation, is inherently subjective. There are implicit value judgements, even if we do not always recognise them explicitly. For instance, is it desirable to provide clean drinking water through taps? Most people will naturally and understandably reply in the affirmative. But what’s the objective, clean drinking water or clean drinking water through taps? In parts of India, clean drinking water is available through streams or wells. HDI believes three strands can capture development -– a long and healthy life, knowledge and a decent standard of living. No one will contest the inclusion of these three strands, but many people will say there should be other heads too. In other words, HDI can be criticised for the omission, not the commission. When HDI originally evolved, there was a lot of discussion and debate. The idea was that there should be a simple measure, one that policy-makers can identify with. If one plugs in too much, matters may become unnecessarily complicated, and after all, various indicators of development are correlated with one another. But the simple point is the following. There is nothing sacrosanct about these three heads. They are subjective, not objective. If someone else comes along with another index, that will be as good or bad as this one.
A head like a long and healthy life is at a general level. One can think of SDGs. Those goals are at a general level. One must quantify and measure them. In the case of SDGs, that’s done through indicators. Here too, a long and healthy life must be pinned down through an indicator. UNDP uses life expectancy at birth. Once again, one can ask questions. Why life expectancy at birth alone? Why not, in addition, a life expectancy of 60 years? That also tells us something about a long and healthy life. Why not some indicator of morbidity, in addition to that on mortality? Similarly, for knowledge, we have two indicators, mean years of schooling and expected years of schooling. Finally, for a decent standard of living, we have per capita income, expressed in PPP (purchasing power parity) US dollars. I have flagged two areas of subjectivity – first, in the choice of heads and second, in the choice of indicators under those heads. Don’t get me wrong. Any such exercise will involve subjectivity, and UNDP is no exception to this principle. Let me now flag the third area of subjectivity, the construction of an index. For example, let’s say India’s life expectancy at birth is 70 years. That figure is not used as it stands. HDI normalises it, as does much such empirical work, to reduce it to an index that is between 0 and 1. In the world today, the minimum life expectancy at birth is 20 years, and the maximum is 85 years. (These are UNDP assumptions. There are countries with a life expectancy of more than 85 years, and, as far as I know, there is no country with a life expectancy as low as 20 years.)
In any event, HDI takes the actual life expectancy of the country (in years) and subtracts the minimum of 20 from this. This number is then divided by 65 years (85 minuses 20, maximum minus minimum) to arrive at an index for long and healthy life. Something similar is done for PPP per capita national income, with a presumed maximum of 75,000 US dollars and a presumed minimum of 100. Again, there are countries with more than 75,000, and I don’t think any country has a figure as low as 100. Be that as it may, we have a long and healthy life index and a decent standard of living index. But there is an additional issue for knowledge or education. There are two separate indicators here, and the arithmetic average (mean) is taken of these. We are still left with three different indices for the three heads, and a geometric mean is taken of these. That, too, is subjective. A geometric mean implies equal weights. Is it necessarily obvious these three heads should have equal weights? Finally, we come to the matter of data. Where do we get data to compute the indices? Are there time lags? If data aren’t available, how are missing numbers computed? My intention is not to lambast HDI as a measure. It makes heroic assumptions and so will every such similar exercise. Therefore, when we look at HDI, we should use it to measure a country’s improvement (or deterioration) over a period of time, where those heroic assumptions cancel themselves out, so to speak. Although everyone loves rankings, I would personally take the cross-country rankings, based on HDI, much less seriously. (In addition, over time, the number of countries also changes, and that, too, affects the rankings.)
Consider HDR for 2021-22. HDR, of course, has much more than HDI. In 2021, we were told India’s HDI value is 0.633, with a life expectancy at birth of 67.2 years, expected years of schooling of 11.9 years, mean years of schooling of 6.7 years and PPP per capita income of 6,590 US dollars. That HDI figure in 2021 declined from 0.642 in 2020 to 0.645 in 2019. Given the Covid pandemic, a decline is understandable. However, in the middle of the pandemic, how did UNDP obtain data for the four indicators under those three heads? That’s not very obvious. For instance, where did we get that life expectancy figure? Where and how did we get the PPP per capita income figure? The same point can be made about education/knowledge indicators. An ostensibly sophisticated methodology, albeit with value judgements, isn’t very credible if it lacks transparency on sources of data. (Notes in HDR aren’t quite forthcoming.)
Bibek Debroy
Chairman, Economic Advisory Council to the PM