Monday, August 31, 2015

Human Development Index (HDI) - Time for an overhaul?

The Human Development Index is composite index prepared annually by UNDP, which tries to give a single numerical score (from 0 to 1) to measure the extent of human development in various nations. Whenever the HDI report is released, the Indian media normally focuses on two aspects – what is India’s rank (135 out of 187 nations), and how does it compare to Pakistan (146 – India is ahead) and a few other nations. But its time to question the validity of the Index itself.
Human Development is multi-dimensional, and no single numerical measure can truly capture the essence of development. This is one of the drawbacks of any such Index. At the same time, a single number does allow people to focus on the issue – simplicity has its advantages. But as Einstein is said to have stated – “Everything should be made as simple as possible, but not simpler”
Lets see if HDI appears a valid score of development.
How is HDI Computed?

HDI looks into three aspects, Long and healthy life, Knowledge, and standard of living (Illustrated below- Technical note to HDI Report). For each of these dimensions, a proxy is used to compute a score between 0 and 1. A geometric mean of the three scores is then taken to arrive at the HDI Score.
An arbitrary cutoff of 0.550. 0.700 and 0.800 is then used to categorise the nations as Very High Human Development (>=0.800), High Human Development (0.700-0.799), Medium Human Development (0.550-0.699) and Low Human Development (<0 .550="" p="">Based on the 2013 report, this neatly categorises the 187 nations into four almost equal sized groups.
The variation of the three individual components of Health, Education and Income which leads to the HDI Score can be seen in the scatter plot below:
The Balancing Act of Income Index
The striking feature in the above scatter plot is the variation in the Index score for Income. Every other proxy for development, the number of years of schooling (average and expected), and the life expectancy at birth contributes linearly to its related Index. However, the Income index based on the Gross National Income GNI per capita on PPP (Purchasing Power Parity) grows on a logarithmic scale, ie it rises very fast initially, even for small increases in GNI. The index score for GNI is designed to scale logarithmically from a baseline income of $100 upto a maximum per capita GNI of $ 75,000. The logarithmic scaling between $100 and $75,000 implies that an individual gets 40% of the benefit that money can buy with an annual GNI of $1500 , which is just 2 % of the maximum upper GNI of $75,000. Does this appear reasonable?
The rationale for these two numbers as given in the HDI report is:

“The low minimum value for gross national income (GNI) per capita, $100, is justified by the considerable amount of unmeasured subsistence and nonmarket production in economies close to the minimum, which is not captured in the official data. The maximum is set at $75,000 per capita. Kahneman and Deaton (2010) have shown that there is a virtually no gain in human development and well-being from annual income beyond $75,000”

The World Bank itself uses a poverty line of PPP adjusted  $1.25 per day. This translates to an annual threshold of $ 456.25.

Then why would UNDP use a minimum of GNI of $ 100 a day to start scoring on the Income Index? Is it to arrive at "acceptable" HDI scores?

The study of Kahneman and Deaton provides valid grounds for two aspects relating to the Income Index as used in HDI computation. One, the subjective well-being is best measured against the logarithm of Income, and not absolute income (in keeping with the basic fact of perception known as Weber’s Law), and two,  the effects of income on the emotional dimension of well-being would satiate fully at some level - ∼$75,000.
However, Kahneman had Deaton had based it on a survey of Household income, and not per capita income as is used here. Further, the survey was based on US respondents. More importantly, as their paper states: "Some observations were deleted to eliminate likely errors in the reports of income. The GHWBI (survey) asks individuals to report their monthly family income in 11 categories. The three lowest categories—0, <$60, and $60–$499—cannot be treated as serious estimates of household income."
The above statement is damning - data relating to people below a monthly stated income of $499 was simply ignored. So how can one  assume that this logarithmic scaling is valid even for incomes below $499? So even if we accept that income satiation level is reached at a per capita GNI of $75,000, what grounds do we have to support the lower level of $ 100? Why not start at $10, or $1,000, or $5,000? Note that the lower limit used for HDI Income index is GNI of $100 per year.
By varying the lower baseline number,large changes can be effected on the Income index score without materially altering the well-being of the individuals in different nations.
If we use the PPP adjusted poverty line of $ 1.25 per day or $456.25 (rounded to $ 460 per year), the number of countries in the lowest HDI Category increases from the current number of 43 to 56, a 30% increase in the number of low HDI countries. 
The number of countries in each category by using a lower income level of $ 460 per year is given below:


And this is how the world looks based on the revised scaling of Income index:
Yes - India now moves to the lowest category of HDI, with a score of 0.474 against an earlier score of 0.586.
The march of HDI - Development over the  years
This is how the HDI score looks over the years for all nations for which data was made available with the HDI report for 2014. (Each line represents one country, showing its HDI score over the decades)
While there is an apparent onward march with increase in HDI score over the decades, I now wonder how much of it is due playing with the model leading to the Index computation?
What is however striking is how few countries are able to break into a different HDI category, delineated by the cutoff scores of 0.550, 0.700 and 0.800.

 (The colours used are based on the HDI category as of 2013 HDI report)