India. Craig Jeffrey

these expert groups – with what outcome is unknown at the time of writing (in March 2019, by which time no official estimates of poverty levels had been published since 2011–12). The upshot has been the proliferation of quite wildly divergent estimates of the extent of poverty in India, produced by different official bodies. The Tendulkar Committee produced an estimate of 37.2 per cent of the population in 2004–05 (407 million people), when the (old) official estimate for that year stood at 27.5 per cent (302 million people). These are ‘headcount poverty ratios’ – referring to the proportion of the population unable to meet the minimum consumption requirements set out in the methodology described earlier. At about the same time, and with reference to the same date, another committee, chaired by N. C. Saxena, argued that for practical purposes – of providing for the welfare of the people – the estimate of the incidence of poverty should be increased to 50 per cent; while the National Commission on Enterprises in the Unorganised Sector, chaired by Arjun Sengupta, estimated that 77 per cent should be considered poor. The World Bank estimate of the incidence of poverty in India in 2005 was 40 per cent; that of the Asian Development Bank, about 50 per cent. S. Subramanian has commented that ‘the official methodology has thrown the door open to complete anarchy’ (2016). And this remark is with reference to the preferred conventional procedures for measuring poverty levels, based on NSSO data on household consumption expenditure. Another whole dimension of debate arises from the fact that observations of mean consumption reported in India’s National Accounts are significantly (and increasingly) above those provided by the survey data – though the survey data are generally preferred, because they measure living standards directly.

That there should be so many different estimates points to the considerable technical difficulties associated with defining and measuring poverty – which involves a whole series of judgement calls. A respected Indian economist, A. Vaidyanathan, who has devoted much of his long professional life to these matters, has written that ‘it is not possible to arrive at a definitive estimate of poverty incidence that can be used as a reasonably robust benchmark’ (Vaidyanathan 2013: 41. See also Krishnaji 2012 for a similar argument by another senior economist). He was referring here especially to the problem of identifying ‘the poor’ for the purpose of allocating welfare benefits supplied by government. An important instance of this is that of access to foodgrains and other basic consumer goods at controlled, low prices, through India’s Public Distribution System (PDS). This has been targeted, in principle, since 1997, to those defined as being ‘Below the Poverty Line’ (BPL). But how are the ‘BPL’ to be identified? The main reason why the poverty line calculations offered by the Planning Commission in 2011 became the subject of so much controversy was that it was suggested that the poverty estimates to which they gave rise would be used for targeting purposes – though the Tendulkar Committee had expressly advised against this. The absurdity of the procedure was pointed up in a cartoon in the English language newspaper The Hindu (22 September 2011), which showed an obviously poor man crossing the road to join his family, living on the pavement, and saying to a passing motorist who was offering him some coins, ‘No thanks. I’ve earned Rs. 25 today. I don’t want to lose my BPL status’.

      This is not to say that the attempt to measure the incidence of poverty in a country, to track trends over time, and to make comparisons across regions, is worthless. These certainly are important data for any responsible government, and in a democracy should be expected to play a part in its evaluation by the electorate (so governments have an incentive to set the poverty line low, in order to enhance the chances of producing a favourable impression). It is for this reason that we go on to review evidence on poverty trends in India, and on the extent of poverty among different social groups, as these are given by the conventional monetary measures. But we should remind ourselves that the poverty line is a construction, not a truth (and, in India, a construction that might be significantly improved – see Subramanian 2014). An aggregated measure does not, of course, take account of differences between individuals, between, for instance, those who are disabled in some way and others, or between people living in different localities with very different needs in terms of clothing and shelter. And the monetary measure of poverty may not take account of important differences that have to do with public services, such as the availability or not of clean water, which exercises such an important influence on nutrition and health. So, as Vaidyanathan has also argued

      Strategies to address the myriad and varied disabilities of the poor cannot be decided on the basis of the overall incidence of income poverty. They need to be based on assessments of the deficiencies of access and realization relative to accepted minimum desirable levels of specific components of living standards such as food intake, unemployment and underemployment, housing, connectivity and indicators of health and education status’ (Vaidyanathan 2013: 41).

Latterly, several different attempts have been made to come up with multidimensional assessments of the incidence of poverty, taking account of direct measures of nutrition and health, and of other indicators of living standards, of which the best known is that of the Oxford Poverty and Human Development Initiative (www.ophi.org.uk). The Multidimensional Poverty Index (MPI), published since 2010 by the United Nations Development Programme (UNDP), is based on the work of the Oxford Group. We consider this, too, in what follows, where we discuss the findings of the mainstream of poverty research relating to India.

3.3 Poverty Trends in India

      Economists associated with the World Bank, Gaurav Datt, Rinku Murgai and Martin Ravallion, point out that among developing countries, India has the longest series of national household surveys suitable for tracking living conditions, and they have reported on their findings from these data for the 60 years from 1957 to 2012 (Datt et al. 2016a, 2016b). Their analysis refers to the definition of the poverty line established by the Planning Commission in the 1970s (the one anchored in the nutritional norms of 2,400/2,100 calories per person per day), updated in the conventional way, using urban and rural price indices. They find that there has been a definite downward trend in poverty since the early 1970s, according to the standard measures, with acceleration after 1991 – when India embarked upon the economic reforms described in chapter 2 – despite increasing inequality. Poverty reduction since 1991 has been more responsive to growth than it was before – there has been a higher growth elasticity of poverty incidence reduction. Still, as Ravallion has shown elsewhere, the growth elasticity of poverty in China, historically, has been about twice that of India, probably because of more favourable initial conditions. With relatively low inequality in access to land and human capital, the poor in China have gained more from growth (though over time these favourable conditions have been eroded with increasing inequality in incomes and in education and health), while opportunities for the poor in India have been restricted by low levels of education and health (Ravallion 2012).

The strong suggestion in this analysis that India’s policy shift at the beginning of the 1990s, and the higher and more consistent economic growth that the country has seen since then, has had positive outcomes in terms of poverty reduction, is apparently confirmed in other analyses that focus more on the trends of the last two decades. Himanshu and Kunal Sen, whose analysis refers to the Tendulkar measure of the poverty line, present evidence that leads them to the ‘inescapable conclusion that poverty has declined faster [over the period 2004–05 to 2009–10, when India was experiencing ‘superfast’ growth] than in the previous period [1993–94 to 2004–05]’ (2014: 78). They estimate that the rate of poverty decline in the more recent period was 1.12 per cent per annum (even though their figures are lower, for reasons that they explain, than those of the Planning Commission), compared with 0.74 per cent per annum over the previous decade. One measure of the intensity of poverty, the poverty gap, shows marginally better improvement in the later period than the earlier one; but according to a second measure, the squared poverty gap, performance in the later period was worse. This suggests the possibility that the poverty decline was associated with an increasing share in the population of those who are vulnerable to falling into poverty (see also Dang and Lanjouw 2015). And Himanshu and Sen find slower improvements in consumption expenditure among the most vulnerable people – those employed as agricultural labourers in the rural