By Jean Drèze
Statistical poverty lines should not become real-life eligibility criteria for food entitlements.
Nothing is easier than to recognise a poor person when you see him or her. Yet the task of identifying and counting the poor seems to elude the country’s best experts. Take for instance the “headcount” of rural poverty — the proportion of the rural population below the poverty line. At least four alternative figures are available: 28 per cent from the Planning Commission, 50 per cent from the N.C. Saxena Committee report, 42 per cent from the Tendulkar Committee report, and 80 per cent or so from the National Commission for Enterprises in the Unorganised Sector (NCEUS).
On closer examination, the gaps are not as big as they look, because they are largely due to the differences in poverty lines. The underlying methodologies are much the same. The main exception is the Saxena Committee report, where the 50 per cent figure is based on an independent argument about the required coverage of the BPL Census. Other reports produce alternative figures by simply shifting the poverty line.
In this connection, it is important to remember that the poverty line is, ultimately, little more than an arbitrary benchmark. It is difficult to give it a normative interpretation (in this respect, the Tendulkar Committee report is far from convincing). The notion that everyone below a certain expenditure threshold is “poor,” while everyone else is “not poor,” makes little sense. Poverty is a matter of degree and to the extent that any particular threshold can be specified, it is likely to depend on the context of the exercise.
What tends to matter is not so much the level of the benchmark as consistency in applying it in different places and years (by using suitable “cost-of-living indexes” to adjust the benchmark), for comparative purposes. It is this consistency that is being threatened by the current mushrooming of independent poverty lines. In this respect, the Tendulkar Committee report does a reasonably good job of arguing for the adoption of the current, national, official urban poverty line as an “anchor.” State-wise urban and rural poverty lines are to be derived from it by applying suitable price indexes generated from the National Sample Survey data. This approach permits continuity with earlier poverty series, consistency of poverty estimation between sectors and States, and some method in the madness from now on.
As it happens, the Tendulkar Committee report’s estimate of 42 per cent for rural poverty, based on this new poverty line, is not very different from the 50 per cent benchmark proposed in the Saxena Committee for the coverage of the BPL Census. In fact, the Tendulkar estimate, plus a very conservative margin of 10 per cent or so for targeting errors, would produce much the same figure as in the Saxena Committee report. Thus, one could argue for “50 per cent” as an absolute minimum for the coverage of the next BPL Census in rural areas.
However, poverty estimation is one thing, and social support is another. The main purpose of the BPL Census is to identify households eligible for social support, notably through the Public Distribution System (PDS) but also, increasingly, in other ways. In deciding the coverage of the BPL Census, allowance must be made not only for targeting errors, which can be very large, but also for other considerations, including the fact that under-nutrition rates in India tend to be much higher than poverty estimates. This gap is not so surprising, considering that the official “poverty line” is really a destitution line. The consumption basket that can be bought at the poverty line is extremely meagre. It was an important contribution of the NCEUS report to point out that even a moderately enhanced poverty line basket, costing Rs.20 per person per day, would be unaffordable for a large majority of the population. How would you like to live on Rs. 20 a day?
Also relevant here is the case for a universal as opposed to targeted PDS. The main argument is that the Right to Food is a fundamental right of all citizens (an aspect of the “Right to Life” under Article 21 of the Constitution), and that any targeting method inevitably entails substantial “exclusion errors.” This raises the question of the BPL Census methodology.
The 2002 BPL Census was based on a rather convoluted scoring method, involving 13 different indicators (related for instance to land ownership, occupation and education) with a score of 0 to 4 for each indicator, so that the aggregate score ranged from 0 to 52. There were serious conceptual flaws in this scoring system, and the whole method was also applied in a haphazard manner, partly due to its confused character. The result was a very defective census that left out large numbers of poor households. According to the 61st round of the National Sample Survey, among the poorest 20 per cent of rural households in 2004-05, barely half had a BPL Card. Any future BPL Census exercise must be based on a clear recognition of this major fiasco.
The Saxena Committee recently proposed an alternative BPL Census methodology, involving a simplified scoring system. Instead of 13 indicators, there are just five, with an aggregate score ranging from 0 to 10. This is a major improvement. Even this simplified method, however, is likely to be hard to comprehend for many rural households. This lack of transparency opens the door to manipulation, and undermines participatory verification of the BPL list. There is no guarantee that the results will be much better than those of the 2002 BPL Census.
Perhaps the proposed method can be further improved. But the bottom line is that any BPL Census is likely to be a bit of a hit-or-miss affair, not only because of inherent conceptual problems but also because of widespread irregularities on the ground. This is the main argument for universal provision of basic services, including access to the PDS. Another strong argument is that targeting is divisive, and undermines the unity of public demand for a functional PDS. It is perhaps no accident that the PDS works much better in Tamil Nadu, where it is universal, than in other States.
A universal PDS would, of course, involve a major increase in the food subsidy. However, universalisation could be combined with cost-saving measures such as decentralised procurement, self-management of Fair Price Shops by gram panchayats, and a range of transparency safeguards. There is no obvious alternative, if we are serious about ensuring food security for all. If someone has a better idea, let’s hear it.
Meanwhile, the government seems to be running in the opposite direction, judging from the recent recommendations of the Empowered Group of Ministers (EGoM) in charge of the proposed National Food Security Act. The EGoM suggested not only that the government’s legal obligation to provide foodgrain under PDS should be restricted to 25 kg per month for BPL families, but also that the Planning Commission’s measly poverty figures should be used as a “ceiling” for the BPL list. This amounts to disregarding at least three official committee reports (Tendulkar, N.C. Saxena and NCEUS), and trivialising the proposed Act.
In a country where half of all children are underweight, the idea that freedom from hunger and under-nutrition can be made a legal right is rather bold and far-reaching. It has a bearing not only on the Public Distribution System but also on a range of other interventions and entitlements, relating for instance to child nutrition, social security, health care, and even property rights. Framing an effective National Food Security Act requires a great deal of creative work, public debate, and political commitment. Alas, seven months after the Finance Minister stated, in his previous budget speech, that work on the Act had “begun in right earnest,” and that a draft would be in the public domain “very soon,” things seem to be moving backward rather than forward. Let us see what the Honourable Minister has to say on this in his forthcoming budget speech.
(The author is Visiting Professor at the Department of Economics, University of Allahabad.)