There has been an upsurge of public discussion on a number of inter-related issues revolving around official assessments of poverty, the linking of welfare entitlements to poverty status, the reasonableness of officially stipulated money-metric poverty lines, the relative virtues of universalisation and targeting of welfare benefits, and the fiscal sustainability of increased public spending in the cause of poverty redress. The present essay offers a very brief evaluation of the methodology of poverty identification advanced by an Expert Group of the Planning Commission in 1993, and undertakes more elaborately what is essentially a critical assessment of the 2009 Tendulkar Committee’s approach to specification of the poverty line.

This essay is based in considerable measure on a part of a much more extensive working paper (Subramanian 2010). I would like to thank D Jayaraj and Sanjay Reddy for many helpful discussions, over a long period of time, on the subject of this paper, and A Arivazhagan for help with the graphics.

S Subramanian (subbu@mids.ac.in) is on the faculty of the Madras Institute of Development Studies, Chennai, and a Visiting Fellow at the Indian Institute of Technology Madras.

Economic & Political Weekly

1 Background

I

In the wake of the public outcry against these poverty norms, and in particular the opposition to linking welfare (especially food) entitlements to poverty caps derived from the headcount ratios of the population living below the rural and urban poverty lines stated above, it has been clarified by the Planning Commission that these lines will not be employed to cap entitlements. One hopes the clarification is seriously and sincerely intended. If it is, then at least one incentive for understating poverty through understated poverty lines would be eliminated. (It is instructive, in this regard, to recall that even the

vol xlvi no 48

reckoning, it was unrealistic to expect that foodgrains could be delivered free to as many as 37% of the Indian population

– which, precisely, is the Tendulkar Committee’s estimate of the national headcount ratio of poverty in 2004-05.)

If the declared determination to dissociate welfare entitlements from poverty caps should afford a reprieve to the overriding concern of confining poverty estimates to “manageable” magnitudes and trends, then this could pave the way for a somewhat more relaxed and rational consideration of the principles by which a money-metric poverty line might be derived. An important guiding principle in this regard is the elementary requirement of an unvarying poverty standard to be employed in poverty comparisons across space or over time. “Resourcist” approaches to specifying a money-metric poverty line have generally tended to identify this unvarying poverty standard in some given level of “real income” (that is, income corrected only for variations in price levels), or in some given basket of commodities (to be valued at the prices that are relevant for the context under examination). “Capabilitybased” approaches, on the other hand, seek invariance of the poverty standard in some identified set of absolute capabilities, seen as being reflected in the achievement of a set of corresponding functionings, such as, for example, freedom from hunger, illiteracy and ill-health: the poverty line is then the total cost of achieving these functionings.

Viewing poverty as a capability deprivation is compatible with drawing a moneymetric poverty line which is not unique across individuals or contexts/environments. Specifically, personal heterogeneities can be reflected in variations across individuals in the ability to convert resources into functionings: disabled individuals or lactating mothers, for example, would typically require more resources than others to achieve some given combination of “minimally satisfactory” functionings. Similarly, variations across contexts and environments could also call forth variable resources to address the requirements of a given set of functionings: people living in cold climates would require more clothing than those living in temperate climates to be adequately protected, for instance, just as people exposed to the risk of gastrointestinal infection may require more nutritional input than others to achieve a given level of nutritional adequacy. This is the basis for Amartya Sen’s (1983) well-known observation to the effect that poverty postulated in an absolute sense in the space of functionings is compatible with relative, or variable, real incomes or commodity bundles or resources in general that are required to achieve these functionings. The reader is referred to Sanjay Reddy (2004) for a lucid explication of the principles underlying the capabilitybased approach to specifying a moneymetric poverty line, and of the conceptual soundness of seeking invariance of the poverty standard in the space of functionings rather than in that of real incomes or commodity bundles.

Official approaches to the specification of a poverty line for India have unfortunately not been inspired by any easily discernible conceptual clarity underlying the identification exercise. In particular, it is hard to confer any meaningful interpretation of the official poverty lines – whether we speak of the lines derived from a methodology advanced by a 1993 Expert Group appointed by the Planning Commission or of the lines implied by the methodology advanced by the recent Tendulkar Committee – as actually constituting the amount of income that would be required in order to achieve a set of basic pre-specified human functionings needed for escaping the condition of deprivation.

In this brief critique, I shall concentrate on aspects of the Tendulkar Committee’s Report, which has served as the basis for the controversial poverty lines mentioned in the Planning Commission’s affidavit. As a preliminary, I present a very short discussion of the 1993 Expert Group’s methodology, the perceived inadequacies of which, presumably, led to the setting up of the Tendulkar Committee.

2 The 1993 Methodology

The Expert Group appointed by the Planning Commission endorsed, in a Report submitted in 1993, a methodology that had earlier been advanced by a task force in 1979. The approach consisted in deriving a “poverty line consumption bundle” as the one revealed by National Sample Survey data on the distribution of both consumption expenditure and calorie intake across expenditure size-classes, as corresponding to that level of expenditure at which a predetermined calorific norm (2,400 kilocalories per person per day for the rural areas of the country and 2,100 kilocalories for the urban areas) was observed to be achieved in 1973-74. The poverty line in other years was, effectively, taken to be simply the expenditure level yielded by the “reference year (1973-74)” commodity bundle valued at current prices.

It is, naturally, difficult to comprehend why 1973-74 – or, indeed, any other particular year – should be accorded the privileged status of a “reference year” by (implicitly) imbuing the pattern of consumption expenditure that obtained in that year with some special normative significance. As it happens, if we employ the 1973-74 poverty line consumption bundle as the unvarying standard for poverty comparisons, and proceed to obtain the poverty lines in subsequent years by simply revaluing the reference year commodity bundles at current prices, then we obtain a pleasing trend of declining headcount ratios over the period 1973-74 to 2004-05.

Elementary demand theory – see Subramanian (2005) – suggests that, under plausi ble conditions relating to the relative rates of price increases in food and nonfood items, changes in preferences in f avour of non-food, and the dwindling free availability of common property r esources (such as fuel for firewood), the declining poverty trend mentioned earlier must be an inbuilt and logically inevitable feature of the Planning Commission methodology – as, indeed, would be a drift over time of actual calorific intake at the poverty line from the original calorific norm one started out with. (For important critiques of the Planning Commission methodology, the reader is also referred to Ray and Lancaster (2005) and Patnaik (2007).)

Indeed, it can be shown that if we were to progressively move the reference year forward from 1973-74 (to 1977-78, 1983, 1987-88, 1993-94, 1998-99, and 2004-05, which are the years for which official consumption expenditure surveys are avail able), then while the declining trend obtained from treating 1973-74 as the reference year is preserved, the magnitudes of the poverty rate keep increasing. One can go further and ask: if any one year may be treated as a reference year, what is the logical bar on treating each year as a reference year? Utsa Patnaik (2007) resorted to just such a procedure, and presented, on the basis of this approach, an increasing trend in India’s rural poverty!

As Sanjay Reddy (2007) points out, none of these wholly inductive approaches to determining the poverty line has anything to commend it. It is, of course, possible to advance the case for a normative poverty line commodity bundle which is not based entirely on any particular year’s pattern of consumption expenditure – but then it is still not clear why one must also demand invariance of that bundle over time or across space. Conceptually, it would appear – and this takes us back to the discussion in the introductory section of this essay – that what is dictated is invariance of the poverty standard in the functionings space. Briefly, the possibility of initiating appropriate anti-poverty policy on the basis of approaches to poverty identification such as those subsumed in the 1993 Expert Group’s methodology is poorly served by the unreliable diagnostics – in relation to magnitudes and trends of a substantively meaningful conception of poverty – yielded by these approaches. Much was therefore expected of the Expert Group to Review the Methodology for Estimation of Poverty which was appointed in December 2005, under the Chairmanship of Suresh Tendulkar. The Report of the Tendulkar Committee was submitted in November 2009 (Planning Commission 2009). It is this Report that will be the principal concern of the rest of this essay.

3 Some Aspects of the Tendulkar Committee’s Report

The present discussion of the 2009 Expert Group’s recommendations will be confined to the identification aspect of the

vol xlvi no 48

This volume investigates the nature of economic growth in India, its pace over time, its relationship to changes in the policy regime and the role of the external sector, and uses data to evaluate the policies that have implicitly underpinned the changes.

Presenting a range of approaches, views and conclusions, this collection comprises papers published in the Economic and Political Weekly between the late 1990s and 2008 that are marked by an empirical awareness necessary for an understanding of a growth history. The articles reflect a certain groundedness in their approach in that they privilege content/context over methodology.

This volume is an important addition to the literature on post-liberalisation economic growth in India. It will be useful to students and scholars of economics and management.

Contributors include Deepak Nayyar • Rakesh Mohan • Atul Kohli • Arvind Panagariya • Kunal Sen • Neeraj Hatekar • Jessica Seddon Wallack • Pulapre Balakrishnan

• Ravindra Dholakia • Ramesh Chand • R. Nagaraj • Montek Ahluwalia • Shashank Bhide • Amit Bhaduri • Pranab Bardhan

Readings on the Economy, Polity and Society

This is the first book in a new series jointly published by EPW and Orient Blackswan. The series is being published as part of a University Grants Commission project to promote teaching and research in the social sciences in India. The project (2010-12) is being jointly executed by the Tata Institute of Social Sciences, Mumbai, and the Economic and Political Weekly. The series is meant to introduce university students and research scholars to important research that has been published in EPW in specific areas. The readers draw on the EPW ’s archive of published articles.

Pp xiv + 454 ISBN 978-81-250-4271-6 2011 Rs 445

Orient Blackswan Pvt Ltd

november 26, 2011 vol xlvi no 48

PERSPECTIVES

India recommended by Food and Agriculture Organisation (FAO) is currently around 1,800 calories per capita per day which is very close to the average calorie intake of those near the new poverty lines in urban areas (1,776 calories per capita) and higher than the revised FAO norm (1,999 calories per capita) in rural areas in the 61st round of NSS.” But, surely, this is a purely fortuitous outcome, and one fails to see the

Assuming, for a moment, some merit to this method of estimating the normative food expenditure level, at least one difficulty with such a “validation check” is represented in Figure 1. We have three (hypothetical) cumulative density functions (cdfs) Figure 1: Determining Normative Food Expenditure Levels

•

age-group) children in the household at the PCTE to school…” (Planning Commission 2009: 9). The normative level of expenditure on health is taken to be given by the expected value of expenditure on treatment/hospitalisation, itself the product of the probability of the onset of illness requiring treatment/hospitalisation (taken to be represented by the incidence of treatment/hospitalisation) and the median cost

of treatment/hospitalisation. A couple of

1.0

H

Cumulative density function of X

signi ficance of these sorts of circumstan

tial detail. On the other hand, if the choice

of the poverty line had indeed been influ

enced by congruence with a nutritional

norm, then we would be back in the world

of the 1993 Expert Group! Also apparently

0

remarks are in order. Why is the median

cost a normative cost? Costs are likely to

H

1

H

2

rise when treatment/hospitalisation tends

towards greater completeness/comprehen

siveness: the median cost in a poor econ

counting in favour of the 2009 Group’s

omy is scarcely likely to be reflective of the

* * *

choice of poverty line is that it “…happens 0 XXXcost that would be incurred in order to

102

Food expenditure X

to be close to, but less than, the 2005 PPP

•

finance a reasonably comprehensive course

$1.25 per day poverty norm used by the World Bank in its latest world poverty estimates” (p 8, op cit). Whether or not this is good advertisement for the poverty line is at least debatable: it must be noted that the World Bank’s global poverty estimates are by no means widely perceived as being acceptable, and are, indeed, severely discredited in certain quarters (the reader is referred to the particularly important work, in this context, of Reddy and Pogge 2010).

4 On the Adequacy of the Poverty Lines

The 2009 Expert Group is also of the view that “external validation checks” carried out by it suggest that the actual expenditures incurred on nutritional, educational, and health outcomes at the prescribed poverty line expenditure levels in 2004-05 are adequate to cover normative levels stipulated by the Expert Group in this regard. In the matter of normative food expenditure, the Expert Group specifies this level by reading off the level of food expenditure, from the cumulative density function (CDF) of expenditure on food, corresponding to the proportion of the population under nutritional stress (interpreted as the simple average, obtained from National Family Health Survey-III data for 2005-06, of the headcount ratio of under-5 underweight children, the headcount ratio of women in the age group 15-49 with low body mass index, and the headcount ratio of men in the age group 15-49 with low body mass index).

for food expenditure, where expenditure is represented in “real”, i e, “constant prices” terms. The cdf labelled 1 refers to the reference year, the cdf l abelled 0 refers to a year preceding the reference year, and the cdf labelled 2 refers to a year succeeding the reference year. Suppose H1 is the headcount ratio of those under nutritional stress in the reference year. From cdf 1 we can see that the expenditure level corresponding to the headcount ratio H1 is X *1. Effectively, the Expert Group employs X *

1

as the normative level of adequacy for food expenditure. Suppose also that X * is

1

the observed level of expenditure on food at the poverty line in each of the years 0, 1 and 2. Additionally, suppose that H0 and H2 are the proportions of the nutritionally deprived populations in years 0 and 2 respectively. Then Figure 1 suggests that the normative levels of food expenditure in years 0 and 2 should be, respectively, X*

0 ( X*) and X * ( X * ). The actual food ex

1 2 1

penditure is adequate to cover the normative requirement in year 1, but not in years 0 and 2. How, in other words, can we tell a priori that just because the poverty line expenditure on food is adequate for a certain “normative” food expenditure level in the reference year, this will also hold for all other years?

The normative level of expenditure on education is taken to be given “…by the expenditure required at state-specific median cost (derived from the 61st round employment-unemployment survey) for sending all school-going (in 5-15 year of treatment or hospitalisation. Second, the proportional incidence of treatment/ hospitalisation is unlikely to be the probability of the onset of illness requiring treatment/hospitalisation: the true incidence of illness requiring treatment will, in an environment of poor affordability, typically be larger than the incidence of illness actually treated. There is therefore good reason to believe that these “normative” expenditure levels on education and health are underestimates.

Indeed, a simple back-of-the-envelope exercise should provide a rather more robust “external validation” (or failure of validation) than is afforded by the convoluted tests proposed by the Expert Group. The 2009 Expert Group’s urban poverty line for Tamil Nadu in 2004-05 is reported to be nearly Rs 560 per person per month, or (without allowing for economies of scale), Rs 2,240 for a household of four members per month. To suggest that such a household would not, in 2004-05, have been in poverty strikes the present author (as a resident of Tamil Nadu’s capital city) as being optimistic in a grimly determined way. In 2004-05, a modest tea-shop singledish meal (such as a plate of lemon rice) would have cost Rs 10. Allowing two meals a day to keep out hunger, a household of four would require Rs 80 per day to spend on food. Let us cut this requirement down to Rs 60 per day to take (exaggeratedly unreal) account of the fact that home-cooked food is cheaper. This works out to Rs 1,800 per month (30 days). A

Economic Political Weekly

EPW

november 26, 2011 vol xlvi no 48

modest (by which one means cramped and depressing) one-room house with attached kitchen and common toilet (for a family of four!) would have easily commanded a monthly rental of Rs 1,500 in Chennai in 2004-05. Adding this to the food expenditure requirement of Rs 1,800 already brings up the total to Rs 3,300 – making for a deficit, in relation to the poverty line of Rs 2,240, of Rs 1,060. And we have not even counted the cost of education or clothing or transport or the occasional movie or cigarette, not to mention such exotica as a broken bone or a bout of typhoid. This does mark a contrast from the situation as projected by the Report of the Expert Group (Planning Commission 2009: 2) when it says:

Even while moving away from the calorie norms, the proposed poverty lines have been validated by checking the adequacy of actual private expenditure per capita near the poverty lines on food, education and health by comparing them with normative expenditures consistent with nutritional, educational and health outcomes. Actual private expenditures reported by households near the new poverty lines on these items were found to be adequate at the all-India level in both the rural and the urban areas and for most of the states.

One is irresistibly reminded, here, of the fate of G V Desani’s immortal comic hero H Hatterr when he recalls the head of his orphanage-school from the days of his childhood:

The sort of loco parentis who’d shower on you a penny, and warn you not to squander it on woman, and wine, and song!1

5 Concluding Observations

The preceding discussion suggests that there is good reason why official assessments of poverty have met with the sort of generalised public dissatisfaction that has been in evidence in the wake of the recent Planning Commission affidavit submitted to the Supreme Court. As we have seen, the identification methodologies advanced by successive task forces and expert groups have little by way of logical appeal or outcome-plausibility to commend them. A major cause for this must be attributed to the absence of any real effort made by our planning agencies to identify a set of reasonably spatially differentiated and discriminating poverty lines based on a costing of minimally satisfactory levels of functioning with respect to an agreed- upon list of basic human capabilities. The case for setting up a permanent poverty monitoring and assessment bureau for this purpose presents itself (Reddy 2007). There is also, of course, a strong case for a direct assessment of multidimensional poverty in functioning space, and an example of an exercise of this nature is available in Jayaraj and Subramanian (2009).

Meanwhile, if what one is after is a purely money-metric indicator of how the relatively income-poor sections of the population are faring, then it makes sense to get out of the standard identificationcum-aggregation mould of conventional poverty measurement. An alternative approach might be to track a statistic such as the “quintile income”, as has been suggested by Kaushik Basu (2001, 2006), which is just the average income (or consumption expenditure) of the poorest 20% of a population. Subramanian (2011) deals with the case for presenting an annual time-series of the quintile income in the Economic Survey, and of contrasting the actual achievement in this regard with targeted levels of the quintile income based on an appropriate reckoning of an inclusive and egalitarian pattern of overall and quintile growth. This should be a helpful stratagem for confining planning exercises and purely money-metric poverty-related assessments to the straight and narrow.

In any event, the worst possible stratagem for addressing the problem of poverty would be to seek to delude oneself and others by defining it out of existence through the mechanism of postulating unrealistically low and substantively meaningless poverty lines. The issue is of far too much moral gravity to be dealt with in such terms. Or, if that sounds needlessly high-minded, there is still available to one the argument based on enlightened selfinterest. Taking poverty seriously should do little harm to the prospects of reelection to power.

Note

1 G V Desani, All About H Hatterr, Penguin Modern Classics Series, Penguin Books, London, 1972 (first published 1948).

References

Basu, K (2001): “On the Goals of Development” in G M Meier and J E Stiglitz (ed.), Frontiers of Development Economics: The Future in Perspective (New York: Oxford University Press).

– (2006): “Globalisation, Poverty, and Inequality: What Is the Relationship? What Can Be Done?”, World Development, 34(8): 1361-73.

Jayaraj, D and S Subramanian (2009): “A ChakravartyD’Ambrosio View of Multidimensional Deprivation: Some Estimates for India”, Economic Political Weekly, 45(6): 53-65.

Patnaik, U (2007): “Neoliberalism and Rural Poverty in India”, Economic Political Weekly, 48 (32): 3132-50.

Planning Commission (2009): Report of the Expert Group to Review the Methodology for Estimation of Poverty, Government of India, New Delhi.

Ray, R and G Lancaster (2005): “On Setting the Poverty Line Based on Estimated Nutrient Prices: Condition of Socially Disadvantaged Groups during the Reform Period”, Economic Political Weekly, XL(1): 46-56.

Reddy, S G (2004): “A Capability-Based Approach to Estimating Global Poverty” in In Focus: Dollar a Day – How Much Does It Say? International Poverty Centre, United Nations Development Programme: Brasilia, September, pp 6-8.

– (2007): “The Great Indian Poverty Debate”, Development, 50(2): 166-71.

Reddy, S G and T W Pogge (2010): “How Not to Count the Poor” in J Stiglitz, S Anand and P Segal (ed.), Debates in the Measurement of Poverty (Oxford: Oxford University Press), For a longer version, the reader is referred to www.socialanalysis.org.

Sen, Amartya (1983): “Poor, Relatively Speaking”, O xford Economic Papers, 35(2): 153-69.

Subramanian, S (2005): “Unravelling a Conceptual Muddle: India’s Poverty Statistics in the Light of Basic Demand Theory”, Economic Political Weekly, XL(1): 57-66.

– (2010): “Identifying the Income-Poor: Some Controversies in India and Elsewhere”, Courant Research Centre “Poverty, Equity and Growth in Developing Countries” – Discussion Paper Number 46, November, Goettingen.

– (2011): “‘Inclusive Development’ and the Quintile Income Statistic”, Economic Political Weekly, XLVI (4): 69-72.