Some Issues in the Measurement of Poverty in India
Professor Tendulkar has been a major contributor to the sizeable literature on poverty in India. And, beginning with our 1983 EPW paper on “Poverty Reduction in Sixth Plan: Population Factor and Rural-Urban Equity,” it has been my privilege to have co-authored with Suresh numerous articles, papers and reports on this subject over a period of a quarter-of-a century and more. Unsurprisingly, I will be drawing on some of our joint papers in the paragraphs that follow. More specifically, I will be drawing on the 1993 Sundaram-Tendulkar paper on “Poverty in Asia Pacific: Conceptual Issues and National Approaches to Measurement” (Economic Bulletin for Asia and the Pacific, UN ESCAP)—hereafter referred to simply as the ESCAP-paper. I will also be drawing on my unpublished paper offering an appreciation of Suresh’s work on Poverty, Employment and Public Policy presented at the 48th TIES Conference (Sundaram 2012).
Recently there has been a lot of debate and criticism—not always fair—of the poverty lines for 2004-05 suggested by the Tendulkar Committee. Much of the controversy has centred around the inadequacy (from the perspective of meeting the basic needs) of Rs 32/per-capita per-day as the price-adjusted value of the 2004-05 poverty line proposed by the Tendulkar Committee.
A few points on this critique are in order:
Let us begin by noting that when the Report came out in November 2009, there was very little concern expressed about the 2004-05 urban poverty-line—at a little over Rs 19 per-capita per-day—about that amount being too low. This fact, incidentally, also vindicates the Committee’s position of this figure being a reasonable starting point for deriving the poverty lines for the urban areas of individual States and for the corresponding rural areas, and, from that for all-India rural by appropriate price indices. More importantly from the point of the criticism of Rs 32/per-capita per-day, the change in the public perception about it would appear to be about the procedures adopted and the data-base used to adjust for inflation since
2004-05 rather than about the poverty-line for 2004-05 as proposed by the Committee.
It is of course entirely possible to re-draw the absolute poverty line to reflect current social sensibilities on this issue. Indeed, the new Expert Group under the Chairmanship of Dr. C. Rangarajan has been set-up by the Planning Commission (GOI 2012) with precisely such an end in view. Two further comments:
First, for assessing the change in poverty between two points of time—say 2004-05 and 2009-10—there is no alternative but to adjust the base year poverty line for the rise in prices over this period of the goods and services comprising the base year poverty line basket (PLB).
The second point to be made in the context of deriving anew a fresh set of poverty lines is something that Suresh and I stated nearly two decades ago in our ESCAP paper. I quote:
The absolute poverty line is not the aggregation of expenditure needed for purchasing the commodities and services required for fulfilling all the basic needs. This follows from the problems of objective norm specification as well as those of aggregation across interdependent needs (discussed earlier in the paper) and from the fact that households are not uniform in their composition, tastes and location across climatic conditions. There is therefore an inherent and irreducible element of arbitrariness in the specification of the absolute poverty line and (there is) no alternative but to treat it as broadly representing a “low enough yet reasonable minimum living standard” (Sundaram and Tendulkar 1993, p. 45).
I would like to underline the idea of “low enough” in specifying, de novo, the absolute poverty line. For, it is always possible to specify a poverty line such that three-fourths or more of the all-India population would be below it. Such a construct would, in my view, rob the concept of poverty of any operational cutting edge in policy formulation and implementation. For starters, such an all-India poverty line would imply that in the relatively-poorer states/regions, almost the entire population is below the poverty line!
Admittedly, making this judgement about what constitutes “a low-enough yet reasonable minimum living standard” is easier said than done. This task is not made any easier by the analytical arguments—going back to Sukhatme (1982), Srinivasan (1992) and Sundaram and Tendulkar (1993)—and the empirical evidence (see Deaton and Dreze 2009) against anchoring the poverty line to calorie norms—even if the latter were to be reworked by reference to the current structure of the population.
Having rejected the anchoring of the poverty line to calorie-norms, it is indeed a mystery why the Tendulkar committee goes into what it calls “external validity checks... with regard to nutritional, educational and health outcomes” (GOI 2009,
As regards nutrition outcomes, except in the case of “weight-for-height” measure, other anthropometric measures of undernourishment reflect an “outcome of a situation where, over a sustained period, the nutritional intake of the individual under reference has remained inadequate. As against this, the focus of measurement by reference to the absolute poverty line is on purchasing power and its disposition in the current period ... (and) one can at best capture the deficiency in intake of nutrients in the current period. There is however, no necessary presumption that those who are identified (on the stated assumptions) as having been in that state over a sustained period in the past as well.” (Sundaram and Tendulkar 1993, p. 47, emphasis added). See also Suryanarayana (2011).
As regards educational outcomes, in our 1993 ESCAP-paper referred to earlier, Suresh and I had noted that: “the observed outcomes, in terms of say the proportion of illiterates in the adult population, would primarily reflect the cumulative impact of past provisioning of public services and private decisions to avail of these services and to utilize a part of private purchasing power for the same and for buying essential complementary goods and services. The role of resources (public and private) currently committed to this end would be important primarily in shaping future outcomes” (Ibid., p. 47, emphasis added).
In respect of health outcomes, the problems of using indicators of these out- comes—mortality/morbidity—are more basic. Mortality indicators are not even defined at the level of an individual person/household. As for morbidity indicators, which also are defined for population groups, they could be seriously misleading and out of alignment with mortality indicators for the same population. This can be readily seen from a comparison of NSS Survey-based morbidity indicators for, say, Kerala and Jharkhand.
This feature of the mortality/morbidity indicators that they are appropriately defined only for population groups would also automatically rule them out as a component of any measure of poverty relevant for the design and implementation of polices targeted at individuals or individual households.
If health status outcomes cannot be a meaningful part of a measure of poverty focused on individual persons/households can we at least define a normative level of private expenditure on health in defining the poverty line? In answering this question a few features of the household expenditure on healthcare need to be kept in view.
First and foremost, household expenditures on healthcare—focusing here on curative healthcare-are in the nature of contingent liabilities in the sense that they arise if and only if and when someone in the household falls ill or suffers an injury. But, when such a contingency does arise, the expenditures can be quite substantial and estimates of average expenditure on health carry little meaning to the affected
Secondly, the prevalence and incidence of several diseases -which underlie the probability of a member of a household falling ill—are crucially dependent on the public provision of a range of goods and services—drinking water, sewage and sanitation facilities, Vaccination services, to name just a few—and the access of the households to them. This factor, in turn, raises questions about the very idea of a normative level of private expenditure on health.
Finally, inferring the “required” health expenditures from household consumer expenditure surveys runs into a different sort of a problem. What gets recorded—in fact what can get recorded—are the “out-of-pocket” expenses incurred by the households. Specifically, it excludes a sizeable component of “costs” of healthcare that are met by publicly provided services at heavily subsidized “user charges”—if not provided free.
From another perspective, deprivation of households in respect of access to a range of public goods and services such as drinking water, sewage and sanitation facilities, vaccination services is indeed measurable. Some of these deprivations are sometimes listed as components of what is referred to as multi-dimensional measure of poverty. Some comments/reservations on this are in order.
First, these and other similar goods and services are public goods and, therefore, their provision to the population must be on a universal basis. Specifically, this public provisioning must not be tied to the “poverty-status” of the household or individual.
A second set of problems would arise if deprivation in respect of services that are in the nature of inputs shaping outcomes such as incidence of malnutrition are aggregated with such outcome indicators.
There is, firstly, the problem of these indicators being not independent of one another. Further, even if the indicators that are sought to be combined are indeed independent of one another and hence aggregable in principle, there is the problem of analytically appropriate rules of aggregation. Also, as I had argued over a decade ago (Sundaram 2003) such an aggregation results in loss of information from the perspective of design and implementation of policies targeting individual components.
Finally, given the public goods character of these services, an efficient design of policy would focus on locations of inadequate supply rather than on the number and identity of “poor” households with poverty being defined by reference to another metric.
Such problems referred to above-especially in respect of aggregation of indicators irrespective of whether or not they are independent of one another-also apply to other indicators such as access to land.
To cut a long story short, there are no clear and easy answers to the issue of defining a poverty cut-off in terms of generalised purchasing power by reference to which one could track changes (variations) in levels of living of the poor over time (across space).
As professor Tendulkar and I had recognised nearly two decades age, this cut-off is essentially, and inherently, arbitrary, with “low-enough yet reasonable minimum living standards” as the only guide. There are however some policy initiatives that in my view can significantly mitigate the consequences of any misjudgements one may make in drawing this arbitrary cut-off to demarcate the poor in the society. These would include a public policy on universal provisioning of public goods and services; support for a universal health insurance to cover the contingent liability of large costs of health care; and, last but not the least, a policy of ensuring access to essential food at reasonable prices to at least the vulnerable sections of the society if not to all. This last is a key component of Scitovsky’s notion of equity: a concept that deeply influenced Suresh’s writings on inequality and poverty in India.
-  The Expert Group to Review the Methodology for Estimation of Poverty set up by the PlanningCommission had Professor Tendulkar as the Chairman and Professor R. Radhakrishna & Dr.Suranjan Sengupta as members who had signed the report. Regrettably, one has seen little in theform of a response to the criticisms of the report from the members of the committee—especiallysince Suresh was no more there to defend the position of the committee.
-  This Expert Group, with Dr. C. Rangarajan as chairman, has as its members Dr. S. MahendraDev, Director IGIGDR, Dr. Mahesh Vyas, MD & CEO of CMIE, Dr. K.L. Dutta, formerlyadviser, Perspective Planning Division, Planning Commission, and, myself. Needless to say, theviews presented here are those made in my personal capacity and should in no way be seen asreflecting the views of other Members of the Expert Group.
-  The case of private expenditure on health insurance is, strictly speaking, a part of financingstrategy of the household to meet such contingent liabilities as and when they arise.
-  The improvements in mortality/morbidity indicators that have been realized over the past twodecades or so cannot be fully understood without factoring-in the increase in public expenditure inthis sector. Also important has been the growth in the expenditures in the provision of a range ofhealthcare services by non-profit institutions serving households—NPISH in National Accountsterminology. At least a part of the explanation for the growing divergence between the NationalAccounts estimate of Private Final Consumption Expenditure and the NSS estimates of householdconsumer expenditure over the same period is tied to this last-mentioned factor.