The variety of information on which social welfare analysis can draw can be well illustrated by the study of poverty. Poverty is typically seen in terms of the lowness of incomes, and it has been traditionally measured simply by counting the number of people below the poverty-line income; this is sometimes called the head-count measure. A scrutiny of this approach yields two different types of questions. First, is poverty adequately seen as low income? Second, even if poverty is seen as low income, is the aggregate poverty of a society best characterised by the index of the head-count measure?
I take up these questions in turn. Do we get enough of a diagnosis of individual poverty by comparing the individual’s income with a socially given poverty-line income? What about the person with an income well above the poverty line, who suffers from an expensive illness (requiring, say, kidney dialysis)? Is deprivation not ultimately a lack of opportunity to lead a minimally acceptable life, which can be influenced by a number of considerations, including of course personal income, but also physical and environmental characteristics, and other variables (such as the availability and costs of medical and other facilities)? The motivation behind such an exercise relates closely to seeing poverty as a serious deprivation of certain basic capabilities. This alternative approach leads to a rather different diagnosis of poverty from the ones that a purely income-based analysis can yield.
This is not to deny that lowness of income can be very important in many contexts, since the opportunities a person enjoys in a market economy can be severely constrained by her level of real income. However, various contingencies can lead to variations in the “conversion” of income into the capability to live a minimally acceptable life, and if that is what we are concerned with, there may be good reason to look beyond income poverty. There are at least four different sources of variation: (1) personal heterogeneities (for example, proneness to illness), (2) environmental diversities (for example, living in a storm-prone or flood-prone area), (3) variations in social climate (for example, the prevalence of crime or epidemiological vectors), and (4) differences in relative deprivation connected with customary patterns of consumption in particular societies (for example, being relatively impoverished in a rich society, which can lead to deprivation of the absolute capability to take part in the life of the community).
There is, thus, an important need to go beyond income information in poverty analysis, in particular to see poverty as capability deprivation. However (as was discussed earlier), the choice of the informational base for poverty analysis cannot really be dissociated from pragmatic considerations, particularly informational availability. It is unlikely that the perspective of poverty as income deprivation can be dispensed with in the empirical literature on poverty, even when the limitations of that perspective are entirely clear. Indeed, in many contexts the rough-and-ready way of using income information may provide the most immediate approach to the study of severe deprivation.
For example, the causation of famines is often best seen in terms of a radical decline in the real incomes of a section of the population, leading to starvation and death. The dynamics of income earning and of purchasing power may indeed be the most important component of a famine investigation. This approach, in which the study of causal influences on the determination of the respective incomes of different groups plays a central part, contrasts with an exclusive focus on agricultural production and food supply, which is often found in the literature on this subject.
The shift in informational focus from food supply to entitlements (involving incomes as well as supply, and the resulting relative prices) can make a radical difference, since famines can occur even without any major decline — possibly without any decline at all — of food production or supply. If, for example, the incomes of rural wage labourers, or of service providers, or of craftsmen collapse through unemployment, or through a fall of real wages, or through a decline in the demand for the relevant services or craft products, the affected groups may have to starve, even if the overall food supply in the economy is undiminished. Starvation occurs when some people cannot establish entitlement over an adequate amount of food, through purchase or through food production, and the overall supply of food is only one influence among many in the determination of the entitlements of the respective groups of people in the economy. Thus, an income-sensitive entitlement approach can provide a better explanation of famines than can be obtained through an exclusively production-oriented view. It can also yield a more effective approach to the remedying of starvation and hunger.
The nature of the problem tends to identify the particular “space” on which the analysis has to concentrate. It remains true that in explaining the exact patterns of famine deaths and sufferings, we can get additional understanding by supplementing the income-based analysis with information on the conversion of incomes into nourishment, which will depend on various other influences such as metabolic rates, proneness to illness, body size, etc. These issues are undoubtedly important for investigating the incidence of nutritional failures, morbidities, and mortalities. However, in a general analysis of the occurrence and causation of famines, affecting large groups, these additional matters may be of secondary importance. While I shall not enter further into the famine literature here, I would like to emphasise that the informational demands of famine analysis give an important place to income deprivation, which have more immediacy and ready usability than the more subtle — and ultimately more informed — distinctions based on capability comparisons.
I turn now to the second question. The most common and most traditional measure of poverty had tended to concentrate on head counting. But it must also make a difference as to how far below the poverty line the poor individually are, and furthermore, how the deprivation is shared and distributed among the poor. The social data on the respective deprivations of the individuals who constitute the poor in a society need to be aggregated together to arrive at informative and usable measures of aggregate poverty. This is a social choice problem, and axioms can indeed be proposed that attempt to capture our distributional concerns in this constructive exercise.
Several distribution-sensitive poverty measures have been derived axiomatically in the recent social choice literature, and various alternative proposals have been analysed. While I shall, here, not go into a comparative assessment of these measures (nor into axiomatic requirements that can be used to discriminate between them), elsewhere I have tried to address this issue, jointly with James Foster. However, I would like to emphasise the fact that we face here an embarrassment of riches (the opposite of an impasse or an impossibility), once the informational basis of social judgments has been appropriately broadened. To axiomatise exactly a particular poverty measure, we shall have to indulge in the “brinkmanship” of which I spoke earlier, by adding other axiomatic demands until we are just short of an impossibility, with only one surviving poverty measure.
Excerpts from Amartya Sen’s Nobel Prize lecture at Trinity College, Cambridge, 8 December 1998. Read the full lecture here.
This is part of ThePrint’s Great Speeches series. It features speeches and debates that shaped modern India.