The title of Amartya Sen’s book is more provocative than it may at first seem. Hunger is associated with poverty in that people who are not poor are not hungry; and famine is simply the extreme of hunger. Yet most people are convinced that the basic cause of a famine is not poverty but a failure of food supply relative to the population. A localized famine is commonly thought of as resulting from a local failure of crops that is not mitigated by importing food, as happened in the Sahel region of Africa in the late 1960s. Countries where hunger is widespread are frequently blamed, moreover, for allowing excessive population growth. The simple Malthusian ratio of food supply to population is further simplified so that the cause of misery is often seen as a matter of overpopulation alone; and we even hear advocates of “lifeboat ethics,” by which countries should be abandoned to their fates.1

Mr. Sen makes a strong case against such views and is highly qualified to do so. He is a scholar of unusually wide interests in an era in which most economists have become highly specialized. As Drummond professor of political economy at Oxford, he holds the oldest chair in the United Kingdom. He has been a student of economic development since his first work that became widely known, his monograph Choice of Techniques, where his cool insistence on proper economic principles argued against simplistic planning and political doctrines in India. He is known for highly technical studies, particularly on the meaning and measurement of social welfare; indeed, he has been elected to be president of the Econometric Society, a worldwide association of the mandarins of mathematical and statistical methods in economics. He is also known among philosophers for his ideas on ethics and the basis of good social action. An understanding of the workings of an economic system combined with rational concern for income inequality, and particularly for poverty, has guided his study of how the economic system contributes to the causes of famine.

In brief, he argues that famine results from the working of the economic system in allocating the ability of people to acquire goods. Famine cannot be explained by a simple relation between food supply and population. Sen illustrates this argument by detailed studies of four famines: the great Bengal famine of 1943-1944, in which perhaps three million people died (mostly by lowered resistance to disease); the famine in several provinces of Ethiopia between 1972 and 1974; the highly publicized drought and famine in the Sahel between 1968 and 1973; and the famine in Bangladesh in 1974 (the same region as the 1943-1944 famine, but under a different political regime).

Most striking are the statistics on the two Bengal famines, the first of which Sen analyzes in the greatest detail. The 1943 crop of rice and other foods was somewhat low, especially in relation to the extraordinarily large harvest of 1942, but it was distinctly higher than the crop of 1941, which was not a famine year. Sen finds that the per capita availability of food supply was 9 percent higher in 1943 than in 1941 and only about 10 percent less than the average of the five preceding years.

Therefore, he concludes, famine cannot be accounted for simply by lack of available food. Relatively small changes in the food supply can be accompanied by dramatic increases in the number of deaths from famine. Why should this occur? Sen points to the simple fact that goods reach people through their ability to “command” that they have goods, as provided by the workings of the socio-economic system. At any given moment, each economic agent has an “entitlement,” a range of different goods that he or she can acquire. This concept can most easily be understood in a private-enterprise economy with little government intervention, though, as Sen emphasizes, the concept is much broader than that.

In a free-enterprise economy every good or service has a price, and each economic agent starts out by owning some goods or services. The rice farmer owns some land, used for producing rice, which can then be sold on the market at the going price or reserved for use by the farmer and his family. The receipts from sales can be spent on other goods—different foods, spices, clothing, and so forth. The agricultural laborer has only his or her labor to sell; the proceeds can be spent on rice or other goods. Similarly the cities contain workers who sell labor for money to buy food, shelter, and clothing, and entrepreneurs who buy goods and labor, produce other goods, sell them, and have the proceeds for personal consumption and investment in business expansion.

People will starve, then, when their entitlement is not sufficient to buy the food necessary to keep them alive. The food available to them, in short, is a question of income distribution and, more fundamentally, of their ability to provide services that others in the economy are willing to pay for.


This, of course, does not mean that the supply of food is irrelevant. A decrease in the supply of food will usually increase its price, as people compete for the scarcer quantity. This will in turn decrease their ability to buy food by using their entitlement and, if they start close enough to the margin of hunger, my drive them to the point of starvation. Further, the entitlement approach, simple as it is, enables the analyst to say something about the distribution of the burden of starvation. Farm owners and, to a lesser extent, sharecroppers, should be less affected than others because the reduction in the amount they sell is at least partly offset by the higher prices. If the reduction in supply is caused by some factor, like flood, that reduces the amount to be harvested, farm laborers are thereby more likely to be seriously affected.

Thus studies of different entitlements and how they are affected by variations in food supply, alone or in conjunction with other shifts in the economy, are capable of giving a much greater insight into the causes of famine than a simple measure of the amount of food available. But the real point of Sen’s analysis is to show that relatively small changes in food supply may nevertheless be accompanied by famine. Indeed the economic theory he presents would allow for the possibility that, with no changes in the food supply at all, famine could be caused by other economic factors.

In his analysis of the Bengal famine of 1943 and 1944, Sen points to just such factors. Predominant among them, in his view, is the effect of the war against the Japanese in increasing demand. (He is careful to add that the evidence is not strong enough to establish one sequence of causes as against another.) Government expenditures, especially on construction, rose sharply. This clearly increased the entitlements of the newly employed urban workers; and since the total supplies were unchanged, the entitlements of the rural groups had to fall. Indeed, rice prices rose sharply even before any evidence that crops had failed.

A parallel situation occurred in the United States at the same time, though at a far higher level of well-being in general and food consumption in particular. As the still very large pool of unemployed workers in 1940 was absorbed into industrial work by the rising demand for war goods, the demand for meat rose sharply. The meat supply did not fall; it actually rose, but the rising demand still drove the prices up. From the viewpoint of the newly employed, there was indeed a shortage. When price controls were imposed, the shortage became a visible social problem; there were bare shelves at the butcher’s, black markets, and rationing.

For the worker in the US, meat was, in the economist’s jargon, “income elastic.” That is, for a newly employed worker at the bottom end of the scale, an increase in income led to a more than proportionate increase in expenditure on meat. But for food generally and even more for the low-priced foods, the grains, even the poorest member of the regular labor force in the United States now or in 1940 has a low income elasticity: the proportion of income spent on food rises only very slowly as income rises.

In Bengal, on the other hand, when conditions are sufficiently desperate, a person will, if necessary, spend the whole of his or her income on food. Even many of those who do not live in fear of starvation will be hungry enough so that a large fraction of any increase in income will be spent on food. Indeed, in the poorer less-developed countries, roughly two-thirds of total income goes for expenditures on food.2 Hence, as groups like urban workers newly employed in war work gain increased entitlements, their willingness to buy additional food is high. Prices rise, and the entitlements of other groups will necessarily fall unless there is a corresponding rise in food supply. If these other groups are already on the margin of sustainable life, then indeed a famine may be created without any decline in food supply.

Sen offers additional explanations. Perhaps the most controversial is that of hoarding, either for oneself or for speculation. When situations of scarcity arise, hoarding is always blamed. The evidence for the degree and the effects of hoarding is usually difficult to come by, but, apart from this empirical question, there is a significant theoretical problem concerning its implications. If the famine is prolonged, then hoarding at the beginning means greater stores will be available later on. In fact, if the hoarder was correct in his expectations, that is, if the farmer does need to consume his own grain at a later date or if the speculator makes money by selling at a higher price, then hoarding will have improved the availability of food later on, at, to be sure, the cost of making things worse initially.


A still better result might be achieved if the government took over some part of the hoarded stores and distributed them to the needy; but it should also withhold some from immediate use. Such a policy would amount to changing the entitlements. Indeed, when a famine occurs it would certainly require an inhuman preference for established property rights over human needs not to redistribute purchasing power and other forms of entitlements. But that does not alter the basic point about hoarding: if we consider the entire period of famine, hoarding will only worsen the famine if it turns out to be excessive, that is, if some of the hoards are retained beyond the end of the famine.

In view of the absence of detailed data on famines, it is not surprising that Sen’s analysis does not explain everything. His point of view would suggest that if the total food supply changed only mildly the terrible losses of some people should be reflected in greater food consumption by others. Yet in Table 6.7 (p. 71), concerning the Bengal famine of 1943-1944, the proportionate increase in destitution is much the same for noncultivating landowners (admittedly a poor group in the best of times) as for peasant cultivators or for those who worked partly for themselves and partly for others. If the rice crop failure is treated as minor, the landowners should have gained; any reduction in quantity would have been more than offset by the rise in the price of rice compared with other prices. However, the table covers only some rural regions; it may well be that greater supplies of grains were to be found in the urban regions. In Calcutta itself, only migrants from the rural areas seem to have suffered from famine.

Many other insights are to be gained from Sen’s emphasis on entitlement, especially concerning the subtlety of the interaction between pastoralists and farmers, a subject he discusses in his chapters on the famines in the Sahel and in Ethiopia.

What policies does an entitlement approach suggest? Sen is surprisingly chary of answers to this question. A brief chapter mostly tends to emphasize the complexity of the analysis introduced by the entitlement approach. He correctly points out that the market mechanism cannot be relied upon, for it is precisely the failure of market power that chooses the victims of famine. He does speak of insurance arrangements, that is, ways by which entitlement of a particular group, for instance rural laborers, could be automatically increased under famine conditions. However, he seems to emphasize the difficulties rather than the advantages of such a device.

For example, he points out that general famine conditions affect only selected persons, and an insurance program based on aggregate food supply, for example, would therefore miss its intended beneficiaries. If, on the other hand, insurances were highly individualized, this would, in my view, affect the incentives people have to work and to plan for the future. Still, it does not seem to me so difficult to propose that, at certain levels, a program for insurance (or, what is equivalent, relief) should go into effect, based on the ability of some designated fraction of the population to buy food; not does it seem excessively difficult to direct aid to the groups most likely to be affected. This will not feed every starving person, since some will receive help who do not need it and some who need help will not receive it; but no system will avoid problems of this sort.

Although famine is a fairly unusual event, and improvements in transportation and communication have made it less and less likely, the need to respond to specific famines is no less urgent. The three famines of the mid-1970s discussed by Sen, though apparently much less severe than the 1943-1944 famine in Bengal, are horrifying enough. Indeed, that famine is becoming less frequent and intense should make it all the easier to provide prompt aid. On a world scale, hunger and malnutrition are far greater problems. Of course, they are intimately related. As is obvious from Sen’s analyses, famine is the result of pushing already hungry populations over the relatively low threshold that separates them from genuine starvation.

The extent of world hunger as expressed in protein-calorie malnutrition has been much disputed. Estimates range from 100 million to 2.5 billion people affected.3 The recent Presidential Commission on World Hunger suggests a range of between 500 and 800 million. In any case, it is certainly very large. Since the various aid programs that nations and international organizations are now willing to support are sure to be inadequate to deal with even the lowest estimates, the exact figure is really of little practical consequence.

Among experts there is an increasing tendency to regard hunger as primarily a problem of income distribution, of purchasing power. In this sense, Sen’s book can be seen as part of a larger effort to make a fresh analysis of hunger. His entitlement approach adds to the emphasis on income distribution by stressing its causes, rather than merely taking the distribution as a given. But since the analysis of causation is necessarily complex and uncertain, the two approaches are similar. The pioneering work in connecting income distribution with hunger is that of Shlomo Reutlinger and Marcelo Selowsky for the International Bank for Reconstruction and Development (the World Bank).4 They estimated (very imperfectly) a relation between calorie consumption and income; then, with the aid of the fragmentary income distribution data available, they estimated the calorie consumption for different income classes. On this basis, they could estimate the number of people whose supply of calories fell below some critical level. Because much of the data is unavailable, the possibilities for error are enormous; but this approach gets at the underlying determinants of hunger.

What has become clear is that hunger and ultimately famine are basically questions of the distribution of income and the entitlements to food. That does not of course mean that food supply is irrelevant, but that it is far from determining who will go hungry. At higher levels of consumption, hunger fades away as a basic problem but is replaced by other needs, including medicine, shelter, and the like. In short, averages are insufficient guides to economic performance, though they cannot be ignored.

The appropriate measures to change income and food distribution, especially those that relieve famine, hunger, and poverty, partly depend on the information available; but they also require conceptual definitions, and these are intimately tied to value judgments and measures of welfare. As a counterpoint to his analysis of famines, Sen discusses some of the problems of measuring poverty. He shows the inadequacy of the usual United States measure, the number of people whose incomes are below the so-called “poverty line.” He would equally reject a measure of hunger based on the number of people whose food consumption falls below a fixed level defined as “adequate.” The distribution of income below the poverty line or of calorie consumption below the critical level may still allow for sharp differences among poor people. In the case of calorie consumption, the difference may be between widespread hunger and appalling famine.

Sen’s book, together with other recent work on the world problem of hunger, should alert us even more strongly to the need for studying the distribution of income and of improving the relative situation of the very poor in order to curb the worst consequences of the economic system. The political implications of this shift in emphasis may be serious indeed, pointing up the responsibilities of the governments of the less developed countries. Poor as they may be, these countries generally have more control over their domestic income distribution than the advanced nations they deal with.