Choice, Welfare and Measurement
Resources, Values and Development
Amartya Sen occupies a unique position among modern economists. He is an outstanding economic theorist, a world authority on social choice and welfare economics. He is a leading figure in development economics, carrying out path-breaking work on appraising the effectiveness of investment in poor countries and, more recently, on famine. At the same time, he takes a broad view of the subject and has done much to widen the perspective of economists. He has made major contributions to moral philosophy, being as much at home writing for the Journal of Philosophy as for the Economic Journal. He enjoys controversy, in TV debate as well as the academic seminar, and is an incisive and entertaining lecturer.
The economist seeking to take economics beyond its conventional boundaries might well find himself out of favor with his own colleagues. Not so Sen, who is a prophet with honor in his own discipline. He is Drummond Professor of Political Economy at Oxford, and was recently appointed a professor of economics and philosophy at Harvard. He has been president of the Econometric Society and of the Development Studies Association. He has recently been elected president of the International Economic Association, the previous holder of this office being Kenneth Arrow, who, together with Maurice Dobb and John Rawls, has had a major influence on Sen’s work.
On what is this formidable reputation based? The two volumes of Sen’s collected papers under review provide an opportunity to answer this question. Many people associate Sen with social choice theory and welfare economics, and these are the main topics covered in the first volume, Choice, Welfare and Measurement. Social choice theory is concerned with the relation between social judgments and the preferences of individuals. In societies such as the Western democracies in which individual preferences are the basis for social judgments, how can the different preferences be combined into a collective judgment? Kenneth Arrow’s Social Choice and Individual Values, published in 1951, contained a powerful mathematical analysis of this problem. His “impossibility theorem” showed that if one requires the procedure of reaching a collective judgment to satisfy four apparently quite reasonable conditions (such as that there are no restrictions on the way in which people rank different alternatives), then there can be no consistent social judgment.
That problems of this kind arise with majority voting had long been known, and most people will have come across the “paradox of voting,” by which in a majority vote for, say, the choice of the nation’s capital, Washington would be beaten by Boston, Boston would be beaten by Chicago, but, in a vote between Washington and Chicago, Washington would win. Arrow’s theorem shows that problems arise not just with majority voting but also with any conceivable method of aggregating individual preferences that satisfies the four conditions.
Sen has said that as an undergraduate at Cambridge University he was “besotted” with Arrow’s book. Recalling my own excitement on first reading Arrow’s work, at Cambridge some ten years after Sen, I can understand how it came like a shaft of sunlight into a climate that was inhospitable both literally (Sen has written with feeling of his first winter in an “astonishingly damp country”) and intellectually. Cambridge economics at that time was sympathetic neither to welfare economics nor to the use of mathematics in social science. However, Maurice Dobb, Sen’s supervisor, was more open-minded and encouraged Sen in his explorations.
In examining how Sen has taken social choice theory further, we should begin with the questions that Sen has posed about the nature of choice at the level of the individual rather than society. The opening essays are concerned with choice and preference, including his inaugural lecture at the London School of Economics in 1973 in which he attacked the “revealed preference approach” to consumer behavior that has dominated the economic theory of choice since the 1940s. To quote from Paul Samuelson’s celebrated paper on the subject, the basic idea of this approach is that “the individual guinea pig, by his market behavior, reveals his preference pattern.” For the consumer, the process runs from preferences to consumption decisions; the economist puts the process in reverse: we observe what a person buys in the supermarket and deduce preferences from these choices.
Sen argues that this revealed preference approach takes an overly narrow view, underestimating the influences on choice other than the person’s own preferences. He stresses the ways in which group norms and the culture of a class or community can affect choices. For example, in choosing an insurance plan or making wage demands, people working in a particular job may be influenced as much by the attitudes of their fellow workers as by their personal preferences. For Sen, “economic man,” concerned only with his own desires, is “close to being a social moron.”
Sen warns against equating human beings with guinea pigs, stressing how ethical considerations can figure in human motives and are relevant even to the purely descriptive goal of explaining economic decisions. Notions of “fairness,” for instance, may enter into the explanation of the rate of output of a group of workers. A person’s statement that he does not buy South African goods may well be relevant in determining what can be deduced from the contents of his supermarket basket.
In attacking the model of economic man as a “rational fool,” Sen is not alone. Others unhappy about the foundations of microeconomic behavior include such well-known economists as Harvey Leibenstein, Janos Kornai, and Tibor Scitovsky. Sen’s particular contribution is his dissection of the different meanings that can be attached to the notion that a person “prefers” a particular thing or action that we can call A. To give some examples:
—the person gets more satisfaction from A
—the person prefers that A is chosen
—the person believes that it would be right to choose A
—the person chooses A
None of these entails any of the others. Professor X may get more satisfaction from choosing South African grapes on the grounds that he likes their taste better, but believe that it would be wrong to buy them on political grounds. He may therefore not buy them in the supermarket. The same person may, however, prefer that his faculty club choose South African grapes (particularly if he can complain about the choice while eating the grapes).
This may appear quite obvious, yet individual choice continues to be treated in economics texts as presenting no such problems. Hal Varian’s Microeconomic Analysis, possibly the most widely used graduate text, is based on the assumption that the first statement listed above implies the fourth, and that the second and third are not relevant. For the economics student, beset with difficulties on every side, this simple portrayal may come as a welcome relief, but it impoverishes the subject. What Sen wants us to do is to take more account of the complexity of human behavior.
Individual preferences enter economics both as determinants of behavior and as the basis for judgments about the welfare of societies or groups. Here we come back to the issue of social choice, but now with the insight that there may be more than one interpretation of the way preferences are aggregated. Sen notes in his masterly 1975 survey the difficulties that arise in the theory of social choice from “a desire to fit essentially different classes of group aggregation problems into one uniform framework.” He distinguishes between the aggregation of individual interests (for example, in sharing the burden of taxation) and that of individual judgments (for example, whether apartheid should be ended). He argues that Arrow’s approach is inappropriate for combining different interests. Majority voting is particularly unattractive as a procedure for arriving at decisions: “Making the poor much poorer and passing on half the plunder [the remainder being lost in the process] to the rest would seem to be both wasteful and inequitous, but [the method of majority decisions] strongly supports such a change.”
What is needed according to Sen is to widen the range of information that we bring to bear in making social choices. This additional information may be concerned with individual welfare, conveying, for instance, the value a person attaches to his situation before and after a proposed tax reform. The use of such information poses a number of problems, particularly when it comes to making comparisons between the welfares of different people. One of Sen’s major contributions has been to clarify the different types of information that can be used in making social choices and their relationship with different principles of justice. If, for example, we can compare the value of an extra dollar to one person with that of an extra dollar to anyone else, without necessarily being able to say that one person is overall better or worse off than another for having the extra dollar, then it is legitimate to apply the utilitarian principle, associated with Jeremy Bentham. This means that we calculate the total of everyone’s gains and losses from, say, a tax reform, much as a government calculates the costs and benefits of a spending project. Judging different choices in this way provides a consistent method of social decision.
On the other hand, the more recent “difference principle” of the Harvard philosopher John Rawls requires that we be able to say that one person is better or worse off overall than another. In particular, we can identify the least advantaged in society, who are the group to whom Rawls’s principle would give priority. This too is a permissible criterion for social welfare. We can judge different choices by their effect on the welfare of the least advantaged, asking simply whether they gain or lose from a tax reform and not seeking to compare the size of this gain or loss with that of other groups. As it was put by Robert Lampman, one of the economists instrumental in founding the Institute for Research on Poverty at Wisconsin, the test of policy according to this principle is “what does it do for the poor?”
These approaches are described by Sen as “welfarist,” in that they assume we can use only information about personal welfares. Such a limitation is in the tradition of welfare economics, but Sen has emphasized the need to take a broader view, considering concepts of justice that depart from this assumption. Indeed, Rawls’s difference principle is concerned with the position of the least advantaged defined not by personal welfare, but by “primary goods,” or “things that every rational man is presumed to want.” This takes us outside the traditional scope of welfare economics. The rejection of the welfarist approach is similarly to be found in Marxist theories of exploitation, relating social judgments to the historical information that capital represents the product of past labor. Robert Nozick’s theory of justice is quite different but equally appeals to historical information. For him, it is not the distribution of income that matters but the process by which it is brought about, people being “entitled” to resources that were justly acquired or that were transferred to them according to a just process, even if this means they will be immensely rich and that their riches may be of no benefit to the poor.
In linking the problem of social choice to the writings of moral philosophers and the longstanding concerns of ethical theory, Sen’s contribution has been of a high order. I have referred to Rawls as a major influence on his work, and Rawls himself acknowledges, in his A Theory of Justice, that Sen has influenced his own thinking. What is remarkable is that Sen, in his forays into philosophy, has sharpened the questions being asked, without seeking to impose the economist’s mode of analysis on other disciplines.
There are, however, respects in which he could have had a more forceful effect as an economist among the philosophers. Take for example the standard economist’s model of competitive general equilibrium by which supply equals demand in all the markets of the economy—including markets for labor, for consumer and capital goods, and for services, etc. While seriously deficient as a vehicle for discussing the causes of unemployment or the behavior of international corporations, it is a considerably richer model than the cake sharing and other examples typically used in discussions of the theory of justice. The model allows for the possibility that the size of the cake will change and this affects the way in which we think about different distributive principles. This may be seen from the fact that if the size of the cake is fixed, and people are identical, then a Benthamite utilitarian, a Rawlsian, and an egalitarian would all divide it equally. But where the size of the cake depends, say, on people’s decisions about work, application of the three principles of economic justice leads to different degrees of redistribution. If the redistribution takes the form of a guaranteed minimum income, financed by a tax on earnings, then the utilitarian will generally choose a lower guaranteed minimum (with a lower tax rate) than a Rawlsian, since the Benthamite is concerned with the welfare of the rich as well as that of the poor. The Rawlsian chooses the tax rate to maximize the advantage to the poor, but this may still leave a gap between rich and poor that the egalitarian regards as unacceptable and wishes to narrow still further. So that all three principles would give different answers.
The practical economic application of theories of justice is illustrated by the measurement of income inequality and of poverty. In the case of poverty, Sen has opened up an important new line of inquiry and stimulated a large theoretical literature. Why, he asked, do we simply measure the extent of poverty by the number below the poverty line? The US Bureau of the Census publishes figures showing that 15 percent of the American population lives below the official poverty standard, but this measure takes no account of how far different people fall below.
Sen goes on to develop a new approach to the measurement of poverty, in which a person’s distance from the poverty line is weighted according to its severity. If you are $1,000 below the poverty line, this gets more weight than a shortfall of $100. What should the weights be? Sen takes the view that the key factor is your position in the income distribution among the poor, or “rank order.” Imagine a parade in which everyone walks past in order of their income; what the rank order measures is how far you are from the person just at the poverty line. However, the rationale for adopting such a rank order in measuring poverty is not clear. Why should the severity of a $1,000 shortfall from the poverty line depend solely on how many people are ahead of you? To answer this, Sen refers to “relative deprivation”: “the lower a person is in the welfare scale, the greater his sense of poverty.” But this line of justification is not really developed. Perhaps because he leaves this question open, Sen’s approach has so far had only limited effect on empirical research. It is striking, for example, that there is not a single reference to Sen’s poverty measure in the recent Fighting Poverty, a 400-page review of twenty years of the US War on Poverty published by the Institute for Research on Poverty. The gulf between the theoretical literature on poverty measurement and empirical practice has not really been closed.
Closing the gap between theory and practice is, however, an activity at which Sen has excelled in another sphere—that of development economics—as is well brought out in the second volume, Resources, Values and Development. I can recall that Sen’s thesis, Choice of Techniques, the introduction to which is reprinted here, gave me as much pleasure as an undergraduate as Arrow’s book. It brought clarity to a confused debate on whether developing countries should use labor-intensive methods of production. Where labor is abundant and capital scarce it is tempting for development planners to adopt highly labor-intensive methods, but this ignores the implications for the future availability of capital, in that the savings rate may be lower if more is paid out in wages. To take the example of cotton weaving studied by Sen, cottage hand looms may have the highest output per unit of capital, but factory power looms may generate a larger surplus of money that can be invested to provide more capital in the future. This point, which presumes that savings are insufficient and that no alternative policies to encourage savings are available, had been made by Maurice Dobb, to whom Sen owed his interest in the subject. Sen’s contribution is to provide an analytical scheme within which these issues can be discussed, one that combines elegance of theory with a concern for practical application.
The solution given by Sen to the problem of choosing between labor-intensive and capital-intensive techniques can be described by his concept of a “shadow price” for labor. A government deciding how much labor to employ should base its calculation not on the actual cost of taking on an additional worker but on the “shadow” cost, which takes account of the full economic implications of employing an additional person. Such implications include the reduction in surplus for investment as well as the loss of the output that the worker would have produced elsewhere (which could be zero). The notion of shadow prices has been known since the debates in the 1930s and 1940s on the economics of planning, but Sen refined it as a practical tool of investment appraisal, to be used to determine the choice of projects undertaken. Together with Partha Dasgupta and Stephen Marglin, he wrote the UNIDO Guidelines for Project Evaluation, which, along with the OECD manual by Ian Little and James Mirrlees, has dominated the literature and practice of investment appraisal.
One of the features of Sen’s analysis in this field is his concept of “control areas.” He criticizes the “concentration in the planning models on a mythical hero called The Planner,” stressing that
the planner…is part of a political machinery and is constrained by a complex structure within which he has to operate. Successful planning requires an understanding of the constraints that in fact hold and clarity about precise areas on which the planners in question can exercise effective control.
The consequences of investing in, say, a chemical plant depend on what else the planner can change. It might be preferable for the chemicals to be imported, but this may be precluded by a political decision in favor of economic independence from foreign suppliers.
Sen’s work on famine and poverty in developing countries has attracted most attention recently. His central idea is that famines are not simply a matter of the balance between total food supply and the population. Large-scale famine can take place even if total food supply is adequate. He argues, for instance, that rice supply was higher at the time of the Bengal famine of 1943 than in 1941 when there was no famine. Famines occur, according to Sen, when there is a collapse for particular groups in their “entitlement” to food. (This terminology is potentially confusing, as he himself acknowledges, since it has no moral connotations, unlike its usage by Nozick.) Sen’s basic notion is that the entitlement to rice of a landless laborer depends on the wage rate he receives for his work relative to the price of food. If for some reason there is a rise in the demand for food by other groups (for example, in the Bengal case, because World War II generated more income in the urban areas), then the entitlement of other groups (rural laborers) may fall sharply. Sen’s analysis of famines has been misinterpreted by some. It certainly does not imply that total food production is unimportant, still less that food aid is unnecessary. What it does do is to direct our attention to the distribution of entitlements—the real ability to obtain resources—and the mechanism by which resources are distributed.
Here we return to Sen’s concern with ethical issues, as in his discussion of national and international distribution. It is both a cheap gibe and a deep question to ask those concerned about poverty in advanced countries why they do not demand the same degree of income redistribution on a world scale. Why should not the US Bureau of the Census poverty line be applied world-wide? Sen’s essay on “Ethical Issues in Income Distribution: National and International” is a valuable attempt to confront different moral theories with this kind of problem. He is critical of approaches that treat nations as entities, noting that this serves to conceal inequalities within countries. He attacks the notion that the existing world order can be justified by applying principles of justice in acquiring and transferring wealth, along the lines advocated by Nozick. This, he argues, would be difficult “given the acknowledged lack of free and competitive exchange in the trade between nations, and given the documented records of colonial economic relations.”
This brief review has covered only part of Sen’s contributions to economics. I have said nothing about growth theory, capital theory, peasant agriculture, the behavior of labor cooperatives, the economics of the family, liberalism, rights, or the profit motive, all of which have been subject to his scrutiny. Nor have I the space to discuss Sen’s concept of “capabilities” and its relation to the definition of poverty, on which he has recently written extensively. He is indeed remarkably prolific and seems to have sped up at just the age when other people begin to relax. When invited to give yet another public lecture, he is not content to repeat old arguments; as the lectures reprinted in these volumes bear testimony, each contains a new twist to the story previously told.
What then are the main qualities of Sen’s extraordinarily impressive contribution? First, there is his constructive approach and his willingness to tackle uncomfortable subjects. One senses his irritation with the sterility of earlier welfare economics and its timidity with regard to issues of distribution. He is not content simply to point out the difficulties; his way is to seek solutions. He himself has emphasized “the danger of falling prey to a kind of nihilism [which] takes the form of noting, quite legitimately, a difficulty of some sort, and then constructing from it a picture of total disaster.”
The second striking feature is the integrated view of the subject that runs through his work in different fields and links the research at different stages of his career. This is apparent from the way in which the essays on development economics interlock with those on social justice. His concern with famine is linked to that with poverty, and his writing on the measurement of poverty draws in turn on his more philosophical investigations. It is also apparent in the link between his work on individual choice and that on social choice. We have seen how he has drawn attention to the role of social values in affecting individual behavior, and argued that recognition of the complexity of individual motivations has implications for the formation of social judgments.
What is more, he has been remarkably successful in arousing the interest of the economics profession in the issues on which he has concentrated. He has significantly affected the evolution of the subject. Indeed it has been suggested to me that one of his key contributions has been to legitimize the investigation of certain topics—such as famine—that had previously been outside the profession’s concerns. (It does not, of course, speak well for present-day economics that this should be necessary.) His central position in the discipline has meant that when he has ignored the conventional lines of demarcation then graduate students have had the courage to follow.
Has he always used this influence well? In some cases this is open to debate. I am probably not alone in believing that the heavy emphasis in graduate research on social choice theory has been excessive when viewed against the weak theoretical underpinnings of other fields of economics. (Sen has recently estimated that the number of books and articles on formal social choice theory exceeds a thousand.) On occasion Sen seems to have allowed his fascination with the intellectual logic of ideas to run ahead of his judgment. He gave the initial impetus to the analysis of poverty measures, but the particular approach he adopted, with its emphasis on a person’s rank in the distribution, may have hindered rather than accelerated progress. In his criticism of the standard model of economic man, he may have failed to persuade economists of the importance of the elaborate distinctions that he draws in his discussion of choice and preference; a more direct attack might have produced clearer results.
This said, his influence on economics has been enormously to the public good. In the introduction to the second volume, Sen criticizes the narrow boundaries of modern economics, “involving partitions that classical economists, such as Smith or Marx, would not easily have recognized.” He has done much to reverse this trend. By emphasizing the richness of human motives, the institutional complexities of development, the subtleties of social goals, he has stimulated research on topics that are not always aesthetically satisfying but are always important (and he often succeeds in treating them with an elegance others failed to attain). His writing has not only helped persuade those in other disciplines that discourse with economists is possible but has also contributed prominently to their own thinking.
In his analysis of famines, Sen has emphasized the importance of the press and public debate in securing effective action, pointing for example to the large-scale famine that remained concealed in China between 1959 and 1961. As he says, “a study of entitlements has to go beyond purely economic factors and take into account political arrangements (including pressure groups and news distribution systems).” And if they are not to repeat conventional platitudes, the press and the interest groups should attend to the cool, yet concerned, analysis of which Amartya Sen is a master.