The history of economics is rich in economists of the first rank who were equally influential as philosophers, sociologists, and political thinkers: Adam Smith, John Stuart Mill, Karl Marx, and Friedrich von Hayek are just four from very different times and of very different political allegiances. Amartya Sen is another. His contributions to the purest and most abstract realms of economic theory make him “an economist’s economist”—and were one reason for his Nobel Prize in economics in 1998. But he is universally known for his work on famines, economic development, the condition of women, and the problems of poverty, work that has earned him the admiration of everyone who works in the field of development. This was another element in his Nobel citation.
Rationality and Freedom focuses on abstract issues in economic theory, but the real world obtrudes enough to remind us what motivates this work. Three or four fascinating pages discuss the surprising fact that the Indian state of Kerala “has by a large margin the longest life expectancy at birth (67.5 years for men and 73 years for women, compared with around 56 years for both men and women in India as a whole),” while more of its inhabitants complain of illness and general ill-health than anywhere else. In contrast, Bihar and Uttar Pradesh have very low life expectancy and “astonishingly low rates of self-assessed morbidity.” As health care improves and people become aware of the contrast between the possibility of good health and the actuality of minor illness, they notice what they would have ignored if there had been no progress. The implication is that objective measures of well-being—longer life expectancy—may not match subjective measures—feeling well. It is an understatement to say that this complicates the evaluation of standards of living across time and in different societies.
Rationality and Freedom consists of twenty-two chapters, most of them essays and lectures published in economics journals during the two decades before Professor Sen’s Nobel Prize. They are on two themes central both to economics and to political theory and moral philosophy. The first is the nature of individual rationality. “Rationality,” says Sen, “is interpreted here, broadly, as the discipline of subjecting one’s choices—of actions as well as of objectives, values and priorities—to reasoned scrutiny.” This is fighting talk from an economist; economics has for many years adopted very narrow conceptions of rationality, concentrating either on the capacity to choose efficient means to what are presumed to be selfish ends, or even more minimally on consistency in choice. What ends we pursue and how we have come to adopt them are widely thought not to be any of the economist’s business.
Sen insists “that it is important to reclaim for humanity the ground that has been taken from it by various arbitrarily narrow formulations of the demands of rationality.” This is not merely because, as we shall see, some transparently crazy behavior is rational by the standards of these narrow formulations; it is also because the prestige of economics among the other social sciences is such that sociologists, legal scholars, and political scientists have been applying its analytical methods to their own disciplines under the label of “rational choice theory.” The first, introductory, chapter of this book is the best discussion of what is wrong with narrow notions of rationality that a lay reader is ever likely to come across; it might with advantage be turned into a pamphlet and given to incoming graduate students in all the social sciences.
The second theme with which Sen is concerned is the question of social choice: Is there a rationally defensible way of getting from the preferences of rational individuals taken one by one to what a whole society should choose? Many readers will complain that that is not the right way to phrase the question, since not least among the issues in social choice theory is the question whether it makes any sense to think of a society choosing at all. Robert Nozick’s Anarchy, State, and Utopia (1974) argued passionately that it did not, that there are only individuals and their choices; James Buchanan, the 1986 winner of the Nobel Prize in economics, has been a fierce critic of the idea of aggregating individual preferences into a “social choice,” and a good deal of Rationality and Freedom is devoted to meeting such criticisms of the very idea of social choice.
But the economist who dominates the entire seven hundred pages of this book is Kenneth Arrow. In 1950, Arrow, a graduate student at the time, published his “impossibility theorem”—its formal title was the more optimistic-sounding “General Possibility Theorem”—which showed that (given a few apparently quite irresistible requirements) there is no rule for combining individual preferences into a social choice that does not generate paradoxes. Suppose, for example, that a society wants to decide whether the proceeds of a national lottery should be spent on education, health, the arts, or sports. We are tempted to think that there must be some way of taking each citizen’s preferences about the outcome and combining them to produce the choice of society as a whole. Arrow’s theorem demonstrated that there is not.
One might think that this result should spell the end of social choice theory. In fact, of course, it did the opposite. It inspired so many economists, including Arrow himself, to look for ways around the impossibility, and to devise alternative ways of getting from individual choice to social decision, that as Amartya Sen observes in a footnote, major economics journals finally had to discourage submissions on the subject because they were swamped. Rationality and Freedom is, then, in part an act of homage to Arrow; when reviewing the first volume of Arrow’s Collected Papers, Sen writes, “This volume is all together a marvelous collection of fine analyses, stimulating ideas, and powerful arguments from one of the greatest of economists. The quality of the reasoning varies all the way from the merely admirable to the spectacularly superb.” Which is just what one wants to say of Rationality and Freedom.
What constitutes individual rationality is a question as old as philosophy. Within the social sciences, the assumption that individuals must be treated as rational agents is less a matter of unwarranted optimism about human beings—since none of us is wholly rational all the time, and most of us are less rational than we would wish to be a good deal of the time—than a condition of being able to offer some explanation of human behavior. Commentators have expressed bewilderment about Saddam Hussein’s irrational conduct both before and during the recent war in Iraq. If he had no weapons of mass destruction, why did he provoke an invasion by not fully meeting the demands of the inspectors? That is, given his presumed goal of remaining in power, why did he not take the obvious steps to avoid an attack? If he wanted to make life difficult for an invading force once war had begun, why did he not mine the oil fields and the bridges over the Tigris? Given what we would all assume his goals to be, why did he not take efficient steps to achieve them? Only when people are rational in the limited sense of acting in the way that best promotes their purposes can we understand what they are doing and predict what they will do next.
The idea that rationality is best defined in this limited way, without getting into contentious issues about what human beings ought to want or to set themselves as their favored goals, is commonplace in social science. Indeed, as Amartya Sen says, the search for austerity in defining rationality has led a good many economists to define the rational person as one who shows consistency in his or her choices, and nothing more. But, as he also says, five minutes’ reflection on the kinds of counterexample that all theorists delight in coming up with shows that such a minimal definition may come at too high a price.
Professor Sen provides a particularly gruesome counterexample. Suppose we find a man busily cutting his toes off with a blunt knife. If he wishes to be without his toes, his actions are consistent with his goals, and his behavior displays internal consistency. The economist who sees in rational behavior nothing more than internal consistency cannot criticize his behavior, except perhaps by suggesting that a sharper knife would attain his goal more expeditiously. What the example shows is that defining rationality with no attention at all to either the justifiability of the agent’s beliefs or the justifiability of the ends being sought is not so much austere as self-defeating. Even consistency, of course, may or may not be a mark of rationality; to think I must eat bacon and eggs at every meal because I ate them at breakfast is not a high point of rational conduct. Even the much-diluted principle that a person’s “values” or “ultimate preferences” must be consistent over time may be wrong. Should we not learn from experience and revise our values and preferences in the light of that experience?
A subject on which Amartya Sen is particularly illuminating is the sense in which rationality involves “maximization,” or getting as much as possible. The most famous figure in the history of arguments about rational choice, and specifically about maximization, is not a person but a donkey. Centuries before Sen’s famous essay “Rational Fools,” the fourteenth-century logician Jean Buridan—a philosopher at the University of Paris—popularized the dilemma faced by a “rational ass.” Buridan’s ass faced two equally attractive bundles of hay; they were not only equally attractive but equidistant from him. Unable to decide between them, he starved to death. It was only because he was a rational ass that he starved; a conservative ass that habitually ate whatever was on the left (or right) of his line of view would have been fine; an impulsive ass that ate the first pile of hay his eye lighted on would have been fine. The trouble was that neither pile was (rationally) preferable to the other, so Buridan’s ass could not choose between them. Buridan’s ass stars in several essays here.
Many moral philosophers have recently argued that rationality has nothing to do with maximizing anything whatever. The right number of chocolate bars to eat may be debatable, but it surely is not “as many as possible”; and there are many activities in life in which “good enough” is what we want—who wants to spend all his waking hours ensuring that he has the very best television set or the very best car? In everyday life, getting the best result is mostly very different from maximizing. Sen recurs several times to the reason why this view (from which he does not in fact dissent) does not impugn his view that maximization is essential to rationality—and that an understanding of this point would have saved Buridan’s ass. There is a point here that is partly verbal—economists and moral philosophers seem to employ “maximization” and “optimization” in opposite senses—and partly not.
“Maximization” in the sense in which welfare economists use it is a mathematical notion; it means that the function relating your actions and your payoff is at a maximum. Less chocolate than that leaves you wanting more; more chocolate than that leaves you wanting less. “Optimization” is a more restricted notion, because it implies that there is a single best solution. All maximization implies is that there is no better solution—but there may be many worse. When we are right to decide “enough is enough” we are maximizing after all; but the poor ass looked for a uniquely best choice, and overlooked the existence of two equally good choices, either of which was good enough and a great deal better than starvation. Haystack A and haystack B were in the mathematician’s sense maxima, and a truly rational ass would have picked one.
What this suggests—aside from the simple pleasure to be derived from what Sen calls “brainteasers”—is that an essential element in rationality is the ability to step back from your present wants and beliefs and reconsider them:
The need for reasoned scrutiny applies not only to accommodating moral and political concerns in personal choices and in social living, but also in incorporating the demands of prudence. Many failures to be adequately prudential arise, in fact, precisely from the absence of an adequately reasoned scrutiny. If, for example, Buridan’s ass died of starvation because of a failure to grasp the discipline of maximization with an incomplete preference (i.e., to choose either haystack, but not neither), the remedy lies in a deeper scrutiny.
This is where rationality and freedom intersect; a person who had no more control over her aims and beliefs than a computer has over running its software would not be “rational,” but would be a slave to a program.
Sen’s guiding principle is that we have to think about human beings in ways that do justice to the complexity of their values and beliefs. If they in fact guide their conduct by high principle, a passion for justice or freedom, simple compassion for the badly off, or whatever else, there is nothing to be said for theories that represent “rationality” as the single-minded pursuit of self-interest defined in the narrowest possible terms. This makes him a particularly deft critic of the contemporary passion for so-called rational choice theory. “RCT,” as Sen usefully abbreviates it, has divided law schools and departments of sociology and political science for many years.
Rational choice theory appeals to academics who hanker after the kind of consensus they see among economists about the standards for good, professional work, even when they disagree about policy issues. In law, Richard Posner is a leading light of the law and economics movement. Indeed his book Economic Analysis of Law (1973) began that movement by arguing that many of the features of our law, both civil and criminal, are best understood as ways of providing incentives to desirable behavior to people who are presumed to approximate the self-interested rational man of economic theory. A later book, Sex and Reason (1992), is better known outside law schools, if only for its suggestion that mothers should be allowed to auction off their newborn babies. It is now fifty years since Anthony Downs’s An Economic Theory of Democracy started political scientists down the track of thinking of politicians and voters as trading policies for votes, an analogy that has been illuminating and frustrating in about equal measure.
Opposed to RCT are those who think that what needs emphasizing are the ways in which politics, law, and social life generally are not like the subject matter of economics. If anyone can bring peace to this scene, it is Amartya Sen; he neither dislikes mathematics and the attempt to set up interesting axiomatic models, as many critics of RCT do, nor does he encumber himself with the assumption that all social interactions must be explained by “rational self-interest” as too many enthusiasts for RCT do.
So Sen says, “To its credit, as I see the issues, RCT has instilled the significant understanding that systematic—rather than ad hoc—explanations are needed for observed behavioral phenomena, that human behavior has many regularities, and that these regularities can be caught within a maximizing framework.” On the other hand, Sen criticizes RCT for systematically impugning the role of ethical, or in other ways principled, motivations. It is not that the enthusiasts for RCT deny that we act for moral reasons, or act on principles that restrict our pursuit of our own self-interest; rather that they reinterpret those reasons so as to make them part of, or elaborate means to, our own self-interest, after all. Sen writes:
Choices based on social or moral or politically integrative reasons have to be reinterpreted, in this approach, within the format of intelligent pursuit of self-interest (with complex instrumental linkages as and when needed for this accommodation). This has given the explanatory role of RCT an almost forensic quality, focusing on the detection of hidden instrumentality, rather than any acknowledgment of direct ethics. Things, it is darkly hinted, are not what they seem (or at least seemed to simple-minded observers like Smith or Kant).
Squeamish readers may feel this even about the work of the most distinguished practitioners such as Gary Becker, whose Nobel Prize in economics in 1992 was awarded for thirty years of extending economic analysis to behavior which seemed particularly resistant, such as marriage, the decision to have children, or divorce. Even when Becker insists on the reality of love, it is never quite clear whether our attachments to each other express, disguise, or simply misrepresent the process whereby a woman trades the secure access to sexual favors that man is thought to want in exchange for the secure access to the man’s earnings the woman is thought to want. Something of the same ambiguity hovers over Becker’s work on the economics of crime, where it is never quite clear whether assaulting people is both wicked and costly to the victim, or merely costly and sentimentally described as wicked by non-economists.
Much of Rationality and Freedom is concerned with the difficulties of “social choice.” Chapter Two began life as Amartya Sen’s Nobel lecture, delivered in Stockholm in December 1998, and the title he gives it here is “The Possibility of Social Choice.” Because the Nobel lecture is both an occasion for explaining to a wider audience what it is that has animated the career of the prize winner and an occasion to remind colleagues what their discipline is most fruitfully concerned with, the lecture takes in much more than the problems posed by Arrow’s “impossibility theorem,” including Sen’s abiding interest in such issues as gender inequality in economic development, famines, and the difficulty of integrating a concern for individual rights into any plausible theory of social choice. To put it summarily, social choice theory starts from the premise that individuals may legitimately have preferences about absolutely anything—I may have preferences for or against your religion, sexual habits, reading materials, and so on—while a theory of rights starts from the premise that there are many things where my preferences about your life do not count—your religion, sexual habits, and what you read, for instance. If you have a right to read what you like, there is at least one unavailable social outcome, namely the one by which you read what I choose for you.
Social choice theory is not easy going for the non-economist. Part of the difficulty is verbal. Social choice theory is not primarily concerned with persons choosing anything, but with aggregating individual preferences into something one could plausibly call a collective preference. And it raises not one but a multitude of issues. They are summed up by Sen as follows:
When would majority rule yield unambiguous and consistent decisions? How can we judge how well a society as a whole is doing in the light of the disparate interests of its different members? How do we measure aggregate poverty in view of the varying predicaments and miseries of the diverse people who make up the society? How can we accommodate rights and liberties of persons while giving adequate recognition to their preferences? How do we appraise social valuations of public goods such as the natural environment, or epidemiological security?… The reach and relevance of social choice theory can be very extensive indeed.
As Sen explains, social choice theory in its contemporary form draws on two very distinct traditions. One is utilitarianism; the other is the interest in voting systems that began with Enlightenment thinkers such as Jean Charles de Borda and the Marquis de Condorcet. They have, as Sen observes, complementary strengths and weaknesses. Utilitarianism provides an unequivocal answer to the question “What is the best social state?” It is the one that maximizes total happiness, otherwise called “utility.” Utilitarianism has deep problems, however. One is that it requires us to make dubious comparative judgments about everyone’s happiness. It is hard even to begin to compare the happiness you get from winning a chess game and the happiness your child gets from eating an ice cream cone. But to justify sacrificing your game to her trip to the ice cream parlor on utilitarian grounds, you have to be able to make such a comparison.
Another difficulty is that utilitarianism is hard on people who are chronically unhappy, even if they desperately want what will do little to cheer them up. A heart patient may very much want the expensive treatment that prolongs an unhappy existence; but if happiness is what we are after, the money would be better spent on cheerful and healthy teenagers. A related objection is that by focusing only on total happiness, utilitarianism ignores the route by which happiness is achieved: $500 may yield you and me equal happiness, but it makes a difference whether you have earned it or I have stolen it.
Utilitarian moral and political theory has ways around some of this, but that is not the point. Professor Sen’s point is narrower; we want a way to assess whether one “social state” is better than another; measuring “total utility” seems a plausible solution, because it corresponds so neatly to the idea that an individual is “best off” when she or he is “happiest.” But, says Sen, by the late 1930s, the whole idea of comparing the happiness, or “utility,” of one person and another had come to seem suspect. Nor was it only the intrinsic weaknesses of utilitarianism that were at issue. The philosophical climate of logical positivism had made it seem less than respectable to appeal to mental states such as happiness in any case; only something visible such as choice or “revealed preference” would do.
This came close to closing the gap with the tradition that began with an interest in voting procedures; its attractions are obvious. If we think of individuals ranking social states or conditions in order of preference, we can bypass questions about why individuals prefer one social state to another—health ahead of education ahead of sport ahead of strong defense, say. We just have to devise a voting system that will always get us from whatever individual preferences there may be to a unique “social preference.” What Arrow’s “impossibility theorem” showed was that it cannot be done. The upshot, taken absolutely literally, is elegantly expressed by Sen:
Two centuries after the flowering of the ambitions of social rationality, in Enlightenment thinking and the writings of the theorists of the French Revolution, the subject seemed to be inescapably doomed. Social appraisal, welfare economic calculations, and evaluative statistics would have to be, it seemed, inevitably arbitrary or unremediably despotic.
Of the escape routes explored in the past half-century, it need hardly be said, despotism does not feature prominently; not only does dictatorship destroy the public good of political participation, it is also the worst of all forms of regime for responsiveness to the interests of a heterogeneous population.
Broadly speaking, Sen’s reaction has been to reintroduce an element of comparison between one person’s well-being and another’s; this is the subject of a great deal of his more applied work.* In Commodities and Capabilities, published some two decades ago, Sen started a particularly exciting line of thought by distinguishing between commodities—measured by GNP—and what he termed “functionings,” human abilities to accomplish aims. The two may well pull apart. In 1982, he thought that China had pulled far ahead of India
in the matter of functionings and living standard, which must not be confused with GNP per head. The capabilities of the Chinese masses are now immensely superior in many vital respects [to] those of the Indian masses. They live a good deal longer, have much safer infancy and childhood, can deal more effectively with illness and disease, can mostly read and write, and so on.
Of course, if other functionings and capabilities were considered, the Indian masses would have come out ahead, including most of those for which civil and political rights are important; and indeed Sen has argued elsewhere that it is precisely the absence of such rights that made it possible for China to conceal a famine whose threat would have been exposed by the press and politicians in India. But Sen’s comparisons provide plausible estimates of objective well-being in an economic sense. None of these comparisons requires us to look at subjective states such as happiness. The importance of subjective well-being may anyway be overestimated: in the same lectures, Sen reported that a year after the Bengal famine of 1943, 45.6 percent of widowers surveyed—men—complained they were in “indifferent” health, while “the proportion of the widows who had that perception was—it is reported—exactly zero!”
Social choice theory is all about moving from individual evaluations to collective evaluations. It is an enterprise that raises doubts in the minds of critics who are concerned with individual rights. One group, of whom James Buchanan is perhaps the most distinguished member, started from a skepticism about the whole idea of a collective evaluation. On this view, social outcomes can be assessed by their legitimacy—whether everyone’s rights have been respected in the process that led to those outcomes—but not by their goodness—whether the society as a whole is doing well. In the long essay “Rationality and Social Choice,” Sen explores the ways in which each side can accommodate the other, with a degree of evenhandedness that, as he says, surprises even himself.
The other anxiety is one that Professor Sen himself has done more to stir up than anyone else in a famous article, “The Impossibility of a Paretian Liberal,” published in 1970, a commentary taking the ideas of the Italian theorist Alfred Pareto as its point of departure. The example Sen discussed—two people trying to reach a collective decision about which of them, if either, should read a book that one thinks is pernicious and the other is prepared to admire—has become rather dated, as Sen observes in a footnote. The issue itself is perennial: How can we ensure that a decision that is made by aggregating preferences can be made to respect individual rights? The Texas law on sodomy may have gratified many Texans and inconvenienced very few, but it was illiberal; legislation that keeps black teachers out of white schools may gratify a racist majority, but it is illiberal; censorship may reassure an anxious majority, but it is illiberal.
On the face of it, the problem stems from the assumption that for the purposes of collective choice, the individual preferences that we aggregate may be about absolutely anything—including what other people read, where they are employed, and what their sexual preferences are. The liberal view is that those kinds of preferences ought not to be counted at all; but what liberals need is a coherent and principled account of which preferences can and cannot be counted, so that social choice works only with suitably sanitized preferences, such as the choice of political leaders or fiscal policies. Thirty years of discussion have shown how hard it is to provide such an account, and to do more than provide a list of objectionable preferences ad hoc. Several of Sen’s essays explore the problem of preserving a concern for rights within the frame of social choice, but none very much advances that more fundamental project.
To say so may sound grudging. It is not. One of the most attractive qualities of Rationality and Freedom is an extraordinary intellectual good nature. Whenever he can express gratitude, Professor Sen does so; whenever he criticizes it is gently—and save on very rare occasions it is only after he has expressed his appreciation for the stimulus provided by the error he uncovers. It would be a poor return for what he offers us here to pretend that everything he writes is equally persuasive; for even when he is unpersuasive he provides intellectual pleasures that few writers can match.
December 4, 2003
There is a very great deal of this, including On Economic Inequality (1973); Poverty and Famines: An Essay on Entitlement and Deprivation (1981); Commodities and Capabilities (1985); The Standard of Living (1987); and Resources, Values, and Development (1997). ↩