Amartya Sen

Rationality and Freedom
(Updated Oct. 22, 2006)

It is centrally important for social choice theory to relate formal analysis to informal and transparent examination. (pp. 73-4)

A deep complementarity between formal and informal reasoning so central to the social sciences is well illustrated by developments in modern social choice theory. (pp. 96-7)

Rationality and Freedom
Amartya Sen
Harvard 2004

Sen's recognition of complementarity here is very welcome -- many formal thinkers aspire to the transcendent status of quantum mechanics, and are thus hostile to the very idea of natural-language interpretation. All of Sen's revisions of neoclassical economics are congenial to me, but even though Sen is classified as orthodox, it seems to me (granted that I'm a systematically informal thinker) that his innovations are actually quite destructive of large parts of the orthodox consensus.

My conclusions are based on the first two chapters of Sen's book, which are the ones intended for a general audience. The first presents Sen's critique of economists' concepts of rationality, and proposes a replacement. The second rescues and revises social choice theory and directs attention to questions of distribution. 

The "rational fool" definitions of rationality Sen criticizes are a.) rationality as internal consistency of choice, b.) rationality as maximization of expected utility, and c.) rationality as maximization of self-interest. To anyone outside economics these definitions have always seemed wrong, first because a prudent psychopath could easily satisfy all three of them, and second because they do not  describe actual human behavior and are merely assumed "to simplify the modeling" (p. 23). For decades economists have relied on Friedman's argument that these simplifying assumptions are, like Newton's, unproblematic ("Introduction to Positive Economics"), but Steve Keen (Debunking Economics, Ch. 7) has shown that this is not the case.

Sen develops his criticisms of these proposed forms of rationality with great diligence and meticulousness. His suggested alternative rationality (p. 4) is "the discipline of subjecting one's choices -- of actions as well of objectives, values, and priorities -- to reasoned scrutiny". This strikes me as quite acceptable, and far superior to the definitions he criticizes, albeit a little thin. (More on this below).

The questions about welfare economics are more complicated. For decades mainstream economics has not tried to evaluate economic from a social or "general welfare" point of view, but only from the point of view of a population of unique and unrelated individuals.  The most aggressive possible statement about general welfare would be Pareto equivalency: the goal should be to attain a state in which no individual can be helped without hurting some other individual. What is specifically excluded is the possibility (accepted by early utilitarians) that certain social changes are good because those helped are helped more than those hurt are hurt (for example, taxing the rich to pay for medical care for the poor). This line of thought questions all governmental attempts to attain reduce poverty and suffering or attain other social goals unless they cost no one anything.

Commentator Henry Evans writes:
It is incorrect to suggest that economists have abandoned the notion of Kaldor-Hicks efficiency (i.e., that there are some changes that increase welfare overall despite decreasing it for some people). This is, after all, the main justification for removing or reducing trade barriers.

The problem with Kaldor-Hicks efficiency is that economics has nothing to say about whether it is just in any particular instance. Justice is just not within the subject matter of economics.

From my point of view, this makes matters worse. In other words, if it's a matter of justifying a practice beneficial to business, the summation and comparison of of utilities is allowed, and social choice is possible. It's only when anyone tries to point out that a dollar has more utility to a poor man than to a rich man, and that a shot of penicillin can have infinitely more value than a glass of wine, that comparisons are disallowed. (I presume that Kaldor-Hicks efficiency compares dollar values and calculates a net, whereas what I was talking about required the interpersonal comparison of actual utility, not defined by dollar value, and I suppose I should have said it that way.)

The fact that justice is not part of the subject matter of economics means that people who use economics as their main window on reality will behave randomly or indifferently with regard to justice, but rationally with regard to self-interest, and in general will tend to be unjust. I do not regard this as a hypothetical or theoretical problem, but as an actual one.

This weakening of welfare economics is theoretically grounded on two principles: Lionel Robbins' conclusion that 'Every mind is inscrutable to every other mind and no common denominator of feelings is possible', (p. 71) and Kenneth Arrow's proof that "even some very mild conditions of reasonableness could not be simultaneously be satisfied by any social choice procedure" (p. 69). Since Robbins has shown that there is no way of precisely measuring relative utility, and since Arrow has shown that there can no algorithm for voting which would make a rigorously fair democratic decision procedures possible, all "social choice" (especially if it ends up mandating redistribution) will be "inescapably arbitrary or intellectually despotic" (p. 95).

These principles are also meticulously analyzed and criticized by Sen, and he replaces them with better principles which make social choice and the comparison of individual welfare possible. To me the details are not terribly important; from my point of view I would say that he simply points out that both Robbins and Arrow were "defeatist perfectionists" who set their standards so high that they could only fail.

I find Sen's conclusions quite acceptable, though not terribly exciting -- the interesting question for me is how Sen's work fits into economics.  His 1998 Nobel Prize shows that he's been quite well-received, but it seems to me that his work should wreak quite a lot of havoc. For example, the  economic notions of rationality rejected by Sen serve quite well to "simplify the modeling" of component individuals for someone trying to define a rigorous formal economic model, whereas Sen's definition ("subjecting one's choices to reasoned scrutiny") hands the economic theorist a can of worms.  Likewise, while Sen's ways of comparing utility or making social choice decisions are completely OK as far as I am concerned, they seem unlikely to be useful to an economist hoping to weave an rigorous, elaborate theoretical structure. Sen has seemingly renounced the mathematical virtuosity and the claim to rigor and certainty which have been the economist's stock in trade for decades, in order to move toward a different model of economics which I would describe as commonsensical.

That's presumably not what Sen would say. While he believes (unlike many technical economists) that the dialogue between common sense and formal economics is valuable, he also believes that his formal argumentation is a necessary part of the debate, and (for example) that rationality is still in some sense maximization. My opinion is that Sen has finally returned us to reasonable positions on rationality, distribution, and social choice which could easily have been attained fifty or sixty years ago (and probably was), and that the long detour through formalization has been harmful. Without speculating about Arrow's and Robbins' personal motives, it seems clear that their work became central to economics because  the profession of economics, especially after 1970, did not want to talk about distribution, but did want to make the best possible case against state intervention in the economy. 

A final question occurs to me. The rational fool Gary Becker was awarded the economics quasi-Nobel in 1992; Merton and Scholes, the not-quite-felons of Long-term Capital Management, received theirs in 1997; Sen received his in 1998. Does this represent some sort of ethical power struggle within the profession, or are economists unaware of any incongruity? Was Sen's prize intended to divert public attention from Merton and Scholes?



Tyler Cowen on rationality (blogged here

Edward Lazear and Steven Durlauf on rationality and other topics

Arrow's impossibility theorem (Wiki)

An old Delong / Sawicki thread on these topics


I am emersonj at gmail dot com.

Original materials copyright John J Emerson

Return to Idiocentrism