Kenneth Arrow has died

by Henry Farrell on February 22, 2017

Arrow was a wonderful economist and from all accounts that I’ve heard, a very good guy. Others are much better able to evaluate the technical contribution than I am. Still, It always gave me a little pleasure that the person who had co-discovered the foundational account of general equilibrium, and more than anyone else, had built the basics of social choice theory was a cheerful social democrat. My old co-supervisor, Colin Crouch, told me about the time that he met Arrow at a conference in the Vatican and wandered off together with him to chat about their bemusement at the odd life chances that had brought two left-wing Jewish boys together to roam the corridors of the Catholic Church’s sanctum sanctorum. Arrow was also a one-time Crooked Timber seminar participant – we’re lucky to have had him, and I’m glad of the contact, however slight and glancing, that editing the piece involved.



John 02.22.17 at 4:25 am

I lived with him for 3 years during college. He rented rooms to poor students. I knew the whole family very well. I have many stories about him but my favorite one was during the during the great Cambridge Blizzard of 1976 or 1977. We had three feet of snow, people cross-country skiing over cars etc. Our heat was out for a week. When the oil man finally came to fill the tank, Kenneth spent an hour with him in three feet of snow keeping him company.
A really fine man. Thank you very much for writing something about him.


fishfush 02.22.17 at 12:15 pm

Whether he’s allowed into heaven shouldn’t depend on whether the angels prefer Shapley to Nash or vice-versa.


Jameson Quinn 02.22.17 at 3:28 pm

A giant whose shoulders I have stood on for most of my life. I complain about those who prefer to sit on his shoulders, who think that the field of social choice is no more than impossibility proofs and possibility demonstrations; but he himself kept striving upwards for his whole life, and the world will miss him.


Patrick S. O'Donnell 02.22.17 at 3:34 pm

I won’t dispute the accolades (and not only because it’s in bad taste), especially the long-standing consensus that he was “a very good guy.” All the same, I’m inclined to believe that Arrow’s undoubtedly clever if not brilliant “impossibility theorem” (Amartya Sen describes it as a ‘result of breathtaking brilliance and power’) had, and speaking generally, a pernicious effect on the discipline of economics, captured in part by Deirdre (né Donald) McCloskey’s comment that it, along with other qualitative general theorems in the discipline, “do not, strictly speaking, relate to anything an economist would actually want to know,” in other words, “axiomatizing economics” (which Arrow alone cannot be held responsible for) was a turn for the worse, no doubt motivated by a desire to bring (natural) scientific respectability and putative “rigor” (of the sort believed to characterize physics) to a field not amenable to same (to put it bluntly if not mildly). For a different sort of critique of his work in this regard in economics and the “social choice” literature, see Hausman and McPherson’s Economic Analysis, Moral Philosophy, and Public Policy (Cambridge University Press, 2nd ed., 2006). There is also a vigorous critique of the use of this theorem by professional economists and political scientists in S.M. Amadae’s Rationalizing Capitalist Democracy: The Cold War Origins of Rational Choice Liberalism (University of Chicago Press, 2003). Sen has a decidedly more favorable assessment of the “impossibility theorem” in his book, Rationality and Freedom (Belknap Press of Harvard University Press, 2002).

Alas, it was mischievous interpretations and application of his famous “impossibility theorem” that unequivocally did enormous harm to the discipline of political science, particularly with regard to democratic theory (and by implication, praxis as well): see Gerry Mackie’s Democracy Defended (Cambridge University Press, 2003).


Donald A. Coffin 02.22.17 at 4:16 pm

Links abound of course. For an excellent discussion of his contributions, this (the first of four posts that will appear this week) is a good place to start.


Glenn 02.22.17 at 5:14 pm

While listening to candidates for office and Trump in particular, I remembered a quote of Kenneth Arrow (from where I don’t remember): “Information is precisely that which reduces uncertainty,” and from this I ascertained to my satisfaction the complete absence of information in American political discourse.


Lee A. Arnold 02.22.17 at 5:45 pm

Great intellect over a wide range. So deep that he took his mathematical analyses to the brink of epistemology. One of the few economists who generated insights for complex systems. Not sure he ever wrote this, but he seems to have taken economics as a set of separated tools, hoping to see how they might be applied — but never presuming that the results describe reality, as so many other economists do. Several of his papers end with statements about why the economics falls short. Showed ways in which kinds of important “information” cannot be transmitted by market prices. Just about the only economist who argued that natural ecological systems (including the processes of climate change) must be considered from outside the market system, and using a discount rate to price them is almost useless for policy.


peterv 02.22.17 at 10:58 pm

Glenn @6

Except that, contrary to Arrow and conventional wisdom in Economics, more information may increase uncertainty. It does so when it reveals more possible futures or more entities in an environment than had hitherto been known. When Steve Jobs returned to Apple, the company’s future options and possibilities increased significantly over the single, doom-laden trajectory that had existed before.


Peter Dorman 02.23.17 at 3:07 am

I remember a nice moment a number of years ago when Arrow and Solow both served on a blue-ribbon panel evaluating contingent valuation in the wake of its use in the Exxon Valdez spill. They were presenting at an ASSA session, and during the question period a young woman rose from her seat to challenge the panel’s conclusion. She didn’t word her criticism very precisely, and Solow responded with a thundering blast that translated into “you don’t know anything and are no position to have an opinion”. Arrow then leaned into the mic, and in a quiet voice, said something like, “Bob, what she means is this, and it’s a reasonable point of view.”

I appreciate academics who’ve risen to the status of uber-experts but don’t judge others by their level of expertise.


Joe Pickard 02.23.17 at 3:54 am

I stopped reading Patrick S. O’Donnell (4) at “All the same…”


Tabasco 02.23.17 at 4:05 am

Many economics Nobels live into their 90s and beyond, far more than the population average. There is a certain amount of self-selection at play because you can’t get a Nobel if you die young, but even so it is intriguing.


Harry 02.23.17 at 4:08 am

Tabasco — social gradient of health/status syndrome. Same is true of Oscar winners. And, I’m hoping (selfishly), of knights of the realm.


John Quiggin 02.23.17 at 4:27 am

A long and rewarding life, but still a great loss to those he leaves behind.


JPL 02.23.17 at 9:16 am

Sorry for the tangential comment, but for what it’s worth (no doubt not much) here are some thoughts I had in response to an idea of Kenneth Arrow’s.

It is not “the will of the people” that is important in democratic systems of governance, but the principle of action and critique of action. My Kantian pragmatist suggestion for democratic systems is that there should be an institutional and permanent mechanism to allow this relation to function in any system of government. Complaints from the people should be able to be registered and considered, not ignored, and citizens should be confident that their complaints are being seriously considered. It should always be possible for changes in policy to be made in response to this critique. (The role of critique should not be left to the commercial media, nor should it be restricted to the opposition party, as in question time.) The meaning of an election or referendum result is always unclear, since there are so many, possibly open-endedly so, different reasons people had for voting the way they did and not the other. As reasons for the action they may be mistaken or ill-intentioned, but the particular complaints may be valid. And of course, some things that voters sometimes want (e.g., suppression of minorities) are not valid things to want. Complaints have to go through a critical evaluative process. One of the most important principles of any constitution is provision of a mechanism to protect minorities (the rights of minorities) from abuse resulting from the majoritarian selection of office-holders (e.g., election), or a plebiscite, including specifying the kinds of ethical violations that are likely and prohibited. The meaning of the constitution is that it specifies laws and protections of rights that apply equally to all citizens and rules out laws that violate this principle. And since these protected rights are from an ethical perspective universal human rights, this protection must be extended to all individuals within the boundaries of the country. What makes a constitution “exceptional” is its recognition of the primacy of ethical principles in the ideal sense over considerations of mere power, such as results from elections. Elections can determine which party is to temporarily hold the structural positions in government; they can not tell us anything definite about policy preferences. (From a consideration of “Arrow’s theorem”.) (Note on the role of “experts”: The preferable role of experts (such as social scientists) is in critique, not in governing.)


LFC 02.23.17 at 3:04 pm

Tabasco @11
Arrow got his Nobel in ’72, when he was 51.

I had a brief but not insignificant academic encounter with Arrow when I was a senior in college (almost 40 years ago), though I wasn’t studying economics. I won’t go into the details b/c it would take too long to explain the context and would not be of that much general interest. Regrettably, I haven’t in the intervening years managed to read any of his work.


LFC 02.23.17 at 3:19 pm

p.s. I mean a personal encounter, in case it wasn’t clear. (It’s sometimes hard to be both cryptic and clear at the same time.)


pnee 02.23.17 at 7:40 pm

peterv @8

I don’t have any particular attachment to Arrow’s statement that information is that which decreases uncertainty, but I don’t see how your example refutes it.

Steve Jobs return as CEO was a significant actual change in the management of Apple, not in “information about Apple”. If anything, it raised uncertainty precisely by making much exiting information about Apple’s plans obsolete.

Now, a rumor a few weeks before that “Jobs may come back to Apple” might count as “information” that “increased uncertainty”. I’m not sure how rigorously Arrow defined “information” and “uncertainty” in the original context. And if you didn’t hear the rumor and were completely surprised by Jobs’ return, did you really have less uncertainty than someone who heard the rumor?


peterv 02.23.17 at 10:57 pm


For us members of the general public, the return of Jobs to Apple was a complete surprise. It was not rumored in any way in any public forum, to my knowledge. The futures for the company considered possible by external observers (ie, non-insiders) were many more than before. Exactly as I said and as you agree, the public announcement of Jobs’ return was new information which increased public uncertainty.


Lee A. Arnold 02.24.17 at 11:45 am

“Information” has different definitions in different disciplines. One of Arrow’s last lectures explains his use of the word, and also his view of the current state of many other things. Only 9 pages, no math:


likbez 02.26.17 at 11:20 pm

Two questions to esteemed commenters here:

1. Is not the idea of permanent equilibrium a fallacy?

2. If not excessive use of mathematics in economics called mathiness?


likbez 02.27.17 at 2:59 am

Reposted from Economict’s View

anne said in reply to anne…
July 25, 2009
Why Markets Can’t Cure Healthcare

By Paul Krugman

Judging both from comments on this blog and from some of my mail, a significant number of Americans believe that the answer to our health care problems — indeed, the only answer — is to rely on the free market. Quite a few seem to believe that this view reflects the lessons of economic theory.

Not so. One of the most influential economic papers of the postwar era was Kenneth Arrow’s “Uncertainty and the Welfare Economics of Health Care,” * which demonstrated — decisively, I and many others believe — that health care can’t be marketed like bread or TVs. Let me offer my own version of Arrow’s argument.
There are two strongly distinctive aspects of health care. One is that you don’t know when or whether you’ll need care — but if you do, the care can be extremely expensive. The big bucks are in triple coronary bypass surgery, not routine visits to the doctor’s office; and very, very few people can afford to pay major medical costs out of pocket.
This tells you right away that health care can’t be sold like bread. It must be largely paid for by some kind of insurance. And this in turn means that someone other than the patient ends up making decisions about what to buy. Consumer choice is nonsense when it comes to health care. And you can’t just trust insurance companies either — they’re not in business for their health, or yours.

This problem is made worse by the fact that actually paying for your health care is a loss from an insurers’ point of view — they actually refer to it as “medical costs.” This means both that insurers try to deny as many claims as possible, and that they try to avoid covering people who are actually likely to need care. Both of these strategies use a lot of resources, which is why private insurance has much higher administrative costs than single-payer systems. And since there’s a widespread sense that our fellow citizens should get the care we need — not everyone agrees, but most do — this means that private insurance basically spends a lot of money on socially destructive activities.
The second thing about health care is that it’s complicated, and you can’t rely on experience or comparison shopping. (“I hear they’ve got a real deal on stents over at St. Mary’s!”) That’s why doctors are supposed to follow an ethical code, why we expect more from them than from bakers or grocery store owners.

You could rely on a health maintenance organization to make the hard choices and do the cost management, and to some extent we do. But HMOs have been highly limited in their ability to achieve cost-effectiveness because people don’t trust them — they’re profit-making institutions, and your treatment is their cost.

Between those two factors, health care just doesn’t work as a standard market story.
All of this doesn’t necessarily mean that socialized medicine, or even single-payer, is the only way to go. There are a number of successful healthcare systems, at least as measured by pretty good care much cheaper than here, and they are quite different from each other. There are, however, no examples of successful health care based on the principles of the free market, for one simple reason: in health care, the free market just doesn’t work. And people who say that the market is the answer are flying in the face of both theory and overwhelming evidence.



pnee 02.28.17 at 12:52 am

peterv @18

Actually, what I said was the actual change to jobs as CEO, not the news of it, is what created the uncertainty. I tried, unsuccessfully it seems, to use the hypothetical rumor as an attempt to differentiate between an event and information about an event.

If you view a event like change in key personnel as “information” then your working definition of information is broader than perhaps I am used to.

In any case, I think we’ll have to agree to disagree on whether your example works.


pnee 02.28.17 at 12:59 am

Sorry to continue my last post in a separate piece, on a digression no less, but I think I should add that I’m used to a definition of “uncertainty” that means roughly the likelihood and degree of being surprised by the actual outcome.

So if Steve Jobs has become CEO, uncertainty about Apple’s future has increased, whether I have heard the announcement or not.

Hearing the news, I would at least be aware that Gil Amelio’s plans are going to scrapped, and this would reduce my uncertainty, in comparison to not knowing about Steve’s new job.


Wylie Bradford 02.28.17 at 2:22 am

Patrick @4

I’m still racking my brain trying to work out what the ‘pernicious effects’ on economics of Arrow’s General Possibility Theorem are meant to be.

Arrow’s result throws up serious (if not, at root, unanswerable) questions about the coherence of the idea of a social welfare function constructed out of individual preferences. Economists were (and still are) rather blase in their assumption that a social ordering of states of nature can be constructed out of individual orderings and retain the same properties of the later i.e. that we can talk about what society wants in the same way as we do about what an individual wants (theoretically speaking) because we can aggregate individual wants into a consistent social ordering. Hence all the ‘social indifference curves’ we are wont to draw for undergraduates when talking about optimal resource allocation.

Arrow’s theorem pretty much says that that can’t be done, that an aggregation of preferences can’t be relied upon to have the same properties as an individual ordering (at least not if it is going to give a non-zero weight to every individual preference ordering).

That’s a big deal. It pretty much wipes out welfare economics based on Bergsonian social welfare functions, it puts the skids under representative agent models (as another aspect of the argument that individual agents can’t validly stand in for collections of agents) and it very seriously undermines cost-benefit analysis (which, at its theoretical core, relies on comparisons of a social welfare function with and without the project). It essentially challenges the meaning of ‘social welfare’ from the perspective of the subjective value theory that underlies the meaning of individual welfare in economic theory.

These are fundamental implications, but they are only ‘pernicious effects’ if the ideal is to keep pretending that a particular theoretical discourse is coherent when it isn’t.

As for Arrow’s theorem not being concerned with ‘ anything an economist would actually want to know’, I would submit that

a) the soundness (or not) of the theoretical bases of welfare economics is something with which economists ought to be greatly concerned, as without welfare economics economists are of very limited use indeed;

b) the time is long since past when McCloskey should be recognised as a huckster trading in unsubstantiated assertions, self-reflexive appeals to authority and niche marketing.

Comments on this entry are closed.