The Road to a System that Works (Without Shooting People)

So “mechanism design” has entered the language of everyday economics, as described by newspapers. It is a truism that most Nobel Prizes are won by researchers who tumble onto their topics in their twenties and often have all but nailed them down by their late thirties. Thus, even before they left Harvard University, where they had been undergraduates and graduate students in the 1970s, Eric Maskin, of the Institute for Advanced Study, in Princeton, and Roger Myerson, of the University of Chicago, were firmly on the trail of the ideas that led to their recognition last week, “implementation theory” and “the revelation principle.” Both men are 56.

How is it that Leonid Hurwicz, of the University of Minnesota, shares the spotlight with two economists more than thirty years younger? At 90, Hurwicz is the oldest person ever to be recognized by the Swedes, in any discipline. There has to be a story in that.

The modern theory of mechanism design, as presented today in microeconomics texts, is a hard-edged and high-tech topic, especially auction theory — a body of knowledge that informs the sale of radio-spectrum licenses and timber-harvest rights; structures cap-and-trade emissions schemes and incentive systems designed to lead expert panelists to tell the truth; and which determines the price of advertising on Google and the mechanics of transactions on eBay. The theory is not complete, of course; far from it. Even the best-understood mechanisms, auctions, are no more than a metaphor for more general forms of competition.

But in fact the roots of our understanding of economic mechanisms trace back to a topic intensely debated by European intellectuals in the years immediately after World War I — could a planned economy, such as that of Germany during the war, succeed? Could patriotic bureaucrats, operating without the benefit of markets, prices and money, be depended on to make better decisions than self-interested entrepreneurs?

With the socialist nature of the Russian economy under the new Soviet Union in the 1920s, the topic became more interesting still. And then the Great Depression brought the possibility of planning to the fore in the West, especially in Great Britain. Friedrich von Hayek, in 1935, edited a collection of essays by Continental writers, including the one by Austrian economist Ludwig von Mises that had most forcefully broached the issue, all designed to introduce English readers to the controversy: Collectivist Economic Planning: Critical Studies in the Possibility of Socialism.

By then, the argument had acquired a name (“the socialist calculation debate”); various champions (Oscar Lange, Abba Lerner); a good deal of interest in mathematical and quantitative planning techniques under development elsewhere; and, finally, in 1945, a concise statement of the argument against planning (and against formal methods) by Hayek, in “The Use of Knowledge in Society:”

The economic problem of society is not merely a problem of how to allocate “given” resources…. It is rather a problem of how to secure the best use of resources known to any of the members of society, for ends whose relative importance only those individuals know. [I]t is a problem of the utilization of knowledge not given to anyone in its totality. This character of the fundamental problem has, I am afraid, been rather obscured than illuminated by many of the recent refinements in economic theory, particularly by many of the uses made of mathematics.

Leonid Hurwicz’s entry onto the scene came that same year, when he was 28. He had been born in Moscow in 1917, two months before the Bolshevik Revolution. His parents returned to their native Poland, by horse cart, in 1919. “It was something you could make a Doctor Zhivago movie about,” he told the economics writer Douglas Clement last year for his article, “Intelligent Designer.” He studied law, graduating from Warsaw University in 1938, then studied for a year with Nicholas Kaldor at the London School of Economics. He was in Geneva when Hitler invaded Poland. Paul Samuelson, who had just left Harvard for the Massachusetts Institute of Technology, hired him for a year. The next year Hurwicz moved to the University of Chicago to teach meteorology (mathematically gifted, he was a quick study). There he quickly fell in with the Cowles Commission, which was then engaged in seeking to build the first calculable models of the US economy. And in 1945 the editor of The American Economic Review asked him to review a book by a pair of central European expatriates, John von Neumann and Oskar Morgenstern.

Hurwicz gave Theory of Games and Economic Behavior a rave.

Had it merely called to our attention the existence and exact nature of certain fundamental gaps in economic theory, the [work] by von Neumann and Morgenstern would have been a book of outstanding importance. But it does more than that. It is essentially constructive: where existing theory is considered to be inadequate, the authors put in its place a highly novel analytical apparatus designed to cope with the problem. It would be doing the authors an injustice to say that theirs is a contribution to economics only. The scope of the book is much broader. The techniques applied by the authors in tackling economic problems are of sufficient generality to be valid in political science, sociology, or even military strategy.

For the problem the von Neumann and Morgenstern addressed was absolutely fundamental, wrote Hurwicz. It wouldn’t be easy to define “rational economic behavior” on the part of any one person “when that very rationality depended on the probable behavior of others.”

A century after the essence of problem had been identified (by Augustin Cournot, writing in French), the necessity of strategic behavior at last had become an inescapable part of economics (in mathematics).

For the next five years, Hurwicz experimented with game theory, linear programming, the concept of strategic equilibrium that had just been introduced by John Nash, and all the other exciting ideas that were in the air. He spent a summer in Santa Monica, California, at the free-wheeling Air Force think-tank known as Rand Corp., and met Kenneth Arrow. They began one of the most remarkable research collaborations in all of economics. And then, in 1950, he found his problem. Since no one over the years has drawn out Hurwicz more successfully than George Feiwell, Retired Alumni Distinguished Service Professor of the University of Tennessee, I will rely here on Feiwel’s work in Arrow and the Ascent of Modern Economic Theory to let Hurwicz do the talking.

My work in thus area started around 1950 when I was still with the Cowles Commission. I was writing a more or less expository paper dealing with activity analysis…and happened to use the term “decentralization,” which was then often applied to the market mechanism as a sort of a selling point. But when I used the word “decentralization” I thought I should explain what I meant. So I made a footnote mark, went to the bottom of the page, and began writing, “By decentralization we mean…” But then it struck me that I did not know what we meant by decentralization. That was the beginning of many years of work trying to clarify the concept, because I thought that if we think this property is so important, we should be able to define what it is.

Already the diaspora of math-econ types from Chicago had begun. Hurwicz was hired by Walter Heller at the University of Minnesota, and, with his wife, settled down to a quiet life of research in Minneapolis. The move was, in effect, one-for-one trade: Milton Friedman had left Minnesota for Chicago four years before. Chicago became a center of one sort of economics; Minnesota of quite another.

Hurwicz’s first notable effort came about the time he turned 40, in the form of a game in which participants send messages to each other and/or to a “message center,” and where a pre-specified rule assigned an allocation outcome to each set of messages received. “Optimality and Resource Efficiency in Resource Allocation Processes” was finally published, in 1960.

My interest had been in a broad class of situations, broader than the advanced industrial market economies, including situations in third world countries, and in countries attempting to have some kind of socialist approach to their problems. I have been interested in how one can construct efficient mechanisms that have the decentralization features similar to a market but that do not necessarily resemble a market. For this purpose, I formulated the notion of an informationally-decentralized economy in which perfect competition was just a very special case….

To carry out such an analysis, you have to have a very general notion of what you mean by decentralization, because you cannot just point to the market system and say, well, that is decentralized. (That is as if someone were to ask you what is a mammal?, and you would point to a dog and say, a dog is a mammal. This, of course, does not help answer the question of whether an elephant is a mammal or not.) You must provide a general description of what would qualify as decentralized. If you don’t have a rigorous answer to that question, how can you know whether it is possible to decentralize in a given situation?

Having defined mechanisms in terms of information flow and decision-making authority, Hurwicz had achieved a generality more satisfying to him and others than vague talk of “central planning” or “competition.” He was able to move on to thinking systematically about the rules under which individual strategies would be pursued. One of these he called “the greed process.”

The basic idea of the greed process is that whatever the other side offers you take as a minimum and then you ask for more. Its origin was an old Polish Jewish anecdote about a young man who went to buy a suit. But he had never bought anything before. So his father told him, “Whatever they ask, always offer half.” So when he was asked, let us say, 100 zloty, he said 50 zloty. When the tailor went down to 80 zloty, he retorted 40 zloty. At the end, the tailor is really disgusted, wants to get rid of him, and tells him he can have the suit for free. The young man then retorts, “Can I have two pair of pants?” The “greed process” is somewhat similar in spirit.

The emphasis on models of information-processing lasted until the late 1960s.

But then I noticed that whenever I was asked to present some of my work, I would start by saying, “Of course the incentive problem is very important, but I will assume that people are angels and whatever you tell them to do, they will do.” Thus I was ignoring the incentive aspect and instead asking the following question, “Could we give the decision makers (say managers) the kind of instructions that, if followed, would make the economy run well?” But at some point I decided that since I know people are not angels, perhaps I should not completely ignore the incentive aspect.

At that stage I tried to see how one could formalize the incentive issue. Initially I was thinking of it in rather informal terms, somewhat along these lines: Let us say a country has some economic problem, for instance its balance of payments is in bad shape, as in pre-war Poland. What would it do? It might, say, introduce exchange controls (you must not export money, and so on). But what happens then? People figure out ways of exporting money: one has an uncle in London, others over-invoice or under-invoice… all the usual tricks. You could of course put them in jail or shoot them. But that is a distinct failure of economics, isn’t it? Because what economists should be able to do is to figure out a system that works without shooting people.

Thus did Hurwicz arrive at the idea for which he won the Nobel Prize last week, the concept of “incentive compatability,” in a paper circulated widely for several years and finally published in 1972, when Hurwicz was 55. He framed the argument in terms of a famous 1954 paper by Paul Samuelson on the nature of public goods, in which Samuelson stated that no decentralized system will work for their production (“Of course as soon as I see the word ‘decentralized,’ I am aroused, especially when such a strong negative assertion is made”); but clearly he was thinking of his native Poland as well:

What I meant by this was a system of rules designed in such a way that people would have an incentive to obey these rules. If the system is incentive-compatible, you do not have to threaten criminal punishment to get compliance. But this does not necessarily mean just maximizing profits. So the question is, “Could one design (a combination of taxes, subsidies, trading rules and whatnot) that would work as one would want it to work (that is, to achieve its goals) even without coercion or compulsion?”

The old debate about “socialist calculation” had yielded a new way of thinking, at once highly general and extremely detailed, about how incentives might be aligned in systems of all sorts, capitalist, socialist, military, religious, whatever. The effect on young researchers just entering the field in the late ’60s and early ’70s. was electric. “Hurwicz gave us the definitions, and we went to work,” says Jerry Green, of the Harvard Business School, co-author of a leading microeconomics text. As Hurwicz’ co-laureate Roger Myerson wrote last year, it had become clear within a decade that incentive constraints in the “social coordination problem” could be sorted into two basic varieties: there were information constraints, that made formal the various “adverse selection” processes that cropped up when people sought decentralized information; and there were strategic constraints, the formal version of various problems of “moral hazard” of controlling behavior that had been noticed and dealt with on a case-by-case basis over the years.

So much, then, for how it is that Leo Hurwicz became the oldest person ever to be awarded the Nobel Prize, and shared the honor with two men who, when he started working on what would become their problem, had not yet been born. He, too, started when he was a young man; he kept working until he got the set-up that he wanted. Indeed, he is still working. “But Who Will Guard the Guardians?,” a meditation on the ancient problem of trust and verification, was on the agenda when economists gathered in Minneapolis last April to celebrate his 90th birthday. He is one of those thinkers, like Arrow and Samuelson, who worked at a consistently high level over many decades.

We will come back to mechanism design in December, when the prizes are actually awarded to Hurwicz, Maskin and Myerson in Stockholm. The Royal Swedish Academy of Sciences, which only in 1969 began awarding a prize in economics (technically the “Bank of Sweden Prize in Economic Sciences in Memory of Alfred Nobel”), is still studying a dwindling backlog of significant work done in the ’50s by economists who are alive, leading their friends to joke that the Swedes engage in an annual game of “Beat the Reaper.” A painful case in point came in 1996, when 82-year-old William Vickrey shared the prize with 60-year-old James Mirrlees (for work in the engineering of mechanisms that anticipated the deepening interest in the principles of their design). Vickey was so pleased that, a few days after the announcement, he jumped in his car to drive overnight to a meeting — and died of a heart attack on the way. Potential laureates also sometimes die young: Jean-Jacques Laffont, who was a sure-fire eventual winner (and the man chosen in 1996 to lecture in Vickrey’s place), succumbed to cancer at 57 in 2004.

The outcome this year is highly satisfactory: Swedes 1, Reaper 0.

* * *

The roots of mechanism design in the socialist calculation debate were nicely laid out last week by Peter Boettke, of George Mason University, in an op-ed page feature in The Wall Street Journal, and in much more detail by Roger Meyerson in a lecture last year at the North American meetings of the Econometric Society. Still the best source on the formalization of economics is Philip Mirowksi’s Machine Dreams: Economics Becomes a Cyborg Science — a saucy, scholarly, brilliant, and ultimately thoroughly wigged-out narrative of events. E. Roy Weintraub’s How Economics Became a Mathematical Science offers a calmer, less salacious version of the story. Bruce Caldwell’s intellectual biography of Hayek is indispensable for understanding the Austrian’s contributions to the debate. You may have to go to a library to find George Feiwel’s remarkable 1987 contribution, Arrow and the Ascent of Modern Economic Theory, but it might be worth it. Useful, too, are these two magazines devoted to Minnesota economics.