Interview with the Laureates in Economics Eric S. Maskin and Roger B. Myerson, 6 December 2007. The interviewer is Adam Smith, Editor-in-Chief of Nobelprize.org.

Eric Maskin, Roger Myerson, welcome to this interview with Nobelprize.org and to Stockholm. You’re co-recipients, together with Leo Hurwicz of the 2007 Sveriges Riksbank Prize in Economic Sciences in honour of the memory of Alfred Nobel. You’ve been awarded it for your work contributing to a mechanism design theory which essentially provides an analytical framework for analysing institutions making collective decisions?

Eric S. Maskin: Right.

Roger B. Myerson: Well put.

I think I’ve just paraphrased your words, I’m sure. We should perhaps start by mentioning Leo who can’t be here because of health reasons. But he’s the founder of the field, he invented the mechanism design theory and he’s also the oldest ever Laureate at 90 years old. He started this field half a century ago in a very different climate. Then people were considering the differences or trying to compare economic institutions in capitalist and communist situations for instance. Was that a very different time to be doing economics?

Roger B. Myerson: I don’t think 1972 was that different a climate. I think he was responding. Much of his career was in response to this early 20th century capitalist versus socialist debate where it can’t be a recognition that all of our wonderful results about the efficiency of the free market didn’t necessarily mean that a socialist command economy was bad. In fact they didn’t say anything about the socialist command economy and that economic models simply weren’t broad enough.

Leo Hurwicz was, I think, largely responding to this and Frederik Hayek’s observation that the debate between capitalism and socialism in an article of 1945 by Hayek argued that the real distinction had to be on understanding the economy. The market is a mechanism for communicating information. People who need things. People who have resources. The supply and demand are communicated in the market and Hayek said the virtues of the free market, that he said in his opinion, that socialism couldn’t imitate, were virtues of communication and this is what needed to be studied. Hayek also said that of all the economists of his time, the people who were worst about this were the mathematical economists.

You are talking to two mathematical economists and in particular Leo Hurwicz was a mathematical economist from the start. Leo understood, as I think we all do correctly, that whatever mathematical or non-mathematical economist may or may not have understood in 1945, if you think your models are not general enough to understand the broader phenomena, mathematics which is versatile and you know it’s the vigour and versatility of mathematics is a good place to go. It’s a useful tool. If economic models were not broad enough to be able to compare capitalism and socialism, mathematical models were worth pursuing and this is the challenge that Leo took.

Eric S. Maskin: Just to amplify what Roger has been saying. The debate between von Hayek and von Mises on the one hand and Lange and Learner on the other about capitalism versus a planned economy was very important and very interesting but also very frustrating in a sense because none of the terms they were using, decentralisation, a command economy. For that matter a market economy were defined precisely enough so that the debate could be assessed and actually Leo got into his great work in part because he was trying to understand just what these terms might mean. The word decentralisation sounds as though it has an obvious meaning, but when you think about it for a moment, you see that it’s really quite a subtle concept. He spent a good many years just puzzling over what decentralisation means.

And then framing it in some kind of mathematical terms?

Eric S. Maskin: The advantage of mathematics is that it’s quite ambiguous. It gives a precision that sometimes ordinary language doesn’t provide.

Roger B. Myerson: Put it another way if you want. I would argue that in some of this debate, you can find Hayek saying things that we can now say in our models. But because Hayek couldn’t say it in formal economic models, he’d tell them to get lost. If you don’t have it in an analytical framework, the words fly by. When you have a systematic framework I think the real breakthrough was in Leo’s 1972 paper when he introduced incentives to communicate. Suddenly then things changed.

That’s the idea that the market is not dominated just by the amount of resource around, it’s also dominated by the incentives that people have to use it, to work with it, is that right?

Eric S. Maskin: Incentives are a critical part, not just of markets, but also of alternatives.

Roger B. Myerson: Exactly. They’re both.

Eric S. Maskin: And to do a proper comparison of a market institution versus a non-market institution one has to take into account the incentives of the individuals in that institution to behave as we would want them to behave.

Roger B. Myerson: In a nutshell, I think in economic theory, the models that people taught their students to work with, in trying to help satisfy the desires and needs of human beings, increased human welfare, the economic problem is constrained by their own infinite resources. And resource constraints that we don’t have enough to make, enough oil to do everything we want in the world. Resource constraints were the essence of the economic problem. Before now the economic problem, as we understand it in our models, and theory also involves incentive constraints, the incentive to communicate information honestly. The incentive to work hard and to exert effort when nobody’s watching. It is these incentive constraints when information is imperfect are the essence of daily life. But now we’ve added to resource constraints in the economic problem we also have incentive constraints, and economists understand that too, not just ordinary people.

Ok, we’ll come back to mechanism design theory a little bit more later, but was Leo Hurwicz’s work instrumental in making you move into this field? What was your path to the field yourselves?

Eric S. Maskin: In my own case, it was a direct influence because I started out in pure mathematics which I still find a very beautiful subject. But wandered almost by accident into a course that one of Leo’s close friends, Kenneth Arrow, was teaching on just this sort of stuff. Ken Arrow called it information economics, but a large chunk of the course was devoted to what we’ve now called mechanism design. I thought this was great stuff because it combined the rigour and clarity of mathematics with a real attention to important, social and economic issues, and I thought this was an unbeatable combination. So, I can truthfully say that Leo Hurwicz changed my life.

Right. So it was a way of capturing both the practical application and the beauty of the theory of your mathematics?

Eric S. Maskin: Right. As it was presented in that course, that it was theoretical. That is that we did not look at particular applications to which the theory might be directed but it was obvious that there would be such applications and in fact, in the years since we’ve seen quite a few of them.

Roger B. Myerson: Both of us are theorists, there’s no question. We start from looking for conceptual frameworks that can help people to understand applied problems better. While Eric in graduate school was reading Hurwicz, I was reading more John Harsanyi. We were at graduate school together, but I’m trying to remember when I first met Leo Hurwicz and I’m sure it was within the first year of being an assistant professor after I finished. But I was at the same time reading Harsanyi who was trying to develop the most general framework for the analysis of competitive behaviour in competitive situations, in games. Harsanyi had this great breakthrough in how to model games where people have different information. So both cases, Hurwicz and Harsanyi you were both working on information.

I think we when we got these frameworks together, we realised Harsanyi was developing tools for analysis of a given game, and Hurwicz was giving us a framework for how do you design a game. I think that stream came together in the late 1970s where two of the groups of then young economists who jumped into the breach with the game theoretic tools in place, and Hurwicz’s mechanism design in place and at last economics was prepared to really deal systematically and analytically with interactions between people who have different information, the creation of institutions where people have different information.

So you actually felt aware of the fact that there was suddenly new tools available to work with?

Eric S. Maskin: Well, not having been around in the days without these tools, I’m not sure that we appreciated, at least I didn’t appreciate it at the time. The extent to which all of this was new. I certainly do now. I realise now, as I didn’t then, that this was a very special time. Maybe you knew at the time then?

Roger B. Myerson: I got into the game theory at the end, I think you may have been in the course, Howard Raiffa taught a wonderful decision analysis course.

Eric S. Maskin: I took that.

Roger B. Myerson: It was analysis of rational decision making. He made it very practical. You wake up in the morning and you have a utility function and how would you assess it? But at the end of course, he said supposing there are two people interacting and they each have utility functions that they’re maximising. That’s game theory and there isn’t very much known about game theory. Now by the way we’re about to celebrate the 50th anniversary of Howard Raiffa’s book with Duncan Luce from 1957. It was the best book we had available but when he said …

Eric S. Maskin: It’s still a very good book.

Roger B. Myerson: It is a good book, but when he said, Actually not enough is known about this, about how to analyse games, how can life go on? I’m sorry, how can we do anything if we don’t understand how to analyse these structures. I had a great sense that fundamental work was needed, and the truth is I was lucky to have been born at a time that got me in graduate school at a time when the breakthroughs were ready. There’s still more work to be done on fundamental game theory, but we’ve made a lot of progress on the tools.

Eric S. Maskin: The landscape has definitely changed. When I started teaching in 1977, my first job was at MIT. The MIT economics department had never had a game theory course before. I taught the first game theory course that the department had ever offered. Now it’s inconceivable that any department in the world worth its salt shouldn’t, not only must these departments offer game theory, but game theory has to be a central part of the curriculum, so you can see how over the years the landscape has changed with it. But I didn’t appreciate at the time that that was going to happen.

Roger B. Myerson: May I amplify that with one? At my 25th college reunion, the president of Harvard University when we came back spoke about the new humanities centre that the university was investing in and he said that Harvard took humanities seriously and not just feels like for example game theory. When we were students there were humanities courses taught at Harvard University. There were no game theory courses taught. It’s nice to … I felt a certain vindication. I thought it was important too. Humanities is important also. I want to be very clear about that.

OK. You came out of the same milieu, you were students together and then you separated and went your own ways. But then reunited as collaborators but presumably also competitors?

Roger B. Myerson: Oh yes, absolutely. Coming up with the same ideas for instance. I just discovered, preparing for this, and looking at Eric’s vitae I realised there’s something I want to do on the soft budget constraint and Eric’s already written the great paper on the subject. There’s more work to be done of course but we have three decades now of history of coming back to the same ideas and that’s good.

Eric S. Maskin: One instance where some collaboration might have occurred early was on auctions. It turned out that at the same time Roger was writing his now universally acknowledged seminal paper on the design of auctions. John Riley and I were thinking along very similar lines. Would have been nice if we could have joined forces.

Roger B. Myerson: Yes, yes.

Eric S. Maskin: Early on. Competition is a great thing, but collaboration is perhaps even better.

Roger B. Myerson: I worked for 25 years at Northwestern. I was very influenced by my colleagues and I really thought the way I framed it was very much a Chicago Northwestern approach. But I was approaching it the same way as Eric. We must have got a lot out of school together or great minds just think alike, but whatever. We have one joint paper, and as I’ve mentioned to you before I owe my Erdős number to that paper.

That’s the source of the Erdős number.

Roger B. Myerson: Yes.

So this is your collaborative distance from Erdős?

Roger B. Myerson: Yes.

OK.

Roger B. Myerson: He’s the close runner up.

And you’re a 2?

Eric S. Maskin: I’m a 2 because Peter Fishburn actually wrote a paper with Erdős and I wrote a paper with Peter.

I read somewhere that there’s a baseball player who signed a baseball at the same doctoral ceremony that Erdős got a honorary doctorate at Emory University.

Roger B. Myerson: Does that make him a collaborator?

That’s the question. Does he have an Erdős number of 1 or not?

Eric S. Maskin: I don’t know. Might need a referee.

Probably not. Let’s think about the mechanism and design theory for a while. It’s something that actually touches peoples’ daily lives in an enormous way, because it’s used for so many decision-making processes that affect people. Auctions being one that’s already been touched on. Would you each like to choose an example of where it’s been applied that would help to illustrate its far-reaching impact?

Eric S. Maskin: Shall I go first? One current example is traffic congestion. I think this an example of what most people can relate to, at least city dwellers. Many cities, New York among them, have a problem with too many cars at the same time, and the question is how to deal with that problem. There are many alternative conceivable policies. One is to ban traffic outright, at least from certain parts of the city. Another is to raise tolls leading into the city. Subway, tunnel tolls or bridge tolls. A third possibility – and this has been pursued here in Stockholm and a number of Australian cities – is to equip cars with transponders and keep a record of where the cars go in the course of a day and charge them accordingly. Not only as to where they go but also the time of day.

A fourth policy is to invest in public transportation, and there are many other possibilities as well. What mechanism design enables us to do, it gives us the tools to evaluate these alternative policies and to see which one of them or combination – it’s usually some combination – is best suited for the particular circumstances of a given city. It gives us an analytic framework to evaluate policy.

How does that work, how do you apply that analytic framework? I’m sure there are obviously many levels of detail to the answer to that question, but one can see how one could work out the income from for instance a toll system, but how do you use mechanism design theory here?

Eric S. Maskin: Income is one consideration, but you also want to understand how the toll system is going to affect peoples’ behaviour. If you change the tolls you’re going to presumably change peoples’ driving behaviour, so mechanism design gives you a method for evaluating the sensitivity of that behaviour to the policing instruments.

In general is it used when people are trying to make these decisions? Are cities such as Stockholm do you think using mechanism design theory when they make their decisions?

Eric S. Maskin: Typically, at least some cities. I can’t speak about the Stockholm case but Melbourne, Australia, for example did seek out the advice of well-trained economists when designing their system, so, yes, I think that’s an example of mechanism design in action.

Excellent, thank you.

Roger B. Myerson: My favourite example to think about is mediation and dispute resolution. When a mediator is going back and forth between two parties and the media is essentially creating a game by which they provide information to the mediator and a deal is put together that has certain characteristics that depend on what they say at various stages or there’s some positive probability of failure of the negotiation. That process should be understood as a mechanism design possibly. The mediators look in the middle. I think there’s some awareness of this. At Northwestern I worked with people in dispute resolution from a variety of methodologies and I think there’s still more that needs to be communicated. It’s very popular to study bargaining, how to negotiate something, but I think from my perspective, pedagogically, I think the study of bargaining should begin with a study of mediation and the theory of mediation begins with the concepts of mechanism design.

On the one hand you need to find out peoples’ trade-offs in order to find what they call win-win opportunities. What am I willing to give up that doesn’t mean so much to me but means a lot to me in exchange for something that you’re willing to give up that means a lot to me. But also, and that’s sort of the old-fashioned economics, the trade-off between two utility curves in an Edgeworth box diagram. Then there’s the problem of getting people not to pretend that they need everything and for that some positive probability of disagreement may be of the essence of the matter, and a mediator who doesn’t understand that is going to make more trouble than help sometimes. I think mediation in dispute resolution is a fundamental problem and it is an area where the ideas of mechanism design have been studied and have been applied to some extent but could be much more.

That seems to open up the ground for an extension of mechanism design theory into political science and legal science or legal theory, and that’s already going on?

Roger B. Myerson: I think the extension of economics methodology with game theory and mechanism design has enabled the boundaries between political science and economics which were academically artificial and to some extent not so productive. Because after all social problems are both in the market and in politics at many times. Those boundaries have been lowered, people can apply ideas from both sides and I think that this is very helpful. Much of what we do in political modelling is not necessarily mechanism design because it’s not necessarily having the kind of universal perspective. It’s often of the form as you change the constitutional structure, you’re changing the rules of the game the politicians play and how will that change political behaviour? That is the way we should approach any specific reform.

The question then is what would be an optimally designed constitution? That’s a very good question. I don’t think we’re very close to that. The most important part of mechanism design I see at this point is an understanding within politics itself, a political leader is creating an organisation, he’s got a faction of supporters, he’s got people he works with, and they all need to trust each other to make the campaign go for high office. Understanding the nature of politics by understanding the political leaders within their factions are solving a mechanism design problem. That I think we have something useful to say.

Dwelling a little bit on the artificial academic separation between political science and economic science, is the separation very exact in university departments? Is a political science department and an economics department, are they very different entities still?

Eric S. Maskin: I’m afraid in many universities they are. In fact, one of my complaints about even such a great university as Harvard – where I was for many years – is that people are very much separated by departments, that there’s not enough opportunity to talk across disciplinary lines. For that matter there are many political science departments which themselves are divided. There are the people within the political science departments who take what’s known as the rational choice route which is relatively close to the economics approach. But then are the institutionalists who eschew that kind of analysis and then there are the political theorists who are largely philosophers. Some political science departments are actually three little departments in one and there may not be very much communication amongst the three.

Roger B. Myerson: Political science departments typically don’t have a unifying analytical methodology the way that economics department do, so they do many things. The rational choice approach, let me say for a moment, I don’t think I have not lived my life in a perfectly rational manner, to the standards that we normally assume in economic models so I don’t see why I should think anybody else is perfectly rational, but the truth is for analysis of institutions, it helps to assume. You want to ask if you’re going to reform the structure of a constitution for example or change the structure of promotions and pay within a firm. You’re changing the rules of the game and it’s very useful to assume that people are rational and selfish, I should say, in the analysis of how the institutional structure will affect individual behaviour.

I think the fact that much of political science, they don’t invest in. There’s been a tradition of most political scientists not investing in the calculus of rational choice analysis. That has meant that analysis of alternative institutional structures. What would happen suppose we designed a constitution that combines presidential and parliamentary aspects and will we get something that’s the best of both worlds or will we get the worst of both worlds? For that we need some theoretical analysis and I think rational choice is the best or almost an essential tool for such. There are many ways of approaching truth in life. Political science departments have typically not gone into comparison of political institutions sometimes in the way that they could with a more rational choice analysis.

There are two ways of getting around that and both are happening. One is political science departments in recent decades have started more and more to have rational choice analysis, and the other is economics departments have promoted professors whose research was using economics, rational choice methodology, to study political institutions. That is to say the questions migrate to the economics department. There’s a third approach which is people in economics and political science departments could actually talk to each other. That does happen too. But I think in Europe in particular there’s been a very good expansion of interest in economics departments of studying questions that are about the nature of politics and political institutions.

I think on the radical idea of economists and political scientists getting together and talking we might draw the interview to a close since the tapes are ending. But thank you very much indeed for taking the time to speak to us and I wish you a very pleasant Nobel Week in Stockholm.

Eric S. Maskin: Thank you so much.

Roger B. Myerson: Thank you.

Watch the interview

Did you find any typos in this text? We would appreciate your assistance in identifying any errors and to let us know. Thank you for taking the time to report the errors by sending us an e-mail.