Robert Wilson On Economic Engineering, Game Theory, & Auction Design

Robert Wilson is the Winner of the 2020 Nobel Memorial Prize in Economic Sciences. He is the Adams Distinguished Professor of Management Emeritus at Stanford University School of Business. Professor Wilson is the Co-Inventor of An Auction Format Which Shaped The Entire Modern U.S. Telecommunications Industry. He is Known As ‘The Dean Of Market Design’.

By Aiden Singh, March 10, 2025

Introduction

In 2020 Professor Robert Wilson was jointly awarded, with his former student Paul Milgrom, the Nobel Memorial Prize in Economic Sciences for ‘improvements to auction theory and inventions of new auction formats’.

In addition to Paul Milgrom, two other former doctoral students of Professor Wilson - Alvin E. Roth, and Bengt Holmstrom - have gone on to win Nobel Prizes in Economics.

Professor Wilson was born in Nebraska where, in his youth, he got his first exposure to auctions watching farm animals get sold at Saturday morning cattle auctions.

After completing his Doctorate in Business Administration at Harvard University, Dr. Wilson made his way to UCLA and then to Stanford University.

There he began scholarly work on game theory and auctions.

In 1993, the U.S. Federal Communications Commission (FCC) was looking to auction off the broadcast spectrum used for wireless communications.

It commissioned Professor Wilson and Professor Milgrom to design an auction format which would achieve the best outcomes.

They put forward a proposal for a new auction design called simultaneous multiple round auctions (SMRA) which was adopted, almost in full, by the FCC.

Professors Wilson and Milgrom thereby played a key role in shaping the entire U.S. telecommunications industry.

- - - - - - -

Theory & Practice

Aiden Singh: Your research can get quite theoretical and abstract. Yet it has also been applied to numerous practical applications.

Professor David Kreps has described you as ‘the founder of the “School of Economics as Engineering’’, by which he meant that he considers you a pioneer of the idea that economics can be used to design better institutions.

Did you always intend for your theoretical work to have practical applications or was that just a fortunate byproduct?

Robert Wilson: I guess I’ve always been averse to theory that was empty of application. Abstract theory is maybe OK if it’s generalizing something practical. But if it’s just theory looking for a practical application - a hammer looking for a nail - I’ve always been averse to that.

I like the theory to be motivated by an application. That is my whole approach to everything.

And with my students and courses, I tried to emphasize that.

A very substantial part of my work emanated from consulting relationships which involved practical problems in some industry. And then from that came ideas that I then tried to formalize in a theory. But there’s never been a time where I worked on a theory looking for an application.

- - - - - - -

Is Economics A Science?

Aiden Singh: There’s a long-running debate involving some of the greatest economic thinkers and going back centuries over what goals economists should set for themselves, whether economics should be viewed as a hard science, and what role mathematics should play in the discipline.

For example, Milton Friedman suggested economics be approached like a hard science while Friedrich Hayek argued economists would be mistaken to intimate the techniques used in the physical sciences.

Both figures were, like yourself, recipients of the Nobel Prize in Economics.

Does your vision of economics as engineering imply that economics should be viewed as something akin to a hard science?

Robert Wilson: Sort of.

I mean we’re not dealing here with immutable physical laws. I agree with Hayek that we [economists] are always dealing with people with different information, different motivations, and different resources. And there’s almost nothing you can do that can possibly anticipate all of the variety of human experience. So you have to take the point of view that there’s a huge amount of variability. And for the participants themselves, they usually face a lot of uncertainty as to what the future holds. So from that point of view, you can’t nail any of the social sciences down to be a science - they’re not predictable from principles to what the outcome will be. This is a case in which the ingredients - the human participants, the information they possess, their motives, and so on - we never know such things.

There are some kinds of problems that supersede that kind of variability so that they have application over a broad area.

So for example, auctions - my area of work - is a case of a market mechanism that is quite robust: getting a good outcome doesn’t depend a lot on what people’s beliefs and resources are.

On the other hand, you mentioned Friedman’s view that economics should be like a hard science. Well, I certainly think that mathematics and theory is good and relevant. It’s good and relevant in the way that a mechanical engineer would certainly have a lot of interest in theories from physics. If you’re going to design a bridge, you want to have theory inform your approach to this practical problem.

I think it’s sort of the same in the social sciences - it’s just that there’s so much more variability.

I mean, you could say that the bridge is designed by the mechanical engineer to anticipate all of the varieties of earthquakes, wind sheer, and so on. There’s some sort of parallel like that.

The term ‘economic engineering’, by the way, comes from Al Roth. He’s the one who interpreted this approach to economics as a sort of engineering approach. I never thought of it that way but I think its a pretty good description.

So you might say that if economics is engineering then it’s sort of connected to scientific methods, mathematics, and so on.

We haven’t mentioned the empirical aspects. It’s certainly true that the sciences depend a lot on observation. Auctions are ancient institutions that go back a couple thousand years. Apparently they were used in Greece about 2,200 years ago.

When I said that I think theory should grow out of applications, I’m thinking of it as an observational thing. You get involved in the practical problems in an industry, see the kinds of problems that they’re deal with, and then you try to build maybe a better mechanism or give them better strategies.

So I think of it as not exactly a laboratory science.

- - - - - - -

Thoughts On Experimental Economics

Robert Wilson: There was actually a Nobel prize awarded to Vernon Smith for his work conducting experiments with human subjects in a laboratory setting where you have a controlled experiment. I guess that always kind of bored me because they were such trivial problems being dealt with. He was sort of trying to verify that Adam Smith’s view was empirically relevant and that you can replicate demand and supply. But that’s sort of mundane and the bigger problems are not really addressed by that sort of thing.

There’s a whole field called experimental economics which deals with experiments in laboratory settings. I just never felt motivated to try to confirm the basic elements of economic theory - that people will work out a way that demand will equal supply, that there will be a clearing price, and so on. I think it’s maybe good as a methodological point to examine such things.

Vernon Smith’s work has these very simple settings. Charles Plott at Caltech has much more complex settings and then it becomes much more interesting - he sets up markets where bubbles develop and then crash. Now you start to see really interest phenomena in a dynamic setting. Those do interest me.

- - - - - - -

Thoughts On Behavioral Economics

Aiden Singh: You’ve discussed experimental economics, do you have any thoughts on behavioral economics?

Robert Wilson: Well, I’m a great believer in Amos Tversky and Daniel Kahneman - I like their work. And I think it’s very true that the human mind has these biases and distortions in the way it perceives things. So from a cognitive psychology point of view it’s really good to kind of document these tendencies.

In a larger setting, where people have a lot of resources to study problems and there’s a lot at stake, one speculates - and a lot of economics is based on this view - that people work really hard to try to overcome the anomalies in human information processing and reasoning.

So I think behavioural economics is really insightful for demonstrating what people who are trying to act rationally need to overcome.

- - - - - - -

Game Theory

Aiden Singh: Can you tell me how and when you developed your interest in game theory?

Robert Wilson: Howard Raiffa, my advising professor at Harvard, was the co-author of a 1956 book on the potential role of game theory in the social sciences. So I was sort of aware of game theory while at Harvard.

But I didn’t actually know very much about it.

I was always interested in a multi-person decision theory, and the natural methodology for studying that is game theory. But since I didn’t know much about game theory, the work I did early on in multi-person decision theory didn’t use game theory. I worked on, for example, trading votes in a legislature without using game theory.

During my doctoral studies, I also got interested in a case study of oil companies bidding where there was a huge disparity of information.

So I got to UCLA and took a job working part-time at RAND. And I continued at RAND part-time even after moving over to Stanford.

At RAND I got to interact with Lloyd Shapley, who shared a Nobel Prize on matching markets. But he actually worked in all aspects of game theory. And he was a wonderful guy to interact with.

And he sponsored a conference which had a big impact on me.

And I came away from that whole experience with a real interest in game theory.

Then I got to Stanford and had a student who did some brilliant work on game theory applied to auctions. His name was Armondo Ortega. He finished his degree and went back to Mexico and just disappeared; he became a consulting engineer and never followed up on this brilliant work he did.

But his work was very influential. He had a working paper which had a wide distribution - in part because I produced about 50 copies of his dissertation and sent them to everybody in the field who I knew would be interested.

So his work had a big impact.

So that’s when I started getting serious about game theory and auctions.

It goes back to my interest in a case that I'd seen written at Harvard by one of the doctoral students, where oil companies were bidding in situations where there was a huge disparity of information.

One company had far far superior information than the others. So my first attempt with that was kind of amateurish. But later, the second attempt, I produced a very fine elegant little article. That's the one that's really cited in the Nobel award.

So then from there on, I guess game theory just became my operational methodology.

- - - - - - -

Basic Types of Auctions

Single-Item Auctions

Aiden Singh: Your earliest exposure to auctions was the cattle auctions you watched as a child. These auctions were organized as what economists call ‘English auctions’, one of the most basic auction formats.

Can you describe for our audience some of the basic auction types?

Robert Wilson: So in the English auction, it could be that the auctioneer keeps asking for higher bids - the current high bid might be 300 and the auctioneer asks if anyone is willing to bid 320. And he looks out at the audience to see if somebody nods or holds up their panel or whatever.

Another format is where the price rises mechanically. There's a clock that just keeps going up. And you, as a bidder, are a candidate to win as long as your thumb is on the button in front of you. As soon as you lift your thumb off that button, you're out. And whoever's still there when everybody else has quit is the winner. So that's format is called an ascending clock auction.

Then there’s the Dutch auction format, which is like an English auction in reverse. You start with a high price. Nobody's willing to pay that amount, but the price comes down according to a clock. The first person to push their button wins.

The flower auctions in Holland are very famous because they go through many lots of flowers very quickly using this Dutch format. So in comes this truck of flowers. People get some short amount of time, like 30 seconds, to look at it. And then there’s a price which continually comes down. The first bidder to push the button stops the clock, and that's the price he pays.

There’s also an all-pay format - everybody pays upfront and whoever paid the most wins. So this is like lobbying. Organizations spend on lobbying activity trying to get a bill passed by Congress - trying to influence senators and representatives. So all of the participants in this process spend a lot of money. When all that's sort of done, Congress enacts some form of legislation and we find out who won.

So those are the basic simple auction formats for bidding on a single item.

- - - - - - -

Capturing Auction Dynamics Via Game Theory

Robert Wilson: It's all computerized now, but I once went to the floor of the Mercantile Exchange in Chicago. There were maybe 200 floor traders out there.

You’ve got dealers and brokers. And let's say there’s a sell order and they're looking for who's willing to bid the most for it. Then you’ll have maybe 30 bidders crowding around shouting with an oral outcry and then the broker or the dealer will decide which offer they'll accept.

So these outcry auctions are incredibly chaotic and noisy. You’ll have a guy start with an offer and all the traders start running across the floor.

Now it's computerized and very sedate. But at that time it was an unbelievable thing to see.

And the intensity of that competition, you can capture it: game theory is sort of predicting the intensity of that competition and how it's going to play out.

- - - - - - -

The Winner’s Curse

Aiden Singh: Often the various bidders in an auction will possess different information or will interpret the same information in different ways.

For example, bidders on cattle at an auction may hold differing views on what the price per pound of meat will be by the time it reaches the market.

Likewise, bidders for oil drilling rights may not know what the price of oil will be by the time the barrels reach the market. Or bidders may not know exactly how much oil is really contained in the deposit whose drilling rights are up for auction. Or perhaps both: oil companies bidding on drilling rights may not know how much oil is truly under the surface nor the what the price of a barrel of oil will be once it’s extracted.

You refined the theory of auctions for items with a common value - that is, items, such as oil deposits, whose value is uncertain before an auction but, in the end, is the same for everyone.

These situations can lead to an outcome economists refer to as the ‘winner's curse’. What is the winner’s curse and how does it affect bidders’ behavior?

Robert Wilson: So I said before that behavioral economics is insightful because it shows us what people who are trying to act rationally need to overcome.

Richard Thaler is a behavioral economist who has published a whole bunch of articles on the winner’s curse.

So take the case of oil companies bidding on drilling rights. There’s this structure under the ground and all the companies have is indirect evidence that about what's in it, maybe via seismographic surveys.

So there’s uncertainty.

And if you’re the company that wins the bid, the very fact that your bid won is telling you that your estimate of how much oil is actually in the ground is the highest of all the estimates. And this winning bid might be higher than the actual value of the oil.

And the result is that this produces very conservative bidding.

But there are actually formulas you can use as a bidder to correct for the winner’s curse.

I consulted for the oil industry, where the winner’s curse had a big impact on firms’ financial returns. Once the firms began to realize - partly through my efforts - that they were missing something, they were very quick to change. And changes pervaded the industry within just a few years.

So I think this is an example where there’s an important behavioural pheneomenon and what you find is that if people become well enough informed about that phenomenon and what strategies they can use to overcome it, they’ll do it.

- - - - - - -

The FCC Spectrum Auctions

Aiden Singh: In the 1990s the U.S. Federal Communications Commission (FCC) was looking to auction off the broadcast spectrum used for wireless communications.

It turned to you and Professor Milgrom to help design auctions that would achieve the best outcomes.

And the proposal you put forward was for a new auction design called simultaneous multiple-round auctions (SMRA).

Can you explain what challenges the FCC faced in designing an auction format for its spectrum sale and how your auction format helped overcome these challenges?

Robert Wilson: There were two basic problems. One was all the uncertainty the bidders face and the role of private information - we talked, for example, about the winner’s curse.

And the other was that the auction was dealing with complementarities. So, for example, a company might want to have a license in Northern California and one for Southern California. But if it buys one for Northern California and then another company comes in and bids up the price for the Southern California license, the already acquired Northern California license is now worth much less because its less optimal not to be able to service the whole region.

This is called the ‘exposure problem’. And firms are extremely sensitive to it.

So bidders were stuck in this terrible problem: they needed a a coherent portfolio of licenses, but how do they assemble a package like that?

So the plan we proposed was one with a dynamic process where they know current high bids and change their bids.

So you have multiple licenses. When we open the auction, they're all available for bid. So, if there are 10 of them, then the bidding for each of the 10 licenses opens simultaneously. And the auction is not over until there's no new bid for any of those 10 licenses.

We had one round in the morning and one in the afternoon.

And we would announce the current high bid for each license: for example, the current high bid for this license is $30,000,000, the current high bid for this license is $42,000,000 and so on.

In the next round, if you’re not the current high bidder for a particular license, you could put in a new bid and try to become the high bidder for it.

And we had the rule that the new high bid must be at least 5% higher than the current high bid.

And at the end of the auction, the current high bidders are the ones who are going to receive those licenses.

So it's a dynamic process that enables them to assemble a package based upon the prices they can see. And it also dissipates the role of private information so that their informational rents are reduced, and that means you get a more efficient outcome.

Two big things were our closing rule and our activity rule. And these are both due to Paul Milgram, who is a genius.

So what was the closing rule? The closing rule was that none of these bids end until they all end. Every license is open to new bids until there are no new bids on any license.

Why is that? It's so that when the firms are assembling their package, they actually see the prices for each license in front of them.

But this can be compromised by the fact that a firm could use a ‘snake in the grass’ strategy of hiding its intentions. So the firm doesn’t tell you that it’s going to be bidding for the Southern California license, and then near the end, it'll drive the price up.

So you have to force them to reveal what licenses they are interested in and what their intentions are.

So we have an activity rule that says you're initially qualified to bid on a certain number of what we call ‘megahertz pops’ - the megahertz of spectrum serving a population of such and such. You can bid on certain millions of megahertz pops. And there was a first phase where you have to bid on at least a third of your entitlement or lose it. And then there was a second phase where you may have to bid on at least 2/3rds or lose it. And then at the very end, you have to use all of it or lose it. That's was the initial activity rule.

Later on much better activity rules were developed based on the principle of revealed preference. Here your bidding behavior has to be consistent with some single sort of utility function for outcomes, some valuation of outcome: it must be that if we see how you're bidding, we can say it is consistent with the following kinds of preferences and information. So that's a more complicated approach to the subject.

So, I began this by saying that the overall problem was the huge uncertainty bidders faced and that you needed to suppress those effects.

So this is the point about game theory: it shows you where there are problems and enables you to use basic design principles to deal with them.

Aiden Singh: The FCC adopted, almost in its entirety, the proposal you and Prof. Milgrom put forward for using an SMRA.

Essentially the auction design you and Professor Milgrom created shaped the entire modern U.S. telecommunications industry.

Subsequently, Canada, Finland, Germany, India, Norway, Poland, Spain, Sweden, and the U.K. have all gone on to deploy different versions of SMRAs in auctioning off their broadcast spectrum. So you and Professor Milgrom have really shaped the telecommunications industry globally.

- - - - - - -

On Teaching Three Nobel Laureates

Aiden Singh: Not only have you won a Nobel, but three of your former students have also gone on to win one.

What’s the secret to producing successful students?

Robert Wilson: All I can say is that they're extremely bright guys. How they happened to come to me was almost accidental in each case. So having brilliant students, that's part of the formula.

But part the other part of the formula is that none of them had a traditional economics training.

Back in, say, 1970 , how did people think about economics? Supply equals demand and where they cross, that's what the price is. Economics had blinded itself by thinking everything could be thought through by this paradigm. We call it the Walrasian paradigm; Walrasian economics.

Nobody was thinking about how prices are formed, who runs the markets, what the organization is, what the incentives are, and so on. They had this magical thinking that the market will clear and we don't need to ask further.

But, Al Roth, for example, is in operations research. Dave Kreps - who was not my student - was from operations research. These are guys with a background in applied math. Bengt Holmstrom had also been in operations research, but he transferred into our program.

Paul Milgram had been an actuary. For 7 years, he was an actuary in an insurance company. So he had all this training in mathematical methods, probability, and statistics in actuarial science.

They weren't being trained in economics. They weren't learning this paradigm that demand equals supply and so on and so forth.

So they came to the subject fresh with a different perspective.

And they were thinking about: How did we get to these prices? Who's running the market? How is the market organized? Different people have different information, so how does that get embedded in prices?

So, they asked different kinds of questions, particularly questions about information. Because a lot of this is all about, in some of huge sense, the role of information, people's private information. Private information such as the estimates of how much oil might be located in a oil field. And this goes back to our earlier conversation about Hayek.

So they focused on the very salient role of information. And that had such a big impact because economic theory subsequently was transformed by this point of view.

- - - - - - -

Advice For Young Economists

Aiden Singh: Do you have any advice for young economists today who are looking to make a name for themselves, maybe win a Nobel Prize of their own someday?

Robert Wilson: They should learn something very substantive about the real world. They should try to identify what's going on or some big problems that aren't being solved and try to build a theory about that.

Aiden Singh: If you were starting over today, your career as a economics professor, scholar, researcher, what problems would you be working on today?

Robert Wilson: How do you sustain cooperation over time through reciprocity?

And unfortunately game theory, in the naive sense, says well you might get a cooperative outcome, but you might not. So there's a very vague basic prediction that everything's possible.

But in my own work, I've studied the effect of bounded recall where people could only remember so much. And I've shown numerically that you didn't get into a repeated prisoner's dilemma - you get cooperation. But I wish I had a theory.

I've been trying to prove a theorem for a dozen years.

So that's the problem I'd work on. It’s the problem I am working on.

- - - - - - -

The Nobel Prize

Aiden Singh: Who is somebody who hasn't won a Nobel Prize yet who you think deserves to?

Robert Wilson: Well the first thought that comes to mind is Dave Kreps, my colleague. I mean, there's so many prominent people in my field who are deserving.

Aiden Singh: What did you think of the most recent award?

Robert Wilson: The award to Daron Acemoglu is a very interesting departure because his work deals with factors affecting entire societies like the legal structure, or commercial arrangements for trade, or the political constitution, and so on.

Why is it that the former British colonies have done so much better than the former French colonies? Questions of that kind.

It doesn't involve technical theory but it’s work that's very insightful and tells us important things.

Aiden Singh: I'm very tempted to ask who is somebody that’s received a Nobel Prize that you think didn't deserve it? But I won’t make you answer that question.

Robert Wilson: Oh I have my opinions on a couple of those. I think I’ll keep that to myself.

Both: (Laugh)

Aiden Singh: Yeah that’s probably best - let’s not get you in any trouble. But it would be funny to hear.

Robert Wilson: My wife knows the answer.

Aiden Singh: Oh, really? I’ll have to ask her.

Robert Wilson: The thing is that she is so unconnected to economics that probably, even though I told her the names, she doesn't remember.

Aiden Singh: That’s a shame - so I guess nobody will ever know who you think didn’t deserve the Nobel.

Both: (Laugh)

——————