The Experimental Economist

Nobel laureate Vernon Smith takes markets places they've never been before.


When Reason interviewed George Mason University economist Vernon L. Smith back in May, one of the first questions we put to him was, "You're always mentioned as a short-lister for the Nobel Prize in economics. What do you think of that?" The 75-year-old Smith, the founding father of experimental economics, responded with a hearty laugh. "It's nice, but whenever I hear that, I always ask, 'Do any of the people saying that have a vote on the selection committee?'" It turns out that they did. Today, the Royal Swedish Academy of Sciences awarded Smith the Nobel "for having established laboratory experiments as a tool in empirical economic analysis, especially in the study of alternative market mechanisms."

What do experimental economists do? "We take propositions of economic theory and we test them with real people in controlled settings," explains Smith, who's also a research scholar in the Interdisciplinary Center for Economic Science and a fellow of the Mercatus Center, both affiliated with George Mason. "Take the idea that there's a tendency for markets to achieve price levels at which they'll clear-at which buyers and sellers agree with one another and no money's left on the table. I'll run experiments where I motivate buyers to buy low and sellers to sell high." Smith helps keep living, breathing human beings at the center of a discipline notorious for transforming flesh and blood into impersonal theoretical equations.

"We're less interested in what people think than in what they actually do in specific situations," says Smith. "The thing that's not very explicit in much of economics is what the rules of trading are and how they affect outcomes. Experimental economics asks how the performance of a market is influenced by its rules."

Over the past 50 years, Smith has overseen thousands of experiments, with everyone from school kids to business tycoons to congressional staffers. He also has been instrumental in developing policies that create true markets where they've never existed, such as in electricity sales.

Smith has found that even when people have no clear idea of how or why markets work, they nonetheless tend to be both savvy and surprisingly generous. "I think we're born traders," he says. "We're social animals, very much into social exchange. This propensity of humans is very likely what led ultimately to trade and markets that produce wealth. The benefits of market exchange are easy to see in personal interactions, where you do something for me and I do something for you. Out there in 'markets,' though, they're not always clear. If the price goes up, the oil companies get more money and I have less. That's the average person's perception and experience. Experimental economics helps put a human face on markets."

Smith's professional odyssey began at Caltech, where he studied electrical engineering before getting a master's degree in economics at the University of Kansas in 1952 and a Ph.D. at Harvard in 1955. Prior to joining George Mason's faculty in 2001, he taught at the University of Arizona, Purdue, Brown, and the University of Massachusetts. En route to winning the Nobel, he served as president of the Public Choice Society and the Economic Science Society, received countless awards, and published widely (Cambridge University Press has published two volumes of his papers in experimental economics).

Smith's ideological journey has been no less wide-ranging. He started out as an avowed socialist who believed that the good society was one in which a few wise men made most social, economic, and political decisions. Over the years, he gravitated toward a libertarian position that holds that individuals should be as free as possible to make their own tradeoffs. "Whether we're talking about politics or economics, or even social interaction," says Smith, "the best systems maximize the freedom of the individual, subject to the constraint of others in the system."

Contributing Editor Mike Lynch and Editor-in-Chief Nick Gillespie interviewed Smith at his George Mason office and a nearby restaurant in late May. This interview will appear in the December issue of Reason.

Reason: Experimental economics stresses the importance of institutions-the rules and regulations of markets and systems of exchange. Yet people often talk about the "free" market and property rights as if they exist in nature the same way a mountain or lake does. Your work suggests that these rules are created for specific people in specific situations.

Vernon Smith: I think a lot of market conventions and property rights come from norms that emerge through people's interactions. Often the state comes along later and codifies them.

Robert Ellickson's book Order Without Law (1991) documents this. He went and looked at what fence law is in Shasta County, California. The law makes it clear that you're liable for the damage your cattle does, so you have an incentive to build fences. But that's not quite the way it works. The reality is that people share fences. If you're a guy who gets known in the community as careless, as someone who doesn't keep your fences up and whose cattle keep getting out, your neighbors are much harder on you than if it's just a one-time-only mistake. Most of these disputes don't even go to court. There's pressure on you to pay up and settle out of court at a more expensive level.

It's interesting because people work out exchange systems that are not necessarily related to formal law. If you read [economist] F.A. Hayek, you know that the early lawgivers were not people who made law. They just wrote down the existing practices. This is what people are doing in Shasta County. It's "discovered law." "Made law" starts to come in later.

Of course, not every transaction is local or face-to-face. That's why you need more formal markets and property rights. What's the old saying? "Everything for a friend, nothing for an enemy, and the law for strangers." Property rights and markets help to extend the gains from trade to strangers by ensuring payment or ensuring delivery.

Reason: People are more likely to cheat people who live far away.

Smith: Definitely. But even then, perhaps not as much as we'd first think. Consider the Internet. What's remarkable about Internet markets is not that some people don't deliver or pay, but that it's so few. There's practically no way to enforce a lot of the commercial activity. The amazing thing about it is how much of what goes on does so on the basis of trust.

Reason: What are the policy goals of experimental economics?

Smith: We know markets are efficient. We'd like to see if we can come up with market mechanisms that can be applied where they've never been used before. In electric power, for example, or in landing and takeoff slots in airports.

My colleagues and I have developed techniques to design market mechanisms that are likely to work in the real world. But we don't trust ourselves without doing experiments. The experiments are the means by which we test our knowledge. We use the laboratory to make our mistakes at low cost-and we make plenty of them. We always learn stuff and end up making changes to our model institutions, to our rules, and to our payoffs.

The next step is to bring in the people who will actually be using the system. They put design elements in, and then we run experiments with them. When they're comfortable with it, we go out in the world with it.

Here's an example dealing with electric power. Let's say you're creating a market for wholesale electricity. You have buyers and sellers of power at different nodes in an electric power network. People put in location-specific asking prices to sell power; they've got to be location-specific because power grids leak, and depending on where you are on the network, your costs will be different. If you're a wholesale buyer, then you put in a bid schedule at which you're willing to pay for power to be delivered to your node. These are all computer-assisted markets. A computer essentially takes all the asks, all the bids, and all the location costs and it maximizes the gains from trade. It does that by finding prices that clear the markets so there isn't any money left on the table.

We did experiments for this sort of system in Australia in 1993 and again in '96. We ran experiments and had a seminar a day for two weeks, and the participants got really good at it. Australia ended up deregulating its electric power industry, and I think they did a pretty good job.

What's interesting is that it's still not all privatized-that varies from state to state. Victoria, for instance, took all its generation assets and sold them off to private industry. In New South Wales, they're still held by public entities. But the important thing is that all power providers have got to get their money in the market. And all new capacity is completely private in Australia.

Reason: So experimental economics can be a tool to show regulators and industry types how truly price-sensitive markets could and would work?

Smith: It is a tool. In 1995 the Progress and Freedom Foundation sponsored a series of workshops at the University of Arizona, where I was at the time. We brought in 25 executives who were high up in their organizations-Florida Power, Duke, Ohio Edison, Mohawk, Pacific Gas & Electric, Southern California Edison, and so on.

They participated in experiments, and they could see and experience what it was like to make a real market in power. One of the things we emphasized was the importance of demand-side bidding, where wholesalers are able to interrupt power flow to some of their customers to try and keep prices down. We had a client in Ohio that took demand-side bidding very seriously. In fact, they're having customers save money by agreeing to undergo voluntary power interruptions in certain circumstances. There are gadgets available that allow individual users to figure out what appliances they want to keep running at different price levels. Customers can decide, "If the price gets up above a certain place, cut off lower-priority uses-air conditioners, hot water heaters, etc." If you do that even 15 minutes an hour, you can really cut peak demand.

The main problem in electric power is that there's almost no capability of interrupting demand in a selective way below the substation level. If you're in an elevator, if you left the porch light on in the daytime, if you've got a toaster running-those are all the same priority in terms of getting electricity in the current system. There's no way to pick and choose which gets cut off when demand surges, prices spike, and supply gets tight.

Both government and investor-based utilities have this huge supply-side bias: They provide whatever electric power people want all the time at a single, constant price. The reality is that the cost to produce and consume electricity fluctuates all the time because of changes in supply and demand and other conditions. It's like anything else, but it's hard to realize that without a visible market.

Reason: It's pretty easy to see how people will respond to observable fluctuations in price: I'll turn off my soothing water fountain at 25 cents a kilowatt, but I'll keep my iron lung going at $100 a kilowatt. But it's easier said than done to create a functioning market, isn't it?

California's famously botched restructuring of its power market underscores that. Not only did power suppliers help set up the market rules to begin with, they must have known that they would be bailed out if they could make a credible case that they couldn't supply power at the "deregulated" price. The state says sellers manipulated the market in all sorts of ways.

Smith: Sure, they're going to try to game it. That's why it's important to have the right set of rules in place. Nobody knows where exactly demand is going to come out, and sellers have an interest in trying to make that price come out as high as possible.

But the main problem is that inability to interrupt demand. You're talking about an industry in which the way they think about their problem and the way the regulators think about the problem is that all demand has to be served all the time. No one should ever lose their lights, and a fair price is a constant price over time. You always meet all demand. Well, it's impossible. You can't always meet all demand. You can't meet it in storms. You can't meet it when supplies are tight.

A big barrier to change is the local electric wires, which are franchise monopolies. For nearly 100 years, companies have been tying in the sale of energy with the wires. There's no reason the provider of the wires has to be the provider of the energy. Do you remember when Ma Bell wouldn't let you connect any telephone unless she produced it? They argued that they had to protect the "integrity" of the network: We just can't let anyone go in and connect whatever they want! It's silly. Why would anyone produce a telephone that wasn't compatible with the system?

That's all changed for phone service providers, but not for companies in the electric power industry. They're saying, "Why should we give access to our wires to alternative energy providers, especially ones who want to put in an interrupt device?"

What we need is a system where entrepreneurial types can come in and compete and try out different things, including the demand-interrupt system. Some will lose money and go broke. Some will hit it big. That's what's happening in telephones, and it's not happening in electric power. It won't happen until those barriers are removed.

Reason: What are other areas where experimental economics is playing a role?

Smith: We're doing work on creating a market for the exchange of landing and takeoff slots at airports. In normal circumstances, those rights have been fully allocated among the airlines at a given airport. But let's say a bad weather front moves in, so there's a ground delay. They've been doing maybe 60 landings and takeoffs per hour, but now they've got to reduce that to 30. What airports tend to do is just stretch out the existing schedule, which leads to cancellations and other problems. What you need is a market mechanism so that the flights that have higher priority get out. What would be a higher priority? Bigger planes, probably, but also full planes and planes with a lot of passengers who have connecting flights.

Suppose we're talking about planes leaving LaGuardia in New York. If a plane's going to Los Angeles, it's probably the final destination for a lot of the passengers. Planes going to Chicago or Dallas probably have a lot passengers who are catching connecting flights. Maybe those flights should have a higher takeoff priority in bad weather. In any case, you need a market mechanism where the airlines can compensate one another-and their passengers-to cancel their flights and trade takeoff slots.

Such a system doesn't exist now. First, the airlines have to be convinced that's the way to go. Then the Federal Aviation Administration has to cooperate. And you really need Congress to approve this, since it gave the original allocation of slots. The danger is that Congress will say, "Wait a minute, we don't like this because they're buying and selling these slots. We gave them to you, and now they're making money by reselling them."

Another area for experimental economics has to with NASA. We worked on the Cassini mission [which in 1997 sent 800 pounds of scientific instruments on a small spacecraft to Saturn to conduct experiments]. We used a trading system to allocate the resources that each separate experiment got to use on board.

Each experiment used three basic elements: energy, mass, and volume. The idea was to come up with the most efficient use of these three resources. We allocated a set amount for each resource. Participants were given tokens that were essentially money that they could trade among themselves. When the bidding started, the price of mass started out very high relative to volume and energy. Then people started to conserve it and, by the end, it collapsed. The price of mass collapsed!

That was the first space mission that came in without a big cost overrun. But they never did that sort of market allocation again. Why? Because there's lot of people that like things the way they are. They can run the current system and they don't care about waste. NASA's space station is now about $5 billion over budget. We've proposed selling enough space on that station to earn $5 billion. I don't know whether there's a market for that, but why not take a look at it?

We met with someone in the White House and there was some interest in it, but it takes a political entrepreneur to change that sort of thing, and it's very risky. If it doesn't work, you're going to get all the blame. If it does work, then everybody's going to claim to have been behind it.

Reason: You started out life as a socialist but now call yourself a libertarian. What do you mean by that?

Smith: For me, libertarianism is tied to a certain set of recognitions: that all organizations have the problem of decentralized information, that decentralized mechanisms are the best way to organize that information to produce good outcomes, and that the best results come when the individual is free to make his or her own tradeoffs while aggregating information. That's true whether we're talking about politics or economics or even social interaction. The best systems maximize the freedom of the individual, subject to the constraint of others in the system.

I was born and grew up in Wichita, Kansas. My parents were socialists and our friends were socialists. We were supporters of the Socialist Party. My mother's first vote when she was 21 was cast for Eugene V. Debs, when he was in jail for opposing the First World War. My mother's father was an engineer on the Missouri Pacific Railroad. He was a great lover of Debs, who had organized the railroad workers. My mother ran for state treasurer on the Socialist Party ticket in Kansas more times than I can count. My first vote was cast for [Socialist] Norman Thomas in 1948, when I was 21. The two votes that I've felt the most comfortable about casting were for Norman Thomas in 1948 and [Libertarian Party candidate] Ed Clark in 1980.

Reason: What happened in between?

Smith: My socialist tendencies were starting to fade as I learned more about economics, first at the University of Kansas and then at Harvard. Then I began teaching at Purdue in 1955 and starting doing experiments in 1956. That really changed completely the way I thought about markets.

At Harvard I had been taught by Ed Chamberlain, the author of The Theory of Monopolistic Competition. He had a little classroom demonstration he did the first day of his class. It was designed to show that competition was not in fact workable.

In the fall of 1955, I was teaching principles of economics at Purdue. One night I woke up and I got to thinking about Ed Chamberlain's experiment. If you wanted to show that markets don't work, I thought, you ought to give them a fairer shot than he did. I thought I'd find out how they trade on the New York Stock Exchange, because if there's a competitive market anywhere, that ought to be it. I got a book on the stock market that contained all the details on how the trading took place.

The following spring, I conducted my first experiment, which lasted about six minutes. I offered no real payoffs. I just told students to think of themselves as making money-the difference between the value and price if you're a buyer and the difference between price and the cost if you're the seller. It converged on a competitive equilibrium. I repeated it and the same thing happened.

I thought there must be something wrong with the experiment. How can this be? There's not complete information, and at that time we all knew you needed that for efficient markets. Every semester after that, I'd do experiments with more and more students, and I'd change some of the parameters.

When I gave talks away from Purdue about what I was up to, I could see there were a lot of people who questioned whether this was even economics. "Why would anyone do this?" they'd say. You certainly couldn't make a living doing experimental economics. You had to do the sort of work that would get you promoted. That's when I learned something important. In economics every paper begins with a previous paper, not with a problem of the world. No one questions your right to do something if it's already been done. I did my first experiment January 1956, and the first paper about it was published in 1962.

These days, experimental economics is a growing subfield with international scope. Even Harvard and Yale have some people doing it. They've been very late to get on board, of course. If it's at Harvard and Yale, you've got to wonder if maybe it isn't time to get out, if it isn't over.