
Vernon Smith on Markets and Experimental Economics 5/21/2007
What this covers
Russ Roberts talks with Vernon Smith about how his career in experimental economics began with a January 1956 classroom exercise. Half the students received private buyer values and half private seller costs, and a two-sided oral auction settled at the intersection of supply and demand although no one knew the aggregate schedules. The conversation follows that thread through the double auction, where untrained subjects reach the competitive equilibrium in three or four periods, and in seven or eight even with two interdependent commodities. Posted-offer markets are slower, at ten or eleven periods. Smith leans on Hayek's papers on knowledge to explain why.
The more contested material comes next. Smith says markets with only four to six traders still converge, and that competition there comes mostly from one's own side, which makes collusion hard. He finds the textbook story of perfect information and many price takers unsatisfying. He is also critical of behavioral economics for studying isolated individuals, and suggests the anomaly literature may simply assume the wrong model of choice under uncertainty. Roy Radner's survival argument is offered as a reading of apparent risk-seeking in losses. Smith also argues that specialization, exchange, property rights, and prices must co-evolve, and that the institutions embodying this learning came from unrecorded trial and error.
The last stretch turns to policy. Smith calls California's electricity deregulation a failure of market design rather than of liberalization. He argues that US regulation leaves consumers bearing investment risk. He points to Australian water markets and to airport landing slots, which are rationed without congestion pricing.
Smith argues that market equilibrium comes from trading institutions and repeated learning, not from the textbook assumptions of perfect information or large numbers, and that laboratory experiments show ordinary people discover prices and specialization through rule-governed exchange.
- Double-auction experiments converge to competitive equilibrium with naive subjects and small numbers of traders
- Institutions and trading rules evolved through unrecorded trial and error and embody the learning that produced wealth
- Behavioral economics focuses on isolated individuals and misses how market and social exchange systems adapt to uncertainty
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Roy Radner's economics of survival shows that an agent who wants to stay in business forever will act as if risk-seeking when wealth falls below a critical level, because a high-variance investment offers the only chance of escaping a poverty trap, so apparent risk-loving behavior in the loss domain is about survival rather than a fixed risk preference.
“there's a critical level of wealth below which you will act as if you prefer risk”
Naive subjects with no training in economics, who were motivated only by their own private buy-low and sell-high incentives, converged to the competitive equilibrium in three or four trading periods, so the result is not magic but a demonstration that people are good at discovering prices through repeated trading and learning.
“after three or four periods this converged to the competitive equilibrium very close to it”
In his first classroom experiment in January 1956, Smith gave half the students private buyer values and half private seller costs, organized a two-sided oral auction, and the market converged to the intersection of supply and demand within a few short trading periods even though no participant knew the aggregate schedules.
“this market converged to the price and volume that's represented by the intersection of those supply and demand schedules”
When supply and demand shift over time, double-auction markets track the moving equilibrium price with lags and occasional overshoots, but they move in the right direction, so the equilibrium concept is a useful reference even though real markets are never perfectly at rest.
“these double auction markets track those changes in equilibrium prices quite well they may lag behind and they may overshoot”
The double auction solves the far harder problem of simultaneous equations: in a two-commodity market where each commodity's willingness to pay depends on the other's price, subjects who cannot solve nonlinear equations still converge to the equilibrium prices and volumes in each market within about seven or eight trading periods.
“in about seven or eight trading periods a two commodity market will converge to the equilibrium price and volume in each of those markets”
Most participants believe they could have done better in the market, but at equilibrium each person is already doing the best he can given what others are doing, which is John Nash's definition of equilibrium, so their sense of missed gains is mistaken.
“the equilibrium has the property that each person is doing the very best he can do for himself given what everybody else is doing”
Posted-offer markets, where sellers post prices and buyers accept the lowest, show more inertia than double auctions: sellers start well above the competitive price and converge more slowly, taking about 10 to 11 periods instead of 3 to 4, though they still reach equilibrium.
“those markets have more inertia they don't adjust as rapidly is a two-sided auction so the convergence is somewhat slower”
Hayek's 1937 paper Economics and Knowledge and his 1945 paper The Uses of Knowledge in Society are essential for understanding experimental markets, because experimental institutions are information-generating devices in which participants not only produce information but adjust their behavior to incorporate it over time.
“these experiments institutions are information generating devices”
Psychology studies how individual minds work and behave, while economics studies how individual decisions aggregate in market settings, which have the Hayekian property of being the product of human action but not of human design, so no one understands or designed the rules that generate their competence.
“the product of human design but they're the product of human action but not of human design”
Markets as Hayek described them are an amorphous aggregation of dispersed information, far from a farmers market or a stock exchange, and the experimental work shows how this vast seemingly chaotic process produces order when people exchange.
“when Hayek was talking about markets you're talking about this vast seemingly chaotic but actually extraordinaire early ordered aggregation of information”
Market convergence does not depend on large numbers of participants: Smith initially used 25 to 50 people, but found that markets with only four to six balanced buyers and sellers still converged, though noisily, so the claim that competition requires large numbers is wrong.
“it turns out if you've got all 4 to 6 people or less balanced on both sides of the market it may be somewhat noisy and volatile a pattern of convergence but these markets still converged”
Adam Smith's division of labor depends on market extent, but specialization, exchange, property rights, and prices must co-evolve simultaneously, so none can emerge without the others, and economics teaches this interaction poorly because it lacks tools for dynamic processes.
“because it had to be simultaneous you see people had in ancient history had to discover trade they had discovered property rights”
The textbook explanations for why supply and demand works, namely that everyone has perfect information and that there are very large numbers of price-taking buyers and sellers, are unsatisfying, and the equilibrium framing says nothing about how people discover the equilibrium in the first place.
“it also is all about equilibrium it kind of wasn't about how people might discover whatever an equilibrium was”
Much of the empirical literature on anomalies in decision making under uncertainty may be misinterpreted because it is not clear that the expected-utility model is the right model, so the decision-making literature needs to be enriched and integrated with how people behave in socio-economic systems that include markets.
“there's a lot of empirical anomalies in that decision-making literature and it's not even clear to me we have the right model”
Behavioral economics, which challenges the rationality of individuals, focuses on isolated individuals making decisions under uncertainty and therefore misses how people adapt to uncertainty together through rule-governed markets and social exchange systems, which include trading favors.
“the literature is primarily concerned with individual decision making and it's individuals that are isolated from the kind of interactive world”
In small experimental markets, competition comes mainly from one's own side of the market: sellers undercut each other and buyers bid prices up against each other, so collusion by suppliers to exploit consumers is very hard even with only four to six participants.
“your main competitions from your fellows so if you're a seller your other sellers who are ruining it for you by bidding prices down”
California's electricity deregulation was not a real market but a highly regulated, stylized system of government-constructed incentives that failed badly, and that failure reflects flawed market design rather than the idea of liberalization itself.
“the California episode I think is a beautiful example of how flawed a market design can be and people don't understand that that is that was the problem”
At the competitive equilibrium of these experiments the sum of buyer and seller surplus is maximized, so no gains from trade are forgone by the group as a whole, which Smith finds surprising given how little any participant understands about the market.
“the total gains to the buyers and sellers combined is maximized there's no foregone benefits to the group as a whole”
The static equilibrium model and welfare theorems contain no dynamics or innovation; distribution falls out of equilibrium as a byproduct, so the model cannot tell a story about how wealth is dynamically created or influenced by distribution.
“the ordinary static equilibrium model doesn't have a mean distribution just falls out of it as a byproduct”
Airport landing slots and airspace are rationed without congestion pricing, which is a problem analogous to the one in electricity: the US deregulated airline routes but not airports, so scarce landing rights remain allocated by non-market means.
“we deregulated the airlines but it was the routes it wasn't the airports and airports can now are becoming a major problem in terms of congestion”
Water is a scarce network good like electricity, and water markets and auctions, which Australia has adopted, would let self-regulating market properties allocate supply, demand, and distribution.
“water networks of supply and demand and distribution I think that's an area where good quality water and and and is becoming scarce”
Expected utility theory predicts consistent maximization, but people are not consistent expected utility maximizers, and the empirical anomalies suggest economists do not yet have the right model of decision making under uncertainty.
“suggests that people are not very good expected utility maximizers not consistently so”
Electricity markets have improved in several foreign countries after liberalization, but in the United States the transition is hard because regulation was designed so that consumers bear the risk of new utility investment, which the utilities recover by raising prices through regulatory fiat rather than through the marketplace.
“the risk of new investment is borne by the consumers not see the firm's because they get a lien current investment”
Economic institutions such as auctions and trading rules evolved through a long history of unrecorded trial and error, with the failures especially lost because people rarely record what did not work, yet those failures are a crucial part of learning, so the rules that survive embody valuable accumulated wisdom.
“those are very important to failures but you see people don't write they don't record very much about the things that didn't work”
Wealth creation is not independent of how wealth is distributed, because the people who discover specialization and exchange benefit themselves while also helping others along the way, a dynamic result that the static equilibrium model cannot represent because it treats distribution as a byproduct.
“wealth creation is not independent of the distribution of wealth”
Experiments treat institutions as information-generating devices: participants generate information through market rules and incorporate it into their behavior over time, so the laboratory can test whether the equilibrium is reached.
“these experiments institutions are information generating devices”
In a laboratory chat-room experiment with Bart Wilson, subjects who discover specialization and exchange can roughly triple their earnings, because the complementary payoff structure makes home production alone yield nothing, while some groups never discover this and others find it only late.
“they can triple their earnings if they discover specialization and exchange”
Hayek's maxim that the proper study of social science is that which is not, meaning institutions and rules that do not exist, can be pursued in the laboratory, where artificial trading rules can be built and their inferiority to surviving institutions can be shown, revealing why the surviving ones have the properties they do.
“the proper study of social science is this study of that which is not”
Experimental economies fall into three types: those that capture essentially all possible wealth, those stuck in autarky (no trade, self-sufficiency), and those that discovered trade late and are still developing, so the experiment can be used to study property rights and how wealth creation depends on distribution.
“we have three kinds of economies we have economies that are by the end of the experiment are capturing a hundred percent of the possible wealth”
Human traditions have developed trading rules that retain successful patterns of behavior, and experiments show these rules have been enormously valuable in facilitating wealth creation.
“somehow in the human career we have devised these methi's traditions that retain certain patterns of behavior and certain rules”
Participants in these experiments typically believe there is no mathematical model that could predict where the market settles, yet if an equilibrium price and quantity are written in a sealed envelope in advance, the outcome matches it remarkably closely, so people fail to perceive the regularity that emerges from their own behavior.
“this widespread denial that that could be the case”
There are three meanings of markets: farmers markets and stock exchanges where buyers and sellers congregate, Hayek's dispersed-information order that emerges from exchange, and government-designed incentive systems that imitate markets, and the third is often mistaken for the first two because designers forget the pieces that create stability.
“there's a third kind of thing we call markets where government uses incentives akin to markets of price and supply and demand”
Margaret Thatcher's privatization of British utilities was largely justified by the need to improve the British Treasury, since the government-owned sector was a fiscal burden, which made foreign liberalizations easier than in the already privately owned US electricity sector.
“Margaret Thatcher sole privatization largely on the basis that we improved the British Treasury”
Buyers profit from exchange as much as sellers do, which is not how people usually think of accounting; in the experiments buyers earn surplus equal to the value they receive minus the price paid.
“buyers profit from exchange just as much as sellers although we normally think of sellers only as profiting”
Hayek's 1937 paper 'Economics and Knowledge' was a precursor to his 1945 paper 'The Use of Knowledge in Society,' and its ideas started in the early 1930s, making them unusually original.
“I read his great 1945 paper on the uses of knowledge in society”
Babylonians used a sealed-bid-style auction around 500 BC, and the double auction likely evolved out of earlier auction forms.
“once the so called English auction actually goes back to the Babylonians 500 BC”
In the early 1980s Smith's group moved experiments onto computer terminals, which allowed programmed trading rules such as bid-ask and posted-offer markets to govern the messages buyers and sellers exchange, enabling much more complex experiments.
“we were starting to do experiments in which the subjects interacted through the terminals of computers”
The textbook and teaching tradition is still presenting Smith's own early-1950s understanding of supply and demand, which Smith argues is wrong, so the invisible hand seen in experiments is less a mystery than something revealed by scraping away the dust of outdated teaching.
“we have people out there still teaching what I learned fifty years ago yeah and you it's unfortunately it's wrong”
Smith has developed great respect for the human ability to solve problems through institutions that humans have created, even though we know little about where those institutions come from.
“I've taken on a tremendous respect for the ability of human beings to solve problems through the institutions that they have somehow created”
Smith's 1962 paper reporting a dozen or so experiments was rejected by referees at the Journal of Political Economy because its author expected Chicago's pro-market reputation to favor it, but the result revealed that committed believers in markets often do not feel the need for evidence, and the paper was published only after a new editor, Harry Johnson, had it re-reviewed.
“if you believe in markets you don't necessarily need evidence that's right”
Behavioral economics tells a negative story about human capacities that is difficult to reconcile with the enormous successes of the human species, because the field's emphasis on individual cognitive limits neglects the exchange and specialization systems that created wealth.
“the behavioral economics story is is tends to be a negative story that's difficult to explain enormous successes of the human career”
Politics makes the transition to liberalized markets very difficult, even when foreign liberalizations show that improvements are possible.
“we still face a political problem in making that that transition”
When Smith began teaching principles of economics at Purdue in 1955, he realized he did not know how people's actual behavior under different trading rules connected to the supply and demand theory he taught, and the standard textbook explanations (perfect information, very large numbers of price takers) were unconvincing.
“I really didn't know what the connection was between what people actually do in markets under various trading rules and the theory of supply and demand that we taught”
Friedrich Hayek's 1945 paper 'The Uses of Knowledge in Society' and his 1937 paper 'Economics and Knowledge' are deeply insightful and can be re-read repeatedly with new insights, since they contain propositions that Smith's experiments confirm.
“I can still read that paper and benefit I could do it tomorrow again and I'd find things in it that I hadn't seen before”
Consumer theory, which includes utility theory, is mainly useful for exam questions, with only one or two useful insights, and cost curves in the theory of the firm are similarly good for generating tricky exam questions but say little about the real world except for engineers.
“the theory of the consumer which utility theory is part of is mainly good for exam questions”
Economics teachers should constantly ask what their theories are good for, and should teach with examples students can relate to from their own experience as economic agents, which is why Smith uses experiments in his teaching.
“it's very important that people be able to relate it to their personal experience as agents”
Harry Johnson, then JPE editor, asked the referees to consider all comments and revise, and he accepted the paper after one referee who had been negative said he had been convinced; Smith argues this kind of editorial independence was rare and that Johnson was the last JPE editor competent to read every submission.
“he says you've convinced me that is all”
George Stigler at Chicago eventually became a great fan of experimental economics, though the field needed some ice-breaking at the time.
“George Stigler at the University of Chicago eventually became a great fan of experimental economics”
Richard Hale's approach to scholarship was to learn everything needed to be truly expert on a problem, including mathematics and several foreign languages, and this model of thoroughness shaped Smith's own work.
“you take a problem that you're interested in and then what you do is learn all the things you need to know in order to to really be good at that”
The lack of dynamism in conventional models of markets is very frustrating to Smith.
“I find the lack of dynamism in our conventional models of markets to be very frustrating”
Whether laboratory experiments can capture enough complexity to base public policy on them is an open question, and Smith says the research is still in progress.
“do you think that the things that you can create in the lab can capture enough of the complexity that we can base policy on it”