There has been an explosion of interest in the last few years in the economics of human beings as they really are; not the “lightning calculator of pleasures and pains” characterized by Thorstein Veblen a century ago, but the habit-ridden, time-bound, emotionally-driven creatures that we know ourselves to be. An extensive interchange among psychologists, political scientists and economists has been growing rapidly for the last ten years, underwritten in the United States by the National Science Foundation and private foundations including Sage, Sloan, and MacArthur. Last year’s Clark Medal for the single most promising economist turning 40 was awarded to Matthew Rabin of the University of California at Berkeley, one of the leaders of the field. The new developments will have far-reaching implications for our business culture.
But “behavioral economics” sometimes is treated as if it were a recent discovery. The fact is that serious investigation has been going on for decades, in various skeins. One of these skeins is game theory, historically a relatively theoretical pursuit. Experimental economics is another. It stemmed in large part from the work of Edward Chamberlin, a brilliant theorist of industrial organization who was all but ignored during his later years at Harvard University because of an unfortunate tendency to insist that he, and he alone, had unlocked economics’ deepest riddles. (He died in 1967.)
Chamberlin was in the habit of organizing trading sessions in his classes with a view to identifying market imperfections. As a graduate student in the early 1950s, Vernon Smith picked the idea up from him — skeptical at first, then sufficiently intrigued as to develop (in 1962) a laboratory double auction, such that bidders could raise their prices and sellers cut theirs, with all bids, offers and transaction prices publicly observed. An enormous volume of productive work followed. Today, along with his former colleague Charles Plott, Vern Smith is on most short lists for a Nobel Prize.
Now a successful financial practitioner (and third-generation student of experimental economics) has written a book that provides a vivid glimpse of where some economists think this sort of thing is headed. Paving Wall Street: Experimental Economics & The Quest for the Perfect Market, by Ross M. Miller, argues that laboratory experiments and computer simulations already are doing for naturally-occurring markets what wind tunnels long have done for bridge and airplane design — making markets safer and better understood, in spite of, or perhaps because of, an occasional disaster. The really significant advances, he says, barely have begun.
Miller’s credentials are not all academic. He established the quantitative finance research group at General Electric, then served for a time as senior vice president and director of research at NatWest Markets in North America at a time when that investment bank was busily spinning out exotic new financial products. He now consults to financial institutions from home in Niskayuna, N.Y (near Albany.) He previously was the author of Computer-Aided Financial Analysis But it was as an undergraduate at Cal Tech that he first collaborated with Smith and Plott, who were his teachers there. And in Paving Wall Street he looks at developments through the eyes of a practicing scientist. (Several basic citations in the experimental literature belong to him.)
The evidence is strong, of course, that this stuff works. To underscore his point, Miller’s book begins with a “Chronology of Major Events” that starts with Galileo’s telescope and Harvey’s demonstration of the circulation of the blood and runs quickly through The Wealth of Nations ,Alfred Marshall’s's neoclassical synthesis and the subsequent mathemastization of the field by Paul Samuelson, John Von Neumann and others. Then come Edward Thorpe’s introduction of Blackjack card-counting, Harry Markowitz’ portfolio theory, William Sharpe’s capital asset pricing model, and the Black-Scholes options-pricing formula and the introduction of the PDP-8, the first mass-market minicomputer.
These first two pages of this timeline give way to two more ever-more detailed pages listing an avalanche of work since 1973: A. Michael Spence’s formal description of market signaling, Amos Tversky and Daniel Kahneman’s individual irrationality experiments, agenda manipulation experiments, preference reversal experiments, bubble experiments, information mirage experiments, information aggregation in asset market experiments, and so on and on, intellectual events punctuated by Sun Microsystem’s introduction of the first “pizza box” work station (SPARCstation 1), the advent of a complicated auction market for space station resources, the first auction of electromagnetic spectrum, the defeat of chess champion Garry Kasparov by IBM’s Deep Blue, and, of course, the failure of Long Term Capital Management in 1998 and the market-induced rolling blackouts in California in 2001.
The great virtue of Miller’s book is that these developments are thoroughly explained, clearly and simply and often with an eye for human detail that enlivens what otherwise would be dull reading. The defect is that the most recent work is still so new that perspective is difficult to maintain. Yes, those insights are like so many tall buildings, but it is all a bit like a walking tour of the Manhattan skyline, too close-up to leave a lasting impression. Miller makes a heroic effort to synthesize the rapid developments of thirty years in real time — ultimately work for many hands.
Some of the promise of wind-tunnel economics is to be seen in the current discussion surrounding the recent Internet bubble, when the prices of dot.com and telecom stocks were bid up to unrealistically high levels before they crashed. It was in 1985 that Miller and his teacher Charlie Plott devised the first market experiments that required explicit learning by market participants. Previously experimental buyers and sellers had traded only identical items; now “Regulars” and “Supers” (with stripes attached) were introduced. Soon enough it became clear that Cal Tech subjects quickly became adept at appropriately valuing stripes.
But experiments at community colleges showed that less pattern-oriented subjects had a hard time figuring the system out. A commonly agreed-upon signaling system failed to take hold. Bids fluctuated wildly. A simple adjustment — recording Regulars and Supers on blackboards with different colored chalk (custom software was still a rarity in the mid-’80s!) — was enough to get the experiment to “come out right.” That is, with clearer signals prices regularly would converge toward equilibrium.
Then in 1988, Vern Smith and colleagues demonstrated the ease with which bubbles formed in laboratory settings and the difficulty in setting them right. A spectacular run-up in asset prices far in excess of the expected dividend payout turned out to be relatively easy to engineer. What was necessary was a certain amount of momentum. A rapid increase in prices would cause nearly everybody to forget the rules of fundamental valuation. The availability of plenty of cash in everyone’s pocket helped too. Once out of line, prices almost never readjusted normally. Bubbles tended to pop rather than slowly to deflate.
Miller’s experience with the improvement in learning occasioned by the colored chalk leaves him a believer that relatively small reforms can lead to big improvements in market processes. Markets that bubble to excess and then crash display a more dramatic instability than markets in which signalling fails to take hold, but the underlying problem is the same — the lack of a connection between current price and an asset’s expected future cash flow. (Miller’s argument is contained in “Can Markets Learn to Avoid Bubbles,” forthcoming in the Journal of Psychology and Financial Markets and posted on his website at www.millerrisk.com.)
Bubbles “in the wild” almost always occur in markets where forward or futures markets are restricted or illegal, he says. During the Internet boom, for example, there was no organized way to trade forward contracts in individual US stocks, thanks to a turf war between the Securities and Exchange Commission and the Commodities Futures Trading Commission that lasted from 1982 until the issue of oversight was resolved in 2000. Relatively illiquid “equity swaps” created by investment banks were the next best way to buy and sell companies’ future earnings — not nearly enough bad news there to keep the bubble from taking hold.
Maybe some new method of relating the intrinsic value to market value can be devised. In that case, the human propensity to create bubbles may be defeated. Such a mechanism seems at least as likely to come from the laboratory as from the markets themselves. Certainly future candidates routinely will be tested there. The “great lesson of the 20th century,” according to Miller, is that no longer can economists expect an Invisible Hand to guide markets along the path to perfection. “Based on the early results that come out of the laboratory,” he concludes, “We have an obligation to participate fully in the economy’s evolution.”