Experimental Finance and Paving Wall Street

I first became interested in the stock market at the age of eleven and the first stock to catch my fancy was IBM--probably because it traded at a much higher price than other stocks. When I was fifteen, I used a primitive computer, the Olivetti-Underwood Programma 101, to help me select stocks for Value Line's contest.

I was more interested in math and computers than I was in the stock market until I was a junior at Caltech. There, I had the incredible good fortune to take a course called "Laboratory Experiments in the Social Sciences" the first time it was offered. Then, in 1974, it was more like a research seminar than a typical college course. It was run by Charles Plott and Vernon Smith, who had been fishing buddies when they both taught at Purdue, and most of the attendees were the rising superstars on Caltech's faculty in economics and political science. I was one of three "paying customers" as Vernon Smith liked to call the students in the class. At the time, I was living across a large parking lot from the largest brokerage office in Pasadena, which had theatre-style seating to watch the electronic ticker tape, and I spent many mornings there trying to make sense of the markets.

I quickly read the few papers published about experimental markets and found  them to be overly simplistic with little bearing on real financial markets. (Experimental economics, the testing of economic theories in a controlled laboratory setting using human subjects paid real money, was considered "disreputable" by the leading lights of the economics profession until the end of the twentieth century.)  I found that  these markets did not allow for what I considered the most important aspect of market--the speculative element. My course project involved the creation of the first true experimental asset market and provided a crude test of Milton Friedman's assertion that speculation exerted a stabilizing influence on markets. After the course, which is still offered at Caltech as Ec/PS 160, was completed, Charlie, Vernon, and I conducted further experiments and wrote them up in a Quarterly Journal of Economics article, Intertemporal Competitive Equilibrium: An Empirical Study of Speculation. Paving Wall Street gives my account of that time at Caltech and  Charles Plott and Vernon Smith have their own accounts.

I returned to Caltech as a visiting faculty member in 1981 where I wondered if an experiment could be devised to test the signaling theory of my main thesis advisor, Michael Spence. Any such experiments would be a radical departure from earlier experiments because the subjects were not merely be led "as if by an invisible hand" to an equilibrium, but would have to "learn" something collectively to get there. In the signaling experiments that Charles Plott and I devised and ran, buyers could signal that they had produced high-quality goods by adding "stripes" to their products. Buyers had to figure out on their own that they could do this and sellers had to learn to associates more stripes with higher quality

It was more difficult to get signaling to work in the laboratory than was indicated by Spence's theory. Indeed, we learned that how data was presented to subjects influenced how quickly they figured out how to signal. Also, we found that when subjects failed to find a signaling equilibrium, they behaved as if one did not exist, a theoretical possibility discovered by Michael Rothschild and Joseph Stiglitz. We published our results in Econometrica in an article entitled Product Quality Signaling in Experimental Markets. In several follow-up studies, Charles Plott and I worked with Michael Lynch and Russell Porter of the U.S. Federal Trade Commission to examine how price could be used as a signal of quality--the key element of George Akerlof's "lemons" theory.

Finding myself at Boston University's finance department, I stopped running experiments and starting looking into how computers could be used in the financial markets. My research took two tracks. The conventional track was how to use computers to determine the value of the increasingly complex derivative securities that were beginning to appear on Wall Street. The more radical track, inspired by my experimental work with Charles Plott and Vernon Smith, was how to treat the market itself as a computational device and to explore what that meant for the design of financial markets.

My conventional work on the construction and valuation of derivative securities took the form of a book, Computer-Aided Financial Analysis, and a companion article that converted a substantial amount of the book's LISP code into Mathematica. I departed academia to try things out in real world at General Electric. My adventures there and how they led to Miller Risk Advisors are recounted in the Risk Management section.

My next book had to wait until work in experimental economics became respectable enough that a nontechnical book on the topic would interest a publisher. (The rumor that the Swedish Academy was planning to award a prize to one or more experimental economists was enough to do the trick.) We managed to get Paving Wall Street: Experimental Economics and the Quest for the Perfect Market into the stores nine months before the prize was awarded to Vernon Smith, who wrote the book's foreword.

Paving Wall Street is my take on the world of finance from the viewpoint of someone coming at the subject from an experimental and computational point of view. David Warsh, the former economics columnist for the Boston Globe, has a nice piece on the book and its significance on his Economic Principals website.

Paving Wall Street is available on Amazon.com. There, one can either browse the sample pages or use the Search Inside This Book feature to look for something specific. (I apologize in advance to all authors of other works whom I unjustly failed to cite.) 

This site contains an expanded table of contents with a description of each chapter and a comprehensive set of links to topics and research papers discussed in the book.