Misbehaving: The Making of Behavioral Economics

by Richard H. Thaler

  Sometime after returning to Ithaca from my year in Vancouver, I was at a conference sitting next to the economist Hal Varian, then a well-known theorist who later went on to become the chief economist at Google. Hal was telling me about a new journal that the American Economic Association was starting called the Journal of Economic Perspectives. Hal was an advisory editor. The editorial board was thinking about commissioning regular features for the journal. The clever Barry Nalebuff would write one on economics-based brainteasers and puzzles. Hal and I came up with an idea for a feature that I might write on anomalies. The editor of the journal, Joseph Stiglitz, who enjoys stirring the pot, was easily convinced, and the concept was approved. Four times a year I had a platform to write about anomalies. These could be documentation that supposedly irrelevant factors actually matter, or any other set of facts that were inconsistent with the standard way of doing economic theory.

  I quoted Thomas Kuhn in the opening passage of the first installment of the series, which appeared in the first issue of the journal, published in 1987.

  “Discovery commences with the awareness of anomaly, i.e., with the recognition that nature has somehow violated the paradigm-induced expectations that govern normal science.”

  —Thomas Kuhn

  WHY A FEATURE ON ANOMALIES?

  Consider the following problem. You are presented with four cards lying on the table before you. The cards appear as shown:

  FIGURE 8

  Your task is to turn over as few cards as possible to verify whether the following statement is true: Every card with a vowel on one side has an even number on the other side. You must decide in advance which cards you will examine. Try it yourself before reading further.

  When I give this problem to my class, the typical ranking of the cards in terms of most to least often turned over is A, 2, 3, B. It is not surprising that nearly everyone correctly decides to turn over the A. Obviously, if that card does not have an even number on the other side the statement is false. However, the second most popular choice (the 2) is futile. While the existence of a vowel on the other side will yield an observation consistent with the hypothesis, turning the card over will neither prove the statement correct nor refute it.

  Rather, to refute the statement, one must choose to turn over the 3, a far less common choice. As for the least popular choice, the B, that one must be flipped over as well, since a vowel might be lurking on the other side. (The problem, as stated here, did not specify that numbers are always on one side and letters on the other—although that implicit assumption is commonly made by solvers.) Two lessons emerge from this problem (based on Wason, 1968). First, people have a natural tendency to search for confirming rather than disconfirming evidence, as shown by the relative popularity of the 2 over the 3. This tendency is called the confirmation bias. Second, the confirmation bias can be accentuated when unwarranted assumptions make some kinds of disconfirming evidence seem unlikely, as illustrated by the unpopularity of turning over the B.

  This feature will report successful searches for disconfirming evidence—economic anomalies. As suggested by Thomas Kuhn, an economic anomaly is a result inconsistent with the present economics paradigm. Economics is distinguished from other social sciences by the belief that most (all?) behavior can be explained by assuming that agents have stable, well-defined preferences and make rational choices consistent with those preferences in markets that (eventually) clear. An empirical result is anomalous if it is difficult to “rationalize,” or if implausible assumptions are necessary to explain it within the paradigm. Of course, “difficult” and “implausible” are judgments, and others might disagree with my assessment. Therefore, I invite readers to submit brief explanations (within the paradigm or otherwise) for any of the anomalies I report. To be considered for publication, however, proposed explanations must be falsifiable, at least in principle. A reader who claims that an alleged anomaly is actually the rational response to taxes should be willing to make some prediction based on that hypothesis; for example, the anomaly will not be observed in a country with no taxes, or for non-taxed agents, or in time periods before the relevant tax existed. Someone offering an explanation based on transaction costs might suggest an experimental test in which the transaction costs could be eliminated, and should be willing to predict that the effect will disappear in that environment.

  I wrote a column in every issue, that is, quarterly, for nearly four years. The articles were about ten to twelve published pages, short enough to make them a quick read, but long enough to give a fair amount of detail. Each article ended with a “Commentary” section in which I tried to explain the significance of the findings.

  I can’t say that I had a grand plan when I started writing these columns. I made a list of topics, and off the top of my head I knew I could write at least ten, so the question was what to write about first and how to get the right tone. Having recently written two papers about what makes people angry, I was fully aware that this enterprise could backfire. It was also incredibly time-consuming. Many of the topics were well outside my field of expertise, so in those cases I recruited a coauthor who was a specialist in the field. But I still had to do a lot of boning up on new topics, since I ended up writing the final versions of all of them. That meant that these columns were taking time away from what most academics would consider to be “real research,” meaning discovering new facts, developing new theories, and publishing papers in refereed journals.*

  The potential payoff, however, was huge. The AEA at one point conducted a survey of its members to see what they thought of the new journal. They asked members whether they read it and specifically whether they read the features. Half the members of the AEA who responded to the survey reported that they read the “Anomalies” feature “regularly,” whatever that means. To put this in perspective, the average article written in a specialized academic journal is probably lucky to find 100 readers. These anomalies articles were reaching over 5,000 economists. When recruiting coauthors, I could truthfully tell them that more people were likely to read this article than anything else they would ever write. The same was true for me, of course. I had eyeballs. What should I put in front of them?

  My goal was to cover a broad spectrum of anomalies and to find examples that relied on a wide variety of empirical methods, including many that used market data, to help dispense with the myth that anomalies only occur in the laboratory. Of the fourteen columns I wrote in those first four years, only five were primarily based on experimental data. The others were far-ranging, though many were related to finance, for the simple reason that those were both the most surprising and most disturbing to the defenders of the standard paradigm.

  I should note that I did not have satisfactory behavioral explanations for every anomaly. Some were just empirical facts that did not line up with theoretical predictions. For example, the first two columns were about “calendar” effects in the stock market. These results are just weird. Consider just a sample of them: Stocks tend to go up on Fridays and down on Mondays. January is a good month in which to hold stocks, particularly the early part of the month, and especially for the shares of small companies. Finally, the days before holidays, often Fridays, are particularly good. A burst of papers had documented these results. All logical, and some illogical, explanations for these effects could be rejected. I had no explanation either, but they were certainly anomalies.

  Another anomaly came from bettors at the racetrack. Racetracks in the United States and in many other parts of the world (excluding Britain) use what are called pari-mutuel betting systems, where the odds are determined by the amount of money bet on each horse, rather than a fixed amount set in advance. In the simplest case of bets to win, the track first removes its predetermined share of the betting pool, typically around 17%, and then those who bet on the winning horse divide the rest of the money. The horse that the crowd thinks has the best chance to win is called the favorite, while the horses with small chances to win, say with odds great
er than 10 to 1, are called longshots.

  If the track takes 17% of the bets and the betting market is efficient, then all bets should have the same expected return, namely minus 17%. If you bet $100, you expect to get back $83 on average, from the odds-on favorite to the longest of longshots. But that is not what the data show. The return in betting on favorites is much better than betting on longshots. For example, a bet made on an even-money favorite will return 90 cents for each dollar bet, but a bet on a 100-to-1 longshot only returns about 14 cents on the dollar. And, remember from our earlier discussion of gambling and the breakeven effect (chapter 10), the return in betting on longshots is even worse on the last race of the day.

  After writing fourteen columns in consecutive issues, I took a break. These columns were lightly edited and published in book form with the title The Winner’s Curse (the title of one of the columns). I then wrote a few more on an occasional basis, though without the quarterly deadline, their appearances becoming increasingly irregular. The last appeared in 2006. Shortly thereafter, the column was officially retired. The editor of the journal at that time, Andrei Shleifer, declared that their purpose had been served. That was a polite way of saying that my job chronicling anomalies had ended. I was fired.

  ________________

  * One of the joys of writing the Anomalies columns was that the editors themselves handled the refereeing process, and every paper also received true “editing” to make it intelligible for non-specialists. Tim Taylor, an economist who can also write, has ably performed that task from the beginning, and he is still at it. At most academic journals the editors make sure the economics is right and a copyeditor checks for typos and style, but no one is making suggestions on how to make the article more readable. Early on Tim caught on to the power of defaults. He would rewrite every article, send his new draft along, and then tell authors they were free to opt out of any of his suggestions. By the way, the Journal of Economic Perspectives is available free online to anyone at www.aeaweb.org/jep, including all the back issues. It is a great place to learn about economics.

  19

  Forming a Team

  The “Anomalies” columns served the purpose of showing the economics profession that there were lots of facts that did not line up with the traditional models. They helped establish the case for adopting a new way of doing economics based on Humans rather than Econs. But economics is a big discipline, and I was one lazy man. To create a new field would require a team. How could I do anything to encourage others to join the fun? There was no field manual available to consult on how to make that happen.

  Of course, new fields emerge all the time, and they usually do so without any coordination. Someone writes a paper on a new topic that opens up new lines of inquiry, such as game theory in the 1940s. Soon others read about it, think that the topic seems interesting, and decide to try to make a contribution of their own. If things go well, enough people are soon doing research in the area to start having conferences on the topic, and eventually a journal dedicated to the subject matter emerges. But this is a slow process, and I was yearning for people to talk to besides Amos and Danny. In the late 1980s, there were really just three people besides me who thought of themselves as behavioral economists. One was George Loewenstein, whose work was mentioned in the section on self-control. Another was Robert Shiller, who appeared above and plays a starring role in the next section, and the third was Colin Camerer.

  I first met Colin when he was on the academic job market. At that point he had picked up an MBA and was nearly done with a PhD from the University of Chicago, and he had not yet turned twenty-one. Colin has made many important contributions to behavioral economics. Two stand out. First, he more or less invented the field of behavioral game theory, the study of how people actually play games, as opposed to standard game theory, which studies how Econs would play games if they knew that everyone else playing was also an Econ. More recently, he has been at the forefront of neuro-economics, which uses techniques such as brain imaging to learn more about how people make decisions.

  Colin has many talents. While still a teenager in grad school, he formed a record company and signed the famously satirical punk band called the Dead Milkmen. One of their “hits” was “Watching Scotty Die.” Colin is also a skilled mimic. His Gene Fama and Charlie Plott are particularly good. Personally, I think his Thaler is only so-so.

  Although the additions of Camerer, Loewenstein, and Shiller to the field were all important milestones, I knew that behavioral economics as an academic enterprise would flounder unless it could acquire a critical mass of researchers with a variety of research skills. Fortunately, there was someone else who had the same goal, and could also contribute some resources. That man was Eric Wanner.

  Eric Wanner was a program officer at the Alfred P. Sloan Foundation when he took an interest in combining psychology and economics. Eric is a psychologist by training, but I think he is an economist by predilection, and he relished the chance to see if these two fields could somehow find common ground. He sought out the advice of Amos and Danny about how he could help make this happen. Danny, who prides himself on being a pessimist, remembers telling Eric that he “could not see any way to honestly spend much money on this endeavor,” but they both suggested to Eric that he talk to me. After Eric and I met at the Sloan Foundation in New York, Eric convinced the foundation to provide the funding to support my year in Vancouver visiting Danny.

  Soon after I returned to Cornell, Eric left Sloan to become the president of the Russell Sage Foundation, also located in New York. Although behavioral economics was not at the core of the stated mission of the foundation—which is to address important social policy issues such as poverty and immigration—the board was sufficiently anxious to hire Eric that they agreed to let him bring his behavioral economics agenda along with him. Naturally, he had no more idea of how to go about nurturing a new field than I did, but we put our heads together and tried to figure it out on the fly.

  Our first idea seemed like a good one at the time. Since the goal was to combine economics and psychology, we decided to organize occasional meetings of psychologists and economists and hope that sparks would fly. We invited three groups of people: distinguished psychologists who were willing to endure a day spent talking to economists, some senior economists who were known to have an open mind about new approaches to doing economics, and the few hard-core folks who were engaged in doing research.

  Eric is a persuasive guy, and as a result of his charm and arm-twisting, the collection of psychologists who showed up at our initial meeting was truly astonishing. We had not just Amos and Danny, but also Walter Mischel, of the Oreo and marshmallow experiment fame, Leon Festinger, who formulated the idea of cognitive dissonance, and Stanley Schachter, one of the pioneers of the study of emotions. Together they were the psychology version of the dream team. Some of the friendly economists who agreed to participate were also an all-star cast: George Akerlof, William Baumol, Tom Schelling, and Richard Zeckhauser. The hard-core group was Colin, George, Bob, and me. Eric also invited Larry Summers to come to the inaugural meeting, but Larry couldn’t come and suggested inviting one of his recent students, Andrei Shleifer. It was at that meeting that I first met the rambunctious Andrei, who would later become my collaborator. Jon Elster, the eclectic Norwegian philosopher who seems to be knowledgeable in nearly every intellectual domain, rounded out the group.

  Given the amazing lineup, the couple meetings we had did not turn out to be very productive. I have two vivid memories. One is of Leon Festinger making wry wisecracks, interrupted only by his frequent trips to the foundation’s patio for a smoking break. The other was a plea from William Baumol for us to move beyond the discovery of anomalies. He thought that our anomaly-mining, as he called the activity, had served its purpose, but that we now had to move on to a more constructive agenda. But he had no suggestion about what that constructive agenda should be.

  I think the real problem we faced was a general one t
hat I have learned with experience. Interdisciplinary meetings, especially those with high-level agendas (reduce poverty, solve climate change) tend to be disappointing, even when the attendees are luminaries, because academics don’t like to talk about research in the abstract—they want to see actual scientific results. But if scientists from one field start presenting their research findings in the manner that the colleagues in their field expect, the scientists from other disciplines are soon overwhelmed by technical details they do not understand, or bored by theoretical exercises they find pointless.*

  Whether or not my gloomy assessment of interdisciplinary conferences is correct, the presence and enthusiastic participation of the collection of all-star psychologists at these meetings, held at the Russell Sage Foundation’s office in New York, were both encouraging and misleading regarding the future of the field—encouraging because such luminaries were taking the time to come and seemed to think that the mission was both worthy and sensible, but misleading because they reinforced the belief we all held at the time, which was that if there were to be a successful field called behavioral economics, it would have to be a truly interdisciplinary effort with psychologists and economists working together. It was natural for Amos, Danny, and I to think that, because we had learned so much from one another and had begun to produce actual joint research.