Misbehaving: The Making of Behavioral Economics

by Richard H. Thaler

  27

  Law Schooling

  I spent the academic year 1994–95 as a visiting professor at MIT’s Sloan School of Management in order to spend some time with France Leclerc, who was then on their faculty in the marketing department. It was during that year that we both accepted faculty positions at the University of Chicago Graduate School of Business (as it was then called), and we later married.* While at MIT I got a call from Orley Ashenfelter—the economist who had let Eldar Shafir and me use his wine newsletter to study mental accounting—asking whether I might give a plenary talk on the applications of behavioral economics to the law at a conference he was organizing. “We need some of that wackonomics,” he said. I told Orley that the topic was definitely interesting, but I knew nothing about the field of law. I said I would look for a knowledgeable collaborator and get back to him.

  One of the participants in our first summer camp, Christine Jolls, was a prime prospect. She was just finishing up both her PhD in economics at MIT and her law degree at Harvard, and was a hard worker. Christine was game, and as we tossed around topics to cover we soon came up with enough material for a decent talk, so I told Orley we would accept his invitation. The basic idea was to think about how the field of law and economics, as currently practiced, should be modified to accommodate recent findings in behavioral economics.

  The traditional law and economics approach was based exclusively on models of Econs. Many of the articles took many pages to reach the conclusion that things would turn out for the best if markets were left alone to sort things out. Many of the arguments depended implicitly on some form of the invisible handwave.

  Our idea was to introduce some of the essential elements of behavioral economics into such arguments and see how they would have to be modified. By this point I had adopted the pedagogical device of calling these essential elements “the three bounds”: bounded rationality, bounded willpower, and bounded self-interest. In law and economics these properties of Humans had heretofore been assumed to be thoroughly unbounded.

  I ended up having to miss the conference, leaving Christine to give the talk solo, but it went over well enough that we thought it was worth expanding into an academic paper. We planned to get busy on that once we both settled into our new jobs. She had been hired by Harvard Law School and would join their faculty at the same time that I was arriving at Chicago.

  The stars must have been in some kind of fortuitous alignment, because when I arrived at Chicago, the first faculty member I met from outside the business school was Cass Sunstein, a professor at the law school. Cass had already been collaborating with Danny and was excited about behavioral economics. In the world of academic law, Cass is a rock star. Although nominally his specialty is constitutional law, he has written articles and books on nearly every branch of the law, and is widely admired. We had lunch a couple times and hit it off well. His enthusiasm is catching, and his encyclopedic knowledge is astonishing. At some point I suggested to Christine that we should consider asking Cass to join our behavioral law and economics project. It was not a hard sell. Adding Cass to your research team is a bit like adding Lionel Messi to your pick-up soccer game. Soon we were off and running. And I mean running, because Cass is fast.

  It took only a few months for the three of us to produce a draft of a paper that we titled “A Behavioral Approach to Law and Economics.” It was the longest paper I have ever written. To law professors, the longer a paper is, the better, and there can never be too many footnotes. The published version of the paper came in at 76 pages and 220 footnotes, and it was only this short because I kept complaining of its excessive length.

  When we had a draft ready to submit for publication, I learned that the process is very different in legal circles than in economics. In economics, you are only allowed to submit your article to one journal at a time. If they reject it, you can try another one. But law reviews allow authors to submit to several at once, which we did. The Stanford Law Review was the first to get back to us with an acceptance, and soon after another law review was expressing interest as well. We had bargaining power, so I made a suggestion. Since the editors were so keen to get this article, and it was bound to be controversial, why not get them to solicit a commentary from a prominent representative of the law and economics inner circle to be published in the same issue, with us having the opportunity to reply? I had in mind how the debate with Merton Miller and his team had attracted a lot of attention to the closed-end funds paper, and I thought this might have a similar effect.

  The obvious choice to provide the critical commentary was the legal scholar Richard Posner. Posner is considered by many to be the founder of modern law and economics, and he has written the definitive treatise on the subject, revised numerous times. The field Posner helped create introduced formal economic reasoning into legal scholarship. From the beginning, law and economics was primarily based on traditional, Chicago-style economics, so he had a considerable investment in the approach to which we were offering an alternative.

  We knew that Posner would find much to criticize in our approach, and we also knew that he could knock off a comment quickly. In spite of his serving both as a part-time law professor and federal judge on the Seventh Circuit in Chicago (one step below the Supreme Court), his research productivity is legendary. As the economist Robert Solow so colorfully put it: “Posner evidently writes the way other men breathe.” Writing a comment on our long article would not take him much time.

  Although we had a good hunch about what Posner might think of our paper, any uncertainty about which parts of the paper he would find to be most objectionable was resolved the day before the three of us were to present our paper at the University of Chicago Law School. That morning we received a letter from him with his comments. The letter, which ran many single-spaced pages, was highly critical, and quite emotional. Posner told us he had written up his thoughts so that he could remain silent during our talk, knowing that others would be anxious to speak as well. Maybe he thought it would serve as a good commitment strategy.

  Before getting to what the arguments were about, some background is required. When Richard Posner and others of his generation started the law and economics movement, there were many legal scholars who were uncomfortable with some of the conclusions of their work, but they lacked the economic training to put up a good fight. At that time, the few law professors that had any formal training in economics were using the traditional approach based on models of Econs, and legal scholars who tried to challenge the conclusions of such papers often felt bullied if they entered the ring against the law and econ crowd, who could brush aside critiques with a condescending “Well, you just don’t understand.” As a result, at our workshop some attendees would be defending the old-time religion, like Posner, while others might be (quietly) rooting for the underdogs to score some points against the bullies.

  Cass and Christine both thought I should present the paper. They argued that I had more battle experience, or at least that was their story. They were nearby, and I kept expecting to look over at them and find them hiding under the table.

  I began by reminding everyone that standard law and economics assumes that people have correct beliefs and choose rationally. But suppose they don’t? How should law and economics change? Our paper offered an illustrative example based on a new policy that had been adopted by the Chicago police department. Parking tickets had traditionally been placed on a car’s front windshield, held down by the wiper blade. The new policy was to issue tickets that were printed on bright orange paper and were attached by some sticky substance to the driver’s side window, where they were highly visible to drivers passing by. We pointed out that such a policy was smart from a behavioral perspective, since it might increase the perceived probability of getting a ticket, thus discouraging illegal parking at almost no cost.† This example may not seem either profound or controversial, but remember that part of the received wisdom in law and economics is that people have correct beliefs, in
cluding about the probability of getting caught committing some crime, and base their decisions about whether to commit a crime, from illegal parking to robbing a bank, by calculating the expected gains and losses. If it were possible to change the perception of the chance of getting caught by just changing the color and location of parking tickets, without changing the actual probability of being caught, then it might be possible to do the same for more serious crimes. This thought was pure heresy.

  Judge Posner remained quiet for about five minutes, but then he could no longer contain himself. Why, he asked out of the blue, were we ignoring evolution? Didn’t evolutionary biology explain many of the odd behaviors discussed in the paper, such as turning down small offers in the Ultimatum Game, or ignoring sunk costs? Couldn’t evolution explain these and all our other “cognitive quirks” (a slyly deprecating term he insisted on using)? His thought was that if humans had evolved to pay attention to sunk costs, or resist unfair offers in the Ultimatum Game, then such behavior must be good for us, in some sense, and therefore rational. Problem solved.

  I assured him that I was not a creationist and accepted evolution as a scientific fact. I added that there was no doubt that many of the aspects of human behavior that we were talking about had evolutionary roots. But, I argued, accepting the theory of evolution as true does not mean that it needs to feature prominently in an economic analysis. We know people are loss averse; we don’t need to know whether it has an evolutionary explanation. (Amos used to joke that there once were species that did not display the endowment effect, but they are now extinct.) Furthermore, the real point of behavioral economics is to highlight behaviors that are in conflict with the standard rational model. Unless we change the model to say that people pay attention to sunk costs, the model will make poor predictions. At this point Posner was completely exasperated. “You are completely unscientific!” he cried, in utter despair. I had resolved to remain calm so I just smiled at this outburst, said, “Okay then,” and moved on. There was much more contentious material still to come, and I was determined not to get into a shouting contest, especially with a federal judge!

  The biggest fight was about something called the Coase theorem. The Coase theorem is named for its inventor, Ronald Coase, who had been a faculty member at University of Chicago Law School for many years. The theorem can be easily stated: in the absence of transaction costs, meaning that people can easily trade with one another, resources will flow to their highest-valued use.‡

  The logic is easy to explain. I will follow Coase’s lead and explain it with a simple numerical example. Suppose that Alexa and Julia are college roommates. Julia is quiet and studious, but Alexa is boisterous and likes to play loud music while she studies, which disturbs Julia. Julia complains to the dorm resident advisor, Hallie, who is empowered to settle disputes like this. Hallie can choose one of two alternatives: she can give Alexa the right to play her music as loud as she likes, or she can give Julia the right to quiet during certain hours. The Coase theorem makes a strong and surprising prediction: Hallie’s decision will have no effect on how much music Alexa will play. Rather, that will depend simply on whether Alexa likes her music more than Julia hates it.

  The result is surprising but the logic is simple. Suppose Alexa is willing to pay $5 per night to blast her music, and Julia is willing to pay $3 a night for silence. If Julia is awarded the right to silence, then, according to the Coase theorem, Alexa will pay Julia some amount between $3 and $5 for the right to play her music, an amount Julia will accept. Both will be happier this way than if Alexa couldn’t play her music but no money changed hands; that is, after all, why they’re both agreeing to the transaction. And if Alexa wins the right to play her music, Julia will be unwilling to pay her enough to stop, since her value of silence is less than Alexa’s joy of music. Either way, Julia will have to find somewhere else to study if she wants quiet.

  The reason this result is important for the law is that judges often decide who owns a certain right, and the Coase theorem says that if transaction costs are low, then what the judge decides won’t actually determine what economic activities will take place; the judge will just decide who has to pay. The article that includes this result, entitled “The Problem of Social Cost,” is one of the most cited economics articles of all time.

  The argument I have sketched up to this point crucially depends on the stated assumption that the costs involved in the two parties coming to an efficient economic agreement are small to nonexistent. Coase is upfront about this. He says: “This is, of course, a very unrealistic assumption.” Although many applications of the Coase theorem ignore Coase’s warning, we wanted to show that the result was wrong, even when it could be shown that transaction costs were essentially zero. To do so, we presented the results of the mug experiments that were discussed in chapter 16, the results of which are summarized in figure 17.

  FIGURE 17

  Recall that the first stage of the experiments involved tokens that were redeemable for cash, with each subject told a different personal redemption value for a token, meaning the cash they could get for it if they owned one at the end of the experiment. The Coase theorem predicts that the students who received the highest personal valuations for their tokens would end up owning them; that is what it means to say that resources flow to their highest valued use. And that is what happened. The market worked perfectly, just as the theory predicted, which also meant that transaction costs must not be inhibiting trade in any meaningful way.

  But the Coase theorem is not meant to be limited to tokens for which people are told their personal values. It says that the same thing should happen when we replace the tokens with real goods, such as coffee mugs. So when we gave every other student a coffee mug, the Coase theorem predicts that the students who liked the mugs most should end up owning them, and since the mugs were randomly assigned, about half the mugs should trade. Yet we found that trading volume was much lower than that: resources were not flowing at the rate predicted. And the reason was the endowment effect: people given mugs valued them about twice as much as people not given the mugs. How goods were allocated did affect who would end up owning the mugs. In other words, the Coase theorem worked in theory, when trading for tokens redeemable for cash, but it did not work in practice, when trading for real-world objects like coffee mugs. Questioning the Coase theorem at a law and economics workshop! That was high treason.

  One of the unfortunate aspects of the University of Chicago at that time, one that is thankfully no longer the case, was that there was an undue tolerance for scholars who would spout the Chicago School traditional lines, loudly and frequently. One example was the economist John Lott, who had strung together a series of visiting appointments allowing him to be at the university for several years. Lott is most famous for writing a book entitled More Guns, Less Crime. As the title suggests, the thesis of the book is that if we just made sure every American was armed at all times, no one would dare commit a crime, a claim that other researchers have strongly disputed.§ Lott was a frequent attendee and active participant at workshops. His style resembled that of a pit bull.

  At this workshop, Lott was present and looking annoyed, so I hoped he was not packing a gun. His wife, Gertrude (also an economist), was in the crowd as well and asked a question about the mugs study. Couldn’t the low trading of the mugs be explained by transaction costs? I explained that the tokens experiment had ruled out this explanation—after all, the tokens had the same transaction costs as the mugs, and the tokens did trade as much as the theory predicted. She seemed satisfied, but then Lott jumped in to “help.” “Well,” he asked, “couldn’t we just call the endowment effect itself a transaction cost?” I was shocked by this comment; transaction costs are supposed to be the cost of doing a transaction—not the desire to do a transaction. If we are free to re-label preferences as “costs” at will so that behavior appears to be consistent with the standard theory, then the theory is both untestable and worthless. So instead of trying t
o reason with Lott, I turned to Posner and asked him whether he would now concede that I was not the least scientific person in the room. Posner smiled, nodded his agreement, and everyone in the room who could see him laughed. But Posner was not in Lott’s line of sight, so I saw him angrily asking people around him what had happened. I quickly moved on to another topic.

  The fact that the strongest resistance to behavioral economics came from those who had the greatest investment in building up the rational actor model raised an amusing possibility. Might their objections be more evidence in support of the sunk-cost fallacy? Of course, I could not say to my critics that by clinging to their beloved theories they were merely paying attention to sunk costs, but I could introduce the one bit of new experimental data we had included in the paper. The data came from a version of the Ultimatum Game.

  In the usual version of the Ultimatum Game, the experimenter provides the money that the participants divide. Now we had created a version in which the experimenter makes money! We asked students to bring $5 to class for the purpose of an in-class demonstration. (Participation was voluntary.) Then each student filled out a form indicating how he would play a $10 version of the Ultimatum Game, with the money coming from the $5 each player had contributed. Players indicated their contingent decisions both as Proposer and as Responder, were told that they would be randomly assigned to one of those roles, and then were paired with another, anonymous student who had been given the other role.¶

  If sunk costs don’t matter, then the outcome of this game should be identical to the one where the experimenter provides the money. The sunk cost of $5 is a SIF. But economists might think that if the students had to provide their own money, they will take the experiment more seriously and therefore act more rationally. We found exactly the opposite. Although the Proposers behaved very similarly to participants in previous versions of the game where the money came from the experimenter, with most offering to share at least 40% of the $10, the Responders, the ones we were really interested in, changed their behavior in a manner that made the results even more inconsistent with the predictions of the standard theory.