Misbehaving: The Making of Behavioral Economics

by Richard H. Thaler

  My poker observations yielded another wrinkle on mental accounts. Players who were ahead in the game did not seem to treat their winnings as “real money.” This behavior is so pervasive that casino gamblers have a term for it: “gambling with the house’s money.” (The casino is referred to as “the house.”) Using this reasoning, when you are ahead you are betting with the casino’s money, not your own. You can observe this behavior at any casino. Watch a (nonprofessional) gambler who wins some money early in the evening and you may see what I call “two-pocket” mental accounting. Take a player who has brought $300 to the casino to bet, and finds himself up $200 early in the evening. He will put $300 into one pocket and think of that money as his own money, and put the $200 worth of chips he has won in a different pocket (or more likely, on the table, ready to bet). When it comes to “house money,” the expression “easy come, easy go” applies. This is about as blatant a violation of the rule that money is fungible as one can find. The money in either pocket will spend equally well.

  Taking money off your colleagues is fun,† but far from scientific. So Eric Johnson, a marketing professor now at Columbia, and I started work on a real paper. This is the one I mentioned in the preface that took a while to satisfy Amos. Essentially, we wanted to replicate in the lab what I had seen at the poker table. But first we had to address the problem that had originally pushed Kahneman and Tversky to run experiments using hypothetical questions. How can you run ethical experiments in which subjects can lose money, and how can you get approval from the university review board that oversees such experiments? We solved this problem by having subjects answer a series of choices between sure things and gambles, some of which involved gains and others losses, and truthfully told them that one of the choices would be selected at random to “count” for the study. But not every gamble was equally likely to be chosen, and by making the favorable gambles the ones that were more likely to be played, we were able to assure the subjects that the chance of losing money was tiny, although we made it clear that we fully intended to collect from anyone who did lose money. If they wished, they could pay off their debt by doing some research assistance. In the end no one lost money, so we did not have to try to collect.

  Here are three of the questions that were included in our study. The numbers in brackets are the percentages of subjects who chose the selected answer. In this example, an Econ who was risk averse would choose the sure thing in each of these problems, since in every case the expected outcome of the gamble is equal to the sure thing.

  PROBLEM 1. You have just won $30. Now choose between:

  (a) A 50% chance to gain $9 and a 50% chance to lose $9.

  [70%]

  (b) No further gain or loss.

  [30%]

  PROBLEM 2. You have just lost $30. Now choose between:

  (a) A 50% chance to gain $9 and a 50% chance to lose $9.

  [40%]

  (b) No further gain or loss.

  [60%]

  PROBLEM 3. You have just lost $30. Now choose between:

  (a) A 33% chance to gain $30 and a 67% chance to gain nothing.

  [60%]

  (b) A sure $10.

  [40%]

  Problem 1 illustrates the “house money effect.” Although subjects tend to be risk averse for gains, meaning that most of them would normally turn down a coin flip gamble to win or lose $9. When we told them they had just won $30, they were eager to take that gamble. Problems 2 and 3 illustrate the complex preferences in play when people consider themselves behind in some mental account. Instead of the simple prediction from prospect theory that people will be risk-seeking for losses, in problem 2 a loss of $30 does not generate risk-taking preferences when there is no chance to break even.‡ But when given that chance, in problem 3, a majority of the subjects opt to gamble.

  Once you recognize the break-even effect and the house money effect, it is easy to spot them in everyday life. It occurs whenever there are two salient reference points, for instance where you started and where you are right now. The house money effect—along with a tendency to extrapolate recent returns into the future—facilitates financial bubbles. During the 1990s, individual investors were steadily increasing the proportion of their retirement fund contributions to stocks over bonds, meaning that the portion of their new investments that was allocated to stocks was rising. Part of the reasoning seemed to be that they had made so much money in recent years that even if the market fell, they would only lose those newer gains. Of course, the fact that some of your money has been made recently should not diminish the sense of loss if that money goes up in smoke. The same thinking pervaded the views of speculative investors in the boom housing market years later. People who had been flipping properties in Scottsdale, Las Vegas, and Miami had a psychological cushion of house money (no pun intended) that lured them into thinking that at worst they would be back where they started. Of course, when the market turned down suddenly, those investors who were highly leveraged lost much more than house money. Many also lost their homes.

  Gambling when behind in an effort to break even can also be seen in the behavior of professional investors. Mutual fund portfolio managers take more risks in the last quarter of the year when the fund they are managing is trailing the benchmark index (such as the S&P 500) to which their returns are compared. And, much worse, many of the rogue traders that lost billions for their employers were taking on ever increasing amounts of risk at the end, in a desperate effort to break even. This behavior may have been rational from the point of view of the rogue trader, who stood to lose his job or worse if he did not recover his loss. But if true, that means management needs to pay close attention to the behavior of employees who are losing money. (Well, come to think of it, management should have been paying more attention before the rogue traders built up their big losses.) A good rule to remember is that people who are threatened with big losses and have a chance to break even will be unusually willing to take risks, even if they are normally quite risk averse. Watch out!

  ________________

  * This was before the beginning of the trend toward winner-take-all poker evenings inspired by the popularity of that form of wagering at poker tournaments.

  † In some cases it was also easy. Bill Green, an econometrician who was a regular in our group, and I noticed that when a certain colleague of ours got a good hand he would start bouncing up and down in his chair. This was the ultimate “tell.” At some point we felt sorry for him and let him know about it, but he could not restrain himself when he got a really good hand. I kept waiting for him to take away a big prize with a fake bounce, but he never did.

  ‡ This means that the prediction from prospect theory that people will be risk-seeking in the domain of losses may not hold if the risk-taking opportunity does not offer a chance to break even.

  III.

  SELF-CONTROL:

  1975–88

  Prospect theory and the insights provided by its value function greatly facilitated my attempt to understand mental accounting, which in turn helped me make sense of many of the items on the List. But one of those examples seemed be in a different category: the incident of removing the cashews while waiting for dinner. To an economist, removing an option can never make you better off. So why were we so happy that the bowl of cashews was safely hidden in the kitchen?

  I started collecting other examples of “cashews” phenomena. Smokers paid more for their cigarettes by purchasing them one pack at a time instead of by the carton. Dieters did not stock any ice cream in the freezer. Academics (including me) would commit themselves to present a paper that was still a work in progress at a conference several months off, to give themselves an incentive to finish it. People who had trouble getting up in the morning put their alarm clocks on the other side of the room so they could not just reach over and switch off the alarm without getting out of bed.

  What these examples have in common is the presence of self-control problems. We want to eat just a few more nuts, but ar
e worried that if the bowl is left on the table, we will submit to temptation.

  This distinction between what we want and what we choose has no meaning in modern economics, in which preferences are literally defined by what we choose. Choices are said to “reveal preferences.” Imagine the following conversation between a Human who just removed a bowl of cashews with an Econ looking on.

  ECON: Why did you remove the cashews?

  HUMAN: Because I did not want to eat any more of them.

  ECON: If you did not want to eat any more nuts, then why go to the trouble of removing them? You could have simply acted on your preferences and stopped eating.

  HUMAN: I removed the bowl because if the nuts were still available, I would have eaten more.

  ECON: In that case, you prefer to eat more cashews, so removing them was stupid.

  This conversation, which is obviously going nowhere, mimics many I had with economists at the time. Although it is never stated explicitly as an assumption in an economics textbook, in practice economic theory presumes that self-control problems do not exist. So my next big project was to study a supposedly nonexistent problem.

  11

  Willpower? No Problem

  Economists have not always been so dense about self-control problems. For roughly two centuries, the economists who wrote on this topic knew their Humans. In fact, an early pioneer of what we would now call a behavioral treatment of self-control was none other than the high priest of free market economics: Adam Smith. When most people think about Adam Smith, they think of his most famous work, The Wealth of Nations. This remarkable book—the first edition was published in 1776—created the foundation for modern economic thinking. Oddly, the most well-known phrase in the book, the vaunted “invisible hand,” mentioned earlier, appears only once, treated with a mere flick by Smith. He notes that by pursuing personal profits, the typical businessman is “led by an invisible hand to promote an end which was no part of his intention. Nor is it always the worse for the society that it was no part of it.” Note the guarded language of the second sentence, which is rarely included (or remembered) by those who make use of the famous phrase, or invoke some version of the invisible handwave. “Nor it is always the worse for society” is hardly the same thing as an assertion that things will turn out for the best.

  The rest of the massive book takes on almost any economics topic one can think of. For example, Smith provided the underlying theory for my PhD thesis, on the value of a life. He explained how workers had to be paid more to compensate them for taking dirty, risky, or unpleasant jobs. The famous Chicago economist George Stigler was fond of saying that there was nothing new in economics; Adam Smith had said it all. The same can be said of much of behavioral economics.

  The bulk of Smith’s writings on what we would now consider behavioral economics appeared in his earlier book The Theory of Moral Sentiments, published in 1759. It is here that Smith expounded on self-control. Insightfully, he portrayed the topic as a struggle or conflict between our “passions” and what he called our “impartial spectator.” Like most economists who find out that Smith had said it first, I only learned about this formulation after proposing my own version, which we will get to later in this section. The crucial feature of Smith’s conception of our passions is that they are myopic, that is, shortsighted. As he framed it, the problem is that “The pleasure which we are to enjoy ten years hence, interests us so little in comparison with that which we may enjoy to-day.”

  Adam Smith was not the only early economist to have sensible intuitions about self-control problems. As behavioral economist George Loewenstein has documented, other early treatments of “intertemporal choice”—that is, choices made about the timing of consumption—also stressed the importance of concepts such as “willpower,” a word that had no meaning in the economics being practiced in 1980.* Smith recognized that willpower is necessary to deal with myopia.

  In 1871, William Stanley Jevons, another economics luminary, refined Smith’s observation about myopia, noting that the preference for present consumption over future consumption diminishes over time. We may care a lot about getting that bowl of ice cream right now rather than tomorrow, but we would scarcely care about a choice between this date next year versus the day before or after.

  Some early economists viewed any discounting of future consumption as a mistake—a failure of some type. It could be a failure of willpower, or, as Arthur Pigou famously wrote in 1921, it could be a failure of imagination: “Our telescopic faculty is defective and . . . we, therefore, see future pleasures, as it were, on a diminished scale.”

  Irving Fisher provided the first economic treatment of intertemporal choice that might be considered “modern.” In his 1930 classic, The Theory of Interest, he used what have become the basic teaching tools of microeconomics—indifference curves—to show how an individual will choose between consumption at two different points of time, given a market rate of interest. His theory qualifies as modern both in its tools and in the sense that it is normative. He explains what a rational person should do. But Fisher also made clear that he did not think his theory was a satisfactory descriptive model, because it omitted important behavioral factors.

  For one thing, Fisher believed that time preference depends on an individual’s level of income, with the poor being more impatient than those who are better off. Furthermore, Fisher emphasized that he viewed the impatient behavior exhibited by low-income workers as partly irrational, which he described with vivid examples: “This is illustrated by the story of the farmer who would never mend his leaky roof. When it rained, he could not stop the leak, and when it did not rain, there was no leak to be stopped!” And he frowned upon “those working men who, before prohibition, could not resist the lure of the saloon on the way home Saturday night,” which was then payday.

  Quite evidently, from Adam Smith in 1776 to Irving Fisher in 1930, economists were thinking about intertemporal choice with Humans in plain sight. Econs began to creep in around the time of Fisher, as he started on the theory of how Econs should behave, but it fell to a twenty-two-year-old Paul Samuelson, then in graduate school, to finish the job. Samuelson, whom many consider to be the greatest economist of the twentieth century, was a prodigy who set out to give economics a proper mathematical foundation. He enrolled at the University of Chicago at age sixteen and soon went off to Harvard for graduate school. His PhD thesis had the audacious but accurate title “Foundations of Economic Analysis.” His thesis redid all of economics, with what he considered to be proper mathematical rigor.

  While in graduate school in 1937, Samuelson knocked off a seven-page paper with the modest title “A Note on the Measurement of Utility.” As the title suggests, he hoped to offer a way to measure that elusive thing Econs always maximize: utility (i.e., happiness or satisfaction). While he was at it, Samuelson formulated what has become the standard economic model of intertemporal choice, the discounted utility model. I will not strain you (or myself) with any attempt to summarize the heart of this paper, but merely extract the essence our story requires.

  The basic idea is that consumption is worth more to you now than later. If given the choice between a great dinner this week or one a year from now, most of us would prefer the dinner sooner rather than later. Using the Samuelson formulation, we are said to “discount” future consumption at some rate. If a dinner a year from now is only considered to be 90% as good as one right now, we are said to be discounting the future dinner at an annual rate of about 10%.

  Samuelson’s theory did not have any passions or faulty telescopes, just steady, methodical discounting. The model was so easy to use that even economists of that generation could easily handle the math, and it remains the standard formulation today. This is not to say that Samuelson thought his theory was necessarily a good description of behavior. The last two pages of his short paper are devoted to discussing what Samuelson called the “serious limitations” of the model. Some of them are technical, but one deserves ou
r scrutiny. Samuelson correctly notes that if people discount the future at rates that vary over time, then people may not behave consistently, that is, they may change their minds as time moves forward. The specific case he worries about is the same one that worried earlier economists such as Jevons and Pigou, namely, the case where we are most impatient for immediate rewards.

  To understand how discounting works, suppose there is some good, perhaps the chance to watch a tennis match at Wimbledon. If the match is watched tonight, it would be worth 100 “utils,” the arbitrary units economists use to describe levels of utility or happiness. Consider Ted, who discounts at a constant rate of 10% per year. For him that match would be worth 100 utils this year, 90 next year, then 81, 72, and so forth. Someone who discounts this way is said to be discounting with an exponential function. (If you don’t know what that term means, don’t worry about it.)

  Now consider Matthew, who also values that match at 100 today, but at only 70 the following year, then 63 in year three or any time after that. In other words, Matthew discounts anything that he has to wait a year to consume by 30%, the next year at 10%, and then he stops discounting at all (0%). Matthew is viewing the future by looking through Pigou’s faulty telescope, and he sees year 1 and year 2 looking just one-third of a year apart, with no real delay between any dates beyond that. His impression of the future is a lot like the famous New Yorker magazine cover “View of the World from 9th Avenue.” On the cover, looking west from 9th Avenue, the distance to 11th Avenue (two long blocks) is about as far as from 11th Avenue to Chicago, which appears to be about one-third of the way to Japan. The upshot is that Matthew finds waiting most painful at the beginning, since it feels longer.