18.1 When Human Beings Fail to Act the Way Economic Models Predict

The actors who populate the economic models we have seen in this book worry about one thing and one thing only: getting the most they can for themselves. Firms maximize profit; consumers maximize utility. They make rational tradeoffs that involve complicated sets of choices and require an impressive degree of computational ability.

Critics of the “normal” economic models have mockingly used the term Homo economicus to describe the creature that inhabits the economic world. They envision Homo economicus to be like a regular human being (Homo sapiens) except that he is able to solve any complicated economic problem, always knows exactly what he wants and how to get it, and never makes any mistakes. In other words, Homo economicus doesn’t seem much like Homo sapiens at all.

693

2Thinking, Fast and Slow by Daniel Kahneman

In this section, we explore five of the most important ways in which behavioral economics says Homo economicus differs from Homo sapiens.2

Systematic Bias 1: Overconfidence

overconfidence

A belief that one’s skill and judgment are better than they truly are, or that better outcomes are more likely to happen than their true probability.

One psychological bias that behavioral economists say real people suffer from is overconfidence. When individuals believe their skill level and judgment are better than they truly are, or when they expect that outcomes that are better for them are more likely to happen than they truly are, they will systematically make mistakes in a rational economic model that assumes people have realistic expectations and base decisions strictly on the facts.

When asked in surveys, 93% of American college students say that they are above-average drivers. It turns out that in their own minds, most people believe they are outstanding on many different dimensions, not just behind the wheel. Other studies have found that people overestimate their intelligence, sense of humor, and many other factors. On at least one Internet dating site, 77% of listed individuals (about equal fractions for men and women) rated themselves as having either “very good” or “above-average” physical attractiveness. In fairness, perhaps they don’t believe their attractiveness is superior, but are just saying this so they can get a first date and then wow their companions with their smarts and sense of humor.

Two economic settings in which overconfidence can play an important role are stock markets and futures markets. Every time someone makes a trade in one of these markets—selling 100 shares of Google, say, or a thousand barrels of oil—there must be another trader willing to take the other side of the transaction and buy the shares. The buyer must believe the value will rise, the seller that it won’t. One of them is wrong. If the person on one side of the trade believes that the other has better information or better judgment, he probably won’t want to make the trade. But, if each trader thinks he is smarter than everyone else, he will want to make more trades.

Likewise, company managers confident in their own abilities will be inclined to make bigger investments and take more risks on the overconfident view that they will succeed.

How Economic Markets Take Advantage of Overconfident People Having a psychological bias opens a person up to being taken advantage of in the marketplace by more rational players. The health club industry keenly understands and takes advantage of people who suffer from overconfidence bias. When individuals sign up for a health club, they tend to be far too optimistic about the prospects of sticking to their exercise goals; they use the health club much less, on average, than they expect to at the outset. Aware of this, health clubs tailor their offerings to exploit such overoptimism. For example, suppose individuals who work out frequently are willing to pay $100 a month for a health club membership, but a membership is only worth $20 a month for those who work out infrequently. Having just made a New Year’s resolution to be more fit in the upcoming year, overly optimistic consumers believe they will go to the gym frequently over the next year. And, indeed, some people will stick to their resolution for the entire year. Most, however, will be heavy users of the gym for only a few months (or days).

In light of this, rational health club owners detected a business strategy. They could have priced their product the way most goods are priced: per unit like a gallon of milk or a ticket to the movie. But health clubs don’t price that way—they don’t usually allow customers to pay, say, $10 for each visit to the club. Instead, health clubs usually require customers to sign a long-term contract that involves a big up-front payment, sometimes with a very short trial period.

694

3Stefano DellaVigna and Ulrike Malmendier, “Paying Not to Go to the Gym,” American Economic Review 96, no. 3 (2006): 694 – 719.

If every customer overconfidently believes he will be a heavy gym user after a week of working out, he might be willing to make an up-front payment of $1,200 for a membership that allows him unlimited use of the gym for two years. That type of contract extracts a lot more revenue from overly optimistic consumers than charging them per visit. Consumers may be willing to pay huge up-front fees even though they would have been better off paying even very high prices per trip to the gym because of how infrequently they end up going. Behavioral economics researchers have documented exactly this situation for health club businesses.3

Systematic Bias 2: Self-Control Problems and Hyperbolic Discounting

hyperbolic discounting

Tendency of people to place much greater importance on the immediate present than even the near future when making economic decisions.

Another psychological bias highlighted by behavioral economics is people’s strong desire to have things NOW and the difficulties that poses for self-control. We discussed discounting the future in Chapter 14, where we learned about net present value, and in Chapter 12’s discussion of repeated games. In those discussions, we used a simple discounting framework in which people discount the future by a set percentage each period, such as 10% per year (i.e., they viewed a payment of $1 a year from today as equivalent to $0.90 today). Behavioral economists argue that although people may think that way about decisions well into the future, they seem to value the here and now by much more than the basic discount rate would suggest. This behavior is called hyperbolic discounting—where people tend to prefer immediate payoffs to later payoffs, even if the later payoff is much greater.

One way to understand this is to think about a person deciding to buy a Ronco Smart Juicer as seen on TV. It costs either $200 up-front or four (easy) monthly payments of $65 each. Either way, Ronco sends the juicer to the consumer upon purchase; there’s no four-month delay if you chose the installment payment plan. A fully rational consumer (one who conforms to economic models’ assumptions of rationality) buying the juicer would do a net present value calculation discounting at a discount rate of, say, 5% per month (this is really high but it makes our point easier to see than if we use a lower discount rate). The cost of the juicer in one payment is $200. The four undiscounted payments of $65 total $260, but the NPV of the four payments is 65/(1.05) + 65/(1.05)2 + 65/(1.05)3 + 65/(1.05)4 = $230.49. Because the NPV of the four payments ($230.49) is higher than the up-front cost ($200), the consumer should pay for the juicer in full upon purchase.

However, because hyperbolic discounters really, really value things that happen now, they have a huge discount rate between now and even the near future. Another way to say this is they need to earn a much higher interest rate to make it worth their while to postpone their instant gratification. In our example, let’s say our hyperbolic discounting consumer discounts the first month at 25%, and the remaining three months at 5%. The calculation then involves an NPV of 65/(1.25) + 65/(1.25)(1.05) + 65/(1.25)(1.05)2 + 65/(1.25)(1.05)3 = $193.61, less than the $200 up-front cost. Someone who really wants things now will therefore opt for the installment plan because it allows him to have the juicer and keep more of his money today. The hyperbolic discounter’s very high discount rate in the first period makes paying the full $200 immediately very painful because he wants to have that money available to buy other things (that he also wants NOW).

Note that hyperbolic discounting is more than just having a high discount rate. Regular discounting applies the same discount rate between any two periods that are equally spaced. For example, consumers who discount the future in the regular way apply the same discount rate between now and one period in the future as they do between one period in the future and two periods in the future. (In the Ronco example, this rate was 5% per month.) Hyperbolic discounting specifically has different discount rates between now and the near future (25% in the Ronco example) and between two future periods (5%).

695

time-consistent

Consistencies in a consumer’s preferences in a given economic transaction, whether the economic transaction is far off or imminent.

One of the problems that this dual-discount aspect of hyperbolic discounting raises is that consumers stop being time-consistent. That is, consumers today will specify their preferred actions now, next year, and the year after that, but then when they get to next year (year 2), they won’t want to stick with the plan they set up. When a decision is a year off in the distance, waiting an extra day might not seem like a big deal for a hyperbolic discounter. But after a year passes, and that same wait is between today and tomorrow, the waiting feels much more costly. What people think they will do two years from now, and what they actually do, no longer match up, even if the world two years from now ends up being exactly how they imagined it.

When people are not time-consistent, analyzing their behavior becomes complicated. The person you are today is different from the person you will be tomorrow, and that fact turns self-control issues into a game theory battle between two different people. “Today you” wants to save a lot of money next year, but “next year you” would rather spend the money and put off saving one more year. If you understand that you are like this, “today you” (who wants to save next year) can take an action that commits your future self (who, next year, wants to spend) to save. You could sign up with your employer to automatically take money out of your salary for a retirement account starting next year. You could also write out a check to an organization you do not like and hand it to a friend who promises to mail it to the organization if you don’t follow your plan (and who will rip up the check if you do). People use less extreme versions of these sorts of commitment devices all the time. Smokers might throw away their cigarettes or dieters their ice cream, so their future self won’t be tempted.

How Economic Markets Take Advantage of People with Self-Control Problems Because hyperbolic discounters put too much weight on the present, they make choices they often regret in the long run, such as not studying hard enough for their final exam in economics, having unprotected sex, or not saving enough for the future.

Clever marketers often try to take advantage of people’s propensity for impulsiveness. Furniture stores let you sit on the futon and offer “no money down” purchases that end up costing more over time. During the housing bubble of the 2000s, many people refinanced their mortgages but with the added kicker that the bank allowed them to take a significant amount of cash out of the home, thus increasing the debt on the house until it equaled the entire value of the house. Some banks went even further, allowing homeowners to go “underwater”: that is, take out enough cash to raise their debt above their houses’ values, hoping or assuming that house values would continue to rise. Credit card companies send out offers of 0% interest for the first 30 days, low minimum payments on your monthly bill, and “free” balance transfers from other cards. In each case, the cards make it as easy as possible for consumers to obtain money or reduce the costs of borrowing right now in exchange for higher costs they will have to pay in the future.

As you might imagine, if you build an economic model in which some people or firms are hyperbolic discounters who put great weight on the present and others are fully rational economic agents who discount at “normal” rates, after a while, the rational discounters often end up with all the money (because they are patient and keep building up their savings year after year), and those who “want it now” end up broke (because they want everything right away and spend all their money each year). So, if you believe you are prone to impulsive hyperbolic discounting behavior, watch out. Try to find ways to stick to your financial plans and avoid temptations.

696

Systematic Bias 3: Falling Prey to Framing

Another systematic bias in decision making is that people often make incompatible (and thus irrational) decisions depending on how a decision or problem is framed. The disease treatment program example at the opening of the chapter was a case in point, but there are many types of framing biases.

endowment effect

The phenomenon where simply possessing a good makes it more valuable; that is, the possessor must be paid more to give up the good than he would have paid to buy it in the first place.

One is the endowment effect, which occurs when the pain a person suffers when giving up something they already have is greater than the pleasure they gained in acquiring it. An example will clarify how the endowment effect works. Suppose a professor decides to give every student in class a gift. Half of the class gets coffee mugs and the other half gets chocolate bars. The gifts are of equal average value and distributed randomly, so some students who prefer candy will end up with mugs and vice versa. Conventional economic models tell us that allowing the students to trade in a free market will increase total welfare (we discussed this in Chapter 15). Because there are an equal number of identically valuable gifts and because they were randomly assigned, something like half the students should want to swap gifts (because we expect half the students to prefer candy and half to prefer mugs).

When researchers tested this and similar scenarios in real life, however, that’s not what happened. When behavioral economist Richard Thaler tried this experiment, only 15% of students (instead of around 50% as the theory predicts) were willing to trade what they had been given initially, regardless of what item that was. As soon as the students received the mug or the candy, they felt it was more valuable to them. Indeed, when asked if they would sell the coffee mug back to the experimenter at the retail price, many people refused to do so, even though they had never chosen in the past to buy such a coffee mug at the store. (Presumably, having passed on the mug at the store means they valued the mug less than its retail price.) Their perception of the utility they would obtain from the mug had been altered by their ownership of the mug. That is the endowment effect.

loss aversion

A type of framing bias in which a consumer chooses a reference point around which losses hurt much more than gains feel good.

The endowment effect is a special case of a broader pattern of behavior known as loss aversion. A loss-averse person has a reference point in mind that takes on special significance. He might choose this reference point based on any one of a number of reasons depending on the setting, such as how much he paid for a good the last time he bought it, how much income he thinks his neighbor earns, or the number of points he scored in yesterday’s basketball game. Whatever the reference point is based on, loss aversion implies that falling below that point (which is perceived by the person as suffering a loss) hurts worse than rising above it (perceived as a gain) feels good. So, for instance, if a restaurant raises prices and the consumer’s reference point is what he paid for the meal last time, he suffers a large drop in utility relative to how good he feels when the restaurant lowers its prices by the same amount.

This is not true for the consumers we model elsewhere in the book; they just care about the bundle of goods they are able to buy. A small price increase lowers those consumers’ utility levels by essentially the same amount as a small price reduction increases their utility levels. What makes loss aversion different is that consumers’ choices depend on where they start. When a loss-averse person is just above the reference point, he is likely to act as if he is very risk-averse. But if he is just a little below the reference point, he may very well act like he’s risk-loving in order to try to get back to the reference point.

anchoring

A type of framing bias in which a person’s decision is influenced by specific pieces of information given.

4Amos Tversky and Daniel Kahneman, “Judgment under Uncertainty: Heuristics and Biases,” Science 185 no. 4157 (1974): 1124-1130.

Another framing bias is anchoring. People tend to base their decisions on the pieces of information they are given. When asked what percentage of countries in the United Nations were in Africa, for example, people responded with a smaller number, on average, if the question were “Is it more than 10%?” rather than “Is it more than 65%?”4 In a market transaction, anchoring bias means that a consumer’s willingness to pay for a kayak, say, is higher if he first sees several high-priced items that frame the kayak purchase and raise his willingness to pay. Thus, the presentation of the products and their prices influences a consumer’s choice, whereas conventional demand theory says that the willingness to pay comes straight from the consumer’s tastes. Framing and labeling are not supposed to matter.

697

mental accounting

A type of framing bias in which people divide their current and future assets into separate, nontransferable portions, instead of basing purchasing decisions on their assets as a whole.

A final form of framing bias is mental accounting, which occurs when individuals divide their current and future assets into separate, nontransferable portions. In traditional consumer behavior theory, a consumer makes rational decisions about savings or about buying various products based on his income, on prices, and on other factors that are part of his utility function. In contrast to this rational consumer, some behavioral economists claim that people make mental accounts in which they divide up their money and their purchases. Instead of considering “savings” as one big category, for example, people keep mental accounts for college money, vacation money, and retirement money, and act as if moving funds between these accounts is difficult or impossible. For consumption, they may have mental accounts for monthly spending on gasoline, clothing, and food instead of one account for “consumption.”

Mental accounting can also apply to different sources of income. If a person finds a $20 bill on the floor and goes out and spends this windfall but saves most of a $20 tip he receives at work, he is thinking like a mental accountant. The standard economic model would say that a $20 bonus is a $20 bonus. It doesn’t matter whether it came from the floor or from a grateful customer.

How Markets Take Advantage of People Who Fall Prey to Framing The number of ways a clever marketer can use a person’s framing bias to make money is almost unlimited. Some cynics might even say that knowing how to do this is the very definition of clever marketing.

To take advantage of people with the endowment effect bias, a firm might offer a money-back guarantee. Before buying a product, the fully rational consumer isn’t always sure whether he will like it, so the option to return it and get his money back has value—the guarantee makes it more likely that he will buy the product to try it out. After buying the product, though, a customer with an endowment bias will experience an exaggerated increase in the product’s value to him, so he will be very unlikely to return it for a refund, even if he is not fully satisfied.5

To take advantage of people with an anchoring bias, a firm might artificially inflate the base price of a good and then advertise a “50 percent off” sale. By anchoring in the consumer’s mind the idea that the good is worth the original inflated price, the half-price good looks like a bargain, even though the “50 percent off” price is what the good normally would have sold for anyway.

A buyer who is a mental accountant must beware of the tactic often used by salespeople in car showrooms across the world: “How much do you want to pay for a car?” Once you put an amount of money into a mental account, you can be fairly sure that a smart seller will find a way to withdraw it for you.

698

Systematic Bias 4: Paying Attention to Sunk Costs

One of the most important themes in economic decision making is that sunk costs do not matter when making a decision. The money is already spent and it cannot be recovered, so it shouldn’t matter for the decision. Rational decision makers think at the margin and only consider opportunity costs. (To review these ideas, see Chapter 7.)

sunk cost fallacy

The mistake of allowing sunk costs to affect decisions.

6Hal R. Arkes and Catherine Blumer, “The Psychology of Sunk Cost,” Organizational Behavior and Human Decision Processes 35, no. 1 (1985): 124 – 140.

In reality, however, behavioral economists say that people are often influenced by sunk costs when making decisions. A classic example of the sunk cost fallacy comes from an experiment published in 1985 by psychologists Hal Arkes and Catherine Blumer.6 The researchers worked out a deal with the theater at Ohio University so that when a person arrived at the ticket window and asked to buy a season ticket (for 10 plays), the buyer was randomly assigned a price for the season ticket. A third of the buyers paid the full price of $15 per play for the season ticket. A second group paid $13 per play, and a third group paid just $8 per play.

Given that the ticket holders all had tickets in hand on the morning of a performance, we should be able to assume that how much people paid for the tickets had no bearing on the size of the marginal benefit from attending the plays. The price of the ticket was a sunk cost, and the marginal cost of attending was just the value of their time. No matter what the purchasers had paid for the ticket, the opportunity cost of attending the play was the same; there was no additional expense. And yet, in the first half of the season, the people who had paid full price attended about 25% more of the plays than did the groups that received discounts. Behavioral economists argued it was because the high-price group felt a greater obligation to attend so that they could get their money’s worth from the tickets. This result illustrates that people just can’t seem to ignore sunk costs, even though they rationally should.7

8Allan I. Teger, Too Much Invested to Quit, Oxford: Pergamon Press, 1980.

Companies and governments have made similar sunk cost mistakes. One example involves the development of the Concorde supersonic jet, which was jointly developed by the British and French governments. The joint venture began in the early 1960s, amidst great optimism. Plans went quickly awry, however. Ultimately, development costs ended up being 6 times higher than projected, cost overruns that were anticipated far in advance. By virtually all accounts, the right course of action would have been to stop the project, but the governments forged ahead because there was “too much invested to quit,” as described in Allan Teger’s 1980 book on the subject.8 Ultimately, only 20 of the airplanes were ever made, and the last plane was taken out of service in 2003.

Application: The Sunk Cost Fallacy and Professional Athletes

National Football League quarterback Robert Griffin III had already built up an impressive résumé by the time he joined the Washington Redskins in 2012. A natural athlete, in high school he had excelled in football, basketball, and track. He even qualified to compete at the 2008 U.S. Olympic Trials in track and field, but later gave up his dream of becoming a track superstar to pursue a career in football. By the end of his playing days at Baylor University, he had won the Heisman Trophy (awarded to the most outstanding player in college football) and was the number two pick in the 2012 NFL draft. The Washington owners were willing to give a lot to snag Griffin as their quarterback in the draft. They sacrificed three future years of first-round draft picks and one second-round draft pick, a huge outlay for a draft player, on top of a $21 million, four-year contract.

During the 2012 regular season, even this high price looked like a bargain. Griffin was a phenomenon on the field, winning the Offensive Rookie of the Year award. Washingtonians went crazy over their new quarterback; some enthusiastic Redskins fans even found out at what online sites he and his fiancée had registered and started sending wedding gifts. That turned out to be the high point, however. In the playoffs following the 2012 season, Griffin injured his knee and his decline began. Since then he has faced additional injuries and has had trouble picking up newer elements of the team’s offensive system. Washington won only 7 games and lost 25 in Griffin’s second and third seasons with them.

Early in the 2014 season, Griffin’s performance had degraded to the point where it appeared that the Redskins’ backup quarterback Kirk Cousins would have been more effective than Griffin. Yet team management was not willing to let Cousins take Griffin’s place as the regular starter. They seemed to be falling prey to the sunk cost fallacy.

The fact that Washington had sacrificed so many of their future draft picks to acquire Griffin in 2012 should not have influenced their decisions in 2014. They couldn’t go back and change their minds, and there was no way the team could recover the expense. Moving forward, then, any decision made by the team about Griffin should have considered only the future costs and benefits. More directly, the price paid during the 2012 draft shouldn’t have factored into the head office’s decision about whether to play Cousins or Griffin. Yet the Redskins’ behavior and statements suggested that they believed it important to give Griffin more of a chance than they otherwise would have, simply because they had paid so much for him in the first place. They don’t seem to be changing their minds, either; in the early summer of 2015, team management announced they would pick up Griffin’s $16 million fifth-year option on his contract, declaring the decision “a no-brainer.”

If the Redskins were using their brains like economists, they would realize that starting the quarterback who was most likely to win games, irrespective of what they had paid for that player, is the right choice. Big purchases allow teams to reap the rewards when a player fulfills his potential, but stories like that of Robert Griffin III illustrate how such purchases can also lead teams to place undue weight on sunk costs and, in turn, to continue investing in a player even when it is no longer profitable to do so.

699

How Markets Take Advantage of Loss-Averse People Attentive to Sunk Costs At the most basic level, the market will leave the losses and the carrying costs in the hands of those who are suffering from the sunk cost fallacy. In the Application above, for example, it’s the stubborn Washington Redskins who suffer: They’re stuck losing games they might have won because they refuse to change quarterbacks. Market participants who do not suffer from the sunk cost bias can end up benefiting from the mistakes of people who do.

Systematic Bias 5: Generosity and Selflessness

altruism

Acts motivated primarily by a concern for the welfare of others.

Economic models are premised on the belief that consumers and firms act in rational self-interest. Perhaps one of the most basic challenges to this assumption is that people often engage in non-utility-maximizing acts of generosity, selflessness, and altruism—acts for which the welfare of others is the primary concern. For example, parents sacrifice for their children; volunteers work long, hard hours for various causes; military personnel fight for their country; and so on. Economists have tried to incorporate this behavior into standard models by adding a taste for generosity (called the “warm glow of giving”) into the utility function or by letting a parent’s utility depend not only on his own consumption but also on his children’s consumption. These are some of the ways to take into account the generosity bias when modeling consumer behavior. But, these methods don’t address the fundamental issue that, by frequently helping others at no benefit to themselves, people don’t always appear to act as rational economic agents.

700

While altruistic behavior almost surely exists in many settings, interestingly (or sadly, depending on your perspective), in the debate to explain acts of generosity and selflessness, many economists have detected the aroma of self-interest or other conventional economic explanations for behaviors that might otherwise be classified as completely selfless, such as charitable giving.

9Jonathan Meer and Harvey S. Rosen, “Altruism and the Child Cycle of Alumni Donations,” American Economic Journal: Economic Policy 1, no. 1 (2009): 258 – 286.

As an example, let’s consider alumni donations to colleges and universities in the United States. Alumni generosity accounts for a tremendous amount of the budget of many educational institutions. The Council for Aid to Education reports that donations totaled more than $38 billion in 2014. We normally think that alumni donors are motivated to give back to their institutions in thanks for what those schools did for their lives. But, when economists Jonathan Meer and Harvey Rosen began to examine the donations to a large private university, contributions looked a bit less selfless than at first glance.9 Alumni with children were dramatically more likely to give than alumni without them. The donations increased significantly when the child reached age 14 and continued to increase through the period when the child applied to college. By age 18 or 19, if the alum’s child is admitted to the same college, donations were more than 10 times higher than for alumni without children. For an alumnus whose child applied but was rejected by the college, donations fell precipitously and were no higher than donations from alumni without children. The seemingly selfless behavior of donating to a nonprofit university seemed instead to be directly influenced by the self-interest of getting one’s child into a good school.

10John A. List, Stefano DellaVigna, and Ulrike Malmendier, “Testing for Altruism and Social Pressure in Charitable Giving,” Quarterly Journal of Economics 127, no.1 (2012): 1–56.

Another look at generosity may be found in a study by John List, Stefano DellaVigna, and Ulrike Malmendier.10 They sent out volunteers to knock on doors and solicit charitable contributions. Many people gave donations, consistent with altruism. But in one part of the experiment, List and his cohorts asked their solicitors to first put flyers on the front doors of those houses they intended to visit, to let the occupants know that someone would be coming by the next day at a specific time to solicit contributions. If the residents were indeed altruistic, one would expect them to make a special effort to be home to contribute to the charity. Instead, List, DellaVigna, and Malmendier found just the opposite. When warned that the solicitors were coming, many people made a point of not being home or else refused to open the door even though they would have likely given a contribution to someone who showed up at their door unannounced. In this experiment, people seemed mainly to contribute not out of altruism, but because of the social pressure associated with not wanting to look selfish when asked for a small donation.

Because economic models assume that rational decision makers are self-interested, the overall subject of charitable giving and generosity poses a systematic puzzle to standard economic models. These examples demonstrate, however, that in many circumstances there still seems to be an important role for self-interest in people’s behavior.

701

Application: Do Animals Care about Fairness?

One kind of altruism that humans sometimes appear to exhibit is an inherent preference for fairness. That is, people will sometimes take costly actions that aren’t profitable from a standard cost–benefit standpoint (at least one that ignores social returns or losses of such actions) in order to get to an outcome that would be viewed as more fair.

Consider, for instance, individuals’ responses in the following game, often called “the ultimatum game.” One player, the proposer, is given an amount of money by the experimenter and then instructed to offer whatever portion of it she wishes to an anonymous second player, the recipient. The recipient then decides whether to accept or reject this offer. If the recipient accepts the offer, each player gets the portion of money specified in the proposer’s offer. If the recipient instead rejects the offer, neither player gets anything.

The traditional models we have been working with throughout this book would predict that the proposer will offer a minimal amount, say, 1 cent, the recipient accepts this offer, and the proposer ends up with virtually all of the initial money. We can use backward induction to see why. Think about the recipient’s choice of whether to accept an offer once it has been made. Because the recipient gets nothing if he rejects the offer, he should be willing to accept any offer greater than zero—something, no matter how small, is better than nothing. Knowing this will be the case, the proposer should offer the recipient a tiny but positive amount (1 cent), keeping almost all of the cash for herself.

It turns out, however, that people who play this game in lab experiments deviate far from the outcomes predicted by backward induction. On average, proposers offer between 40% and 50% of their endowment, and recipients often reject offers that are below 20% of the endowment. It seems that recipients see low offers as being somehow unfair, and they are willing to give up some cash of their own (whatever was offered to them by the proposer) to punish the proposer for selfish behavior. Proposers, knowing this, make more generous offers for fear of getting nothing because a recipient has rejected an offer viewed as unfair.

11Darby Proctor, Rebecca A. Williamson, Frans B. M. de Waal, and Sarah F. Brosnan, "Chimpanzees Play the Ultimatum Game," Proceedings of the National Academy of Sciences 110, no. 6 (February 5, 2013): 2070–2075.

Scientists have wondered if this behavior is special to humans, or whether other animals also act in ways that indicate a sense of fairness. Researchers Darby Proctor, Rebecca Williamson, Frans de Waal, and Sarah Brosnan set out to explore this question with chimpanzees.11 They conducted a standard ultimatum game with chimpanzees, except rather than money, they gave proposer chimpanzees tokens that could be exchanged for bananas. (The chimps had already been trained to understand what the tokens could be used for and how the ultimatum game worked in their context.)

The chimpanzees matched human behavior in one respect. Proposer chimps offered something near an even split nearly 75% of the time, suggesting they valued fairness. However, unlike human recipients, recipient chimps never rejected an offer, no matter how small it was. Not rejecting lopsided offers seems to indicate these recipient chimps didn’t care about fairness. But if that’s the case, why were the proposers often making even-split offers?

image
“Crummy cucumbers. No fair. Grapes, we want grapes.”
Video still of the experiment at the Yerkes National Primate Research Center, courtesy Frans de Waal

It turns out that the recipients did value fairness; they just chose to express it in a different way than their human counterparts. Rather than reject offers they saw as unfair, chimp recipients sometimes responded to these offers with threats, such as spitting water at the proposer or pounding on the barrier separating the two chimps. The high number of offers with close to even splits, especially later on in the experiment as the proposers became familiar with the recipients’ typical reactions, suggests that proposers understood the point of these threats, even if their offers were never rejected.

702

12Sarah F. Brosnan and Frans B. M. de Waal, "Monkeys Reject Unequal Pay," Nature 425 (September 18, 2003): 297–299.

A second experiment by Sarah Brosnan and Frans de Waal, two of the authors of the chimpanzee study, conducted a somewhat similar experiment with pairs of female capuchin monkeys.12 Each of the two monkeys had its own cage, but the cages were made of plastic, so each could see what was happening in the other’s area. In one version of the experiment, both of the monkeys would receive a cucumber slice if they gave a researcher a token. In another version, one monkey would get a cucumber slice for a token, while the other would get a grape (capuchins like grapes a lot more than cucumbers). In a third version, one monkey would get a cucumber slice for a token, while the other received grapes without even having to exchange a token.

The researchers found that in the second and third versions of the experiment, the “cucumber monkeys” would sometimes refuse to accept the cucumber when the researcher offered it to them. (A video of the experiment shows one particularly irate monkey throwing the cucumber slice back at a researcher.) In other cases, the cucumber monkeys refused to even hand over the token in the first place. This refusal behavior happened more often the longer the experiment went on and was most pronounced in the version of the experiment where the other monkey received grapes for “free.” Because monkeys very rarely refuse food when not participating in this type of experiment, the researchers concluded that the refusals suggested a preference for fairness. The cucumber monkeys, unhappy with what they viewed as unfair treatment, were willing to express a preference for fairness, even if it meant giving up food they would normally eat.