Misbehaving: The Making of Behavioral Economics

by Richard H. Thaler

  Costco, the discount retailer mentioned in the previous chapter, also uses a version of this strategy. In order to shop at Costco a customer must become a “member,” which currently costs a household $55 a year. It seems likely that members view the annual fee as an “investment” and make no attempt to allocate that cost over the various purchases they make during the year. Rather, it serves as a sunk cost, offering up yet another reason to shop at Costco. Similarly, Amazon charges customers $99 a year to become a “prime member” which entitles them to “free” shipping. Again, the cost of the membership may well be viewed as an investment that does not “count” toward the cost of a particular purchase.

  It is time for two confessions. Although I mostly advocate for thinking like an Econ, when it comes to mental accounting I have some notably Human tendencies. I am usually pretty good about ignoring sunk costs, especially if the sunk costs are purely monetary in nature. But like most people, if I have put a lot of work into some project I can find it difficult to give it up, even if it is clearly the right thing to do. In writing this book, for instance, my strategy for getting a first draft done was to keep writing and not worry about whether a particular passage or section would make the final cut. This process did produce a first draft, but one that was obviously too long. Some bits were going to have to be cut, and I fielded suggestions for which parts to drop from my friends and editors who read the initial draft. Many mentioned the advice, often attributed to William Faulkner, but apparently said by many, that writers have to learn to “kill their darlings.” The advice has been given so often, I suspect, because it is hard for any writer to do.

  When it came time to revise the manuscript, I decided to create an “outtakes” file of material that was in the first draft but was cruelly murdered. My plan is to post some of these precious masterpieces of glorious verbiage on the book’s website. I don’t know how many of these passages will actually get posted, but the beauty of this plan is that it doesn’t matter. Merely having a place where these pieces are stored in a folder on my computer labeled “outtakes” has been enough to reduce the pain of cutting some of my favorite passages, a pain that can hurt as much as wearing those expensive, ill-fitting shoes. The bigger lesson is that once you understand a behavioral problem, you can sometimes invent a behavioral solution to it. Mental accounting is not always a fool’s game.

  My second confession regards wine, which, as you have guessed by now, is one of my vices. Although I fully understand the concept of opportunity cost, I admit to falling victim to a version of the same thinking articulated by the respondents to our questionnaire. If I take out an old bottle that I have stoically refrained from drinking for many years, the last thing on my mind is the price I could get for the wine if I were to sell it at an auction. In fact, I do not want to know that price! I end up like Professor Rosett. I would not dream of buying a thirty-year-old bottle of wine, but I am happy to drink one on special occasions. Just call me Human.

  ________________

  * The song is actually titled “Waist Deep in the Big Muddy,” and the lyrics illustrate the concept of escalation quite vividly as the verses go from knee-deep to waist-deep to neck-deep.

  † From early on Orley has been a supporter of me and my misbehaving fellow travelers, including during his tenure as the editor of the American Economic Review. Nevertheless, to this day, Orley insists on calling what I do “wackonomics,” a term he finds hysterically funny.

  ‡ There is an interesting side note to this experiment. The lottery offering up the $75 bottle of wine generated 178 respondents from a relatively affluent group of readers. That is 42 cents per reply, and they had to pay their own postage! If you want to get people to do stuff, lotteries can be very effective motivation.

  9

  Buckets and Budgets

  In those interviews with families that I used to inform my thinking about how households manage their finances, we learned that many households, especially those on a tight budget, used explicit budgeting rules. For families that dealt mostly in cash (credit cards were just coming into use at this time in the late 1970s), many would often use some version of an envelope system. One envelope (or mason jar) for rent, another for food, another for utilities, and so forth. In many cases the particular method used was one they had learned from their parents.

  Organizations do something similar. Departments have budgets, and there are limits for specific categories within those budgets. The existence of budgets can violate another first principle of economics: money is fungible, meaning that it has no labels restricting what it can be spent on. Like most economic principles, this has strong logic behind it. If there is money left over in the utilities budget because of a mild winter, it will spend perfectly well at the children’s shoe store.

  Budgets exist for sensible, understandable reasons. In an organization, the boss does not want to have to approve every expenditure made in the organization, and budgets serve as a crude way to keep costs under control while giving employees discretion to spend as they see fit. Still, budget rules can lead to silly outcomes. Anyone who has worked in a large organization has run into the problem where there is not enough money in the assigned budget to take care of some urgent need, and there is no way to dip into money sitting idle in another budget. Money should be spent in whatever way best serves the interests of the organization or household; if those interests change, we should ignore the labels that were once assigned to various pots of money. But we don’t. Labels are SIFs.

  Individuals and families set their own rules, of course, but they use budgets in much the same ways. Just how explicit the budgeting rules are will often depend on how much slack is in the budget. A study by psychologists Chip Heath and Jack Soll found that most MBA students had weekly food and entertainment budgets and monthly clothing budgets. Once they had graduated and started earning more money, these budgets probably became more relaxed.

  But while in graduate school, the budgets and the resulting violations of fungibility influenced their behavior. For example, Heath and Soll asked two groups of subjects whether they would be willing to buy a ticket to a play on the weekend. One group was told they had spent $50 earlier in the week going to a basketball game (same budget) while another group was told they had received a $50 parking ticket (different budget) earlier in the week. Those who had already gone to the game were significantly less likely to go to the theater, presumably because their entertainment budget for the week was already spent.

  A study by economists Justine Hastings and Jesse Shapiro offers the most rigorous demonstration of the effects of mental budgeting to date. The question Hastings and Shapiro investigated is what happens to the choice of regular versus premium gasoline when the price of gasoline changes. In the United States, gasoline is typically sold in three grades based on octane: regular, midgrade, and premium. Although a question remains whether any car really requires something other than regular, a higher grade is recommended for some models, and some consumers buy a higher grade for other reasons, such as the probably erroneous belief that it is better for the engine. The authors studied what happened to the sales of premium grades of gasoline when the price of gasoline fell in 2008 by roughly 50%, from a high of about $4 a gallon to a low just below $2. Hastings and Shapiro were able to study this because they had customer purchase data from a grocery store chain that also sold gasoline.

  Let’s first think about what an Econ would do in this situation. Suppose a household is spending $80 a week on gasoline when the price is $4 and is buying the regular grade. Six months later the price has dropped to $2 and the household’s cost has dropped to $40 a week. An Econ would think this way: First, gasoline is cheaper, so we should take more road trips. Second, we have gained the equivalent of $40 a week in take-home pay, and we can spend that on anything we want, from more date nights to higher quality beer. The $40 in extra income would be spent in the way that maximizes utility. Some of that money might be spent on improving the grade of gasoline,
but only a minuscule amount. On average, if a family’s income goes up by $1,000 a year, their propensity to buy something other than regular grade gasoline increases by only 0.1%. So a family of Econs might decide to treat their car to one tank a year of mid-grade gas, and spend the rest of their windfall on things more valuable.

  Suppose instead a family of Humans has a gas budget, possibly kept in a mason jar in the kitchen. Like the Econ family, they will spend some of that money on taking more road trips, but they might also think, hey, gasoline is so cheap now I might as well buy the good stuff. That is exactly what Hastings and Shapiro found. The shift toward higher grades of gasoline was fourteen times greater than would be expected in a world in which money is treated as fungible. Further supporting the mental accounting interpretation of the results, the authors found that there was no tendency for families to upgrade the quality of two other items sold at the grocery stores, milk and orange juice. This is not surprising, since the period in question was right at the beginning of the financial crisis of 2007, the event that had triggered the drop in gas prices. In those scary times, most families were trying to cut back on spending when they could. The one exception to that tendency was more splurging on upscale gasoline.

  Wealth, too, is often separated into various mental accounts. At the bottom of this hierarchy sits the money that is easiest to spend: cash. There is an old expression that money burns a hole in your pocket, and cash on hand seems to exist only to be spent.

  Money in a checking account is slightly more out of reach than cash, but if there is money in an account labeled “savings,” people are more reluctant to draw that money down. This can lead to the odd behavior of simultaneously borrowing at a high rate of interest and saving at a low rate, for example by keeping money in a savings account earning virtually no interest while maintaining an outstanding balance on a credit card that charges interest at more than 20% per year. There is what seems to be an obvious financially attractive opportunity, which is to pay off the loans with the savings. However, people may be anticipating that the strategy will backfire if they never repay the money “borrowed” from the savings account.

  The most sacred accounts are long-term savings accounts, generally those dedicated for future spending, such as retirement accounts or children’s education accounts. While it is true that some people do borrow from retirement savings accounts such as 401(k) plans, typically these loans are relatively small and are repaid within a few years. More dangerous to the accumulation of wealth than loans are job changes. When employees switch jobs they are often offered the chance to take their account balance in cash. Even though such cash-outs are taxable income and are subject to a 10% surcharge, many employees take the money, especially if their balance is small. This leakage can and should be addressed by making the option of rolling the account over into another retirement account as easy as possible, preferably the default.

  Home equity offers an interesting intermediate case. For decades people treated the money in their homes much like retirement savings; it was sacrosanct. In fact, in my parents’ generation, families strived to pay off their mortgages as quickly as possible, and as late as the early 1980s, people over sixty had little or no mortgage debt. In time this attitude began to shift in the United States, partly as an unintended side effect of a Reagan-era tax reform. Before this change, all interest paid, including the interest on automobile loans and credit cards, was tax deductible; after 1986 only home mortgage interest qualified for a deduction. This created an economic incentive for banks to create home equity lines of credit that households could use to borrow money in a tax-deductible way. And certainly it made sense to use a home equity loan to finance the purchase of a car rather than a car loan, because the interest was often lower as well as being tax deductible. But the change eroded the social norm that home equity was sacrosanct.

  That norm was eventually destroyed by two other factors: the long-term decline in interest rates and the emergence of mortgage brokers. In the past three decades, interest rates in the United States have declined from double digits to essentially zero (or less, if you adjust for inflation). Adding mortgage brokers to the mix then proved fatal to the old, unwritten eleventh commandment: “Though shalt pay off thy mortgage.” The role these brokers played in eroding the norm of paying off mortgages as soon as possible was to make the process of refinancing much easier. They had the relevant information in their computers, and with interest rates dropping, they had numerous opportunities to call and say, “Hey, do you want to lower your mortgage payment?” When the housing bubble arrived and drove up prices, homeowners were told they could lower their mortgage payment and take out a bit of extra cash too, to refinish the basement and buy a big-screen television.

  At this point, home equity ceased to be a “safe” mental account. This fact is illustrated by a change in the borrowing behavior of households with a head that is aged seventy-five or older. In 1989 only 5.8% of such families had any mortgage debt. By 2010, the fraction with debt rose to 21.2%. For those with mortgage debt, the median amount owed also rose over this period, from $35,000 to $82,000 (in 2010 dollars). During the housing boom in the early 2000s, homeowners spent the gains they had accrued on paper in home equity as readily as they would a lottery windfall.

  As documented in House of Debt, a book by economists Atif Mian and Amir Sufi, by 2000 increases in home equity had become a strong driver of consumption, especially of consumer durables. For example, in cities where house prices were booming, automobile sales also jumped, as homeowners borrowed against the increased equity in their home and used the proceeds to finance a new car. Then on the way down, the reverse happened; automobile sales crashed along with home prices, as there was no way to finance the new car purchase if a homeowner had zero home equity or was “underwater,” meaning that the outstanding mortgage exceeded the value of the home. This phenomenon helps explain why the burst of the tech bubble in 2000–01 did not cause the same deep recession as the pricking of the housing bubble. Most non-wealthy households only hold stocks in their retirement accounts, which are still relatively sticky places to keep their money, especially for those with non-trivial account balances. This means that the fall in stock prices did not impact spending as much as the fall in home prices.

  It remains to be seen whether the norm of paying off the mortgage before retirement will ever reemerge. If the long-expected trend of rising interest rates ever gets started, we may see people resume the habit of paying off their mortgage because refinancing will be less enticing at higher rates. Otherwise, home equity might remain a leaky bucket.

  Like most aspects of mental accounting, setting up non-fungible budgets is not entirely silly. Be it with mason jars, envelopes, or sophisticated financial apps, a household that makes a serious effort to create a financial plan will have an easier time living within its means. The same goes for businesses, large or small. But sometimes those budgets can lead to bad decision-making, such as deciding that the Great Recession is a good time to upgrade the kind of gasoline you put in your car.

  10

  At the Poker Table

  During my time at Cornell, a group of economics faculty members met periodically for a low-stakes poker game. It was rare that anyone won or lost more than $50 in an evening,* but I noticed that some players, in particular ones who reported the game’s outcome to their spouse, behaved differently when they were winning versus when they were losing. How you are doing in the game that night, especially for stakes so small relative to net worth, should be irrelevant to how a particular hand is played. Compare someone who is down $50 in that night’s poker game to another who owns 100 shares of a stock that was down 50 cents at market close. Both have lost a trivial portion of their wealth, but one of the losses influences behavior and the other does not. Losing money in the poker account only changes behavior while you are still playing poker.

  This situation, in which a person is “behind” in a particular mental account, is tricky to handle in
prospect theory, something Kahneman and Tversky knew well. In their original paper, they discussed a similar case at the racetrack. Because the track takes about 17% of each dollar wagered, bettors at the racetrack are collectively losing money at a rate of 17% per race. When the last race of the day comes along, most of the bettors are in the red in their racetrack mental account. How does this affect their betting? The normative prediction is “hardly at all.” Just as with the poker game example, a bettor should be no more concerned with a loss of $100 in bets at the racetrack than he would be by a similar loss in his retirement savings account, which would go unnoticed. Yet Danny and Amos cite a study showing that the odds on long shots (horses with little chance of winning) get worse on the last race of the day, meaning that more people are betting on the horses least likely to win.

  Kahneman and Tversky explained this finding by relying on the feature of prospect theory that people are risk-seeking when it concerns losses. As was discussed in chapter 4, if you ask people whether they would rather lose $100 for sure or choose a gamble in which they have a 50% chance of losing $200 and a 50% chance of breaking even, a majority will choose the gamble. These results are the opposite of those found when the choice is between a guaranteed gain of $100 and a 50-50 gamble for $0 or $200, where people prefer the sure thing.

  As I watched how my poker buddies played when they were behind, I realized that the Kahneman and Tversky explanation was incomplete. Suppose I am down $100 at the racetrack and I would like to get back to zero to avoid closing this account in the red. Yes, I could bet $2 on a 50-to-1 long shot and have a small chance of breaking even, but I could instead bet another $100 on an even-money favorite and have a 50% chance of breaking even. If I am risk-seeking (meaning that I prefer a gamble to a sure thing that is equal to the expected outcome of the bet) why don’t I make the $100 bet on the favorite and improve my chances of breaking even? Prospect theory is silent on this question, but my poker experiences suggested that Amos and Danny had the right intuition. My impression was that players who were behind were attracted to small bets that offered a slim chance for a big win (such as drawing to an inside straight) but disliked big bets that risked a substantial increase to the size of their loss, even though they offered a higher probability of breaking even.