Nudge: The Final Edition

by Richard H. Thaler

We have also added a new chapter on what we call sludge, which is nasty stuff that makes it more difficult to make wise choices. (Sludge is everywhere; you’ll see.) The use of Smart Disclosure is one way to reduce sludge. So is sending everyone a tax return that has already been completed and can be filed with one click. So is reducing the length of those forms you have to fill out to get licenses, permits, visas, health care, or financial aid, or to get reimbursed for a trip you take for your employer. Every organization should create a seek-and-destroy mission for unnecessary sludge.

A nudge, as we will use the term, is any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives. To count as a mere nudge, the intervention must be easy and cheap to avoid. Nudges are not taxes, fines, subsidies, bans, or mandates. Putting the fruit at eye level counts as a nudge. Banning junk food does not.

Those who reject paternalism often claim that human beings do a terrific job of making choices, or if not terrific, certainly better than anyone else would do (especially if that someone else works for the government). Whether or not they have ever studied economics, many people seem at least implicitly committed to the idea of Homo economicus, or economic man—the notion that each of us thinks and chooses unfailingly well, and thus fits within the usual depiction of human beings that is offered by economists.

Suppose a chess novice were to play against an experienced player. Predictably, the novice would lose precisely because he made inferior choices—choices that could easily be improved by some helpful hints. In many areas, ordinary consumers are novices, interacting in a world inhabited by experienced professionals trying to sell them things. More broadly, how well people choose is an empirical question, one whose answer is likely to vary across domains. Generally, people make good choices in contexts in which they have lots of experience, good information, and prompt feedback—say, choosing among familiar ice cream flavors. People know whether they like chocolate, vanilla, coffee, or something else.

Biased assessments of risk can perversely influence how we prepare for and respond to crises, business choices, and the political process. When technology stocks have done very well, people might well buy technology stocks, even if by that point they’ve become a bad investment. People might overestimate some risks, such as a nuclear power accident, because of well-publicized incidents such as Chernobyl and Fukushima. They might underestimate others, such as strokes, because they do not get much attention in the media. Such misperceptions can affect policy, because some governments will allocate their resources in a way that fits with people’s fears rather than in response to the most likely dangers.

But Sherif found big conformity effects when people were asked to act in small groups and to make their estimates in public. Here the individual judgments converged and a group norm, establishing the consensus distance, quickly developed. Over time, the norm remained stable within particular groups, thus leading to a situation in which different groups made, and were strongly committed to, quite different judgments. There is an important clue here about how seemingly similar groups, cities, and even nations can converge on very different beliefs and actions simply because of modest and even arbitrary variations in starting points.

An important implication is that if senior members of a group want to obtain the actual beliefs of their junior coworkers, they will ask for those beliefs independently (so coworkers don’t influence each other) and, most important, before they state their own opinion.

For those who want to enlist social influences, an important challenge, as well as a real opportunity, is pluralistic ignorance—that is, ignorance, on the part of all or most, about what other people think. We may follow a practice or a tradition not because we like it or even think it defensible, but merely because we think that most other people like it. Many social practices persist for this reason, which means a small shock, or nudge, can dislodge them.21 A dramatic example is Communism in the former Soviet bloc, which lasted in part because people were unaware of how large a share of the population despised the regime. As people became aware of what others actually thought, they were emboldened to say what they believed, and to act accordingly.

The discussion thus far suggests that people may most need a good nudge for choices that require memory or have delayed effects; those that are difficult, are infrequent, and offer poor feedback; and those for which the relationship between choice and experience is ambiguous.

If you are designing a drug and you have complete flexibility, how often would you want your patients to have to take their medicine? If we rule out a onetime dose administered immediately by the doctor (which would be best on all dimensions but is often technically infeasible), then the next-best solution is a medicine taken once a day, preferably in the morning. It is clear why once a day is better than twice (or more) a day, because the more often you have to take the drug, the more opportunities you have to forget. But frequency is not the only concern; regularity is also important. Once a day is much better than once every other day, because it can be incorporated into the morning routine. And since people tend to take more pills as they age, adding another is not a problem. (Pill container boxes, one for each day, are useful nudges.) By contrast, remembering to take your medicine every other day is beyond most of us. (Similarly, meetings that occur every week are easier to remember than those that occur every other week.)

When we face a small number of well-understood alternatives, we tend to examine all the attributes of all the alternatives and then make trade-offs when necessary. But when the choice set gets large, we must use alternative strategies, and these can get us into trouble.

Social science research reveals that as the choices become more numerous or vary on more dimensions, people are more likely to adopt simplifying strategies. The implications for choice architecture are related. As alternatives become more numerous and more complex, choice architects have more work to do, and are much more likely to influence choices (for better or for worse). For an ice cream shop with three flavors, any menu listing those flavors in any order will do just fine, and effects on choices (such as the order in which they are listed) are likely to be minor because people know what they like. As choices become more numerous, though, good choice architecture should provide structure, and structure will affect outcomes.

Perhaps the most basic principle of good choice architecture is our mantra: Make It Easy. If you want to encourage some behavior, figure out why people aren’t doing it already, and eliminate the barriers that are standing in their way. If you want people to obtain a driver’s license or get vaccinated, make it simple for them, above all by increasing convenience. Of course, this principle has an obvious corollary: if you want to discourage some behavior, make it harder by creating barriers. If you want to make it harder for people to vote, forbid voting by mail and early voting, and reduce the number of polling stations (and place them far away from public transportation stops). While you’re at it, try to make people spend hours in line before they can vote. If you don’t want people to immigrate to your country, make them fill out a lot of forms and wait for months for good news in the mail (not by email), and punish them for answering even a single question incorrectly. If you want to discourage poor people from getting economic benefits, require them to navigate a baffling website and to answer a large number of questions (including some that few people can easily understand).

The sludge in the tax system can come simply from the straightforward task of attempting to follow the law as written, but it can also extend to the legal efforts to minimize one’s tax bill by taking advantage of myriad breaks to which the law says you are entitled. One of the problems is that whereas everyone is in favor of the principle of making the tax code simpler, groups organize to oppose getting rid of the particular tax breaks from which they benefit.

The simple cases are when most wealth is held in liquid market securities such as shares of publicly traded stock. The wealth that Jeff Bezos holds in shares of Amazon stock is easily measured. But what about the rest? Senator Warren was fond of saying on the campaign trail that she would tax the jewelry, art, and yachts owned by billionaires.21 Let’s take art as an example of how hard this would be to do. There are two critical problems: we don’t know what art people have, and we don’t know what it is worth. There is no national (much less international) registry of artworks. In order to tax wealth, the IRS would have to know each piece of art held by every rich family, and its current market value.fn2 Now think about jewelry, or stamps, or sports memorabilia. Our goal here is not to have you shed a tear for the accountants and art dealers of billionaires. Rather, just think of the staffing requirements that would be needed to administer such a tax. Now, you might think we could reduce the sludge by limiting the tax to wealth held in liquid assets, but of course that would encourage the rich to put even more of their money into other forms of wealth, and might also induce more large companies to become privately held like Cargill or Fidelity Investments. They would be encouraged to do so because it is difficult to determine the value of a privately held business whose shares do not trade.

Saving for retirement is one of the hardest tasks Humans face. Just doing the computations is hard enough, even with some good software, but then implementing the plan involves a lot of self-discipline. Also, it is a task that is relatively new for our species, so we are still figuring out how best to do it. It is new because for most of our time on earth, we Humans did not have to worry much about saving for retirement; most people died before the problem arose. Those who were lucky enough to make it to old age were generally cared for by their extended families. Only recently did the combination of rising life expectancy and geographical dispersion of families make it necessary for people to think about providing for their own retirement income rather than depending on their children. In the span of human history, this is a very short period of time.

Consider two extreme “dream” scenarios. In the first dream, one being dreamt by a free-market economist with a peaceful smile on his face, advertisers are helping to educate consumers by explaining the benefits of lower costs, diversification, and long-run investing, as well as the folly of extrapolating recent returns into the future. In this dream, ads help each consumer discover his own ideal location on what economists call the “efficient frontier”—the place all rational investors want to find. In other words, the advertising helps consumers make better, smarter choices. The other dream is more of a nightmare, one that keeps psychologists and behavioral economists tossing and turning. In this dream, advertisers are encouraging participants to think big, not to settle for average (by indexing), and to think of investing as a way to get rich. In this nightmare, ads almost never mention fees. But they do talk a lot about past performance, even though there is essentially no evidence that past performance predicts future performance. (People who like to bet on sporting events will recognize a parallel in advertisements telling people about “locks” on upcoming games and about the amazing and nearly infallible forecasts over, say, the past three weeks.)

Economists agree about the right way to think about insurance. The most important principle is to get protection against rare but significant mishaps that can lead to financial ruin. The sorts of risks that should be insured are homes being destroyed by floods or fires, major health problems, the death or disability of a family income earner, and the crash of the family car (if it is still worth anything). Any of these events can put a household in debt for years, or even lead to bankruptcy. Paying a company to share those risks makes good sense. We should not be insuring against the possibility that our team will lose the championship game, that our coffee machine dies, or even that we dent our car backing into a parking spot at night.

If you are tired of reading about money mistakes, you can skip the rest of this chapter if you promise to follow this one rule of thumb: when purchasing insurance, choose the largest deductible available. Of course, there are exceptions to any rule of thumb, including this one. If you are offered a deductible so large that paying it would cause a serious financial hardship, we give you permission to choose a lower one. But in general, people choose a deductible that is too low. In fact, we have a name for this mistake: deductible aversion.

In the United States, households that elect a high-deductible plan have the option of creating something better than a notional On My Own Account, which is only a mental account, after all. They can create something called a health savings account (HSA), contributions to which are tax-sheltered. Many employers try to encourage employees to choose the high-deductible (HD) plan by making contributions to the HSA for any employee who does so. Any money contributed to the HSA that is not spent at the end of the year simply rolls over to the next year and eventually just turns into a retirement savings account. The existence of these accounts deepens the mystery as to why high-deductible plans get so little take-up, even in cases when they are a certain money saver.

To begin with, we think that people have drawn the wrong conclusion from that famous Johnson-Goldstein graph. The demonstrably correct conclusion to draw is that defaults have a huge impact on the elicitation of preferences. We have seen evidence to support that conclusion throughout the book, and this example remains one of the best empirical tests of that finding. But it is wrong to conclude that adopting a presumed consent policy and obtaining a low opt-out rate will necessarily save lives. That would be a reasonable conclusion if countries were implementing a hard presumed consent rule, meaning that Families would not need to be consulted. Donor consent would literally be presumed. Such a policy would completely eliminate the time and effort needed to get the permission of Families in a process that is highly time sensitive, as well as remove the risk that Families will block the donation. However, we have found that few if any countries actually implement that policy. To give just one example, in Italy, which is nominally a presumed consent country, the official website says this about the role of Families (our translation): “In the case in which the deceased person did not leave a statement regarding organ donation, the donation of an organ is allowed only if the relatives (in order: spouse, partner, adult children, and parents) are not opposed to the donation. For minors, it is always the parents who decide; if even one of the two is against, the donation will not t...

Some basic lessons of behavioral economics help explain why nations have not done more. Sadly, we have what amounts to a perfect storm, a confluence of factors that make collective action difficult. Here are the main obstacles: 1. Present bias. As we have seen, people tend to be much more concerned with now as opposed to later. Although scientists have been warning about the perils posed by climate change for decades, the most serious risks have often been seen to be off in the future, perhaps decades away. People have bills to pay now! It’s true that all over the planet, people are now facing risks and problems related to climate change. But at important moments, many leaders, and many voters, have seen climate change as a challenge for future generations, and they don’t vote. There is a big contrast here with COVID-19, in which the toll, in terms of illnesses and deaths, was felt in the here and now as friends, families, and political leaders got sick. 2. Salience. People can see smog, and they really don’t like it. Dirty air and water are both visible and scary. In many nations, including the United States, the public has demanded (and often received) cleaner air and water. By contrast, greenhouse gases are invisible in the air. If you can’t even see them, you might not worry about them. 3. No specific villain. For some threats, there is an identifiable perpetrator—a wrongdoer whose terrible deeds capture public attention. It’s not so hard to mobilize public attention a...

This is what is often called a “tragedy of the commons.” Each dairy farmer has an incentive to add more cows to his herd, because he obtains the benefits of the additional cows while suffering only a fraction of the costs; but collectively too many cows will eventually ruin the pasture. Dairy farmers need to find some way to avert this tragedy. Similar problems plague the fishing industry, and they help to explain air pollution and climate change, with the latter sometimes described as a “wicked” commons problem because of the gravity of the threat and the sheer number of people and nations involved.2 When a tragedy of the commons is involved, the standard remedy is coercion, agreed to by all. For example, dairy farmers might all agree to limit the number of cows they will add to their herds—and they might agree to punish those who violate the deal.

The behavioral economics literature does offer an idea about one possible way forward. Recall that people playing the public goods game are conditional cooperators; they cooperate if others do. In a series of experiments, the behavioral economists Ernst Fehr and Simon Gächter have shown that cooperation in repeated public goods games can be increased if players are allowed to punish noncooperators at their own expense.10 The way this works in the game described earlier is that if Player A observes that Player D (whose identity is masked) is not contributing, A can punish D. For each dollar A is willing to pay, D’s payoff is reduced by three dollars. Note that A gains nothing directly from punishing—and indeed has to pay for the privledge of punishing a noncooperator. As a result, Econs would never punish anyone in this game. But well-meaning (or possibly spiteful) Humans do punish! It turns out that the introduction of an option that no Econ would use greatly increases cooperation. In fact, with these rules in place, cooperation increases as the game is repeated—just the opposite of what is observed under the normal rules.

A particularly significant success story for disclosure requirements is the Emergency Planning and Community Right-to-Know Act, a law enacted by Congress in 1986 in the aftermath of an industrial chemical accident from a U.S. plant in Bhopal, India.19 Understood as a modest and uncontroversial step, it was essentially a bookkeeping measure, intended to give the EPA and local communities a sense of what hazardous materials might be lurking undetected. The statute ended up doing a lot more. In fact, the requirement of disclosure, captured in the Toxics Release Inventory, counts among the most unambiguous success stories in all of environmental law. To create the Toxics Release Inventory, firms and individuals must report to the national government the quantities of potentially hazardous chemicals that have been stored or released into the environment. The information is readily available on the EPA website to anyone who wants it. Tens of thousands of facilities now disclose detailed information on hundreds of chemicals, covering billions of pounds of on-site and off-site disposal or other releases. Users of hazardous chemicals must also report to their local fire departments the locations, types, and quantities of stored chemicals. And they must disclose information about potential adverse health consequences.

Note in this regard that mandatory cooling-off periods make best sense, and tend to be imposed, when two conditions are met: (a) people make the relevant decisions infrequently and therefore lack a great deal of experience and (b) emotions are likely to be running high. These are the circumstances in which people are especially prone to making choices that they will regret.