Printer Friendly

Behavioral economics and paternalism.


     A. Two Systems in the Mind: Of Humans and Econs
     B. Behavioral Market Failures
        1. Present Bias and Time Inconsistency
        2. Ignoring Shrouded (but Important) Attributes
        3. Unrealistic Optimism
        4. Problems with Probability

     A. Working Definitions
        1. Choices and Welfare
        2. The (Important but Troubled) Distinction Between Means
           and Ends
        3. The (Important but Troubled) Distinction Between Hard
           and Soft
        4. A Very Quick Summary
        5. On Welfare
     B. The Paternalist's Large Toolbox

     A. Five Welfarist Objections: An Antipaternalist's Quintet
     B. Welfare: Normative Issues
     C. Welfare: Empirical Problems
     D. Imaginable Worlds and Rule-Consequentialist Antipaternalism
     E. Choice Architecture and Inevitable Nudges

     A. Autonomy: The Thin Version
     B. Autonomy: The Thick Version
     C. Thin, Again
     D. Thick, Again
     E. An Accounting

     A. Of Transparency and Political Safeguards
     B. Of Easy Reversibility
     C. The Legitimate Claims of System 1
     D. A Real Concern: Impermissible Motivations


[T]he only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others. His own good, either physical or moral, is not a sufficient warrant. He cannot rightfully be compelled to do or forbear because it will be better for him to do so, because it will make him happier, because, in the opinion of others, to do so would be wise, or even right.

--John Stuart Mill (1)

The central conundrum has been referred to as the Energy Paradox in this setting (and in several others). In short, the problem is that consumers appear not to purchase products that are in their economic self-interest. There are strong theoretical reasons why this might be so.

[1] Consumers might be myopic and hence undervalue the long-term;

[2] [Consumers] might lack information or a full appreciation of information even when it is presented;

[3] [Consumers] might be especially averse to the short-term losses associated with the higher prices of energy efficient products (the behavioral phenomenon of "loss aversion");

[4] [E]ven if consumers have relevant knowledge, the benefits of energy efficient vehicles might not be sufficiently salient to them at the time of purchase....

--U.S. Environmental Protection Agency (2)


From 2009 to 2012, I was privileged to serve under President Obama as Administrator of the White House Office of Information and Regulatory Affairs, sometimes described (hyperbolically to be sure) as the nation's "regulatory czar." When I served in that position, the President stressed the need to consider flexible approaches that reduce costs and maintain freedom of choice for the American people. In fact, the President specifically charged me with promoting that goal, emphasizing its special importance in a period of serious economic difficulty.

In the Obama Administration, many of us were concerned about reducing regulatory costs, but we were alert both to the existence of standard market failures and to the findings of behavioral economics. We sought to identify approaches that would remedy those failures and respond to those findings, while surviving cost-benefit analysis and without imposing unjustified burdens on the private sector. (3) We knew that there were many opportunities for using regulation to save both lives and money, by, among other things, promoting safety on the highways, cleaning the air, reducing smoking, increasing the fuel economy of cars, combating childhood obesity, and reducing health risks from food.

Executive Order 13,563--a document of signal importance, a kind of mini-constitution for the regulatory state--contains a key provision called "Flexible Approaches," which states in no uncertain terms that "each agency shall identify and consider regulatory approaches that reduce burdens and maintain flexibility and freedom of choice for the public. These approaches include warnings, appropriate default rules, and disclosure requirements as well as provision of information to the public in a form that is clear and intelligible." (4) This provision can be understood as an explicit recognition of the potential value of low-cost, freedom-preserving approaches, or nudges. (5)

Behavioral economists have emphasized that in important contexts, people err. (6) Human beings can be myopic and impulsive, giving undue weight to the short term (perhaps by smoking, perhaps by texting while driving, perhaps by eating too much chocolate). (7) What is salient greatly matters. (8) If an important feature of a situation, an activity, or a product lacks salience, people might ignore it, possibly to their advantage (perhaps because it is in the other room, and fattening) and possibly to their detriment (if it could save them money or extend their lives). Human beings procrastinate and sometimes suffer as a result; (9) they are greatly affected by default rules, potentially to their detriment. (10) They can be unrealistically optimistic and for that reason make unfortunate and even dangerous choices. (11) People make "affective forecasting errors": they predict that activities or products will have certain beneficial or adverse effects on their own well-being, but those predictions turn out to be wrong. (12)

It is important to emphasize that free markets provide significant protection against such errors. Most important, markets often deter exploitation of human fallibility. If companies provide unhelpful default rules, steering consumers in directions that harm them, they may be punished as a result of competition. Companies that shroud expensive attributes, costing consumers a lot of money, may find themselves without customers before long. In addition, companies offer countless services to help people counteract self-control problems. The market itself creates strong incentives for companies to respond to these and other behavioral problems. With new technologies, those responses will become increasingly helpful, frequent, inventive, and personalized; (13) helpful "apps," of countless sorts, are proliferating, (14) and in the future, we will see unimaginably more. The market for protecting people against their own mistakes is flourishing. (15)

But there is another side. In free markets, some sellers attempt to exploit human errors, and the forces of competition may turn out to reward, rather than punish, such exploitation. In identifiable cases, those who do not exploit human errors will be seriously punished by market forces, simply because their competitors are doing so and profiting as a result. Credit markets provide many examples in the domains of cell phones, credit cards, and mortgages. (16) More generally, some policies will not be designed well if they are not informed by what we know about human behavior. (17)

It is true, of course, that a great deal remains to be understood about the nature of human error in disparate contexts. Research is continuing, and more is being learned every day; some behavioral findings are highly preliminary and need further testing. There is much that we do not know. Randomized controlled trials, the gold standard for empirical research, must be used much more to obtain a better understanding of how the relevant findings operate in the world. (18) Even at this stage, however, the underlying findings have been widely noticed, and behavioral economics and related fields have had a significant effect on policies in several nations, including the United States and the United Kingdom.

In the United States, a number of initiatives have been informed by relevant empirical findings, and behavioral economics has played an unmistakable role in numerous domains. These initiatives enlist tools such as disclosures, warnings, and default rules, and they can be found in multiple areas, including fuel economy, energy efficiency, environmental protection, health care, and obesity. (19) As a result, behavioral findings have become an important reference point for regulatory and other policymaking in the United States. (20)

In the United Kingdom, Prime Minister Cameron has created a Behavioural Insights Team with the specific goal of incorporating an understanding of human behavior into policy initiatives. (21) The official website states that its "work draws on insights from the growing body of academic research in the fields of behavioural economics and psychology which show how often subtle changes to the way in which decisions are framed can have big impacts on how people respond to them." (22) The team has used these insights to promote initiatives in numerous areas, including smoking cessation, energy efficiency, organ donation, consumer protection, and compliance strategies in general. (23) Other nations have expressed interest in the work of the team, and its operations are expanding. (24)

Behavioral economics has drawn attention in Europe more broadly. The Organisation for Economic Cooperation and Development has published a Consumer Policy Toolkit that recommends a number of initiatives rooted in behavioral findings. (25) In the European Union, the Directorate-General for Health and Consumers has also shown the influence of behavioral economics. (26) A report from the European Commission, called Green Behavior, enlists behavioral economics to outline policy initiatives to protect the environment. (27)

These developments, and the relevant findings, raise a natural question, which is whether an understanding of human behavior opens greater space for paternalism, supplementing the standard accounts of market failures by providing grounds for government action even in the absence of harm to others or some kind of collective action problem. (28) We know, for example, that people are greatly affected by choice architecture, understood as the social background against which choices are made. (29) Such architecture is pervasive and inevitable, and it greatly influences outcomes. In fact, it can be decisive. It effectively makes countless decisions for us, or at least affects our decisions. (30) Choice architecture exists whenever we enter a cafeteria, a restaurant, a hospital, or a grocery store; when we select a mortgage, a car, a health care plan, or a credit card; when we turn on a tablet or a computer and visit our favorite websites; and when we apply for drivers' licenses or building permits or social security benefits. For all of us, a key question is whether the relevant choice architecture is helpful and simple or harmful, complex, and exploitative.

Should choice architects, including those in the public sphere, be authorized to move people's decisions in their preferred directions? Would any such efforts be unacceptably paternalistic? Who will monitor the choice architects, or create a choice architecture for them? (31) From various empirical findings, it is possible to identify a set of behavioral market failures, (32) understood as a set of market failures that complement the standard economic account and that stem from human error. Is it unacceptably paternalistic to use such failures to justify regulation, even when externalities are not involved? Is it legitimate to use choice architecture to counteract behavioral market failures?

My goal here is to explore these questions. My basic answer is that behavioral market failures do, in fact, justify paternalism. (33) When such failures exist, and are significant, there are good (presumptive) reasons for a regulatory response even when no harm to others can be found. But because of heterogeneity and the risk of government error, it is usually best to use the mildest and most choice-preserving forms of intervention, such as nudges. (34) We might even venture a general principle, which might be called the first (and only) law of behaviorally informed regulation: in the face of behavioral market failures, the best responses usually are disclosures of information, warnings, default rules, and other kinds of nudges, at least when there is no harm to others. But there are exceptions to the general principle, and the choice of response depends on an analysis of costs and benefits. (35) In some cases, no response at all may be best, because the costs exceed the benefits. In other cases, stronger responses, even mandates, may turn out to be justified, because the benefits exceed the costs. Social welfare is the master concept, and when social welfare calls for a stronger response, we should give it serious consideration. (36)

It is useful to begin, I suggest, by distinguishing among varieties of paternalism. Some varieties respect people's ends and try only to influence their choice of means; other varieties attempt to affect people's choices of ends. Means paternalists might encourage (or perhaps even require) people to save money with refrigerators that are inexpensive to operate, when saving money is exactly what they want. Ends paternalists might forbid people from engaging in certain sexual activity, even though engaging in such activity is exactly what they want. Behavioral economists generally favor paternalism about means, not ends. Most of their key findings involve human errors with respect to means; their goal is to create choice architecture that will make it more likely that people will promote their own ends.

Moreover, some varieties of paternalism are highly aggressive, or "hard," while others are weaker, or "soft." Soft paternalism is libertarian, in the sense that it preserves freedom of choice. (37) A jail sentence and a criminal fine count as hard paternalism, whereas a disclosure policy, a warning, and a default rule count as soft or libertarian paternalism. Some forms of paternalism impose material costs on people's choices in order to improve their welfare; other forms of paternalism impose affective or psychic costs. Behavioral economists have generally favored soft rather than hard paternalism. (38) Means paternalism can be hard or soft, and the same is true of ends paternalism. My topic here extends far beyond libertarian paternalism and nudges, understood as approaches that affect choices without coercion, but it is important to see that nudges generally fall in the categories of means paternalism and soft paternalism. (39)

My central claim is simple: behavioral market failures are an important supplement to the standard account of market failures, and in principle they do justify (ideal) responses, even if those responses are paternalistic. As in the case of standard failures, however, the argument for a government response must be qualified by a recognition that the cure may be worse than the disease, and that all relevant benefits and costs must be taken into account.

I offer four additional conclusions:

1. Choice architecture is inevitable, and hence certain influences on choices are also inevitable, whether or not they are intentional or a product of any kind of conscious design.

2. Some of the most intuitively appealing objections to paternalism rely on autonomy, but as applied to most efforts to remedy behavioral market failures, those objections lack force, because such efforts do not interfere with autonomy, rightly understood. In fact, some such efforts promote autonomy, in part because they open up time and resources for more pressing matters. (40) There is also a risk that some of these autonomy-based objections are rooted in a heuristic for what really matters, which is welfare.

3. The most powerful objections to paternalism are welfarist in character. In many contexts, those objections are a good place to start and possibly to end, especially insofar as they emphasize the importance of private learning and the risk of government error. But they depend on normative claims that are complex and highly contested, and on empirical claims that are often false. There is no sufficient abstract or a priori argument against paternalism, whether hard or soft. (41)

4. The welfarist arguments against paternalism, new or old, are irrelevant insofar as choice architecture, and nudges, are inevitable. But insofar as paternalism is optional (and it often is), there is an intelligible rule-consequentialist objection to paternalism--though the strength of the argument depends on the form of paternalism. There are plausible rule-consequentialist arguments against (optional) paternalism, but those arguments depend on strong empirical assumptions, involving extreme optimism about markets and extreme pessimism about public officials, that are unlikely to hold in our world. The objections to paternalism are weakest when it is soft and limited to means; especially in such cases, there are many opportunities for improving welfare without intruding on freedom of choice.

The remainder of the discussion comes in five parts. Part I discusses human errors, with particular emphasis on those errors that are most likely to matter for purposes of regulatory policy. Part II explores the nature of paternalism, distinguishing among various forms, and emphasizing the wide range of tools that paternalistic choice architects might use. Part III turns to welfarist objections to paternalism. Part IV explores autonomy. Part V discusses several independent objections to soft or libertarian paternalism, particularly those that emphasize the potential lack of transparency, the risk of manipulation, and the limits of reversibility. The discussion ends with a brief conclusion.


In recent decades, there has been an outpouring of empirical work on human cognition and the risk of error. (42) As noted, this work has been noticed by policymakers, (43) and its influence is likely to grow in coming decades. My goal here is to provide a brief summary, acknowledging that research is continuing and that a great deal remains to be learned, and emphasizing those findings that have special importance for exploring regulation and the question of paternalism.

A. Two Systems in the Mind: Of Humans and Econs

Within recent social science, authoritatively discussed by Daniel Kahneman in his masterful Thinking, Fast and Slow, it has become standard to suggest that the human mind contains not one but two "cognitive systems." (44) In the social science literature, the two systems are unimaginatively described as System 1 and System 2. (45) System 1 is the automatic system, while System 2 is more deliberative and reflective. System 1 can be understood to reflect the behavior of Humans, whereas Econs think and act in accordance with System 2. (46)

System 1 works fast. Much of the time, it is on automatic pilot. It is driven by habits. It can be emotional and intuitive. When it hears a loud noise, it is inclined to run. When it is offended, it wants to hit back. It certainly eats a delicious brownie. It can procrastinate; it can be impulsive. It wants what it wants when it wants it. It can be excessively fearful and too complacent. It is a doer, not a planner. System 1 is a bit like Homer Simpson, James Dean (from Rebel Without a Cause), and Pippi Longstocking.

System 2 is more like a computer or Mr. Spock from the old Star Trek show (or the android Data from the somewhat-less-old Star Trek show). It is deliberative. It calculates. It hears a loud noise, and it assesses whether the noise is a cause for concern. It thinks about probability, carefully though sometimes slowly. It does not really get offended. If it sees reasons for offense, it makes a careful assessment of what, all things considered, ought to be done. It sees a delicious brownie, and it makes a judgment about whether, all things considered, it should eat it. It insists on the importance of self-control. It is a planner more than a doer.

At this point, it might be asked: What, exactly, are these systems? Are Humans and Econs agents? Do they operate as homunculi in the brain? Are they little people? Are they actually separate? In the case of conflict, who adjudicates? The best answer is that the idea of two systems is a heuristic device, a simplification that is designed to refer to automatic, effortless processing and more complex, effortful processing. When people are asked to add one plus one, or to walk from their bedroom to their bathroom in the dark, or to read the emotion on the face of their best friend, the mental operation is easy and rapid. When people are asked to multiply 179 by 283, or to navigate a new neighborhood by car, or to decide which retirement or health insurance plan best fits their needs, the mental operation is difficult and slow.

Identifiable regions of the brain are active in different tasks, and hence it may well be right to suggest that the idea of "systems" has physical referents. An influential discussion states that "[a]utomatic and controlled processes can be roughly distinguished by where they occur in the brain." (47) The prefrontal cortex, the most advanced part of the brain (in terms of evolution) and the part that most separates human beings from other species, is associated with deliberation and hence with System 2. The amygdala has been associated with a number of automatic processes, including fear, (48) and can thus be associated with System 1.

With respect to intertemporal choice (an especially important topic for behaviorally informed regulation), it has been found that when impatient people are thinking about their future selves, the particular region of the brain that is most active when people are thinking about themselves is significantly less active. (49) In patient people, by contrast, that region of the brain is significantly more active when they are thinking of their future selves. (50) Here, then, is a neurological basis for distinguishing not only between Humans and Econs but also between different members of the human species. This finding has clear implications for myopia, in the form of neglect of the future, and time inconsistency. (51) In neural terms, impatient people think of their future selves in the same way that they think of strangers--raising the possibility that they may not be sufficiently concerned about their own future well-being. (52) Neural evidence also suggests that when people's emotions are strongly engaged, in a way that makes them motivated to accept certain political conclusions, identifiable features of the brain are active--and that when people do not have a significant emotional stake, those regions are relatively inactive. (53)

On the other hand, different parts of the brain interact, and it is not necessary to make technical or controversial claims about neuroscience in order to distinguish between effortless and effortful processing. The idea of System 1 and System 2 is designed to capture that distinction in a way that works for purposes of exposition (and that can be grasped fairly immediately by System 1).

Here is a striking demonstration of the relationship between System 1 and System 2: some of the most important cognitive errors, including several of relevance here (framing and loss aversion), disappear when people are using a foreign language. (54) Asked to resolve problems in a language that is not their own, people are less likely to blunder. In an unfamiliar language, they are more likely to get the right answer. How can this be?

The answer is straightforward. When people are using their own language, they think quickly and effortlessly, so System 1 has the upper hand. When people are using another tongue, System 1 is a bit overwhelmed, and may even be rendered inoperative, while System 2 is given a serious boost. Our rapid, intuitive reactions are slowed down when we are using a language with which we are not entirely familiar. We are more likely to do some calculating and to think deliberatively--and at least on some questions, to give the right answers. (55) In a foreign language, people have some distance from their intuitions, and that distance can stand them in good stead. In a foreign language, Humans recede in favor of Econs.

There is a lesson here about the importance of technocratic approaches to law and regulation, including those that emphasize the need for careful consideration of costs and benefits. (56) Such approaches do not (exactly) use a foreign language, but they do ensure a degree of distance from people's initial judgments, thus constraining the mistakes associated with System 1. People do not naturally think about risk regulation in terms of costs and benefits, but the effort to do so can weaken or eliminate the effect of intuitions, in a way that leads to greatly improved decisions. (57) There is also a point here about the hazards of relying on intuitions as a foundation for political or moral theory--a point to which I will return. (58)

The defining feature of System I is that it is automatic, but I have said that System 1 can be emotional, and when it is, its emotional character creates both risks and opportunities. People may be immediately fearful of some risk--say, the risk associated with terrorism, or the risk of losses in the stock market--whether or not reality, and the relevant statistics, suggest that there is cause for alarm. A great deal of work finds that people tend to assess products, activities, and other people through "an affect heuristic." (59) When the affect heuristic is at work, people evaluate benefits, costs, and probabilities not by running the numbers, but by consulting their feelings. They might hate coal-fired power plants or love renewable fuels, and those feelings may influence their judgments about the benefits and costs of coal-fired power plants and renewable fuels. (60) System 1 is doing the key work here.

In fact, some goods and activities come with an "affective tax" or an "affective subsidy," in the sense that people like them more, or less, because of the affect that accompanies them. Advertisers, and public officials, try to create affective taxes and subsidies; consider public educational campaigns designed to reduce smoking or texting while driving. Some political campaigns have the same goal, attempting to impose a kind of affective tax on the opponent, and to enlist the affect heuristic in their favor. Many political campaigns appeal directly to System 1, not System 2. The same is true for some lawyers involved in trials or even appellate litigation. If System 1 can be enlisted, it may run the show, with System 2 operating as a kind of ex post helper. (61) In many cases, System 2 acts as lawyer for the cause, and System 1 is a most demanding client. (62)

One explanation for the operation of heuristics is that people decline to answer a hard question and answer a simpler one instead. (63) For political candidates, people might not ask, "Do I agree with Candidate A or Candidate B on economic policy?" (a potentially complex question) but instead, "Do I like and trust this person?" or, "Is this person like me?" (potentially much easier questions). (64) Something similar is at work in educational campaigns that attempt to trigger fear (in the context, for example, of smoking, obesity, and texting while driving), and thus to engage System 1 rather than to offer statistical analyses.

B. Behavioral Market Failures

I now turn to four sets of mistakes that can lead to significant harms and that should be counted as behavioral market failures. As we shall see, all of these mistakes are firmly rooted in the operations of System x. The unifying theme is that insofar as people are making the relevant errors, their choices will not promote their own ends. It follows that a successful effort to correct these errors would generally substitute an official judgment for that of choosers only with respect to means, not ends. (65) There are, however, some complexities in this claim. The distinction between means and ends raises a number of difficult puzzles, some of them involving the identification of people's ends over time.

1. Present Bias and Time Inconsistency

According to standard economic theory, people will consider both the short term and the long term. They will take account of relevant uncertainties; the future is unpredictable, and significant changes may occur over time. People will appropriately discount the future. It is probably far better to have money, or a good event, a week from now than a decade from now. People may, rationally and reasonably, select different balances between the present and the future. With respect to present and future consumption, people who are twenty-five make different tradeoffs from people who are sixty-five, and for excellent reasons.

In practice, however, some people procrastinate or neglect to take steps that impose small, short-term costs but produce large, long-term gains, and at least some of the relevant actions seem hard to justify. (66) While System 2 considers the long term, System 1 is myopic, and, in multiple ways, people show present bias. (67) People may, for example, delay enrolling in a retirement plan, (68) starting to exercise, ceasing to smoke, or using some valuable, cost-saving technology. (69) In many cases, inertia is an exceedingly powerful force. (70)

One implication is that some people fail to make choices that have short-term net costs but long-term net benefits--as is the case, for some, with choosing more energy-efficient products, including appliances and cars with good fuel economy. (71) Another implication is that some people make choices that have short-term net benefits but long-term net costs, including a significant risk of causing premature death (as is the case, for many, with smoking cigarettes). Procrastination, inertia, hyperbolic discounting, (72) and associated problems of self-control (73) are especially troublesome when the result is a small short-term gain at the expense of large long-term losses. There is a close connection between procrastination and myopia, understood as an excessive focus on the short term. (74)

The problem of time inconsistency arises when people's preferences at Time One diverge from their preferences at Time Two. (75) At Time One, people might prefer to eat a great deal, to smoke, to spend, to become angry, to drink, to procrastinate, or to gamble. The resulting choices might have serious adverse effects on the same people at Time Two, leading to a significant welfare loss. As I have suggested, an identifiable region of the brain is most actively engaged when people are thinking about themselves, and for impatient people in particular, this region is less active when they are thinking about their future selves. (76) Studying the ventromedial prefrontal cortex (vMPFC), psychologist Jason Mitchell and his coauthors state that this "neural signature" suggests that "shortsighted decision-making occurs in part because people fail to consider their future interests as belonging to the self." (77) Thus, for those who are shortsighted, the "vMPFC response was nearly identical when people tried to predict their future enjoyment ... and another person's present enjoyment," suggesting that such people think of their own future selves in the same way that they think of strangers. (78)

Strikingly, Mitchell and his coauthors find the following:
   [T]he magnitude of this vMPFC difference between judgments of
   present and future enjoyment predicted the impatience or
   shortsightedness of people's intertemporal choices. Those
   participants in whom vMPFC activity most differentiated between
   predictions of present and future enjoyment tended to make the most
   impatient decisions, preferring small present rewards to large
   future rewards. In contrast, participants in whom vMPFC did not
   differentiate between predictions of present and future enjoyment
   tended to make the most patient decisions, preferring large future
   rewards to small present rewards. (79)

Some behavioral economists have emphasized the problem of "internalities" (80)--problems of self-control and errors in judgment that harm the people who make those very judgments. We can think of internalities as occurring when we make choices that injure our future selves. Of course people can use various techniques to overcome this problem, including precommitment strategies; consider Ulysses and the Sirens. (81)As I have noted, private markets are perfectly capable of creating products and practices to help overcome self-control problems; in fact there are countless such products and practices. But it is at least plausible to suggest that regulatory approaches that address internalities can produce large welfare gains, in some cases by saving lives. (82)

Such approaches might take the form of disclosure requirements or warnings, designed to promote self-control. Flexible approaches of this kind have the advantage of maintaining freedom of choice and thus respecting heterogeneity, which is especially important in light of the fact that reasonable people can trade off the present and the future in multiple ways based on the particulars of their situation. But in imaginable cases, an economic incentive or a mandate might be the best solution; consider, for example, efforts to promote healthy foods or bans on texting while driving, if understood to protect drivers (as well as those whom they endanger). With respect to internalities, energy policy includes many examples, such as energy-efficiency requirements for appliances and fuel-economy requirements for vehicles. (83) Under imaginable assumptions about costs and benefits, the best approach to a palpable neglect of the long term might turn out to be a ban. (84)

2. Ignoring Shrouded (but Important) Attributes

What do people notice? What do they miss? In the late 1990s, social scientists Christopher Chabris and Daniel Simons tried to make some progress on these questions by asking people to watch a ninety-second movie, in which six ordinary people pass a basketball to one another. (85) The simple task? To count the total number of passes.

After the little movie is shown, the experimenter asks people how many passes they were able to count. Then the experimenter asks: And did you see the gorilla? A lot of people laugh at the question. What gorilla? Then the movie is replayed. Now that you are not counting passes, you see a gorilla enter the scene, plain as day, and then pound its chest, and then leave. The gorilla (actually a person dressed up in a gorilla suit) is not at all hard to see. In fact, you can't miss it. But when counting passes, many people (typically about half) do miss it.

Behavioral economists have been quite interested in the gorilla experiment, because it shows that people are able to pay attention to only a limited number of things, and that when some of those things are not salient, we ignore them, sometimes to our detriment. Magicians and used-car dealers try to hide gorillas; the same is sometimes true of those who provide credit cards, cell phone service, and mortgages. (86)

Attention is a scarce resource, and attention is triggered by salience; it follows that salience greatly matters. One reason is that System 1 does not closely survey all aspects of social situations, and System 2 may be working hard on other business. When certain features of a product or an activity are not salient, people may disregard them even if they are important, and the result may be individual harm. Complexity and information overload are problems in part because of the importance of salience. When hidden amidst complexity, important features of products and situations might be missed, thus creating real problems. In fact, a lack of salience can be a serious kind of market failure, producing individual and social harm.

Why, for example, do so many people pay bank overdraft fees? One answer is that such fees are not sufficiently salient to people, and some fees are incurred as a result of inattention and neglect. A careful study suggests that limited attention is indeed a source of the problem and that once overdraft fees become salient, they are significantly reduced. (87) When people take surveys about such fees, they are less likely to incur a fee in the following month, and when they take a number of surveys, the issue becomes sufficiently salient that overdraft fees are reduced for as much as two years. (88)

In many areas, the mere act of being surveyed can affect behavior by, for example, increasing use of water-treatment products (thus promoting health) and the acquisition of health insurance; one reason is that being surveyed increases the salience of the action in question. (89) In the same vein, a field experiment finds that simple textual reminders that loan payments are due have a significant effect on payments--indeed, the same effect as an economic incentive in the form of a twenty-five-percent decrease in interest payments! (90) A field experiment shows that reminders have a strong effect on people who are due for a dental checkup. (91) Reminders and checklists are effective because they promote salience.

A more general point is that many nontrivial costs (or benefits) are less salient than purchase prices. They are "shrouded attributes" to which some consumers do not pay much attention. Such "add-on" costs may matter a great deal but receive little consideration because they are not salient. (92) An absence of attention to energy costs, which may be "shrouded" for some consumers, has significant implications for regulatory policy. The clearest such implication involves the importance of providing cost-related information that people can actually understand. In 2012, the Department of Transportation and the EPA produced new fuel-economy labels with this goal in mind; the new labels explicitly draw attention to the economic effects of fuel economy. (93)

An understanding of the problem of shrouded attributes also helps to identify a potential justification for regulatory standards in the domains of fuel economy and energy efficiency, involving a behavioral market failure. Of course such standards reduce social costs by reducing air pollution and promoting energy security. But from recent rules, the strong majority of the relevant benefits are private; they come from consumer savings. (94) On standard economic grounds, it is not simple to identify a market failure that would justify taking account of such benefits. A plausible argument is behavioral. The basic idea is that such standards might help produce a set of outcomes akin to those that would result if relevant attributes were not shrouded.

This point has not escaped official attention. In explaining the new fuel-economy rules issued in 2012, the Department of Transportation referred to
   phenomena observed in the field of behavioral economics, including
   loss aversion, inadequate consumer attention to long-term savings,
   or a lack of salience of relevant benefits (such as fuel savings,
   or time savings associated with refueling) to consumers at the time
   they make purchasing decisions. Both theoretical and empirical
   research suggests that many consumers are unwilling to make
   energy-efficient investments even when those investments appear to
   pay off in the relatively short-term. This research is in line with
   related findings that consumers may undervalue benefits or costs
   that are less salient, or that they will realize only in the
   future. (95)

So justified, fuel-economy standards are a form of hard paternalism, but they need not question people's ends. The idea is that people want to minimize all relevant costs, and if they are not taking account of some such costs, properly designed fuel-economy standards promote, and do not override, their ends. It is true that if the problem is a lack of attention and salience, the most natural and presumptively appropriate response is disclosure, not a mandate-and on one view, fuel-economy labels, and not a mandate, are the better option. But if such a mandate has benefits far in excess of costs, it would appear to be justified as well. (96)

3. Unrealistic Optimism (97)

System 2 is realistic, but System 1 is not. (98) A great deal of work in behavioral psychology and economics suggests that most people are unrealistically optimistic, in the sense that their own predictions about their behavior and their prospects are skewed in the optimistic direction. (99) Indeed, the tendency toward unrealistic optimism seems to be hardwired. (100) And if people are unduly optimistic about their future behavior, they may select financial packages (say, for credit cards, mortgages, health care plans, and cell phones) that result in significant economic losses. (101) In addition, they may run risks (say, by texting while driving) that can lead to serious harm. The most general point is that if people are unduly optimistic, they may fail to take optimal precautions against serious dangers. An obvious response is a disclosure strategy, perhaps including graphic warnings, that helps to counteract unrealistic optimism. (102)

When people imagine their own future, they tend to see it as very good, even if the likely reality is far more mixed. (103) The "above average" effect is common; (104) many people believe that they are less likely than others to suffer from various misfortunes, including automobile accidents and adverse health outcomes. A study found that while smokers do not underestimate the statistical risks faced by the population of smokers, they nonetheless believe that their personal risk is less than that of the average smoker. (105) Unrealistic optimism is related to confirmation bias, which occurs when people give special weight to information that confirms their antecedent beliefs. (106) To the extent that people show this bias, and to the extent that it affects their behavior, they may be led in directions that produce serious welfare losses.

What makes people unrealistically optimistic? How can people maintain such optimism in the face of repeated experiences with reality, which should press them toward greater realism? One reason involves a remarkable asymmetry in how people process information. (107) In brief, people give more weight to good news than to bad news. Tali Sharot and her collaborators find that when people receive information that is better than expected, they are likely to change their beliefs-but when what they learn is worse than expected, their beliefs are more likely to remain constant. In the first stage of the experiment, people were asked to estimate their likelihood of experiencing eighty bad life events (such as robbery and Alzheimer's disease). In the second stage, they were given accurate information about the average probability for similarly situated people. In the third stage, people were asked to state their view about their personal probability in light of what they had learned.

The central finding is that updating is more likely when people get good news than when they get bad news. More specifically, people were more likely to move their personal probability estimate upward when they learned that the population average was above the number they gave than to move their personal probability estimate downward when they learned that the population average was below the number they gave. Here, then, is clear evidence of selective updating. The authors conclude that the impact of a learning signal, or new information, "depends on whether [that] new information calls for an update in an optimistic or pessimistic direction." (108)

The authors also studied fMRI data to explore what happens in identifiable regions of the brain-more particularly, the right inferior prefrontal gyrus (IFG), a region of the prefrontal cortex. This is an important question, because the IFG is the region that corrects errors in estimation. Does the IFG react differently to negative and positive information? The answer is yes. The authors' basic conclusions are technical but worth quoting:
   We found that optimism was related to diminished coding of
   undesirable information about the future in a region of the frontal
   cortex (right IFG) that has been identified as being sensitive to
   negative estimation errors. Participants with high scores on trait
   optimism were worse at tracking undesirable errors in this region
   than those with low scores. In contrast, tracking of desirable
   information in regions processing desirable estimation errors
   (MFC/SFG, left IFG and cerebellum) did not differ between high and
    low optimists. (109)

A subsequent study found that people's ability to incorporate bad news into their judgments can be improved by disrupting the functioning of the left (but not the right) interior frontal gyrus; this disruption eliminates the good news/bad news effect. (110) The conclusion, with neural foundations, is that people are unrealistically optimistic, in the sense that they are more responsive to desired than to undesired information--a point that obviously raises challenges for regulatory policy and disclosure requirements in particular. Perhaps the most important point here is that disclosure requirements may turn out to be ineffective with respect to optimistically biased consumers. Any such requirements should be devised so as to reduce that risk; graphic warnings are a possibility here.

4. Problems with Probability

For various reasons, System 1 does not handle probability well. One problem is the availability heuristic. When people use that heuristic, their judgments about probability are affected by whether a recent event comes readily to mind. (111) If an event is cognitively "available," people might well overestimate the risk. If an event is not cognitively available, people might well underestimate the risk. (112)

In deciding whether it is dangerous to walk in a city at night, to text while driving, or to smoke, people often ask about incidents of which they are aware. While System 2 might be willing to do some calculations, System 1 works quickly, and it is easy and even fairly automatic to use the availability heuristic. Instead of asking hard questions about statistics, it asks easy questions about what comes to mind. "Availability bias" can lead to significant mistakes about the probability of undesirable outcomes. (113) The bias can take the form of either excessive fear or complacency.

A distinct but related finding is that people sometimes do not make judgments on the basis of the expected value of outcomes, and they may neglect the central issue of probability, particularly when emotions are running high. (114) Especially in such cases, people may focus on the outcome and not on the probability that it will occur. (115) If there is a small chance of catastrophe--the loss of a child, a fatal cancer--that outcome, rather than the statistical likelihood that it will happen, may dominate people's thoughts. If there is a small chance of something wonderful--the best vacation ever or a fabulous job opportunity-people's enthusiasm about that outcome may crowd out the statistics.

Those who sell insurance trade on people's fear of the worst-case scenarios; so do terrorists, who aim to convince civilians that they "cannot be safe anywhere" in their daily lives. When people are making mistakes about probability, well-designed disclosure strategies, including warnings, could help. Here too, the government would be respecting people's ends. When officials (or private institutions) correct people's mistakes about risks, they are affecting means, and helping people to achieve their goals.


A. Working Definitions

Do the findings just outlined justify paternalism? The initial task is to produce a working definition of paternalism. Of course paternalism can come from diverse people and institutions. Employers, professors, doctors, lawyers, architects, bankers, rental car companies, and countless others are capable of paternalism. All of these, and many others, may attempt to influence System 1 or to educate System 2, and those efforts, along with social pressures, can greatly affect individual choices. My narrow focus here, however, is on paternalism from government. Though the underlying issues deserve careful attention, and though the discussion here bears on those issues, I do not explore behavioral justifications for paternalism from nongovernmental actors, such as doctors, teachers, lawyers, and employers. (116)

There are many recent examples of arguable or actual paternalism from public officials. Consider, for example, the controversial decision in 2012, initiated by New York Mayor Michael Bloomberg, to ban the sale (in certain places) of sodas in containers of more than sixteen ounces. (117) Mayor Bloomberg sought to reduce obesity, and he believed that the ban would promote that goal. Some people choose drinks in large containers, and Mayor Bloomberg's proposal would not merely influence that choice but make it unavailable. And indeed, much of the negative reaction to the proposal stemmed from the view that it was paternalistic and unacceptable for that reason. (118) Why--critics asked and sometimes raged--should Mayor Bloomberg make the decision about the size of soft drink containers, rather than consumers themselves? His proposal was certainly taken as a form of paternalism. (Note that it was a mild form; people could still drink as much as they like; they simply had to buy two containers rather than one. I will return to the question of how best to characterize it below.)

1. Choices and Welfare

It is tempting to suggest that the government acts paternalistically when it overrides people's choices on the ground that their choices will not promote their own welfare. But there is an immediate problem with this suggestion. The idea of "overriding" is ambiguous. Government has a series of tools for influencing people. Some of the strongest tools involve incapacitation, with capital punishment and life imprisonment counting as the limiting cases. Insofar as we are speaking of these particular penalties, and of imprisonment more generally, it may be fair to speak of overriding choices. Other tools are more subtle, ranging from monetary penalties, large and small, to the use of education, warnings, default rules, and time, place, and manner restrictions. Even criminal and civil bans are often accompanied by monetary penalties. When the government imposes penalties on certain choices, it puts people who make those choices at some kind of risk or in some kind of jeopardy. Choices are not overridden, strictly speaking. If people are told that they will have to pay a fine if they engage in certain behavior, they remain free to engage in that behavior and to pay the fine. Paternalistic policies may influence rather than override choices.

The unifying theme of paternalistic approaches, however diverse, is that government does not believe that people's choices will promote their welfare, and it is taking steps to influence or alter people's choices for their own good. (119) In acting paternalistically, government may be attempting (1) to affect outcomes without affecting people's actions or beliefs; (2) to affect people's actions without influencing their beliefs; (3) to affect people's beliefs in order to influence their actions; or (4) to affect people's preferences, independently of affecting their beliefs, in order to influence their actions. Automatic enrollment would fall in the first category insofar as it affects outcomes; but it need not lead to any change in people's actions. (120) The power of automatic enrollment stems from the fact that it works on those who are passive. (121) A civil fine would fall in the second category insofar as it affects what people do without affecting their beliefs. (122) An educational campaign or a set of factual warnings, specifically designed to alter beliefs, would fall in the third. A graphic warning campaign, designed to affect preferences but without necessarily affecting beliefs, (123) would fall in the fourth category.

From the standpoint of those who oppose paternalism, all of these effects may be objectionable, but perhaps for different reasons. For example, efforts to affect people's preferences might seem especially insidious except insofar as such efforts are limited to the provision of truthful information. Provision of such information is certainly a nudge, but it may or may not qualify as paternalistic. I will explore that complex issue below.

2. The (Important but Troubled) Distinction Between Means and Ends

I have noted the importance of distinguishing between means paternalism and ends paternalism. In acting paternalistically, government might well accept people's ends but conclude that their choices will not promote those ends. A GPS provides information about how to get from one place to another. People can ignore what the GPS says and try their own route, but if they do so, there is a serious risk that they will undermine their own ends (and people know that). Means paternalists see their proper domain as building on the GPS example. If, for example, people want to make a sensible tradeoff between up-front costs and long-term fuel costs, but sometimes fail to do so (perhaps because long-term costs are not salient), means paternalists might take steps to steer people in the direction of considering all relevant costs at the time of purchase.

We have seen that disclosure is the most natural solution here, but we have also seen that means paternalists would consider a fuel-economy mandate if they could be convinced that such a mandate would promote consumers' ends. The analogy here would be to a GPS that forces cars to take the best or most sensible route--not an entirely attractive idea (what if people enjoy certain scenery, or are nostalgic about longer routes ?), but perhaps appealing for some people and at some times and places. The idea of the coercive GPS can be seen as a model and a test for hard paternalism with respect to means.

Ends paternalists have more ambitious goals. They might think, for example, that longevity is what is most important and that even if people disagree, and are willing to run certain risks for reasons they believe to be good and sufficient, paternalists should steer them toward longevity. Or ends paternalists might believe that certain sexual activity is inconsistent with people's well-being, suitably defined, and hence they should not be allowed to engage in that activity. Behavioral economists have not sought to revisit people's ends. They have generally emphasized human errors with respect to means, and hence means paternalism is their principal interest and also my main focus here.

While the distinction between means paternalism and ends paternalism captures something important, it raises a number of questions, and the line between the two is not always sharp. Some of the most straightforward cases of means paternalism involve shrouded attributes, optimism bias, and availability bias. Suppose that people want a refrigerator that will perform well and cost as little as possible. If government ensures that people have accurate information about cost, it is not revisiting their ends in any way. Indeed, it is not even acting paternalistically, in the sense that it is informing people's choices, rather than (independently) influencing them. The same can be said if people underestimate the risks of distracted driving or of smoking. If the government corrects people's unrealistic optimism, or counteracts the effects of the availability heuristic to produce an accurate judgment about probability, it is respecting their ends, and we might not want to characterize its action as paternalistic at all.

So too if, for example, people are ignoring certain product attributes because those attributes are shrouded. If those attributes would matter to people if they attended to them, then efforts to promote disclosure do not question people's ends. Of course there may be hard questions here in determining whether people are in fact ignoring shrouded attributes (as opposed to not caring about them), but thus far, at least, there is no problem of ends paternalism, and indeed there might not be paternalism at all. If the relevant steps are harder--if they involve economic incentives designed to discourage the relevant behavior, or flat bans--then they would qualify as paternalistic. But they would seem to count as means paternalism if they are designed only to ensure that people achieve their own ends. (124)

Even in the apparently easy cases, however, there are complications. Consider a fuel-economy label, designed to inform people of the cost over a year or a five-year period of particular cars. If the government provides this information through a vivid letter grade--say, an "A" or a "B," as was in fact proposed (125)--it is not merely providing people with facts. To be sure, this is not the most aggressive form of paternalism about either means or ends. But formal grades might be taken as a form of paternalism not merely about means, but also about ends, insofar as government is singling out the particular variable of fuel economy and attempting to focus people's attention on that variable, as opposed to numerous other variables that would remain ungraded. And indeed, the government declined to require letter grades in part on the ground that such grades might be taken, wrongly, to suggest that the government was giving "all things considered" grades to cars. (126) But I am making a different point here: even if this risk did not exist, a fuel-economy grade could be taken to be paternalistic, and to involve a degree of paternalism about ends as well as means, insofar as it would focus and heighten people's attention with respect to one of innumerable features of cars. The government does not, after all, give serious consideration to requiring letter grades with respect to speed, or acceleration, or brightness of color, or stylishness, or coolness (actual or perceived).

Even without letter grades, any fuel-economy label itself has at least a degree of paternalism, certainly about means and indeed about ends as well, insofar as it isolates fuel economy, rather than other imaginable features of cars, for compulsory display. Consider a thought experiment or perhaps a little science fiction. We should be able to agree that the government would focus only on means, and would not be paternalistic, if it could have direct access to all of people's internal concerns and provide them with accurate information about everything that concerns them. And perhaps in the fullness of time, government, or the private sector, will be able to do something like that. But insofar as the government is being selective, it is at least modestly affecting people's ends, and perhaps intentionally so.

Of course people want to save money; that is one of their ends. But the government chose a fuel-economy label, rather than an acceleration label or a coolness label, for a reason--to focus consumer attention on that particular feature of cars. (To be sure, it is also possible that fuel economy is more shrouded than other features, but that is hardly self-evident.) To these points we might add the more familiar one, which is that any disclosure requirement has to be framed in a certain way, and the choice of frame may well affect people's decisions and even their ends.

It is reasonable to say that the government would be focused solely on means if it provided people with accurate information about everything that they cared about. In that event, disclosure would not be paternalistic at all. It would be means focused, and it would not attempt to influence choices except insofar as it would promote accurate beliefs, which is not a paternalistic endeavor. But if the government frames a disclosure policy with the purpose and effect not only of informing but also of influencing people's choices, it is engaging in a form of soft paternalism--not only about means, but also about ends, insofar as it is attempting to affect them. And if the government's disclosure policy is selective, in the sense that it requires disclosure with respect to one attribute (that people care about) but not others (that people also care about), it is again engaging in a form of soft paternalism about means and also ends, insofar as it is attempting to affect them--unless it can be shown that the selected attribute is, distinctly, one on which people now lack and need information.

But we should not be too fussy or clever here, and we really should avoid tying ourselves into conceptual knots. If framing or selectivity is at work, there may be a form of ends paternalism, but it is likely to be of a very modest kind. If the characteristic is one that people antecedently do care about--like money--then it is fair to say that any paternalism is at least centrally about means, and that the intrusion on people's ends is modest and possibly even incidental.

In the domain of procrastination and time inconsistency, however, the distinction between means paternalism and ends paternalism is more troubled still. In addressing those problems, are paternalists addressing means or ends? If ends, at what time? At Time 1, the person sought to smoke, to drink, and to eat a lot; at Time 2, the (same) person wishes that none of these choices had been made. (127) To know whether a paternalistic intervention is about means or about ends, we may have to identify the level of generality at which people's ends are to be described. If the end is "for life to go well," then all forms of paternalism, including the most ambitious, seem to qualify as means paternalism, since they are styled as means to that most general of ends. But if the end is very specific--"To buy this product today!" or "To smoke this cigarette right now!"--then many and perhaps all forms of paternalism qualify as ends paternalism. If ends are described at a level of great specificity, there may be no such thing as means paternalism.

In the hard cases of procrastination and time inconsistency, the best solution may be to decline to answer the "means or ends" question directly, on the ground that it is not tractable, and instead to ask about people's aggregate welfare over time, on the theory that aggregate welfare (taking all relevant values into account) is the end that people really do care about. If an effort to overcome unjustified procrastination promotes people's welfare on balance, it responds to a behavioral market failure and hence is plausibly justified, at least on welfare grounds. The word "plausibly" is important; there are many objections, and I will get to them in due course. And of course public officials may face formidable problems in deciding what promotes aggregate welfare over time (128)--a point that argues in favor of soft rather than hard paternalism, and one to which I will return.

3. The (Important but Troubled) Distinction Between Hard and Soft

Let us dispense with the idea of "overriding" choices and emphasize the different tools that paternalistic officials are using. We can imagine actions of government that attempt to improve people's own welfare by threatening to imprison those who make certain choices. We can also imagine actions of government that attempt to improve people's own welfare by threatening to fine those who make certain choices. If the government imposes criminal or civil fines on those who smoke marijuana, refuse to buckle their seatbelts, or gamble, and if it does so because it disagrees with people about what would promote their own welfare, it is acting paternalistically.

There is of course a continuum here between paternalistic actions that impose high costs and paternalistic actions that impose low costs. A small monetary fine--of, say, five cents--falls within the definition of paternalism, but it may not have a significant effect on behavior. Note, however, that some sanctions have expressive functions and may be effective for that reason, even if the actual size of the sanction is small. (129) A modest criminal fine (say, for smoking, failing to buckle one's seatbelt, texting while driving, or gambling) may have a large deterrent effect. A paternalistic intervention with such a sanction, however modest, might be found highly objectionable by those who abhor paternalism. Note in addition that even very small costs--say, a five-cent charge for a bag at a grocery store--may have a significant effect on behavior. (130) And indeed, a careful analysis shows such an effect, in part because of the power of loss aversion. (131) If such small costs are imposed in order to protect people against their own bad or harmful choices, they count as paternalistic.

In fact, it might be best to understand paternalistic interventions in terms of a continuum from hardest to softest, with the points marked in accordance with the magnitude of the costs (of whatever kind) imposed on choosers by choice architects. On this view, there is no sharp or categorical distinction between hard paternalism and soft paternalism; all we have are points along a continuum. But we should agree that there is a significant difference between, say, a severe criminal ban on smoking marijuana and a nominal civil fine, and between a prison sentence for failing to buckle your seatbelt and a graphic educational campaign offering vivid warnings.

Under this approach, a statement that paternalism is "hard" would mean that choice architects are imposing large costs on choosers, whereas a statement that paternalism is "soft" would mean that the costs are small. And under this approach, all costs, material or nonmaterial, would count, and to assess the degree of hardness, we would inquire into their magnitude. For example, psychic costs, as produced by graphic warnings, could move an intervention along the continuum toward hard paternalism, as long as those costs turned out to be high. Nudges would count as soft paternalism because and insofar as they impose no or very small costs on choosers.

There are significant advantages in seeing a continuum here rather than a categorical distinction. But if a categorical distinction is what is sought, we should focus on the existence of material costs. On this approach, we would understand the term "hard paternalism" to refer to actions of government that attempt to improve people's own welfare by imposing material costs on their choices. By contrast, the term "soft paternalism" would refer to actions of government that attempt to improve people's own welfare by influencing their choices without imposing material costs on those choices.

If the government engages in an advertising campaign designed to convince people to exercise more than they now do, it is engaging in a kind of soft paternalism. If the government requires employers automatically to enroll workers in health insurance plans, or requires warnings to accompany certain products, soft paternalism is involved. Soft paternalism is libertarian insofar as it does not impose material costs on people's choices. (Of course, material costs are being imposed in all of these cases; the focus is on whether those costs are being imposed on the choices of end-users.) We can understand soft paternalism, thus defined, as including nudges, and I will use the terms interchangeably here.

In a careful and highly illuminating book, Riccardo Rebonato offers a provocative and different definition of libertarian paternalism, or nudges:
   Libertarian paternalism is the set of interventions aimed at
   overcoming the unavoidable cognitive biases and decisional
   inadequacies of an individual by exploiting them in such a way as
   to influence her decisions (in an easily reversible manner) towards
   choices that she herself would make if she had at her disposal
   unlimited time and information, and the analytic abilities of a
   rational decision-maker (more precisely, of Homo Economicus). (132)

This definition is useful, but it is imprecise in three respects. First, the universe of nudges is far broader than the definition suggests. Soft paternalism includes interventions (such as warnings and default rules) that may be helpful, but that need not specifically counteract biases and decisional inadequacies. Second, the word "counteracting" is better than "exploiting." Nudges can counteract biases (such as unrealistic optimism) without exploiting anything. Third, the words "easily reversible" are imprecise, because they could capture (for example) small civil penalties, even though they do not count as libertarian.

Emphasizing the idea of a continuum, however, we should recognize that approaches that impose (high) psychic costs, and thus target System 1, may have a greater effect, and in that sense turn out to be less soft, than approaches that impose (low) material costs. Moreover, an approach that does not impose high material costs may have a major effect on choices. Indeed, it may greatly affect both beliefs and actions, and hence make all the difference. People may well change their behavior when psychic costs are high even if material costs are close to zero. An emphasis on material costs may be useful for purposes of taxonomy, but it should not be taken to suggest that such costs are all that matter, or even that tools that impose such costs are the most influential ones in the toolbox.

4. A Very Quick Summary

Summarizing these various points, we can imagine the following possibilities, with illustrative examples:

SOFT PATERNALISM   Fuel-economy labels      Automatic enrollment in
                                            particular political party

HARD PATERNALISM   Fuel-economy standards   Criminal ban on same-sex

Where behavioral market failures justify corrective action, the government should be inclined to stay in the upper-left quadrant, unless strong empirical justifications, involving relevant costs and benefits, support a more aggressive approach. Recall the first law of behaviorally informed regulation, which is that in the face of behavioral market failures, nudges are generally the right response. Moreover, those who emphasize behavioral market failures would seek to avoid both quadrants on the right-hand side.

5. On Welfare

My account of paternalism raises an immediate question. What counts as people's welfare? Does it mean happiness, narrowly conceived? Might it include whatever makes lives good and meaningful, even if happiness, strictly speaking, is not involved? I will return to these questions. For now, and to keep the focus on the issue of paternalism, I am going to understand the term "welfare" very broadly (and in a way that clearly separates the capacious idea of welfare from the narrower one of utility). Let us also notice the importance of distinguishing between "welfare" from the standpoint of the chooser and "welfare" from the standpoint of the paternalist.

With respect to the chooser, let us understand the term to refer to whatever choosers think would make their lives go well. (133) Choosers might, for example, care about the taste, amount, and nutritional content of food and drink. They might be happy to eat a lot of high-calorie foods, every day, simply because they enjoy them so much. (They might dislike or even hate calorie labels, on the ground that they detract from the enjoyment.) Or they might care not only about the economic benefits of fuel-efficient cars but also about the environment.

Their principal concerns might be religious; they might believe that fidelity to God's will is what is necessary to make their lives go well. When they think about their own lives, they may want to make choices that benefit other people--not only their friends and families, but strangers as well. They may want their lives to be meaningful, not merely full of pleasure, and they might sacrifice material and other benefits to achieve that goal. They strike their own balance; different people will choose differently. They may or may not enjoy exercising or smoking. They may or may not care a lot about health effects or aesthetics.

With respect to the paternalist, we can understand "welfare" in the same way, to refer capaciously to whatever the paternalist thinks would make choosers' lives go well. The paternalist might believe that choosers have the right ends, but that some kind of action is needed to ensure that they actually achieve those ends (perhaps because of the operation of System 1). Alternatively, the paternalist might believe that choosers have the wrong ends--perhaps choosers do not focus enough on health, or sexual abstinence, or on what makes life meaningful, or on obedience to God's will--and that some kind of response is needed, with respect to actions or beliefs, to ensure that the right ends are achieved. Though paternalists might have any number of views about what would make people's lives go well, my focus throughout is on paternalists who respect people's own views about their ends, and who seek to ensure that their decisions promote those ends. (134)

My working definition of paternalism does not include government efforts to prevent people from harming others--as, for example, in the case of assault or theft, or air pollution. There is nothing paternalistic about preventing people from beating you up, stealing your car, or making the air unsafe to breathe. Nor does the definition include government efforts to produce certain familiar and widely held social goals; consider laws designed to protect endangered species, or to prevent discrimination on the basis of race, sex, disability, and sexual orientation. None of these is fundamentally rooted in paternalistic considerations. (135) By contrast, the definition includes government efforts to override people's judgments about whether it is best for them to drink alcohol, to gamble, to drive while talking on their cell phones, or to eat a dozen chocolate peanut butter cookies before, during, or after dinner.

True, and importantly, some of these cases may involve harm to others. If you drive while talking on your cell phone, you might endanger other people, and perhaps a restriction could be defended for that reason. If you are making yourself drunk or even sick, you might affect others. In the cases just described, it is possible that regulation can be justified on grounds that have nothing to do with paternalism. To see some of the complexities here, recall recent rules that require increases in fuel economy. Such rules produce substantial social benefits by reducing air pollution and by increasing energy security; producing these benefits does not involve paternalism. But as we have seen, the strong majority of the benefits of such rules come from private fuel savings, (136) and producing these benefits might well be thought to involve paternalism. To get clear on the underlying issues, let us put third-party effects entirely to one side.

If we begin with this definition, the central concern about paternalistic interventions, elaborated most famously in John Stuart Mill's On Liberty, (137) is that people must remain free to choose as they see fit. The focus is on preventing certain action by the state, unless harm to others is involved. We should be able to see that while the principal objection is to ends paternalism, means paternalism can raise serious problems as well. Even in the face of behavioral market failures, why should public officials be authorized to interfere with people's judgments about the best means to promote their ends? Mightn't they err as well, and possibly more damagingly? These are important questions, but one of my principal goals here is to suggest that insofar as the Millian view neglects the existence of behavioral market failures, and the wide range of behavioral findings about human errors, it points in exactly the wrong direction. (138)

B. The Paternalist's Large Toolbox

To know whether and what kind of paternalism is involved, and to get clearer on the underlying concepts, we need to be more specific about the set of tools that government might use. Consider some possibilities:

1. Government says that no one may smoke cigarettes and that the sanction for smoking cigarettes is a criminal penalty--of $500.

2. Government says that no one may smoke cigarettes and that the sanction for smoking cigarettes is a criminal penalty--of $0.01.

3. Government says that no one may smoke cigarettes and that the sanction for smoking cigarettes is a civil fine--of $500.

4. Government says that no one may smoke cigarettes and that the sanction for smoking cigarettes is a civil fine--of $0.01.

5. Government does not say that no one may smoke cigarettes, but instead imposes a tax on cigarette purchases--a tax of $2.00 per pack.

6. Government does not say that no one may smoke cigarettes, but instead creates a program that provides a financial subsidy to smokers who quit for six months- a subsidy of $500.

7. Government does not say that no one may smoke cigarettes, but instead engages in a vivid, frightening advertising campaign, emphasizing the dangers of smoking. (139)

8. Government does not say that no one may smoke cigarettes, but instead requires packages to contain vivid, frightening images, emphasizing the dangers of smoking. (140)

9. Government does not say that no one may smoke cigarettes, but instead engages in a public education campaign designed to make smoking seem deviant, antisocial, or uncool.

10. Government does not say that no one may smoke cigarettes, but instead engages in a truthful, fact-filled educational campaign disclosing the dangers of smoking.

11. Government does not say that no one may smoke cigarettes, but instead requires packages to provide truthful information disclosing the dangers of smoking.

12. Government does not say that no one may smoke cigarettes, but instead requires cigarette sellers to place cigarettes in an inconspicuous place, so that people will not happen across them and must affirmatively ask for them.

13. Government does not say that no one may smoke cigarettes, but instead requires cigarettes to be sold in small containers, each having no more than five cigarettes. (Cigarette packs usually have twenty cigarettes now.)

Those who begin with the definition I have offered should acknowledge that all of these cases are not the same. If we are focused on leaving freedom of choice unaffected by government, and use the definition offered above, approaches (1) through (5), involving penalties, would count as forms of paternalism. Approach (6) might be seen as more difficult. Is it paternalistic to subsidize behavior? Does paternalism include not merely penalties but also subsidies? What about selective subsidies, as in, for example, a decision to allow recipients to use food stamps to pay for almost all food and drink, but not soda or chocolate bars? (In 2011, Mayor Bloomberg asked the United States Department of Agriculture for permission not to allow food stamps to be used to pay for soda; the Department denied the petition. (141)) Insofar as a subsidy is designed to influence a person's choices on the ground that those choices would not promote his or her welfare, it should be counted as paternalistic.

By contrast, disclosure of truthful information is not ordinarily understood as paternalistic. As we have seen, the basic reason is that disclosure requirements are meant to inform, not to displace, people's understanding of which choices will promote their welfare. But we have also seen several complexities here. First, disclosure of information will often affect that understanding, especially if--and because--it is selective. Second, the framing of information much matters, (142) and any disclosure requirement will inevitably include a certain kind of framing. It may be disputed whether a given disclosure requirement is simply informing choices; some forms of disclosure can certainly fall within the category of soft paternalism.

What about approaches (7) and (8), involving the use of vivid, frightening images? I have emphasized that psychic costs, no less than material costs, can alter behavior. Some people might think that efforts to frighten people, and thus to go beyond mere disclosure of facts and to grab the attention of System 1, can be taken as a form of (soft) paternalism. Under the definition I have offered, it is more than plausible to hold this view. Indeed, at least one court has drawn a distinction of this kind for First Amendment purposes, suggesting that compelled disclosure of facts is different from, and more acceptable than, compelled graphic warnings. (143) Any efforts to stigmatize a product, and to do so through emotional appeals, might be seen as imposing a psychic or affective cost on purchase or use. Imposition of affective costs is paralleled by the creation of affective benefits, which could come, for example, by efforts to portray certain activities, such as exercise or eating vegetables, in a positive light; such approaches could also be characterized as soft paternalism.

Approaches (12) and (13) also involve forms of soft paternalism. If officials put a product in an inconspicuous place, and if their goal is to discourage its purchase, they are steering people in a certain direction because they distrust people's own judgments about what would promote their welfare. No monetary penalty is involved, but time and effort must be expended to find the relevant goods. And if government requires a product to be sold in small containers so that people will consume less of it, it is behaving paternalistically insofar as it is malting it harder for them to make the choices that they prefer. True, many people may prefer that private or public institutions impose such costs, and some or many smokers may themselves share that preference because they would like to quit--but the point remains.
COPYRIGHT 2013 Yale University, School of Law
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Introduction through II. Paternalisms, p. 1826-1867; Storrs Lectures
Author:Sunstein, Cass R.
Publication:Yale Law Journal
Date:May 1, 2013
Previous Article:Rethinking the federal eminent domain power.
Next Article:Behavioral economics and paternalism.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters