How Change Happens

by Cass R Sunstein

  4. See Lauren Willis, The Financial Education Fallacy, 101 Am. Econ. Rev. 429 (2011).

  5. Friedrich Hayek, The Market and Other Orders, in The Collected Works of F. A. Hayek 384 (Bruce Caldwell ed., 2013); italics added.

  6. See Sarah Conly, Against Autonomy (2012).

  7. See Stephen Darwell, The Value of Autonomy and the Autonomy of the Will, 116 Ethics 263, 269 (2006).

  8. See T. M. Wilkinson, Nudging and Manipulation, 61 Pol. Stud. 341, 347 (2013).

  9. Id. at 351.

  10. See Conly, supra note 6.

  11. Wilkinson, supra note 8, at 345.

  9

  Control

  In the late, great television show Lost, one of the central characters, with the telling name of John Locke, exclaimed at crucial moments, “Don’t tell me what I can’t do!”1 Locke liked doing things, but perhaps even more, he disliked being told that he could not do things. At every turn, he insisted on his own capacity for agency.

  The Lockean Exclamation, as I shall call it, is related to the psychological idea of reactance,2 which refers to people’s tendency to do something precisely because they have been told not to do it. It attests to a general phenomenon of which reactance is an example: much of the time, control, understood as liberty of action, has intrinsic and not merely instrumental value. When people feel that their control is being taken away, they will often rebel, even if exercising control would not result in material benefits or might produce material harms.

  Tocqueville offered a memorable account: “What has always kindled such a powerful love of liberty in the hearts of certain men is its intrinsic attractiveness, its inherent charm, independent of its benefits. It is the pleasure of being able to speak, act, and breathe without constraint under the sole government of God and the law. Whoever seeks in liberty anything other than liberty itself is born for servitude.” 3 As a matter of political science, Tocqueville may or may not be right. But he is certainly capturing an important aspect of political psychology: an insistence on personal agency, or the value of control, for its own sake, “independent of its benefits.” As he immediately added, what many people “love about” liberty “is not the material goods it gives them. They consider liberty itself a good.”

  Both psychologists and economists have investigated the intrinsic value of control, but we need to learn much more, not least because of its immense importance to both politics and law. Whenever a private or public institution blocks choices or interferes with agency, or is perceived as doing so, some people will rebel. In extreme cases, violence is possible. Consider the highly publicized example in 2012 when New York Mayor Michael Bloomberg tried to ban the sale of containers of sodas and other sugary beverages larger than 16 ounces. This was a modest initiative, but the effort created a significant backlash, in part because of what appeared to be a loud, collective Lockean Exclamation from New Yorkers. Any form of paternalism—with respect to food, health care, and cigarettes—risks running into a similar backlash. Of course some nations are quite different from others. The United States is especially prone to such backlashes. But we are one species: in every human heart, you can find the Lockean Exclamation.

  Because nudges preserve freedom of choice, they do much better than mandates and bans; we have seen that in diverse nations, they turn out to be approved by strong majorities. But even when majorities embrace nudging, minorities reject them. To the extent that some nudges create concerns, it is because they too trigger a version of the Lockean Exclamation, at least in some circles. For example, default rules (such as automatic enrollment in pension plans or green energy) allow people to opt out and to go their own way. But some people see them as putting a weight on the scales, and to that extent as diminishing agency and control. More broadly, it is even plausible to think that political movements—not only the movement for sex equality and for same-sex marriage but also for Brexit and Donald Trump—have been animated, in significant part, by the Lockean Exclamation. In the United States, the Lockean Exclamation played a large role in Trump’s presidency, not least in his attitudes toward immigration, NATO, and the Paris agreement on climate change.

  The intrinsic value of control matters for business as well as politics and law. To what extent should companies preserve freedom of choice—and emphasize to consumers and employees that they ultimately have the freedom to go their own way? Banks, grocery stores, soft drink companies, cell phone companies, and automobile manufacturers face that question every day; so do employers of every kind. The answer is greatly complicated by the fact that the intrinsic value of control is countered, in many cases, by a preference for preserving scarce cognitive and emotional bandwidth—and hence by an explicit or implicit choice not to choose. Inside everyone’s mind, there is an occasionally desperate plea: “You figure it out!”

  The plea might be heard when people purchase a home in a new city or enter a store with countless options, or when a government website asks them to register their preferences on lengthy forms as they (try to) apply for benefits, permits, or licenses. You might even want government to ban certain foods and drugs, not because you do not prize your own agency, but on the ground that it knows what is healthy and what is dangerous. Citizens might well be willing to say to their government: “We want you to decide which medicines are safe and effective for their intended use.” In view of the informational advantages of public officials, that could easily be a rational delegation.

  We can call the underlying plea (“You figure it out!”) the Madisonian Exclamation, after James Madison, who emphasized the many advantages of a republic over a democracy—in part because in the former, people delegate authority to relative specialists, who could make better decisions.4 In the history of political and economic thought, many people have emphasized that point, including (in very different ways) Max Weber and Walter Lippmann. As for the Lockean Exclamation, so too for the Madisonian Exclamation: We know all too little about when we will hear it, and when it is justified.

  Principals, Agents, and Rational Choice

  The most objectionable denials of control occur when someone deprives you of freedom to act or to choose—say, when you are forbidden to use your money as you see fit or to take the job that you want. But it is important to emphasize that any principal can voluntarily choose to relinquish some of that freedom. Indeed, that is an exercise of control, and it can be exceptionally important to both freedom and well-being. You might rely on or appoint an agent, who might have superior knowledge, might be immune from various biases, and might relieve you of the obligation to devote scarce time and limited cognitive resources to making choices that impose some kind of cognitive or hedonic tax. For these reasons, the Madisonian Exclamation has its appeal. Consider the areas of pension law and health care, in which many people are often willing to delegate a little or a lot of authority.

  At the same time, appointing an agent might be a big mistake. Any agent might have inferior knowledge, be ignorant of the principal’s real concerns, have her own biases, or be influenced by her own self-interest. Aware of those problems, even a Madisonian principal might decide not to appoint an agent or to appoint an agent with limited powers. Such a principal might think: I want to maintain control over my own pension and my own health care. Those with Lockean dispositions will be suspicious of the relinquishment of control that any such appointment might involve. As a matter of psychology, they tend to have faith in their own judgment and to distrust those who seek to make decisions on their behalf (hence the terms micromanagers and control freaks).

  With respect to the public sphere, it is often useful to think of the citizen as the principal and the agent as the government. Consider the questions of food safety, environmental protection, worker safety, and national security. I have suggested that regulation might be an implicit or even explicit delegation; citizens can decide to delegate the power to choose or even to nudge.5 In any society, citizens will differ intensely on whether such a delegation is advisable. />
  Many debates about what public officials should do involve an opposition between the Lockean Exclamation and the Madisonian Exclamation. I speculate that those who are suspicious of relinquishing control often have an immediate, intuitive, even automatic response to anyone who asks them to do so—and that response is intensely negative. In the terms of current psychological work, their automatic system (System 1) insists on maintaining control even if their deliberative system (System 2) counsels in favor of a delegation.

  In theory, the decision whether to choose, or instead to delegate the power to choose, should be a fully deliberative one, based on some form of cost-benefit analysis. Choosers might begin by thinking in terms of expected value: Would the payoff be higher with or without a delegation? Choosers might also ask about the value of saving limited time and attention. If the savings would be substantial, choosers might be willing to sacrifice something in terms of expected value. It also matters whether choosing itself has benefits or costs, in the sense that choosers enjoy, or instead dislike, the time that they devote to choosing.

  For some people, it may be interesting or fun to think about the best investments or the right health care plan. For other people, those choices are unpleasant and tiring, a kind of hedonic tax, and it is a great relief if someone else can be persuaded (or paid) to make the choice for them. Compare experiences in restaurants, where some people like to linger over the possibilities and to engage with the waiter, whereas others dislike the process and would just like the waiter to decide.6 There is a great deal of heterogeneity on this count with respect to relatively trivial and far more important decisions. Some people greatly enjoy thinking about their pension and the best investments; other people abhor that process. Some people like the idea of investigating potential mortgages; others really do not.

  Choosers might also consider whether the pleasure of a reward, and the pain of a loss, are amplified or instead reduced if they are personally responsible for the outcomes. Studies have shown that people value items they have selected themselves more than identical items that were selected for them.7 It is not hard to imagine a situation in which choosers would prefer (1) gaining $100 if that gain came from their own efforts to (2) gaining $110 if that gain came from someone else’s efforts: the subjective value of the self-attained $100 may be greater than that of the $110 that was attained as a result of an agent’s choice. On the other hand, personal responsibility might amplify the adverse effect of a bad result, because people might blame themselves more acutely than they would if the result came from the decisions of a delegate (even if the choice of the delegate was made personally).

  Debates over the legislative grants of authority to executive officials often turn on issues of this kind, as legislators might seek to avoid responsibility by giving discretion to administrative agencies (but might be held responsible for the grant). In areas that range from occupational safety to climate change to immigration, legislators might benefit if they can delegate authority to others—and blame them. They are making their own version of the Madison Exclamation.

  The Control Premium

  Consistent with the Lockean Exclamation, it has been found that people prefer options that permit further choice to those that do not.8 Similarly, people are willing to pay to control their own payoffs, rather than delegate, when faced with potential rewards: the control premium.9 A sense of control has also been shown to reduce stress and anxiety in the face of unwanted outcomes.10 For that reason, making a choice may reduce the aversive utility of a loss, leading people to prefer agency over delegation.

  Tali Sharot, Sebastian Bobadilla-Suarez, and I tested whether people will pay, or demand payment, to be choosers.11 We conducted an experiment with several trials. On each trial, we asked participants to make a simple choice between two shapes to maximize reward and minimize loss. The task took skill, and participants could see how they were performing. On “gain trials,” a correct choice would result in a monetary gain and an incorrect choice in no gain. On “loss trials,” a correct choice would result in no loss and an incorrect choice in a monetary loss. After performing the task for an extended period of time on their own, participants were given an opportunity to delegate the decision-making to an advisor. They could also see how the adviser performed; hence the expected value of the advisor was disclosed on each trial and participants’ perception of their own expected value was also elicited. This allowed us to examine whether participants made “rational” delegation choices given their beliefs when faced with potential gains and with potential losses. In other words, participants could compare their own performance with that of their advisers, and they could choose whether or not to delegate, seeing who was likely to do better.

  Our central finding was simple: Participants are willing to forgo rewards for the opportunity to control their own payoffs. In their own way, they were making, and acting on, the Lockean Exclamation. This preference was observed not only when participants were faced with potential gains but also when they were faced with potential losses. In one of our experiments, for example, a value maximizer would delegate 50 percent of the time, but participants’ average delegation rate was significantly lower. As proof of the control premium, participants were also much more likely to retain agency when this was not the optimal decision (i.e., failure to delegate) than to delegate when this was not the optimal decision (i.e., failure to retain agency). As a result of these failures, participants earned a lot less than they could have if they selected optimally.

  Importantly, we asked participants to make assessments of the (sub)optimality of their delegation choices. Their assessments were quite accurate, suggesting that they were entirely aware of the premium they were paying to maintain control. Our finding of a control premium is consistent with those in several other studies.12

  Why Control? When Control?

  The existing literature is instructive, but it remains sparse, and it raises many puzzles. One of the most important involves the sources of the control premium. In our own study, it is reasonable to think that people were making some kind of mistake. The task was not exactly fun, and so there was no hedonic value in making choices on one’s own. On one interpretation, people were using a kind of Lockean Heuristic, in the form of a mental shortcut, or presumption, that generally works well but that creates severe blunders: Do not delegate consequential choices to other people. Undoubtedly people do not use the Lockean Heuristic in all domains—but they certainly do in some important ones.

  Probably the most important reason for the Lockean Heuristic, associated with John Stuart Mill’s great argument on behalf of liberty, is that outcomes that we select ourselves often suit our preferences and values better than those that have been selected for us. Recall Mill’s insistence that the individual “is the person most interested in his own well-being” and that the “ordinary man or woman has means of knowledge immeasurably surpassing those that can be possessed by any one else.”13 Behavioral economics raises some serious questions about Mill’s argument to this effect. But to the extent that it retains (enough) validity, it argues on behalf of a heuristic that would lead to a control premium. Note that in our experiment, the Lockean Heuristic went wrong—but on plausible assumptions, it usually goes right, at least in many domains in which people find themselves.

  We might also speculate that a biologic system that provides higher intrinsic reward for things we have obtained ourselves could be adaptive. (Recall that people who win $100 through their own efforts might well enjoy higher subjective welfare than people who win $110 from the efforts of an agent.) If we learn that an action results in a reward, we can repeat that action in the future to gain more of the same. But if we do not execute an action to obtain reward (or avoid harm), we lose the opportunity to acquire a “blueprint” of how to gain rewards (or avoid harm) again in the future. Like most evolutionary arguments, this one is highly speculative—but it is plausible. The increased value of outcomes we have obtained ourselves emerges both from
their immediate utility and from the information they offer us for later choices.

  But we have also seen that for some decisions, people do not much care about control, and for others, they actually abhor it. In a restaurant in a foreign country, you might tell the waiter, “You choose.” When you are making a difficult medical decision, you might ask your doctor to figure it out for you. Control can be a cost, not a benefit.

  For rational actors, the relevant considerations can be organized into a simple framework, involving the costs of decisions and the costs of errors. Suppose that a website constantly asks you, “Would you like to change your privacy settings?” Many people do not welcome that question; it is a kind of cognitive tax (it imposes a decision cost), and it is not likely to reduce the number and magnitude of errors (because you are fine with your settings). But if you are deciding what place to buy or rent in a new city, you might welcome a question of that kind, and you might not do a lot of delegating: you have privileged access to your preferences and your values, and any agent might well make costly errors. You will welcome information and advice, but the ultimate decision will be yours. If someone intrudes, you might well offer the Lockean Exclamation.

  For boundedly rational agents, heuristics and biases, and supposedly irrelevant factors, will introduce complications. Take the case of the nonchooser offering the Madisonian Exclamation. Perhaps she is refusing to make a worthwhile investment—say, in the acquisition of knowledge. Perhaps she is myopic and is excessively influenced by the short-term costs of choosing, which might require some learning, while underestimating the long-term benefits, which might be very large. A form of present bias might infect the decision not to choose. Perhaps the nonchooser suffers from a kind of self-control problem. She is unable to slow down and to take the time to focus on something that is not especially interesting, at least not in the short-run, but from which she can ultimately benefit a great deal. Perhaps she does not realize that her own tastes are malleable. Or a nonchooser might be unduly affected by availability bias because of an overreaction to a recent situation in which her own choice went wrong.