Abstract
Introduction
Since Cass Sunstein and Richard Thaler coined the terms “libertarian paternalism” 1 and “nudging,” 2 the questions whether and when governments are justified in treating their citizens paternalistically have gained renewed attention from both academics and politicians. In his latest book Why Nudge? The Politics of Libertarian Paternalism, Sunstein rehearses and refines some of his arguments in favor of nudge paternalism and further analyzes implications for both theory and practice. More than in his previous work, and contrary to what critics often assume, Sunstein moves with caution and subtlety, arguing that only under specific conditions can specific forms of paternalism be justified. This is also what Julian Le Grand and Bill New do in their book Government Paternalism: Nanny State or Helpful Friend? While engaging with the general arguments in favor of and against paternalistic interventions by liberal, democratic governments, they do not attempt to settle the debate at some abstract level. Like Sunstein, they thankfully redirect the discussion to specific cases in which specific paternalistic interventions can be legitimate.
As such, both books are of interest to everybody involved and interested in the question whether democratic governments should protect citizens from themselves. The paternalist issue is an urgent one, given the magnitude of lifestyle-related health problems, traffic accidents and suicides, which all belong to the leading causes of death in developed contemporary societies. It deserves the careful scrutiny and insightful discussion that readers can find in both books. Even though their arguments and conclusions often run remarkably parallel, both books do not refer to each other. While Sunstein uses a less academic style and approach, Le Grand and New more systematically analyze the issues at hand with a rigor that scholarly political philosophers and theorists will appreciate.
In what follows, I will highlight some of the similarities between both books with respect to both the definitions and the justifications of paternalism they provide. I will summarize the nuanced positions they defend, frame them in the larger debate and end with some critical remarks.
Definitions of Paternalism
The standard definition of paternalism is that of Gerald Dworkin, according to whom a person or institution (Y) intervenes paternalistically towards a person (X) if (1) Y interferes with X’s liberty/autonomy, (2) against X’s will/consent (3) in order to promote X’s well-being/welfare/good/happiness/interests/needs/values. 3 In line with his earlier work, however, Sunstein refers to examples of nudging—such as highlighting some of the options on a menu—that are paternalist but do not violate people’s liberty. This implies the need to drop Dworkin’s first two conditions. Le Grand and New (12–13) add the example of subsidizing activities that are believed to promote people’s well-being but that expand rather than restrict people’s freedom of choice.
Keeping only Dworkin’s third condition (to promote people’s well-being), however, reduces paternalism to beneficence. 4 This why we need an improved definition, which can be found in both books. According to Sunstein (54) a paternalistic “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.” Obviously, these steps can include laws, incentives, taxes, nudges, campaigns and even arguments. Like Sunstein, Le Grand and New (22–23) drop Dworkin’s first two conditions but add that Y is paternalistic if Y judges that X will suffer from reasoning failure and, without Y’s intervention, will fail to make the “right” choice. Crucial in both definitions is thus Y’s distrust of X’s abilities to devise and promote X’s own conceptions of the good.
In their third chapter, Le Grand and New provide a neat typology of different types or paternalism (soft versus hard, means versus ends, volitional versus critical, legal versus moral). Interestingly, the distinction between ‘soft’ and ‘hard’ paternalism, widely discussed in the literature, can be understood in two ways. First, Le Grand and New (26–27) follow the approach most philosophers take in understanding soft paternalism as directed at people who lack autonomy or voluntariness, such as children, the mentally ill, et cetera. 5 Hard paternalistic interventions then intervene with people who (are about to) autonomously and voluntarily harm themselves. In this sense, soft paternalism is more justified because the oft-heard criticism that paternalism violates people’s liberty and/or autonomy, does not apply. The people it targets are, after all, not genuinely free and autonomous in the first place.
In contrast, Sunstein (55–59) follows the approach most economists take in understanding soft paternalism as imposing no substantial (material or financial) costs to the person at hand (e.g., installing a default option with an easy opt out), 6 whereas hard paternalism is more costly (e.g., heavily taxing or fining some activity). Sunstein’s continuum from softer to harder paternalism is thus based on the magnitude of the costs imposed on Y, not on Y’s degree of autonomy or voluntariness. Here too, soft paternalism is considered less pernicious than hard paternalism, because it better preserves people’s liberty and autonomy.
Interestingly, nudges can be understood as “soft” paternalism in both meanings. Not only do they impose negligible costs, they also make use of deeply rooted psychological, cognitive and motivational mechanisms that influence people’s behavior. Empirical research from psychologists and behavioral economists reveals a multitude of biases and heuristics that, nudge enthusiasts argue, choice architects can tap into in order to steer people’s choices so that they are more likely to promote their own well-being. Here too, there is no such thing as (full) autonomy or (unrestricted) voluntariness to be violated anyway. While Sunstein does not formulate it as such, the general idea behind nudging is that people quite often do not make genuinely autonomous choices. Because their decisions are influenced by these psychological mechanisms (internal), nudgers can design the choice architecture (external) to improve them.
This brings us to another distinction, that between “means” and “ends” paternalism, which “has rarely been made in the literature, but . . . is of great importance” (Le Grand and New, 27). Take John Stuart Mill’s famous example of a person approaching a dangerous bridge. If we know or can reasonably assume that he wants to get to the other side safely, he is obviously mistaken about the means to achieve that goal and will predictably fail to promote his own good. As means paternalists, we would distrust his judgement and intervene (by warning or even forbidding him to cross the bridge). Now, if we know that this person wants to commit suicide, means paternalists would not intervene but ends paternalists would. The latter believe people can be mistaken not only about the means to their goals but also about their goals themselves.
Justifications of Paternalism
All this has obvious normative repercussions. If we are indeed often not as rational and autonomous as we think we are and want to be, then why should government not (be allowed to) help us make the choices that we would make if we were rational and autonomous (soft paternalism in both meanings of the term)? Again, both books develop remarkably similar arguments, stressing the desirability of government engaging in soft means paternalism in circumstances where people predictably make mistakes.
Sunstein (34) identifies a number of “mistakes that produce significant harms and problems, and that should be counted as behavioral market failures.” Amongst these are the status quo bias, time inconsistency, self-control problems, salience, forgetfulness, unrealistic optimism, the availability bias and other problems people experience when dealing with probabilities. Le Grand and New (82–101) refer to empirical evidence of roughly the same biases and problems that can affect decision making: limited technical ability (biases in estimating probabilities, anchoring and other framing effects), limited imagination (status quo bias, endowment effect, salience, myopia), limited willpower (akrasia, visceral decision making) and limited objectivity (confirmatory bias, overoptimism, hindsight bias).
Both Sunstein (50) and Le Grand and New (34–35, 101–4) argue that this evidence provides a rationale for government to be means paternalist, because it shows that people often err about the appropriate means to achieve their goals: “paternalistic interventions should be confined to substituting its judgement for the individual’s only both where there has been a failure of the means by which the individual tries to achieve his perception of the good, and where the government could do better” (Le Grand and New, 23). In these cases, there is an objective standard, such as logic, to assess reasoning failures, which enables psychologists and behavioral scientists to study them empirically. However, because “experimental evidence does not exist for identifying where people might fail to have the right ends” (Le Grand and New, 104), government should refrain from ends paternalism. Imposing ends on other people is what anti-paternalists rightly deem moralistic and deplorable.
Interestingly, both books engage in this respect with John Stuart Mill, according to whom individuals should be sovereign in devising their own standards and leading their own lives accordingly. According to Mill’s infamous harm principle, “the 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. . . . Over himself, over his body and mind, the individual is sovereign.” 7
Now, according to Le Grand and New (177), “John Stuart Mill was wrong.” In specific circumstances (when people’s choices of means are known to be biased), specific kinds of paternalism (correcting people’s means with a minimal impact on their autonomy) can be justified. In contrast with ends paternalists, means paternalists do respect people’s (sovereignty over their) goals and actually help people achieve them. Sunstein (5) also argues that Mill’s principle “cannot get off the ground” whenever people are prone to error and government can intervene to make their lives go better. Through a critical analysis of the justifications of Mill’s principle in terms of knowledge, diversity, learning, freedom and dignity, Sunstein answers some of nudging’s critics. Instead of defending a meddlesome nanny state that systematically uses psychological tricks to steer people in directions it deems desirable, he talks of “occasions for paternalism” (25). Take the epistemic argument for Mill’s harm principle, according to which individuals know best—better than others—what is good for them. According to Sunstein, people do not always know best but predictably make mistakes. Think of people having to assess probabilities (e.g., underestimating the chance their smoking behavior will lead to lung cancer) or making choices on the basis of irrelevant aspects of the choice situation (e.g., picking whatever is at eye level in a cafeteria). Obviously, Sunstein is careful to avoid the “government knows best” reproach aimed at paternalists in general but also at his “nudge squad” in the Obama Administration. As a liberal, he therefore avoids ends paternalism and prefers “softer” over “harder” paternalism.
Techniques of Paternalism
Even if we grant government such occasions for paternalism, the question remains which kinds of interventions are most desirable. Sunstein obviously favors nudging, because it preserves people’s liberty. Without coercing people or significantly altering their options, nudges steer their behavior towards “their welfare (as they themselves understand it)” (Sunstein, 107). Both books provide a number of original proposals such as requiring smokers to buy a permit every year (Le Grand and New, 157–58) or offering personalized default rules for savings and insurances (Sunstein, 99). Such nudges are useful additions to the paternalist’s toolbox, which already consists in legal restrictions and fines (seatbelts), taxes (tobacco), or the imposition of substantial costs (speedbumps). Le Grand and New but also Sunstein argue that such more far-reaching measures can be justified as well (and are in fact deemed legitimate by most people to whom they apply).
Answering “the challenge from autonomy,” Le Grand and New (105) believe that paternalist interventions can be justified, even if they offend people’s (perceived) autonomy or (perceived) capacity for self-rule and self-determination. This (perceived) autonomy loss can be outweighed by benefits in terms of well-being. Le Grand and New (132) thus rightly stress the “need to trade off the values of autonomy and well-being against each other.” While non-paternalistic governments that treat “autonomy as a lexicographic value” (Le Grand and New, 129) or as “a trump card” (Sunstein, 163) see their citizens dying from coronary diseases caused by obesity, lung cancers caused by smoking and traffic accidents caused by over-optimism, paternalistic governments can easily lower these numbers. If the stakes are high indeed, the balance might well tip to protecting people’s well-being over protecting their autonomy (which in practice is often not as pristine as we think it is).
This willingness to trade off well-being and autonomy—Sunstein (164) speaks of a “careful consideration of benefits and costs”—distinguishes nuanced paternalists from anti-paternalists. If we would all make autonomous choices that systematically make us well off, then paternalism would be superfluous. And surely, respecting people’s autonomy is generally a good way of identifying what exactly makes people well off. But this does not take away that people sometimes can go wrong and welcome some help, especially when the stakes are high.
Of course, we should acknowledge the possibility of government, consisting of policy makers and public servants, being fallible as well. They might go wrong in identifying our well-being or have incentives to promote their own well-being rather than that of the people they are supposed to serve. While Sunstein (115) simply remarks on this issue, Le Grand and New (167–175) devote a whole chapter to it. Only if we can safely assume, they argue, that government can identify people’s mistakes and knows how to redress them, will paternalistic techniques actually improve people’s well-being.
Critical Remarks
Before concluding, let me formulate some critical remarks. As for Sunstein, it is quite disappointing to see that he does not take much time to answer his critics. A lot of the well-known criticisms of nudging do not figure in the book at all, 8 leaving open important questions surrounding the definition and justification of nudges. Can we really understand informing people as nudging (Sunstein, 17, 139) or does that not even count as paternalistic (Le Grand and New, 15)? And if the use of mechanisms, biases and heuristics is what characterizes nudges, does this make them inherently manipulative? Are they not patronizing towards the adults they target? Can they be called genuinely libertarian if they are so successful in influencing behavior? What does it mean for governments to be transparent about nudging?
As for Le Grand and New, my main quarrel lies in the distinction between “means” and “ends” paternalism, which is pivotal to their argument. The interventions they support—seatbelt laws (151), automatic enrollment in pension schemes (159–61) and opera subsidies (153–54)—are systematically framed as means paternalist, but can also (or even better) be understood as ends paternalist. Because most of us deem it undesirable if people prefer recklessness in traffic over concern for their own safety, we think it is okay to fine them and heavily campaign to change their views (about their goals, not their means). The same goes for pension saving (people should resist their short-term gratifications pushing away their more prudent concerns) and for opera subsidies (it is good for people to develop a taste for culture).
People’s failure to realize their fundamental goals is not only due to them being mistaken about means. Assume that I am weak-willed and procrastinating when it comes to some longer term project that I find valuable, such as finding a job or losing weight. My behavior clearly goes against my goal, opening an occasion for paternalism. If we deem it worrisome if people systematically favor their short term goals over their cherished long term goals, why should we not endorse ends paternalism? If I am too lazy to get out of bed for a job interview, I really want a paternalist (mother, alarm clock or whatever) to help me achieve what I really want. My desire to stay in bed is not some inappropriate means to achieve the goal I want but another goal that interferes and conflicts with it and that I want to overcome.
Biases and heuristics can cause people to err with respect to both their means and their goals. While Le Grand and New are right in stressing that ends paternalism inevitably refers to a standard that is not objective, I believe a subjective standard can suffice to assess people’s errors and to justify paternalism. We can criticize a person’s goals by referring to his other, more fundamental goals. Even most machos do not want to die in a car crash. And people’s preference for smoking can be deemed problematic if they have a fundamental desire to lead a long and healthy life. One’s view of the good life constitutes a higher-order preference that rightly gains precedence over the multitude of preferences that are being formed on the fly under the influence of framing and other effects.
In this respect, Sunstein (70) acknowledges that everything depends on “the 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 and abstract of ends.” In my view, it makes sense to be more specific about people’s goals and try to assess with which of these people identify and which they regard as alien to or distracting from their overall life plans.
Conclusion
From their book’s subtitle, we can already see that Le Grand and New see a possibility for government to become a “helpful friend” rather than a “nanny state.” Like Sunstein, they believe it can do this, as long as it respects people’s capacity to devise their own life goals and thus limits itself to means paternalism. To claim that paternalists inevitably disrespect us because they substitute their goals for ours and/or because they violate our liberty and autonomy is therefore all too crude. To claim that government always knows best is indefensible as well. Both of these books present nuanced arguments and insightful examples that show why a middle road is both possible and desirable. As with the basic question parents themselves struggle with—how to treat their children—questions surrounding government paternalism and the appropriate relationship between government and its citizens will remain contentious. While some favor a looser approach, believing that trial and error will eventually lead to better choices, others favor a stricter approach to speed up the process of improving choices. There is no reason not to tackle these issues head on, in full recognition of the often inevitable conflicts between key values such as liberty, autonomy and well-being.