Cass Sunstein’s Extraordinarily Weak Case for Paternalism

Mario J. Rizzo*

In a recent article in the New Republic, Cass Sunstein tells us why paternalism is our friend. I shall be concerned here with both the actual content of his individual arguments and with the structure of his argumentation. The latter is quite important because, oddly enough, most of the arguments he brings up are those against paternalism.

He admits that all of the following arguments have substantial weight, at least generally speaking:

1. People might genuinely want to eat sweets, smoke, gamble and so forth. So we must keep that in mind. We don’t want to interfere with the true ends of individuals.

2. However, because of systematic biases, people make mistakes in satisfying their true preferences. They do not absorb information presented to them well enough; they get fooled by the irrelevant aspects of how a problem is framed; they suffer from weakness of will; they systematically over or under estimate probabilities. Thus their actual welfare is not improved by their freely-chosen behavior. But, on the other hand, it is true that government should not short-circuit the individual and social learning processes that making these mistakes often generate. People will only learn if they bear the consequences of their mistakes.

3. Of course, public officials (academics?) have their own motivations and biases. They may be affected by special interest groups (rent-seeking and all that) and by their own cognitive and behavioral failures.

4. The market provides all sorts of options: healthier food, health clubs, advice books, motivational coaches, and more. In the long run free competition will help people make better choices. We do not want simple paternalistic solutions to short-circuit all this.

“These objections have a lot of force,” Sunstein affirms.

In addition, freedom of choice may be a separate distinct goal. It can be an element of welfare insofar as coercing, nudging and even preaching may annoy people. But is also can be a trump against welfarism insofar as freedom of choice is an autonomous end, that is, an end in itself.

And yet: Paternalism can still be my friend.

This is because, fundamentally, Sunstein sees all of these objections as assessable on a case-by-case basis. They may be determinative in some cases but not in others. So now we take out our trusty cost-benefit tools.


In the first place, the trump-argument is, by definition, not subject to cost-benefit analysis. Its role is to protect an important value against the Benthamite calculus. And the idea of freedom of choice as a part of welfare is cute, but how do we measure its value and trade it off against other elements of welfare?

This might be especially difficult because Sunstein (and Richard Thaler) are constantly reminding us that we cannot trust superficial preferences because they might be affected by ignorance, lack of attention, availability biases, framing illusions and on and on. Therefore, we must measure the true preference for free choice to strike the correct cost-benefit balance.

We are told that the pursuit of welfare is plagued with numerous and fatal mistakes. But the behavioral economics mistake-industry is, on its own terms, quite out of control. The last time I checked the Wikipedia entry on cognitive biases the number of claimed biases was over one-hundred and growing every day. These biases move often in opposite or in otherwise conflicting directions. It would take a fantastic intellectual achievement to lay the ground-work for a serious empirical evaluation of all this.

For example, people may “protect” themselves from bad information by an optimism bias whereby, even if they were to know the true probabilities of smoking-related disease, they would feel subjectively protected from the impact. The probabilities apply to thee but not to me. Keep in mind, however, that cigarette smokers over-estimate the probabilities of health problems associated with smoking, according to research by Kip Viscusi and others. So maybe this all balances out. But maybe not.  In the event it all results in too much smoking relative to true preferences, we can appeal to another bias to rescue us.

Paternalist policy-makers can use public money to finance scary commercials about smoking diseases so that horrible scenarios about possible disease-filled futures become more available to the mind. This might offset the interaction of the optimism bias with the overestimation of probabilities. Perhaps now everything is at the paternalist optimum. We don’t really know.  I have never seen figures estimating these effects and totaling them all up. 

I believe that Sunstein and others really don’t think they really need any of this. They have direct intuition that the optimal amount of smoking is near zero.

Sunstein would insist “even if these points are compelling in the abstract, they are not decisive in all cases.” Leave it to the experts to consider everything and come up with the right answer.

But, of course, the experts do not determine policy. The political process does. Surely the process is rationally-ignorant and irrationally-biased.  Sunstein knows all this; he admits it. And yet it has little impact on his views.

Wave all of this away. Can we embark upon a policy of paternalism and restrict it only to those cases that pass “cost-benefit” analysis? I put the term in quotation marks because what goes under that name here must be the vaguest and most unreliable of exercises.  I cringe, for example, at the use of such evidence as the statements of former smokers that they are happier now. (Sunstein raises this gem in his article.) We can be forgiven for asking whether their “happiness” is a function of these people having been socially and morally beaten down by paternalists and are now grateful for psychological release. Is this a true welfare improvement by autonomous individuals? In any event, these are the people who stopped, not the ones who haven’t. No one has denied that people who want to stop make efforts to stop and may be happy with their decision.

The fundamental linchpin of Sunstein’s argument is that he claims to advocate a paternalism of means, not a paternalism of ends. The individual tells us what his ends are (his true, informed, autonomous ends) and paternalists simply tell him how to get there. Paternalists are our guides in this world of confusion and irrational biases.

Strangely, paternalists of the Sunstein-Thaler variety accept the hyper-rationality view of what human beings ought to be, while they reject this as a description of what people are really like. The fault lies not in the model of hyper-rationality but in the real human beings who must be fixed. Salvation lies in approaching hyper-rational behavior as closely as possible.

The idea that their paternalism is one of means only presupposes that someone somewhere is able to identify the purified preferences of the rational individual. These must be purified of the contaminants of cognitive and behavioral biases. Aside from netting out the conflicting influences of a serious bias analysis, the paternalists must not mistake simple desires for economic preferences. Economic preferences are simple desires confronted with true opportunity costs. For example, we all want more savings – but do we want it at the cost of present consumption? A preference for more saving accepts and bears the opportunity cost.

In listing many “legitimate concerns” and general objections that have “a lot of force,” Sunstein hopes to limit their impact. He can say, in effect, “I admitted these problems. I am way ahead of you.” But what has he really done? He has simply stapled together his paternalism program with the admissions of various possible difficulties. He does not deal with the latter in a serious way. He does not have to tools to do so. He can only deal with simple stylized examples that are designed to be easy.

He says in his conclusion: “But legitimate concerns about illegitimate paternalism should not be allowed to prevent officials from seeking to identify the best ways to improve people’s lives, even if they end up influencing people’s choices.” To the half-asleep reader this may sound reasonable.

For the awake among us: Sunstein has not solved the knowledge problem of new paternalism. He has not figured out a reliable way to determine people’s true, autonomous preferences (if they exist, of course) nor how to negotiate the thicket of biases and conflicting quantitative effects. He trades on the superficial appeal of contemporary bromides about how we should live. He hides behind the fiction that paternalists are only humble efficiency experts to who tell us how to get where we want to go.

*Dr. Mario J. Rizzo is an associate professor of economics and co-director of the Austrian Economics Program at New York University. Professor Rizzo is also a member of the Journal of Law & Liberty Board of Advisors.  Academic articles by Dr. Rizzo relating to paternalism include the following: