Summary

When you begin the final chapter of Escaping Paternalism, Rizzo and Whitman (RW) seem ready to rest their case.  They neatly recap their nested argumentative strategy:

We began this book with an extended critique of the neoclassical model of rationality, which behavioral economists have rejected as a positive model of human behavior but nevertheless have accepted as a normative standard…

The rest of the book could easily be read as a series of “even if” arguments: even if we accept neoclassical rationality as true rationality, behavioral science has not advanced far enough to answer a number of crucial questions for policymaking, such as the generalizability of behavioral results across different contexts and the applicability of laboratory results in the wild. Even if we had better research on those questions, policymakers would face practically insurmountable local and personal knowledge gaps that would hobble attempts at crafting paternalist policies that are effective and cost-benefit justified. Even if we could acquire such knowledge, policymakers have little incentive to do the hard work of crafting good policies, especially when interested parties (such as rent-seekers, moralists, and bureaucrats) can tilt the legislative and regulatory process in their favor – and so much the worse if policymakers are afflicted by any of the cognitive biases attributed to regular people. And even if we set all of these concerns aside, behavioral paternalism creates the risk of a slippery slope toward more extensive and intrusive policies that go beyond what has been justified by theory and evidence, resulting in ever greater restrictions on individual choice.

They also fault their opponents for dire intellectual negligence:

The reality of behavioral policymaking stands in sharp opposition to the rhetoric. Behavioral paternalists have frequently emphasized the need for evidence-based policy (Thaler 2015b, 338) that should be implemented in a cautious and disciplined manner (Camerer et al. 2003, 1212). Their confident tone often suggests that it is anti-paternalists who eschew evidence and rely on gut instinct for policymaking. To paraphrase the Bible, they behold the mote in the anti-paternalists’ eyes while neglecting the beam in their own. Consider this telling passage from Sunstein that purports to recognize the need for better evidence:

With respect to errors, more is being learned every day. Some behavioral findings remain 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 far more to obtain a better understanding of how the relevant findings operate in the world. Even at this stage, however, the underlying findings have been widely noticed, and behavioral economics, cognitive and social psychology, and related fields have had a significant effect on policies in several nations, including the United States and the United Kingdom. (Sunstein 2014, 11–12)

Notice the speed of the transition from the need for cautious collection of more research to a congratulatory discussion of policy impact. If the need for better research were taken seriously, surely the “significant effect on policies in several nations” would be cause for concern, not approbation.

Instead of heading home, however, RW return to the stage for a stunning encore.  They list, explore, and critique an encyclopedic list of intellectual “escape routes” for paternalists.  In order of appearance: revert to objective welfare paternalism; appeal to obviousness; shift the burden of proof; loosen the definition of paternalism; rely on the “libertarian condition”; invoke the inevitability of choice architecture; focus on the irrational subset of the population; rely on extreme cases,;treat behavioral paternalism as a “toolbox”; and invoke fiscal externalities.

Each of these sections is rich and wise.  Yes, if you make paternalism vague enough, “We’re all paternalists now.”  A silly game.

Is GPS really a form of paternalism? Sunstein thinks so. In fact, GPS is one of his favorite examples, and not just when it’s a gift. He even calls GPS an “iconic nudge” that should be seen as “a form of means paternalism,” and one that paternalists should seek to build upon (Sunstein 2015, 61–62). GPS is, of course, almost always self-adopted… Is consulting a map also a form of paternalism? A map does, after all, simplify the territory that it depicts in order to ease the process of finding things.

But GPS is not the most trivial example of alleged paternalism. According to Sunstein, a restaurant providing a low-calorie menu for its customers is (or can be) a paternalist nudge (2014, 2). According to Thaler, giving someone accurate instructions on how to get to the subway is a form of paternalism (2015b, 324). Text-message reminders from doctors (Thaler 2015b, 342) and credit card companies (Sunstein 2015, 518) are also apparently paternalism… As Sunstein and Thaler see it, any time someone gives helpful advice, provides useful information, or gives a friendly warning, that’s paternalism. The paternalist bar seems to be remarkably low.

Escaping Paternalism then provides a series of “friendly warnings” about how to spot dangerous paternalism:

We want to focus on the characteristics that distinguish innocuous interventions from more problematic ones, irrespective of the label attached. Here are several factors, often overlapping and highly correlated with each other, that will help us to distinguish the harmless activities from more troubling ones…

RW name the following criteria: self-imposed versus other-imposed; invited versus uninvited; competitive versus monopolistic environment; coercive versus voluntary; public versus private; and informative versus manipulative.  On the latter point:

When a Swedish maker of snus, a form of smokeless tobacco, petitioned to modify the warning label on its product to say that it carries “substantially lower risks to health than cigarettes,” the FDA rejected the petition even though the claim is true given current medical knowledge. Why? The primary concern seems to have been that “labels that indicate lower risk may tempt people, particularly young people, to use tobacco products that they might not have tried otherwise” (Tavernise 2015). We don’t know whether any behavioral paternalists weighed in on this particular issue, but it is indicative of how providing truthful information, even that which is clearly relevant to some consumers, takes a back seat when the regulatory focus is on changing behavior.

In short, behavioral paternalism has a complicated relationship with the truth. Truthful disclosures may be useful from a paternalist standpoint . . . but not necessarily. To judge whether a given piece of information is desirable, the paternalist must have some notion of how the targeted agent should behave, all things considered. Then information can be delivered – or obscured – in the manner most likely to nudge the agent in the supposedly correct direction.

None of these “escape route” sections is without its charms, but RW’s critique of the “inevitability” argument is my favorite.  Highlights:

If paternalism is inevitable, it’s pointless to discuss whether or not to be paternalistic, and instead we should focus on how and how much to be paternalistic. But in fact, only choice architecture is inevitable. That means we can ask about other ways, besides paternalistically, that choice architecture might be chosen.

Even this wording may be too narrow, as “chosen” implies a level of intentionality… It would be a mistake, then, to assume that undesigned choice architecture is simply arbitrary or random. Here is one simple example: goods displayed in public view in a store, particularly those with price tags, are available for sale to anyone who can afford the price. When this rule is violated, merchants will usually say so explicitly (“Display items not for sale”). This simple default rule, a kind of choice architecture, minimizes confusion and eases communication between potential buyers and sellers. As far as we know, this practice was never explicitly chosen by anyone.

Consider a more detailed example: the case of refunds and exchanges. In the United States, there is no general legal rule requiring merchants to let customers return goods for a refund or exchange. Although there are some exceptions, especially with regard to door-to-door sales, for most transactions merchants are free to make all sales final (Ben-Shahar and Posner 2011, 115–116)… Consumers assume goods can be returned in good condition during a reasonable period, unless sellers explicitly say otherwise…

Furthermore, the usual default rule is suspended in a number of familiar cases, undergarments and perishable foods being well-known examples. In these cases, most customers are aware that refunds cannot be taken for granted, even if the merchant hasn’t said so explicitly. These goods presumably differ from other goods because the goods in question “depreciate” quickly after sale or use. What all of these cases together suggest is an ongoing market for default rules. If all-sales-final were the universal default, we might suspect that consumers aren’t thinking carefully and sellers are simply taking advantage of them. But the variation of default rules tells a different story – one in which market default rules are responsive to the needs of both buyers and sellers.

Furthermore:

Even when defaults are chosen deliberately rather than evolving, there exist ways to decide among default rules that do not necessarily involve paternalism. Here are a few possibilities:

* Defaults may be chosen in line with conventional expectations. This has the advantage of not surprising or confusing people who have become accustomed to the usual rules…

* Default rules may be chosen in a minimalistic fashion – i.e., assuming that people are not making promises or exchanges unless they explicitly consent to them. This would rule out, for instance, a default rule that assumes for-cause termination, as that constitutes an additional promise on top of the simple agreement to exchange labor for compensation…

* Defaults may be chosen in line with the goal of minimizing transaction costs, which usually would mean minimizing opt-outs…

* Defaults could be chosen to minimize how much choosers infer from the rule. In other words, people are more inclined to infer advice or information from some defaults (or framings) than others…

Analysis

1. Big picture: Rizzo and Whitman’s Escaping Paternalism is probably the best book about paternalism ever written.  The authors demonstrate an old-school mastery of their subject, and pack the work with so much insight that you’ll keep learning new things if you read it five times.  This book is a more valuable contribution to human knowledge than the latest paper in the AER, or even the whole latest issue of the AER.

2. While RW insist otherwise, I reluctantly conclude that their “inclusive rationality” verges on unfalsifiable.   Given their level of expertise, their refusal to name a single definite example of irrationality in the world is telling.  They’ve raised their bar so high even they can’t meet it.

3. Still, RW rightly point out that mainstream economists are too quick to call others irrational.  Yes, saying, “People changed their mind,” is an unsatisfying explanation of human behavior.  Yet people really do change their minds, and they’re hardly “irrational” to do so.  Indeed, stubbornness is a greater cognitive flaw than flightiness.  To quote Emerson:

A foolish consistency is the hobgoblin of little minds, adored by little statesmen and philosophers and divines. With consistency a great soul has simply nothing to do. He may as well concern himself with his shadow on the wall. Speak what you think now in hard words, and to-morrow speak what to-morrow thinks in hard words again, though it contradict every thing you said to-day.

4. While reading the book, I kept thinking about (a) kids; and (b) opioid addicts.  RW cover an immense range of topics, but gloss over the human beings almost everyone thinks should be treated paternalistically.  RW could have bitten the bullet and declared kids and opioid addicts to be “inclusively rational.”  Or they could have backed off and said, “Paternalism for such people clears the burden of proof.”  Disappointingly, they decline to take a stand.

5. RW have little patience for the “revert to objective welfare paternalism” escape route:

From our perspective, the very idea of objective welfare is implausible. No such thing as “welfare” exists until an individual mind comes into being. The individual mind generates values, desires, and preferences (typically through interaction with many other minds). And as it turns out, different minds can generate very different values, desires, and preferences. People are idiosyncratic; they want different things. And despite the many wants they have in common, they want the same things to a different extent. We see no plausible grounds for stepping outside of the mind to define what is good for it.

Despite my shared hostility to paternalism, I staunchly disagree.  The very idea of objective welfare is not merely plausible, but compelling.  (Even “inevitable”!)  Like everyone else, the homeless act on their own preferences; but almost all of them would have much higher objective welfare if they adopted a sober bourgeois lifestyle.  The same goes for children, alcoholics, drug addicts, and so on.  Left to their own devices, these impulsive humans generate “very different values, desires, and preferences.”  And left to their own devices, they ruin their lives.

6. RW convincingly accuse the “new paternalists” of being crypto old paternalists.

The apparent willingness of behavioral paternalists to favor some expressed preferences over others also colors the policy debate. They lend the veneer of science to what are in fact subjective judgments, giving policymakers the cover they need to implement policies based on prejudices and moralistic attitudes, such as the universal desirability of conventional virtues such as patience, moderation, and temperance.

So far so good.  But aren’t the new paternalists wrong for the right reason?  Namely: Despite RW’s incredulity, the conventional virtues of patience, moderation, and temperance are indeed universally desirable.  If someone taught their kids to be impatient, immoderate, and intemperate, you would baffled.  And if you had to quickly summarize the “root causes” of the miseries of homelessness, drug addiction, and so forth, isn’t the obvious answer that those who suffer fail to practice these bourgeois virtues?  This is no “veneer of science”; this is common-sense.

7. RW almost grant this point in their “Rely on Extreme Cases” section:

If you want to demonstrate the irrationality of human beings, one very simple strategy is to point to extreme cases: drug addicts whose actions destroy their lives, compulsive gamblers who lose everything they have and more, morbidly obese people who cannot even leave their homes. It is hard to believe that such people are acting rationally, even by the most permissive definition.

In these cases, perhaps we can safely indulge our intuition that their biases are truly damaging in terms of their own well-considered well-being, and if only they could see their situation globally they would truly wish to behave differently.

I agree that “intuition” (or just “common-sense”) says this.  The reason, though, is that “well-considered well-being” is a thinly-veiled version of objective well-being.  The morbidly obese are plainly acting in accordance with their own preferences, but they are acting contrary to their own long-run happiness.

Couldn’t you simply say, “The morbidly obese care about food more than happiness”?  You can and you should.  I’ve argued much the same about the “mentally ill.” But this doesn’t show that objective well-being is a myth.

8. If we concede that objective welfare exists, doesn’t this open the door to paternalism?  Sure, in the same sense that conceding that China exists opens the door to protectionism.  Logically speaking, the cleanest way to prevent a trade war with China is to deny that China exists.  This, however, is an absurd claim, and will at best convince your most dogmatic allies.  Similarly, the cleanest way to prevent paternalism is to deny that objective welfare exists.  But this, too, is an absurd claim, and will at best convince your most dogmatic allies.

9. What then is the reasonable position on paternalism?  Per Michael Huemer’s The Problem of Political Authority, we should begin with a strong but defeasible moral presumption against using coercion to increase objective welfare.  This is hardly an exotic libertarian position; in their personal lives, almost everyone holds such a presumption.  To justify paternalism, you have to show that the net benefits of paternalism are large, even after factoring in knowledge problems, public choice problems, slippery slopes, and all the other practical difficulties RW detail.

10. This Huemerian position explains why it is typically OK to treat your own young children paternalistically.  Why?  Because they’re incompetent, you really do know better, and you really do have their best interests at heart.  (And don’t forget, “My house, my rules.”)  The same goes for elderly relatives with dementia, though the bar should be much higher because the love of the old for the young is so much stronger than the love of the young for the old.

11. Doesn’t this justify actually-existing paternalistic government policy?  Hardly.  As RW explain, governments habitually use their powers with gross negligence – and plenty of earlier researchers document the collateral damage of policies like Prohibition and the War on Drugs.  Yes, the world is complex; but as long as there is a strong moral presumption against coercion, the complexity of the world tips the moral scales strongly toward inaction.

12. To return to my own nagging concern, what about the opioid addicts?  I say government should leave them – and their suppliers – in peace.  Why?  Yes, they’re acting strongly against their own objective welfare.  Yet this pales before (a) the classic economic arguments against prohibition, (b) RW’s pragmatic concerns, and (c) the strong moral presumption in favor of leaving strangers alone.  The case against parents forcibly placing their adult children in rehab for their own good is weaker, though even there we should be skeptical.  As the old joke goes, “How many therapists does it take to screw in a lightbulb?  Just one, but the bulb must want to change.”