As the VAPI crisis emerged in August and September, we were reviewing the final page proofs for our book, Escaping Paternalism: Rationality, Behavioral Economics, and Public Policy. The news put us in a quandary, as our book included a section in which we had used vaping as an illustrative example of paternalism in practice. Our analysis relied, in part, on then-current evidence that vaping was most likely safer than traditional smoking.
Knowing that the science might change, we had already included the following sentence: “Genuine health harms from e‑cigarettes may yet be found. But thus far, and to the best of our knowledge, no study has shown genuine and systematic health problems among direct users of e‑cigarettes, to say nothing of second-hand users.” Nevertheless, we worried that the VAPI crisis could be used to discredit our overall argument. We therefore prevailed on the editors to insert a last-minute footnote acknowledging VAPI (and also noting its tentative link to vitamin E acetate, an additive used primarily in black-market vape capsules). We hoped the footnote would serve to blunt any criticism.
But as time passed, we realized that the VAPI episode instead strengthened many key arguments in the book. If we were writing the book now, we wouldn’t omit the vaping section — we would expand it. But to see why, we need to explain the broader argument of our book.
THE BIAS OF BIASES
Escaping Paternalism documents the rise of a new form of paternalism, built on evidence from behavioral economics indicating that people are affected by irrational biases. The “behavioral paternalists,” as we call them, argue that policies to correct those biases could therefore result in better personal decisionmaking. The purpose of our book is to rebut this argument.
If biases can indeed justify paternalism, then modern-day paternalists have an embarrassment of riches. Wikipedia lists about 175 different cognitive and behavioral biases. Many of these are just scientific names for phenomena known since ancient times: impatience (“hyperbolic discounting”), greater attention to potential losses than gains (“loss aversion”), being affected by strong emotions (“hot-cold empathy gap”), and so forth. Other alleged biases are entirely novel. For very few is there a clearly defined mental process that produces the bias; for most, it is just a label given to a phenomenon.
The growing list of supposed biases serves as a grab bag of possible deviations from “rationality,” and hence justifications for interventions. What is rarely explained, however, is that the “rationality” in question is a special and technical concept created by economists for model-building purposes. It is certainly not synonymous with reasonability or justifiability, and deviations from it are not necessarily self-defeating or counterproductive. More often, deviations simply indicate preferences that differ from what economists have historically assumed.
For example, it is said that in certain circumstances people are “loss averse”; that is, they value a potential loss of $50 more than a potential gain of $50. In terms of traditional economic rationality, they “should” weight both equally. But why? Where, in general, is the harm in this? It is true that loss-averse individuals may fail to maximize their lifetime wealth — but given loss-averse preferences, maximizing wealth is apparently not their goal. There is no error to correct here.
Similarly, in some cases people may exhibit inconsistent preferences over time. On New Year’s Day, they resolve to eat less chocolate, but by the end of January (or sooner), they revert to their previous level of consumption. But why? The paternalists, assuming longer-term preferences are the “true” ones, blame weakness of will. But this is not the only interpretation available. Perhaps it is easier to give something up in the abstract moment of planning than in the concrete moment of doing. Thinking about the benefits of reducing chocolate consumption when one is full of food and good cheer is not the same as reckoning the cost. So maybe it is short-term preferences that are more reflective of the truth. Or maybe individuals simply have conflicting preferences within themselves, which they have not yet resolved (and possibly never will). Why should the state resolve the dispute for them?
Yet there are no doubt cases where people really want to change their behavior but require help. Reasonable people have many resources at their disposal. There are calorielimited packages of snack food, smaller cans of soda, and diet plans. Some people ban fattening foods from their cupboards. Others announce their plans to other people to generate support or subject themselves to shame if they depart from the plan. Not all deviations from a plan are failures, however. Rigid plans may not be the most effective. Rewarding oneself with periodic treats often makes the overall effort more sustainable — or not; it all depends on the specific individual and the context. Outside observers are apt to underestimate the degree to which individuals engage in selfregulation. If so, then they overestimate the degree to which these individuals need additional help.
Examples like these raise the question of how much corrective medicine for supposed biases is appropriate. Much of the research on this important quantitative issue is seriously flawed. In part this is because, as we stated earlier, would-be paternalist planners don’t know the extent of self-regulation. Perhaps more importantly, studies tend to look at one bias at a time, even though an individual can exhibit multiple biases, not all of which move the individual in the same direction or to the same degree.
Concerns like these should drastically complicate the process of designing corrective interventions, especially given the behavioral paternalists’ promise of a “careful, cautious, and disciplined approach” to policymaking. In practice, policymaking rarely approaches that optimistic ideal.
An example should make this clear. In the case of cigarettes, at one time it was considered sufficient simply to warn people about the dangers of smoking tobacco; all they lacked was information. But today, behavioral economists say that mere information is not enough because people suffer from optimism bias. Even if smokers and potential smokers are aware of the statistics showing the increased health risks from smoking, the same people may feel that somehow they are protected; it won’t happen to them. To counter optimism bias, behavioral paternalists suggest harnessing yet another bias: availability bias. This is the idea that an event will be considered more likely if it stands out in a person’s mind. So the proposed solution — already implemented in many places — is to present smokers (and potential smokers) with graphic images of sick people and diseased organs. The images may also be accompanied by “risk narratives” describing real instances of bad outcomes. Note that these presentations are highly biased, inasmuch as they typically represent the worst possible outcomes — not the median or most common outcomes. This is deliberate. The purpose is to use exaggeration to trigger an offsetting bias that will make people think, “Yes, this can happen to me.” In effect, the policy raises the subjective probability of a bad outcome to the smoker.
So far, so good, you may think. However, as economist W. Kip Viscusi and others have shown, smokers tend to believe that smoking is more dangerous than it really is. So now we have at least three relevant biases — optimism, availability, and overestimation — which somehow must be balanced so as to approximate the (potential) smokers’ “true” preferences, all things considered. Does anyone believe that regulators have in fact engaged in a calculation along these lines? Or have they simply mandated policies intended to reduce smoking, irrespective of people’s genuine preferences?
Another smoking-related example: The FDA has admitted that certain tobacco products that are not smoked — like snus — are safer than smoked tobacco. However, the FDA won’t allow the sellers to advertise this because some people might think “safer” means completely safe. So here we have the government purposely depriving people of relevant and true information because of how they might respond. Again, we ask: Did these regulators really “do the math” to determine that this policy would advance people’s genuine preferences, or did they simply wish to reduce tobacco use, full stop?
Old-fashioned paternalists would not care about this. Their objective is to reduce the incidence of “bad” behaviors. But modern paternalists say they’re just trying to help you do what you really want to do or to implement your “true preferences.” Furthermore, they admit that vices have their virtues; people do get pleasure from things that have costs. So what modern, scientific paternalists want is a correct costbenefit analysis in terms of people’s real preferences, somehow stripped of all bias. This means they should care about how policies are crafted in the real world.
PATERNALISM IS FOR CHILDREN
And so we return to the vaping/VAPI panic, which turns out to be emblematic of paternalistic policymaking in general. By December 2019, the CDC had concluded that VAPI was indeed a new phenomenon, not an older syndrome that had previously gone unnoticed, and that it was strongly associated with vitamin E acetate.
A judicious regulatory approach might have controlled the use of vitamin E acetate (a substance rarely if ever used in legal vape products anyway). But what did the actual public policy look like?
A panic-driven rush to action, of course. As we write in the book, “When disaster strikes, politicians react. From high-profile terrorist attacks to the latest mass shooting, the immediate response is nearly always a demand for action, often with little regard for efficacy or unintended consequences.” If we want to speak the language of biases, we could call this an “action bias.” The VAPI crisis bolstered support for a new federal law raising the minimum age for all tobacco products, vapes included, from 18 to 21. Some states and localities banned flavored vapes in the immediate wake of the crisis, and the FDA has now announced its intention to ban flavored vapes nationwide (with an exception for the less popular “tank” delivery systems). All of this, despite no evidence whatsoever of a link between flavors and VAPI.
Much of the vaping discussion has focused on young people because sweet flavors are presumed to appeal to them. If there is any group of people for whom paternalism is appropriate, surely it is children, as the word itself suggests. The problem with paternalism is not treating children like children, but treating adults like children. But notice that the actual policies are not directed solely at children. Prior to VAPI, vaping was already illegal nationwide for people under 18. The new laws target flavored vape capsules irrespective of consumers’ ages. Notably, the vast majority of adult users — approximately 90 percent — prefer nontobacco flavors, usually sweet ones. And, of course, the higher minimum age affects a significant group of nonminors: 18–20-year-olds. If advocates were truly concerned about protecting minors while respecting the choices of adults, surely more-targeted policies were available.
We have to speculate that, in truth, many advocates of vaping regulation harbor paternalistic motives across the board, not merely for children. Indeed, this was a primary point of our book’s section on vaping: that antivaping campaigns have exposed the underlying paternalistic motives behind smoking regulation in general. The public justification for smoking regulation often relies on seemingly nonpaternalistic goals, such as the protection of bystanders from secondhand smoke. But for vaping, the evidence of harm to third parties is close to nonexistent, in part because evidence of harm even to direct users is (at the moment) scant, and in part because vapes emit very little sidestream vapor that could affect nonusers. Absent the bystanders justification, we argued, paternalism is all that’s left. However, the current VAPI-driven focus on flavors suggests that paternalists have found a different fig leaf to cover their across-the-board paternalism: the hackneyed plea to “think of the children.” This is a common feature of paternalistic regulation in general: that it interacts with other arguments in a way that makes true motives difficult to discern.
A second way the response to VAPI mirrors paternalistic intervention in general is the seeming indifference of advocates to evidence. Even if you agree that it’s the state’s job to protect individuals from themselves, evidence of actual harm would seem to be necessary. But for paternalists, the mere potential for harm is enough. The campaign against vaping began long before any genuine evidence of harm had emerged, with arguments resting on the unknown effects of chemicals in vape fluid — or sometimes on the known effects of these chemicals when consumed at levels unseen in actual use. Then, when VAPI emerged, anti-vaping advocates used the new syndrome to justify interventions entirely disconnected from the most likely suspect: vitamin E acetate used primarily in blackmarket capsules. When and if evidence emerges of other genuine harms from vaping — which is certainly possible — we should expect more of the same: hasty policy interventions, uncoupled from evidence, motivated by a preexisting anti-vape agenda.
A third way the response to VAPI follows the pattern of other paternalistic interventions is in neglecting how the affected people may react, often in ways that thwart policymakers’ intentions. Higher taxes on vape products will tend to push smokers toward the most obvious substitute: traditional cigarettes. A recent National Bureau of Economic Research study indicates that vape taxes tend to reduce quitting rates for cigarette smokers. Given the widespread preference for flavored vapes, including among former smokers, flavor bans could easily have a similar effect. Bans and high taxes will both tend to encourage the growth of black markets — precisely the context where the additive responsible for VAPI was most prevalent.
But policymakers — driven by the desire to do something and facing pressure from activists — paid little attention to such concerns. Again, this is a common feature of modern paternalism grounded in behavioral economics. Models of “optimal sin taxes” rarely consider the possibility of consumers switching to other harmful products that serve similar needs. Proposals for default (or mandatory) enrollment in savings plans rarely account for people who offset their now-higher savings by incurring greater consumer debt or making early withdrawals. Proposals intended to focus people’s self-control on some goal, such as weight control or smoking cessation, fail to consider that doing so may divert selfcontrol resources (like attention and focus) away from other goals, such as studying or working productively. In general, behavioral paternalists employ relatively simple models that include only one or (if we’re lucky) two alleged biases at a time, with little attention to how biases interact.
Taken as a whole, the VAPI episode should lead to greater skepticism about both vaping regulation and paternalism in general. Even in the abstract, the behavioral case for paternalism is weak, resting on conceptual confusions and unfilled empirical gaps. Wouldbe paternalist planners simply lack the knowledge needed for the “careful, cautious, and disciplined approach” they promise. It should come as no surprise, then, that actual policy falls short. As we write in our book, “In the rough-and-ready world of practical politics, policy is shaped in a maelstrom of idealism, activism, ignorance, time constraints, power struggles, and special-interest pressures. It would be genuinely shocking for real-world policies to resemble those imagined by hopeful academics.”
