Summary

Rizzo and Whitman (R&W) begins with a primer on the “new paternalism” – the influential policy reform movement powered by the engine of behavioral economics.

For most of history, paternalists have drawn their support from religious or moral notions of goodness. They have claimed special knowledge, from God or some other source, about how people ought to live. The overtly religious character of temperance movements in the nineteenth and early twentieth centuries exemplifies this kind of “old” paternalism. Given their moral convictions, the old paternalists did not usually appeal to the preferences and desires of those whose behavior they wished to regulate. The very desire for alcohol – or drugs, or deviant sex – was condemned directly.

[…]

Behavioral economists, like most economists, still accept subjectivism. However, they have rejected the notion of accurate preference revelation. Experimental evidence suggests that individuals may systematically deviate from the behavior that would best satisfy their own preferences. The list of alleged deviations from strict rationality includes – but is not limited to – status quo bias, optimism bias, susceptibility to framing effects, poor processing of information, and lack of willpower or self-control. To the extent that these phenomena cause people to make errors, paternalist policies can in principle help them to do better – not by some exogenous standard, but by their own standards. This is the defining feature of new paternalism that distinguishes it from the old: the new paternalists claim to help people better achieve their own preferences, not someone else’s.

Cass Sunstein and Richard Thaler are the popular and intellectual leaders of the new paternalists.  As R&W succinctly explain:

Cass Sunstein and Richard Thaler, coming from an economic perspective, support policies that will make people better off “in terms of their own welfare.” That welfare is best advanced by the choices people would make “if they had complete information, unlimited cognitive abilities, and no lack of willpower” (Sunstein and Thaler 2003, 1162). In other words, Sunstein and Thaler define the correctness of choices in terms of what people would choose for themselves – if only they were not afflicted with cognitive biases.

New paternalists then present novel rationales for a wide range of government interventions.  For example, they advocate “sin taxes” to help not society, but the “sinners” themselves.  Fat taxes help fat people weigh what they “really want to weigh.”

R&W close the chapter by explaining their multi-stage argumentative strategy.

Our case will consist of a series of challenges – in effect, hurdles that behavioral paternalist proposals must clear in order to be justified as a matter of policy. We will begin with the most abstract and conceptual, then proceed to more pragmatic and applied challenges.

The most abstract: Behavioral economists take simple models of rational choice too literally.

We refer to the traditional economic definition of rationality, adopted by mainstream and behavioral economists alike, as “puppet rationality.” It is a brand of rationality well suited for building models of how the world works. Models are not unlike stage plays, and puppets are the players. The puppets are always well behaved. They play the roles they were designed to play. They follow the rules. They have no motive force of their own. Real human beings, however, are not puppets. Their preferences and behavior may deviate from what is expected of agents in a model. But such deviation does not provide sufficient warrant for deeming them irrational.

But even if standard notions of rationality are solid, the much-hailed experimental evidence is sorely lacking in external validity:

Much of the evidence for “failures of rationality” derives from experimental settings that are effectively context-free. Such experiments may identify “raw” or unmodified propensities in human behavior in the laboratory. But they do not tell us how strong those propensities are “in the wild,” where people make real decisions…

We also argue that people are often (though not always) aware of their cognitive biases, and sometimes they act to counteract or control them. People “self-debias” in myriad and idiosyncratic ways…

Pragmatically, R&W argue that the new paternalism suffers from a massive yet neglected knowledge problem.  Most fundamentally: “If actions do not reveal preferences, then what does?”

Last, the book explores public choice and slippery-slope problems with paternalism in great detail.

This is only Chapter 1.  In Chapter 2, R&W appeal to “inclusive rationality” to critique behavioral economics at its root.

Are human beings rational? …But it is not easily answered, because the word “rational” has many meanings. Rationality can simply mean purposefulness, that is, trying to use the best means available to satisfy your goals or preferences given your beliefs about the world. It can mean taking an abstract approach to solving problems, applying universal systems of thought and inference, and following scientific methods. It can mean avoiding errors of logic and reasoning. It can mean revising one’s beliefs in accordance with Bayes’ Rule. It can mean having preferences that conform to certain axioms – transitivity, completeness, and so forth – which together guarantee the preferences are internally consistent and have a certain structure. In neoclassical economic theory, it has historically meant all of the above.

[…]

In this book, we will defend what we call inclusive rationality. Inclusive rationality means purposeful behavior based on subjective preferences and beliefs, in the presence of both environmental and cognitive constraints. This notion of rationality preserves the core notion of purposefulness, and in that sense it should seem familiar. But unlike other notions of rationality – many of which were invented for modeling purposes but have since taken on a life of their own – inclusive rationality… allows a wide range of possibilities in terms of how real people select their goals, form and revise their beliefs, structure their decisions, and conceptualize the world.

They continue:

[R]eal people may do all of the following and still qualify as inclusively rational:

  • Experience internal conflict that has not yet been (and may never be) resolved;
  • Have preferences that change over time;
  • Have preferences that are indeterminate or incomplete – i.e., that do not specify attitudes over all possible decisions at all possible times and states of the world;! Have preferences that are in the process of being created or discovered;
  • Have preferences that depend on context, including both the options available and the way in which decisions are framed;
  • Hold beliefs that serve purposes other than truth-tracking, such as providing motivation or intrinsic satisfaction;
  • Make inferences based not on the strict rules of classical logic, but on contextual and linguistic cues they have learned from human interaction;
  • Indulge “biased” modes of decision-making when the costs are low and rein them in when the costs are high;
  • Economize on scarce mental resources by refusing to impose perfect consistency on their preferences and beliefs;
  • Structure their environments, possibly in ways that constrain their own choices;
  • Adopt personal rules and resolutions that create internal incentive systems;
  • Enlist the help of friends, family, and other groups to assist in attaining goals;
  • Rely on institutions, social customs, and market structures to assist in attaining goals;
  • Employ heuristics that minimize cognitive effort and/or informational input.

R&W are well-aware of the methodological objection:

Because our notion of inclusive rationality is very broad, we might be accused of offering a theory that cannot be falsified. We should therefore clarify that we do believe positive claims should, in principle, be falsifiable. But some claims are more easily tested than others, and there is no guarantee that the most easily tested claims are also normatively relevant.

In Chapter 3, R&W flesh out this critique in great detail.  To take one example: In real life, people often seem to change their preferences.  Yet behavioral economists either refuse to consider this possibility, or treat changing preferences as ipso facto “irrational.”

The second interpretation is that the agent’s preferences have simply changed over time. In other words, if someone chose A over B on Monday but B over A on Tuesday, that could be the result of consistent (complete and transitive) preferences on Monday, and a different set of consistent (complete and transitive) preferences on Tuesday.

In general, economists resist preference change as an explanatory strategy because it feels ad hoc. Virtually any change in behavior can be rationalized as resulting from changing preferences, but economists usually wish to show the reasons why behavior may change even if tastes do not – for example, because of changing relative prices. To resort to saying preferences have changed seems like cheating. These are valid concerns from a positive perspective. But from a normative perspective, they hold little weight. There is no reason a person’s preferences should remain fixed. Rational people may change their minds – and we have not encountered anyone arguing otherwise.

Critical Comments

1. Escaping Paternalism focuses almost entirely on the “new paternalism.”  Yet the more I read the book, the more I concluded that “new paternalism” is largely an attempt to repackage “old paternalism” for an elite, secular audience.  Even today, no more than 5% of people who want to forcibly reduce alcohol consumption think they’re helping heavy drinkers “achieve their true preferences.”  Instead, they seek to give drinkers “what they need, instead of what they want.”  The main audience for the new paternalism is: meddlers who also have an ideological ax to grind against the want/need distinction.

2. On present bias, behavioral economists don’t go far enough.  It’s not enough to have exponential discounting rather than hyperbolic discounting.  Instead, I say the rational discount rate for utility is no time discounting at all.  Contrary to what you may have heard, this would not lead you to starve to death.  And you should still discount for uncertainty, finite life, and so on.  But, pace Hume, ’tis contrary to reason to prefer a present pleasure to a future pleasure solely because the present is now and the future is later.

3. R&W repeatedly mention behavioral economists’ support for mandatory “cooling-off periods.”  I wonder how many behavioral economists have ever examined how often people even take advantage of this opportunity to cancel a contract.  The last time I refinanced, we had to sign paperwork apprising us of our right to renege.  When I asked the banker how often people exercised this right, he furrowed his brow and laughed, “Never.”  I believed him.

4. R&W repeatedly discuss behavioral economists’ support for “employee-friendly” labor contract defaults.  Given all the evidence that unemployment breeds misery, a wise paternalist would strive to “nudge” labor’s total compensation down, not up!

5. R&W insist that “inclusive rationality” is falsifiable, but I don’t think they name a single concrete counter-example.  This is arguably intentional, for they often say things like:

We do not, however, claim that everyone is always fully rational. We are happy to concede that they are not. But it is one thing to say people make mistakes; it is another to clearly and definitively identify which actions are, in fact, mistakes.

Strikingly the word “opioid” does not appear in the text.  (“Heroin” does, but sans empirics).  I sincerely wish they had provided an “inclusive rationality” take on the putative opioid crisis.

6. One of my favorite passages:

We don’t expect that every reader will find all of our challenges to paternalism equally compelling. Some may find our conceptual objections persuasive, others may see the knowledge problem as definitive, etc. But our hope is that, taking the gauntlet of challenges as a whole, readers will recognize just how tenuous the entire new-paternalist enterprise is. Behavioral paternalism’s proponents often present it as just common sense – a set of smart, simple, straightforward corrections that will make us all better off. All we’re trying to do is correct some mistakes; who could be opposed to that? But the reality is far more difficult, complicated, and even dangerous.

As you’ll see, I find the conceptual objections only moderately persuasive, but the rest of their critique is strong indeed.

7. R&W go through a lengthy list of defensible “mistakes” people make in experiments.  Let’s consider them one-by-one:

When real people take tests designed to test neoclassical rationality norms, they do not leave their inclusive and contextually shaped rationality at the testing room door. Instead, they may behave as they do in the real world. For example:

  • They may assume, in accordance with ordinary conversational norms, that experimenters provide only information that is relevant to solving the problem – i.e., no irrelevant or “tricky” information. They do not immediately assume the experimenters are trying to fool them.

Reasonable.

  • They may resist the distinction between the validity of a syllogistic inference (e.g., “People with red hair are Martians, John has red hair, therefore John is a Martian”) and the truth of a conclusion itself (John is not a Martian). Normally, in everyday life, it is the truth that is more important.

This isn’t entirely wrong, but the ability to recognize such logical errors really is a crucial real-life skill.  Anytime you propose an explanation for anything, logical reasoning opens your theory up to testing.  Even for an explanation as simple as, “If I go to McDonald’s before 7 AM, the line will be less than 5 minutes long.”

  • They may not assume that prior probabilities about something – such as the likelihood that someone has a disease – must be equal to the “base rates” from the population provided to them. Instead, their priors may be affected by their evaluation of the significance of the base rates to a particular problem in front of them – say, whether a specific person who chose to visit the doctor and chose to take a test has the disease. Treating priors in this way is fully consistent with the subjectivist Bayesian view that prior probabilities are subjective – a fact frequently ignored in the rush to deem subjects “irrational.”

Reasonable.

  • They may not agree with model-builders on the informational equivalence of different descriptions of a situation. Instead, they may infer implicit information or advice from how a problem is presented. For example, they may perceive an important difference between a stated probability of success equal to 0.7 and a stated probability of failure equal to 0.3. Perhaps the former conveys greater optimism, despite the formal mathematical equivalence of the two statements. Conversational norms and expectations do not always align with logic and probability theory. The former can be adaptive in the real world while the latter is adaptive on experimental tests. Which is more important?

Seems pretty crazy to me.  Sure, if you’re marketing a product, you’re better off saying “70% success” rather than “30% failure.”  Yes, the former phrasing suggests “greater optimism.”  The only reason such rhetoric works, however, is that some potential customers are confused.

  • They may attach satisfaction or utility to things other than what the analysts expect. For instance, they may value an object more because it is theirs already. Or they may care about feelings of gain and loss experienced during the experiment, not just how much money they have when they leave the laboratory. Or they may gain satisfaction purely from having a particular belief, irrespective of its truth (“My wife is beautiful and my children are gifted”).

Somewhat reasonable, though it’s usually tempting to call such behaviors “childish.”  Picture a kid who wants his teddy bear, not an identical teddy bear that’s a hundred miles closer.

8. While we’re on the subject, R&W never discuss children.  Are they, too, “inclusively rational”?  If so, should parents stop treating them paternalistically?  Or what?

9. Suppose someone has inconsistent preferences.  How are we supposed to identify their “real” preference?  While this might seem like nitpicking, R&W show that behavioral economists have been cavalier  at best.

Some defenses have been offered for favoring some preferences over others in cases of conflict, but we find these defenses weak:

Verbal Statements and Survey Responses: When asked, people may say they would rather behave differently or have different preferences. For instance, smokers may say they would rather not smoke, and overweight people may say they would like to eat less. It is indeed possible that these statements reveal “true” preferences. However, the incentives for speech differ from the incentives for other kinds of action. Behavioral research has cast doubt on many economic principles previously taken as given, but the principle that talk is cheap remains intact. Speakers who say one thing while doing another may simply be expressing what they regard as socially approved attitudes – a phenomenon known as social desirability bias (King and Bruner 2000; Grimm 2010). Or their statements may simply reflect “experienced opportunity cost,” i.e., the dissatisfaction that always results from options the agent has forgone.

Note: If we take Social Desirability Bias seriously – as I think we should – we can readily identify the “real” preference.  Namely: Contrary to the new paternalists, the real preference is what people really do!

Regret: A person may feel, and express, feelings of regret about the choices they have made: “I wish I had not done that.” Although regrets are real, they do not necessarily reflect all costs and benefits associated with an action.  specially for intertemporal choices, such as getting inebriated last night and having a hangover today, the regret is typically experienced while the cost is being experienced in the present and the benefit is already in the past. That does not mean the costs outweighed the benefits at the moment of choice – only that the remaining costs outweigh the remaining benefits. In addition, it’s worth noting that regret can also be felt about the kinds of choices that behavioral paternalists favor. When approaching death, people often express regret at not having lived a more spontaneous and present-oriented life (Ware 2012).

They’ve got them there!

If regret may be experienced regardless of the action taken, then it offers little guidance to the paternalist about which preferences are “true.” As with verbal statements, regret can simply reflect the experience of opportunity cost…

Self-Constraint and Commitment Devices: People will often use various devices and strategies to try to keep their vices under control: planning automatic deductions for savings, avoiding locations where they will be tempted to smoke or drink, etc. These activities do provide further evidence of conflicting preferences, and we will discuss them more in future chapters. They do not, however, show which preferences are superior. Commitment devices reveal one set of preferences at work – but other choices show other preferences at work. Furthermore, the outside observer has no means of knowing whether the right amount of self-constraint has been performed.

Planned versus Unplanned Choices: Behavioral paternalists often favor the preferences of a “planning self” over the spontaneous or “acting self.” The idea is that the planning self is more likely to take all costs and benefits into account and render a considered decision. But the planning self does not necessarily represent a disinterested party; rather, the planning self may represent only the longer-term and more self-denying parts of one’s personality (Cowen 1991). This becomes most apparent in the case of extreme behaviors such as anorexia, where the planning self dominates an acting self that might wish to indulge more often.

10. When I began Escaping Paternalism, I skipped over chapters 1-5, fearing they’d be long-winded, half-baked Austrian philosophy.  I was so impressed with chapters 6-10, however, they I started reading the book from the beginning as soon as I finished chapter 10.  I swiftly concluded that unlike so many other Austrian-inspired works, Escaping Paternalism was not only conceptually thoughtful, but eager to engage with an array of empirical literatures.  If you’re impatient with economists waxing philosophical, I urge you to make an exception here.