Libertarian Paternalism and Choice

Is it worth noting that one cannot simply “opt out” of Thaler and Sunstein’s endorsed retirement programs (see endnote 15, chapter 6 of Nudge)? If employers wish not to participate in auto-enrollment, then they are exposed to costly legal liability and must expend costly effort filling out forms to attenuate this liability. If employees wish not to participate in retirement plans like a 401k, they are exposed to a tax penalty. So, even as the authors might prefer to see things, this intervention does indeed reduce the decision space of market actors.

It might be responded that this is merely another failure of the authors to live up to their stated principles – not an indictment of those principles themselves. But I have trouble imagining a situation in which “libertarian paternalism” is enacted by government yet nobody’s decision space is narrowed.

Perhaps the claim is that, if only the employer’s decision space is narrowed, then an intervention is “libertarian paternalism”, because only employees are actual market participants and only their choices matter for utilitarian considerations. Let us fashion a hypothetical law that attempts to respect this conception in order to expose the underlying fallacy. We might coerce businesses into auto-enrolling their employees in retirement accounts that had no tax-benefits over normal bank accounts. This would, supposedly, have no negative effect on employees because they would remain free to opt out of auto-enrollment.

The supposition is inadequate. First, we have the “single-click” cost. Thaler and Sunstein frequently profess the desire to guide consumers to certain outcomes while granting the right to opt-out by methods as simple as a single click of a computer’s mouse. A trivial cost is a cost nonetheless – it does not leave the decision space unchanged. I’m tempted to set this point aside, but the authors seem eager to impose quite a few “single” clicks, involving just about every decision people can make. As the clicks accumulate, they will fatigue individuals who consider opting out – anything but reflexive default acceptance becomes costly. Or if opting in is made just as costly as opting out, the clicks will introduce generally an unnecessary friction to human action.

More importantly, impositions on the choices of employers implicitly create costs for their employees. Thaler and Sunstein argue that people have strong status quo biases. In a market system, there could be good reasons for this bias. If companies tended to offer a certain compensation default, employees could fairly assume that other people in fact preferred to receive compensation in those terms. If this compensation package were not the market equilibrium, workers with status quo biases would experience greater satisfaction at competing firms offering a more preferred default.

If the government mandates a certain default, businesses lose the ability to communicate market default preferences to their employees. Not only do employees lose access to this information, they may be deceived into thinking that the new default is in fact the market default, rather than governmental compensation-fixing. Nor must market information be destroyed to make space for “expert” information. If people want to know how Thaler and Sunstein want them to receive their compensation, they can just read Nudge. But market information can only be communicated through the market.

So to say that these employees can reach the same decision point is true in a sense, but misses the bigger picture. One might as well say that a dart player could make all of the same throws while wearing a blindfold, or that monkeys on typewriters could technically reproduce Shakespeare. Deciding on a particular decision point is in itself a costly procedure, and libertarian paternalism deprives people of important information. The imposition of a rigid, un-chosen, government default interferes with the optimal operation of human choice.

I have difficulty unpacking arguments to the effect that we should think carefully about existing choice architecture, because choice architecture cannot be avoided. These seem to be two separate claims without any important connection.

It is the case that prices must exist in a market. It is not the case that “we” or the government must think about what those prices should be. Nor is it the case that the government is creating choice architecture or fixing prices when it abstains from creating choice architecture or fixing prices. There need not be any government decision in arranging cafeteria food locations or employee retirement, any more than cafeteria managers need to plan their customers’ retirement plans or retirement planners need to design cafeterias for their customers! Some decisions need not be made by people with no connection to those decisions.

As I argued at great length below, paternalists have no grounds for picking out “better” outcomes. They do not need to make most choices for consumers. And if they need not make choices that cannot on objective grounds be said to improve outcomes, then they ought to abstain from interfering.


  • Jacob

    That’s a very flattering picture of businesses. I find it more plausible to consider businesses as profit-maximizing entities. They don’t structure default compensation schemes to maximize utility for their employees, they do it to maximize their own profits. The degree to which these overlap can vary based on how attentive the business thinks potential employees will be (and hence how they will be able to attract employees and maximize profits).

    As an employee, would you rather have the default compensation scheme set by someone who was attempting to maximize your utility, or the employer’s utility? Even if it’s an unsuccessful attempt, you’re probably better off with the first. If you’re one of the few who actually wants to take an active role in structuring the compensation scheme, you will incur the same cost switching from the mandated default as you would from the employer-determined default.

    Going back to earlier in the post, you argue that default enrollment in a retirement savings account imposes a cost. You’re comparing it against the wrong benchmark. Without a default retirement account, a cost is imposed on anyone who wants to save for retirement (single-click if there is a plan they can join, and many more clicks if there isn’t one). There are costs associated with any choice architecture–accounting for the increased costs by switching to a different architecture while neglecting the decreased costs skews the calculation.

  • wallaceforman

    My description of employers is not meant to be flattering or assume any goodwill on the part of business owners. I, too, consider businesses to be profit-maximizing entities. I think your last paragraph above gives a pretty good account of the incentives that picky marginal employees have to gravitate to employers offering the proper default. Given this tendency, businesses must compete in their defaults, just as they compete in prices, in order to maintain employment.

    If the costs of setting a default (or put another way, the amount that wages must be reduced to pay for determining a default) are higher than the net benefits, the efficient (utility-maximizing) outcome will be to not set a default. Employees will demonstrate the value they put on choice architecture to the extent that they choose default-setting employers over default-less employers with higher wages.

    I suppose that the next step in the argument will be to say that choice architecture is a “public good” and that markets under-supply these goods. Inevitably, theoretically valid objections to market outcomes devolve to a very few categories: monopoly, externality, and public good arguments. It is a pity that Thaler and Sunstein chose not to phrase any of their arguments in these terms. Instead they rely on paternalistic arguments about experts knowing objective preferences better than individuals – arguments I am comfortable dismissing out of hand.

    In any event, I refer to my arguments about the subjectivity of ends to dismiss the public good argument. It is theoretically true that paternalists might supply a public good in choice architecture, but because of subjective ends, they have no earthly way of determining what the preferred choice architecture is.

  • Jacob

    I think our fundamental disagreement is now clear: I’m not making a theoretical argument (neither are Thaler and Sunstein, I think). I tried to make it theoretical in the last comment, but I really need to get back on my own turf. Empiricism trumps theory, and the empirical truth is that changing choice architecture can improve outcomes (for example, keeping it private sector, it can improve doctor decision making, patient compliance with medical advice, people lying on insurance forms, pretty much every sphere of life).

    There is a class of choices for which this is incontrovertibly true. Non-compliant patients (in normal cases) are not maximizing utility when they don’t take their medicine, and nothing you say will convince me otherwise. I hope you don’t disagree with me that thinking about choice architecture in these cases (whether you are a private actor or a government) is worthwhile. Of course, this is an argument you are “comfortable dismissing out of hand,” so maybe I shouldn’t be so sure.

    Then there is a class of choices where the effect of choice architecture is unclear, and setting a different default path might decrease utility for some individuals (those who aren’t paying attention and whose situation dictates that the new default is worse, or people for whom switching from the new default imposes a new cost).

    The unclear cases range from ones where we can be pretty sure we know what we’re doing to ones where we have absolutely no idea. Maybe in some of these cases, because ends are subjective and we don’t really know what an individual wants, we will theoretically impose costs on people who have to switch from the default. But as long as we aren’t preventing them from switching from the default, I think the gains far exceed those costs. But I think the empirical finding is that experts can improve subsequent evaluation of outcomes by setting choice architecture in many of these cases. If you disagree about where the line is drawn, I’m fine with that. If you’re opposed to any attempt on theoretical grounds, I think we are at an impasse.

  • wallaceforman

    The debate is not just theoretical or empirical, I think, but definitional. What are we willing to define as the ends that people aim at? I am uncomfortable describing something as an end if it is not chosen by the person whose end it is supposed to be. Thaler and Sunstein are quite comfortable with this – either because they “know” it to be the case, or there is survey data supposedly supporting this conclusion.

    Briefly – utility maximization. I ought to have said that the market (absent monopoly/public good/externality considerations) would yield an economically or pareto efficient outcome – where no one could be made better off without making someone worse off. Individual maximum utility is a more or less nonsense consideration, as one can almost always make someone better off just by transferring someone else’s stuff to them – which is not, I think, what we are talking about.

    Pill compliance ought be brought to an economically efficient outcome by the market just like anything else. By all means, private sector choice architecture will and should be a part of this process. Whether pill noncompliance derives from forgetfulness, ignorance, or a simple death wish, overcoming these features have costs that people either do or don’t feel are worth paying. It may be true that beyond an economically efficient point, some subset of people can be made even better off by mandating people engage in a more expensive default (i.e. different defaults will be better for some people and worse for others). But choosing between these points requires us to engage in progressivist distribution shopping. One might as well argue about what level of welfare is best – the choices are simply arbitrary value judgments.

    I do think that some engagement with public goods is beneficial for society. Public roads and police stations, for example, are obviously very useful (though I would hope for a trend toward privatization of the former). But these are very high cost items. Choice architecture has the benefit of appearing to be a pretty (relatively) low cost invention, so there is no reason to think that there is a critical need for the government to figure out what choice architecture should be set by private enterprise. The costs of choosing the wrong architecture – which I consider quite likely without any market feedback in these decisions – would very easily outweigh the costs of waiting for the private sector to fill in the perceived gaps. In the meantime I betake upon myself to supply a public good in skepticism with new government interventions. Perhaps I can get a subsidy for that?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.