Libertarian Paternalism
Jonah Goldberg has a fascinating series of posts reacting to Thaler and Sunstein’s article in the LA Times on “choice architectures”. I think his reactions get into two separate questions: (1) Does libertarianism require a set of fixed moral principles as a foundation?; and (2) Does the idea of authorities manipulating choices that face us for the purpose of making us happier really have anything to do with libertarianism, properly understood?
As per prior posts, I think that the unequivocal answer to the first question is ‘yes’. The second is a little trickier.
“Choice architectures” and “libertarian paternalism” are just fancy ways of saying that authorities make it more expensive (in the broad sense of money, time, convenience, etc.) to make some choices rather than others. Consider a widely-discussed example that the authors provide: requiring that employees opt-out of default enrollment in 401K programs. Sure, I have the choice to opt out of the 401K, but the operative issue is how hard this is to accomplish. On one extreme, as part of my first day of work I might have to go to a company intranet website to enter various information that includes my decision about 401K, and the yes/no drop-down list might default to “yes”. In this case, it would take two mouse clicks to opt-out. A more restrictive choice architecture might automatically enroll me, and then require me to make a series of appointments with uncooperative HR staff, fill out numerous paper forms and dig up all kinds of information that involves going through files buried somewhere in my garage in order to opt out. In the extreme, even if the law required me to enroll, I would still have the choice not to do so – I would simply have to be unemployed or work on the black market and risk prosecution. This is closely analogous to the concept that the power to tax is the power to destroy.I think that at a practical level, the real question is one of how hard it is to exercise “undesirable” choices, at two levels of abstraction. First, within the specific choice architecture of a 401K plan sign-up or menu choices at a school lunch counter, how much are the dice loaded in favor of some choices relative to others, and how much additional total choice burden have the authorities added?. While there is no absolutely neutral choice architecture, defaulting to one choice on a two-item drop-down list is surely far more neutral than creating huge administrative hassles for getting out of the plan. Second, to what degree does the government mandate that all relevant choice architectures be identical? For example, does the federal government require by law that all companies have a specific opt-out method for 401K plans, or does it provide tax breaks and other economic incentives for those that do this, or does is just use the bully pulpit to encourage this?
I’ll do all the supplemental readings later (so sorry if this is addressed), but I do want to point out that no matter what an employer does in this situation it is making a normative choice. The employer is doing exactly what you are worried about the government doing. If the default is to have a zero contribution, then the employer is making a statement of what is desirable/undesirable of what the employee’s optimal saving/consumption rate should be, in the same way Thaler et al are suggesting the government do. The employer can’t get around this – if he roles a die, he is saying the employee’s contribution should be 3.5. Etc.
I don’t see how, since Thaler et al are insisting that their default rates are, on average, better than the employers, makes it any more/less oppressive to consumer choice than whatever default rate my HR rep hands me. (Remember employers match employees contributions in many circumstances; they have a vested interest in keeping contributions low.) More specifically, since consumer choice isn’t being limited (the horrors of dealing with HR aside), I don’t see what isn’t libertarian about these policies.
— Mike · Apr 3, 09:35 PM · #
Mike:
Sure, I agee there are no neutral choice sets. My only point was that, as a practical matter, some make it fairly easy to chose among alternatives and some harder, and further that a broader system that let’s me “choose my choice sets” (e.g., let’s different employers have different choice sets and defaults for 401k plans) gives people more choices, allows more experimentation and so on.
BTW, love the AD&D post.
— Jim Manzi · Apr 3, 09:52 PM · #
“Choice architectures” and “libertarian paternalism” are just fancy ways of saying that authorities make it more expensive (in the broad sense of money, time, convenience, etc.) to make some choices rather than others.
This makes sense in the HR-bureaucracy 401K example, in which you know what you want but it’s a pain to get it, but I’m not sure it well it fits a situation where, for example, there are there are ten options and we’re trying to decide how to arrange them on the page or on the screen.
They’re all equally expensive to actually select (touch the screen! check the box!), but you might actually notice them with differing probabilities. To say that those lower on the list and less likely to be noticed (or whichever ones the brain finds easier to ignore) are “more expensive” seems like we’re taking the agency metaphor too far.
We can sort of save ourselves by thinking of choosing our choice sets, choosing how much thought we put into each choice set, how much time we spend considering all options and how much satisficing we do. But if this meta-choice is another choice, it means there’s another choice architecture that influences our selections, and, well, I guess it quickly becomes turtles all the way down.
— Consumatopia · Apr 4, 12:13 AM · #
Consumatopia:
Great comment, thanks.
I actually cut out a paragraph (yes, beleive it or not, I edit posts here) that went into another of the hypotheticals that the authors went into: a schools adminstrator who experiments with different orderings of the food items in the school cafeteria to find the sequence that leads students to have a greater probability of selecting healthy foods. It turns that large supermarket chains have done well-designed random-assignment trials that find the sequencing of beef, prok and chicken in the meat aisle that maximizes store profits. It is actually (it is believed) because the brain processing “costs” of holding open intial options for consideration lead people to satisfice even in the 3 – 10 second walk along the meat aisle. So it’s not really a metaphor, though obviously it’s not exactly condeming people to death camps either.
I think, in theory, it is turtles all the way down in exactly the sense you mean it. In practice, there is a finite number of such nested choice sets (and in fact, it’s some kind of nesting structure that’s actually more complex than that). It’s Hayek in action.
— Jim Manzi · Apr 4, 12:53 AM · #
I’m still not completely comfortable with the relationship between the expense we pay after deciding on an option (calling HR) and the expense we pay to in order to make the decision in the first place (reading all the way to the last option in the menu). And, to be clear, to the extent that they are different, I find the latter more coercive than the former per unit of expense. At least with the former you know what you’re trying to get as you make the payment, whereas with the latter you have to pay in order to know what you’re paying for.
To put it in algorithmic complexity terms, the Incompressibility theorem puts a wrinkle in our attempt to use meta-choices to deal with the fact that we have more options in life than our three pounds of meat can handle. If we can only handle X choices, we can only decide from among 2^X possibilities, and no amount of meta-choosing ultimately lets us get more than 2^X from X.
OTOH, I guess we could have a nearly infinite number of turtles, if we personify our DNA (like The Selfish Gene), so that even when the wiring of our brains biases us in one direction (say, to satisfice after 3 seconds instead of 4), that represents a choice on the part of our biology. Neural cognition just can’t be expected to go very many levels deep (at least mine doesn’t), but natural selection gets to add adaptations (a sort of choice, I guess) with every reproduction.
I guess that only gets us so far, though, as something analogous to the Invariance Theorem should let the grocery store design choice architectures that are comparatively more “expensive” for nearly all choice-making Turing machines. And, heck, reengineering our brains to accommodate larger sets of choices would probably make us hungrier, as we need more nutrients to maintain our larger brains. This plays right into the grocer’s hands. Clearly, this game cannot be won.
— Consumatopia · Apr 4, 03:11 AM · #
Jim, near the end of your comment there’s a clear flaw that I’ll expl — HUH?
PORK? ME HUNGRY! NO READ MORE. EAT NOW!
— Matt Frost · Apr 4, 11:16 AM · #
IIRC, the issue with 401(k) plans is that under the relevant law, there is a higher standard of liability for people who give investment advice and a lower standard for persons/entities who are involved in administering 401(k) plans without offering investment advice. Under that analysis, if the employer automatically enrolls employees, that’s sort of like providing investment advice and therefore risks exposing the employer to the higher standard of liability. To encourage default enrollment, the law has been clarified to assure that providing default enrollment does not mean an employer is stuck with the higher standard of liability. It’s not an issue of micromanaging the HR department’s web site.
— alkali · Apr 4, 03:01 PM · #
>Liberty Dad – a Libertarian Paternalist in Slovakia < There’s a longer Brookings paper on this which I read years ago and convinced me that this label fits my beliefs more than Conservative Libertarian.
Professor Bainbridge (years ago?) objects to Libertarian Paternalism because of its potential use in justifying additional regulations “with some choice”. This objection is weak for 2 reasons.
First, the main target of any Lib-Pat action would be to increase choices in a currently regulated/ non-choice gov’t area.
Second, adding choices to any power-freak’s control plans are certain to weaken the desire to implement the control; there’s less point to it if folks can choose otherwise.
In general, I’d guess I’m like most Lib-Pat guys who will accept living in a very non-Lib, democratic mildly fascist society, and push to increase the choices.
As with school vouchers, whereas the full Libertarian position is no compulsory attendence nor any gov’t money for schools, I think vouchers allow society 80-90% of the benefits in choice that would be available, and are much more likely to be passed. Though passage, plus 50% or more success, means it becomes less likely to fully scrap the (now-working better) gov’t.
The fascist desire to control others is not fought directly with such choices, but its blows are deflected towards causing much less damage at much less effort in defense. To parry a swordstroke by allowing the stroke but making it miss, rather than fully stopping it in mid air (so much more photogenic / manly / macho – flash).
— Tom Grey - Liberty Dad · Apr 11, 04:27 PM · #