Julian Sanchez header image 2

photos by Lara Shipley

Libertarian Paternalism: The Revenge

May 31st, 2007 · 6 Comments

Cass Sunstein and Richard Thaler’s concept of “libertarian paternalism” is generating renewed discussion on the heels of a Wall Street Journal debate between Thaler and my quondam mentor in the occult mysteries of libertarianism, NYU economist Mario Rizzo.

For all that I’m a fan of linguistic precision, I think Will Wilkinson’s semantic critique of the phrase “libertarian paternalism” splits some hairs that could’ve gone unsliced. Even on a strict reading of the component terms, it’s not obvious that paternalism necessarily entails coercion, and there aren’t some kinds of paternalist default-setting that even the strictest libertarian would permit. The law must decide, for instance, when a valid contract has been entered into, and asking (for instance) whether someone has really given consent will mean ruling out certain kinds of apparent agreement—assent given while drunk, say—that we think are unlikely to reflect a person’s true or reflective interests. And this does, albeit in a pretty minor way, restrict people’s freedom insofar as it rules out entry into any contract under any circumstances. Leaving that aside, though, the terms in compound phrases do frequently modify each other from their strict independent meanings. Someone who objects that boxer-briefs aren’t quite boxers and aren’t quite briefs has missed the point of adjoining the terms. I think what’s intended in this case is clear enough to most readers: “Libertarian paternalism” is “paternalistic” in the sense that it seeks to shape people’s behavior for their own benefit—if not coercively, then often by exploiting cognitive bias in a way that differs from ordinary rational persuasion—and “libertarian” in the sense that it seeks not to seriously burden individual choice when it differs from the paternalist’s assessment. This, too, is scarcely a wholly novel use: Replacing public schools with a voucher system or preserving the FDA in a strictly advisory capacity might not be the pure libertarian position, but we all know what is meant by the claim that this would be more libertarian than the status quo.

Glen Whitman objects to certain specific policies Sunstein and Thaler class as instances of “libertarian paternalism,” which he may be right about, but I don’t know that it goes to the heart of the concept. Indeed, to reject certain specific cases is precisely to acknowledge that the concept has intelligible content. If the point is that Thaler and Sunstein are shoehorning some cases of old-fashioned paternalism under their new rubric, fair enough, but then it’s worth debating whether there are some other sorts of policies that would pass muster. One I’ve found intriguing, not mentioned in this debate, is a policy alluded to in this Economist article [subsc.]: People who know they have gambling problems can request to be placed on a list of people barred from casinos, against which gamblers must be checked at the chip window. If I recall correctly, they can fairly easily remove themselves, though not, as it were, on the spot with a phone call, creating just enough of a hurdle to frustrate spontaneous attacks of akrasia. This is, of course, coercive toward casinos, who are required to check players against the list, but I think it works as an instance of “libertarian paternalism” in the sense that it seeks only to help individuals adhere to their own considered judgments about their best interests.

Ilya Somin at Volokh objects if that if people are plagued by cognitive biases in their decisionmaking at the individual level, they will a fortiori be subject to the same biases in their capacity as voters. And that’s certainly true, but the kinds of narrow policies we’re talking about here are almost certainly going to be designed “by unelected experts” rather than referenda, a possibility with which Ilya also has problems. Here, too, I think the problems may be less severe than Ilya makes it sound. Somin contrasts market consultation with experts, whom one may ignore, with state experts. But if we’re talking about cases of genuine libertarian paternalism, the programs will be such that it is, after all, quite easy to ignore the expert judgment. If it is not, then we’re talking about old-fashioned paternalism of the kind we all agree is a bad idea. True, the program designers still need to be at least indirectly chosen by the political process—appointed by legislators who may be held responsible if the programs prove harmful or burdensome—but this is one case where I’m not sure I find the “bias, bias, everywhere” concern as pressing. Yes, short-term bias may be ubiquitous, but it’s also context activated. It’s likely to be a more significant factor, for instance, in the decision to step into the casino tonight than it is in the selection of a person to design a voluntary no-gambling list, or in that person’s deliberate construction of a program to combat short-term bias.

The best objection, I think, comes from


6 responses so far ↓

  • 1 jadagul // May 31, 2007 at 4:49 pm

    There are certain types of libertarian paternalism that don’t bother me at all. The main example that comes to mind is default contract rules: for instance, I have no problem assuming that any contract that doesn’t explicitly include “will perform sexual favors for coworkers/boss” as a job responsibility includes a tacit provision saying that no such favors are required (as a proponent of legalized prostitution I obviously think contracts that explicitly mandate the performance of sexual acts should be valid). Similarly, I wouldn’t really be upset by something like Glen’s 1-4, such as the presumption of a certain amount of vacation time, as long as overriding that presumption really is as simple as including a “vacation time” clause in the contract.

  • 2 Glen Whitman // May 31, 2007 at 5:24 pm

    I think your metaphors are in conflict, Julian. If you think the slippery slope is the most serious objection to libertarian paternalism, you can’t follow that by saying you’d like a little bit of soft paternalism as a vaccine against hard paternalism. These are contradictory mechanisms. The slippery slope mechanism says that innocuous policy A increases the likelihood of intrusive policy B. The vaccine mechanism says that innocuous policy A decreases the likelihood of intrusive policy B.

    I also think you should take the semantic arguments more seriously, because it is precisely through semantic manipulation that people like Sunstein & Thaler are greasing the slope. For more on this, see my follow-up post:


  • 3 Dave W. // May 31, 2007 at 6:44 pm

    Nice post!

  • 4 Julian Sanchez // May 31, 2007 at 8:10 pm

    Glen: Well, not for cases where we’re already at the bottom of the slope. Something like the no-gambling list might make the legalization of gambling in more areas more palatable; you can imagine some parallel system for recreational drugs.

  • 5 Arkady // Jun 1, 2007 at 12:48 am

    I think you’re both right: innocuous policy A could be a good option if it moves you away from a pre-existing intrusive policy B, or if it diffuses political support for policy B (although given the history of slippery slopes and reductio creep, I’m not sure the second scenario really occurs in the long run).

    But if the examples in Glen’s post are at all representative, then voluntary casino blacklists and different retirement plan defaults aren’t really of the same degree (plus, you can imagine how either of them might arise — and have before — without state coercion). If Sunstein and Thaler “would be willing to impose real [I assume here they mean large] costs on workers and consumers who seek to do what, in the paternalist’s view, would not be in their best interests,” then I don’t understand how that is any different from the same old normal paternalism.

  • 6 Dave W. // Jun 1, 2007 at 9:27 am

    My favorite form of libertarian paternalism is government mandated nutritional labeling. I see this form of paternalism as high preferable to food bans.