For a long time, the nation has been split between two types: old-style Democrats, favoring mandates and bans, and new(ish)-style Republicans, insisting that markets and free choice should be respected. Richard Thaler and I think that there is a way to avoid mandates and bans, and to respect free choice, while also helping people to make better decisions.
In short, we hope that libertarian paternalism might provide a real third way, one that recognizes the best in Hayek and Friedman while also noticing the work of Simon, Kahneman, and Tversky (and Thaler), which shows that human beings often choose poorly. Thus, for example, libertarian paternalism offers fresh ways of thinking about the mortgage crisis, credit card reform, savings for retirement, prescription drugs, health care, environmental law, and even marriage. In all these contexts, a few nudges could help a lot.
I'm most grateful for the many excellent comments and objections here; I've read them all with care. By way of conclusion, let me try to respond to just a few of the most prominent, with the hope of a continuing discussion.
1. Objection (made on the Becker-Posner blog as well): Libertarian paternalism is an oxymoron.
Answer: It is not. The approach is libertarian in the sense that it preserves freedom of choice. It is paternalistic in the sense that it tries to produce good outcomes for choosers. See, e.g., Save More Tomorrow.
2. Objection: Libertarian paternalism is paternalism is lamb's clothing.
Answer: Because libertarian paternalism preserves freedom of choice, it is different from standard paternalistic approaches. We favor "one-click paternalism," in the form of easy, simple opt-outs.
3. Objection: Libertarian paternalism is a tool of the left; it is the left's newest thing.
Answer: The tool (or tools) can be used by many different types. Many of our own proposals will be more popular with the right, eg school choice and medical malpractice waivers.
4. Objection: Who decides what choices are best?
Answer: The choice architect (private or public). Because choice architecture is inevitable, the "who decides" question is not a good one, at least for our purposes. Cafeterias have to be designed. Default rules have to be developed. Starting points are inevitable.
5. Objection: What about the public choice problem? Aren't public officials subject to the same heuristics and biases as everyone else, and pushed around by interest groups too?
Answer: Agreed, entirely. That's one reason this form of paternalism respects freedom. Wherever a program is in place, or inevitable, libertarian paternalism provides an appealing approach. Recall too that choice architecture, even by government, cannot easily be avoided, in the form of property law, contract law, and tort law. Existing law governing environmental protection, credit markets, labor law, and the family could be improved with a few nudges. Sure, government should privatize in some domains, and libertarian paternalists sometimes call for privatization, to increase freedom.
6. Objection: Libertarian paternalism starts us on a very slippery slope.
Answer: Not if the libertarian condition (easy opt-out) is respected. That should reduce the slipperiness by a lot.
7. Objection: In some domains, we need bans and coercion.
Answer: Sure, at least if third-party effects are serious. And we don't want to repeal the criminal law involving (say) murder and assault. But in most domains, freedom of choice should be the rule (in part for reasons that Hayek developed so powerfully).
8. Objection: Libertarian paternalism is an ugly and confusing term. Ugh!
Answer: Maybe. Probably. Originally we wanted to call the book Libertarian Paternalism, and no one wanted to publish that book. So we called it Nudge.
Thaler and I regard this as a continuing project, in some ways in its early stages, and we're aware of legitimate questions and of room for much more work on these topics. Many thanks to Eugene Volokh for hosting me and to all of you for your terrific comments and emails. We'll keep pondering these issues.
Related Posts (on one page):