Thaler on Libertarian Paternalism

While there has been a lot of dissing about the Thaler/Sunstein libertarian paternalism, I'm listening to Thaler talk at Google right now, and it seems pretty convincing. He's here to talk about their book Nudge: Improving Decisions About Health, Wealth, and Happiness.

His basic claim is that we have to set up options in some fashion (he calls this "choice architecture"), and we might as well do so in a way that increases welfare. The "increasing welfare" is paternalism, and doing so without limiting choice is libertarian.

An example he gives is that defaults tend to be sticky, and there is tons of data about this. In the rational model for human behavior, as long as the cost of switching is low, the default shouldn't matter much. In practice, the default matters enormously. (Perhaps one way to look at this is that even when switching is physically easy, it is mentally difficult, so the total cost is always high.)

Another example is feedback mechanisms, like installing a light which glows based on current energy usage. By making usage more salient, people conserve more. He contrasts this idea with the standard economics which just says "get the price right", behavioral economists also want to make the price salient.

It's hard to see what there is to disagree with in setting good defaults and making costs more obvious. Why do Thaler and Sunstein generate so much ill-will from libertarians?

Share this

Where to Start?

Perhaps here.

Or maybe here.

Cheers...

I'd be more sympathetic to

I'd be more sympathetic to libertarian paternalism if Sunstein/Thaler put forward proposals making currently mandatory government programs voluntary but "opt out". For example, you might be able to opt out of social security but have to sign a zillion forms saying you understand you won't get a government pension in old age, and so forth (of course, the trick there is having the government threat be credible).

But I have yet to see that out of libertarian paternalists (though I haven't read their book yet); instead, it seems to be more along the lines of "adding additional regulations, but making it opt out." Well, fine, but I want to opt out of some of the programs currently in place. That's why I can't get enthused about their program, though I agree with you that it has some merit.

UPDATE: I'm reliably informed that the Sunstein/Thaler book does in fact have some arguments for privatizing social security (I really do need to read it). So that makes me feel better what their proposing, maybe it really would lead to an increase in liberty.

It's hard to see what there

It's hard to see what there is to disagree with in setting good defaults and making costs more obvious. Why do Thaler and Sunstein generate so much ill-will from libertarians?

Part of it is a general libertarian knee jerk reaction which Will Wilkinson (ironically) pointed out in a different context. It goes something like "gubmint iz bad mmk." Plus there are a lot of anarcho-capitalists who think the government is unjust by its nature. The former probably don't understand the argument and its probably irrelevant to the latter.

I don't think it's

I don't think it's irrelevant to the latter; even in ancapistan, contracts need to have defaults, and welfare-maximizing defaults are preferable to the alternative.

Ancapistan defaults are private

In ancapistan, the law itself is private, the judges are private, and the defaults are consequently set by the market same as price. Or if you want to be more specific about what happens, private judges, who individually have no monopoly power, decide the defaults that will hold in their own court, and their customers by opting into those courts (by hiring them and not hiring competing courts), opt into those defaults. The nice thing about that system is that if Thaler becomes a judge, he's got about the same power to decide my defaults as McDonald's has to decide whether I'm going to have a hamburger or sushi for lunch today. It's a whole different ballgame when Thaler bends the ear of the state, with its monopoly on law.

Clearing up our terms

Or if you want to be more specific about what happens, private judges, who individually have no monopoly power, decide the defaults that will hold in their own court, and their customers by opting into those courts (by hiring them and not hiring competing courts), opt into those defaults.

Either you are making an assumption about what anarchocapitalism would look like or you're using a different definition of it than I am. I may be in the minority here, but the definition of anarchocapitalism that I default to unless told otherwise is a social system in which no individual, including those in 'governments,' is allowed to violate the rights of others or something similar. Under this definition, it is possible that a judge have a legitimate territorial monopoly, in which case it would be possible for a judge to have monopoly power.

You might have defined anarchocapitalism as a social system in which political entities- courts, defense organizations, etc.- don't have monopoly power. There's nothing wrong with this definition - I think I've seen others use it before - it's just not what I was talking about. Also note that rights violations can be institutionalized under this sort of anarchocapitalism which is why I shy away from this definition - I don't think it's what the anarchocapitalists are talking about, they just think that anarchocapitalism would lead to this sort of nonmonopolistic structure.

I think Constant is

I think Constant is referring to a coercive territorial monopoly of law. Originally "monopoly" always implied coercive, the meaning has changed though.

Indeed

Also, I'm not entirely sure what he means by,

Under this definition, it is possible that a judge have a legitimate territorial monopoly, in which case it would be possible for a judge to have monopoly power.

I'm not sure what he means by "territorial monopoly". How does the judge manage to have such a monopoly? Suppose that a second judge sets up shop with no connection to the first judge. Exactly what happens? Is the second judge blocked from operating in some fashion?

As far as I can tell, Matt actually does mean that the judge has a coercive territorial monopoly. After all, a lot of people do have a kind of blind spot when it comes to the service that judges provide. They think that as long as a judge judges justly, then nobody's rights are being violated. It never even occurs to them that forcing people to use a certain judge violates their rights. It is as though they think of the court system as itself outside of and above the law. Matt may not take this view - he may think that the judge is just so damned wonderful that he blows away the competition in fair, open competition. But if that's what he thinks, I don't know what he can possibly mean by "territorial monopoly". It sounds very state-ish.

non coercive territorial monopolies

I'm not sure what he means by "territorial monopoly". How does the judge manage to have such a monopoly? Suppose that a second judge sets up shop with no connection to the first judge. Exactly what happens? Is the second judge blocked from operating in some fashion?

In a word, yes. Just like I'm blocked from operating a factory on your front lawn. It isn't logically impossible for a judge to justly acquire the right to judge all cases within a given territory. Suppose, for example, that a judge legitimately acquires the state of Oregon, then sells the property off piece by piece along with a covenant built into the contract specifying that only he can adjudicate disputes occurring completely within Oregon. In this case he has a territorial monopoly over adjudication in Oregon, but no one has been coerced. It may not be a likely scenario, but the issue isn't likelihood, it's possibility.

It leaks

He is in effect the state. You've reconstructed something that walks, talks, and quacks like a state. It's a state.

Attempts to reconstruct the state by purely voluntary means sound plausible on the surface but there's usually a leak. Nozick tried to do it but he made a move which many libertarians don't accept as consistent with liberty. You're trying to do it by way of land ownership. I don't think it works.

Property rights are themselves consequences of law. If I have a property right in something, it is because my right is recognized in law, which means (if law is private) specifically that a court somewhere will respect my right. That court (or those courts) upon which my ownership depends can't be (just) my own court with me as judge, because then my ownership of the property would mean only that I consider myself the owner, and this is simply not enough to constitute ownership.

So my ownership of property is itself subject to the decision of judges other than me. Consequently, a number of disputes concerning my property are possible which, necessarily, must be resolved in a court other than my own.

We can approach the matter from a different direction. Start with this formula:

a social system in which no individual, including those in 'governments,' is allowed to violate the rights of others

Suppose, then, that someone's rights are violated, and that the territorial monopolist judge falsely rules that no rights were violated and furthermore that the same thing (i.e. the violation, declared by the judge to be "not a violation") must be done to the victim ten more times in order to teach the victim a lesson. Then what? By assumption, this is not allowed to happen. By assumption, then, there is some constraint on the judge.

But what is the nature of the constraint? If we assume that there is truly no judge but that judge, then that constraint must be internal to the judge - it must be psychological. The judge must be a supremely virtuous person. But this violates the assumption that what disallows the violation is "a social system". Furthermore, if the constraint is psychological, then the fact is not that the judge is not allowed, but rather that he chooses not to. This is not the same as "not allowed".

So this contradicts the formula. Twice.

When I say that the judge

When I say that the judge has a territorial monopoly on adjudication, I don't mean he is unconstrained. He is still subject to appeals. The sense in which he has a territorial monopoly is that if you want to live, do business, or otherwise be within the territory, by covenant, you have to go to this judge to adjudicate any disputes. He may make a mistake and you may want to appeal his mistake to some mutually acceptable third party, but you have to go through him, and consequently pay him, first. His monopoly over adjudication is no different than your monopoly over building factories in your backyard. If anyone wants a factory in your backyard, they have to pay you. If anyone wants adjudication in Oregon, they have to pay the judge.

Instead of obtaining a monopoly over adjudication in this way, the judge could obtain a monopoly over lawmaking. Suppose the covenant gives the judge the right to make up all the rules in any way he sees fit within Oregon. In effect, he has obtained a right to tyranny. He still isn't a state, however, because his power rests on the enforcement of external courts. Does this mean he can legitimately kill whoever he wants within? Well, if he specifies this in the covenant, I see no reason why not, at least assuming away any potential problems with inalienability. He would have to let people know that they are agreeing to this when they enter his territory, but in principle it's possible. If the murder park thought experiment violates no one's rights, I don't see why this would. If you do have problems with inalienability, then simply ignore this paragraph.

But then he's irrelevant

He may make a mistake and you may want to appeal his mistake to some mutually acceptable third party, but you have to go through him, and consequently pay him, first.

But then his monopoly is irrelevant to the issue. If his decisions can be appealed, then he does not get to decide law unilaterally, and if he does not, then his monopoly is simply irrelevant to the question of tinkering with law, which is the question at hand.

Instead of obtaining a monopoly over adjudication in this way, the judge could obtain a monopoly over lawmaking. Suppose the covenant gives the judge the right to make up all the rules in any way he sees fit within Oregon. In effect, he has obtained a right to tyranny. He still isn't a state, however, because his power rests on the enforcement of external courts.

What you are describing is a contract which gives one party to the contract the power to arbitrarily make up new requirements of the other party as he goes along. I am not sure that such an open-ended contract is even valid. The whole point of a contract is that it specifies agreed-upon conditions in advance. And what is not specified is subject to legal defaults.

This much more extreme than an employment contract. An employment contract is somewhat open-ended in that it does not specify in detail what the employee will do - he agrees, rather, to work for someone else, who will decide as he goes along what is to be done. But such a contract is not fully open-ended in the sense that a monopoly on lawmaking is openended.

Don't appeals need to end

Don't appeals need to end somewhere? Is it really true that a contract cannot legitimately have a single, final arbitrator? This arbitrator need not be a party to the contract.

Conflicts end, they don't have to be forced to end

Conflicts end whenever the parties decide that the cost of pursuing the conflict is greater than the benefit. Usually this happens fairly quickly. You don't need to bake a hard end into the system of conflict resolution in order for the vast majority of conflicts to come to a quick end. And I'm not sure you actually can.

A court case is a kind of negotiation between two parties - one assisted by a mediator. In principle I can always approach someone with whom I have finished a negotiation and re-open negotiations. Even if we agree not to negotiate further, the very next day we may find ourselves negotiating further (e.g. we might both change our minds, or one of us might offer the other one a sufficiently tasty incentive). In practice at some point I decide that I have better things to do and need to move on with my life.

If you try to force an end to conflict, either or both of the parties can always come into conflict with your attempt to force and end.

If you are worried about addressing the occasional conflict that never comes to an end, in the real world I don't think you can absolutely put an end to those no matter what you try. Long-running feuds will probably exist for as long as the very possibility of conflict exists.

A simple example showing that there is no such thing as finality. Suppose you define the final arbiter is Bob. Bob is adjudicating a dispute between neighbors who are arguing about where to build a fence, or something really minor and mundane. Bob rules that Andy (one of the parties) is obligated to kill his own family, burn down the houses of all the rest of the neighbors, and then commit suicide. My guess is that, even if everyone had previously agreed firmly that Bob was going to have the absolute last word on the conflict and that the two parties were going to do whatever Bob said, whatever it was, nevertheless if this is Bob's ruling, guess what, he wasn't the absolute last word after all.

If whether someone has "the absolute last say" depends on what he says, then he doesn't in fact have the absolute last say.

Repairing a faulty example

But then his monopoly is irrelevant to the issue. If his decisions can be appealed, then he does not get to decide law unilaterally, and if he does not, then his monopoly is simply irrelevant to the question of tinkering with law, which is the question at hand.

His monopoly is relevant in one way: you still have to pay him, but my example turns out to be faulty. A better example will show it relevant in another way.

Instead of the covenant making our judge the sole adjudicator, assume it gives him the right to create and enforce law within Oregon as well. There are limits, of course. He can't simply murder you, he has to give reasonable notice of law changes, etc. Instead of Oregon, let's make his territory Portland, just to make it more intuitively plausible. What we have is a private city. If you live, work, do business, or otherwise willfully make your way into the city, you agree to abide by the laws of the city, however good or bad they are. There are all sorts of little taxes in the city to fund it's operation - retail, income, etc., except they are called "subscription fees." There might even be a toll at the border for any nonresidents who enter the city.

Now his monopoly is relevant in a second way - he can provide really terrible law within Portland and the only thing anyone can do about it, short of buying the city from him, is leave. If our judge makes Portland a terrible place to do business because of European style labor controls, that is his prerogative. His law ultimately rests on appeals courts again, but they should enforce it - everything is stipulated in contracts which any disputant presumably agreed to. If the contract is too open ended for your liking, there's no reason why every single law could be stipulated in the covenant beforehand with absolutely no way for our judge to change them.

The point of all this is that our judge has a territorial monopoly in the same way that the electric company does - within the given territory, there are no competitors, so the both can charge high(er) prices for a sub par good/service. There's also no reason why the territory has to be the size of a city. Theoretically, the judge could acquire the entire world in this way, giving him the ultimate monopoly power.

A long post deserving a long

A long post deserving a long answer but in a few words : no ! A state is an organization that engages in institutional coercion through a territorial monopoly on law. If I own land and I decide that I shall be the only judge at home, I am not coercing anyone, no one is forced to live on my land. I am not a state. And property rights do not depend on law, private or not. They don't need to be "recognized". They exist. Good law recognizes them, bad law doesn't, that's it.

Really?

Not even natural law? You think that private property rights are logically prior to natural law and, therefore, to natural rights? I thought that rights were, well, rights. They're spelled the same. r-i-g-h-t-s. Or do you distinguish between natural law and natural rights? If you do, then you are using jargon that I am unfamiliar with, as natural law simply is the principle that natural rights may not be violated, and natural rights simply are that which are protected in natural law - flip sides of the same coin. As Wikipedia points out:

The theory of natural law, a law whose content is set by nature and therefore has validity everywhere, is derived from natural rights.

Anyway, that seems to take care of the natural rights point of view. I actually dealt with the two views of anarcho-capitalism in my previous comment but I will make the distinction more explicit here. The natural rights point of view holds that there simply is a set of rights which may not be violated.

An alternative point of view of anarcho-capitalism is David Friedman's point of view, which does not take either natural law or natural rights as a given but imagines competing courts. Without getting bogged down in details, this alternative point of view does not recognize natural law or natural rights, this is not part of its scope, but it recognizes only the law as decided by courts. And courts in fact decide property disputes. Therefore, considered from this viewpoint, a person's property rights themselves are subject to the courts.

I addressed the two viewpoints, first David Friedman's more positivistic viewpoint, and second the natural rights viewpoint.

Furthermore, if we adopt the natural rights/natural law viewpoint, then the whole idea of tinkering with the law to set defaults is immediately placed into question, because you can't tinker with natural law.

Because of this, really the only interesting point of view here is David's positivistic approach, which considers the actual law as decided by private courts as being where it's at.

Patri can of course step in and correct me, if he reads this comment. Or, for that matter, David. The problem with talking about the ideas of living people is that they can correct you.

The problem with talking

The problem with talking about the ideas of living people is that they can correct you.

I hear-I heard what you were saying. You-you know nothing of my work. You mean my whole fallacy is wrong. How you ever got to teach a course in anything is totally amazing.

Well, assuming that the

Well, assuming that the anarchocapitalist cares about welfare maximization. A large group of them (the majority?) probably don't. What they want private courts and contracts to do is give every relevant party something they can agree upon, including the defaults. If that means welfare maximizing defaults, ok. If not, so much for the welfare maximizing defaults. In other words, when I said 'probably' above, I meant that if you pick a random ancapitalist, he probably falls into this category.

Who decides

I think that a lot of lib resistance to Thaler/Sunstein paternalism (and please, let's not beg the question by calling it "libertarian paternalism" just yet) comes from the perception, shared by many on the left, that the "choice architect" must be the government or acting under government decree. If that's the case then the whole T/S paternalism model just becomes a stalking horse for hard paternalism --

step 1, government decides what you should do
step 2 government makes what you should do the default
- if many people opt out then step 3a the government mandates the program because the soft paternalism wasn't working;
- if few people opt out then step 3b the government mandates the program because few people object and the choice system is expensive/inadequate/flawed.

I think there's a strong case for private T/S paternalism -- opt-out 401(k) contributions for instance, might in the long run improve employee happiness and so give an advantage to a company that used them. I think the case for governmental or government-mandated T/S default rules is much more troublesome, though, because it goes right back to the old "the government knows better than you do" problem -- the defaults chosen are chosen by people just as biased and flawed as us proles -- and the problem that government power tends to expand.

Some additional points against it

I think Will Wilkinson nailed some of the key problems (see Kip's links).

Some random points, mostly overlapping

1) What part of "my life and my choices are none of your business" is so hard to understand?

2) The paternalist mindset (you're foolish and I know better and it is in fact my business to nudge you in the right direction) is not universally shared and is found offensive by some, some of those being libertarians.

3) There's a reason we sometimes call people "nudges" when we're annoyed by them and want them to go away and leave us in peace.

4) There's a reason for the tension between teenagers and parents. Parents are paternal toward their children, and people once they begin to enter adulthood start to dislike that, sometimes intensely.

All of (1) through (4) are independent of whether coercion is applied. The great thing about private nudges (including, say, paternalist employers) is that you can get away from them, e.g. by quitting your job and getting another one. The state is harder to get away from.

5) The argument that it's libertarian purports to be independent of the content of the defaults - since, whatever the default is, the person is free to opt out. But the example defaults that I have so far seen are something that a reasonable and thoughtful person might consciously opt into. The argument that the option to opt out makes the approach liberty-preserving isn't really being put to the intuitive test by these reasonable-seeming examples. What would clinch the argument is to make up examples with over-the-top evil defaults, and then see whether the claim that liberty is preserved passes the smell test. For example, suppose that the default in some middle eastern state is that Jews consent to being killed by anyone for any reason, and must fill out and submit a form to opt out of this default consent. If Thaler is right about inertia, then a significant number of Jews won't think to opt out. Is it really morally all right to kill Jews who neglect to opt out, because, after all, they had the choice? Is their liberty truly being preserved?

It's Simple Status Quo Bias

Thaler and Sunstein seem perfectly reasonable in trying to bring immediately applicable policies to the public sector. From what I've read their whole theory seems to be based off implementing policies with the status quo bias in mind. I recently took a class on experimental economics (unfortunately, a lot of the stuff is not functional outside the lab), and the theory basically goes: people follow the status quo even when they would truly prefer something else if they were forced to choose. So simply because they are given a default they follow it but would truly want another plan which they can still change to but do not because it is not the default.

I wonder if one could use other anomalies of irrational human behavior such as the endowment affect, loss aversion, intertemporal choice, or preference reversals to try and reform the public sector.

Electricity Would Be Too Cheap To Meter

>Why do Thaler and Sunstein generate so much ill-will from libertarians?

Probably because most of us realize that the next asteroid isn't going to be diverted by a bunch of people standing around watching their electric meters, separating their trash (which then all ends up in the landfill anyway), or doing any other useless rituals whether Green, Red, or Mauve. Ritual makes people FEEL that they're "doing something", while we march back into the Dark Ages. This is what we need to do:

http://www.lewrockwell.com/walker/walker14.html

Libertarian paternalism

"I'd be more sympathetic to libertarian paternalism if Sunstein/Thaler put forward proposals making currently mandatory government programs voluntary but "opt out""

A number of people in blog discussions such as this raised the question of just how libertarian, in several respects, libertarian paternalism is. Cass Sunstein is an ex-colleague, so I asked him--both about the freedom of private choice architects and about rolling back current non-libertarian paternalist policies. A summary of his response, written by me then amended by him:

Yes, private choice architects should have freedom of choice. Some, but not necessarily all, current paternalist measures should be rolled back; in particular the book explicitly supports school vouchers and freedom of contract to allow doctors and patients to contract around medical malpractice liability.

My conclusion is that although Cass is not currently a hard core libertarian, his position is libertarian in the weaker sense that adds up to 10-15% of the electorate--he wants less (but not necessarily zero) government compulsion in both economic and social matters than now exists.