Discussão Thaler e Rizzo sobre Comportamento Humano

O Wall Street Journal promoveu uma discussão entre Thaler e Rizzo sobre o comportamento humano. A seguir a reprodução da discussão:

Should Policies Nudge People
To Make Certain Choices?

Driven by research in behavioral economics that suggests people don't always act in their own best interests, some economists are arguing for new policies that would challenge traditional "hard" tools for changing behavior, such as sin taxes and outright bans.

Such policies would often rely on default options that nudge, steer and coax -- but don't force -- individuals to make certain choices. Is this sensible governance? The Online Journal asked Mario Rizzo, a professor of economics at New York University and director of NYU's Program on the Foundations of the Market Economy, and Richard Thaler, professor of economics and behavioral science at the University of Chicago's Graduate School of Business, to hash it out.

What do you think? Share your comments on our discussion board.

* * *


Richard Thaler writes: Behavioral economics is founded on the unremarkable observation that human beings are imperfect decision makers. They have limited information-processing abilities, willpower, memory and attention. As a result, they make predictable errors by their own lights. Many Americans think of themselves as overweight; most 401(k) participants think they are saving too little; and nearly everyone thinks of himself as forgetful.

In light of human limitations, Cass Sunstein and I argue for policies that we call libertarian paternalism. Although the phrase sounds like an oxymoron, we contend that it is often possible to design policies, in both the public and private sector, that make people better off -- as judged by themselves -- without coercion. We oppose bans; instead, we favor nudges.

Consider two examples, both designed to increase savings. The first is to enroll people, automatically, into savings plans -- while allowing them to opt out. The second is the Save More Tomorrow plan, which allows employees to commit themselves now to increasing their savings rates later, when they get raises. Both approaches have been remarkably successful.

Well-chosen default rules are examples of helpful "choice architecture." Since it is often impossible for private and public institutions to avoid picking some option as the default, why not pick one that is helpful?

* * *


Mario Rizzo writes: The decisions of individuals may be imperfect but can the legal paternalist successfully steer them toward better decisions?

I say legal paternalist because Cass Sunstein and Richard Thaler clearly do not object to using coercion at the level of framing decisions. For example, they seem to approve of employers' being legally required to provide automatic 401(k) enrollment unless the employee opts out. But they also approve of mandatory cooling-off periods for consumer purchases; these cooling-off periods absolutely prevent individuals from concluding immediate exchanges even at lower prices.

It is a good thing to help people make better decisions. But law requires us to go beyond intention. What is the appropriate standard for better decisions? Thaler and Sunstein say it's what people would do if they had "complete information, unlimited cognitive abilities, and no lack of willpower." This is a very ambitious standard that could tax the abilities of even well-meaning policymakers.

Can we discover "true" preferences through individuals' statements that they are too fat and save too little? Talk is cheap. These could be expressions of mere desire, not a real willingness to make trade-offs between values. We all want to have more savings and more consumption, too.

Moreover, the public sector is not governed by science or even by behavioral economists, but by ambitious people with limited cognitive abilities, lack of willpower, and faulty memories, not to mention expanding waistlines. Whom should we trust more: individuals who face the costs and benefits of their own choices, or politicians and bureaucrats who do not?

Richard Thaler writes: It is both wrong and misleading to characterize libertarian paternalism as primarily an activity of governments. Automatic enrollment and Save More Tomorrow have been adopted in thousands of companies with no governmental involvement. When the government does get involved, we prefer nudges to requirements. A good example is the 2006 Pension Protection Act. Under that law, firms that offer to at least partially match their employees' contributions, enroll their employees automatically, and automatically escalate their contribution rates are given a waiver from some burdensome paperwork. No coercion is involved.

We agree that it can be difficult to determine people's true preferences. And it is an axiom of behavioral economics that intentions do not always -- or even usually -- lead to action. But statements of good intention can signal a desire for help in following through. Many employees have voluntarily signed up for the Save More Tomorrow program, and very few subsequently quit. We do not think that many people would sign up for "Smoke More Tomorrow," or "Eat More French Fries Tomorrow."

We agree that government workers are human. We are also happy to go on the record stipulating that politicians are boundedly rational. Some are also dishonest. Some are even fat. However, what are the implications of these obvious facts? Do we want to charge our political leaders with the task of making people worse off?

Often nudges are inevitable. Where they aren't, we agree that unless there is a good showing of need, government might do best to stand aside.

* * *


Mario Rizzo writes: Is New Paternalism primarily about advising private individuals and firms? If so, why use a political term -- libertarian -- to identify it?

It's true, some firms have adopted automatic 401(k) enrollment policies -- while fewer have adopted the Save More Tomorrow program. The market allows for this diversity while eliminating ineffective or inappropriate plans. No libertarian I know of has ever opposed privately adopted options. If this is all Thaler is saying, what's new?

But Thaler and Sunstein do go beyond this. Elsewhere, they've argued for costly contractual provisions like vacation time, allowing only termination for cause, non-waivable cooling-off periods, maximum 40-hour work weeks, and -- presumably -- the legal requirement of automatic savings plan enrollment, if not enough firms voluntarily adopt it. All this is said to be consistent with "libertarian paternalism."

Thaler says that those automatically enrolled in 401(k)s haven't quit, so they must be benefiting. This is an odd claim for a behavioral economist. Why would failure to change indicate a benefit? When the default is non-enrollment, Thaler says that individuals tend to stay in it because they're irrationally biased toward keeping the status quo.

The mistakes of bureaucrats, politicians and voters aren't as likely to be corrected by individual or social processes as errors made in the private sphere. Why? Because bureaucrats and politicians don't care about private welfare as much as individuals do and voters don't have much incentive to become informed. So people -- who make imperfect decisions -- tend to do less harm if they "nudge" only themselves and not others through policy choices.

* * *


Richard Thaler writes: I am glad Mario agrees with our private initiatives on retirement savings in which firms have nudged rather than required employees to take certain beneficial actions. Surveys of employers suggest that a majority will be offering automatic features by next year, a big step forward. Of course, the fact that few employees opt out is partly due to inertia, but most employees do get around to joining the plan under opt in, so the main gain from automatic enrollment is to get people to join sooner, something they appreciate.

Mario's major complaints are with positions that we do not advocate, namely what he calls the "legal imposition of costly contractual presumptions." We have never suggested that any particular contractual form be imposed, including automatic enrollment. See my previous post. Instead we stress that when governments do write laws, especially those mandating -- rather than nudging -- some action, they should do so with an eye toward making people better off. We do admit to liking some mandated -- and thus non-libertarian -- cooling-off periods under certain circumstances, as when buyers are especially likely to have made decisions under undue selling pressure. Who amongst us has not bought something under pressure that he would like to undo the next morning?

Mario's main misconception is that government can avoid nudging. It can't. The rules of the common law are legal rules that governments write. Whether governments are more or less corrupt than the private sector is an empirical question, and there are surely many examples of dumb or unethical behavior in both sectors. But this is beside the point. We favor better government, not more government. We urge both sectors to adopt libertarian paternalistic policies.

* * *


Mario Rizzo writes: I repeat: "Is New Paternalism primarily about advising private individuals and firms? If so, why use a political term -- libertarian -- to identify it?" It is simply a management-consulting philosophy.

If a firm chooses a default option, it is by no means inevitable that it must choose on the basis of paternalistic criteria. Under purely voluntary conditions, it will choose so as to enhance the attractiveness of its compensation package, that is, according to the actual preferences of its employees. Its goal is to maximize profits. To anticipate what employees or consumers want is the market principle, not paternalism. I repeat: "If this is all Thaler is saying, what's new?"

If automatic enrollment proves popular in the long run, then, at least most people must be aware of their procrastination bias -- assuming it exists -- and want to overcome it. The previously hostile legal environment had prevented employers from responding to this de-biasing preference.

As to the more intrusive examples of paternalism mentioned in my previous post, Richard and Cass Sunstein have indeed argued that they are consistent with libertarian paternalism. Let the reader decide.

The standard Richard advances that "when governments do write laws, especially those mandating -- rather than nudging -- some action, they should do so with an eye toward making people better off" may seem innocuous, but it is actually dangerous. This is because just about anything can slip by. (Satisfaction of informed preferences is an obscure criterion of "better off.") Such a standard could set in motion a slippery slope to much more intrusive interventions, especially in a world of boundedly rational individuals who tend to view the world in a narrow frame. It does not take corrupt public officials to go down this road. Self-interest and bounded rationality are quite enough.

* * *


Richard Thaler writes: Let's recapitulate. People make mistakes, so sometimes they can be helped. It is possible to help without coercion. That is libertarian paternalism. The concept can be and is used in both the public and private sectors. For example, in London, pedestrians from abroad are reminded by signs on the pavement to "look right" because their instincts from back home are to expect traffic to approach from the left. No one is forced to look right, but fewer pedestrians are hit by trucks.

Another example comes from Sweden, which launched a partial privatization of their social security system in 2000. The plan was open to any fund, which meant that participants faced 456 options. There was also a very well-designed default fund -- using private managers selected by the government -- that offered global diversification at very low fees (16 basis points). By any standard, both ex ante and ex post, the participants who selected their own portfolio of funds did worse than those who took the default plan. The main mistake the government made in designing this plan was to discourage participants from choosing the default fund, perhaps thinking, as Mario does, that choosing for oneself is always the best approach.

Mario thinks we are naïve about government. We think he is naïve about firms. Does he think that the companies that offered stock options to student loan officers to induce them to feature their loans had the "actual preferences" of the students at heart? Maximizing profits does not always mean maximizing the welfare of the customers.

Finally Mario seems to have a phobia about slippery slopes. I guess he thinks that if governments start with signs that say "look right," the next thing you know we will have Prohibition coming back. By the same logic, we should worry that if libertarians succeed in eliminating rent control that we will be soon down the slippery slope toward anarchy. Slippery slope arguments should be avoided unless there is proof that the slope is greased. In our case, by insisting, as we do, on only libertarian paternalism, the slope runs into a brick wall before it even gets started. And besides, what is the alternative? Inept neglect?

* * *


Mario Rizzo writes: Libertarianism is a political philosophy that seeks to reduce the activities of the state to a very low level. It is very much about less government. Paternalism is a political or moral philosophy that seeks to override the actual or operative preferences of individuals for their own benefit, however defined, according to Donald VanDeVeer's 1986 book on the subject. When applied to the actions of government, paternalism cannot be libertarian. It can only be more or less intrusive.

Does Richard wish to reduce his "libertarian paternalism" to the appropriate management of government-owned streets or other enterprises? In the London case, what people want is obvious: They don't want to get hit by cars. London is doing what entrepreneurs generally do: satisfying actual preferences. London is mimicking the market.

In Sweden, the government actively discouraged people from relying on the default investment option. People probably interpreted this as meaning the default option was not very good. They succumbed to this unfortunate inference because they viewed the government as an authoritative investment adviser. Government provision of investment advice is not consistent with libertarianism. But if it does provide advice, is it paternalistic to provide it in such a way that people make reasonable inferences? If Vanguard provides good advice, is that paternalism? In each case, this is just satisfying actual preferences for advice. (Note that none of this requires reference to the idea of "true" or "informed" preferences about which so big a deal was made.)

Richard wants to use the word "libertarian" to differentiate his paternalism from the traditional variants. Yet he uses the word in a fuzzy way. He wants to define libertarian along a continuous variable -- the cost of exercising the exit option. However, libertarianism, as every libertarian understands it, uses a bright-line test -- who imposes the cost? The authors of the concept of "libertarian paternalism" have said that clearly intrusive/coercive interventions are consistent with it. See my previous post. And they have also said, explicitly, that there is no sharp line between libertarian and non-libertarian paternalism. Thus, Richard cannot claim that his standard creates a bright-line rule that would help us resist the slippery slope.