Libertarian Paternalism: A Test
Existing Regulations and Programs Could Be Altered to Conform to Libertarian Paternalism
JULY 01, 2007 by DWIGHT R. LEE
Behavioral economics is a growing subfield of economics based on the finding that people are not as rational as economic models have traditionally assumed. Numerous experiments have shown that people’s choices are systematically altered in response to changes in how those choices are framed, even though the framing is irrelevant to the consequences of those choices.
Some behavioral economists now seem intent on bringing their own rationality into question by advocating what appears to be a loopy idea—giving government more power over our decisions to make us better off, and doing so in the name of libertarianism. This “libertarian paternalism” sounds like an oxymoron, but two of its leading advocates—Cass Sunstein and Richard Thaler—assure us it is not. (Their 2003 University of Chicago Law Review article is titled “Libertarian Paternalism Is Not an Oxymoron.”)
Their argument is superficially plausible. But on closer examination it becomes clear that a big leap of faith in the integrity of the political process is required to embrace libertarian paternalism as a reasonable proposal. Fortunately there is a test of the compatibility of the political process with libertarian paternalism. I shall argue that this test should be given, and must be passed, before taking libertarian paternalism seriously. But, first, let’s consider what it means and the arguments for it.
The case for libertarian paternalism begins with a finding by behavioral economists of what can be described as the default bias. In many circumstances, most people will accept a default option rather than choose another even when the stakes are high and the other options seem far better. A commonly mentioned example is a matching, tax-deferred 401(k) plan at work. When employees have to explicitly decide to join, typically more than half accept the default of not participating even though signing up is easy and, with the employer’s contributions, assures an attractive return. On the other hand, studies show that if employees are automatically signed up unless they opt out, most remain in the program.
The next step in the case for libertarian paternalism gets us to the paternalism. Since in every choice there is a default option, it is a good idea, say advocates of the principle, to make it the option that is best for most people. And fortunately it is often possible for impartial experts with training in behavioral economics to know what the best option is. Therefore, if government frames the choice by requiring that the “best” option be made the default option, then the default bias would result in most people accepting this choice. But the libertarian part of the argument acknowledges that what is best for most people is not best for all, so anyone may choose something other than the default.
For example, an employee who is independently wealthy, or who isn’t expected to live long enough to retire, or who just doesn’t want to start a savings program can choose to have her name removed from the 401(k) plan. No one is forced to bend to the will of an authoritarian paternalist. The government experts, under the guidance of libertarian paternalists, will behave like libertarians by leaving the ultimate decision to the individual, but they will also be paternalists by exploiting the default bias and framing the decision in a way that makes people better off (in their view).
The obvious leaps of faith behind this proposal are in believing 1) that the political power required to implement libertarian paternalism would be limited to setting default options when the best default for most people is known and 2) that the choice of default will be guided only by the interest of the people whose choices are being affected. Without this faith in the political process the case for libertarian paternalism is wholly unconvincing. Can we actually expect that the political process will know what option is really best for most people in different situations and how a different default would affect decisions? Maybe behavioral economists are convinced that they know when they cannot pick a better default than already exists and what the best default is when they can. But even if this is true, behavioral economists will not be doing the picking. They may be brought in as advisers to the politicians and bureaucrats in charge of designing and implementing libertarian paternalism, but their advice will not always be taken. Others with more political influence and more narrowly focused interests than behavioral economists will also be giving advice—and will be doing so aggressively.
For example, if the authority to set the default on 401(k) plans is transferred to government, the financial industry will quickly recognize the potential profits in influencing how that authority is exercised. Indeed, the legislation giving that authority to government will be drafted with help from the financial industry. And there are ways to put some devil in the details of such legislation. The legislation could cover workers who, because they are young, part-time, and receive low pay, would not normally be offered 401(k) plans. Opting out could be made to require more effort than the libertarian paternalists have in mind. The type of investment people are put in if they accept the default option would be subject to political influences that have little to do with what is best for the employee. And, assuming the behavioral economists are correct and most people stay with the default, the political clout of financial firms (likely with the help of political restrictions on competition) could result in higher financial fees. (Concern over high financial fees was one argument used by those who successfully opposed President Bush’s attempts to partially privatize Social Security.)
Placement of Food
Libertarian paternalists use examples other than savings plans to illustrate what they have in mind. They often mention the placement of food in cafeterias, buffets, and grocery stores. Again they argue that the food items have to be placed somewhere, and behavioral economics has apparently shown that choices of what to eat are influenced by where food is placed. So food encountered sooner in cafeterias and most conveniently located in grocery stores can be thought of as default options—it is still possible, of course, for people to choose other options. As far as I know, no advocate of libertarian paternalism has yet recommended that government regulate the location of food in cafeterias and grocery stores. But if the principle achieves political traction, there are sure to be pressures in that direction—though maybe cautious and subtle at first. Who would have thought not long ago that governments would soon be imposing restrictions on where and how cigarettes are displayed in stores or what type of fat is permitted in fast food?
If government began moving toward using libertarian paternalism to encourage people to choose healthier diets, political competition among suppliers of grocery-store products would intensify. The fruit and vegetable suppliers, along with the organic-food farmers, would lobby regulators to have their products moved to the front of the store. The producers of soft drinks, candy, chips, bean dip, and a host of other processed foods would use their significant political influence to maintain their locations. Even the magazine publishers would join the fray. No one knows how this political competition would play out. But don’t expect that fruit, vegetables, tofu bars, and maybe cigarettes made with organically grown tobacco would soon dominate the shelves at the checkouts, with the candy and gum there now being moved as close as possible to the loading dock. Or that you would find serious reading material such as The Economist, Barron’s, The Nation, National Review, and The Chronicle of Higher Education in the checkout racks, with the magazines on dieting, fashion tips, and the latest activities of Hollywood celebrities relegated to where the tofu used to be.
Of course, a move in this direction cannot be ruled out. What we can be confident of is that more resources that could have been used to produce goods and services valued by consumers would be devoted to influencing political decisions. The financial well-being of government regulators and lobbyists would certainly be improved more than the dietary well-being of the public.
Some may think that this depiction of an overreaching political process powered more by organized interest than by the public interest is overstated. And perhaps even if the political process does possess the special-interest tendencies just discussed, an idea like libertarian paternalism has so much potential to serve the public interest that the ability of narrowly focused interests to sabotage it will be severely limited. If libertarian paternalists believe this, then they should not object to a test of the political process’s affinity for implementing libertarian paternalism primarily to benefit the public. There is such a test, and libertarian paternalists should require the political process to pass it as a prerequisite for their continued advocacy.
The test requires considering the many paternalistic regulations that governments now enforce in unlibertarian ways and recognizing that they could be made consistent with libertarian paternalism. A political process compatible with new regulations in the spirit of libertarian paternalism should be able to pass the test by modifying existing regulations to make them consistent with that spirit. Several examples of existing regulations and policies that can be used in this test come to mind.
First, the Bush administration’s plan to partially privatize Social Security would be a move in the direction of libertarian paternalism. The current Social Security system would remain the default option, but workers could reduce their payroll tax up to some specified amount that would increase over time by an amount equal to each dollar they put in a private savings program—with restrictions on accessing the savings before a certain age. There are obviously transitional issues involved in such a policy shift, but if libertarian paternalists and others cannot convince the federal government to establish even a limited private option to Social Security, it will not be an encouraging sign for the political feasibility of libertarian paternalism.
Second, taxpayers are now required to help the poor by having a certain percentage of their tax dollars transferred to them (a smaller percentage than most people believe) by various levels of government. Instead of requiring that those transfers be made through governments, we could make government transfers the default option, but with taxpayers allowed to reduce their tax payments by some significant percentage of the money they donate to private charities dedicated to helping the poor. This libertarian-paternalism approach has real potential for improving the way the poor are helped, since private charities are better than government agencies at assisting the poor without inducing dependency. But this potential depends on government’s being willing to accept competition from private organizations by allowing people to reject the default option. If governments reject this shift to a libertarian-paternalism approach, it would have to be seen as a failure on the test of governments’ willingness to properly implement libertarian-paternalism policies.
Third, smoking regulations suggest another way of testing governments’ willingness to exercise the tolerance for alternative approaches required of libertarian paternalism. An increasing number of state and local governments have outlawed smoking in privately owned establishments that serve the public. The purpose is to protect people against being involuntarily subjected to secondhand smoke. Of course, many people who frequent restaurants, bars, pool halls, bowling alleys, and other establishments enjoy their experience more if they can smoke, with secondhand smoke being an unobjectionable and trivial supplement to firsthand smoke. The preferences of these people are largely, if not completely, ignored by existing antismoking regulations. This situation would be improved by the libertarian paternalism approach. No-smoking could be the default policy, but the owner of a restaurant, for example, could opt out of that policy by making it clear to potential patrons that smoking is allowed. Plenty of options would remain for people to dine out in nonsmoking restaurants, but there would also be options for those who enjoyed smoking while dining, drinking, shooting pool, or bowling. If governments are sufficiently libertarian to offer this alternative to the default policy of no-smoking, it would be an encouraging sign for libertarian-paternalist policy. If governments resist options that let people enjoy smoking while in the company of other smokers and those who don’t mind secondhand smoke, it has to be considered a failure on the test.
Fourth, regulations against insider trading are another area where libertarian paternalists should be able to get government to liberalize its approach if their principle is in harmony with political incentives. The purpose of insider-trading prohibitions is to prevent outsider investors in corporate stock from being harmed by the trading of those with inside information. The objective is a good one, but prohibiting insider trading may not be the best way to achieve it in all cases. For example, regulation against half the perceived problem is unenforceable, since prohibitions against using insider information to decide not to trade cannot be enforced. Also, the price information created by insider trading can be useful to outsiders and can increase the general efficiency of financial markets. Instead of an outright prohibition, no trading on insider information could be the default policy for corporations, but they could opt out in favor of a policy allowing insider trading by making that policy clearly and widely known. This libertarian-paternalism approach would allow corporations and investors greater choice in how they operate and invest. In particular, outside investors could protect themselves against any harmful effects they perceived from insider trading by not investing in the firms that allowed it. And to the extent that investors did perceive harmful effects from insider trading, the relative price of stocks would compensate for those effects. How willing the Securities and Exchange Commission is to move to this approach would give useful information on how appropriately libertarian-paternalist policies in general would be implemented by government.
A final example of a political libertarian-paternalism test (out of the many that could be considered) involves increasing the supply of organs for transplant. Currently in the United States people can choose to donate their organs on their death by indicating a willingness to do so, usually when they obtain or renew their driver’s licenses. Not making the “donate” choice means that a person has automatically chosen the default option not to donate. As behavioral economists predict, most people choose the default option. Libertarian paternalists don’t think this is the best choice for most Americans, and so they want to switch the default option to “donate.” Being the libertarians they are, however, libertarian paternalists would allow people to opt out by explicitly choosing not to donate.
As evidence from other countries suggests, this would significantly increase the number of Americans who choose to donate their organs. But one would think that real libertarians would want to allow people to sell their organs as another nondefault choice. This is more libertarian (indeed a policy restricting our choices to either keeping our organs or giving them away is not even remotely libertarian) and would probably increase the number of organs available to those who need them. Another advantage libertarian paternalists should see in adding the selling choice is that it would provide another measure of how much the political process could be trusted not to sabotage the other applications of the principle they are recommending.
None of these libertarian-paternalist alterations would bring perfect policies. But how good would they have to be to be better than what they replaced? And certainly they would result in policies more in the spirit of libertarian paternalism. If anything, libertarian paternalists should want to change some of the defaults from what is required by existing policies to options that would be paternalistically superior. But this is a test, and leaving the defaults where existing policies have placed them surely makes it easier for government to pass.
If those recommending libertarian paternalism really want to improve public policy, they should be anxious to subject governments to this and other libertarian-paternalism tests. If the governments pass, existing policy will be improved by bringing them more in line with libertarian paternalism. On the other hand, if governments consistently fail, this will provide libertarian paternalists with important information. Governments unwilling to alter existing regulations and programs so they conform to libertarian paternalism are unlikely to create new ones that satisfy the principle’s requirements. Far more likely is that the rhetoric of libertarian paternalism will be used to justify the expansion in government’s authoritarian paternalism.
Of course those claiming to be libertarian paternalists may object to the test because they think the existing regulations are acceptable as they are. But this would suggest that they are really authoritarian paternalists trying to pass themselves off as libertarian paternalists. So my test is not only a test of the political process’s compatibility with libertarian paternalism. It is also a test of the libertarian paternalists’ commitment to libertarianism.