Cato Unbound has an excellent symposium on “libertarian paternalism,” the theory that argues that government should intervene to protect people against cognitive biases that lead them to make decisions that ultimately reduce their ability to achieve their own objectives. Advocates of libertarian paternalism argue that their approach is different from and superior to traditional paternalism, which imposes the paternalists’ own values on those subject to regulation. Overall, I largely agree with the criticisms of libertarian paternalism in the Cato symposium by Glen Whitman (here and here) and Jonathan Klick. However, I wish to focus on a different weakness of libertarian paternalism: the implicit assumption that voters and government regulators are not subject to serious cognitive biases of their own.
It may well be that private citizens acting in markets and civil society often make decisions that they later regret because of cognitive errors. However, regulators and voters are people too. They also might make bad decisions because of cognitive errors. Libertarian paternalist scholars generally ignore this possibility by implicitly comparing perfectly rational regulators with often irrational consumers. But there is no a priori reason to believe that the former are more rational than the latter.
I. The Cognitive Biases of Regulators.
Indeed, there are good reasons to believe that regulators are likely to be more susceptible to cognitive biases than private sector consumers. This is so for at least three important reasons. First, regulators are making decisions for others, not for themselves. As a result, they have less incentive to get them right. If regulators in the proposed Consumer Financial Protection Agency ban financial products that are of great value to consumers, the regulators themselves won’t suffer (unless they happen to want to purchase those products themselves). The less people have at stake in the decisions they make, the less incentive they have to control their cognitive biases.
Second, we are naturally more ignorant of the preferences of others than our own. Regulators have no reliable way of estimating the benefits that consumers derive from potentially risky products. When making decisions for other people, we are therefore prone to the cognitive bias of assuming that what they “really” want is what we ourselves would prefer in their place. It may, for example, be difficult for a health-conscious upper middle class regulator to believe that a consumer might genuinely prefer the pleasures of eating large numbers of cheeseburgers to the health benefits of a more balanced diet. Thus, he will be likely to put down decisions to consume huge numbers of cheesburgers to consumer “irrationality” and favor paternalistic anti-obesity regulations.
Third, regulators will be making decisions for thousands or even millions of consumers. This requires much greater information and analytical skill than the individual consumer’s task of deciding for himself or perhaps also his family. The more complex the task, the greater the temptation of trying to simplify it with cognitive shortcuts that are prone to bias and may well turn out to be misleading.
II. Voters May be Even Worse.
Of course expert regulators aren’t the only people with influence over paternalistic policies. In a democratic society, voters will have a lot influence too. And, as I have pointed out in previous critiques of libertarian paternalism (see here, here, and here), voters have strong incentives to be both ignorant about public policy and highly irrational in the way they analyze the limited political information they do have. Because the chance that any one voter will influence an electoral outcome is infinitesmally small, most voters have little incentive to either acquire much information about the choices before them or make a strong effort to control the irrational biases they may bring to its evaluation. By contrast, when consumers purchase products in the market, they know that their decisions are decisive and therefore have much stronger incentives to make rational choices.
Once we recognize that voters and regulators are also subject to cognitive biases and that they have only weak incentives to combat those biases, the case for libertarian paternalism is significantly weakened. What I find strange, however, is that prominent libertarian paternalist scholars have paid so little attention to this problem. Cass Sunstein, one of the leading academic advocates of libertarian paternalism, has written some brilliant work on regulatory irrationality in other contexts, including his excellent 2002 book Risk and Reason and an important 1999 article coauthored with economist Timur Kuran.
Lastly, it’s important to note that everything I have said above assumes that voters and regulators designing libertarian paternalistic policies have good intentions; that both are genuinely trying to adopt only those regulations that will help people correct their cognitive biases and more effectively achieve their goals. Once we recognize, as Whitman and I have pointed out elsewhere, that regulatory agencies implementing these policies are subject to interest group “capture” and slippery slope effects, the case for such regulation becomes weaker still.
UPDATE: I know some will argue that regulator and voter biases don’t matter much because libertarian paternalists advocate only noncoercive “nudges” that still leave the final decision up to individual choice. However, as Whitman notes in the Cato symposium and here, they in fact advocate many policies that go well beyond that. Moreover, one important consequence of voter ignorance is that voters are unlikely to make fine-grained distinctions between “libertarian” paternalistic policies and more heavy-handed ones. As a result, libertarian paternalist policymakers may find it very difficult to limit the scope of government intervention to the types of “nudge” policies they initially envisioned.