The recent wave of behavioral economics has led some theorists to advocate the possibility of “libertarian paternalism,” in which regulators designing institutions permit significant individual choice but nonetheless use default rules to “nudge” cognitively biased individuals toward particular salutary choices. In this Article, we add the possibility of a different kind of nudge: temporary law.

The case for temporary law arises from a particular regulatory rationale. In some cases, the best normative defense of regulation against the libertarian critique—the best response to the claim that free market competition produces efficiency—is path dependence, the idea that market institutions can become trapped or locked in to a suboptimal equilibrium, even when some better equilibrium exists. For our purposes, it suffices to define an equilibrium as a behavioral outcome that is stable in this way because no one individual gains by changing behavior given what the other individuals are doing. Some situations allow for multiple equilibria: multiple behavioral patterns that, once reached, are stable. When this is true, there is no reason to expect that the outcome that market competition produces will inevitably be the best one, the global social optimum. Instead, it may be merely a “local maximum.” The outcome that occurs therefore depends arbitrarily on the behavioral starting point; different paths do not all lead to efficiency.

A few legal scholars have explored the relevance of pathdependent legal evolution, particularly in corporate law. The fields of intellectual property and antitrust are also concerned with the path dependence of technological change. But the general literature on regulation has, quite surprisingly, not appreciated the importance of the concept to discussions of market failure, a shortcoming we hope to correct. Most importantly for our purposes, the literature has failed to note this rather surprising implication: temporary law may have a significant advantage over permanent law. When the rationale for regulation is to overcome path dependence, there is no need for a permanent restriction on liberty and there are several critical reasons to make the restriction temporary.

We thus propose imagining regulations that include an expiration date. Our principal example for illustrating these points is the regulation of smoking in public places, a field that has seen substantial change in recent years. Libertarians and other market optimists assert that, in the absence of government regulation, competition among private suppliers would produce the optimal number of nonsmoking establishments—malls, restaurants, bars, apartment buildings. Yet when the government did not regulate, nonsmokers felt that there were an insufficient number of nonsmoking options. In our personal observations across many jurisdictions, there were literally no nonsmoking bars, meaning that there was no good option for nonsmokers.

What does it mean for nonsmokers to complain that, without government intervention, the market is underserving their needs? One possibility is that the only equilibrium consisted of a small number of nonsmoking options because smokers cared much more about the issue than nonsmokers. In other words, if the only choice is smoking, nonsmokers patronize the establishment and grumble, but if the only choice is nonsmoking, smokers stay home and withhold their patronage altogether. This is the intuition of the libertarian, who explains that nonsmokers are being hypocritical because they are not willing to pay sufficiently to induce bars, restaurants, and other establishments to switch to nonsmoking.

Yet there is a second possibility: path dependence. For reasons explored below, rational mechanisms and behavioral biases could have created a situation in which the same set of preferences and levels of wealth permit at least two equilibrium outcomes, one with a high proportion of smoking establishments and the other with a low proportion. In this context, an equilibrium means that no owner of an establishment has any incentive to change the smoking status of the establishment because she is making as much or more profit with the smoking policy she has. If there are multiple equilibria, then it is possible that the low-smoking equilibrium is optimal, and we have reached the high-smoking equilibrium only because of the happenstance that our starting point from decades ago, when preferences and beliefs about smoking were different, involved high smoking rates and near-universal tolerance of smoking. Had history been different, the same preferences (the ones that existed before smoking bans) could have sustained a different and lower level of smoking establishments. If freedom of choice and market competition are consistent with two behavioral patterns, we should want to reach the efficient pattern, not the one that happens to emerge first.

Given path dependence, it may be desirable to use law to shift society from the high-smoking to the low-smoking equilibrium. Across a large domain of issues besides smoking, the best argument that can be made for legal intervention and the most charitable interpretation of the arguments that are made is exactly this point: that the status quo is trapped in an inefficient equilibrium and that law will shift the system to a more desirable equilibrium, one that is also consistent with individual choice to satisfy existing preferences.

The possibility that multiple equilibria exist in a variety of regulatory contexts has never been thoroughly considered. Part of this Article’s contribution is to identify a list of mechanisms that can produce multiple equilibria. Our main point, however, is to explain why, when multiple equilibria exist, the best response is usually a temporary law. If the problem is path dependence, a temporary law will often be both necessary and sufficient to move behavior to the more efficient outcome. For example, suppose the status quo among a city’s restaurants is a high-smoking equilibrium (95 percent permit smoking) and we believe there is a more efficient low-smoking equilibrium (10 percent permit smoking). If the temporary law bans smoking in all restaurants for a certain time period (say, two years), it pushes toward a zero-smoking outcome. When the law is removed, restaurant owners will decide whether to allow smoking again; many will. But the implication of there being a lowsmoking equilibrium is that the number of restaurants allowing smoking will rise from zero to the number the low-smoking equilibrium represents (10 percent) and then stop. In short, the concept of path dependence identifies the importance of arbitrary starting points; temporary law offers a new “starting” point, resetting the system to allow the emergence of the equilibrium with the lowest smoking levels.

It should be immediately apparent that the temporary law cannot be a first-best solution. The first-best solution would be to move directly to the more efficient equilibrium. In the smoking example, part of the cost of the temporary ban is the inefficiency of having too few restaurants—zero—that allow smoking during the period the law is in effect. If the efficient low-smoking equilibrium is that 10 percent of restaurants allow smoking, then the state could just create licenses equal to 10 percent of the restaurants and allocate them by auction or lottery, enforcing a ban against only unlicensed restaurants. If this is the situation, there is no advantage to making the law temporary, as a permanent law merely requires people to do what they already want to do in equilibrium.

The problem, however, is that this first-best, direct solution demands costly or unobtainable information. We might have no good way of estimating the exact location of the low-smoking equilibrium. And here we see the possible advantage of a temporary law. If we are uncertain what the low-smoking equilibrium is—perhaps it is 10 percent of restaurants, but it could be as high as 35 percent or as low as 5 percent—we will likely grant too many or too few licenses, thereby forcing an inefficient level of smoking indefinitely. With the temporary law, the short-term inefficiency is likely greater—requiring 0 percent smoking restaurants is too low—but lasts for only a limited period, after which voluntary exchange produces the low-smoking equilibrium.

This revelation—what might be called “equilibrium location”—is only the first informational advantage of temporary law. For the second, assume there is also uncertainty or ambiguity about the entire situation just described. There may be multiple equilibria, but there is also some chance that there is really only one behavioral pattern consistent with existing preferences and free exchange. If so, the libertarian has a good reason to assert that the status quo already represents the efficient outcome. The licensing scheme then imposes a severe inefficiency (for example, 10 percent of restaurants are smoking when 95 percent is efficient) for an indefinite time. It also offers no mechanism for revealing whether the licensing scheme represents a suboptimal outcome. But when the temporary law expires, if the premise on which it was based were false, and there were only one equilibrium, then restaurants would return to their initial level of permitting smoking (95 percent). We will then learn that there was no market failure to be solved. Thus, temporary law works like an experiment. The information it reveals is both equilibrium location and what might be called “equilibrium verification.”