Learning through policy variation.

• Author: Listokin, Yair

ARTICLE CONTENTS INTRODUCTION I. THEORIES OF PUBLIC POLICYMAKING II. CHOOSING POLICIES FROM AN OPTIMAL SEARCH PERSPECTIVE: THEORY AND APPLICATIONS A. The Theory of Optimal Search B. Factors Influencing the Optimal Degree of Policy Variance C. The Optimal Search Approach: Applications 1. Choosing Between Reversible Regulations 2. The Optimal Search Approach and Penalty Default Rules in Contract Law 3. Increasing Shareholder Democracy in Corporate Law 4. Federalism and the Optimal Search Approach III. THE OPTIMAL SEARCH APPROACH: OBJECTIONS, RESPONSES, AND MODIFICATIONS A. Burkean Objections B. The Costs of Changing Policies C. Irreversibility, Real Options, and the Optimal Search Approach 1. The Real Options Approach to Policymaking 2. Burkeanism, the Precautionary Principle, and the Real Options Approach 3. Sticky but Reversible Policies D. Optimal Variance in Different Policymaking Contexts E. Reversibility and Institutional Design 1. Sunset Clauses and Legislative Entrenchment 2. Separation of Powers 3. Stare Decisis IV. PUBLIC POLICYMAKING INCENTIVES AND THE OPTIMAL SEARCH APPROACH A. High-Variance Policies and Reelection/Reappointment Incentives B. High-Variance Policies and Incentives for Political Advancement C. Federalism and Incentives To Innovate V. IRREVERSIBILITY AND POLITICAL INCENTIVES: APPLICATIONS AND RECOMMENDATIONS A. Reversible Regulations B. Contract Default Rules and Increased Judicial Policymaking C. Shareholder Power D. Federalism and Preemption E. Other Sources of Variation: Direct Experimentation CONCLUSION INTRODUCTION

How should policymakers choose laws and regulations when outcomes are uncertain? The answer initially seems simple: they should choose the best policies--the ones with the highest average payoffs along some metric. Burkeans have a different view. They are skeptical of human ability to divine the best policies. Instead of encouraging policymakers to choose the policies that seem best, Burkeans urge policymakers to choose policies that change the status quo incrementally rather than drastically. When policymakers can learn from the results of their laws and make changes, however, both the common sense position--choose the best policy--and the Burkean position--choose new policies cautiously and incrementally--are often wrong.

When learning is possible, (1) innovative high-risk policies with lower average outcomes but the potential for greater outcomes become preferable. (2) If a high-risk policy proves a failure, then the policy can be changed, and the policy with the highest average payoff can be pursued. If the policy succeeds, then policymakers will have achieved an ideal outcome and will no longer need to search for alternatives. Learning allows policymakers to limit the downside of a high-risk policy but still enjoy the upside, making a high-risk policy with a lower average payoff a better initial choice in many cases than a low-risk policy with a higher average payoff.

In other words, policies serve two functions. Their primary function is to achieve some outcome in the current period. But the information provided by observing a policy's outcome also assists the search for better policies for the future. And the best policy from a search perspective often differs from the best policy for the current period. The "optimal search" for a policy seeks an excellent policy that will enable policymakers to end the policy search. Thus, optimal search theory favors high-variance policies, because variance increases the probability of finding excellent policies. The average outcome of a policy matters less from an optimal search perspective than the upside of a policy because it is unlikely that a reasonable but suboptimal outcome will end the search for a good policy. A bad policy, moreover, can be changed in the next period.

The economics literature on optimal search focuses on the choice between two new possibilities. Policymaking, however, often involves the choice between a new policy and the status quo. Generally, new policies have higher variance in outcomes than existing policies. The optimal search approach, therefore, suggests that new policies should be implemented frequently. Even when choosing between alternative untried policies, the policy with greater variance is the better choice, other things equal.

The implications for public policymakers are wide ranging. Take contract law, for example. A number of scholars have proposed complex rules that aim to improve the status quo. (3) Critics argue, however, that the efficiency of the rules depends on questionable assumptions about individual behavior. (4) Given these defects, most of these proposals have never been tried. The optimal search approach, however, suggests that these policies should be tested. If the alternative default rules fail to improve social welfare, (5) then the policies can be discarded. If the default rules succeed, however, then policymakers will have achieved a significant, long-lasting improvement in the quality of law.

Corporate law involves similar disputes. It typically changes little, and every proposed change is met by critics who praise the status quo. The critics may be right that the status quo is better on average than are proposed reforms, but the optimal search approach suggests that even if the critics are right, new policies should be tried so long as they have some chance of constituting a significant improvement over the status quo.

Reversible regulations provide a third area where the optimal search approach illuminates policymaking. At present, debates on regulation are between those who favor the cost-benefit approach, which advocates choosing the best law on average, and those who favor the precautionary principle, which advocates caution in many regulatory choices. The best risk regulation from an optimal search perspective constitutes a modification of cost-benefit analysis that contrasts with the precautionary principle even more sharply than conventional cost-benefit analysis.

The optimal search approach also buttresses the argument of those advocating for policy experimentation at the state level and against wide-ranging federal preemption of state laws. In a federalist system, policy variance becomes even more desirable than in a national system as the learning benefits of variance are shared through space as well as through time. While nationally applicable policies will often maximize per-period outcomes, the preemption that often accompanies these laws stifles learning through variation to a degree underappreciated even by those who argue that states are the laboratories of democracy. As a result, the optimal search approach favors extremely limited preemption of state law.

While illuminating, the optimal search approach, which favors high-variance policies, relies on a number of assumptions. It assumes that policymakers can learn from their laws and that they can change these laws in response to their learning. When policies have irreversible effects, the benefits of variance in policies are greatly reduced. Indeed, when new policies are irreversible, the dynamic analysis emphasized by the optimal search approach indicates that variance is no longer positive, or even neutral, but rather negative. (6) Burkean approaches thus have continued salience for policymakers when policy is examined in a dynamic context because Burkean approaches are optimal when policies are irreversible. Similarly, expected-value maximization rationalist approaches, which ignore variance, become more attractive for policymakers when policies are sticky but reversible, as the learning benefits of high variance and the flexibility benefits of low variance partially offset each other. In total, the choice of optimal policies depends critically on policy reversibility.

Policy reversibility has two sources. Some degree of irreversibility is inherent in all policies, while other sources of irreversibility arise from policymaking institutions. To gain the maximum benefits of the optimal search approach, this Article recommends institutional mechanisms that maximize reversibility, such as sunset clauses, unicameral legislatures, and a reduced emphasis on stare decisis.

In both federalist and uni-jurisdictional settings, the optimal search approach assumes that policymakers aim to maximize social welfare. The Article later relaxes this assumption and examines public policymakers' incentives to innovate in uni-jurisdictional and federalist contexts. Because policymakers' incentives to innovate are often lower than optimal in each context, several recommendations, such as subsidizing federalist innovation or emphasizing innovation in contexts with electoral insulation, may be justified.

This Article is organized as follows. Part I summarizes the Burkean, classical liberal/rationalist, and "experimentalist" approaches to public policymaking. Part II develops the optimal search perspective and demonstrates that high-risk policies with relatively low average outcomes often should be instituted before policies that other policymaking rationales classify as superior. Additionally, Part II examines how this optimal search insight relates to different variables, such as the choice of discount rates and the time required to evaluate a policy. It also develops the optimal search approach through idealized applications to risk regulation, contract law, corporate law, and federalism.

The remaining Parts of the Article address impediments to learning through policy variation via the optimal search process. Part III evaluates Burkean and rationalist objections to the optimal search perspective and discusses how the optimal search approach applies to policymaking and institutional design when the effects of policy changes are irreversible. Part IV relaxes the assumption that policymakers pursue socially beneficial policies and examines the optimal search approach from a public choice perspective...