The Expressive Powers of Law: Theories and Limits 9780674735965

Citation preview

The Expressive Powers of L aw

The Expressive Powers of Law Theories and Limits Richard H. McAdams

Cambridge, Massachusetts London, England 2015

Copyright © 2015 by the President and Fellows of Harvard College All rights reserved Printed in the United States of America First Printing Library of Congress Cataloging-­in-­Publication Data McAdams, Richard H., author. The expressive powers of law: theories and limits / Richard H. McAdams. pages cm ISBN 978-­0 -­674-­04692-­4 1. Law—​Language.  2. Law—​Methodology.  3. Expression.  4. Rhetoric.  I. Title. K213.M425 2014 340'.14—​dc23 2014014454

For Anna

Contents

List of Figures

Introduction: Alternatives to Deterrence and Legitimacy

ix

1

1 Expressive Claims about Law

11

2 The Focal Point Power of Expression

22

3 Law as Focal Point

57

4 Law’s Focal Power in Dynamic Perspective

94

5 Legislation as Information

136

6 Revelation of Information by Legal Enforcement

169

7 The Power of Arbitral Expression

199

8 Normative Implications

233

Conclusion: Law’s Expressive Powers

260

Notes

263

Acknowledgments

315

Index

317

Figures

1.1 Categories of Expressive Claims about Law

13

2.1 The Classic Prisoners’ Dilemma

30

2.2 The Prisoners’ Alibi Game of Pure Coordination

32

2.3 A Prisoner’s Assurance Game

33

2.4 A Prisoner’s Battle of the Sexes Game

35

2.5 A Prisoner’s Hawk-­Dove Game

36

2.6 A Generic Symmetric 2×2 Game

39

2.7 PD with Embedded BOS Game

40

2.8 The BOS Intersection Game

46

2.9 Another Hawk-­Dove Game

55

4.1 A Public Good Game as PD

130

4.2 A Public Good Game as Assurance

130

7.1 Another Battle of the Sexes Game

205

7.2 A Final Hawk-­Dove Game

209

The Expressive Powers of L aw

Introduction Alternatives to Deterrence and Legitimacy

I

n the United States, traffic law requires adherence to the yield sign and the solid center line. These are the sort of prosaic traffic rules one learns when studying for a driver’s license. The yield sign requires slowing down or stopping to give way to drivers on the other road or lane. The solid center line on a two-­lane road forbids a motorist from crossing the line to pass another vehicle; the dashed line permits passing. Frequently, there are two center lines and each indicates the passing rule for a different lane. Basic traffic rules of this kind are so mundane that they rarely make an appearance in the grand theoretical discussion of why people obey the law. Yet there is much to be learned by asking what motivates drivers to comply with these rules (to the extent they do). When a person is in a hurry, and would prefer the other driver to yield, or would like to pass a slower moving vehicle on a two-­lane road (and the oncoming lane appears to be clear), why forgo the opportunity to proceed first or to pass? Why comply with the law? No doubt, many people comply with many traffic rules out of habit. But habit is only a proximate rather than ultimate explanation. First, people usually act out of habit only when there is no great advantage or possibly no time to reconsider their habitual action. Given enough construction delay, for instance, drivers will rethink and abandon their habitual routes. Yet drivers sometimes obey traffic rules when there is an apparent advantage to disobedience and an occasion to reflect. When the driver is stuck behind a slower vehicle, there is time to consciously consider whether to 1

The Expressive Powers of Law

obey the law and a motivation to disobey—​to minimize the delay from not passing. Why obey law in this context? More fundamentally, it takes time to acquire a habit, so the explanation only postpones the deeper question of why people bother to comply before they acquired the habit of obeying a particular law. Presumably, newly licensed drivers have not yet complied for a sufficient time to have developed the habit of complying with the yield sign or solid center line, yet they will never develop the habit if they do not, at first, consciously decide to comply with the rule. Why do they do so? Before I identify the expressive explanations that are the central focus of this book, let us consider the two conventional accounts of legal compliance. Economists famously emphasize deterrence, that legal sanctions change the costs of behavior, making compliance cheaper than noncompliance. The prototypical example is the deterrence of criminal punishment, but economists focus on the deterrent effect of monetary damages throughout the law. The literature is vast.1 When I have made this point to legal scholars, I sometimes hear the response that economic analysis has long recognized a facilitative or enabling role for law. There are many nonmandatory rules in the law, such as contract default rules or the rules defining the consequences of choosing certain business organizations or familial relationships (e.g., to be a partner or adoptive parent). Here the law offers ready-­made devices for achieving certain ends without using legal sanctions to compel their selection. Yet even when one can choose to opt out of particular legal rules, or not to opt into them, the question I am asking is why the rules—​once adopted—​ affect behavior. The economic answer is implicit but obvious: because the government stands ready to use legal sanctions to enforce such rules against anyone who consents to them. By the economic logic, the reason that contractual default terms facilitate commercial projects is that courts enforce those terms against contracting parties who fail to opt out. Otherwise, “opting in” would have no bite. Returning to the traffic example, deterrence theory posits that the new driver complies with the yield sign and the rule against passing over a solid center line, even before developing the habit of complying, in order to avoid getting a ticket (a fine), which might also have the effect of raising her insurance rates. For the experienced driver, the fear of legal sanctions backs up the habit, motivating compliance even when the annoyance of the yield sign or solid center line causes the driver to consciously consider 2

IN T RO D U C T ION

violating the rule. To a lesser degree, economists also discuss the incapacitation effects of legal sanctions, as where incarceration makes it physically impossible for inmates to commit certain crimes (e.g., bank robbery).2 Those in fear of losing their license—​knowing that the sanctions for driving without a license would deter them from driving—​might comply for that reason. In both cases, however, the economist emphasizes legal sanctions. On this view, the law matters only because the legal sanctions matter. The second conventional account of legal compliance is legitimacy. Max Weber said that a person’s actions might be guided by “the belief in the existence of a legitimate order,” such that “its violation would be abhorrent to his sense of duty (of course, in varying degrees).”3 Indeed, Weber claimed that legitimacy is a more stable source of order, compared to self­interest and habit, because legitimate order “enjoys the prestige of being considered binding.”4 “[T]he most common form of legitimacy,” Weber observed, “is the belief in legality, the compliance with enactments which are formally correct and which have been made in the accustomed manner.”5 Many contemporary legal scholars share this view.6 A standard claim of legal psychology is that “[p]eople are more likely to obey the law when they view the law generally as a legitimate moral authority.”7 When law merely reflects an existing moral consensus, telling people to do what they already feel obligated to do, then it might have no independent effect on behavior. The moral consensus might motivate the behavior, not the law reflecting the consensus. Yet on more contested matters, with no clear social consensus, law might be able to leverage its legitimacy to persuade members of the public to change their moral view, thus affecting their behavior.8 There is a lively contemporary exchange on the sources of legitimacy. The psychologist Tom Tyler emphasizes procedural sources, finding evidence that people are more likely to obey the law (and cooperate with law enforcement) if they perceive that courts and police treat them fairly and with respect, more generally, if the legal processes are fair.9 By contrast, John Darley, Janice Nadler, and Paul Robinson emphasize the substantive sources of law’s legitimacy, which depend on how well or badly the legal rules and outcomes align with the public’s moral intuitions.10 Regardless of its source, where there is legal legitimacy, people are more likely to have internalized a preference—​unreflectively or consciously perceived as a 3

The Expressive Powers of Law

moral obligation—​to obey the law. And that preference or obligation generates greater compliance. In our example, the driver obeys the yield sign and the center line’s ban on passing, because she perceives the government’s traffic rules, or law generally, as legitimate and worthy of obedience. To some degree, social scientists endorse more than one theory. Psychol­ ogists and sociologists do not invariably deny the existence of deterrence.11 Nor do all economists ignore the role that legal legitimacy plays in compliance. Relevant here is evidence that the perception of fair tax procedures or fair tax burdens explains much of the compliance with tax laws.12 The economists Raymond Fisman and Edward Miguel generalize the point in a striking way.13 They looked at the compliance with New York parking rules by United Nations diplomats. Because of diplomatic immunity, these individuals faced absolutely no threat of legal sanctions for parking violations (at the time; there is now a voluntary agreement to submit to sanctions), and yet there was enormous variation in compliance rates among diplomats from different nations, with some never running afoul of the law. Fisman and Miguel found that the greater the corruption in a diplomat’s home country, which plausibly means the lower the legal legitimacy that diplomat experienced before moving to New York, the greater the diplomat’s violation of New York parking laws. So the economists identify some role for legitimacy or at least some role for legal influence not dependent on legal sanctions. Nonetheless, this kind of theoretical pluralism is distressingly rare. The main drama of the empirical study of legal compliance is a long-­running conflict between the social sciences, a battle between the rival hypotheses of deterrence and legitimacy.14 The dominant struggle diverts our attention away from the possibility of other explanations. The result is unfortunate because legal compliance is a matter of fundamental concern. We often want more compliance than we have. If the issue were less important, we might be content to know (if it were true) that sanctions and/or legitimacy generate most of the legal compliance we observe without worrying about what generates the rest. Yet because compliance is of paramount concern, we should seek to understand all the causal mechanisms that produce it. I hope to demonstrate that, in some contexts, an alternative, expressive mechanism plausibly causes more of the compliance we observe than deterrence or legitimacy. But primarily I want to identify the expressive mechanisms so we can begin the work of empirically isolating their effect. 4

IN T RO D U C T ION

Return to the driver, the yield sign, and the center line. It should be obvious that there is more going on with compliance than fear of legal sanctions or deference to legitimate authority. An overwhelming motivation of drivers is to avoid automobile accidents either from colliding with other drivers or running off the road. Complying with traffic rules offers drivers a way to avoid these accidents. In two respects, the law’s expression creates an incentive for compliance. The first expressive power is what I call law’s coordinating function. Driving is a situation in which individuals seek order. Avoiding a collision is a matter of coordinating one’s movements with those of the other motorists to avoid driving one’s car into a space at the same time it is occupied by another car. Traffic law facilitates this coordination when it specifies an orderly means of driving, a set of priority rules. The yield sign is one example (as are stop signs, traffic lights, one-­way signs, etc.). When two motorists wish to drive across each other’s path or to occupy the same lane, the law offers a means of avoiding a collision when it expresses the rule that one driver is to yield to the other. Because each driver has an incentive to coordinate and there is no other obvious means to do so, the government’s proposed solution possesses a natural attraction, a power of suggestion. The driver told to yield is less likely to expect the other driver to yield; if the second driver is not going to yield, the first prefers to yield, so as to avoid a collision. Because compliance is the most obvious way to avoid a collision, the law is, to some degree, self-­enforcing. In simple game theory terms, legal expression provides a “focal point” that solves the coordination problem. The center line works, in part, in the same way. The law says that motorists should stay to one side of the road, in one’s own lane. By marking the road with a center line, the state creates the different lanes, giving clear meaning to “one side of the road.” The clarity makes it easier to comply with the legal requirements, which drivers wish to do to avoid an accident. When drivers approach a hill, curve, or other obstruction, they have a limited view of oncoming traffic. By the time the drivers see one another, and realize that they are on a path to a collision or sideswipe, there may not be time to safely avoid each other, given that swerving presents its own dangers. The center line offers a simple solution, a focal point. If each motorist stays on her side of the line, they will pass each other without incident. When a driver is speeding around a curve so that centrifugal forces edge the car towards the other lane, the line gives immediate feedback on how 5

The Expressive Powers of Law

far it is safe to venture in that direction and at what point one risks disaster. Again, by offering the drivers a means of coordinating, a mode of order, the traffic rule is (to some degree) self-­enforcing. Yet the solid center line is interesting for an entirely different reason, which brings us to the law’s second expressive power. Law also has an information function. The fact that the government allows motorists to pass on some parts of a two-­lane road but prohibits passing on other parts is itself information. Unless one believes that governmental agents are completely arbitrary or perverse, a reasonable inference is that the bureaucrats in charge of road safety believe that passing is relatively dangerous on the part of the road where it is prohibited. These traffic engineers are in a position to know in detail the road’s grade and curvature and other obstacles that determine the probability of an accident while passing. Drivers should update their beliefs about the safety of passing based on the fact that the law, in this location, prohibits passing. Indeed, at night or in other situations of limited visibility, drivers not already familiar with the road may have almost no basis for estimating the risks other than the existence of this prohibition. As a result of these inferences from the law’s existence, the desire for self-­preservation creates an incentive not to pass, which is to comply with the legal prohibition. To some degree, informational updating makes the law self-­enforcing. Thus, legal expression has at least two effects that generate compliance: coordination and information. One might think that coordination is also about information, but it is not, at least not in the same way. The yield sign need not reveal any information about the physical circumstances of the driving situation. Given two equally sized merging roads, for example, one could put the yield sign on either road. If the choice is arbitrary, the fact that it is on one road rather than the other does not imply any facts about the physical situation. Instead, a driver makes an inference not about the physical situation but about how the other driver will behave. By contrast, the placement of a sign prohibiting passing is not arbitrary; traffic engineers select the locations based on the road conditions—​the angle of the curve, tilt of the road, obstructions to vision—​that determine the risks of passing, so the important inference is about those physical circumstances that determine risk, not about how other drivers will behave. What these examples demonstrate is that law has expressive powers independent of the legal sanctions threatened on violators and independent of the legitimacy the population perceives in the authority creating and 6

IN T RO D U C T ION

enforcing the law. That is the central thesis of this book. My aim is to describe and explore these two largely overlooked causal mechanisms—​ coordination and information—​by which legal expression influences behavior, usually in the direction of compliance. Using rational choice assumptions, I hope to convince economic thinkers that we must amend the conventional wisdom of legal compliance. Law deters and incapacitates, but it also coordinates and informs. As part of my effort to persuade the economist to inquire systematically about law’s expressive effects, I even show that legal sanctions owe their power entirely to the law’s ability to facilitate coordination expressively. I also seek to convince the legitimacy theorist that the law’s behavioral effects not attributable to deterrence or incapacitation cannot necessarily be attributed to law’s legitimacy or moral authority. For example, when we observe tribunals successfully resolving disputes despite lacking any power to sanction the disputing parties (even indirectly), we can no longer assume this is evidence of legitimacy, because the tribunal’s influence—​the reason the declared loser concedes—​may be due instead to its expressive powers. The expressive theories I offer are not only rivals to these conventional theories of legal compliance; they are also supplements. Deterrence and legitimacy turn out to be more potent because of their interactions with law’s expressive powers. As should be evident from this claim, I do not seek to repudiate or depose deterrence or legitimacy theory. I am instead advocating a theoretical pluralism about compliance, the proposition that law brings to bear multiple powers at the same time.15 I criticize alternate theories only because and to the degree it is necessary to recognize the distinct power of law’s coordination and information powers, to show where these theories provide the best explanation of compliance. Finally, I work to persuade many legal commentators to be less exuberant and more cautious in making expressive claims. There may be a rhetorical advantage to defending or attacking a law or legal action by saying it will send a message of the right or wrong sort, but the indiscriminate assertions of expressive consequences lowers the apparent value of all such analysis. The theories I offer entail clear limits to the plausibility of expressive claims, which should helpfully constrain expressive claims to those worthy of sustained attention and empirical testing. In short, deterrence and legitimacy dominate the social science discussion of law’s effect on behavior, while other important mechanisms of influence are neglected, an omission I hope to correct. I find the law’s 7

The Expressive Powers of Law

expressive powers at work in constitutional and international law; property and contract disputes; criminal punishment; the regulation of smoking, voting, and driving; race and sex discrimination; the historic success of informal tribunals lacking the power of sanctions; the relationship between law and social movements; and the legal codification of custom. The collection of examples hopefully carries the thesis more convincingly than any one example can. These assertions and the traffic example raise a great many questions and objections. At a general level, what are the conditions necessary for law to have either a coordinating or informational effect? How commonly do these conditions arise and why do they permit so much noncompliance to remain? And how do the two expressive mechanisms I describe compare to other expressive claims in the existing literature? The book addresses these questions and many more in developing the expressive theories of law. After this introduction, the book proceeds in the following order. Chap­ ­ter 1 places the argument of this book in the context of the various discussions about the expressive dimension of law. I identify two positive and two normative branches of the scholarly literature on legal expression. The book primarily concerns the first branch of the literature, though I engage all four. The next three chapters (2, 3, and 4) describe and elaborate the claim that legal expression can provide a coordinating focal point for behavior. Chapter 2 lays out the elementary game theory of coordination and focal points, which identify a situation in which “mere” expression can influence behavior. Chapter 3 applies this theory to law, identifying the circumstances where legal expression works by creating a focal point around which individuals coordinate. Through a series of examples, this chapter shows the large domain in which law has this effect. Both chapters 2 and 3 review the relevant experimental literature on the expressive construction of focal points in situations of coordination. The empiricism here is in the early stages; it does not definitively prove the extent of the law’s coordinating function. As I said previously, I want to identify the expressive mechanisms, to render them plausible in a large array of legal contexts, so as to justify further empirical study of their effect. Chapter 4 further extends the theory and applications by considering the law’s focal point power in a more dynamic setting. When the law seeks to change behavior, it often faces the challenge of competing with an 8

IN T RO D U C T ION

existing social norm, custom, or convention, which also operates as a focal point. I show how law’s focal point power is nonetheless important when social movements are already unsettling an extant convention or where the law seeks only to clarify some ambiguity in a custom. The following two chapters (5 and 6) describe and elaborate the second expressive mechanism: that legal expression can convey or “signal” information, which affects beliefs and behavior. In both cases, the behavioral effects are largely, but in important ways not entirely, in the direction of legal compliance. Chapter 5 focuses on two types of information revealed in legislation: (1) the current state of public attitudes and (2) some collective evaluation of the risks or rewards of regulated behavior, either of which can cause people to update their beliefs and change their behavior. Chapter 6 discusses the informational effects of judicial and executive enforcement, with a particular emphasis on criminal law. Again, I note the relevant empirical studies along the way, but the larger goal is to render the theory sufficiently plausible and clear so as to identify, as I do, a variety of testable implications. Chapter 7 discusses the power of arbitral expression, particularly the ability of arbiters to resolve disputes without wielding the power of sanctions or legitimacy over the disputants. In history and around the world, there have been many successful tribunals without enforcement powers. I contend that the expressive theories offer a useful way to understand this noncoercive dispute resolution and that my account works better as an explanation than legitimacy theory. This chapter stands apart from the preceding ones because only here do I combine the coordination and infor­ ­mation theories, finding a synergy between them when an arbiter declares how a dispute should or must be resolved. In sum, most of this book explicates law’s function in providing coordinating focal points and information, functions I aim to place alongside deterrence, incapacitation, and legitimacy (henceforth, I will drop the reference to incapacitation, which is mostly limited to certain criminal punishments and is therefore less general than deterrence or legitimacy). The bulk of the book is therefore positive, focusing on the effects of law’s expression. But Chapter 8 is normative. It discusses the implications of the law’s expressive effects for both the optimal use of resources and the structure of legal doctrines that resolve expressive disputes. For example, the power of law as a coordinating focal point offers a new advantage of rules over standards; where coordination is required, rules may be specific 9

The Expressive Powers of Law

enough to align expectations when standards are not. As another example, the power of law to reveal information offers important insights into the need for the Establishment Clause, as we can now understand how the government’s symbolic endorsement of religion can have behavioral effects on religious practice. In general, compliance is a central issue for law; expanding our understanding of the mechanisms generating compliance produces a variety of normative insights for law.

10

1 Expressive Claims about Law

L

egal scholars and political actors make broad claims about the expressive dimension of statutes, judicial opinions, prosecutorial decisions, jury verdicts, and criminal sentencing decisions. Expressive theories span topics as diverse as rape shield laws, property law, patents, the regulation of CEO compensation and corporate directors’ duties to creditors, the legal concepts of the appearance of impropriety and standing to sue, cyber­stalking regulation, the fourth amendment exclusionary rule, international law, and bank regulation.1 The nature of the claims varies greatly. Commentators say that legal change will “send a message” of some sort, for example, that legalizing medicinal marijuana will convey to teenagers the harmlessness of smoking pot, that strengthening drunk driving or domestic violence laws will articulate the grave harm those behaviors cause, and that a jury verdict in favor of a rich but undeserving defendant will communicate a message of unequal justice. Some legal scholars claim that people comply with certain laws, such as seat belt mandates and smoking restrictions, because of the expressive (or symbolic or educative) effect of these rules.2 Or that the law can work by changing the “social meaning” of a behavior, as where the historic introduction of a law against dueling created a new meaning to the decision to refuse a duel: not that one was a coward without honor, but that one felt an honorable duty to obey the law.3 Legal scholars also offer to explain the existence of certain laws by their symbolism. For example, there may be little or no behavioral effect from the legislative repeal of constitutionally unenforceable segregation or 11

The Expressive Powers of Law

sodomy laws, from laws recognizing English as the “official” language, or from local ordinances declaring a nuclear-­free zone, but voters may demand and support such laws for the values they are understood to express. Even when a law has some behavioral effect, the politics of its enactment may be dominated by its symbolic importance. Pro-­life voters may like what a law against “partial-­birth abortion” expresses even if the law, by permitting other abortion procedures, has no effect on the total number of abortions.4 Commentators explain legal rules of market inalienability such as the ban on selling human organs, sexual services, or electoral votes by the public’s desire to express the incommensurability or pricelessness of certain values.5 There are also many normative claims about legal expression. Some constitutional theorists claim that the Equal Protection Clause of the Fourteenth Amendment should be read to forbid laws that express the inferiority or subordination of a racial group or sex, because such laws create “expressive harms” regardless of whether there are further consequences from the law.6 The Supreme Court arguably took an expressive stance when it ruled that the clause prohibits racial gerrymandering that creates bizarrely shaped electoral districts because it sends a “message” that “members of the same racial group . . . ​think alike, share the same political interests, and will prefer the same candidates at the polls,” and when it ruled that the clause prohibits public single-­sex higher education and gender-­ based jury selection because they reinforce gender stereotypes.7 Not surprisingly, the meaning of the term “expressive” is not constant across the scores of articles discussing all these expressive claims. What might at first appear to be a single legal literature about the expressive theory of law is really a set of distinct literatures employing the same term. There is no grand unified “expressive theory” that encompasses all of these literatures. This observation is not a criticism of any of the expressive literatures. Instead, the word “expressive” has enough flexibility to cover an array of inquiries. One can believe in the value of each inquiry without thinking that the inquiries are essentially the same. Thus, I have no interest in trying to limit the use of the word “expressive” to the use I make of it and no quarrel with legal scholars describing their theories as such even though they differ from the ones that concern me. (Indeed, if the reader is unable to tolerate my using the term expressive in this book as I do, please feel free to imagine a different term, perhaps “communicative” or “educative.”). 12

EXPRESSIVE C L AI M S A B OU T L AW

The contrary assumption—​that there is a single field of “expressive law”—​has generated confusion. Without a clear understanding of the distinct projects using the term “expressive,” there is a tendency to lump together articles that address different topics. To avoid further confusion, I am at pains to observe that this book concerns, at least primarily, only some of the expressive law literatures. To explain, I must distinguish the various expressive literatures, which I do by offering a simple typology.

Four Types of Expressive Claims about Law I count four categories of expressive claims in the legal literature: (1) that law influences beliefs, emotions, or behavior by what it expresses, an expressive theory of law’s effects; (2) that expressive politics determine the content of law, an expressive-­politics theory of law; (3) that the normative status of law depends on its meaning, a normative theory of expressive law; and (4) that the normative status of the private behavior the law regulates depends on its meaning, a normative theory of expressive conduct.8 Only the first of these categories (the one in italics in Figure 1.1) is the main subject of this book. The coordination and information theories are expressive theories of law’s effects (category 1). Although I emphasize behavioral consequences throughout, there are other consequences. In my behavioral theories, for example, legal expression first changes beliefs, which in turn changes behavior. Yet belief change by itself is an expressive consequence. Indeed, perhaps the most obvious expressive effect is the emotional reaction to the beliefs the law inspires. If one subjectively feels respected by the law, that gain is an expressive consequence. If one feels disrespected, that loss is an expressive harm. I shall occasionally refer separately to these reactive emotions, but the main event here is behavior. Now let’s consider the other three categories. A prominent example of an expressive-­politics theory of law (category 2) is Joseph Gusfield’s explanation Positive Claims

1: expressive theories of law’s effects

2: expressive-­politics theory of law

Normative Claims

3: normative theory of expressive law

4: normative theory of expressive conduct

F i g u r e 1.1  

Categories of Expressive Claims About Law

13

The Expressive Powers of Law

of American Prohibition.9 A simple causal theory might say that Americans adopted Prohibition because they thought it would alleviate the social and medical problems of alcoholism. More pessimistically, one might posit that the law was merely “therapeutic” in the sense that people demanded “something be done” about the problems and there were no good alternatives. Gusfield claims instead that in the early twentieth century, alcohol consumption came to be associated with new, mostly urban and Catholic, immigrant groups, so the prohibition of alcohol worked to express disapproval of those groups. The law’s advocates—​descendants from earlier waves of immigration, mostly rural Protestants—​sought this disapproving expression, valuing it as a symbol of their higher status. Thus, on Gusfield’s account, law is a symbol over which political groups struggle; it is the meaning the competing groups attach to the legal symbol that explains what law is adopted.10 He explains: “We have always understood the desire to defend fortune. We should also understand there is the desire to defend respect.”11 The work of Dan Kahan and Don Braman generalizes this result. Their “cultural cognition” theory suggests that symbolic effects are pervasive in the regulation of risk.12 Because social groups view regulation as a political test of their cultural values, the enactment of handgun regulation or energy conservation standards, the ban on extra large sizes of sugary drinks, or the mandate of an HPV vaccine all depend on what those laws mean to the dominant political groups. There being symbolic competition among social groups, the members of one group will favor the laws they perceive as expressing their social standing and oppose laws the they perceive as expressing the social standing of other groups. Thus, Kahan and Braman, like Gusfield, treat law as the dependent variable and use legal meaning as the independent variable that explains why a particular law exists. By contrast, my (category 1) expressive theories treat behavior as the dependent variable and use legal expression as the independent variable that explains why a particular behavior (especially compliance) exists. The causal claims do not constitute normative judgments; it might be good or bad that law influences behavior expressively and good or bad that the political struggles over meaning influence the enactment of law. Of course, for a consequentialist, it is easy to argue that expressive effects are good when they generate more compliance with legal rules that promote welfare, or that expressive political struggles are sterile and wasteful. 14

EXPRESSIVE C L AI M S A B OU T L AW

Yet the positive claims being made do not strictly entail any normative commitment. By contrast, an example of a normative theory of expressive law (category 3) is the work of Richard Pildes and Elizabeth Anderson, who propose to evaluate law by whether it expresses appropriate equal respect for individuals, regardless of consequences.13 This view favors interpreting the Equal Protection Clause of the Fourteenth Amendment as condemning laws that express superiority or inferiority of a class of individuals and favors interpreting the Establishment Clause of the First Amendment as condemning state action that expresses an endorsement of religion or some particular religion.14 Another example of a category 3 theory is Joel Feinberg’s famous article about punishment.15 Feinberg defines punishment as requiring hard treatment (that an authority imposes on an individual on account of her rule violation) that carries with it a certain expression, namely, “the expression of attitudes of resentment and indignation, and of judgments of disapproval and reprobation.” Given that understanding, Feinberg makes a normative claim, which is that proportionality of punishment should mean that the level of expressive condemnation, rather than the harsh treatment used to express condemnation, is proportionate to the wrong. Thus, one can normatively evaluate a legal scheme for punishment based on how proportionately it expresses the wrong being punished. Although my category 1 theories are positive, not normative, Chapter 6 discusses Feinberg’s account of punishment and Chapter 8 more generally explores the normative implications of the coordination and information theories. I do not there offer an expressive normative theory; I instead use a non-­expressive normative theory—​a concern for consequences—​to argue for certain legal policies based on their expressive effects. I wind up addressing some of the same subject matter Pildes and Anderson cover, discussing topics like the establishment of religion. Finally, there are normative theories of expressive conduct (category 4), which ascertain the moral status of private conduct by examining what that conduct expresses. These theories relate to law because they may justify the legal regulation of private conduct by the conduct’s expressively immoral character. An example is Deborah Hellman’s theory of private discrimination.16 She claims that the discrimination is wrongful if, but only if, it demeans another by expressing her inferiority. The wrong of discrimination thus depends on what the discriminatory act expresses. The theory recommends that law should ban discrimination that, by virtue of 15

The Expressive Powers of Law

its expression, is wrongful. Similarly, Dan Kahan argues that a bias crime is worse than the same crime without the bias because it expresses a more wrongful message, showing greater contempt for the dignity and humanity of the victim.17 The greater expressive wrong justifies greater expressive condemnation, in the form of harsher punishment. Again, my category 1 claims are different because they are causal, not normative. Despite the distinctions I draw, there are important connections between my primary topic—​law’s expressive effects—​and other expressive claims about law. These connections should make the theory here of broad interest. First, an expressive effects theory helps to explain why the stakes in expressive politics are so high. Christians and atheists may struggle politically over laws about a moment of silence in public school just because they prefer a symbolic victory over their opponents. Supporters and opponents of abortion rights may conflict over a new law they perceive as symbolizing the wrongness of abortion, even though they do not believe it will prevent any abortions. But if we understand that these symbolic laws also have expressive effects on behavior, the stakes are higher. As we shall see, government endorsement of religion may not only irritate some people, by reminding them of their outsider status, but also strengthen social norms of public religiosity. Even if abortion regulations contained enough exceptions that they fail to directly lower access to abortion, they might strengthen anti-­abortion social norms. Stronger norms mean stronger informal sanctions against the counter-­normative behavior, which can change behavior. Thus, the expressive power of law explains why symbolic struggle is so fraught, why it is so rarely ever merely symbolic.18 Similarly, expressive effects interact with the normative theories of expressive law, or so I argue in Chapter 8. The basic claim is that normative objections are stronger when they involve behavioral effects. For example, the normative theorist may be correct to condemn a law because it implicitly endorses, say, gender-­based stereotypes, regardless of consequences. Yet the objection to the law is distinct and more powerful if the expression will also have the consequence of strengthening norms of inequality and producing more discriminatory behavior.

The Limits of Expressive Claims As noted, the primary goal of this book is to demonstrate that law influences behavior expressively, that legal compliance is not explained entirely 16

EXPRESSIVE C L AI M S A B OU T L AW

by deterrence and legitimacy. This thesis is addressed to social scientists who work with these dominant theories. In particular, to convince economists who focus on deterrence, I build the theory of this book with simple rational choice tools. Yet I also wish to address a different audience for the purpose of making a different point. There is a significant problem of expressive overclaiming. Where economists are generally too skeptical of expressive effects, some legal academics, political actors, and members of the public are not skeptical enough, being too quick to perceive expressive effects around every legal corner. Claims of expressive effects are too easily made in the absence of a theory that specifies the conditions in which such claims are plausible. The defender or critic of a law who wishes to avoid the burden of serious consequential analysis may find it easy to make a sweeping rhetorical claim about the law “sending a message” of the right or wrong kind. In popular dialogue, the claims are legion: sex education will send a message condoning teenage pre-­marital sex; allowing medicinal use of marijuana will endorse recreational drug use; having the president speak to foreign leaders of hostile nations expresses the acceptability of those leaders and the hostility they show us; affirmative action programs communicate the message that their intended beneficiaries cannot compete on an equal footing; motorcycle helmet laws convey the paternalistic idea that people are not responsible enough to care for their own welfare. A bolder move is to assert that, if government does nothing to stop private conduct of an expressive nature, that inaction condones the private message. For example, if the government did not prohibit the sale of sex or human kidneys, then the government would send the message that human beings are commodities, the same message sent by those engaged in such transactions. Essentially, not to speak is to speak; not to condemn is to endorse. There are three theoretical problems with many of these exuberant claims about legal expression, all demanding greater attention to theory. First, there is nothing remotely inevitable about expressive claims of this sort. Consider some counter-­examples. Most drivers who lose control of their car (especially in good weather conditions) were at least partly at fault by driving too fast, not paying attention, driving while intoxicated or sleepy, etc. Yet most people don’t think that the government’s putting guardrails or soft objects on the side of highways sends a message that bad driving is acceptable. Most people don’t think that government acts to reduce violence or HIV in prison condones the criminal acts that place people in 17

The Expressive Powers of Law

prison, much less that the refusal to torture a criminal endorses his crime. Because many Americans understand the value of protecting free speech, they accept that the government does not endorse every false, offensive, or stupid utterance it permits. We thus require some theory to tell us when to take seriously claims of expressive effects. By articulating the expressive theories in detail, this book will help us to identify expressive effect claims that are too weak to merit attention. Second, there is a serious problem of ambiguity in many expressive claims. Sending the wrong message could be bad for the effects the law has on the world—​which depend on a causal claim—​or it could be bad for sending the wrong message regardless of its consequences. When commentators say that some law sends a message of inequality or reinforces or reifies a racial or gender stereotype, do they mean that the law, by its message, changes the amount of stereotypical thinking and behavior in the world? Or is the claim that the logic that the law embodies relies on a stereotype, whether anyone notices the logic and without regard to its changing anyone’s thinking or behavior? It is often hard to tell. I worry that there is a strategic, rhetorical advantage to the ambiguity, which allows some commentators to hedge about behavioral effects. On the one hand, expressive claims that include effects tend to be more compelling. Sending a message of inequality is worse if it actually strengthens the structures of inequality than if it just communicates an idea people already constantly receive, and understandably find irritating, but will not affect how egalitarian people actually are in their behavior. Similarly, sending a message of paternalism is worse if it actually causes people to take less responsibility for their lives than if it just communicates a message people ignore. On the other hand, effects claims are empirical and therefore subject to testing and falsification. One might encounter an embarrassingly strong counter-­argument, such as the observation that people who live with the inegalitarian or paternalistic law are no different in their general behavior than people who live without the law. So maybe it is strategically sensible, as a rhetorical matter, to hint at expressive effects so as to invoke some of their power, while preserving plausible deniability if called upon to redeem the consequential claim. If we have a catalogue of mechanisms by which legal expression influences behavior, we have a way to identify which send-­a-­message claims are plausibly behavioral and which are not. By drawing attention to the difference, I hope to encourage legal scholars to make less ambiguous claims. 18

EXPRESSIVE C L AI M S A B OU T L AW

The third and final problem is that legal commentators sometimes make an explicit claim of expressive effects, but simply fail to make a plausible case. Most of my argument for this claim, with examples, will have to await Chapter 6, after the development of the coordination and information theories from which to measure such claims. Yet I note here one clear limitation that both of my theories place on the claim that law expressively influences behavior. Causal claims necessitate a focus on the meaning of the law for the audience. Throughout the book, this point imposes some discipline on the kind of expressive claims one can plausibly assert. To explain requires a brief detour. Simplifying greatly, linguists discuss at least three perspectives by which one can judge meaning. Communication requires a speaker, an audience, and a conventional means of expression, and these three components give rise to three types of meaning: speaker’s meaning, audience meaning, and conventional (or sentence) meaning.19 For our purposes, it is sufficient to understand “speaker’s meaning” as “what the speaker intends” by his utterance.20 The audience meaning is what the actual audience understands the speaker’s utterance to mean, i.e., the audience’s “uptake.”21 Conventional meaning is the meaning derived from applying the existing language conventions to the speaker’s utterance. It is, roughly speaking, what the linguist H. P. Grice called “sentence meaning.”22 While speaker’s meaning requires an actual speaker to have that meaning, and audience meaning requires an actual audience to understand that meaning, conventional meaning is a meaning one could hypothetically construct from an utterance by using the relevant conventions (sometimes described as semantic and pragmatic norms).23 Of course, this meaning might merge with speaker’s meaning, if the speaker intends by his utterance a perfectly conventional meaning, and both might merge with audience meaning, if the audience interpretation is also perfectly conventional. But for many reasons, the interpretive practices of actual individuals may deviate from the conventional, so a particular speaker may intend and the particular audience may understand something distinct from what is conventionally meant by the speaker’s utterance. For present purposes, what I take from this literature (which is less settled and more complex than I’ve made it appear24) is the usefulness of distinguishing these perspectives on expression: first party perspective (speaker’s meaning), second party perspective (audience meaning), and third-­party perspective (conventional or sentence meaning). Identifying 19

The Expressive Powers of Law

the perspective clarifies the theory. When I claim that law facilitates coordination (Chapters 2, 3, and 4) and provides information (Chapters 5 and 6), I am using a second-­party perspective, based on audience meaning. For me to claim that the law influences behavior of some population, I must claim that this population—​the audience—​makes certain inferences from the existence of the law. Of course, there might be more than one actual audience; a law might have one expressive effect for one subpopulation based on the meaning that audience received, but a different expressive effect, or no effect, for a second subpopulation that understands the law’s meaning differently. Henry Smith emphasizes the distinction between legal audiences in his theory of property as language, drawing our attention to the communicative benefits of legal formalism when the audience is large and heterogeneous.25 We will return to this insight, but for present purposes, the relevant point is simply that, under the coordination and information theories, the focus is always on some actual audience. It is not relevant what meaning the authors of the law intend, nor what conventional meanings are plausible or best, if no audience receives those meanings. This point is not peculiar to my theories. Any expressive theory of effects invariably turns on audience meaning. If there were no audience meaning because the audience is unaware of the law, it could not possibly affect their behavior (the same is true not only for expressive theories, but also for deterrence and legitimacy theories, as we shall see). If the audience does perceive the law, its effect on their behavior will surely depend on its actual understanding of the law and not on the lawmaker’s intended meaning nor on some sentence meaning a third party could construct from the relevant conventions. The perspective being used in an expressive theory usefully constrains the evidence one should consult.26 If you want to know whether a local anti-­smoking law causes the general public to infer the existence of smoking risks or anti-­smoking public attitudes, you can reasonably look to public opinion polls before and after the ban is enacted. But such polls are not likely to be useful to identifying what (if anything) the lawmakers intended to express by the enactment, nor what the conventional meaning of the law is. By contrast, etymological evidence is usually relevant to conventional meaning, but may not be relevant to audience meaning. A few years ago a controversy broke out over whether the word “niggardly,” used as a synonym for miserly, was a racial slur.27 For the third-­party perspective, based 20

EXPRESSIVE C L AI M S A B OU T L AW

on convention, it is relevant that the word has an origin and derivation unconnected to the slur. But etymology is not relevant to speaker’s meaning or audience meaning unless the speaker or audience was actually aware of the etymology. This lesson offers an important limit to expressive claims: esoteric knowledge, known only to academics or experts, may be helpful or indispensable for determining conventional meaning, or the audience meaning for a small and specialized subpopulation, but is usually not helpful to determining the meaning that that the general public receives. Even though the same message can have different meanings for different individuals, to entertain the possibility of an expressive effect of law on behavior, we must begin with a message actually received by a substantial number of people. For large groups, it is ordinary knowledge and not esoteric knowledge that will determine the meaning. When commentators nonetheless claim expressive effects using esoteric knowledge, without claiming that such knowledge is widely known by any group, I sometimes get the impression that a fourth perspective of meaning lurks behind the claim, an egoistic perspective in which a critic essentially argues that because a law means something offensive to him or her, the law is normatively bad. I assume no one would explicitly make such a transparently flawed argument, but if the knowledge required for an interpretation is sufficiently esoteric, then it may be that no one but the critic shares that interpretation. To avoid this reductio of a normative theory of expressive law, it again helps to nail down what perspective one is taking for one’s assessment of meaning. This book uses the second party perspective, focusing relentlessly on audience meaning, which I show provides some discipline on the expressive effects claims one can plausible make. In sum, this book seeks to challenge both skeptics and over-­claimers of the expressive effects of law.

21

2 The Focal Point Power of Expression

B

ecause the law is full of requirements, we can easily miss its suggestive influence. In this chapter, I use the game theory concept of a “focal point” to describe that influence. Some of the earliest and most informal game theory shows that, if individuals share an interest in coordinating their behavior, they tend to engage in the behavior they find mutually salient—​the focal point.1 In these circumstances, I claim that law facilitates coordination by making a particular outcome salient; law’s requirements focus individuals’ attention on one way to coordinate, channeling their behavior in that direction.2 Obviously, there are situations where law does not have this power of suggestion and strictly requires sanctions, legitimacy, or a different expressive power in order to change behavior. But when individuals need to coordinate, highlighting the required behavior tends to create self-­fulfilling expectations that it will occur. In this chapter and the next two, I argue that the domain for law’s power of suggestion includes parts of constitutional law, international law, traffic and smoking regulation, property disputes, and the enforcement of custom. Let’s start with a well worn but ultimately inadequate example: a law that tells everyone to drive on the right side of the road. The legal announcement may be self-­enforcing because everyone has an incentive to do what everyone else is doing—​to coordinate—​so as to avoid a collision. The announcement makes driving on the right the salient solution to the coordination problem and its salience creates self-­fulfilling expectations that everyone else will drive on the right. 22

T HE FO C A L POIN T POWER OF EXPRESSION

This classic example involves a pure coordination game because, when the issue first arises, there is no conflict over whether to drive on the left or the right side, only the common interest in driving on the same side. However, if law’s focal point power exists solely in such situations, with zero conflict, then it is very limited, even trivial, because law almost always addresses situations of conflict. Indeed, we would have to make the criticism broader. If focal points generally matter only in games of pure coordination, then they are a behavioral anomaly of no great import, a mere curiosity, because pure coordination games are rare. People are usually not indifferent to how they coordinate. Yet The Strategy of Conflict, where Nobel Laureate Thomas Schelling first described focal points, has been called “one of the great classics of the theoretical social sciences.”3 Robert Sugden says that Schelling’s two chapters on focal points “rank among the most brilliant contributions to economics of the last half-­century,”4 while Roger Myerson calls the concept, “one of the great fundamental ideas of social philosophy.”5 Focal points are fundamental because they influence behavior not only in pure coordination games, but also in what Schelling called “mixed motive” games, a broad sweep of life in which there exists both common and divergent interests. Individuals sometimes have purely common interests and sometimes have purely conflicting interests, but frequently have a bit of both, a mutual desire to coordinate behavior but disagreement over how to coordinate. To illustrate, I offer Schelling’s Bystander Example, which I use throughout as a metaphor for how law facilitates coordination. A failed traffic light causes gridlock: “The bystander who jumps into an intersection and begins to direct traffic at an impromptu traffic jam is conceded the power to discriminate among cars by being able to offer a sufficient increase in efficiency to benefit even the cars most discriminated against; his directions have only the power of suggestion, but coordination requires the common acceptance of some source of suggestion.”6 Very likely, the drivers will not entirely ignore the Bystander’s hand gestures. Possibly, there will be full compliance with those gestures, despite the fact that the Bystander lacks the power of sanctions or legitimacy. We don’t have to leave matters purely at the hypothetical. A few years ago, the New York Times reported how a private individual successfully coordinated traffic in Haiti even after a major earthquake. Driving in Port­au-­Prince was extremely chaotic, “a 10th ring of hell” according to the story, but Levy Azor—​“a freelancer with a passion for order”—​successfully 23

The Expressive Powers of Law

directed traffic at a major intersection, making it run like a “symphony orchestra.” Working “only for tips,” he “has quickly become a symbol of hope; a whistle-­blowing reminder of the creativity that blossoms in a stateless void.”7 Azor had been successfully directing traffic long before the earthquake, as well as after. The source of Azor’s power was neither coercion nor legitimacy. Obvi­ ously, drivers did not fear that Azor would chase them down and sanction them for disobedience. Nor did he possess, at least at first, any of the indicia of legitimacy that sociologists and psychologists attribute to law (unless one stretches the notion of legitimacy so far as to be meaningless). There was no procedural fairness to his selection (he selected himself) and no guarantee that he followed a fair procedure in deciding who can proceed first (one might reasonably suspect that he favored those who tip). Yet there is something in the fundamental structure of the situation that gives Azor a power of suggestion. What is the structure? The drivers seek order; they wish to coordinate to avoid the outcome where neither driver yields, both proceed, and they crash in the intersection. For all but the most idiosyncratic driver, that is the worst possible outcome. Moreover, the drivers also wish to coordinate to avoid the situation where both parties yield and remain stopped, which not only needlessly delays them, but also recreates the initial problem. Each stopped driver would then have to decide whether to proceed immediately or wait for the other driver to proceed, where they again wish to avoid the outcomes where both proceed and both wait. These are coordination problems, not the more commonly discussed problems of cooperation, as illustrated by the Prisoners’ Dilemma. A Prisoners’ Dilemma would disappear if “players” (the individuals in the game) were sufficiently altruistic that they sought to maximize the sum of all players’ payoffs. But altruism cannot solve a coordination problem; it exists despite the fact that drivers share a common interest in avoiding certain outcomes. The obstacle is not self-­interest, but the fact that there are multiple ways to match behavior in a way that is best for both or at least better than some alternative each wants to avoid. Given two or more plausible matches, the problem is predicting which one the other player(s) will use. Sometimes matching behavior means doing the same thing—​as where successful radio communication requires everyone to use the same 24

T HE FO C A L POIN T POWER OF EXPRESSION

frequency—​and sometimes matching involves doing different things, as where one driver stops at the intersection and the other proceeds. What is important about the Bystander Example is that, unlike the decision to drive on the right side of the road, the problem here is not solely one of coordination. The drivers also have conflicting motives because each wishes to proceed ahead of the other, especially if yielding to the other will require stopping for several minutes for a long line of traffic. But this is why the example is a good metaphor for the focal point power of law. The Bystander wields a power of suggestion over drivers despite the fact that they have conflicting interests; the Bystander frustrates the driver whom he motions to stop, but she tends to stop anyway. Why does she do so? The Bystander’s ability to influence the drivers depends on the simple fact that, by waving on driver A and gesturing to driver B to stop, in full view of each, she raises the mutual salience of the outcome where the drivers do as she suggests. A and B may be perfectly selfish and rational; they may give no weight to the authority of the bystander. Nonetheless, driver B starts to worry that driver A, having been waved on, is more likely to proceed, which makes B infer that proceeding will produce a collision. So driver B stops. The result is compliance with the focal solution the Bystander creates. I pause here to explain what I mean by the nonstandard term “mutual salience” of the outcome the Bystander endorses. Game theorists typically posit that various parameters are “common knowledge” to the players in the game, which means that they are not only known by each player, but are known to be known, known to be known to be known, and so forth.8 On a common view, to solve a coordination problem (except by accident), drivers A and B must have common knowledge of the Bystander’s recommendation “R.” That is an extremely demanding requirement. Just for A, it would mean that A knows R; knows that B knows R; knows that B knows that A knows R; knows that B knows that A knows that B knows R; and so on, ad infinitum. Common knowledge would also require that B have the parallel forms of knowledge (knows R, knows that A knows R, etc.). 25

The Expressive Powers of Law

The fact that the Bystander gestures from a spot—​the middle of the intersection—​in full view of the drivers, while the drivers are in full view of each other, makes common knowledge possible. Political scientist Michael Chwe has observed that many human rituals succeed in creating common knowledge of the events of the ritual (such as marriage) by the fact that people face each other in view of the ritualized events.9 We assume that Drivers A and B can each see the Bystander and each other, can see that the other can see the Bystander and oneself, and so forth. I shall not always refer to the technical requirements of common knowledge. There is a controversy over whether full-­fledged common knowledge is strictly necessary to coordination.10 One solution is to emphasize that people have common knowledge when they have reason to believe in all the higher-­order beliefs, even though they do not consciously consider them all (as boundedly rational individuals cannot consider infinite levels of belief). In other words, common knowledge might exist with a few lower order beliefs (e.g., A and B each know R and each knows that the other knows R, but there are no further beliefs) combined with the potential for higher-­order beliefs, were the individual to consider the matter further. Given the absence of an alternative means of coordinating, the lower order beliefs should be sufficient to motivate the focal behavior. That is sufficient for my purposes, which is not to contribute to the game theoretic understanding of focal points, but to use the empirical fact of focal point influence (experimental evidence is reviewed below) to explore the effects of law. For the most part, I shall refer to the “mutual salience” of the third­party expression to acknowledge that something less than common knowledge may suffice. Returning to the drivers in the intersection, note that a traffic light, like the Bystander, exploits the power of a coordinating focal point. Of course, there are also sanctions involved, but the motivation of the driver is not merely to avoid a ticket for running a red light, but to avoid an accident. When driver A faces a red light, she believes that driver B on the intersecting road faces a green light and that (if she thinks about it) that B believes (if he thinks about it) that she faces a red light. The green light “waves on” a line of traffic, which makes it more likely that drivers in the intersecting line will cause a collision if they ignore the red and proceed. The law of traffic regulation harnesses the power of the Bystander to create expectations that one line of traffic will proceed, making it in the interest of the other line to stop. 26

T HE FO C A L POIN T POWER OF EXPRESSION

None of this is to say that the focal point influence is inevitable, merely that it is common. There are societies in which drivers routinely ignore traffic rules (a failure of deterrence and legitimacy, as well as law’s focal point power).11 If no one else is coordinating by the traffic signs, then a particular driver cannot use them for coordination either. In this case, either the coordination of driving is impossible or the drivers are using some other means of coordinating, and the new driver looks to discover what that is. Yet it is possible for people to decide, for reasons of pure self-­ interest, to coordinate on the basis of the rules and signs offered for that purpose, in which case the new driver wants to do so as well. Indeed, unless a society’s law is generally powerless, the mutual interest in coordination would initially favor obeying traffic signs; if one did not already know of their particular failure, they usually offer the most salient means of coordinating. My central claim (in this and the next two chapters) is that other law functions in the same way. A relatively small economic literature has explored the idea that law supplies a focal point for behavior.12 In the remainder of this chapter, I describe the general theory of focal points and review the experimental literature supporting the theory. Chapter 3 shows how the theory applies to law, generally and in specific areas: constitutional law, international law, the regulation of smoking, property disputes, the enforcement of custom, and of course traffic regulation. Indeed, we will see that the focal point power is strictly necessary to the application of legal sanctions and may be the mechanism at work in cases usually attributed to legitimacy. Nonetheless, there are limits to the focal point power of law that I also describe in the chapter, to identify the domain to which it does not apply. Chapter 4 extends the focal point theory to a dynamic setting where the law competes with social norms and customs.

The Pervasiveness of Coordination If you’ve ever lost track of your traveling companion when at least one of you lacks a cell phone, you have faced a “meeting place” problem, which requires coordinating on a time and place to find each other. If you’ve participated in a social protest, then you’ve faced a larger meeting place problem. Those who want to march in protest achieve greater influence if they appear together, which means that the interested individuals must coordinate on (among other things) a time and place to march. Such large­scale coordination is difficult, especially if one is trying to stay a step ahead 27

The Expressive Powers of Law

of the government one is protesting against, so in the Arab Spring we observed that new social media could assist the protesters (though perhaps less than some commentators claimed). Yet if you want to communicate via digital technology, you need to have “compatible” software, which means you must coordinate on a format. If you don’t all speak the same language, you will need to coordinate on which one to use and if you don’t want the authorities to observe your communications, you may need to coordinate on a code.13 Or consider ordinary business. If you own a shop, you have identified the place for meeting customers and suppliers, but you still must coordinate on time. You want to coordinate your hours of operation with customers and suppliers, so you will be open when your customers want to shop and when your suppliers can deliver. You will rely on a common time standard, which before the nineteenth century was just the sun: each town considered noon to be the time when the sun was directly overhead. When American railroads first connected distant towns to one another—​itself a major feat of coordination—​it became difficult and confusing to publish train schedules using each town’s local time. The railroads pushed for coordination around regional time standards, such as the four time zones in the continental United States. As a result, the sun might not be directly overhead at noon, but it is easier to read a train schedule and to schedule a meeting or phone call among residents of different towns.14 Of course, with the Internet, a business’s hours of operation matter far less because its virtual location is always “open.”15 The Internet itself is a stunning achievement of coordination, a “network of networks.” Even some kinds of competition involve coordination. In one sense, your goal in a contest is to prevent your opponent from coordinating; you gain by being unpredictable, not by letting the other player find the best strategy to counter your own. But in a more fundamental sense, people seeking a competitive “game” desire coordination on the terms of competition.16 Chess players do not want to get to the middle of a game and then discover they have each relied up to that point on different understandings about what pieces can make what moves or how the game comes to an end. Poker players do not want to lay down their hands and discover disagreement at the table about which hand wins the pot. Soccer and basketball players want to have the same conceptual understanding of what it means to be “out of bounds,” to “foul,” and to “score.” The players want to coordinate on the rules constituting the game because the game is more 28

T HE FO C A L POIN T POWER OF EXPRESSION

thrilling if there can be a decisive outcome, a clear winner who prevails by playing the game better under rules accepted by both sides. What I’ve said about coordinating on places, times, communication, and games is true about a great many parts of life. Many of these situations are recurrent and the settled rules for coordinating are often called a “convention.”17 Though a series of examples in this and the next two chapters, I show that life is permeated by the need for coordination by convention, including the situations of social conflict that law addresses. In Chapter 3, my examples are concrete, but to clarify the concept it will help to begin more abstractly by describing a few coordination “games.” I focus on three simultaneous two-­person, two-­strategy (two-­by-­t wo) coordination games. A reader already familiar with these classic games may skip this section, but for others, the analysis is helpful because I refer back to these games in subsequent sections. More familiar than the games I am about to describe is the Prisoners’ Dilemma (PD), a construct of enormous influence in law and the social sciences. I have argued elsewhere that legal scholars overuse the game, as it has no more real world significance for law than coordination games, but it receives far more attention.18 Nonetheless, the PD is a useful benchmark for comparing coordination games, so I begin with it. The Prisoners’ Dilemma Game and the Prisoners’ Alibi Game In the iconic narrative from which the game gets its name, a prosecutor suspects two prisoners of a felony, but can currently prove their involvement only in a misdemeanor. The prosecutor credibly offers each prisoner the same inducement to confess to the felony: “If you are the only one to confess, I will reward you by dropping all charges,” which is represented by the payoff of 0. “If you are the only one not to confess, I will use your confederate’s testimony to convict you of the felony and obtain for you the maximum five years in prison (–5); if neither of you confesses, you each get one year for the misdemeanor (–1); if both confess, I will convict you both of the felony but give you an intermediate sentence of three years (−3).” Altruism can of course change the game, but the standard assumption is that the prisoners care only about their own punishment and have no way to enforce a promise not to confess. Figure 2.1 summarizes the game (in normal form).19 The PD literature generalizes the example by understanding the strategy of not confessing (deny) as a form of “cooperation”; confessing is a form of “defection.” 29

The Expressive Powers of Law

Prisoner 2 Prisoner 1

Deny

Confess

Deny

−1, −1

−5, 0

Confess

0, −5

−3, −3

F i g u r e 2 .1  

The Classic Prisoners’ Dilemma

With these payoffs, if Prisoner 2 chooses Deny, Prisoner 1 is better off with Confess (receiving a payoff of 0) than Deny (–1). (In all matrices, the payoffs for Prisoner 1 are on the left in each cell and the payoffs for Prisoner 2 are on the right). If Prisoner 2 chooses Confess, Prisoner 1 is better off with Confess (–3) than Deny (–5). Therefore, in the one-­shot version of the game, Prisoner 1 has a “dominant” strategy of choosing Confess; it is her best move regardless of what Prisoner 2 does. Because the payoffs shown are symmetric, Prisoner 2 has the same dominant strategy. Thus, the only “equilibrium” is Confess/Confess, a form of mutual defection. The concept of an equilibrium is central to game theory and I will refer to it frequently. The standard “Nash” equilibrium is a pattern of individual behavior “that may be rationally sustained as unique best responses to each other.”20 Thus, in equilibrium, no one has any incentive to switch strategies unilaterally because each is doing as well as he can given what the others are doing.21 (Throughout, I indicate an equilibrium by underlining the payoffs.) The game here is termed a “dilemma” because the theoretically inevitable equilibrium Confess/Confess is worse for each prisoner than another outcome, Deny/Deny. Yet the latter outcome is not an equilibrium because, at that point, each prisoner is better off unilaterally switching strategies (i.e., the best reply to Deny is Confess). When the game is repeated indefinitely, if the players care sufficiently about the future, the Folk Theorem shows that mutual cooperation becomes a possible outcome (as each individual threatens to retaliate against a defection with a future defection), but mutual defection remains an equilibrium as well. The enthusiasm legal scholars have for the PD has led them to apply it to just about every area of law. Part of the attraction, outside the adversarial context where the government is interrogating suspects, is the simple normative story the PD offers. If citizens are “stuck” in the Defect/Defect 30

T HE FO C A L POIN T POWER OF EXPRESSION

equilibrium of a PD game, the state can make both players better off, improving social welfare, by using legal sanctions to change the payoffs to force a Cooperate/Cooperate outcome. Yet what was at one time a relentless focus on the PD game diverted attention from the pervasiveness of equally important situations, especially coordination games. The allure of the PD game has even caused some scholars to shoehorn a non-­PD situation into the PD model, resisting a more apt coordination description. A simple example is a “run on a bank.” Quite a few articles claim that “[b]ank runs represent a classic prisoner’s dilemma.”22 On this view, “[d]epositors will be better off individually if they could beat their fellow depositors to the bank and reclaim their deposits whenever there is the slightest bit of uncertainty about the value of a bank’s assets.”23 Yet this doesn’t really work. A good model of a bank run should include both the equilibrium outcome where the bank is stable—​as banks usually are—​and also the equilibrium where there is a run. The Prisoners’ Dilemma game can—​at best—​apply only after some uncertainty arises about the bank. By contrast, the simple game of Assurance, discussed in the next section, describes both the efficient “deposit” equilibrium and the inefficient “run” equilibrium and shows how that uncertainty or a lack of “assurance” can tip the situation from the former to the latter. The PD game is a poor model of a bank run for another reason. Even after “the slightest bit of uncertainty” arises in a bank, it is not necessarily the best strategy for each depositor to “reclaim” her deposit. Depositors incur costs in removing deposits and finding another place for their money. If some uncertainty arises about depositor A’s bank, and yet others will not reclaim their deposits, then A will have no interest in incurring the costs of reclaiming hers. It is only when she expects others to withdraw their deposits that she wants to withdraw hers first. The difference between wanting to take some action no matter what the others do and wanting to take some action only if others also do the same may seem small, but the Prisoners’ Dilemma is strictly limited to the former case. The latter situation is about coordinating one’s behavior with others. As our first coordination game, consider a variation on the PD narrative, the Prisoner’s Alibi Game. The prosecutor places, in different interrogation rooms, two prisoners who jointly committed some crime. The prosecutor’s case is so weak that the prisoners can defeat it and free themselves if, but only if, they can give a consistent alibi for their whereabouts at the time of the crime. It won’t work for the prisoners to say they were each 31

The Expressive Powers of Law

alone at the time; they each need someone to confirm their alibi and, because they are guilty, each knows that she can rely on no one except the other prisoner. But they did not agree on an alibi in advance and now must give an alibi without consulting the other. Each prisoner thinks of various possibilities: they were at one of their homes or the other, fishing or motorcycling at a secluded location, etc.24 Each of these possible alibis represents an equilibrium in that, if the prisoners manage to match alibis, neither has any incentive to switch to another alibi. The problem is that this common interest does not guarantee success because neither knows what alibi the other will choose. I illustrate this pure coordination game in Figure 2.2, which assumes there are only two possible alibis, A or B, where the payoffs from matching alibis are 0 and the payoffs from failing to match are –5. The game is “purely” coordination because the players are indifferent to which alibi they use as long as they match. Prisoner 2 Prisoner 1

Alibi A

Alibi B

Alibi A

0, 0

−5, −5

Alibi B

−5, −5

0, 0

Figure 2 . 2 

The Prisoners’ Alibi Game of Pure Coordination

Now let’s consider some impure coordination games, where there is a mix of conflict and coordination. I limit myself to three classic examples: (1) the Assurance (or Stag Hunt) game, (2) the Battle of the Sexes game, and (3) the Hawk/Dove (or Chicken) game. The Assurance Game Consider again a prosecutor bargaining with two prisoners. Suppose that, unlike the PD game, the prosecutor has so little evidence that she cannot convict either prisoner of any crime if neither confesses. Unlike the Alibi game, they need not even give an alibi to avoid punishment; they need merely to remain silent or to deny committing the crime. Against this best outcome, the prosecutor credibly promises a prison term of five years each 32

T HE FO C A L POIN T POWER OF EXPRESSION

Player 2 Player 1 Deny Confess Figure 2 .3 

Deny

Confess

0, 0

−10, −1

−1, −10

−5, −5

A Prisoner’s Assurance Game

if both confess, one year for being the sole confessor, and ten years for being the sole nonconfessor. Figure 2.3 depicts the strategic situation in a form that parallels the PD game. With these payoffs, there are two (pure strategy) equilibria: where both confess and each gets a sentence of five years and where neither confesses and each goes free.25 I derive this result as follows. If Prisoner 2 selects Deny, then Prisoner 1 is better off selecting Deny, and receiving no sentence, than selecting Confess and receiving a one-­year sentence. If Prisoner 2 selects Confess, however, then Prisoner 1 is better off selecting Confess and receiv­ ­ing a five-­year sentence, instead of selecting Deny and receiving a ten-­year sentence. Because the payoffs are symmetric, Prisoner 2 has the same preferences. Thus, the players want to match strategies, as in the Alibi game. But there are two important differences from the Alibi game. First, the difficulty in coordinating on an alibi is that each alibi is equally good, so it is difficult to predict which alibi the other prisoner will select. By contrast, in this game, both prisoners regard one equilibrium—​Deny/Deny—​ as superior to the other equilibrium—​Confess/Confess. The shared preference might make it easier to coordinate on the preferred outcome, except for the second difference. In the Alibi game, if you fail to coordinate, it doesn’t matter how you fail; any mismatched alibi leads to the same bad outcome. But in Assurance, one way to miscoordinate is worse than the other. The worst possible outcome is to select Deny when the other prisoner selects Confess. The strategy Deny is therefore risky, producing either the best (0) or the worst (–10) outcome, while the strategy of Confess is safer, producing the second best (–1) or third best (–5) outcome. It is hard to predict what people will do because they may rationally aim at the best outcome or at avoiding the worst outcome. 33

The Expressive Powers of Law

More generally, this coordination game is called Assurance (or Stag Hunt) because solving the game requires each player to assure the other that she is going to play the riskier strategy—​here, Deny—​so the other should as well.26 If Player 1 believes she has, by her assurances to play Deny, convinced Player 2 to play Deny, then Player 1 also wants to play Deny. Knowing this, Player 2 has some reason to believe Player 1’s assurances.27 Neither wants to confess if the other denies. As much attention as there is to the “Prisoners’ Dilemma,” real world prisoners must find themselves in an Assurance game at least as often, because the police often question pairs of suspects in situations where they lack the evidence to convict either, absent a confession.28 For example, suspects in an antitrust conspiracy often face this “Prisoners’ Assurance game” because prosecutors frequently can’t prove a conspiracy without at least one confessor.29 Thus, the popularity of the PD game in legal scholarship has the potential to obscure the importance of coordination.30 Not that we should use police interrogation scenarios as the ultimate measure of the frequency of coordination. The example is merely convenient for comparing to the PD. Another illustration of the Assurance game is the “bank run” just discussed, although that obviously involves more than two players. Everyone wants to keep their savings in the bank if (nearly) everyone else does, but to remove their money if enough others are going to remove theirs. The efficient equilibrium is where depositors gain the advantage of pooling their resources, but the inefficient equilibrium results when everyone seeks to avoid the risk of pooling and go it alone. The game captures the fact that, in times of uncertainty, leaving money in the bank is risky and the depositors need to assure each other that they will not panic. Unlike the PD, if no one else will withdraw their deposit in fear, there is no reason for you to do so. The Battle of the Sexes Game Now suppose a prosecutor has enough information to charge two prisoners with both a substantive offense, such as burglary, and a conspiracy to commit that offense (where the jurisdiction allows sequential punishment for both crimes). Suppose that the prisoners could each reduce their sentences only by each asserting that a single one of them committed the offense alone. If one accepts full blame, then neither prisoner can be convicted of the crime of conspiracy (which requires two or more participants); the confessor will be convicted only of the substantive offense, and 34

T HE FO C A L POIN T POWER OF EXPRESSION

the other prisoner will go free. The game requires coordination because the prisoners will fail to be credible and therefore fail to defeat the conspiracy charge if they each point the finger at the other or each point the finger at himself or herself. But now there is a direct conflict of interest: each prisoner prefers that the other prisoner “take the rap.” Each prisoner wants to be the one who walks free. Figure 2.4 illustrates this “Battle of the Sexes” (BOS) game.31 I have labeled the strategies “Take Rap” (for confessing that one did the crime alone) and “Accuse” (for saying that the other prisoner did the crime alone). To determine the equilibrium outcomes, note that if Prisoner 2 selects Accuse, Prisoner 1’s best reply is Take Rap (–1 instead of –3). If Prisoner 2 selects Take Rap, Prisoner 1’s best reply is Accuse (0 instead of –3). Thus, unlike a PD, Prisoner 1 does not have a dominant strategy, but wants to “match” strategies by accusing the same person that the other person does. Because the payoffs are symmetrical, it is easy to see that the same is true for Prisoner 2. Thus, there are two (pure strategy) equilibria: Take Rap/ Accuse and Accuse/Take Rap. As a result, the prisoners have common and conflicting preferences. They each rank as the worst outcome the mismatches of strategies where both try to accuse the other or where both try to take the rap alone. They share the desire to coordinate to avoid these outcomes (where each is convicted of both crimes and serves three years). But the prisoners have a conflict over which equilibrium is best, each preferring to have the other take the rap. This classic coordination game describes a great many situations, including some already discussed. For example, theorists usually say that the choice of driving on the left or the right side of the road is a pure coordination game. But it is likely that drivers would not be indifferent between the equilibria—​drive left or drive right—​if the issue arose today in, say, a Player 2 Player 1 Take Rap Accuse Figure 2 . 4 

Take Rap

Accuse

−3, −3

−1, 0

0, −1

−3, −3

A Prisoner’s Battle of the Sexes Game 35

The Expressive Powers of Law

newly formed nation populated by citizens who had formed their driving habits in both left-­driving and right-­driving nations. Instead, we would get a BOS game where those with experience driving on the left would prefer the left-­driving equilibrium, while right-­experienced drivers would pre­ ­fer the right-­driving equilibrium, but all drivers would prefer a uniform rule. The game also models the setting of national or regional time zones (and other standards, such as units of measurement or technical standards for interacting goods). Citizens from different towns may each prefer that the regional time standard adopt their local time (noon is when the sun is directly overhead in my town), but each may also prefer a uniform regional standard that differs from their own to the anarchy of a different time standard in every town. And we shall see how the game may represent a bargaining situation. The Hawk-­Dove Game To complete our survey of prosecutorial bargaining scenarios, suppose a prosecutor has so much evidence that she can convict both prisoners for a serious offense without a confession. But imagine that either prisoner can, by “snitching,” reveal compelling evidence that the true perpetrator is a previously unsuspected person C, thereby exonerating both prisoners. The problem is that the prisoners each prefer that the other one snitch because a snitch may suffer retaliation from C or other criminals.32 Figure 2.5 quantifies the outcomes. In Silence/Silence, each prisoner is convicted and receives a five-­year sentence (−5, −5). The outcome from Silence/ Snitch or Snitch/Silence is 0 for the silent prisoner and −2 for the snitch because neither is convicted but the Snitch incurs an expected cost of −2 from the risk of retaliation. If both snitch, they share the risk of retaliation and the result is −1. With these payoffs, if Prisoner 2 selects Snitch, then

Player 2 Player 1

Snitch

Silence

Snitch

−1, −1

−2, 0

Silence

0, −2

−5,−5

Figure 2 .5 

A Prisoner’s Hawk-­Dove Game 36

T HE FO C A L POIN T POWER OF EXPRESSION

Prisoner 1’s best response is Silence (receiving 0 instead of −1 from Snitch). If Prisoner 2 selects Silence, then Prisoner 1’s best response is Snitch (receiving −2 instead of −5 from playing Silence). Because the payoffs are symmetric, the converse is true for Prisoner 2. As a result, there are two (pure strategy) equilibria: Snitch/Silence and Silence/Snitch. These prisoners face what is called a Hawk-­Dove (or Chicken) game.33 There is conflict because each equilibrium has unequal payoffs, one favoring Player 1 and the other favoring Player 2. But there is also a common interest in avoiding what each regards as the worst possible outcome—​ Silence/Silence. The game involves conflict with a need for coordination. There are subtle differences from the BOS game. For one, BOS describes situations where the players seek to match outcomes where any failure to match is equally bad. The Hawk-­Dove (HD) game describes situations where there is a particular mismatch that is the worst for each player (while the other mismatch could actually be better for both players). In some cases, the difference is more important. The game gets its name from situations in which a player chooses between an aggressive strategy—​ Hawk—​and a peaceful strategy—​Dove. Imagine that Hawk is to insist on getting one’s way (here, having the other prisoner snitch) and Dove is to defer (being the one to snitch). The best outcome is to insist while the other player defers (victory). Intermediate outcomes are to defer when the other also defers (mutual compromise) and to defer when the other insists (capitulation). But the worst outcome for each is when both players insist and produce a costly conflict (mutual damage).34 We saw this game in Schelling’s Bystander example at the beginning of the chapter: two drivers on intersecting streets prefer the other to stop, but if neither stops, they crash, which is worse than being the one to stop. Indeed, the alternative name “Chicken” comes from a fictional game depicted in the film Rebel Without a Cause, where two teenagers drive their stolen cars towards an abyss and the first one who jumps out loses face, but the failure of either to jump in time is catastrophic. As I explain in the next chapter, disputes often have the structure of a HD game; not when an individual wishes to win “at all costs,” but when the failure to peacefully resolve the dispute is worse for each than giving in. The Overall Frequency of Coordination Problems As I said previously, my claim (in this chapter and the next two) is that legal expression influences behavior in situations requiring coordination, 37

The Expressive Powers of Law

so it is important that I establish that such situations are pervasive. The discussion of the prior three games illustrates the ubiquity of coordination problems. In the next chapter, I offer the more important demonstration, a series of concrete legal examples involving coordination. For now, however, consider a final abstract point, a comparison to the PD game. One might suppose that legal scholars give much more attention to the PD game than coordination because the PD is substantially more common. Yet I explore here three reasons to think otherwise. First, the PD represents only a small fraction of the possible games that arise in the simple two-­by-­t wo setting. Second, the payoff structure that gives rise to the PD game is trivially different from the structure that produces the HD and Assurance games, thus making it likely that all three games are equally common. Third, even within situations described by the PD, there is frequently a strong element of coordination. In its simple form, the PD game has two players, and each has two discrete actions they can take, where the decisions are made simultaneously. Long ago, game theorists mapped out all the possible games with this two-­by-­t wo structure. If the players only make ordinal rankings between outcomes and are never indifferent, there are seventy-­eight “strategically distinct games.”35 The PD game represents one of these possibilities. Although there are a few “close variants” of the PD game, there are many more cases and close variants of just the three coordination games just discussed—​A ssurance (Stag Hunt), BOS, and HD (Chicken).36 We don’t know that payoff structures occur randomly in the world, but if they did, we would expect these three coordination games to occur more frequently than the PD game. Certainly, no one has ever explained why the PD should occur more frequently. Second, very minor changes in the payoffs of a PD can transform the game into either an Assurance or HD game. To illustrate, Figure 2.6 represents a (partially) generic two-­by-­t wo game, where specific payoffs are replaced by the variables a, b, c, and d. The particular game that exists in Figure 2.6 depends on the relationship between those variables. The one­shot PD arises if, for both players, b > a > d > c. If b > a, each player wants to respond to strategy A (which, given this ordering of payoffs, we could call “Cooperation”) with strategy B (which we would call “Defection”). Because d > c, each player wants to respond to strategy B with strategy B. So each player will inevitably choose B and the sole equilibrium is B/B. Yet because a > d, each player regards this outcome as worse than the outcome A/A. That’s what makes the game a dilemma. 38

T HE FO C A L POIN T POWER OF EXPRESSION

Player 2 Player 1

Strategy A

Strategy B

Strategy A

a, a

c, b

Strategy B

b, c

d, d

PD: b > a > d > c Assurance: a > b > d > c HD: b > a > c > d Figure 2 .6 

A Generic Symmetric 2×2 Game

Yet Figure 2.6 represents the Assurance game if the payoffs take the form a > b > d > c. If a > b, then each player wants to respond to strategy A with strategy A. If d > c, then each player wants to respond to strategy B with strategy B. As a result, there are two equilibria: A/A and B/B, where both players prefer A/A (because a > d), but each regards strategy B as safer (it necessarily avoids the worst outcome c). This is Assurance.37 Yet the difference from the PD can be trivial, based only on a change in the relative positions of payoff a and payoff b. If b is slightly greater than a, we have a PD; if a is slightly greater than b, we have Assurance. Similarly, the HD Game arises if the payoffs take the form b > a > c > d. If b > a, then each player wants to respond to strategy A with strategy B. If c > d, then each player wants to respond to strategy B with strategy A. The two equilibria are A/B and B/A; each player prefers to play strategy B against strategy A (because b is the highest payoff), but if both play B they get the worst possible outcome (because d is the lowest). This is HD. Again, the difference can be trivial; all that is required to flip the PD game to HD is a change in the relative positions of payoff c and payoff d. If d is slightly greater than c, we have a PD; if c is slightly greater than d, we have HD. Thus, if one takes the PD to be a pervasive feature of social life, as many legal theorists appear to do, then there is good reason to think that the Assurance and HD games are also a pervasive feature of social life, given how little the payoffs have to change to flip one game into the other. If cooperation is a common problem, so is coordination. 39

The Expressive Powers of Law

Coordination is pervasive for a final reason: There is frequently a coordination problem embedded within an iterated PD. A theorist usually posits a PD by specifying that the players have only two possible moves—​one constituting cooperation and one constituting defection. But in the real world, the PD is usually complicated by the fact that there is more than one plausible way for the parties to cooperate. The Folk Theorem shows that the indefinite repetition of the PD will, if the players sufficiently value the future, create an equilibrium of mutual cooperation, though repetition does not guarantee cooperation because mutual defection also remains as an equilibrium. Yet with repetition, it will prove difficult to sustain mutual cooperation if the parties have different understandings of what counts as cooperation. At some point, a party will engage in behavior she believes is cooperative but that the other party views as noncooperative. The second party will punish what it views as defection, while the first views the punishment as unjustified defection requiring retaliatory defection. The resulting recriminations likely end cooperation.38 Thus, one step to solving the iterated PD game, frequently neglected in the legal literature, is coordinating on one particular means of cooperation. The resulting game is more complex than a two-­by-­t wo game. Instead of each player choosing one of two actions, there are three choices: Defect, Cooperate According to Plan A, or Cooperate According to Plan B. The result is a three-­by-­three game, as in Figure 2.7, adapted from Garrett and Weingast, the political scientists who first noted the legal significance of coordination within an iterated PD.39

Player 2 Player 1:

Cooperate A

Cooperate B

Defect

Cooperate A

3, 2*

1, 1

0, 4

Cooperate B

1, 1

2, 3*

0, 4

Defect

4, 0

4, 0

1, 1

* Equilibrium Possible Only With Iteration F i g u r e 2 .7  

PD with Embedded BOS Game 40

T HE FO C A L POIN T POWER OF EXPRESSION

Figure 2.7 represents a PD game. If Player 2 cooperates in either way, A or B, Player 1 gains more by defecting (4) than by either form of cooperation (3 or 1). If player 2 defects, then player 1 gains more by defecting (1) than either form of cooperating (0). Player 1’s incentives are the same. If the game is repeated indefinitely, however, the Folk Theorem says that it is possible to sustain cooperation; each party may cooperate to avoid the other side’s future defection. But in Figure 2.7, there is no gain over mutual defection and therefore little chance of sustaining cooperation unless the two players use the same form of cooperation, A or B. The need to match cooperative form is a matter of coordination. The type of coordination can vary, but the particular coordination problem embedded in Figure 2.7 is a BOS game. If iteration makes cooperation possible, then there are two cooperative equilibria: A/A and B/B. Each player prefers either equilibria to any other outcome, but Player 1 prefers cooperating at A/A, while Player 2 prefers B/B (each receives a payoff of 3 at her preferred equilibrium and 2 at the other equilibrium). Garrett and Weingast use the embedded BOS game to model treaties. Two nations, for example, may agree to limit their tariffs (against domestic interest groups that push for them) and sustain this agreement by threatening to breach if the other breaches. But the parties must define precisely what trading behavior constitutes “cooperation” for purposes of their conditionally cooperative strategies. If one nation eliminates its tariffs but enacts health or labor legislation that impedes imports from the other nation, is that “defection”? Under what circumstances is nontariff legislation that impedes trade consistent with “cooperation”? Perhaps there is a single answer to this question, but if there are two or more ways to define acceptable nontariff legislation and each nation prefers a different standard, they face a situation like Figure 2.7. Unless they first solve their BOS game by agreeing on standard A or standard B, they will not be in a position where long-­term gains can sustain cooperation (because “cooperating” on different standards pays so poorly). If the nations begin to cooperate, but are applying different standards, eventually there will be a case where one nation’s effort to cooperate under standard A is perceived by the other nation, operating under standard B, as a defection. The latter nation is likely to retaliate by defecting in some subsequent period, creating a risk that cooperation unravels. The example generalizes because cooperation is rarely self-­defining, as we will see in the next chapter. Those using the PD usually simplify by ignoring 41

The Expressive Powers of Law

these embedded coordination problems, but they may constitute the biggest obstacle to long-­term cooperation. In sum, there is no reason to think that the PD is more common a problem than coordination. Now let us consider how “focal points” represent a means of coordinating.

The Power of Focal Points Game theory seeks to predict and explain behavior based on a relatively simple set of assumptions. Crucial to the analysis is the payoff a player receives from reaching each outcome, reflecting the utility that outcome provides that individual. In single-­equilibrium situations, like the one-­shot PD, we can say that the payoffs alone create expectations that the sole equilibrium outcome will occur. Given the payoffs (and other variables, like information), rationality yields but a single behavioral outcome. Yet the more common situation is where the payoffs and rationality fail to constrain the outcomes and there are multiple equilibria, as we saw in the Assurance, BOS, and HD games. Here, we cannot say that the payoffs rationally determine any single set of expectations about how the game will be played. The vital implication is this: When the payoffs do not uniquely determine expectations and behavior, then influences other than payoffs can. Schelling called these other influences focal points. The economic focus on legal sanctions is justifiable when one seeks to change behavior in the presence of a single equilibrium, but not when one seeks to influence behavior in the presence of multiple equilibria. Instead, mere legal expression is one of the factors other than payoffs that can create a focal point and influence behavior. Schelling introduced the focal point concept with pure coordination games. If you ask two people to name the same positive number as the other without communicating, there are an infinite number of ways to match answers. Yet when Schelling posed the problem to his Yale colleagues and students, forty percent named the same number—​the number one.40 One might be tempted to say there is a purely rational process that produces this selection, but Schelling’s next example better illustrates the cultural contingency of focal points. Schelling modified the number problem by asking individuals what amount of money they would name if, by matching their answer with someone else trying to match them, they would each win that amount of money. Again, random matching should almost never occur, while matching at the amount of one dollar would 42

T HE FO C A L POIN T POWER OF EXPRESSION

produce almost no monetary return. Instead, a plurality of 29 percent of his respondents named one million dollars (though respondents knew they would not actually receive the money).41 It is difficult to imagine a culturally neutral explanation of this result.42 An equilibrium is focal because it differs in some salient way from the other equilibria for reasons not captured in the model. As Schelling puts it, some “symbolic or connotative characteristics that transcend the mathematical structure of the game” will tend to draw attention to one solution, making it “stick out” from the others.43 And human beings have learned that when faced with just this sort of problem, the best way to proceed is to pick the salient or prominent possibility. Schelling summarizes: “Finding . . . ​a key—​any key that is mutually recognized as the key becomes the key—​may depend on imagination more than on logic; it may depend on analogy, precedent, accidental arrangement, symmetry, aesthetic or geometric configuration, casuistic reasoning, and who the parties are and what they know about each other.”44 Given the variables involved, Schelling concluded that “[t]his corner of game theory is inherently dependent on empirical evidence.”45 Schelling asserts that what is true of the pure coordination game is also true of mixed motive games. We can see his point by introducing a slight degree of conflict into the situation. Suppose we tell two individuals that they will receive a monetary payoff if they “match” in naming a positive number and nothing if they fail to match. But suppose we tell both individuals that one of them—​Player A—​will receive $100 if they match on an odd number and $99 if they match on an even number, while the other—​ Player B—​will receive $100 for an even numbered match and $99 for an odd numbered match. Thus we create a BOS game. The conflict here (the difference in $99 and $100) is trivial compared to the common interest in coordination (the difference between $0 and at least $99), so we should not necessarily expect it to prevent coordination at the focal equilibrium. If the number one is sufficiently salient, B will be better off naming that number and getting a high probability of $99 than naming a nonfocal even number and most likely receiving nothing. Although the size of the focal point effect is a contingent and empirical matter, there is no reason a priori to think that it disappears entirely as the magnitude of the conflict grows. Where nothing about the features of a situation render a particular outcome focal, Schelling contended that expression could create a focal point 43

The Expressive Powers of Law

and thereby influence behavior. One possibility is that the players in the game create the focal point by talking to each other and agreeing on a means of coordinating. Because there are multiple equilibria, an agreement can influence behavior even though it is “cheap talk,” meaning, by assumption, that there are no means of enforcing the agreement, not even informal sanctions or an internal sense obligation for keeping one’s promise. The power of agreement seems obvious in a pure coordination game. If A and B want to meet and “agree” to a particular location, they have solved their coordination problem by agreeing to a location, which is now focal. Yet the same is true given some conflict. If, in the number-­naming game from the prior paragraph, A and B agree to name the number forty-­t wo, they are likely to do so, given the difficulty of coordinating any other way. (I discuss the relevant experimental evidence for this point below). Yet many of Schelling’s most interesting examples focused on the expression of third parties, who are not “in” the game of coordination. When a third party suggests or demands that the individuals in the game coordinate in a particular way, she makes that outcome salient and tends to create self-­fulfilling expectations that the recommended or demanded behavior will occur. The third-­party expression can influence behavior even though it is also “cheap talk” that does not change the payoffs nor reveal pre-­existing information. We see this in Schelling’s Bystander-­in-­the-­Inter­section example just discussed. Ordinary speech (and gestures) can carry the power of suggestion. In sum, Schelling starts by noting how salience influences behavior in pure coordination games. He then extends this basic point in two directions: (1) salience influences behavior even in games that mix conflict with a mutual desire for coordination and (2) third-­party expression is one means of creating salience or, put differently, of constructing a focal point. Combining these points, my claim is that, whenever individuals share an interest in coordinating, law can influence their behavior expressively by making salient the behavior the law requires. Before we consider legal focal points, however, let us examine more deeply the sources of focal power. The Rationality and Robustness of Focal Points Focal points might be entirely subconscious. Perhaps individuals coordinate by mindlessly selecting the option that first comes to mind. The psychological literature on “anchoring” suggests that individuals can be subconsciously influenced by rationally irrelevant stimuli in the way they 44

T HE FO C A L POIN T POWER OF EXPRESSION

make decisions.46 Being exposed to a number selected by a roulette wheel influences individual estimates of the percentage of United Nations member states that are African.47 So it could be the case that the expression of a particular coordination solution influences how individuals coordinate merely through subconscious anchoring. Yet this nonconscious account is not the full story. In situations where individuals need to coordinate, rational self-­interest gives the focal point an additional power. For example, when the Bystander attempts to direct traffic, we imagine that each driver may consciously reason that the other driver will do what the Bystander suggests, which makes each driver want to respond accordingly by also doing what the Bystander suggests. An interdisciplinary group of researchers ran an experiment that demonstrated this kind of reasoning. They began by asking one group of British subjects to name a flower; approximately one-­third named “rose,” which was the most popular answer.48 When the experimenters asked another group of subjects to name a flower that matches the flower named by a second subject in another room who is trying to match their selection, with higher payoffs for matching than not matching, two-­thirds of the subjects named “rose.” Thus, the conscious need for coordination doubled the number of individuals naming the option “rose.” At least some subjects did not just thoughtlessly choose whatever they found salient, but reasoned about what is likely to be mutually perceived as the salient solution. As John Maynard Keynes put it, in situations of this type, “we devote our intelligences to anticipating what average opinion expects the average opinion to be.”49 One cannot derive what is salient in a situation merely from rationality; Schelling’s point was that there is an inherent residuum of the empirical. Yet once one realizes the psychological fact that one solution is mutually salient (or more strongly, that there is common knowledge of the psychological prominence of a particular solution), rational self-­interest encourages the players toward its selection. At least where there are more than two ways to coordinate, as in the selection of a flower, it is rational to select the psychologically most prominent one, which stands out from the rest, because there is no better way to coordinate. When there are only two ways to coordinate, as in the effort to match on heads or tails, the psychologically nonsalient outcome does, strictly as a logical matter, “stand out” as the only nonsalient outcome. But, even here, the psychological salience of heads correctly predicts that more people would choose it, which gives 45

The Expressive Powers of Law

the calculating individual an incentive to select heads as well (when trying to match). Perhaps as an empirical matter, there is a meta-­convention in many societies that, when individuals are better off coordinating but would otherwise be lost about how to do so, they pick the most salient solution. It is important to see the advantage the players gain by coordinating, so let us examine the point with a numerical example. Reconsider the Bystander scenario, where two drivers meet at an intersection with a broken traffic light and the drivers observe the mutual salience of the solution where they behave as the Bystander suggests. Figure 2.8 offers one possible illustration, where the payoffs create a BOS game. The Proceed/Proceed outcome is the worst for each driver because it involves a fender bender. One possibility, represented here, is that the Wait/Wait outcome is the second worst outcome, worse than the payoff from waiting alone. That might be true if we considered the payoff here to incorporate the expectation that, when both drivers stop, each must play this negative value game for yet another round (negative because traffic delay is a cost), deciding again in the next time period whether to stop or proceed.50 Because each player prefers either means of coordinating (Wait/Proceed or Proceed/ Wait) to the two uncoordinated outcomes, this is a BOS game. Part of what is stylized about the example is that I am restricting the driver’s choices to “stop” or “proceed.” Drivers might instead pick the option of “creep slowly and be prepared to stop.” Indeed, we might imagine a different action for every speed by which a driver could choose to “creep” into the intersection. That would greatly complicate the example without affecting the basic point, which is that the uncoordinated action still runs the risk of an accident (even if at low speed) and the serious possibility of wasting the drivers’ time. The Bystander offers a means of coordinating to eliminate these two costs. Driver 2 Driver 1

Wait

Proceed

Wait

−2, −2

−1, 0

Proceed

0, −1

−9, −9

Figure 2 .8 

The BOS Intersection Game 46

T HE FO C A L POIN T POWER OF EXPRESSION

We can now quantify how self-­interest favors the outcome the Bystander suggests. The key is that the drivers know that the Bystander will not signal for both drivers to stop, nor for both drivers to proceed (if he did, they would have no reason to follow him). Thus, following the Bystander eliminates the worst two outcomes for each, giving each driver a chance of getting the best outcome of 0 (playing Proceed against Wait) and a chance of getting the second best outcome of –1 (Wait against Proceed). Let us start with the simple assumption that either outcome is equally (50 percent) likely. The result is an expected value of –0.5. What is the outcome without a Bystander? The answer is that the drivers will be worse off, because there is some probability of achieving each of the two worst outcomes, where both Wait (and waste time) and where both Proceed (and they crash). Averaging in these worst outcomes, the expected outcome is worse. Game theory offers a particularly stylized way to quantify the expected outcome without the Bystander, though the relevant concept requires some explanation. Let us first imagine a more realistic setting with a population of drivers. On occasion, two random drivers encounter each other at an intersection (not knowing anything about the other driver). According to the theory, the drivers can play a mixed strategy, which involves selecting a probability of each action—​Wait or Proceed—​where the probabilities sum to one. It may seem odd to imagine a person deciding to Wait with some probability and to Proceed with some probability, but a mixed strategy seems sensible if one considers making choices over time in a series of identical situations. Real world game players, for example, will try to be unpredictable by selecting different actions in the same situations, as a tennis player “mixes up” the placement of her shots and a poker player “mixes up” bluffing or folding when he has a bad hand. Drivers are not trying to be unpredictable, but depending on what other drivers are doing, they still might maximize their returns at the intersection, not by doing the same thing on every occasion, but by doing each action with some probability. We can see the point by imagining the extreme alternatives. If every driver selects Proceed on every occasion, then there are a great many crashes and many drivers would benefit by switching to Wait. Yet if every driver selects Wait on every occasion, no one ever gets anywhere once they reach an intersection with another driver. It is possible to reach an equilibrium where a certain number of drivers always play Wait and a certain number always play Proceed, but this too would imply that 47

The Expressive Powers of Law

when two drivers who always play Wait arrive together at an intersection, neither one ever moves. So the more likely result is that each driver plays each action with probabilities summing to one. For the payoffs in Figure 2.8, some algebra reveals a unique mixed strategy equilibrium in which the average driver waits with a probability of 80 percent and proceeds with probability of 20 percent. Given that the other drivers are, on average, doing the same, one expects a payoff of –1.8 from either waiting or proceeding.51 Now we get to the bottom line, which is to measure the value of the Bystander to the drivers. Based on the calculations, the Bystander improves the expected outcome from –1.8 to –0.5. If they could, the drivers would want to pay for an individual (up to 1.3) to help them coordinate in this manner, as actually occurred in the Haiti example mentioned previously. In Chapter 7, I suggest that parties to a dispute will pay and obey an arbitrator when the same coordination dynamic is at work. In that chapter, I also consider some objections and complications to this analysis (e.g., that the “losing” party—​here, the one told to wait—​might try to resist the focal point and that there might be “intermeddlers” who provide a competing focal point), but the basic logic remains: There is a mutual advantage, at least ex ante, to having a focal point. The mutual interest in coordinating creates a mutual interest in creating and attending to a focal solution. The salient solution creates self-­fulfilling expectations that it will occur. Legitimacy as Alternative Explanation In conversations, some readers have resisted the significance of focal points by suggesting that what is “really” going on is explained by fairness and legitimacy. Perhaps the Bystander is legitimate by virtue of being a Good Samaritan. Perhaps the force of the example derives from the assumption that the Bystander will fairly alternate waving on the cars and that people will defer to a legitimate process. There is no reason to assume that power of legitimacy is all the Bystander example entails. To a significant degree, the point is empirical, which is why I will review some experimental literature that finds evidence of the focal point power, distinct from legitimacy. But here I offer some final theoretical observations, starting with two more of Schelling’s thought experiments, to establish the conceptual distinction between focal points and legitimacy. 48

T HE FO C A L POIN T POWER OF EXPRESSION

First, before the era of cell phones, Schelling imagined two individuals accidentally separating and losing each other in a large department store. There is a sign posted throughout the store stating: “The management suggests that all persons who become separated meet each other at the information booth in the center of the ground floor.”52 It is easy to imagine that this third-­party expression influences their behavior by making focal one means of coordinating. But is that influence entirely reducible to legitimacy? Surely not. A fair process can create legitimacy, but it doesn’t really matter what process the management used to select the particular meeting place to recommend. Legitimacy can also be substantive, which here might mean that the particular location selected is fair. Yet even if the location is unfair in some way (as by exposing patrons to obnoxious ads or perfumes), the parties who prefer meeting each other anywhere in the store to not meeting each other in pleasant locations will latch onto whatever place is salient. Finally, the speaker—​store management—​might be a source of legitimacy, but even if the lost parties are in the store precisely to protest the illegitimacy of the management’s labor or environmental policies, the salience of the recommended meeting place gives them both a reason to go there. Legitimacy might strengthen the focal point effect, but is separate from it. The distinction between legitimacy and focal influence persists when we introduce conflict. Recall the example from the introduction, the white or yellow “line down the center of the road.” I discussed how the line regulates passing, but Schelling discussed how the center line coordinates driving by the simple creation of two distinct lanes.53 Even the division of a road into lanes involves some degree of conflict, at least for drivers who the road in only one direction (as a long distance driver). Such drivers would prefer a little more space for her lane than the other, especially when going around a curve at high speeds. The line drawn on the road creates a powerful focal solution—​if you want to avoid crashing when you cannot see oncoming traffic around a curve or over a hill, stay on your side of the line. Is obedience to the line solely a function of its fairness? Schelling claimed otherwise, noting that the line “very likely . . . ​can err substantially toward one side or the other before the disadvantaged side finds advantage in denying its authority.”54 This seems right. The traffic line need not lose all of its influence over drivers merely because it isn’t painted precisely down the middle or was set down by a corrupt government. 49

The Expressive Powers of Law

Again, in Chapter 7, discussing dispute resolution, I use game theory to measure just how much the white line or any other focal point can deviate from being “fair” before there is no longer a mutual advantage to coordinating on that basis. Depending on the payoffs, a disputant might tolerate considerable bias in the arbiter who will offer a focal point resolution if the parties share an interest in coordinating and there is no other available arbiter (and no reputational considerations). Yet even with some overlapping interests, there is a limit to how much bias a party will tolerate; as the bias grows, there is a point at which she is better off proceeding without any focal point. Ultimately, we shall see in Chapter 7 that self-­interest can explain why bias matters: one attends to and follows the focal point creator when doing so will, in expectation, increase one’s payoffs; one ignores and resists the focal point creator when doing so will, in expectation, decrease one’s payoffs. None of this analysis implies that legitimacy is not behaviorally important. For now, however, let us turn to the empirical support for the independent importance of focal points.

The Experimental Literature An experimental game theory literature supports the focal point theory both in general and as it applies to law. The experiments I review here mostly follow the economic method of motivating behavior in the laboratory by paying subjects actual money according to the outcome of the decisions they and their anonymous counterparts make. At the most basic level, researchers have demonstrated more rigorously than did Schelling the basic idea that focal points influence behavior in games of coordination.55 Next, considerable research demonstrates that “cheap talk” communication between the players in a coordination game—​a proposal or agreement—​can influence the players’ behavior. The communication in these experiments is mere cheap talk because the statement that one is going to take a particular action is not binding; the incentives to play the various strategies are the same after the statement as before. Nonetheless, when two parties are put in a setting requir­ ­ing coordination, and one player can send a message identifying one of multiple equilibria, the expressed equilibrium is significantly more likely to result.56 Finally, and of greatest interest for law, several experiments establish ­specifically that third-­party expression can influence behavior in coordination situations.57 These experiments demonstrate that the players 50

T HE FO C A L POIN T POWER OF EXPRESSION

attempting to coordinate will be influenced by the expression of someone “outside” their game. There is one dissent from this research, however, an article by Vincent Crawford, Uri Gneezy, and Yuval Rottenstreich (CGR) boldly titled: The Power of Focal Points is Limited.58 Most of the experiments just described involve pure coordination games, but CGR find that focal point influence can collapse entirely as one moves from a pure coordination game to a mixed game with asymmetric payoffs. They use a Battle of the Sexes game, which creates a divergence of interests because the two players prefer different equilibria, though the players share an interest in coordinating to avoid the nonequilibrium outcomes. Following some other researchers, CGR create an experimental focal point by the label they attach to the strategies. In one (pilot) study, for example, conducted with subjects in Chicago, CGR label the two possible strategies “Sears Tower” and “AT&T Tower,” where the former is a much more famous Chicago building than the latter (or was before it was later renamed the Willis Tower). In other studies they used “X” and “Y” as labels (where X turns out to be focal) or a geometric pie figure with just one of the three equilibria shaded. CGR show that the focal label successfully produces coordination in a pure coordination game, where the equilibrium payoffs are the same for each player, but entirely loses its influence in the BOS game where the equilibrium payoffs are unequal (recall each player receives her best payoff in a different equilibrium). In some of the conditions, the degree of asymmetry and hence the degree of conflict is very slight, as little as $5 vs. $5.10, and yet the focal point power evaporates. Thus, their subtitle is Even Minute Payoff Asymmetry May Yield Large Coordination Failures. These titles are, however, overstatements. The evidence that CGR present is far too narrow and contingent to conclude anything remotely as strong as “the power of focal points is limited,” given the results of other experiments. Indeed, even their results are not uniform. In two conditions of their “pie” experiment (AM1 and AL1 results reported in Table 5),59 the focal point continues to influence behavior significantly despite payoff asymmetries. As the authors concede: “label salience remains powerful, even in the presence of very large payoff asymmetries.”60 Apparently, some focal point labels work in the face of conflict and some don’t. The fact that some focal points collapse in the fact of asymmetric payoffs does not ­generalize. 51

The Expressive Powers of Law

Relevant here, Subhasish Dugar and Quazi Shahriar introduce the idea of “strong” versus “weak” focal labels.61 Previous research had found that one could create a focal point by using the current year at the time of the experiment—​e.g., 2014—​as a label, as compared to other years.62 In a pure coordination game, Dugar and Shahriar found that the focal influence of the current year depended on whether the other label was the prior year or a year from the more distant past. The current year generated more coordination—​was a stronger focal point—​when paired with a year from a decade before (2014 vs. 2004) than when paired with the prior year (2014 vs. 2013). The researchers used the same pairs of labels in an asymmetrical game. Instead of a BOS game, like CGR, Dugar and Shahriar use a Stag Hunt (or Assurance game, as in Figure 2.3), where the strategy necessary to reach the efficient equilibrium is riskier than the alternative strategy. Dugar and Shahriar found that attaching the “strong” focal label continued to influence behavior in the asymmetric game, even when the weaker label did not. Perhaps, therefore, CGR obtained their results merely because they stumbled onto some labels creating weak focal points, weak enough to lose all power under the greater stress of an asymmetric game. Indeed, note that the two CGR conditions in which the focal point continued to influence behavior in the asymmetric game both involved a game with three equilibria. Recall the distinction I previously made between situations of two equilibria and situations of three or more equilibria. A given focal label should be stronger with three or more equilibria. With only two, as a logical matter, the fact that one equilibrium is salient means that there is exactly one equilibrium that is not salient. As an empirical matter, individuals might always aim for the most salient equilibrium, but there is a logical reason to do so when there are three or more equilibria, because then there is more than one nonsalient option and no way to choose between them. To be clear, CGR got the same results—​payoff asymmetries nullifying the focal point—​with most of their pie games. Yet the fact that they obtained consistent failure only in games of two equilibria supports the idea that the results are limited to weak focal points in the most difficult circumstances for coordination. Janice Nadler and I tested the relevance of the number of equilibria by using a wide variety of focal labels in a game where the players choose among four actions and there were, as a result, four equilibria (where the two players matched on the same action).63 We found that a variety of 52

T HE FO C A L POIN T POWER OF EXPRESSION

focal labels retained their power to facilitate coordination when switching from the pure coordination game to a BOS game.64 The focal point should grow in power as one increases the difficulty of coordination by increasing the number of options beyond four. Indeed, Antoni Bosch-­Domènech and Nicolaas Vriend get very strong results in a game with thirty equilibria.65 In the real world, as we shall see in the next chapter, there are usually many more than two possible strategies and two equilibria, creating better conditions for focal influence. A recent bargaining experiment is particularly instructive. The economists Andrew Isoni, Anders Poulsen, Robert Sugden, and Kei Tsutsui (IPST), presented pairs of subjects with a tacit bargaining problem (a core concern of Schelling in his work on focal points).66 The challenge to the experimental subjects was, without communicating, to reach an “agreement” over the division of resources by “claiming” one or more disks represented on a grid on their computer screen—​the “bargaining table.” Each disk had a number representing its monetary value and there were always at least two disks. If the experimental subjects claimed the same disk—​i.e., if their claims overlapped—​then they failed to reach agreement and earned nothing. Otherwise, they would earn (in a randomly selected round) the amount of money represented by the disk. This is not a game we’ve seen before, but one can readily see the need for coordination. The players have conflicting interests in that each would like to have all the disks, but common interests in avoiding overlapping claims. There is a natural focal point of splitting an even number of disks, but how do the players coordinate on which disk to take? For some of the games, IPST made focal a particular division. The researchers had assigned each pair of subjects a color, red or blue, and identified for each subject a “base” consisting of a matching colored square on the bargaining table. In some conditions, ISPT placed one disk closer to the blue base and the other disk closer to the red base. Note that this proximity does not change the payoffs in any way; it is entirely extraneous to the structure of the coordination game. But it does make focal the division where each subject claims the disc that, on the grid, is closest to his base. Territorial proximity is likely to be salient. Indeed, the focal points here are quite strong. Over a series of games with varying conditions, subjects claimed the disc proximate to their base from 64 percent to 82 percent of the time, and the nonproximate disk from 7 to 21 percent of the time, statistically, an overwhelmingly significant 53

The Expressive Powers of Law

­ ifference. By contrast, the players had great difficulty coordinating in the d controls, where the disk placements were equidistant from the bases. Of interest for the present discussion, IPST varied the payoffs from the disks. In some games the two disks were of equal value (which still is not a pure coordination game because one person could claim both disks). In other games, the values of the disks were unequal, by ratios of 6:5, 8:3, or 10:1. The results here are complex, but the bottom line is that the focal effect did not entirely disappear for any of the splits, though it had progressively weaker effects as the inequality rose.67 Strategy labels are not the only experimental means of creating a focal point. Two other mechanisms also produce different results than CGR. First are asymmetries in the players. Erik de Kwaadsteniet and Eric van Dijk experimentally manipulated the status of participants by telling them how they did on a preplay examination, designating the superior performer the “leader,” which caused the participants to perceive him or her as having superior status.68 Reversing CGR, de Kwaadsteniet and Van Dijk find that status differential had no effect on behavior in pure coordination games, but significantly increased coordination and payoffs in a BOS game, not surprisingly by having the players reach the equilibrium favored by the high status player. In a similar experiment, Hakan Holm did not seek to create status differences in the laboratory, but merely made subjects aware of the sex of the person against whom they were matched in a BOS game with real monetary payoffs.69 When matched against a woman, a subject was significantly more likely to play the strategy associated with his or her preferred equilibrium than when matched against a man. Thus, gender facilitated coordination and mixed-­sex groups earned more on average than unisex groups. Yet, predictably, men earned more than women.70 Another means of creating an experimental focal point, and perhaps the closest in form to legal expression, is cheap talk. One example comes from my work with Janice Nadler.71 We randomly assigned subjects to the payoffs in the HD game in Figure 2.9. The (pure strategy) equilibria are C1/R2 and C2/R1. The game creates conflict because Player 1 prefers the R2/C1 equilibrium, while Player 2 prefers the R1/C2 equilibrium. Yet the players also share an interest in avoiding the R2/C2 outcome, which is worst for each. We compared the participants’ interactions with and without third-­ party cheap talk. In the control, there was no expression. The expressive 54

T HE FO C A L POIN T POWER OF EXPRESSION

Player 2 Player 1

C1

C2

R1

1, 1

0, 2

R2

2, 0

−1,−1

Figure 2 .9 

Another Hawk-­Dove Game

condition consisted of one of two messages describing the two equilibria: either “R1/C2” or “C1/R2,” delivered in one of two ways: (1) by engaging a mechanical device (spinning a spinner) that would randomly point to a space containing one message or the other (where each message took up exactly half of the space to which the spinner could point); or (2) by keeping one subject—​a “leader”—​out of the HD interaction and assigning him or her instead the task of writing one message or the other on a blackboard. All messages were in plain view of all the participants in the decision making, so there was common knowledge (or mutual salience) of the message. The instructions told the participants that they could consider or not consider the expression as they wished.72 Like a BOS, the equilibrium payoffs in HD are asymmetric, the situation where CGR says the focal point has no effect. This HD is arguably a stricter test than a BOS because a subject who receives a message to play Dove in HD receives 0 with certainty, while playing without third-­party cheap talk would have positive value (the mixed strategy equilibrium payoff is 0.5).73 Nonetheless, the expression influenced the behavior of the subjects. The R player was significantly more likely to play strategy 2 (Dove) when exposed to the message R2, compared to the no-­expression control, while the C player was significantly more likely to play strategy 1 (Hawk) when exposed to the message C1, compared to the control. The opposite message had the opposite effects (also significant). We found a stronger effect when a human leader delivered the message, but even the mechanical device produced a significant effect on behavior. If a simple mechanical expression can influence behavior in this HD game, we reasoned that something as salient as legal expression can do so at least as powerfully. As a final point, a variety of experiments introduce the possibility of learning, which CGR noted might change their results.74 Nadler and I also 55

The Expressive Powers of Law

conducted a learning experiment.75 As in CGR, the first subjects played a BOS game with no information about how anyone else played the game. Here, the strategy with a focal label appeared to be more popular, but the effect of the label was statistically insignificant. Nonetheless, we provided a second group of subjects the exact numerical results from the first group. To casual inspection of these results, the focal label had a small effect on the first group’s behavior. Apparently, noticing this (statistically insignificant) difference, the second group played the focal strategy even more frequently, making the difference statistically significant. These results are sensible. The asymmetric payoffs might make the game confusing to many subjects, obscuring the focal point or the advantage of coordinating at the focal point. But the information about prior players—​a plurality of whom had coordinated at the focal point—​works to make the associated strategy more focal. In the real world, of course, we expect that people can usually observe whether prior players have coordinated around the legally created focal point (the required behavior), so that even a small difference could be magnified over time by learning. In sum, the experimental evidence supports the idea that mere expression can influence behavior by creating a focal point for coordination. Even if law were merely cheap talk, it would remain a powerful way to make salient the outcome it demands and thereby generate self-­fulfilling expectations that it will occur. Salience is not assured, but the experiments verify that mutually observable cheap talk can have the influence the theory predicts. Next, I apply the focal point theory to law.

56

3 Law as Focal Point

T

he last chapter showed that coordination problems are pervasive and that, in such situations, expression can influence behavior by creating a coordinating focal point. This chapter contends that legal expression is one way to create a coordinating focal point and that law often works in just this way. I offer various examples, ranging across traffic regulation, anti-­smoking laws, property disputes, constitutional law, and international law. The accumulation of examples demonstrates the wide scope of the theory, though also its limits. I begin, however, with a puzzle about legal sanctions.

The Puzzle of Legal Sanctions Consider for a moment the nature of financial currency. In mature economies, with “fiat money,” one takes for granted the validity of cash. Only in a crisis do citizens wonder why they should exchange valuable goods and services for pieces of paper with no intrinsic value. There may be some formal legal obligation to accept legal tender for discharge of debt, but that ordinarily is not the immediate reason people accept it. Outside of a crisis, one accepts the standard currency as payment because one expects everyone else to accept the currency in future transactions. Even if the future transaction will involve professional criminals, who are generally beyond the reach of state legitimacy (e.g., illegal drug sellers), those criminals value cash because they expect to be able to transact with cash in the future, even with other criminals. It is these entrenched expectations that give currency its value.1 They also solve a coordination game. Moving from a barter economy to a 57

The Expressive Powers of Law

­ onetary economy produces substantial gains for nearly everyone, but m there are an infinite number of currency forms. Once the expectations settle on a particular form and everyone expects everyone else to accept that currency in exchange for goods and services, there is no incentive to deviate from the currency convention (i.e., no incentive to switch to a different form or to refuse to accept or use legal tender; there is an incentive to counterfeit the existing form).2 The expectations are so stable that we tend to reify the value of currency, as if it had worth independent of the expectations underlying it. Money is valuable, we think, because it is money.3 The same mental error occurs in law. In mature legal systems, one may lapse into thinking that the sanctions backing law are an exogenous and independent force. As paper currency may seem to have intrinsic value, the papers embodying legislation and judicial decrees, each threatening sanctions for noncompliance, seem to have a power that itself motivates behavior. But sometimes a large or small crisis challenges the expectations underlying sanctions, reminding us of the fact that judges and legislatures and the law they create have no power except for the expectations for how people will react to them. President Andrew Jackson is famously misquoted for creating such doubt by saying, after the Supreme Court decided Worcester v. Georgia, “Well, [Chief Justice] John Marshall has made his decision; now let him enforce it!”4 The power of the courts is more settled today. Yet in Arizona a few years ago, a courtroom security officer defied the judge who ordered him to apologize for flagrantly violating a defendant’s attorney-­client privilege (by rifling through papers on the counsel’s table during a hearing).5 When the judge ordered the officer to be jailed for contempt, the officer’s boss, the controversial Sheriff Joe Arpaio, publicly criticized the judge and said there was no certainty that the officer would voluntarily show up to report to jail, implying that the sheriff would also not execute a directive to force him to appear. After the officer did appear and submit to being jailed, fellow officers protested by calling in sick and refusing to escort prisoners to the judge’s courtroom.6 The case illustrates that a judge’s power rests entirely on expectations of how others will respond to his orders. Had the security officer not reported to jail, it is not clear that anyone would have carried out the judge’s order to arrest him. Why then do law and economics theorists say, without elaboration, that it changes the expected payoffs of a behavior when a legislature or 58

L AW AS FO C A L POIN T

adminis­trative agency assigns a legal sanction for the behavior or when an ­executive official threatens penalties? Why should one fear sanctions when the court announces a judgment of damages or a sentence of imprisonment? To ask why the law has the power of sanctions is to ask why enforcement agents who impose sanctions obey officials who pronounce the law. Here, there is a potential for infinite regress. One can say that the sheriff fears that, if he refuses to carry out a court order to seize property, then the court will hold him in contempt. One could then say that the sheriff fears being held in contempt because, if he is, the bailiff will seize him. But why will the bailiff do that? Who does he fear? Obviously, the bailiff does not comply because the individual who issues the legal pronouncement—​the judge—​will personally, physically enforce it. So the problem arises at each level—​the legal pronouncement that Y is to sanction Z does not ensure that Y fears sanctions for failing to sanction Z. The pronouncement that X is to sanction Y for failing to sanction Z does not ensure that X fears sanctions for failing to sanction Y. And so on. As the economists Mailath, Morris, and Postlewaite put it: “Words written on a piece of paper don’t alter the laws of physics.” Thus, the law against theft doesn’t “affect what physical activities (such as attempts to lock an individual in a cell) individuals are capable of . . . . [M]y struggle to keep from being locked in a cage by you will result in an outcome that is determined by our relative strengths and martial arts abilities, but not by the existence or nonexistence of a law.”7 This problem of power is not limited to legal power. It applies to the crime boss, pirate captain, guerilla leader, or political dictator, all of whom depend on obedience for power. The simplistic explanation is that the others obey because they fear the dictator’s punishment, as by death. But the dictator rarely imposes punishment himself; he orders others to do so. And here the same circularity arises as for the courts: the dictator’s order to Y to kill Z does not ensure that Y fears punishment for failing to kill Z. The pronouncement that X is to kill Y for failing to kill Z does not ensure that X fears punishment for failing to sanction Y. And so on. The path out of these puzzles is the power of expectations in a coordination situation. Take the crime boss, who has power only by virtue of everyone expecting everyone else to obey his orders. One can imagine an individual acquiring that power from the fact that, when the group first formed, he was the best fighter, the most lethal killer. Yet those expectations may continue long after everyone knows he is not the best fighter/ 59

The Expressive Powers of Law

killer (like Don Corleone in The Godfather, who maintains power in his dotage). The power exists because any one person expects other mob members to kill him if he defies the boss’ orders just because he would expect the boss to respond by ordering his death and would expect others to follow the boss’ order. Wherever there is effective leadership, these interlocking expectations exist that one must obey the leader (at least in certain domains) because the leader controls the group and the group has more power than any one individual. Why does this leadership situation occur? Again, the basic answer is coordination. As Schelling says, “The coordination game probably lies behind the stability of institutions and traditions and perhaps the phenomenon of leadership itself.”8 The first level of coordination is merely that there is some advantage for the group to have different people doing different tasks at different times. Fighting a war, sailing a ship, playing a symphony, running a business (legal or illegal), all require careful coordination of the labor of multiple individuals.9 Frequently, the best action to take depends on circumstances that are changing rapidly, making time of the essence. For this reason, coordination is unlikely to occur in a decentralized, bottoms-­up fashion. (In Coasean terms, the transactions costs for anticipating and contracting over all possible circumstances in advance or having everyone spontaneously adjust to them as they occur, are prohibitively high). Even though everyone in the mob or guerilla army benefits from coordinating an attack on rival forces, it will not naturally occur that everyone decides on his own to implement the individual parts of the same (much less the best) overall strategy of attack. Instead, an individual needs to give orders. The group requires a leader. Now we reach the second level of coordination: who will the leader be? If everyone has an opinion he is willing to express about how the group should proceed, upon which expression will everyone coordinate? To solve this problem, the group needs to designate a particular speaker whose expression, being the focus of everyone’s attention, will be salient. But frequently there are too many volunteers to lead. Here the situation is like a BOS game: possibly every individual would like to be the boss (commander, captain, etc.), but if each holds out to be the leader, then the result is chaos. Because there are gains from coordination—​e.g., the profits of criminal organization—​each person gains more from being in a mob where someone else is boss than from the chaos where everyone claims to 60

L AW AS FO C A L POIN T

be the boss and no one is. Precisely how the group solves the problem should not detain us. The process need not be fair and need not endow the leader with legitimacy.10 For when everyone benefits from coordination, it is enough merely that everyone recognizes that one individual is salient, such that everyone is more likely to heed his coordinating instructions than those of any other individual. Once that happens, the leader works like the Bystander in Chapter 2: his instructions create self-­fulfilling expectations of compliance. The same is true of legal actors. First, society has basic needs for coordination that it accomplishes through law. Building roads, fighting crime, regulating pollution, creating currency, waging war, all require coordination. Second, to solve the basic coordination problem, societies will in some way select a political and legal leader or, more likely, an array of them. This selection will render focal the orders of these individuals. Once in place, the orders of legislators or judges are obeyed merely because everyone expects everyone else to obey them (and everyone expects everyone to expect everyone to obey them, etc.). Given those interlocking expectations, the legal actor can use expression to influence the enforcers who impose sanctions. Like obedience to the mob boss, everyone comes to expect that everyone else (or enough to make it matter) will obey the executive’s decree, the judge’s order, or the legislature’s mandate, including directives to sanction individuals for violating law. The legal actors have the power, by expression, to create self-­fulfilling expectations that their demanded behavior will occur. Legal sanctions come to be as reified as a stable monetary currency. To return to game theory, when legal sanctions work, it is not by changing the payoffs for given actions, but by changing expectations about what actions will be taken.11 In a formal sense, the act of stealing always pays x if no one subsequently seizes the thief’s property and always pays x − y when someone, including the state, later seizes the thief’s property. Law does not affect these payoffs. Instead, because the political and legal leaders—​legislators and judges—​declare that the thief’s property will be seized, the law changes the would-­be thief’s expectations about whether her property will be seized if she steals, increasing the likelihood of achieving x − y rather than x. This is the foundation of legal sanctions. I return to this point in the next chapter, where I argue that the focal point theory helps to explain how legal sanctions first arise. 61

The Expressive Powers of Law

The Focal Point Theory of Expressive Law Once we have robust legal institutions, the state can usually credibly threaten individuals with sanctions for violating law. At this point, perhaps it seems that one can safely ignore the fact that the state imposes sanctions only because of expressively manipulated expectations. Once legal institutions are established, perhaps there is no analytic loss from ignoring the focal point origin of sanctions. The rest of this chapter and the next argue to the contrary that the focal point power matters to contemporary compliance issues. Economists err in treating legal sanctions as the only mechanism by which law changes behavior. Given multiple equilibria, legal expression is one of the factors other than payoffs that can influence expectations and therefore behavior. By publicly endorsing a particular behavior, law tends to make that behavior salient, thereby producing self-­fulfilling expectations that it will occur, much like Schelling’s Bystander directs traffic. Any legal expression can have this effect—​a constitution, statute, judicial opinion, executive order, or administrative agency decision. Yet the scope of the theory is limited. Janice Nadler and I have specified four conditions: (1) The situation the law addresses includes an element of coordination—​the most important limit, addressed extensively in this and the following chapter; (2) the law is sufficiently clear; (3) the law is sufficiently public; and (4) there are no stronger, competing focal points.12 The necessary conditions for the focal effect obviously do not always hold. First, law may address situations of pure conflict, where there is no element of coordination. For example, there is no element of coordination in a one-­shot PD. The law would not decrease consumer fraud, for example, merely by making focal the outcome where contracting parties did not cheat one another. Second, the publicity of the law usually depends on media coverage or advertisement. Law cannot create a focal point if the content of the law is generally unknown. Third, the content of the law is often unclear to the public. Law cannot align expectations unless it is sufficiently clear that most individuals have the same interpretation of it, that is, believe it “points to” the same outcome. Finally, even if the rule is clear, law may face competition from factors that make another outcome salient. Most commonly, the law might attempt to change an existing norm that, as precedent for past behavior, continues to make salient the behavior that adheres to the norm. 62

L AW AS FO C A L POIN T

Nevertheless, the necessary conditions frequently do hold. Indeed, we might see law as the form of third-­party expression for which these conditions are most likely to hold, making law a society’s general purpose mechanism for constructing a focal point. First, law frequently addresses situations of conflict that contain an element of coordination (as the many examples in this and the next chapter demonstrate). Second, great publicity often attends legal rules, from either media coverage or direct government advertising (as by public service announcements or the posting of signs). In any event, while it is true that law will have no focal point effect if it is not generally known, it is also true that law will have no sanction or legitimacy effect among those who are not aware of its existence. Third, though complexity renders some laws opaque, others are fairly simple, e.g., a “no smoking” ordinance in restaurants or the right-­of-­way goes to the driver with the green light. As discussed in Chapter 8, one implication of the focal point effect is a new value to legal clarity because this particular behavioral effect requires clarity. Finally, law often avoids having to compete with other stronger focal points, as in newly developing situations where there are no settled expectations or after social movements have unsettled previous expectations, two points I discuss in the next chapter. Here, I discuss examples where law works because there is no powerful competition to the focal point it creates.

More Experimental Evidence Let us first return, however, to the experimental evidence. The experiments in the last chapter were abstract; they showed that expression could solve a coordination problem, but they were not focused on legal expression. Now let us consider some experiments explicitly testing the effect of legal focal points. Such experiments are difficult. As soon as one mentions “law” in an experimental narrative, one creates a problem: the perceived legitimacy of law may cause individuals to prefer to obey it for reasons other than its focal point effect. A similar problem may infect experimental designs that test the effect of legal “requirements” or “punishments”—​these morally loaded terms could trigger beliefs and motivations of legitimacy and therefore do not cleanly prove the existence of law’s focal point power.13 With this problem in mind, Janice Nadler and I designed an experiment to identify the contribution of legal legitimacy to compliance so we could separately measure the focal point contribution. As I will explain, we did 63

The Expressive Powers of Law

this by testing two versions of the narrative: one creating a game without coordination—​a one-­shot PD—​and the other creating a game with coordination.14 Because the narrative already involves an imagined rather than real scenario, we merely asked participants what they would do in the situation rather than pairing them against other participants and paying them according to their actual joint decisions. This design is more complex and requires greater explanation. The first of our two vignettes involved a property dispute over the ownership of a cat and the contrast of a PD with a HD game. We used a cat because we wanted the subjects to imagine caring a lot about the outcome without having the property be so valuable that it would be worth hiring a lawyer and going to court. The question we posed was whether the participant would continue to demand the cat knowing that, if the other claimant did the same, their failure to resolve the dispute would lead to a bad outcome. In our narrative, the cat had come into the possession of a third party who had the power to dictate the terms to the two claimants of the cat. Those terms were Solomonic: the third party would give the cat to one claimant if the other claimant conceded, but that if both claimants continued to insist on the cat, the third party would send it away to an unknown destination. If both claimants conceded, the third party would send the cat to a known destination, a friend who would allow both claimants to visit the cat. For the PD and HD versions, the narratives were identical except for the cat’s destination if the claimants both insisted. In the PD version, the unknown destination was a distant but responsible pet owner; the narrative said that each claimant preferred losing the cat to the unknown pet owners than losing the cat to the other claimant. This narrative description (and an accompanying matrix with illustrative payoffs) created a PD game where each claimant preferred to demand the cat no matter what the other did. In the HD version, however, the unknown recipient was a distant animal shelter, from which the cat might be euthanized (if not adopted). The narrative said that the participant preferred losing the cat to the other claimant to having it go to an unknown animal shelter. Thus, the description (and a matrix with illustrative payoffs) created a HD game where each claimant wanted to concede if the other insisted, but to insist if the other conceded. Within both the PD and HD versions, we had a law and no-­law condition. In the control “no-­law” conditions, we did not mention any legal 64

L AW AS FO C A L POIN T

rule. In the treatment “law” conditions, we told the participants that, although the police would not get involved and neither claimant could afford to sue in court, an old judicial precedent in the jurisdiction favored the other party’s claim to the cat.15 Our caveats about the police and courts were designed to avoid concern for legal sanctions. The possible legal influences, therefore, were legal legitimacy and the focal point effect. Our prediction was that the behavioral influence of law would be greater in the HD game than the PD game. Because there is no element of coordination in the one-­shot PD game, if the law has an effect in this scenario, inducing the participant to concede, it would have to be from the law’s perceived legitimacy. In the HD scenario, by contrast, there is an element of coordination and therefore two reasons the participant might comply with the law and concede the cat: legal legitimacy and the focal point effect. The behavioral influence of legitimacy plus focality should exceed the behavioral influence of legitimacy alone. Our results confirmed our predictions. In the PD scenario, there was no significant difference in the participants’ decision to insist in the law and no-­law conditions (66 percent vs. 64 percent). I do not believe that law generally lacks the power of legitimacy, but our narrative apparently created a situation where the perceived legitimacy was too weak to have an effect on the participants’ decision, given the stakes. Recall the “law” was merely an old judicial precedent. By contrast, in the HD scenario, this mention of law favoring the other claimant, despite being unenforceable, did significantly increase the willingness of a participant to concede the cat (decreasing the “insist” choice from 44 percent to 27 percent).16 Thus, law mattered in the HD scenario but not the PD scenario. The results are evidence that law generates more compliance in a condition requiring coordination, where the focal point effect can operate, than in an otherwise identical condition without coordination. The focal point effect is independent of legitimacy or sanctions. In a final experiment, Nadler and I implemented the prior design but changed the narrative from a property dispute to a contract negotiation. To accommodate the change, we switched from a HD game to a BOS game. In our narrative, the participant had reached the final stages of negotiating a contract, but had yet to agree on the remedy in the event of a breach (lost profits or out-­of-­pocket loss). We stated that each side preferred a different remedy and that the parties would not agree to have a contract unless they could agree to the remedy. To create the PD version 65

The Expressive Powers of Law

of the negotiation, we specified that the participant and the counterpart each preferred to have no contract rather than to reach a contract with the other side’s preferred remedy. Thus, the only equilibrium was for each to demand their preferred remedy and for there to be no contract. To create the BOS scenario, we stated that, while each party preferred a different remedy, each also preferred to have a contract with the other party’s remedy to having no contract. Each party therefore prefers to demand whatever remedy the other party will demand. For each of the two remedies there is an equilibrium where both parties demand that remedy. In the no-­law control, we had participants make their choices without any statement of the law’s preference for a remedy. In the treatment condition for law, we made clear that the law gave the parties the right to specify either of the remedies under negotiation, but we indicated that the law stated a default remedy, i.e., a remedy that would apply if the parties decided to create a contract but failed to specify a remedy. In particular, we told the participant that the default rule favored the other party, his or her counterpart. Because the law here is only a default, we automatically eliminate the role of sanctions (since there is no sanction for agreeing to a term other than the default). Our prediction, again, was that the law would have a greater effect on the choice of remedy in the BOS game, where the need for coordination made focal points relevant as well as legitimacy, than in the PD game, where only legal legitimacy could influence the participants’ decisions. Again, our results confirmed our prediction. In the PD scenario, there was no statistically significant difference in the participants’ willingness to accept the counterpart’s preferred remedy in the no-­law and law conditions (the demand for one’s preferred remedy fell from 90 percent to 78 percent). Again, I believe that the law’s legitimacy usually influences behavior, but our use of a mere default rule has no significant legitimacy effect. By contrast, in the BOS scenario, the introduction of a default favoring the counterpart did significantly increase the willingness of participants to concede on the remedy issue, selecting the counterpart’s preferred remedy rather than their own (demanding one’s preferred remedy fell from 72 percent to 42 percent).17 Thus, law again mattered more in the coordination scenario than the PD scenario. In other words, the default rule was focal.18 The results are evidence that a sanctionless law generates more compliance in a coordination situation—​where the focal point can operate—​than in a situation 66

L AW AS FO C A L POIN T

without coordination. The focal point effect is independent of legitimacy or sanctions.

Law’s Focal Power in the Absence of Sanctions Now let us move from the experimental to the actual. In the real world, it is easy to overlook the focal point effect of law because at least one and frequently both of the conventional explanations for compliance—​ sanctions or legitimacy—​are plausible. For that reason, we should begin by examining some law where at least one theory—​sanctions—​is clearly inapplicable because the law is sanctionless, by which I mean the government does not impose sanctions for legal violations. This is true for a considerable amount of international and constitutional law. International Law’s Focal Power Thomas Schelling introduced the idea of focal points primarily to explore what he called “tacit” bargaining, which occurs by actions rather than direct communication. Oligopolists, for example, might attempt to agree tacitly on a noncompetitive price through a series of price changes, because antitrust law makes it riskier to reach a face-­to-­face agreement. Schelling’s primary concern in The Strategy of Conflict was tacit bargaining among nations, especially in regards to military objectives.19 He posited that focal points could be essential to the avoidance of war. In one of Schelling’s informal experiments, he asked two individuals to play the role of opposing generals separated by an expanse of land. He directed each “general” to move his troops to occupy the maximum amount of territory without overlapping the other side, because overlap would trigger war, the costs of which exceeded the benefits. Each respondent acted simultaneously knowing the other was doing the same. Without a focal point, it is nearly inevitable that the respondents will select lines that overlap or that leave large amounts of territory unclaimed, which would lead to another round of the game, again risking war. Yet the respondents frequently succeeded in claiming all the territory without overlap by identifying a salient line on the map, such as a road or river, and moving their troops up to that line.20 The players found a focal point solution to their (imaginary) high stakes game. (The geographic focal point is much like the Isoni, et al. bargaining experiment of the prior chapter, where individuals claimed disks representing cash based on the 67

The Expressive Powers of Law

proximity of those disks to their “base” on a bargaining table). Moreover, if we imagine that the parties want to avoid war and succeeded in moving their troops to the opposite sides of some line on the map, occupying all the space without overlap, then we can see that the result would have a certain stability. Once there is no unclaimed territory, the focal solution is for each side to hold its ground, not moving backward or forward, thus continuing to avoid war. What is tacit can be made explicit. When the nations write down an agreement to a boundary (before or after their armies have moved up to that boundary), we have a treaty. In this scenario, we have the simplest reason for complying with the treaty: by hypothesis the parties wish to avoid war and the treaty expresses the focal solution, any deviation from which involves the ceding of territory or war. Either the treaty negotiation first creates the focal point boundary, by cheap talk, or the focal point boundary that existed before the negotiations determine the negotiation’s course. Either way, the fact that there is no outside enforcer does not mean that the treaty lacks power. The international law contained in treaties can be partially self-­enforcing. Some readers may prefer a different but equivalent description. We might say that the two nations on the brink of war themselves threaten the sanctions that enforce the bargain because each may go to war if the other nation violates the treaty by claiming more territory than the treaty allots. Of course, the war is also costly to the nation who obeys the treaty and merely defends its territory. Yet, one might say that the risk of war is, in some broad sense, a sanction. If so, it is the whole point of this chapter and the next to identify strategic situations where each party fears this kind of second-­party sanction because the worst outcome for each is uncontrolled conflict, a failure of coordination. What remains different, however, from the usual focus of the economic analysis of law is that the sanctions are not legal sanctions. They are not imposed by a third-­party, a government who is external to the conflict. In the domestic context, as we shall see, the examples involve citizens obeying the law from fear of the reaction of other citizens, not from fear that courts or police will fine or arrest the violator. In this international context, the nations need not fear that the International Court of Justice or UN Security Council will sanction them. Thus, my claim that the law provides a coordinating focal point independent of legal sanctions is fully consistent with the idea that the law works by harnessing the sanctions that parties may impose on each other. 68

L AW AS FO C A L POIN T

Back to the international context, we can generalize the result far beyond the military setting. There are two points. First, the interaction of states frequently presents a game involving coordination. Second, a written arrangement or adjudication between states may influence their behavior by virtue of creating a focal point and creating self-­fulfilling expectations of how to behave. As discussed in the last chapter, political scientists Geoffrey Garrett and Barry Weingast provide perhaps the first model of this sort.21 As depicted in Figure 2.7, they emphasize the existence of a coordination game—​ BOS—​embedded within the PD game because there is more than one way to cooperate. Instead of agreeing to a territorial boundary, their example was a trade treaty, an agreement to reduce tariffs that requires some definition of and restriction on nontariff trade barriers. One might add other typical treaties on foreign investments, the use of common pool resources (shared fishing area), or extradition. In each case, the situation might be an iterated PD, where one side cooperates to avoid future defections by the other. Yet there is a coordination aspect if there are different ways of defining cooperation: different understandings of nontariff trade barriers, of how to measure the permissible level of fishing, or of the conditions justifying extradition. The nations need to make sure they have the same understanding of these matters. If they prefer different ways of cooperating, we have a BOS game, and the conflict could easily unravel the agreement unless a deliberate effort is made to resolve differences about how to cooperate. After which, the treaty specifying the form of cooperation is potentially self-­enforcing. (Because the agreement is necessarily incomplete, a judiciary helps to resolve disputes as they arise, as I discuss in Chapter 7). There are even simpler examples of international law, involving only a BOS game (not embedded in a PD). Consider standard setting.22 When national or international parties seek to standardize weights and measures, communications protocols for air traffic control, the right of way among vessels in international waters, the international exchange of mailed and telephonic communications, or the exchange of fingerprints by police departments, they are solving a coordination problem. In each case, coordination on a common standard creates mutual gains. If the states were indifferent between different standards, the problem would be purely one of coordination. More likely, the states disagree over which standard is best, each preferring its own national standard. But despite this disagreement, 69

The Expressive Powers of Law

the benefit from “matching” standards with others may mean that the worst outcome will often be failing to coordinate. There is no necessity that states will reach an agreement on standards. They may hold out for better terms or the expected transition costs may exceed the benefits of a shared standard. When revolutionary France ­promulgated the metric system, it also adopted a decimal calendar and clock. Yet neither the ten-­day week nor the ten-­hour day caught on elsewhere and France abandoned them after a few years.23 The United States famously refuses to adopt the metric system. Perhaps the material costs of transition are too high, but, more likely, Americans sense a loss of identity or prestige by adopting “foreign” standards (an example of an expressive-­ politics theory of law). Yet nations sometimes do reach agreement on a new standard. When they put an agreement in writing—​a treaty or convention—​they create a focal point, which generates self-­fulfilling expectations that those who agree will switch to the new standard. Once everyone is using the same standard, e.g., for air traffic control or the right of way in international waters, there is no incentive to deviate; the standard becomes self-­enforcing. Other states left out of the process may later decide to adopt the standard of the states that made an agreement. Once a critical mass of states joins a given standard, the remaining states may realize that there is no chance of getting agreement to any other standard, so the focal point becomes the dominant standard over time. On this view, the effort to harmonize the domestic law of different states is merely a type of standard setting. Here the “standard” is the domestic law of say, contracts, securities, or antitrust. The greater the divergences between nations, the more difficult it may be to transact across national borders because at least one contracting party has to learn the law of the other nation (usually, as a matter of due diligence before agreeing to arbitrate any dispute under the laws of that nation). Policy convergence and harmonization save transaction costs when firms seek to do business internationally. At the same time, a state incurs costs in switching from one legal standard to another. When nations would benefit from policy convergence or harmonization, but each would prefer that other nations shift to its legal standard, the result is again a BOS game. Once the legal agreement is struck, it works as a focal point; absent some exogenous shock, everyone expects everyone else to use the standard they have agreed to. Given that one expects others to do so, there is no incentive to deviate. 70

L AW AS FO C A L POIN T

Constitutional Law’s Focal Power Hobbes famously described the horrors of life in the anarchic State of Nature: without government, a war of “all against all” where life is “nasty, brutish, and short.”24 Hobbes may have been wrong about what kind of government the war of all against all justifies—​perhaps not Leviathan—​ but on most accounts he was right in thinking that there is a wide range of governmental types that are preferable to the State of Nature.25 If so, one might model the State of Nature as a PD, as many political philosophers once did, with the creation of government necessary to avoid mutual defection.26 Yet a better model is an Assurance game because, on Hobbes’ view, everyone is better off with the equilibrium of government, but the worst outcome is to contribute to creating government (perhaps by forsaking private violence) when no one else does.27 Even an Assurance game oversimplifies the centrality of coordination, however. Besides choosing between the State of Nature and Government, one must also choose between many possible government types. The creation of a particular government is mixed motive because nearly everyone shares a common interest in avoiding anarchy, but there are divergent interests in exactly what government structures are selected.28 If the government’s structure is defined by a constitution, then the selection of a constitution is also a coordination game. Those creating a constitution are likely to disagree about which constitutional version is best, even though they may agree that quite a few versions are better than the failure to create a constitution. Political theorist Russell Hardin was perhaps the first to explicitly characterize constitutions in this way, initiating a political science literature that views constitutions as coordination devices.29 He models the constitutional convention as a multiplayer BOS game between the bargainers who are better off “matching” their strategies, by agreeing to the same structure of government, than they are if they fail to agree.30 If powerful individuals and groups are represented at the constitutional creations, the agreed-­to constitution gets its power from directing their expectations. The terms of the written constitution are focal. On this account, the constitution works like a treaty between the parties who create it, a cheap talk agreement creating self-­fulfilling expectations among participants.31 This explanation refers to the horizontal relationship between the private parties or states at the constitutional convention 71

The Expressive Powers of Law

and possibly those who later share political power in state and federal governments. The parties will insist on getting at least the power the focal point declares them to have. If the written constitution says that each state gets two Senators, a state will relentlessly insist on having at least that much representation in the Senate. Like a nation defending its territory, the state will play a Hawkish strategy—​doing everything it can to secede perhaps—​if it is denied this power. The other political players expect as much and give in to what seems inevitable, given the constitutional focal point. Similarly, when the constitution clearly allocates power between the federal branches, each institution will tend to insist on at least as much power as is clearly allocated to it and defer to the clear claims of others. The House will insist on its right to select its “Speaker and other officers,” individual Senators on their right to hold office for six-­year terms (longer than House members or the President), the Vice President on the right to cast a vote in an evenly divided Senate, the President to grant pardons, and so forth. Institutions might not exercise or claim “powers” when the account­ ­ability costs of holding the power exceeds the political benefits (e.g., the congressional power to declare war),32 but they will predictably insist on retaining the powers they regard as valuable. Given the costs of unresolved conflict, other institutions will tend to give in to such demands. The trickier question concerns whether the constitution operates as focal point for the vertical relationships, between government and citizens. An old idea is that democratic constitutions are an implicit bargain struck by citizens with each other, a social contract. This fiction seems a bit hollow and has been subject to various normative criticisms.33 My point, however, remains positive. What is the reason that political leaders respect popular limits to their power? Why respect individual rights? We think of the courts as enforcing such rights, but why do the executives or legislators heed judicial declarations of unconstitutionality, given that the courts have no means of enforcing their judgments except through agents of the executive and no funding except through agents of the legislature? We sometimes marvel at the fact that Presidents have ignored the rulings of federal courts, such as Abraham Lincoln’s refusal to heed a writ of habeas corpus, but the more fundamental question is why they don’t always ignore decisions contrary to their interests.34 Why are “parchment barriers” ever enough?35 Or, as Matthew Stephenson puts it: why do the parts of government with “the money and guns” listen to the part with neither?36 72

L AW AS FO C A L POIN T

Barry Weingast offers an answer founded in focal points. He claims that the stability of democracy depends on “the people” being willing to ­challenge—​one might say, “sanction”—​official action that transgresses democratic principles, as by a politician attempting to stay in office after being defeated in a lawful election.37 He models the problem as a game where citizen groups can maintain democratic rule only if they coordinate their challenges to officials. Each group prefers to challenge the official if the other group does the same, because only joint action will work, but would rather acquiesce if the other group acquiesces, because unilateral action is costly as well as ineffective. They must coordinate. The coordination problem is that different citizen groups have different views about the appropriate limits to state power. If each group seeks to oust government officials only when (and whenever) that group views the official as having overstepped her authority, the citizen response will never be sufficiently united to threaten authoritarian officials (but yet may cause constant turmoil). It is therefore essential that citizen groups coordinate their efforts to challenge government officials around a “social consensus” of state authority. In a large diverse society, that consensus is unlikely to arise in a decentralized fashion, so a written constitution is the means by which some actors create a consensus.38 On this account, the constitution is “third-­party” cheap talk. The citizens do not all convene and agree to the written constitution, but its wide availability and notoriety allows them to use it as a means of coordinating. As examples, consider constitutional provisions with numbers in them: the United States constitution sets age requirements to be a President, Senator, or Representative, establishes the number of years for different terms of office, the frequency of the census, the number of days the President has to veto a bill, and the supermajority votes necessary for a constitutional amendment. It is difficult to imagine a large and diverse citizenry reaching a timely consensus on all of these limitations, as a step towards enforcing them, without a centralized articulation. Once the constitution provides a focal point resolution, a specific numerical limit, the public can coordinate its enforcement activities around that boundary. Yet citizens face obvious difficulties in coordinating over more complex matters, such as the scope of individual rights, especially given the effort of political leaders to “divide and conquer” them.39 Perhaps the leaders of modern nation-­states are too powerful and the citizens too diffuse for focal 73

The Expressive Powers of Law

points to explain the enforcement of individual rights or the rule of law. Yet Tiberiu Dragu and Mattias Polborn deepen Weingast’s analysis and offer some cause for optimism by introducing a third category of actors: not just political leaders and citizens groups, but also bureaucratic administrators.40 For a complex society with a large government, the political ruler requires administrators to implement her policies; she and her immediate circle cannot do all the work themselves. But these agents, like all agents, have different incentives than their principal, the ruler. In particular, the administrators are more vulnerable to future sanctions for illegal or unpopular action; when the ruler loses power (by death, election, revolution, or otherwise), the future ruler may have no incentive to protect the administrators from civil or criminal liability or other popular retaliations. If so, then today’s ruler may discover that his administrators are more resistant to implementing illegal than legal policies. Agents always “shirk,” but in this model, bureaucrats shirk more if the task is illegal, given the risk of punishment in the long term. The possibility of foot-­dragging gives rulers some incentive to keep their policies within the boundaries of the law. Of course, one might say that sanctions actually underlie this theory: it is the administrator’s fear of future sanctions that drives her to resist the illegal policies. Yet Dragu and Polborn offer a second point that is expressive. They observe that the administrators face a coordination game because of a feedback effect on their decisions.41 The administrators’ decisions about implementation affect the likelihood that the ruler maintains power. If administrators give high effort to implement the ruler’s policies, the ruler is more likely to maintain power. The higher effort the administrators give for illegal policies, the more likely they are punished if the ruler leaves office, but also the more likely it is that the ruler stays in power and protects them. So an individual administrator might want to use high effort if most other administrators will, to protect the incumbent and themselves against the effect of being out of power. But if other administrators will not give high effort to enforce the illegal policy, the regime is less likely to survive, so an individual administrator has a stronger incentive to resist implementing the illegal policy. Suppose the ruler issues an order for police to use violence to suppress peaceful protests, a violation of individual rights. Here, administrators have an incentive to coordinate, to “match” strategies. Roughly, their choices are “Implement with high effort any policy that will preserve the 74

L AW AS FO C A L POIN T

ruler” or “   ‘Implement legal orders with high effort and illegal orders with low effort.” There is an equilibrium where all the administrators adopt the first strategy and a different “rule of law” equilibrium where the administrators adopt the second strategy. But while the first strategy of following all orders is fairly simple, the second is not. The rule of law strategy works only if the administrators all have roughly the same understanding of what “the law” requires, or at least what the citizenry thinks the law requires. So we return to the problem of ambiguity the citizens face in Weingast’s model. They cannot coordinate without a common understanding and the simplest way to provide that is by a centralized, written statement of the law. For the most basic law, that is a constitution. Thus, a well-­publicized document that comes to be known as “the constitution” for a regime supplies citizens with a focal point, a means of coordinating to insist upon certain rights, motivating bureaucrats and political leaders to respect them. There is no guarantee that the citizens will coordinate sufficiently to preserve individual rights. Authoritarianism remains a possible equilibrium. But there is no mystery in the fact that political leaders may feel bound to respect “parchment barriers” despite the absence of externally imposed sanctions. The situation is like that of treaties enforced by the reactions of the parties to a treaty violation. (One can use instead the explanation of legitimacy, but as I argue in Chapter 4, legitimacy seems like a poor explanation for the power of an institution that has not already generated some compliance. At the origin, the focal point offers the best explanation for why a written constitution influences the behavior of political actors). A different way to understand the constitution’s influence is to return to the discussion of leadership and legal sanctions: If political leaders have no power other than the expectation that people will follow their orders, then it is an important limitation if the actual expectations are that people will follow only certain orders. Nothing in the focal point analysis of leadership implies that the expectations have to be all or nothing—​that the putative leader is either a Leviathan or a fraud. So if the public’s expectations—​ including the expectations of those who normally enforce the leader’s orders—​are that no one would follow an order to do X, then the leader lacks the power to do X. The constitution—​its creation and embodiment in text—​ helps to create a common set of expectations about what those limits are. The theory explains the importance of text and perhaps why the American constitution holds a place of reverence for people who worry 75

The Expressive Powers of Law

about governmental encroachments.42 Parchment barriers can be real if the citizens take them seriously enough to pose a coordinated obsta­ ­cle to political leaders. We see the tool of writing used not only in nation states, but also by organizations that enact by-­laws that constrain the powers of different actors in the organization, a topic I return to in the next chapter.43 (When the text is ambiguous, the designated interpreter—​a court—​offers a focal resolution of the ambiguity, as I discuss in Chapter 7.) Constitutional law is constitutive of a legal order. A skeptic might therefore say that law provides a coordinating focal point at the foundational but not the marginal level. That is, the theory explains how people create a basic order and how that order imposes sanctions, but that, once those foundations exist, the rest of the work is done by sanctions without the further influence of focal points. On this view, the focal point theory has a narrow scope because it never helps to explain compliance with new laws. In the rest of the chapter, I give examples refuting this view.

Law’s Focal Influence in Everyday Disputes: Supplementing Legal Sanctions in The Hawk-­Dove Game There is a special clarity in the previous examples because it is so implausible that legal sanctions explain much of the compliance we observe with constitutional or international law. But even where legal sanctions exist, law’s focal point power may have some independent influence on behavior.44 For various reasons, sanctions are always an imperfect deterrent in practice, leaving room for the focal power to produce some additional increment of compliance. Indeed, in coordination situations, we would expect law enforcers to save on enforcement costs by exploiting the focal power. Here, I focus on the legal resolution of a certain class of disputes. The domain is the class of disputes sharing the features of a HD game, where Hawk means to insist on getting one’s way, Dove means to give in to the other’s demand, and the worst outcome for each player occurs when both parties insist. Traffic Regulation Begin with traffic. The subject is mundane but important given that car accidents kill a million people a year worldwide and injure a great many more.45 Traffic involves a constant stream of priority “disputes” between drivers, pedestrians, and bicyclists. For example, two drivers on ­intersecting 76

L AW AS FO C A L POIN T

streets seek to make turns that cut across the path of the other. Drivers on merging roads or lanes each seeks to get ahead of the other or those traveling in opposite directions approach a part of the road narrowed to one lane by construction, parked cars, or an old bridge. These countless priority disputes are prototypical coordination problems. They have the structure of a HD game because each driver wants to proceed ahead of the other (to insist on getting one’s way) but there is a common interest in coordinating to avoid the worst outcome—​a collision—​that occurs if both proceed (and to avoid the outcome where both give in and sit waiting for the other to proceed). Given this structure, where the state succeeds in influencing driving behavior, we should take seriously the possibility that a contributing cause is the coordinating power of law’s expression. Traffic signals, posted signs, and the hand signals of the traffic cop modify behavior, in part, in the same way as the hand signals of Schelling’s Bystander (from the last chapter): by making mutually salient (or creating common knowledge about) one way of coordinating the driving behavior. In particular, traffic signals are very much like the randomizing device—​a spinner—​that Janice Nadler and I used in a HD experiment described in the last chapter. People tended to comply with the spinner’s recommendation. In the introduction, I used the example of a “yield” sign. Assume the sign is placed so it is clearly directed at the drivers on one of two merging roads, but is also visible to drivers on the other road (the ones not directed to yield). If a driver in traffic approaches the intersection for the first time and sees the yield sign, the driver is more likely to expect a collision if he fails to yield. Because he knows that the drivers on the other road can see his yield sign and see that he can see his yield sign, etc., the expectation that the other drivers will not yield is particularly strong.46 Even without the fear of state sanctions or respect for state legitimacy, the driver complies with the yield sign to avoid the collision. Note that the point I am making is descriptive: the traffic sign directly influences the behavior of drivers. Whether this outcome is normatively good depends on a consideration of externalities. While drivers usually benefit, it is possible that their coordination allows them to speed through populated areas in a way that increases pedestrian and bicyclist anxiety and mortality. The signs regulating driving may also be an eyesore. Thus, in certain locations, the automobile speed that driver coordination enables might worsen coordination with pedestrians and cyclists and make society 77

The Expressive Powers of Law

worse off.47 This seems to be the idea behind the innovation of “Shared Spaces,” in which the state removes traffic signs and lane markings in certain traffic circles shared with pedestrians and bicyclists, forcing everyone to slow down and rely on conscious negotiation instead of habit to navigate carefully around everyone else.48 I discuss normative analysis in Chapter 8; for now, the positive claim is that traffic regulations influence behavior by creating a coordinating focal point. Having mentioned habit again, let us consider it as a competing hypothesis. One may habitually obey a yield sign and the other rules of the road. Yet, as stated in the introduction, people make choices about what habits to develop by consciously engaging in the behavior before the habit exists. Prior to forming a habit, people comply for other reasons, one of which is coordination. Note, for example, the paradoxical observation of Leonard Evans, a traffic-­safety researcher; he describes a person who routinely speeds in traffic but sits still at a red light at two a.m. even though there is no traffic.49 Such a combination is common because the norm of obeying red lights is stronger than the norm against speeding. The puzzle is that the driver poses a greater risk to himself by speeding than by proceeding through the red light in this limited circumstance. The coordination theory offers an explanation.50 The need for coordination creates a powerful incentive to form the habit of stopping at red lights, but not the habit of observing posted speed limits. At intersections, the costs of running red lights can be catastrophic, while the benefits are usually limited. Thus, a rigid habit seems quite sensible. By contrast, the costs and benefits of speeding are highly dependent on context. To minimize the risk of collision, one wants to coordinate with other drivers by matching the speed of traffic. Instead of a rigid habit of obedience, one wants to pick a speed either consciously, without habit, or through the habit of mimicking nearby drivers, even when they violate the law. That the same person routinely speeds when speeding is the norm but habitually obeys a red light (at least one that is red before he nears the intersection) is not a puzzle, but evidence that coordination matters. People also decide sometimes to deviate from a habit. Consider stop signs. In general, stop signs generate a lot of compliance; even if rolling through the stop sign is technically noncompliant, the sign still works, in most cases, to determine traffic priority and avoid collisions. Yet the conventional wisdom among traffic engineers is that stop signs don’t work 78

L AW AS FO C A L POIN T

well, and should not be used for the different purpose of slowing down (“calming”) traffic. For example, residential areas often have “t-­intersections,” where one road terminates at another. It is easy to grasp the almost inevitable custom that the driver on the terminating road, who must turn left or right, gives way to a driver proceeding straight on the nonterminating road. Putting a stop sign on the terminating road might desirably reinforce this custom. Yet some neighborhoods use their political clout to get their local government to erect two more stop signs at the intersection on each side of the nonterminating road, creating a three-­way stop. The sole purpose is to slow down drivers in the residential area. Studies show, however, that compliance on these nonterminating roads greatly degrades over time.51 For this reason, the Manual on Uniform Traffic Control Devices tells road engineers to use stop signs only to avoid accidents from uncertain rights of way, not merely to slow traffic down.52 Note the difference. At most intersections, where compliance does not degrade to zero, drivers have at least three reasons to obey the stop sign: (1) they are deterred by legal sanctions; (2) they defer to legitimate authority (for its own sake or to avoid harming others); and (3) they wish to avoid a collision in which they will suffer property damage or personal injury. On the nonterminating road at a t-­intersection, the first two reasons apply but the third does not. The drivers on the nonterminating road do not fear a collision; they expect the driver on the terminating road to stop, especially if they observe a stop sign on that road. So their only reasons to comply are legitimacy and deterrence. And these substantially fail. These stop signs don’t help drivers coordinate with other drivers and so they have no power of suggestion. And it turns out that this power is perhaps a more important incentive than any other to comply with this traffic rule. It is certainly an error to assume that all traffic compliance occurs because of deterrence or legitimacy. To extend the HD model beyond traffic, as I propose to do, the key question is what other disputes, when unresolved, lead to an outcome as mutually bad as the automobile collision is for the drivers who fail to coordinate. There are two conditions: (1) that the expected costs of unresolved conflict, for both sides, are high relative to the costs of giving in to the other’s demand (mutually high conflict costs) and (2) that the situation does not offer either player the opportunity to precommit to “insisting” 79

The Expressive Powers of Law

without risking the mutually worst outcome (effectively, that the game is roughly simultaneous, not sequential). The first condition is usually met in traffic cases because the costs of a collision for most drivers are much higher than the cost of letting the other driver go first. The second condition is usually met because the drivers’ decisions are sufficiently close in time (though not always exactly simultaneous) that a driver knows that if he proceeds through an intersection, he may not have time to avoid a collision if the other driver does the same. To clarify these conditions, consider two cases in which they do not hold. First, suppose a person drives a “clunker” of a car, which has so much damage to the body that no observer can imagine that he cares about incurring one more dent or scratch. He does not regard a minor collision as being worse than “giving in” and so he may be unmotivated by the salience of the yield sign directed his way (as long as any collision would be at low speeds). Second, suppose there is a traffic light at an intersection where a driver wants to turn left but is blocked by a long line of oncoming traffic. If he waits in advance of the intersection until the oncoming traffic stops, when the light turns red, he then expects everyone on the perpendicular street to start proceeding in front of him, at which point his best reply will be to continue to wait, i.e., to comply with the red light. But he has another choice, a common driving tactic: pull into the intersection to wait for the light to change. Thus, even when the light turns red, the drivers on the perpendicular street will wait for him to execute his left turn. What he has done is to seize a “first mover advantage,” which is only possible because the game allows some sequential moves rather than all simultaneous moves. Note that the moves don’t have to be literally simultaneous for the game to retain multiple equilibria, so as to make a focal point effective. Suppose there are stop signs at an intersection only on the secondary road, thus allowing drivers on the primary road to proceed, and the topography is flat so drivers can see the cars on the other road from a considerable distance. Suppose Driver S is on the secondary road and approaches the intersection so that he will reach it at approximately the same time that Driver P reaches it on the primary road. The situation does not force their stopping decisions to occur at the exact same moment. S might “test the waters” by seeing how P reacts if he doesn’t slow down; he might try to seize a first mover advantage by speeding up. P might respond in various ways at slightly different times. 80

L AW AS FO C A L POIN T

But the key is this: S knows that any move other than stopping carries some risk of a collision (greater than P expects from proceeding). He knows he may err in calculating when the two cars will enter the intersection, that P might also be in the midst of accelerating, and that P might not be paying attention to S’s speed because he assumes S will stop. Because of these risks, the expected costs of the Hawk strategy—​including even a small probability of a collision—​may still vastly exceed the expected benefits. So the stop sign can work expressively, even if the moves are not precisely simultaneous. What matters is that the situation does not offer S an opportunity to precommit to “insisting” without at least risking the mutually worst outcome. The examples generalize. The clunker example represents one instance in which the condition for coordination—​mutually high conflict costs—​ does not exist. If only one party has low conflict costs (again, relative to the amount at stake), then that party has a dominant strategy of being aggressive and the other party, realizing this, will back down. For example, in the international arena, most nations might regard the costs of fighting over certain stakes as outweighing the benefits, but a superpower might regard the costs of fighting a weak nation to be so low that there is no element of coordination. Like the driver of the clunker, the superpower can proceed regardless of what others do; knowing this, the weak nation defers. A more extreme case exists where the “conflict” is beneficial to one party, as a nation that stands to benefit from global warming. The nations who stand to lose from climate change might regard the absence of a negotiated solution as very costly relative to the stakes in getting what is, from their perspective, the best deal. But a nation that will gain from warmer climate has no interest in coordinating to prevent it. The first mover example also generalizes. Sometimes people are face-­ to-­face when one of them threatens to engage in an activity that will impose an external cost on the other. Smoking is an example (discussed later), when a smoker is about to start smoking in the presence of an objecting nonsmoker. The parties may then have an interest in coordinating to limit the costs of conflict. Yet in other cases, the externality is imposed prior to any face-­to-­face encounter. If a nonsmoker enters a bus terminal right after the smoker has finished a cigarette and is starting to leave, instigating conflict cannot now prevent the room from being filled with second-­hand smoke. If the situation is not likely to be repeated with the same smoker, the nonsmoker has nothing to gain by confronting the 81

The Expressive Powers of Law

smoker. By moving first, the smoker ended the uncertainty over how they would coordinate. The same is true of others who impose an externality (pollution, accidental property damage, etc.) before the other can object and threaten escalation. In the international arena, where military defense is often much easier than offense, the first to claim a territory, to occupy and fortify it, may cause the other to defer. Where at least one party has low conflict costs or is able to move first, there is no remaining element of coordination for law to facilitate. The theory here predicts no expressive effects. These examples show why we may not want to rely solely on the focal point effect, for traffic regulations and other rules. Inevitably, some individuals will not be playing a coordination game, so society will need some other mechanism—​sanctions or legitimacy—​to bolster compliance for these individuals who are unmoved by focal points. Nevertheless, where the conditions exist, the focal point still generates much of the actual compliance we observe. Ordinary Disputes Now I can offer one my major arguments for saying that the focal point theory has a wider scope than has previously been understood. Quite a few real world disputes have the structure of a standard traffic conflict. In many disputes, each shares an interest in avoiding unresolved conflict because, while each side prefers to get its way, each regards the worst outcome as occurring where neither side gives in and the conflict escalates. And in many of these settings, neither party can claim a first mover advantage with­­out risking the outcome both parties regard as the worst. Thus, there is room for law to influence these disputes by providing a coordinating focal point. To justify this analogy to traffic priority disputes, as modeled by a HD game, there must be some reason that each side regards as the worst outcome the situation where both aggressively insist on their position, some sense in which each side sanctions the other when the two fail to control their dispute. There are a number of reasons this can be true. One is violence. Disputants may find themselves in something like a HD game because of the risk that unresolved conflict would lead to violence, the expected result of which makes it the worst possible outcome. In many such situations, neither side can wholly avoid the risk of violence by trying to play Hawk before the other. 82

L AW AS FO C A L POIN T

Violence is a background risk of many disputes. Much of the violence in ostensibly ordered societies involves individuals engaged in “self-­help” remedies against someone whom they regard as having infringed on their rights.53 We see this in scholarly literature on the enforcement of informal property norms and the reactions to personal insult.54 We also see this risk in news reports of shootings over trivial disagreements, like talking or text­ ­ing in a movie theatre, or the “keying” of cars left in the wrong parking space. Perhaps this is an American perspective because so many Americans are armed, but one cannot have a serious dispute without rationally assessing the risk of violence as greater than zero. That is one reason many Americans would give for being armed. If two disputants each regard the expected costs of possible violence as exceeding the expected costs of giving in to the other (which will be true if the costs of fighting are high relative to the value at stake), each may regard fighting as the worst possible outcome. This is true ex ante even if the party who “won” the fight would do so without injury (e.g., in a gunfight, where the winner may be unscathed) as long as there is uncertainty about who the winner will be and costs from losing. Of course, there is no element of coordination for a party who is sufficiently confident of winning the fight without injury. But there will be a class of disputes in which, like a HD game, each disputant regards violence as the worst possible outcome but hopes the other will give in without a fight. Nations may dispute territory but neither may believe it is worth warring over. The same goes for neighbors. When the Hawk-­Hawk outcome is violence, there is frequently no opportunity for a first mover advantage, no chance to precommit to playing Hawk so the other will play Dove. One might take the first step toward a fight—​“take a swing” or brandish a weapon—​to show that one is committed to playing Hawk, but even if neither side believes the stakes are worth fighting for, one cannot be sure that the other side will back down. The reasons are familiar. An assault victim may respond violently, despite the expected net costs, either because he becomes enraged and acts irrationally (in the short term) or because he believes that the reputational costs—​a loss of face—​from fleeing the fight are too large.55 There might be a series of steps short of fighting that one takes to demonstrate a serious willingness and ability to fight. Or there might be a convention of a “token” amount of fighting, in which no one gets hurt. But 83

The Expressive Powers of Law

the person unwilling or unable to fight seriously will try to mimic these steps and the other party will never be certain the first party is not bluffing. Talking tough or shoving someone is like trying to speed up through a red light; there is always a risk of causing the worst outcome, a serious fight, and that risk makes the game like HD whenever the expected costs of showing oneself willing to fight carries a risk of fighting that is not worth the stakes in the dispute. The problem here is computationally complex, so we should not be surprised that sometimes violence breaks out even though both sides preferred giving in to fighting. Violence is not the only reason that conventional disputes may have the structure of a HD game. A second example of costly unresolved conflict is a heated shouting match or exchange of profane insults (perhaps where each side bluffs about his willingness to fight and neither bluff is called). Both sides may regard the row as emotionally draining and acutely embarrassing or humiliating, especially if it occurs in public. If it occurs in private, the odds are that the row involves two people who know each other, such as neighbors or co-­workers, so the costs may include a permanent rupture of a social relationship. Or contracting parties who prefer to get their way in a disagreement may each expect that heated conflict will damage their reputations as reasonable trading partners, costing more than the amount at stake in the dispute. In general, if the stakes are low enough, the parties may regard a shouting match or their equivalents as the worst possible result of a dispute. As a third and final example, consider the bargaining impasse. Parties may bargain over the resolution of a dispute or they may find themselves disputing in the midst of a negotiation. There are many ways to model bargaining, but I simply want to point out that, whenever there are mutual gains from trade and the parties walk away from a deal, they are harming themselves in a way very similar to the physical and verbal abuse in the situations just described. Various divisions of the possible gains from trade would be better for each than the outcome where no deal is reached. Yet as each attempts to gain a larger share of the contractual surplus, sometimes the parties reach no deal at all. All of this analysis points to a focal point influence. Given a mutual desire to avoid unresolved conflict, and the absence of a first-­mover advantage, third-­party cheap talk may influence how the parties resolve the dispute. As in the foregoing discussion of constitutional and international law, a salient expression endorsing one means of resolving the dispute is 84

L AW AS FO C A L POIN T

likely to create self-­fulfilling expectations. Like the “yield” sign or the gestures of Schelling’s Bystander, a legal expression favoring one disputant is likely to tip the balance of expectations in favor of that party, making the disfavored party believe the favored party will not back down. Of course, the law uses normative language to say that one party is entitled to prevail. But aside from legitimacy is the rational consequence of expectations: in a HD game, if the other party will play Hawk, the best response is Dove. Independent of legal sanctions, the law harnesses the power of second-­ party sanctions, the willingness of the disputants to inflict costs on each other if the dispute remains unresolved. Consider some example, starting with smoking regulation. Suppose two individuals want to sit in the same public area for a time and one wishes to smoke a cigarette and the other wishes not to be exposed to cigarette smoke. They have a dispute. Any such case carries a low level risk that the dispute will escalate into a humiliating shouting match or violence. Almost all smokers and nonsmokers will think it better to give in to the other side (by leaving the area or conforming to the smoking preference of the other) than to have a shouting match or physical brawl. Thus, the smoker wants to give in if the nonsmoker will insist, but to insist if the nonsmoker will give in; so too for the nonsmoker. There are two equilibria: where the smoker defers and nonsmoker insists and vice versa. If so, then a “smoking prohibited” sign can influence expectations by making focal the outcome where the nonsmoker insists and the smoker gives in. In the designated no-­smoking area, the law “waves on” the nonsmoker to proceed as he wishes and the smoker believes it is more likely the nonsmoker will insist. In a designated smoking area, the law (or other designator) creates the opposite expectations. Especially if people understand the law to create “rights” and believe they appear weak if they fail to stand up for their rights, the law emboldens those it favors to insist on getting their way.56 The law makes it seem more likely that the legally disfavored party can avoid the ugly shouting match only by backing down. Like the driving examples, smoking regulations often involve signage. But the government might make one outcome focal without posting a physical sign merely by endorsing it in a statute or judicial decision. The newspaper may carry the news: city council bans smoking in shopping malls or bus stations. Like any compliance theory, the claim requires that the law be well publicized. But if the legal rule is mutually salient to the disputants, then the claim of influence is exactly the same. 85

The Expressive Powers of Law

As an example of this kind of influence, without signage, consider a nonlegal example from a public statement about a private conflict over the use of public space. Former Chicago Mayor Richard M. Dailey once defended the rights of those who dig parking spaces out of the snow in the morning to return to those spaces in the afternoon by saying “This is Chicago. Fair warning.”57 This statement “waves on” the person who dug out the space to insist on his informal rights to it (even predicting property damage or violence). In all likelihood, Mayor Daily changed the expectations of Chicago citizens about the likelihood of a costly result if one were to take a space someone else dug out, making it more likely that people would avoid spaces others dug out. If instead the city council had passed an ordinance (without sanctions) stating drivers have no right to residential parking when others have recently cleared the space of snow, my claim is that it would have had essentially the same expressive effect. The effect does not depend on the existence of public signage.58 As a legal example, consider state laws declaring the right of mothers to breastfeed their infants.59 When a government employee tells a woman in a courthouse that she may not breastfeed in a hallway or a restaurant manager tells a patron that she must take her infant to the restroom to feed him, we have a dispute over the use of space that is much like the smoking dispute. The conflicting preferences may be the employee’s or manager’s or these agents may be acting out of concern for others in the courthouse or restaurant who object to this type of public exposure. There is now the potential for a mutually bad outcome, if the disagreement leads to a shouting match (or in this case, perhaps, a crying baby). Like the Bystander directing traffic, these laws recommend a particular solution to these competing claims, one that favors the mother.60 If the law is well publicized, it can make focal the outcome where the mother insists and the objector gives in (or never complains), creating self-­fulfilling expectations that this result will occur. Consider more conventional property disputes. Following custom, a beachgoer wants to sit on a dry sandy part of the beach formally owned by a private party, who objects. Or the beachgoer wants to walk over a small part of the owner’s land, far from any structure, to access a public part of the beach and again the owner objects. Or consider disputes unique to neighbors: There is a tree planted on one side of the property line, but the neighbor on the other side wishes to cut branches hanging over the line onto his property. One neighbor wishes to build a structure that will block 86

L AW AS FO C A L POIN T

the scenic view of the other. Or one neighbor refuses to change the steep grade of his property, which is causing rainwater runoff and destructive erosion to the other. The law has a rule for each of these conflicts, but let us set aside the effect of legal sanctions by assuming the frequent situation where the police will not get involved and the courts are too expensive for either neighbor to consider. Even in mature legal systems, people often know that they will work out their dispute without the state.61 Despite the absence of sanctions, law can work expressively as it did in the experiments and examples previously discussed. If the neighbors both know (and believe that the other knows, etc.) that the law says, for example, that a property owner has the right to cut tree branches hanging over his property, even when the tree grows on neighboring property, then it is as if the state put up a sign near the property line telling the other neighbor to “yield.” When the dispute involves an element of coordination, the disputants are subject to focal influences and the law makes focal the outcome it demands. Legal expression can create self-­fulfilling expectations of what the players will do, generating compliance independent of sanctions or legitimacy. By describing property disputes with a metaphor involving Hawks and Doves, I am implicitly analogizing between human and nonhuman behavior. One might guess that the analogy is inapt because nonhuman animals do not respond to symbolic features of their environment, but only to simple factors, like relative body size, that predict who will win a fight. As it happens, however, even nonhuman animals use naturally occurring conspicuous landmarks to determine the territory they will fight to defend.62 These territorial boundaries are too persistent to be explained fully by the fighting ability of the animals. “Boundaries often remain essentially unchanged when one neighbor is replaced. In territories defended by mixed-­species flocks of birds, the boundary may move little, if at all, despite eventual turnover of the entire membership of one of the resident groups.”63 As two biologists conclude, “it is possible that the previous position of the boundary marks a mutually obvious solution to the coordination problem that aggressive neighbors face and that the residents continue to respect this boundary as a convention.”64 There is something so fundamental about salient boundaries that it affects how some nonhuman animals behave, how they avoid disputes. Some readers are put off by analogies between human and nonhuman behavior. But the virtue of animal studies is that the rival hypothesis of 87

The Expressive Powers of Law

legitimacy is not so plausible an explanation of nonhuman behavior. Indeed, if we didn’t observe this pattern of behavior in animals, it would count against the focal point theory. A legitimacy theorist might reasonably say that if humans alone perceive legitimacy and humans alone defer to purely symbolic territorial boundaries, then legitimacy must explain the power of such boundaries. There are, of course, fundamental differences in the causes of human and nonhuman behavior. Rationality allows humans to grasp the problem of coordination. Humans can reason consciously toward a solution by considering the problem from the perspective of other humans. As noted before, when humans try to match naming a flower, they reason about what others will think to name. There is an academic debate over whether any nonhuman animals possess the “theory of mind” necessary to engage in even the most rudimentary form of this type of reasoning.65 If not (as seems reasonable for at least some of the animals that are sensitive to natural boundary markers), then the behavior is more likely instinctual. As instinct governs a spider spinning a web, animals may respond to natural boundary markers without cognition, much less by forming beliefs about the beliefs of others. Yet the animal studies are still revealing for human behavior. There is a debate about whether focal points can be reduced entirely to rationality. I take the position that rationality is insufficient to explain what is focal or psychologically “prominent.” That one million dollars is more salient than other amounts of money depends, as Schelling claimed, on arbitrary (payoff neutral) factors from psychology, culture, and history. Rationality does not guarantee salience because a rational being might look at the coordination alternatives and find them all equally salient, and expect others to find them equally salient, in which case there is no basis for reasoning about what other human minds will perceive as salient. There would be no focal point. Yet human psychology is hardly exhausted by rationality; humans can predict what arbitrary features others will find noticeable or prominent in a given situation. It is this predisposition to notice particular elements in the environment that humans share with other animals. Constituting neither a threat nor a source of food or material for a nest, the natural objects that serve as boundaries in these studies are generally irrelevant to the animals’ lives. Outside of their capacity to facilitate coordination, they do not affect payoffs. Yet natural markers are nonetheless focal for many birds. Humans too can 88

L AW AS FO C A L POIN T

perceive some arbitrary features of their environment as being salient and, more importantly for coordination, humans tend to find the same features to be salient (despite not being determined by rationality). Indeed, the animal studies are particularly interesting because the human scientists who conducted the studies, like those of us who read the studies, immediately grasp what features of the natural environment the birds are reacting to. Thus, even though humans may be distinctive in their ability to reason about the use of arbitrary features of the environment, humans are like other animals in finding certain arbitrary features to be mutually salient. Consider then a final data point, an astonishing experiment of symbolic landmarks on animals sharing a space.66 Biologists placed two male/ female pairs of blockhead fish of similar size in an aquarium, each pair at a different end that included a cave structure. Initially, they divided the aquarium in half by an opaque barrier so each fish pair could develop territorial attachment without interacting with the other pair. In the control, there were no landmarks on the bottom, which was a consistent layer of gravel. In the landmark condition, the researchers placed a row of three large flat rocks along the bottom of the aquarium that formed a line dividing the aquarium in half. The rocks look different than the gravel, but were level with it, so they were purely symbolic. The biologists then removed the barrier between the pairs of fish. The results were striking. In the control, in fourteen of fifteen cases, there was sufficient fighting that one pair took over the entire aquarium within about thirty minutes, driving the other pair to swim near the surface, which is complete surrender because these fish are bottom feeders. With landmarks, however, all fifteen pairs established territories within their half of the aquarium, mostly staying on their side (92 percent of the time) with very little fighting. None was driven to the surface. The creation of the simple asymmetry of a visible landmark does not change the expected outcome of a fight, nor is it plausible the fish defer to the landmark out of a sense of legitimacy. But the visible boundary nonetheless dramatically changes the territory the fish end up with, shrinking by half what the stronger pair would have obtained by fighting. The authors note that the results confirm other published field observations finding that “territorial individuals that use visual landmarks generally appear to set their boundaries at conspicuous features of the landscape.”67 Thus, boundaries are focal points for resolving a game of conflict and coor­ d ­ ination. Where nonhuman animals take naturally occurring landmarks 89

The Expressive Powers of Law

as boundaries, humans consciously create boundaries and construct landmarks. Some boundaries, like high walls, also change the payoffs by making it harder to cross the space. But a low stone wall or a row of trees may work, without changing the payoffs, by symbolizing the existence of a boundary. Recall that Schelling’s subjects, acting as “generals,” used topographical features on a map to determine where to stop advancing their armies, so as to avoid war. In the Isoni, et al. bargaining experiment reviewed in the last chapter, mere physical proximity created a focal point for the division of resources.68 More relevant for my purposes, legal language creates conceptual boundaries. The law may clearly proclaim that the property rights of a landowner include the right to cut tree branches hanging over his property and to harvest fruit from them, even when the tree grows on neighboring property. Here, the state has erected a metaphorical landmark letting each property owner know how far he is expected to go to fight for his property. In this case, the landmark actually works like the fish experiment, where the two-­dimensional landmark on the bottom suggested a three-­dimensional division of space. This formal legal rule—​ad coelum—​ similarly says that the boundary of the plots of land, already in the minds of the two owners, extends up into space, perpendicular to the earth, bisecting the tree branches at a spot that defines the border of each property-­owner’s claim.69 For other property disputes, the crossing of an imagined extension of the border will not fully suffice to define a wrong, but the intelligence of humans allows them to create conceptual boundaries. The law of nuisance deals with conflict that arises without “trespass,” that is, without a simple and observable crossing of the boundary of property.70 Suppose an individual plays loud music in the middle of the night, preventing neighbors from sleeping. Without an intruder crossing the boundary line, there are two simple but extreme regulatory choices. The law might say that there is no tort—​individuals on their side of the line can make all the noise they want. Or the law might say that the sound waves are like an intruder, their crossing the property is a trespass, and so it is a tort to make any sound perceptible on another property (without consent). Instead, through the tort of nuisance or an ordinance, the law draws a conceptual line defining “loud” noise—​informally or by decibel level—​ which is further refined by temporal lines defining the customary times 90

L AW AS FO C A L POIN T

for sleeping. As a result, the law can articulate a rule that tells the property owner that he may not play loud music from midnight to dawn, despite the fact that he does so on her side of the property boundary and despite the fact that the crossing of “quiet” sound waves is not a tort. Using loudness and time, the law creates a metaphorical boundary around the playing of music and other noisy activities. If the law is sufficiently salient to the neighbors, it works like the no-­smoking rule by “waving on” one party to insist and signaling the other to defer. If the parties mutually regard insist/ insist conflict as the worst outcome, the conceptual boundary is a coordinating focal point that avoids conflict. There are many disputes of this sort—​whether a neighbor creates a nuisance by emitting noxious odors, harmful vibrations, or plants that attract harmful insects. Can an individual erect a structure that blocks the view of a neighbor? Can she regrade her property in a way that directs a harmful amount of groundwater runoff onto her neighbor’s property? In cases of this sort, the law does not always permit the owner to decide on uses within the physical boundary line, but sometimes supplies a conceptual limit on one’s use of property.71 At least where the parties clearly understand the conceptual boundaries defining these restrictions, the metaphorical line they draw creates a focal point solution in the same way that a physical boundary does. A statute or court decree that clearly states a dispositive legal rule can align expectations—​the one favored by the rule will fight and the other will therefore not. Finally, to link this discussion back to international law, one can use something similar to the HD game to explain the puzzle of successful international adjudication, which exists without a credible threat of sanctions. No one would question that disputing states face some risk of armed conflict if they allow a dispute to continue too long. If the stakes of the dispute are low or the cost of violence particularly high, then the worst outcome for each is a violent conflict. If so, an arbiter can influence their behavior by making one equilibrium a focal point (the one where the party the arbiter declares to be the loser gives in to the arbiter’s declared winner). Tom Ginsburg and I use this theory to explain international adjudication, including the puzzling fact that 68 percent of the decisions of the International Court of Justice (ICJ) produce prompt compliance, despite the fact that the ICJ cannot credibly threaten sanctions.72 The compliance rate is high because nations tend to bring to the ICJ those disputes for 91

The Expressive Powers of Law

which a focal point will resolve their dispute (as discussed in the next chapter; I discuss adjudication more generally in Chapter 7).

Conclusion: The Limits of Coordination It bears repeating that, despite my effort to give the focal point theory a broad domain of application, it has substantial limits. Legal commentators make many expressive claims. But most who do so make no effort to identify any element of coordination in the behavioral setting and cannot therefore be claiming a focal point effect. The theory manifestly does not support many such claims because the necessary conditions do not exist. If an individual has a dominant strategy—​best no matter what the other individual does—​then he is beyond the influence of a focal point. For example, in the dispute setting, a person who has no fear of unresolved conflict (as someone who is certain to win a fight) has a dominant strategy to insist on their position no matter what outcome a third party expressively endorses. Or if a thief believes that the probability of detection is low enough, he may have a dominant strategy of stealing whatever he can. There is also no element of coordination if one person can move first without risking the costs of unresolved conflict; in a dispute, she has a dominant strategy of insisting, so the party moving second must back down. This point explains why we often require sanctions to deter an individual from imposing external costs on others: a factory can pollute or a fisher can “overfish” without risk of contemporaneous hawkish responses. By the time victims discover the externality, there is a collective action problem in which each victim prefers to free ride while others punish the polluters or overfisher. Because of these limits, when legal commentators say that new laws against drunk driving or texting while driving “send a message” about the unacceptability of the behavior, or that a new law distributing clean needles to addicts or free condoms to teenagers “sends the wrong message,” it is not because the law (or sanctions) creates a coordinating focal point. The focal point theory not only justifies the need to take expressive claims seriously, but also demands of those making such claims an argument that the conditions of the focal point theory are met in the case under discussion. Or that they specify some other expressive theory, perhaps the information account I explore in Chapters 5 and 6. Or that they concede that their claim has no bearing on behavior. *   *   * 92

L AW AS FO C A L POIN T

There is a broad but not universal scope for the focal point theory. Individuals need to coordinate in the creation of a legal order and the system of legal sanctions. They need to coordinate in international relations, traffic, property, smoking avoidance, and standard setting. The focal point theory explains all of the law’s influence in some of these cases and part of the influence in others.

93

4 Law’s Focal Power in Dynamic Perspective

H

aving set out the focal point theory of law, I now consider and respond to a possible critique. One might object that my analysis assumes a static rather than dynamic perspective. If the law’s focal point power is significant, it must exist in common, recurrent coordination situations, but if the situations are recurrent, then people will find other, non-­legal ways of coordinating. Individuals will not indefinitely fail to coordinate in the absence of legal expression. Furthermore, once individuals find a customary or conventional way of coordinating in these recurrent situations, the convention or custom will be focal and resistant to change. The law may fail to influence behavior when there is a competing focal point, one that has the power of already being “how things are done.” To illustrate, we don’t imagine that drivers would, absent law, never coordinate their driving on a single side of the road. And once a convention emerges (say, driving on the right), it may seem that the law’s power to displace the status quo focal point would require legal sanctions or legitimacy, i.e., that legal expression alone would be insufficient. In this chapter, I address these two concerns. First, if focal points are not necessary to predict behavior in a dynamic setting, then perhaps focal points don’t actually influence behavior in that setting. Second, even if focal points can influence behavior before an equilibrium emerges from evolutionary processes, perhaps legal expression is too weak a tool to change an equilibrium that has emerged. After briefly addressing the first issue, most of the chapter concerns the second, the reply requiring an extended discussion of how law interacts with social movements and custom. 94

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

Reaching Equilibrium Without a Focal Point As others have noted, Schelling’s brilliant insights on the influence of focal points, more than fifty years later, remain outside the mainstream of game theory. It has proved difficult to model focal points formally. This result is not surprising when one recalls that Schelling said that focal points are “inherently empirical”1 and “may depend on imagination more than on logic, more on poetry and humour than mathematics.”2 A few theorists have offered what I regard as partial models of focal points, applicable to a limited domain, and other theorists have convincingly argued that focal points cannot be entirely reduced to a mathematical model.3 Yet in the last half-­century, game theory has found other ways to predict the selection of an equilibrium when many are possible. In the models of evolutionary game theory, involving recurrent situations, evolution typically does not occur through increased sexual reproduction of players who enjoy greater success in the recurrent situation, but by learning, as players mimic the strategies that produced more utility to other players in recent rounds (which may occur without assuming complete rationality). A key move is to give up on predicting which equilibrium will first emerge and instead to focus on which equilibrium will occur most commonly over time, given additional properties of stability only some equilibria possess. To identify those properties, the evolutionary theorist typically assumes that there are constant random shocks or perturbations that threaten any equilibrium. With some small probability each round, an individual deviates from his equilibrium strategy and randomly picks some other strategy. With extraordinarily small probabilities, enough individuals in a population randomly deviate at the same time that the whole system will flip out of the existing equilibrium and into a new one.4 The important result is that some equilibria are much more resilient to these “stochastic shocks” than others; they have a larger “basin of attraction” (in a three-­dimensional graph of the parameters). This new stability is purely a function of player payoffs. One can use payoffs to predict that the system will in the long run spend the great majority of its time in the most stochastically stable equilibrium, rather than any other equilibrium. These theorists then use evolutionary game theory to explore the conventions and customs that constitute informal order. As noted, the evolutionary turn in game theory could pose two challenges for the focal point theory of law. The first is the claim that these 95

The Expressive Powers of Law

theoretical advances might raise doubts about the effectiveness of focal points, that they are not sufficient to influence behavior. Brian Skyrms, for example, raises questions about the perplexing origins of salience and observes that evolutionary dynamics avoid the need for salience or common knowledge.5 This concern requires only a short reply. None of the evolutionary analysis contests the empirical claims I have made about focal points. Evolutionary theory offers no reason to believe that drivers will ignore the hand signals of Schelling’s Bystander, nor to dispute the experimental data reviewed in the prior chapter, including the finding of greater focal influence when current players were told how past players coordinated. The theory just says that these results won’t matter in the “long run” because, in the long run, evolutionary theory predicts behavior by identifying the most stable equilibrium. Yet the span of time necessary for the most stable equilibrium to emerge gives new meaning to Keynes’ quip that “[i]n the long run we are all dead.”6 The time spans could be in the billions of years. Robert Sugden illustrates by asking “How long would it take Britain to switch to driving on the right if we waited for a coincidence of random mistakes by individual drivers?”7 Thus, if focal points influence initial equilibrium selection, that is of enormous practical importance, because the initial equilibrium might endure for generations, even if it has less stochastic stability than some other equilibrium. For human beings, that is enough. Finally, game theory shows that the equilibrium favored by evolution need not be efficient; there is reason therefore to believe that the law could, by focal point (as well as sanctions or legitimacy), produce a more complex and efficient outcome than would otherwise occur.8

The Persistence of an Existing Equilibrium Evolutionary theory offers a second and potentially more serious objection to the focal point theory: once an equilibrium emerges and everyone is playing a best reply to everyone else, that behavioral pattern is focal and may be difficult to dislodge by mere expression. None of the existing experimental evidence tests whether third-­party cheap talk can influence how the players coordinate after they have already settled into an equilibrium. Indeed, the last condition of the focal point theory specified in Chapter 3 is the requirement that there be no focal point competing with the legal focal point, or at least that any competing focal point is weaker 96

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

than the legal one. The objection, however, is that the past practice—​a social norm, custom, or convention—​supplies a competing focal point. So, it might seem that some nonexpressive power, such as a legal sanction, is necessary. This objection identifies an important limitation. The focal point is most likely to work in situations of disequilibrium because a long-­established equilibrium supplies a means of coordinating that may be more focal than a merely expressed alternative. I would not advise the British government to try to change the convention of driving on the left merely by nonbinding expression. Expression alone may fail to convince everyone that everyone else is really going to switch sides of the road, so it is certainly preferable also to invoke sanctions. Let us call this point the “dynamic effectiveness objection.” It requires a more elaborate response than the last. In this chapter I offer two replies. My first reply is relatively brief: There is no a priori reason that an existing equilibrium will be more focal than the one endorsed by a third party (in which the players have mutual salience or common knowledge of the endorsement). It is standard in game theory to assume that once the players reach an equilibrium, they will stick with that equilibrium (absent some event), but this too is a contingent empirical claim. Suppose there are two (pure strategy) equilibria in an iterated coordination game, A/A and B/B. In all rounds before n, the players fail to coordinate. In round n, each player selects A, so they coordinate. What will the players do in round n + 1? Does rationality require that they each choose A? No. Formally, they are in exactly the same position as they were before round n, which is to say that it would be fully rational to select B if you thought the other player was going to select B. It is only because of the precedent of A/A—​where the players coordinated last time—​that the game now seems trivial to solve. The precedent supplies a focal point, so each player may expect the other to be influenced by the precedent.9 But if the continuation of the same equilibrium round to round is merely a matter of focal influence, not pure rationality, then there is no a priori reason that other focal influences—​ such as third-­party cheap talk suggesting the B/B equilibrium—​could not be stronger and dislodge the A/A equilibrium. Gerry Mackie describes a situation I interpret as being this case: in some parts of Africa, the village-­by-­village abandonment of the cultural practice of genital cutting.10 Mackie first describes the element of coordination in the practice (as well as in an older practice of foot-­binding): parents of girls 97

The Expressive Powers of Law

want to follow the convention because they expect parents of boys to allow marriage only to girls who have followed the convention; if most boys follow the convention, parents of girls will fear the deviance of not doing so, given the social and economic costs their daughters incur from not marrying. He then documents that many villages have abandoned cutting by mere expression—​at a public commitment ceremony that plausibly creates common knowledge (or at least mutual salience) of the fact that everyone intends to end the practice at that time—​which flips the village into a new equilibrium. An American example of a mass political ritual is when tens of millions of people watch the televised swearing in of a new president at the inauguration. If leadership is a matter of coordinated expectations—​everyone obeys an individual’s orders on a particular subject because they expect everyone else to do so—​then it is no surprise that leadership transition benefits from a public ritual in which expectations are reformed. Part of the change is to abandon the expectations of obedience to the existing president, to transform him into a former president, while creating new expectations for the president-­elect. The constitution proscribes one element of the ritual: the new president takes the oath of office. By tradition, the Chief Justice of the United States administers the oath at the United States Capitol in the physical presence of Congressional leaders. Cus­ tomarily, the outgoing president, if alive, is in attendance, acknowledging his loss of leadership. There is common knowledge of the ritual among those present. Television viewers can at least see that there is a large crowd in attendance, including various leaders and the former president, all of who are accepting the individual as the lawful president. The events are salient to a large number of Americans, as much as any periodic political event can be. There is no inherent reason why law cannot serve as a similar commitment ceremony to place in doubt the prior equilibrium. Obviously, a large society does not gather around at the announcement of a new law, in most cases, not even by watching its passage on television. But some of the most important laws do gather a significant audience and some of these address matters of coordination. Consider two examples. First, in a state with a substantial coastline full of beaches, quite a large number of citizens, particularly those who own land along the beach and those who do not own land but desire access, would pay close attention to a legal announcement by the state legislature or Supreme Court regarding 98

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

beach access.11 If all the media give prominent coverage to the legal story, they might create mutual salience (or even common knowledge) to the new rule. The pronouncement might then reshape expectations away from the current practice, emboldening one side in particular disputes to demand whatever rights (exclusion or access) the legislature or court recognized (much like the smoking example in the prior chapter). Constitutional amendments are a second example. The well-­known difficulty of amending the federal constitution makes it more likely that any resulting amendments are publicly salient. Consider the twenty-­second amendment. After Franklin Roosevelt was elected to the presidency for four consecutive terms, that amendment sought to re-­establish the previously unwritten constitutional convention limiting presidents to two terms.12 The amendment has been successful, no doubt in part to its simple reshaping of citizen expectations. Recall the prior chapter’s discussion of leadership. The amendment raises doubt whether everyone would accept as the executive leader an electoral winner whom the law designates as ineligible; the doubts are themselves enough of a political liability that the amendment is largely self-­enforcing. In any event, the possibility that law-­as-­ritual might, by expression alone, supplant an existing practice is only my first point. My second and longer response to the dynamic effectiveness objection is that, even if the focal point power is limited to situations of disequilibrium, such situations are exceedingly common. Schelling’s Bystander is a case in point—​the failure of the traffic light disrupts the normal coordination pattern and creates chaos, in which the Bystander’s expression influences driving. More substantially, I previously discussed the effect of new constitutions, which frequently arise in a period of social and political anxiety and change. Technological and cultural change constantly upset previously settled expectations about what others will do. These states of disorder are the kind of situations in which legal change occurs, so a new law need not always compete with an otherwise unchallenged set of settled expectations. For example, Europeans largely drive on the right side of the road. At one time, the custom was for horse-­drawn carriages to travel on the left side of the road, while pedestrians kept to the right, facing oncoming carriages. Yet after the French Revolution unsettled (to put it mildly) a great many conventions, the government decreed the more “democratic” rule that carriages should drive on the right so the common man could walk on the high status side.13 More recently, a new appreciation of the health 99

The Expressive Powers of Law

costs of smoking combined with a decline in smokers created uncertainty about the continuation of the pro-­smoking convention that generally prevailed. It was at this point that the state started to enact anti-­smoking laws. The norms of public breast-­feeding were also unsettled—​neither definitively for or against—​when states started to enact laws declaring the right of young mothers to breast feed their infants in any public place in which they are lawfully present.14 If legal change follows a certain amount of social change, then it will often be plausible that law can work as a focal point to refocus expectations about how to coordinate. This second response merits further attention; indeed, the rest of this chapter is devoted to it. There are two common situations of disequilibrium in which the law’s focal point remains strong. Legal expression can change behavior dynamically: (1) when a social movement seeks to destabilize an existing convention, and (2) when the precise boundaries of a convention are uncertain or ambiguous. I consider these points in turn.

Law’s Expressive Power and Social Movements Consider first the role of law in facilitating social movements that seek to destabilize an existing convention. We are perhaps most familiar with social movements that targeted discriminatory norms, such as the civil rights, women’s rights, and LGBT rights movements. But I mean to refer broadly to the efforts of groups to change social norms and institutions, including law. Some social movements are well organized. The pro-­life movement seeks to preserve and strengthen social norms against abortion, including the overruling of Roe v. Wade. The environmentalist move­­ment seeks to change production and consumption patterns that damage natural resources, while the environmental justice movement seeks to protect the poor from bearing a disproportionate share of pollution harms. Other social movements are more diffused. Some Christians push to expand the recognition of faith or their faith in the “public square.” Some liberal consumers seek to pressure corporations to adopt more progressive labor practices. I return now to a diffused and prosaic “movement” from the last chapter: the successful effort to change pro-­smoking norms that once existed in the United States, partly through the enactment of laws against certain forms of public smoking. I have used the HD game to describe the conflict between a smoker and nonsmoker. In a dynamic setting, the recurrence of the confrontations between pairs of individuals could produce a social 100

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

norm in which the individuals avoid the Hawk/Hawk outcome: for a given public space, there is a possible equilibrium where the nonsmoker always gives in and another equilibrium where the smoker always gives in. A couple of decades ago, the public norm in many parts of the United States always favored the smoker, meaning that most nonsmokers always accepted the activity of smoking. Let us assume that nonsmokers in, say, 1995 wanted to flip the pro-­ smoking norm into one where the smoker gave in, at least for certain public places. What is required for private action to flip the HD norm is for nonsmokers to insist on getting their way with sufficient frequency that smokers come to see it as in their interest to defer, as a way of avoiding Hawk/Hawk outcomes. This particular social movement strategy requires that nonsmokers cause and endure many unpleasant Hawk-­Hawk interactions—​mostly angry looks, rude remarks, or shouting matches—​in the short run in the hope of getting a better norm in the long run. (Smokers might act collectively as well, seeking to maintain the existing norm by showing a willingness to endure Hawk-­Hawk outcomes. Interestingly, the New York City group against smoking regulation is named CLASH, for Citizens Lobbying Against Smoker Harassment.15) The short run altercations are the cost the “committed” individual must bear to produce social change. Let us assume that some ardent nonsmokers are willing to engage in a finite level of these costly confrontations, assuming that enough other nonsmokers will join them that the effort will succeed (otherwise, if everyone free-­rides, the movement obviously fails). The key point is that uncoordinated individual confrontations are less likely to succeed than coordinated individual confrontations. Coordination matters for several reasons, but I will focus on two simple points. First, coordination may matter because there is “strength in numbers.” That is, it is more costly for one nonsmoker to confront five smokers in a public place than to confront one smoker. And it is less costly for smokers to refuse a request to refrain when there are more smokers around and more costly when there are more nonsmokers. In each case, the “strength of numbers” arises because the few expect greater disapproval or perceive a greater threat of verbal abuse or violence from the many. Nonsmokers, for example, are more likely to insist if they appear to outnumber smokers in a particular place than if the smokers outnumber them. If so, then both sides face a classic coordination game—​the “meeting place” problem—​described in Chapter 2. That is, the confrontation 101

The Expressive Powers of Law

between smokers and nonsmokers remains something like a HD game, but the group of nonsmokers (smokers) wants to coordinate on a meeting place so as to achieve the strength of numbers. At an airport, restaurant, or courthouse, where the norm is pro-­smoking, nonsmokers want to congregate together in a particular place (a room, cluster of chairs, etc.) where their greater numbers will allow them to insist on nonsmoking despite the broader norm. My use of the term social movement might conjure up the idea of nonsmokers solving their coordination problem by consciousness-­ raising and direct action. The simplest way to solve a meeting place problem is through direct communication and one might similarly imagine coordinated confrontations occurring because a bunch of nonsmokers meet at a park or gym, talk to each other, and then travel en masse to the airport, restaurant, or courthouse to assert their rights against smokers. For the most part, this is not the way it happened. The nonsmokers generally remained strangers to one another and confronted smokers individually. At the locations of these confrontations, no one knew for sure if a stranger without a cigarette was a nonsmoker—​an ally—​or a smoker who is currently refraining—​the opposition. Without direct communication, how could nonsmokers coordinate? One solution was and is legal expression. If the law forces restaurants, air­ ­ports, and courthouses to designate “no-­smoking” sections, then the signs demarking those sections will facilitate the “meeting” of nonsmokers in a single place. The nonsmokers will seek open seats in the nonsmoking ­section. Thus, even if everyone knows that the state will not sanction individuals who smoke in the no-­smoking sections, the congregation of nonsmokers in an area will drive down the costs of nonsmokers insisting on that area being smoke-­free, and thus drive up the costs of smoking in that area. If smokers and nonsmokers expect that nonsmokers will outnumber the smokers in these areas, the nonsmokers are more likely to insist and the smokers are more likely to defer. Legal expression works by harnessing private enforcement of the law. This expressive theory fits well with some of the actual incremental change in smoking laws. At first, some laws merely sought to segregate the smokers and nonsmokers by creating “no-­smoking” sections. Based on my memory, those sections were initially small and sometimes constituted less desirable seating. Given the existing norm favoring smokers, their smallness and lack of appeal was an advantage because it was unlikely that anyone would ask to sit there except for nonsmokers and that there would 102

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

be enough nonsmokers to fill the space. That selection made it more likely that nonsmokers concentrated there, where they would confront any smokers who wandered in. The norm tipped from “smoking allowed everywhere” to “smoking allowed only outside nonsmoking areas.”16 Once that change occurs, unsettling the existing norm, the law can incrementally expand the size of the no-­smoking sections and still gain compliance.17 At some point, smokers will get used to the fact that they can’t always get seating in the smoking section and they have to refrain despite their preferences. No doubt this change was aided by the fact that smoking rates were declining. Eventually, the law declares the whole inside of the restaurant a no-­smoking area and creates a credible risk that committed nonsmokers (those willing to engage in costly confrontations) will confront anyone in the restaurant, a result not plausible at the outset, when the long-­standing norm unambiguously favored smokers.18 Thus, the state’s ability to provide a coordinating focal point harnesses the private power of informal social sanctions. The law has this expressive power independent of its formal sanctions or legitimacy. Not all social movements, however, could gain such an advantage merely by coordinating on a meeting place. Consider a second, more general reason that coordinated efforts to produce social change are more likely to succeed. I have posited that committed nonsmokers were willing to bear costs for social change but only up to a point. Assume that this limitation has a temporal limit: the nonsmokers are willing to incur only one Hawk-­Hawk outcome per time period. Sane people do not want to be confrontational all the time; they “pick their battles.” If so, then the movement participants may want to coordinate on the timing of their confrontations. To illustrate, suppose that each smoker has some threshold of uncertainty that affects their decision to smoke in a public place. No doubt the individual does not consciously identify a particular numerical threshold, but is sensitive to when the context makes confrontation more or less likely. For the convenience of illustration, however, I will say that the average smoker in some community will insist on smoking if she perceives the risk of a Hawk/Hawk confrontation to be less than 20 percent, but will refrain from smoking or give in to any demand not to smoke if she perceives the risk to exceed that level. Quite clearly, the nonsmokers need to engage in enough confrontations to raise the probability of smokers suffering Hawk/Hawk outcomes to some point above 20 percent. But suppose 103

The Expressive Powers of Law

the number of committed nonsmokers, willing to engage in costly con­ frontations, is simply not large enough to accomplish that goal. Without coordination, committed nonsmokers raise for smokers the probability of confrontation to only, say, 10 percent, which is too low to change smoking behavior. The obvious solution, however, is for nonsmokers to coordinate the timing of their confrontations so as to concentrate their efforts in some way. In theory, they might just choose a time period, say, three o’clock to four o’clock each afternoon. More realistically, they choose a place, such as a bus station or shopping mall. The place is not a meeting place; the nonsmokers are not seeking to congregate there. Instead, they are simply “picking their battles” by deferring in every other setting and confronting smokers only in the designated place, even when they are the only nonsmoker. Outside of the bus station or shopping mall, the smokers see the probability of confrontations decline, but inside that space, the probability jumps well above their threshold of uncertainty and they switch to a Dove strategy, deferring to nonsmokers. As before, the question is how nonsmokers can coordinate on a location if they mostly remain strangers to one another. Again, one solution is legal expression. If the law forces bus stations and shopping malls to designate themselves as “no-­smoking,” then it will provide a coordinating focal point for nonsmokers. Nonsmokers will not waste their confrontational efforts in other places, but concentrate on those designated by law. Even if every­ ­one knows that the state will not sanction individuals who smoke in the no-­smoking sections, and even if the state lacks legitimacy, the willingness of nonsmokers to “battle” in this one area will drive up the costs of smok­­ing in that area. Again, legal expression works by harnessing private enforcement. Of course, this focal point effect is not unique to law. If the anti-­smoking forces were well organized, their leadership would identify the places or times in which to concentrate their confrontational energies. That is the way it worked for many movements, most famously for the sit-­ins and boycotts of the civil rights movement. But for many dispersed or leaderless groups, like nonsmokers, there will be no one who can communicate to enough willing participants to easily create a focal point for concentrating effort. In these cases, the law can provide the expressive leadership and organize the diffused movement. 104

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

This mechanism—​“effort concentration”—​is quite general. It explains “soft law” standards for “green” production and “living wages.”19 There are committed consumers who are willing to spend their money to encourage firms to engage in certain labor and environmental policies. But they can only boycott so many products. If boycotts are uncoordinated, they are completely ineffective. If half the motivated consumers boycott corporation A and buy goods from A’s main economic rival, corporation B, while the other half boycott B and buy from A, each side is undoing the efforts of the other. One way consumers can be unfocused is that there are many different concerns: child labor, degrading and unhealthy labor conditions, discrimination against women or LGBT workers, the appropriation of indigenous art into its designs, animal welfare, etc. Boycotting might be effective only if it is concentrated on one cause. Once elected, another coordination problem is selecting a performance level that identifies the firms to be confronted. Soft law standards do not guarantee coordination, but they offer an expressive means of concentrating effort. The particular examples I have used in this section might imply a normative claim about focal points, that they promote social welfare. I reserve the normative discussion mostly for Chapter 8, but it is here worth noting a counterexample, involving the opposition to a social movement, to emphasize that there is nothing inherently desirable about focal power. Rick Brooks and Carol Rose explain how white homeowners in Chicago used restrictive covenants as a focal point for resisting racial integration in the early-­and mid-­t wentieth century. Brooks and Rose suggest that many white segregationist homeowners faced not a PD—​where they wanted to move from the neighborhood no matter what other whites did—​but a Stag Hunt or Assurance game, where each wanted to “stay put” if their white neighbors did, but sell to the highest bidder—​including black buyers—​if their white neighbors did that. At least for those who don’t have an independent reason to move (like a job transfer), the mutually best outcome was for all to stay put, but staying was risky because, if others sold to black buyers, given white discriminatory preferences, the property values would fall before the white homeowner could sell. As with the bank deposit game in Chapter 2, if everyone assures each other that they will stay, there will be no panicked rush to leave. At one time, the tools for assurance included restrictive covenants that legally forbade homeowners from selling to nonwhites. One might 105

The Expressive Powers of Law

think that covenants worked by legal deterrence or incapacitation, but Brooks and Rose say otherwise. Given the legal costs and rarity of judicial enforcement, the covenants “appear to have been more significant as expressive focal points than as legal enforcement devices.”20 Part of the argument for this interpretation is what happened after Shelley v. Kraemer held enforcement of these covenants unconstitutional.21 At this point, white segregationists in old, close-­knit neighborhoods could still assure each other and exclude nonwhites, relying on informal mechanisms, sometimes violence. But in newer or less cohesive neighborhoods, the unenforceable racially restrictive covenants were of continued significance in supporting and stabilizing segregation. The covenants continued to demark the boundaries of a neighborhood committed to racial segregation, attracting segregationist whites (like nonsmokers in the previous example), repelling more tolerant white and black buyers (like smokers expecting less approval in the previous example), and coordinating the actions of real estate agents and federal housing agencies (who steered home buyers and selectively provided loans), all towards decades of continued racial exclusivity (until other changes in federal hous­ ­ing law).22 In sum, evolutionary game theory leaves plenty of room for the dynamic influence of focal points, when law interacts with ongoing social change. Now consider a second dynamic possibility.

The Legal Clarification of Custom Law commonly intervenes in disequilibrium when it renders a convention more precise.23 The vast literature on the relationship between law and custom addresses a variety of claims about the virtue or defect of custom, its influence on or rejection by law, and the manner of proving custom in particular cases.24 My claim is narrow and distinct: that law aids compliance with custom by expressively clarifying its demands. As a baseline, considers what happens if the legislature or court adopts and incorporates a custom in all its particulars into the law. The new law might matter for behavior because legal sanctions and/or legitimacy (or the information effect described in the next two chapters) increase obedience to the customary rule. The law is causing an added increment of compliant behavior, but I do not mention the focal point here as a contributing factor because, by being well enough known to be a “custom” before it was incorporated into law, the customary rule was, in all likelihood, 106

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

already focal. To bring in the focal point effect, we have to consider a different case, one where the law does not incorporate the custom in all its particulars, but roughly adopts the customary rule while introducing some modifications that sharpen its meaning. Where the custom was previously ambiguous, there is no powerful focal point for behavior competing with the new legal rule, so the clarifying rule can influence behavior by enabling coordination. Consider two non-­legal illustrations that set up the analysis of this section. In the introduction, I discussed Schelling’s deceptively simple example of the white or yellow line down the center of the road. In most places, the custom of driving on the right (or left) side of the road predates the legal requirement. The line the state paints on the road “expresses” where one lane ends and the other begins. By defining a boundary, the line clarifies the custom. No doubt, the custom is clear enough most of the time. But when the road is unmarked and people drive cars around curves, or can’t see the road well at night, or just fail to pay close attention, they may take up more than half the road and cause an accident. The center line defines more clearly what the custom demands, making it easier to avoid accidents. A closer analogy for the law’s clarification function is the publication of a dictionary. The philosopher David Lewis gave a brilliant and seminal account of language as convention, saying that language solves a recurrent coordination game.25 There is a common interest in communicating and yet an infinite number of ways that human sounds and symbols could map onto the elements of the world. In most cases, everyone wants to use the same sounds and symbols that everyone else uses for a given referent, so as to understand others and be understood. Language first emerges as a complex type of informal order. Imagine the result when someone publishes the first dictionary and it is popular. A dictionary can influence linguistic behavior—​usage, spelling, and pronunciation—​by resolving ambiguity and conflict at the margin of the convention. Whenever there is disagreement over the linguistic rules, the dictionary supplies an easy way to resolve them and thereby to continue to coordinate in matters of communication.26 Past linguistic practice obviously determines the content of the dictionary, but my point is that the causation is inevitably in both directions. For a coordination task as complex as communication, there are always ambiguities in the convention: variations across subpopulations, old usages 107

The Expressive Powers of Law

bordering on desuetude, and neologisms that may or may not take hold. So informal order does not fully solve the coordination game to begin with. If it did—​if the language had settled into an equilibrium in which everyone used every sound or symbol in the same way that everyone else did—​then no native speaker of a language would ever need to consult a dictionary for that language. Where there is ambiguity and disagreement, a word compendium reveals to offers a focal point solution. (The dictionary also works by revealing information, a theory discussed in the next two chapters). Recall from Chapter 1 the experiments showing that individuals can learn over time how to coordinate by observing how others have coordinated. The dictionary can show individuals that some of their usage was statistically deviant; most want to switch to the dominant practice in order to maximize their chances for effective communication. Given multiple dictionaries, each one may influence linguistic behavior in proportion to its sales. Most obviously, the dictionary’s influence is not due to the author’s threatening to sanction individuals for deviating from his dictionary. Nor, in the normal case, is the influence a matter of legitimacy (unless that concept is stretched so far as to be meaningless). The author may make claims for the authoritativeness (completeness and accuracy) of his dictionary, but he is ultimately selling a product. The author usually has no pre-­existing authority, is not selected by some fair process, and is not necessarily motivated by the public welfare but a desire for profit or fame. Nor do members of the public know that the author followed some legitimate process for deciding among competing usages or spellings. Perhaps the author arbitrarily favored some region or his own preferences. Instead, the dictionary works by providing a coordinating focal point. By picking among plausible options and writing them down, a popular text causes people to converge on what the dictionary author prescribes. Finally, note that a dictionary has this effect even though communication is not a pure coordination game. People usually prefer to retain the usage or spelling they have previously adopted, so language standards are like a BOS game in which each person prefers that others adopt their usage rather than the other way around. Yet people often give in to the dictionary. They use dictionaries to settle bets over spelling (of words like “liaison” and “phlegm”), where there is a clear divergence of preferences. 108

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

My thesis is that the written formalization of customary rules into law influences behavior in the same way that the creation of dictionaries influences language. The starting point for this idea is the large rational choice literature on informal order—​social norms, customs, conventions, private rule making, and the like.27 From this literature, I take a standard idea: in repeated interactions between individuals, a complex behavioral equilibrium can emerge in which most people follow certain “rules” most of the time. The rules are, to some degree, self-­enforcing, that is, there are informal sanctions for violations that sustain the rule. To this point, I add the fact that, inevitably, there remains a degree of disequilibrium at the margin of informal order, a setting where expectations about behavior are not perfectly aligned. In these situations, there is rule-­following behavior, but the rules being followed are partly ambiguous, leading to occasional disagreements about application. Given ambiguity, there is no focal point, which creates an opportunity for expressive influence. A legal actor can influence the behavior of the parties following the rule merely by announcing an unambiguous sub-­rule, one that clarifies the ambiguous custom. Legal codification restates the informal rules more precisely and common law courts offer a mechanism for ongoing rule refinement. The same informal mechanisms that produce compliance with the decentralized convention can then produce compliance with the formally clarified custom, the more precise focal point. In sum, there are two points: (1) the inevitable ambiguity in informal order and (2) the law’s ability to clarify informal order. The Inevitable Ambiguity in Informal Order Despite the presence of a custom, there always remains the possibility that a situation will occur in which the players lack common expectations about what each other will do. As Gerald Postema explains, while “the standard situations needing coordination are provided for,” “[t]here may . . . ​ be substantial differences of opinion in the community regarding what the convention requires in some specific instances.”28 Without common expectations, the players will fail to coordinate their behavior in these instances. The inevitability of ambiguity is an old idea in law. We see one version in the standard claim that contracts are “incomplete.” There are several reasons to expect ambiguity not only in intentional agreements, but also 109

The Expressive Powers of Law

in unintentional, spontaneous order. The first problem is local variation, that different communities follow slightly different customs. As Henry Smith puts it, custom, like language, “evolves into various dialects.”29 As a consequence, actual efforts to incorporate custom into law involve a necessary transformation—​generalizing and formalizing the custom so it can be applied to a wider population than the one in which it originated.30Even within a given community, there is a second problem. At least at the margin, a custom is inherently ambiguous because (i) it is based on fuzzy concepts and (ii) there is uncertainty about its completeness. I illustrate with a simple property convention based on possession. Hume was the first to state that property might arise as an informal and decentralized convention in which everyone defers to the claims of first possessors.31 Modern theorists have modeled property as the result of an iterated HD game, where a convention emerges in which (almost) everyone defers to first possessors in order to avoid the mutually destructive Hawk/ Hawk outcome.32 Possession is not the same as property, so the theory requires refinement, but the conventional respect for possession may start in motion a process that ends up as the institution of property.33 This theory is useful for explaining why those claiming to own things often seek to communicate their claim to the world by creating boundary markings or other physical markers on land.34 This sort of communication avoids misunderstandings about who is in possession, which is necessary to avoid a mutually worst outcome like violence. Law is often formalistic in this context because the audience for property claims is extraordinarily large and heterogeneous—​the entire world—​and the costs of communicating to such a population favor simple rules less dependent on context.35 As Henry Smith describes the point: “[T]he use of a thing as a focal point for the right allows those in a right-­duty relationship of potential conflict to interact in an anonymous, informationally unintensive fashion.”36 Despite these efforts, possession often remains ambiguous.37 First, its boundaries may be fuzzy.38 What possession is depends on the conceptual categories the individuals use, but those categories are less than perfectly defined. For example, there may be rights given to first possessors, but disagreement over when possession first occurs. One of the more famous property cases in Anglo-­American law poses this question regarding a wild fox. In Pierson v. Post,39 Post hunted a fox for a time and was about to shoot it when Pierson appeared, killed the fox, and took it. If possession required killing or physical grabbing, then Pierson was the first possessor; if some 110

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

broader concept of control (or the inevitability of physical grabbing) is sufficient, Post was the first possessor.40 The same issue arises where individuals attempt to transform into personal property other wild animals—​whales in particular—​or previously unclaimed gas reserves or even manure.41 Similarly, nations disputing a territory often have different versions of the precise point at which possession of the territory first occurred. One can imagine different answers: when a nation’s citizen first discover and state a claim to the land, when the national flag is first planted on some part of the land, when some of the nation’s citizens first settle on some part of the land, or when the nation’s courts first exercise jurisdiction over individuals living on the land.42 In disputes of this sort, the boundary defining the convention is fuzzy, and therefore, the convention is, for the moment, ambiguous. In certain cases, the ambiguity prevents there being any clear expectation about what a given person will do. Thus, the individual may fail to coordinate and end up in a mutually bad outcome (like mutual defection in a PD game where there is more than one way of cooperating or a Hawk/Hawk fight in a HD game). If the ambiguity in the roles of “original possessor” and “original nonpossessor” were sufficiently severe, it would be unlikely that any convention would arise based on that asymmetry. But the convention might arise because the asymmetry works in what Postema calls the “standard case,” while ambiguity arises only in the residual case. Ambiguity exists for a second reason. Even if the boundaries of the underlying concepts are perfectly clear, the convention is always potentially incomplete. That is, there may be uncertainty about whether the set of expectations constituting a convention is subject to exception. Whenever we seek to apply a rule we followed in the past to the present situation, we must attend to distinctions. There will always be distinctions because no two “situations” are ever exactly alike, but the open question is whether the distinctions matter.43 Suppose that A is unquestionably the first possessor of a plot of land, but that he vacates the area, after which B occupies the land openly for many years. Suppose A is silent about B’s use of the land for this time, but then reasserts his claim, demanding that B vacate the plot. Given the first possession convention, will the nonoriginal possessor (B) give in (play Dove)? B will expect A to aggressively insist (play Hawk) if he assumes that the only circumstances relevant to A’s behavior are those embodied in the concept of first possession. But it is possible that the players will mutually 111

The Expressive Powers of Law

take account of A’s long absence and acquiescence in B’s possession. The convention that eventually emerges might instead be deference to the first possessor except when the current possessor has claimed and occupied the territory for more than, say, twenty years, in which case everyone defers to the current possessor. In other words, we might get the convention similar to what the common law terms adverse possession and what international law calls “acquisitive prescription.”44 There is, however, no guarantee that the extended occupancy exception will arise or, if it does, that the crucial time period will be twenty years. As a result, when these facts first occur, there is uncertainty about whether the new circumstance matters.45 Although repetition of the situation will eventually resolve an ambiguity, conflicting beliefs about the relevance of the new circumstances will produce conflict for some time period, as where each plays Hawk expecting the other to play Dove.46 Law Resolves Ambiguity and Influences Behavior Where law incorporates a self-­enforcing custom, it may generate compliance merely by the act of clarification, i.e., by removing ambiguity in how the custom applies to a particular situation. The law clarifies a custom by providing supplemental rules. If the legal supplement is salient, it works as a focal point, aligning the parties’ expectations and influencing behavior, all independent of sanctions or legitimacy. We saw this clarification effect in the dictionary example, where individuals adjust their linguistic behavior to the dictionary, even when they prefer a different convention. The law works this way as well. Garrett and Weingast use this idea to explain international law—​treaties are contracts that define what constitutes cooperation in an iterated PD.47 Because contracts are incomplete, some event for which the contract is not fully specified may cause the parties’ expectations to diverge, but legal expression can re-­align them. This section offers additional examples of legal clarification. Let us begin with organizational rules, what we might call private “soft law.” Organizations dedicated to games or sports frequently create a statement of rules. These rules can affect gaming behavior just as a dictionary affects linguistic behavior, by offering a coordinating focal point. According to Hoyle is a book stating the rules of whist and other card games, which became authoritative despite its lack of pre-­existing legitimacy or an external enforcement mechanism.48 The book was so influential that the phrase “according to Hoyle” has come to mean in strict accordance with the rules. 112

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

The dominant international chess organization—​Fédération Interna­­ tionale des Échecs (FIDE) provides an authoritative codification of the “Laws of Chess.”49 FIDE can sanction members in competitions, but it is implausible to think that the Laws of Chess have no influence on casual players for whom FIDE has no sanctioning leverage. The private codification works by making focal a particular clarification of the custom in a situation where the parties seek some basis for coordination. For example, there is a customary rule that a game is drawn after a certain number of consecutive moves not involving a pawn and without a capture. But how many moves? One would still be playing the game of “chess” even if one ignored the answer in rule 5.2(e) of the FIDE laws of chess—​fifty such moves—​but that answer is still likely to influence disputing players who seek some way to conclude their game on mutually acceptable terms. Nor would it likely matter to the players how the rules were created, fairly or otherwise, or whether FIDE were regarded as an exemplary or corrupt organization. One takes one’s focal point where one finds it. In the United States, perhaps the oldest example of privately provided soft law is Robert’s Rules of Order (RRO), which purports to define parliamentary procedure for the meetings of professional associations, fraternal organizations, and local governments. RRO sometimes influences the conduct of organizational meetings even when those organizations have not adopted the rules in advance of some procedural dispute.50 Even if the organization is too informal to have bylaws, e.g., a meeting of neighbors to discuss a recent zoning change, invocation of RRO could easily cause an angry participant to back down in some procedural dispute. Obviously, there is no centralized enforcer of RRO. Nor is there any obvious reason for individuals to defer to its legitimacy if the organization has not adopted it in advance. Few people have any idea of the motives or methods of the creator and publisher of RRO. Instead, if protracted conflict is costly, then the highly salient codification of custom influences behavior merely because most people recognize that there is no better way to proceed. A government may also codify soft law.51 An obvious example—​because there are no sanctions—​is the federal statutory law describing customs for the display and treatment of the American flag. The provisions introduce themselves as a “codification of existing rules and customs pertaining to the display and use of the flag . . . ​for the use of such civilians . . . ​as may not be required to conform with regulations promulgated by” executive departments.52 What follows are detailed regulations on the “time and occasions 113

The Expressive Powers of Law

for display,” “position and manner of display,” “respect for the flag,” and “conduct during hoisting, lowering, or passing of flag.”53 The statute speaks to those who desire that they and others engage in conduct that honors the flag, which means they seek to coordinate on a set of behavior that expresses respect. Customs work to some extent on their own, but inevitably there are regional differences and marginal ambiguities, which the codification resolves.54 Those unmotivated to respect the flag will be unaffected, but for those who seek coordination, the code improves upon the custom. The point about codification works also for ordinary “hard” law, the kind with legal sanctions. A statute could incorporate the custom of first possession—​property—​but in so doing also define more clearly what it means to be the first possessor of a wild fox or a parcel of land. The same law could clarify the custom by stating and defining the relevant exceptions. In stating the parameters of the adverse possession exception, a statute could delineate precisely the amount of time in which a person must possess real property—​say, twenty years—​in order to divest the first possessor of ownership. The codification also removes uncertainty over incompleteness by stating that the exceptions to the first possession are limited to the enumerated exceptions.55 The same forces that created and enforced the custom will now apply to the clarified custom, even though it is also enforced by legal sanctions. Sometimes legislation simply adopts custom by reference, without specifying it, leaving to adjudication the task of clarification.56 Judicial expression offers a second way to sharpen the common understanding of what formal or informal rules require. In Chapter 7, I focus on how adjudication can, if sufficiently publicized, influence not only the parties whose dispute is resolved, but also the behavior of others in the community in which the convention exists who may in the future find themselves in a similar dispute. Thus, Pierson v. Post, if publicized, tends to support the expectations that a hunter in the position of Pierson (who is the first to kill the animal) will play Hawk and that a hunter in the position of Post (who had started the hunt and nearly captured the animal) will play Dove. If the case genuinely was one of first impression, the opinion could create such expectations in parties involved in similar disputes in the future.57 In the international arena, we see private NGO and governmental groups create soft law.58 The theory presented here gives an explanation for the effectiveness of such law, as least where it clarifies some existing custom or provides a basis for the parties’ intended cooperation. I would 114

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

even say that privately written legal treatises might influence behavior by clarifying custom. The content of customary international law is notoriously ambiguous. Perhaps no treatise is as influential as RRO is in its domain. But at different times in the past, individuals have published compendia of particular areas of international custom, which have influenced the judicial understanding of customary international law.59 The theory here predicts that those treatises may influence behavior merely by clarifying the existing custom. Stating a clear rule of territorial possession may cause nations to avoid territorial conflict.60 Once conflict begins, one might think there is no role for coordination, but, as my final example, consider the soft law of the “Lieber Code” of 1863. Written by Francis Leiber and promulgated by Abraham Lincoln, this document (formally known as “General Orders No. 100, Instructions for the Government of Armies of the United States in the Field”) provided perhaps the first comprehensive codification of the laws of war.61 It is difficult to prove that the codification affected military behavior (especially as it was written to permit the Union to fight the Civil War as Lincoln desired), or if it did, that the reason was the expressive power of law, as opposed to sanctions or legitimacy. Yet there are elements of the Code that are obviously about coordination and could work as the focal point clarification of custom. Article 115 provided: “It is customary to designate by certain flags (usually yellow) the hospitals in places which are shelled, so that the besieging enemy may avoid firing on them.” When two parties want to communicate, even adversaries in war, they must coordinate on a common language. On a battlefield, one requires particularly simple and clear symbols. The Code here supplies the coordinating symbol, so that there might be no ambiguity to either side. Fascinating is the fact that Lieber only states the “usual” color—​ yellow; the article offers no prescription. As in my experiments with Janice Nadler, however, one can increase coordination merely by stating to the parties how others have achieved coordination in the past. But why would anyone use or respect a yellow flag, or any other means specified by a Code, for designating a battlefield hospital? Even military adversaries might cooperate on certain narrow aspects of their war, the most obvious being the taking of prisoners of war for later exchange or release. The conventional game theory explanation is that the opposing armies are in an iterated PD in which each side fears that defection will cause the other side to defect in the next round.62 One keeps prisoners 115

The Expressive Powers of Law

alive and refrains from attacking hospitals to induce the other side to keep prisoners alive and to refrain from attacking hospitals. Yet there might be more than one way to define the reciprocal obligations towards prisoners or the wounded. And there might be different customs among different peoples, or in the American Civil War, among the same peoples. A well-­ publicized code (Lieber’s Code was sent to the Confederacy) might clarify the custom and align expectations. The Code doesn’t provide the reasons to comply—​that comes from the structure of repeated interaction—​but it works by facilitating the cooperation the parties mutually desire; it helps the parties coordinate on the means of cooperation. As another example, Article 104 of the Leiber Code states: “A successful spy . . . ​, safely returned to his own army, and afterwards captured as an enemy, is not subject to punishment for his acts as a spy . . . .” The background here is the Code’s strong distinction between ordinary combatants and spies, which mostly depends on the visible marker of a uniform. There is an obligation (nonabsolute) to maintain the life of ordinary prisoners, but a right to execute spies. Initially, the distinction allows each side to maintain cooperation on prisoners of war despite the killing of prisoners designated as spies, by making clear that the latter is permissible; it is not a form of “defection” in the PD game. Yet there remains ambiguity: What about former spies, who having completed their spying mission, rejoined the regular forces and are captured in uniform? This is the kind of exceptional case that might not arise very often, making it easy for different parties to have different expectations, so that one side executes a former spy believing it consistent with protecting ordinary prisoners, while the other side regards the action as a violation, which it must reciprocate by retaliating against its prisoners. To head off this unraveling of cooperation, the Code states a solution, resolving the ambiguity and aligning expectations. Not all, but many of the provisions seem to have this purpose. For example, while respect for flags of truce is vital—​Article 114 says that the “character of a flag of truce is sacred”—​Article 112 states that “Firing is not required to cease on the appearance of a flag of truce in battle” (emphasis added). Article 113 provides: “If the bearer of a flag of truce, presenting himself during an engagement, is killed or wounded, it furnishes no ground of complaint whatever” (emphasis added). Again, to preserve reciprocal cooperation on flags of truces, it is essential to have common expectations for when such flags need not be honored. 116

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

No doubt, the focal effect of these and other rule clarifications is difficult to quantify. The abstract experiments discussed in the prior chapter provides some evidence of the magnitude of these effects, but the literature is at too early a stage to say how strong the effect is in particular real world settings. But the theory here does suggest that mere legal expression could influence that behavior of state actors, even those at war, by clarifying the customs they were already inclined to follow. Now that I have fully stated the focal point theory, I will revisit the conventional theories of legal compliance—​sanctions and legitimacy—​and show the importance of focal points to each. Both of these discussions return us to the beginning of the last chapter where I argued that the foundations of a legal order are a focal point solution to a coordination game.

Coordination and Endogenous Legal Sanctions Now that we have discussed informal order, we can give a better account of how sanctions arise endogenously. The literature on informal order gives various reasons that individuals might reach an equilibrium where they sanction each other for certain behavior, as, for example, individuals socially ostracizing or refusing to trade with thieves.63 What I have discussed so far is “second-­party” sanctioning—​as where the victim punishes the thief (perhaps in the process of preventing the theft). All my examples analogizing to the HD game involve some interaction where one party to a dispute inflicts some harm on the other (also harming himself, at least in expectation). But sometimes the sanctioning evolves to a “third-­party” system because even those who don’t suffer directly from the norm violation punish it.64 Various norms theories explain why third-­parties might cut off social and economic ties with the thief or seize his property. Here then is the relevance to the present discussion: Effective third-­party sanctioning requires coordination. As Gillian Hadfield and Barry Weingast draw the distinction, all the prior examples I have given involve coordination in “primary behavior.”65 Now we are turning to coordination in secondary behavior—​the behavior of sanctioning primary behavior. Imagine that A and B each claim ownership of a ring, which is now in the possession of B. Perhaps A, the original owner, lost the ring and B found it; or perhaps a thief stole the ring from A and sold it to B, who believed in good faith that the thief was the owner (and the thief has since disappeared). In a population in which third-­party sanctioning exists to enforce property norms, each member wants to sanction B if his refusal is 117

The Expressive Powers of Law

to give back the ring to A is a violation of the norm. Similarly, if A should use stealth or trickery to regain possession of the ring from B, each member will want to want to sanction A if B is the true owner. There are various arguments sounding in fairness and efficiency to choose between A and B in these circumstances. The selection of one rule (finders, keepers) over another (original owner prevails) is important. My point, however, is the selection among rules is itself a coordination game. A simple model for this sort of standard-­setting—​defining who owns something in cases like this—​is a BOS game because some members of the population prefer one standard and others prefer the other standard. Yet most members may have a strong interest in coordinating on a single standard defining whom to sanction, to avoid the situation where half the population sanctions A for retaking the object and half sanctions B for not giving it back promptly, thus providing no more punishment of the “thief” than the “owner” (under either view) and enforcing no coherent pattern of behavior. When it is important to people that they agree on enforcement, there is room for expressive influence. A single public announcement that the original owner prevails and B is the thief offers a way to coordinate sanctions, directing collective efforts toward punishing B. Of course, A and B might also try to use cheap talk, each seeking to direct punishment towards the other. Or third-­parties might state their preferred rule. The problem with cheap talk solutions is that, being cheap, there will be too many of them. Society can coordinate its third-­party sanctions only if it can solve the higher-­order coordination problem: Upon which of the many third-­ party expressions should everyone base their sanctioning behavior? This is the problem of leadership, discussed at the beginning of Chapter 3. Too many people would like to be the leader, but everyone sees the advantage of having a single leader. Solving the second-­order problem, a leader emerges, one whose expression, being the common focus of attention, is salient. The leader might have legitimacy, but her power is not limited to legitimacy because a leader’s proclamations offer to the group a way to coordinate its third-­party sanctioning. Given ambiguity, the designated speaker creates the necessary focal point to coordinate third-­party sanctions; like a music conductor, he directs the sanctions. At first, his power may be strictly limited to the infrequent cases of ambiguity, where the followers know they would otherwise sanction in an inconsistent fashion. But we should not be surprised if, over time, 118

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

individuals ensure the smooth coordination of third-­party sanctioning by always checking with the “focal speaker” and then come to associate the appropriateness of sanctioning with whatever the focal speaker dictates. At this point, the focal speaker (oracle, shaman, elder, judge, etc.) effectively wields the power of sanctions. If the actor is recognized as a government official, then the government wields the power of sanctions and we call these legal sanctions. Given a theory of third-­party sanctioning, focal points can explain how government first comes to wield the power of legal sanctions. I do not deny that, in a particular context, the real world origin of legal sanctions might have simply been coercion—​as where an individual successfully uses violence to force others to heed his commands and this leader came to be thought of as “government.” My point is that legal sanctions can arise more gradually and without an initial grounding in coercion. In Chapter 7, I examine some historic cases of adjudication where the judge acquired this power to control third-­party sanctioning without the judge or any other leader coercing people to accept his decrees.

Coordination and Legitimacy Having examined how the focal point effect can explain the origin of legal sanctions, now I reconsider legitimacy. Up to now I have focused on establishing the independence of the focal point effect, but here I discuss the relationship between focal points and legitimacy. Although the two theories are rival hypotheses for some purposes, I discuss how they are also complements or, in some cases—​but I stress: not all cases (not the ones discussed previously)—​merely different ways of describing the same phenomenon. Two Synergies of Focal Points and Legitimacy Let’s start with potential synergies. There are two: (1) the focal point power might help to explain how law can first obtain legitimacy and (2) legitimacy might help to explain what it is about law that is focal. f o c a l p oi n t or igi n s of l e gi t i m ac y

First, how does a new legal rule first acquire legitimacy? One possibility is that the procedures for creating the new law were themselves fair, endowing the law with legitimacy from the beginning. What this means in practice is that the legal institution creating the rule—​the legislature or courts—​ 119

The Expressive Powers of Law

possess legitimacy so their new creations start off with a presumption of legitimacy. But there is a deeper origin puzzle when the institution itself is new. Take the International Court of Justice (ICJ), first discussed in Chapter 3, where I noted that the high rate of compliance with this court’s decisions seem not to be the product of sanctions. One might posit that legitimacy is the explanation. An adherent of procedural legitimacy theory might posit that states defer to the ICJ because it uses fair processes in reaching its rulings. But this makes it seem too easy to generate legitimacy. Other international courts have failed to generate compliance, even though they have followed the basic ideas of legal process. The ICJ replaced a court ironically named the Permanent Court of International Justice, which followed similar procedures.66 Then there is the fact that the Security Council can refuse to take action to enforce ICJ rulings if a single member exercises its veto power, as members have done when the ICJ has ruled against them.67 Why is that a fair process? One answer—​the realist view—​is that the ICJ has not and does not now work because of legitimacy. Many nation states are purely realist in orientation; in international affairs, they are unconstrained by moral notions and, in any event, the nations that conduct foreign affairs by moral principle are not likely to share a cross-­cultural sense of procedural fairness. On this account, the ICJ works only because the Security Council threatens sanctions or possibly because the court provides a coordinating focal point as discussed in Chapter 3. As to the latter possibility, in my work on the ICJ with Tom Ginsburg, where we found high levels of compliance (68 percent of merits decisions), we also observed that a great many of the cases plausibly involved coordination. A majority of the cases reaching the merits—​t wenty-­six of forty­one—​were disputes over borders, incursions into air space, or property.68 In a majority of these cases—​seventeen of twenty-­six—​neither nation filed objections to jurisdiction, so each essentially consented to jurisdiction. We can plausibly infer, therefore, that the stakes were sufficiently low that the worst outcome for each was to risk war; otherwise, the side preferring war would have waged it or at least objected to jurisdiction. These consensual cases most plausibly fit the HD model just discussed. In all seventeen of these cases, there was compliance. (I return to the ICJ in Chapter 7). By contrast, compliance was not nearly so high in other types of cases. There was still compliance in a majority of cases—​six of nine—​when one 120

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

side raised jurisdictional objections to adjudicating a dispute over a border, aerial incursion, or property, perhaps indicating that there was still an important element of coordination. Perhaps one side thought it would still do better without adjudication, but still wanted to avoid risking war and therefore found it difficult to ignore the ICJ’s focal point. Finally, in the remaining fifteen ICJ merits decisions, involving disputes where the coordination element is less certain (not concerning a border, aerial incursion, or property), there was compliance in only six cases. Neither a sanctions nor legitimacy theory can easily explain these results. This characterization, however, merely repeats the idea of focal points as a rival hypothesis to legitimacy. Here is a slightly different idea, a synergy. My claim is that, in the usual case, a law or legal institution or regime cannot have legitimacy if no one complies with its demand. The law must be respected to be respectable; it cannot be perceived as a joke. We see this most clearly in an old law that embodies norms that have long since been abandoned and are now violated with impunity. But this is also a problem at the beginning: a new law needs to generate compliance or it will never “earn” legitimacy.69 If so, legitimacy cannot be the initial explanation of compliance because, without some other mechanism, the institution would never generate the compliance necessary to earn legitimacy (and, in turn, generate further compliance). If so, then another compliance mechanism—​sanctions or focal points—​ can lay the foundation, creating some initial effectiveness, which earns legitimacy. In the absence of sanctions, the focal point power answers the puzzle. The ICJ’s ability to influence behavior by providing a coordinating focal point allows it to generate some compliance from the beginning. The effective institution attracts the attention of other disputants, giving it a chance to further demonstrate its effectiveness. Once the compliance created through the focal power of ICJ expression “earns” legitimacy, that legitimacy might allow the ICJ to resolve disputes that do not contain an element of coordination. The two powers might be additive. It is, however, difficult to test this possibility for two reasons. First, I cannot discern anything in legal legitimacy theory that predicts an amount of time for a new legal institution to acquire legitimacy. From my study with Ginsburg, there is no clear evidence that ICJ compliance has increased over time, but perhaps the legitimacy-­creation period is longer than the almost seventy years since the ICJ’s creation. Second, even if the ICJ has earned legitimacy, which now contributes to compliance, that 121

The Expressive Powers of Law

would not guarantee that compliance would increase. The ICJ might overstep, as some critics have argued, and try to resolve disputes that it has no power to resolve, as when it purported to decide the legality of the use or threat to use nuclear weapons.70 The ICJ expression there offered no coordinating focal point as there is no clear incentive for nuclear powers to forego their weapons if others do. And, given the stakes, the ICJ’s power of legitimacy, whatever it is, may simply be insufficient. For a domestic example, consider a new ban on smoking in restaurants. Legitimacy theory alone does not offer a good explanation of the high levels of compliance we observe. The problem is that, with the same procedural legitimacy to courts and police, some laws are extremely ineffective; there is low compliance with speed limits, copyright laws, marijuana and obscenity bans, and restrictions on work hours by hospital residents. The process followed in creating these laws is essentially the same as it is for smoking or property rules; the perception of the police and courts are the same. So it is difficult to say that procedural legitimacy explains the greater success of smoking-­in-­restaurant laws than these other laws. Certainly, it would be better for a theory of compliance to explain variation in compliance across laws. Again, I would point to the law’s role as a coordinating focal point, particularly by “waving on” nonsmokers to insist on nonsmoking when they are in areas designated for them (and other mechanisms discussed earlier in this chapter). Once there is compliance, the law may acquire legitimacy. Conversely, laws that lack both sanctions and the focal point power may be so ineffective that they never acquire the power of legitimacy, and are widely flouted. This explanation predicts a correlation between legitimacy and compliance: People would be more likely to comply with laws they perceive as legitimate. But the reason is not that the legitimacy produces all the compliance but that legitimacy follows from compliance produced by other mechanisms. Once achieved, the perception of legitimacy would create another independent increment of compliance. l e gi t i m ac y c o n t r i bu t io n s t o f o c a l p oi n t s

Now let us turn to the second point of synergy: how legitimacy enhances the focal quality of a law. When I presented the focal point theory of law, the final condition was the absence of a more powerful competing focal point. This is one reason that Schelling’s examples work so well. When the traffic light breaks, there is only one Bystander who steps into the 122

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

i­ntersection to direct traffic. When two shoppers are lost in a department store, the signage works because it conveys a uniform message: “The management suggests that all persons who become separated meet each other at the information booth in the center of the ground floor.”71 Yet, as we have seen, the law often does have expressive competition. Much of this chapter dealt with the common problem that arises if the government wants to displace an existing social norm. As a salient precedent, the customary behavior is also focal point. My main response was that disequilibrium is common, in which case the prior precedent is less inevitable. There are other sources of expressive competition. In particular, where there are social movements, norm entrepreneurs who oppose the government’s new law or prefer a different law may attempt to influence behavior by creating a competing focal point. Perhaps the leader of a “smokers’ rights” group declares that smokers should vandalize no-­smoking signs. Perhaps a neighborhood leader declares, contrary to the legal rule, what residents should consider an acceptable level of noise or that fruit overhanging a property line should belong to the tree owner. Or a private organization purports to clarify the customs regarding the rules of the road for bicycles or the rules of firearms possession, again contrary to law. In each case, the private expression can have a focal effect. The result is cacophony. While the law might work expressively in the absence of the private competitor and the private rule might work expressively in the absence of a state competitor, it might be that no expression works when both are present. The competing expressions fail to create mutual salience of any one equilibrium. In this case, the presence of multiple expressions introduces a second level of required coordination, the same as we saw in Chapter 3’s discussion of leadership. The first level is how individuals will coordinate their behavior (in my examples, to sustain cooperation and avoid mutually destructive conflict). The answer is to look to a third party such as the state to create a coordinating focal point through expression. Now that there are too many expressions, the second level coordination problem is for the individuals to select one of the many expressions that they will use to coordinate. When there are many speakers, whose voice will provide the focal point? So far, my answer has simply been that law is usually “louder.” It typically generates more publicity and therefore is more likely to be mutually 123

The Expressive Powers of Law

salient than alternative expressions. Yet this point is contingent. Here is where sanctions and legitimacy offer assistance. As rival hypotheses, they offer to solve the compliance problem where the focal point fails. But they also make legal expression focal. Suppose, for example, that the expected sanctions are nondeterring—​that is, the sanctions themselves or their frequency of imposition are too low to deter a violation. Nonetheless, the fact that legal expression comes with sanctions might distinguish legal expression from other expression: it is the only expression with sanctions. One might say that every private expression has something unique about it if you look hard enough. The expression of the smokers’ rights leader might be the only expression written on green paper or the neighborhood association’s expression might be the only expression sent by email in a PDF. There is a chance that nondeterring sanctions make legal expression focal, especially if people have to think about the law before they can conclude that its expected sanctions are too small to be deterring.72 But almost anything else could be focal as well. Yet legitimacy is different. By definition, a legitimate authority is the quality to which people defer. The expression of a legitimate speaker may then stand out—​be focal—​in a way that no other speaker is. It means the speaker is someone to whom one “ought” to listen and defer. That a speaker or expression has legitimacy seems inevitably a salient distinction, in a different dimension from the fact that the expression is the one on green paper or in a PDF file. But is this a synergistic account or, in this context, is the focal point merely parasitic on legitimacy? There is a complementarity as long as we acknowledge that there are limits to the power of legitimacy. Suppose individual A fully believes that the law against embezzlement is legitimate (substantively fair) and the institutions that create and enforce the law are legitimate (procedurally fair). Suppose A therefore feels an obligation to obey the law against embezzlement. It is important to acknowledge that we can still imagine this person violating the embezzlement law. Legitimacy is a reason and a motive to obey law, but like every other motive, it can be overcome by stronger incentives to commit the crime. Suppose we use some numbers to illustrate the point. Even with no perceived risk of sanctions, A won’t embezzle unless the opportunity is greater than $100,000. Legitimacy is working quite well here, but there are limits because A will embezzle a larger sum. 124

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

Now consider a case to which the focal point theory applies: compliance with a designated no-­smoking rule. For simplicity, let us assume there is no state enforcement of the ban in some context, so the expected sanctions are zero. The smoker may perceive the smoking ban to be legitimate (procedurally and substantively) and yet his need to smoke and his desire to stay where he is while smoking may overwhelm his desire to comply with the law. Again, let us use numbers to illustrate. Solely because of legitimacy, a pack-­a-­day smoker might obey the smoking ban unless he has been without a cigarette for more than three hours. Legitimacy is powerful enough to outweigh his nicotine cravings below three hours, but not powerful enough after that. The examples reveal the complementary effect of a focal point. In the embezzlement case, there is no element of coordination, so legitimacy acts alone. In the smoking case, however, the smoker has a second incentive to comply: to avoid what I have described as Hawk/Hawk confrontations with nonsmokers, mostly heated verbal exchanges. A no-­smoking sign for the area “waves on” the nonsmokers to play Hawk, insisting on getting their way. The focal point effect is additive with legitimacy, so now it is plausible that the smoker will comply with the law even if he has been without a cigarette for more than three hours. Perhaps with both incentives, he now complies unless he has been deprived of nicotine for more than five hours. Yet there is more than a merely additive relationship between focal points and legitimacy. Suppose some smoker’s rights advocate has written on the “no smoking” sign in ink a phrase like “smokers unite and resist,” or “half a cigarette never hurt anyone.” Now we have squarely presented the problem of competing expressions. There is one message waving on the nonsmokers and another message waving on the smokers. The competing messages may fail to align expectations, neutralizing the law’s focal power, leaving only the legitimacy effect. It is at this point that legitimacy may serve a second function, which is to make the state’s expression more salient than the smokers’ rights graffiti message. The original legitimacy effect is direct—​the smoker defers because the law deserves his obedience. The second effect is indirect—​the smoker defers because he believes the law’s message is the one everyone will pay attention to, thus putting him in the position of a Hawk/Hawk outcome unless he defers. Without a competing message, we imagined 125

The Expressive Powers of Law

the combination of legitimacy and focal point effects would raise the duration of the smoker’s compliance to five hours. With a competing message, the amount might fall back to three hours, but for this indirect effect by which the legitimacy of the state’s message preserves its focality. Thus, legitimacy makes the law’s focal effect more resilient than it would otherwise be. Without legitimacy, the law’s focal power depends entirely on its publicity. It may be the only message or the loudest message in a given context requiring coordination. With legitimacy, the law’s focal effect can exist even if it is not the only or loudest message. Without the analysis of focal points, we might then miss an important effect of ­legitimacy. We can see this last point better by focusing on the Holmesian “bad man” who only wants to comply with law to the extent it is within his narrow self-­interest to do so.73 Such a sociopath would presumably never defer to law because of its legitimacy. He would comply with law because, in a coordination situation, the law provided the only focal point, causing him to expect other people to act in a way that makes his compliance selfishly desirable. At a busy traffic intersection, the Holmesian bad man complies with the traffic light because he expects otherwise to wind up in a collision. Now suppose there was some expressive ambiguity—​an interloper steps into the intersection and attempts to direct traffic contrary to the working traffic light, for no apparent reason. For the bad man, there is no reason other than salience to defer to the state’s traffic light rather than the interloper. But the bad man will recognize that other drivers may not be sociopaths; they will view the traffic light as legitimate and be more likely to comply with it, and so he will be more likely to comply with the traffic light. Here, legitimacy has no direct effect on the bad man, but it has an indirect effect because the situation requires coordination with ordinary folk. Focal point theory shows that legitimacy counts twice—​one directly, the other via the focal point channel. (Conversely, a person who was moral but lacked a theory of mind, who could therefore never anticipate what others will do, might comply with law because of legitimacy’s direct effect, but would never consciously comply with law as a means of coordinating, including this indirect legitimacy effect). Let me summarize the point of this section as follows. In Chapter 2, I said that we might think that people use the salient option as a coordinating focal point, not from pure rationality, but as a meta-­convention for 126

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

coordinating. Now consider a second possible meta-­convention to address the second level of required coordination: when there is uncertainty about mutual salience, as where multiple messages or contextual features point in different directions, coordinate on the legal expression. We might be more inclined to use law to solve our higher-­order coordination problem to the extent it is legitimate and to the extent it is more legitimate than the alternative expressions, but its legitimacy is a means to the end of coordination. People do not just aim to do the right thing by deferring to legitimacy, but also to do the selfish thing by finding the surest way to coordinate. One Case of Equivalence: The Translation of Legal Legitimacy into Focal Points I have so far treated the legitimacy and focal point theories as alternatives, either rival hypotheses or complementary mechanisms for legal compliance. The distinction is necessary for me to show a separate role for coordination. And the distinction is entirely real as long as one focuses on the coordination aspect of any particular legal rule. Yet in this final section, I am going to consider coordination at a much higher level of generality, the element of coordination at stake in the general social order known as the “rule of law” (or one version of it). At this societal level, I contend, legitimacy and focal points are different ways of describing the same phenomenon. Specifically, I describe legitimacy theory with an Assurance Game. By using game theory to describe legitimacy (in this one instance), I hope not to reject the independent importance of legitimacy, but to break down some barriers to interdisciplinary work. Economists, psychologists, and sociologists may, at this high level of generality, be using different language to refer to the same phenomenon. Of course, I do not aim to review or resolve all the methodological and substantive conflict between the social sciences, only to show one surprising parallel in the description of legal compliance. As representative of legal legitimacy theories, consider the work of two sets of scholars. John Darley, Janice Nadler, and Paul Robinson argue that criminal law’s influence depends on its “moral credibility,” which depends on how well the law embodies the substantive moral intuitions of citizens.74 Essentially, people will be more obedient of law n, if they agree with the moral content of the law on subjects a, b, c, etc. Tom Tyler and his co-­ authors emphasize procedural legitimacy: An individual will be more 127

The Expressive Powers of Law

obedient of the substantive law if she feels fairly treated and heard in the interactions she and her acquaintances have with local officials—​police and courts.75 Let us take one or both of these claims to be true. I want to show that we can explain their claims using the language of game theory. Until now, I have generally employed the conventional assumptions of rational choice—​that people are fully rational and self-­interested in the narrow sense that excludes considerations of morality or altruism. A large behavioral economics literature challenges these assumptions in various ways, including the finding that people are motivated in broader and more complex ways. I will use these broader motivations to argue that citizens are in a coordination game with government officials, so it is in their interest to comply with law if, but only if, the citizens perceive other citizens and government officials as contributing to the public good in the way they enact, enforce, and obey the law. c o n t r i bu t io n s t o p u bl ic g o od s a s a n a s s u r a nc e g a m e

Contributions to public goods sometimes have the structure of an Assurance game, though they are frequently described as a PD. Public goods are nonexcludable and nonrivalrous goods, meaning that (1) once they are produced, neither the producer nor anyone else can exclude others from consuming them, and (2) consumption by one person does not diminish the consumption by another.76 Examples are the fruits of individuals participating in a neighborhood crime watch, voting in political elections, or conserving on carbon consumption. Once created, it is not possible to exclude anyone from enjoying the reduction of crime (creation of democratic government, prevention of climate change, etc.). And the fact that one person enjoys this benefit does not diminish the enjoyment of others. The positive externalities these goods create means that their creator may not be able to recoup fully the costs of their creation, so individuals may create a less than efficient quantity of public goods.77 Analysts sometimes use the PD game to describe the situation, predicting that individuals will each decide to contribute nothing to public good creation, even though they are worse off than they would be if all contributed. Yet the PD game is not obviously the right model for a public good, given the results of the experimental literature.78 In the standard design of these experiments, each subject decides whether to contribute some part of a monetary endowment, provided by the experimenter, to the group, where it will be multiplied and then allocated back in equal shares 128

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

to each member of the group. The multiplier captures the efficiency of creating a public good; the efficient outcome is for everyone to allocate all of their endowment. But the self-­interested decision is to give nothing. For example, suppose an experimental group consists of three people with $3 each and that their money will be doubled if contributed to the group. If all three contribute 100 percent, the $9 donated will be doubled to $18, and then divided equally so that each individual receives $6 back. Yet even though this is the best overall outcome, a self-­interested individual is still presumably better off keeping his $3 than receiving back only one-­third of its doubled value, i.e., $2. According to this logic, the best outcome occurs when an individual keeps her $3 and gets one-­third of the $12 the other two group members create by donating their endowment. The sum of $7 the sole defector gains is more than the $6 earned by an individual under universal cooperation. Despite this simple prediction, scores of public goods experiments have found that some individuals will contribute in these circumstances, at least at first. The simple PD prediction is thus falsified. Typically, however, if the game is repeated, the contribution rates fall quickly towards zero. One plausible way to explain why there are any contributions is that, for some reason, the players are not in a PD but a coordination game. Because of non-­self-­interested or nonmaterial motivations that the experiment cannot exclude, some people act as reciprocators. Unlike narrowly self-­interested individuals who refuse to contribute no matter what others do, and unlike “Kantians” who always contribute (in keeping with the categorical imperative) no matter what others do, reciprocators want to give when others give and withhold when others withhold.79 Why would people act as reciprocators? First, there may be external social incentives that make it better for an individual to contribute when others contribute (or worse not to contribute when others contribute). The simple desire for esteem provides a general social incentive to engage in behavior that others approve or to avoid behavior that others disapprove.80 Individuals may be self-­interested but expect to earn approval they value (even from the strangers) by contributing when others contribute and/or to incur disapproval they disvalue by withholding when others contribute. Second, individuals may have internalized conditional obligations to match what others do.81 “Homo reciprocans” may gain utility from reciprocating contributions or lose utility from the guilt of exploiting another player by failing to reciprocate that player’s contribution. 129

The Expressive Powers of Law

Player 2 Player 1

Contribute

Withhold

Contribute

3, 3

0, 4

Withhold

4, 0

2, 2

F i g u r e 4 .1  

A Public Good Game as PD

Whatever the cause, when the players are reciprocators, what would be a PD game may become an Assurance game. Figures 4.1 and 4.2 illustrate this transformation. The first figure is a PD game, with Withhold/With­ hold being the only equilibrium (as explained in Chapter 2). The second figure differs only by the addition of utility for reciprocation, that is, from contributing when the other player contributes, raising it from 3 to 5. There are now two equilibria: Contribute/Contribute or Withhold/Withhold (again, for reasons explained in Chapter 2). The former equilibrium is mutually better, but the riskiness of contributing, which leads to the worst outcome when others withhold, may cause the players to prefer to withhold. To overcome this risk, each player needs to assure the other that they can reach the superior contribute/contribute outcome. The public goods problem is therefore an Assurance game when one wants to contribute if others contribute but withhold if others withhold. The game is slightly more complex than these figures indicate if one can give differing amounts. The reciprocator wants to contribute the same amount that others give. But making any contribution is risky because the worst outcomes involve giving (a lot) more than others give. Moreover, we Player 2 Player 1

Contribute

Withhold

Contribute

5, 5

0, 4

Withhold

4, 0

2, 2

Figure 4 . 2 

A Public Good Game as Assurance 130

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

now have a more complicated coordination problem: not just to coordinate on contributing or not, but also to coordinate on contributing at a particular level. This model of the public goods game offers one explanation for the experimental results. Reciprocators give according to what they believe these other strangers will give and their estimates range widely, so their contributions in the first round range widely. At this point, the riskiness of giving when others don’t causes a downward spiral in contributions. If you gave the most in the first round, you will suffer the worst outcome. You might optimistically think that everyone will match you in the second round, but simple risk aversion causes you to avoid the worst outcome by decreasing your contributions down to, at most, the average level of contributions, or less to be safe. Yet if you gave the least in the first round (but are a reciprocator), you likewise do not match the highest contribution. Given risk aversion, you also aim for the average, or a little lower. As a result, the average in the second round is lower than the average in the first. Now risk aversion causes participants to expect contributions to be even lower in the third round and those expectations are self-­fulfilling. Everyone avoids risk by trying to avoid being the sucker who gave more than everyone else. It is therefore possible that the observed decline and elimination of contributions is a failure of coordination. If reciprocators start at widely different levels in a game where the worst outcome is to contribute more than anyone else, they unsurprisingly fall into a downward spiral. There is a failure of morale. But if they could assure each other they would start at some particular level, and did start at that level, they might be able to sustain even a high contribution effort over time. Thus, there may be a high contribution equilibrium, in which the past success of uniform high contributions assures each participant that there will be high contribution in the next round. If so, reciprocators should be subject to a focal point influence. Third-­party cheap talk declaring a particular contribution level might facilitate coordination at that level. Roberto Galbiati and Pietro Vertova ran an experiment that validated this prediction.82 They put six-­player groups in a linear public goods game, where individuals earned real monetary payoffs depending on the inter­­ action of their decisions with other group members. The rational strategy for purely selfish players was to contribute nothing, but their control subjects initially contributed nontrivial amounts, though, as is standard, their 131

The Expressive Powers of Law

contributions declined over subsequent rounds to near zero. When the experimenters included in the instructions a law-­like statement that a specific “minimum contribution” of eighty percent of the per round endowment “is required,” contributions were significantly higher than the control, and remarkably stable, even though there was still no selfish monetary incentive to contribute in either condition.83 In imposing a requirement, there is a risk of “demand effects,” in which subjects make a contribution to avoid some unspecified penalty or merely to satisfy what they believe the experimenter wants. Galbiati and Vertova try to avoid this by (a) clearly indicating to subjects that they had the power to make a contribution less than what is “required,” and (b) specifying that the effect of contributing less than the requirement is to incur a small possibility of a small sanction. They designed the expected sanction so it is insufficient and nondeterring; for self-­interested actors, the maximizing option was still to contribute nothing. Thus, one way to interpret the experiment is that the subjects contributed at a stable level because they were reciprocators seeking to coordinate and the expression created a focal point that allowed them to do so. Third-­party expression solves the Public Good/Assurance game. As a real world example of this analysis, consider contributions to a social movement. One might use a PD to model the decision of whether to participate in a social movement, but Dennis Chong convincingly argued that the Assurance Game is often a better model.84 He focused on the 1960s civil rights movement. Even though one could not exclude civil rights from those who failed to contribute to the social movement that created them, Chong argued that social incentives raised the payoffs for participation if a particular movement event succeeded, but not otherwise. As a result, the payoff from participating when enough others participated to make the movement event successful was higher than the payoff from not participating in a successful event. But the payoffs from participat­ ­ing in failed efforts were still worse than nonparticipation. As Chong explains, the Assurance game captures an important dynamic of social movements like the civil rights movement—​the need for leaders to assure potential participants that there will be enough participation to succeed. Given a baseline of nonparticipation in the absence of a social movement, charismatic leaders must communicate optimism and publicize (even exaggerate) successes, so that everyone will assume everyone else will ­participate. 132

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

r e c i p ro c a l l e g a l c om p l i a nc e a s a “ru l e of l aw ” a s s u r a nc e g a m e

Now consider the public goods of legal enforcement and compliance. Scholars have applied the idea of reciprocity to tax compliance, finding that people are more willing to pay their taxes if they believe that others are paying their taxes, a result that might seem puzzling if one thought that the only incentive was the expected sanction from nonpayment.85 But one might see this as a multiparty Assurance game, where one feels some obligation to reciprocate the taxpaying of other citizens (because of intrinsic or extrinsic social incentives). For homo reciprocans, there may be no intrinsic benefits to supporting a government that others do not support, but a positive feeling of pride in doing one’s part and belonging to a successful political constituency when most others do their part. The same might be true of many other public goods sometimes regulated by law: voting in political elections, recycling, serving as a juror, not littering or overconsuming common pool resources, etc. One wants to do one’s share if, but only if, others are doing theirs; the law works as a focal point by defining what “doing one’s share” is. Now we have enough background for an informal game theoretic account of legal “legitimacy.” That is, we can explain the observation that people are more likely to obey the criminal law if they (a) perceive that legislators enact statutes consistent with their moral intuitions (Darley and Robinson’s empirical desert theory) or (b) that local police and judges respect them in legal interactions (Tyler’s procedural justice theory). Instead of a game between citizens, as in the tax example, suppose that the game involves citizens and government enforcers, the officials who make and enforce law, i.e., legislators, police, prosecutors, chief executives, judges, etc. The citizens and officials together contribute to the creation of public harmony and order. The citizens can make the enforcers’ job easier by not committing crimes and by aiding the apprehension of those who do offend. The government enforcers can make the citizens’ life easier by not shirking, i.e., by defining the criminal law as citizens prefer, apprehending and punishing (only) the guilty, and making the effort to show citizens respect. Thus, citizens decide whether to obey law and aid enforcement; government enforcers decide what to criminalize and enforce and how hard to work. Each has an incentive to “cheat”: citizens sometimes have opportunities to offend that, in expectation, produce 133

The Expressive Powers of Law

positive payoffs; enforcers have substantial opportunities to shirk by indulging their own preferences for punishing rather than punishing according to preferences of citizens, and, most simply, by being rude rather than exerting themselves to be respectful. This citizen-­official interaction might be a PD game, in which citizens have a dominant strategy of always offending (when the expected benefits exceed the expected costs) and never incurring net costs to aid enforcement, while enforcers have a dominant strategy of always shirking. This is a high crime/high corruption equilibrium. Yet internal or external incentives can make the game one of Assurance, where the mutual shirking remains a possible outcome, but so does mutual contribution to the rule of law. Because homo reciprocan citizens gain utility (pride or a sense of belonging) by contributing when others contribute or perhaps lose utility (guilt, alienation) by free-­riding, they want to obey law and aid enforcement if (a) other citizens generally obey law and aid enforcement and (b) enforcers generally refrain from shirking and do a good job. Homo reciprocan enforcers want to work hard rather than shirk when (a) other enforcers generally work hard and (b) the citizens generally obey law and aid enforcement. In short, citizens will be more likely to contribute to public order by doing their part if most other citizens and officials do their part. We can explain Tom Tyler’s finding that people are more likely to obey the law if they have experienced fair and respectful treatment from police. Arbitrary and rude treatment is a form of shirking; the police are failing to do their part towards the creation of public order. Similarly, Robinson and Darley claim that people are more likely to obey criminal law if it generally corresponds to their moral intuitions, but that is because the legislators who enact laws frustrating common moral intuitions are indulging their own preferences (or those of some narrow but powerful interest groups) rather than those of their constituents, which is also a failure to do their part. In sum, if one modifies rational choice theory by positing a utility function in which people tend to reciprocate pro-­social and anti-­social behaviors (for reasons of the external benefit of esteem or the intrinsic value of reciprocation), then it is possible to use coordination theory to explain the concept(s) of legal legitimacy. The procedural and/or substantive fairness of law produces legal compliance because citizens are, by complying or violating with law, reciprocating the contributions that legal enforcers make or fail to make to the legal order. 134

L AW ’ S FO C A L POWER IN D Y NA M I C PERSPE C T IVE

I do not claim that coordination theory displaces or obviates legitimacy theory. The Rule of Law Assurance Game leaves completely unspecified the particular details of which citizen behaviors are used to reciprocate which enforcer behaviors. I know of no way of filling in those details except the kind of painstaking empirical research that legitimacy theorists have been doing for decades. The bottom line is simply that there is no necessary inconsistency in the different social sciences on this point and there may be some new room for academic collaboration (coordination!) rather than conflict. The focal point effect is an independent expressive mechanism for law to generate compliant behavior. This chapter has primarily concerned what I termed a dynamic effectiveness objection to the focal point theory. When behavior has evolved to a stable equilibrium, one we might call a social norm, convention, or custom, it may be hard to disrupt things with mere expression (although there are contrary examples). But disequilibrium is sufficiently common to offer a wide domain for focal point effects. I stressed two cases: where social movements disrupt existing patterns of behavior and where there is ambiguity at the margins of customs and conventions. In a dynamic frame, we can also see how the focal point theory explains the endogenous emergence of legal sanctions and interacts in interesting and synergistic ways with legitimacy. Finally, in the Rule of Law Assurance Game, I have shown how legitimacy theory can be reconciled with and explained in the terms of coordination theory. Yet this book is not just about focal points. There is another expressive effects theory of law I wish to describe, starting in the next chapter.

135

5 Legislation as Information

A

s explained in the introduction, a focal point works by directly influencing individuals’ expectations about how others will behave, which occurs even if the focal point reveals no pre-­existing information about the nature of the situation. On the road, for example, the yield sign and solid center line work by making salient a particular means of coordinating traffic, one in which the driver (respectively) slows in deference to other drivers and does not pass. Salience is sufficient; the driver need not infer from these signals any pre-­existing facts, such as the nature of the road or the traffic in this area. Yet, aside from the focal point, the driver might also make such an inference, as by taking the solidity of the center line as evidence that the traffic engineers thought that passing from this side was particularly dangerous at this point in the roadway. This chapter turns away from focal points and begins to examine this second expressive theory of law’s effect: that law reveals information. Whether or not the situation requires coordination, legal rules have informational content because, on average, they reflect the beliefs and values of the political actors who created the rules. When people learn of the rules’ existence, they update their beliefs in light of the new information. These new beliefs can change an individual’s behavior, usually in the direction of greater compliance, though sometimes less compliance and sometimes in ways orthogonal to compliance. In short, law provides information; information changes beliefs; new beliefs change behavior. Law is informative. 136

L EGIS L A T ION AS INFOR M A T ION

Of course, law is not informative when an individual is unaware that the law exists, which is often the case.1 As with any theory of law’s influence on behavior, an individual must know of the law in order for it to be influential. As I said in discussing legal focal points, people will not comply with an unknown legal rule (except by accident), regardless of the theory of compliance. Not knowing a particular law exists, they will not fear legal sanctions for violating it, not defer to its legitimacy, not use it to solve a coordination problem, and, I now add, not update beliefs in light of the information the law reveals. Or if they know the law exists, but significantly misunderstand its content, these mechanisms will produce compliance with what people think the law is, rather than what it actually is. Thus, the information theory of law applies only in those cases where people know the law exists and is limited to what they believe the law requires. As I discuss in this and the next chapter, some expressive claims about law are implausible precisely because people are not likely to know the law or legal action. With that proviso, the theory focuses attention on the information embedded in law, which depends on the incentives facing lawmakers; only certain motives and structures make it “incentive-­compatible” for actors to reveal information in the law they create. To begin, this chapter focuses on legislators and legislation. Exactly what information legislation conveys depends on context. There are many possibilities, but I discuss three types of information that are most relevant to legal compliance. The first two are attitudes and risk. With attitudinal signaling, law conveys information about public attitudes concerning the regulated behavior. (Here, “signaling” does not refer to the original use of signaling in economics to refer to a revelation of one’s “type.”).2 Prohibitions tend to reveal attitudes disapproving the prohibited conduct. Laws creating privileges or mandates to engage in behavior tend to reveal attitudes approving the behavior or disapproving its omission. In light of this information, people update their beliefs about what others in their community approve or disapprove. Because people seek to avoid disapproval (for reasons explained below), their beliefs about attitudes can change their behavior, usually toward compliance. For example, by banning employment discrimination against LGBT workers, the legislature may communicate pervasive attitudes against such employment practices, which may convince employers that their potential customers and 137

The Expressive Powers of Law

non-­LGBT employees are less likely to prefer, and more likely to avoid, firms that engage in such discrimination. By restricting abortion, legislators may signal the community’s disapproval of the procedure, which may create or strengthen perceptions that a permissible abortion procedure will incur social stigma.3 Attitude signaling usually increases compliance with law. Apart from my concern with legal sanctions or legitimacy, I may seek to gain the approval or avoid the disapproval of others in my community. If so, then when the new law banning employment discrimination or restricting abortion convinces me that more of the people around me disapprove of those activities, I have a stronger interest in avoiding the behavior the law prohibits. With risk signaling, however, a law conveys information about the costs and benefits of the legally regulated behavior. For example, by prohibiting the use of cell phones while driving, legislators may reveal their beliefs that this combination of activities seriously risks a traffic accident. Mandating the use of seat belts or motorcycle helmets may reveal the legislators’ beliefs that such devices provide significant protection from accidents. Risk signaling usually increases compliance with law. Apart from my concern with legal sanctions or legitimacy, I am motivated to protect my health. If the new law causes me to believe, to a greater degree than I believed prior to the law, that cell phone usage increases the risk of a car accident, then I will have a stronger incentive to avoid the behavior. If I think seat belts provide greater security, I am more likely to wear them. My new beliefs make me more likely to act as the law requires. The third type of information revelation is violations signaling, which occurs where a legal change reveals information about the rate of compliance with a pre-­existing legal rule. A legislative increase in sanctions for copyright violations, tax evasion, or drunk driving may communicate that more people are “getting away” with those transgressions, that is, not being detected or sanctioned. Police and prosecutors may send the same signal when their enforce­­ment operations reveal a surprisingly large set of violations. Typically, the more widespread one perceives violations to be, the weaker the perception of the law’s legitimacy or threat of sanctions. Violations signaling is one explanation of the “crowding out” phenomenon, where an increase in sanctions decreases compliance by “crowding out” other motives for obey­­ing the law. On this view, raising sanctions need not invalidate other ­nonsanction motives for compliance but gives 138

L EGIS L A T ION AS INFOR M A T ION

information that one who violates the law is less likely to receive a legal sanction or social disapproval. In this chapter, I discuss law-­as-­information by focusing on legislatures.4 The next chapter considers the information revealed by executive and judi­ ­cial enforcement of law. In all cases, the influence here is not dependent on a situation of coordination; individuals who update their beliefs about attitudes, risk, or violations may want to change their behavior even if no one else changes theirs.

Attitudinal Signaling For law to reveal information, the law’s existence need only be correlated with certain facts. If so, then observing the law makes more likely the facts that are positively correlated with its existence. In a democracy the obvious starting point is the possible correlation between legislation and public attitudes.5 As I explain, the primary mechanism correlating public attitudes and legislation is legislators’ concern with and expertise on the attitudes of their constituents. Belief change will not affect behavior, however, unless those beliefs affect the individual’s perception of behavioral consequences. So, I begin with the point that people are motivated by public approval and disapproval, which is the first condition for attitudinal signaling. I then discuss the second condition: that legislation affects people’s beliefs about what others approve and disapprove. Background: Why Perceived Attitudes Affect Behavior Commercial actors and political leaders often appeal to popularity as a means of persuasion.6 They know that perceptions of public attitudes can influence behavior. Perceived patterns of approval affect behavior for three reasons. First, individuals place intrinsic value on gaining the esteem and avoiding the disapproval of others, even strangers. This is the claim of a substantial literature on the human desire for esteem.7 The concept is connected to the value individuals place on status, prestige, honor, and fame,8 and why individuals are susceptible to shame.9 Several theorists explain social norms as arising because disapproval works as an intrinsic sanction, prodding people to conform to the behaviors other approve. Second, even if some people don’t value approval intrinsically, most will value it instrumentally, as a means to some material end. Most obviously, a businessperson strives to achieve a reputation for providing quality goods or services. But because people are highly opinionated and have 139

The Expressive Powers of Law

emotional reactions to differences of opinion, an individual’s reputation matters on endless other dimensions. In the presence of one’s exist­ ­ing or potential boss, customer, romantic partner, or in-­laws, one worries that expressing or implying the “wrong” opinion on a controversial subject—​e.g., guns, global warming, or gay marriage—​will damage one’s actual or potential relationship. The point applies obviously to acquaintances, but even among strangers there is some possibility that a stranger is intrinsically motivated (by morality or idiosyncrasy) to incur costs to express disapproval in an unpleasant way. There is always some low risk that a person who takes serious offense at one’s conduct will react with violence. Whether people value approval intrinsically (esteem) or instrumentally (reputation), and whether they seek approval from acquaintances or strangers, they will seek information about what others approve and disapprove. By definition, we have no information about the approval patterns of strangers except for our knowledge of the approval pattern in the population from which they are drawn. Even for people we know, our information runs out, especially for mere acquaintances (rather than close friends and family members). The point of the previous examples is that we may know our boss disapproves strongly of lateness and mumbling, but have little basis to gauge her views on abortion or the death penalty. The same is true for a customer or potential romantic partner. If not, then dealing with acquaintances is like dealing with strangers. The safe response is to “play it down the middle,” though that strategy depends on knowing where the middle is. Attitudes might also motivate behavior for a third reason. When people internalize a social norm, they may feel “guilt” if they violate the obligation, regardless of what others know or believe about the matter.10 This description makes it seem that another person’s belief has nothing to do with whether an individual feels guilt. But guilt inclinations are frequently based on abstract obligations that are not sharply defined, such as “doing one’s share” as a citizen or being “a good parent” (or a “good mother” and “good father”). Once one has internalized an abstract obligation, societal attitudes may affect one’s beliefs about how one goes about satisfying the obligation. For example, does being a good parent require using an expensive and inconvenient safety seat when driving one’s child or keeping inconvenient trigger locks in one’s firearms stored in the home? Does being a good 140

L EGIS L A T ION AS INFOR M A T ION

parent require that a smoker never smoke inside the house, to avoid exposing her children? Here, individuals may take the beliefs of others as evidence for what concrete behavior their internalized but abstract obligations demand. If people were moral philosophers, they might not put much stock in the prevailing view, but people know they are prone to rationalize away their obligations, so they look to what others think as a way of judging what their obligations mean. If so, then the attitudes of others may be an important determinant to whether one feels guilt. A parent who believes that her community considers child safety seats (in part) to define good parenting will use the seats to avoid guilt. Of course, committed iconoclasts might reason in the opposite direction. Some people are motivated to defy conventionality, including what they perceive as a shallow bourgeois morality. Instead of internalizing a role as “good parent,” or caring what most people think of the content of that role, they have internalized a role of social critic, rugged individualist, or outcast. Nonconformist social groups sometimes reward members for the very behavior the dominant society regards as deviant, including criminal behavior. For those who value general social disapproval, the law’s revelation of information will have the opposite effect on behavior, encouraging its violation.11 Yet the general point remains: the statistically dominant element of society values conformity to the common morality and they seek information on what that common morality is. The Basic Claim: Law Reveals Attitudes Although people care about patterns of public approval and disapproval, their daily lives provide only anecdotal and unrepresentative evidence of what current public attitudes are. They know a lot about the attitudes of close friends and family members, but not so much about the attitudes of strangers and mere acquaintances. Law provides one source of information to fill this gap because of lawmaker expertise in public opinion. According to a significant (but not unchallenged) strand of political science research, legislative policy is responsive to the median voter.12 If democracy pushes in the direction of a correlation between law and public attitudes, the former becomes a signal of the latter. This is most obvious in the case of state laws enacted pursuant to popular initiatives or referenda,13 but I will focus on the more common case of legislation. Legislators may or may not have expertise on substantive policy issues, but they and their advisers are experts on what 141

The Expressive Powers of Law

the median voter in their district wants, as well as what issues inspire the most intense levels of approval and disapproval. Because they value re-­ election and expect voters to punish or reward them for their votes, legislative votes signal legislator beliefs about voter attitudes. Depending on the issue, the legislator’s vote may reveal his belief about the attitude of the median voter (or perhaps the size of the subset of highly motivated voters who hold particularly intense views on the subject). The legislator’s vote itself reveals information, but frequently an individual observes only the existence of a law (especially if she entered the jurisdiction sometime after the law passed) and infers the direction of the legislative vote. Because of majoritarian rule, an individual knows that the law reflects a majority of legislative votes in its favor. The majority of legislative votes signals attitudes across the entire jurisdiction of the legislative body. In addition, most American legislatures consist of members elected by separate districts (instead of “at large” representation). Absent any more specific knowledge about one’s representative’s vote, the passage of a law makes it more likely than not that one’s own representative voted for the law, which specifically reveals attitudes in one’s electoral district. As an aside, notice that the combination of these effects means that attitudinal signaling is strongest at the local level. If one is likely to have the most information about the attitudes of one’s close friends and relatives, then law is an important signal only for strangers and acquaintances. Except for public figures, an individual typically earns approval and disapproval from strangers only when the strangers are in one’s immediate environment and can directly observe one’s actions. Federal or state legislation will, overall, reflect the attitudes of the national or state populations, including distant strangers whose approval or disapproval one could never incur. Federal or state legislation is therefore only weakly correlated with the attitudes of the strangers whose attitudes matter. By contrast, at the local level, all the votes in the legislative assembly may be informative, even those outside one’s electoral district. An individual in a town or small city may care about the approval patterns throughout the area because the individual travels between home, work, shopping, recreation, and other local destinations encountering people from different electoral units. The same is true for acquaintances but the analysis is more complex. People obviously do care about and receive approval and disapproval from geographically dispersed acquaintances. Yet many of one’s acquaintance 142

L EGIS L A T ION AS INFOR M A T ION

are still likely to be local. Statistically, local law should correlate more strongly with the attitudes of one’s local acquaintances than national law will correlate with the attitudes of one’s national acquaintances. That is simply because one’s local acquaintances are likely to be a much bigger percentage of the local electorate than one’s national acquaintances are a percentage of the national electorate.14 The smaller the percentage, the greater the chances that one’s acquaintances differ from the median voter whose attitudes the legislation reflects. In short, the law’s signal of attitudes is diluted by the information it contains about people whose approval or disapproval one will never receive; that “noise” is proportionately greater when the legislation is national or state rather than local. One might counter this claim with reasons that Congress signals other information more powerfully than state and local legislatures. While I have been discussing legislative expertise on public opinion, later in this chapter I discuss a second reason that legislative votes are informative—​ because legislative voting aggregates information on risk. As will become clearer after that section, there is more aggregation at the federal level because the number of legislators is larger (among other reasons). Yet because each legislator represents a different constituency, Congress does not aggregate information on attitudes. For example, if legislators vote on smoking regulation based on their assessment the health risks to nonsmokers, they are all voting on the same fact (health risk) and their votes are an aggregation of their beliefs about that fact. But if members of the House of Representatives vote based on their assessment of attitudes, each one is likely to vote on the attitudes of her constituents, who are a separate population from other constituents. Representatives are not experts in nor strongly concerned with the attitudes of citizens outside their own district. (The Senate is only trivially different with two Senators aggregating information attitudes in their state.) None of this is to say that state and federal legislation has no attitudinal power. There may be a significant correlation between the attitudes of median voters across districts and states. More likely is a correlation in the direction of attitude change among districts and states. A signal that the attitudes in other electoral units are all moving toward greater disapproval of public smoking or drunk driving, or greater respect of LBGT couples, is likely to be informative about the direction of attitude change in one’s own electoral unit, even if the static numbers are quite different. 143

The Expressive Powers of Law

Objections and Extensions: Public Choice, Public Polling, and Pluralistic Ignorance The attitudinal theory runs into two objections. First, one might read public choice theory as demonstrating that legislative votes are the result of interest group politics and therefore not informative of constituent attitudes. Second, one might claim that public polling data provides such strong information about public attitudes that there is no marginal effect of law. By addressing these objections, I further elaborate the claim that law signals attitudes. Public choice theory demonstrates how easily legislation can reflect the interests of a well-­organized minority rather than the diffuse majority.15 Corn subsidies, for example, probably do not reflect widespread attitudes favoring agri-­business. The copyright ban on copying DVDs may also reflect, not pervasive public attitudes, but intensive lobbying by industrial beneficiaries. Clearly, not all laws signal popular attitudes. Yet despite the influence of interest groups, there are several reasons to expect that, over­ ­all or in some identifiable contexts, enough people believe laws are correlated with public attitudes to allow the law to reveal attitudes. I begin with the broadest claims for the relevant correlation and move to successively narrower ones. Although I will not always distinguish between specific legislative bodies, the following points apply to federal, state, and local legislatures. First, despite concentrated interests, legislation might remain positively correlated with public attitudes. Legislators value re-­election. While they enact some laws benefitting narrow interest groups, they must also enact popular laws that they can tout in their re-­election campaign. The net effect may be that a given law is more likely than not supported by a majority of the voters in the jurisdiction. Second, if the correlation does not exist for all legislation, it is likely to exist for the category of well-­publicized provisions for which public compliance is an issue. Rent-­seeking tends to occur in low-­salience legislation, beneath the radar of popular media. Yet it is difficult to imagine any theory of expressive effects that applies to laws of which the public is largely ignorant (nor do sanctions or legitimacy theory fare better). So consider only high-­ salience legislation. Moreover, much of rent-­seeking occurs in narrowly applicable provisions of budgetary and tax legislation, for which there is no issue of general public compliance. So consider only legislation that 144

L EGIS L A T ION AS INFOR M A T ION

purports to regulate the general public or large parts of it. If we focus on the subset of (a) well-­publicized legislation (b) that proscribes rules for the general public, we have examples like criminal prohibitions on sexual assault, identity theft, drunk driving and public smoking; restrictive regulation of handguns, pornography, and abortion; legal liability for employment and housing discrimination and sexual harassment; laws mandating seat belt or motorcycle helmet use, childhood vaccines, and the payment of taxes; and laws asserting the right of mothers to breastfeed infants in public. Here, it seems plausible that there is a positive correlation between the legislation and public attitudes. If the public perceives this correlation, it perceives the legislation as a signal of public attitudes. Third, even if the correlation does not otherwise exist, there is a special category where the legislature enacts a salient statute against and despite opposition by concentrated interests. Many legislative bans are like this because there is usually some industry supplying the good or service that is the subject of the ban. One example is the tobacco industry’s early opposition to public smoking bans.16 Another is the NRA’s opposition to gun regulations. Where it appears that there is well funded and organized opposition to a legislative rule, and no concentrated interests in its favor, the rule’s enactment is, and will be seen as, strongly correlated with public attitudes. Finally, consider categories of people rather than legislation. Suppose that the population were divided into idealists and cynics. The ideal­­ ists optimistically believe that democracy ensures that legislation always reflects public attitudes. The cynics pessimistically believe that no such positive correlation exists. If so, then the law signals public attitudes for the idealists, and can so influence their behavior, but cannot have such an effect for cynics. But will the cynics go so far as to believe not only that there is no positive correlation between public attitudes and legislation, but also that a negative correlation exists? Will they think that the legislature raising the punishment for drunk driving and repealing the marital exemption for rape laws reveal that the public, on average, approves of the condemned behaviors? I doubt it. In any event, if the cynics did draw this inference, then there would still be an expressive effect, although it would be one that worked against compliance with the law. More likely, the cynics see no correlation and the idealists see a positive one, so if there are any idealists, the law works, on average, to signal public attitudes. Now consider a different objection: that public polling data provides better evidence of public attitudes and therefore displaces any potential 145

The Expressive Powers of Law

for law to provide unique information about those attitudes. One might say that polling data sends such a strong signal that, once it is received, the law provides no reason to update one’s beliefs any further. Consider several responses. First, the claim dramatically overestimates the availability of polls at the local level where attitudinal signaling is otherwise greatest. Polling is common for national issues and sometimes state issues, but there are many issues left unpolled. Second, even where a poll exists, it matters only if an individual learns of the poll. A legal rule, like a ban on smoking in restaurants or a required use of child safety seats is likely to remain visible long after the law is enacted, while the poll may quickly vanish from most people’s view the day after it is published in a newspaper or on a website. Third, even when an individual learns of a poll, she may discount it heavily. Most people know that polls are highly sensitive to the exact wording of the survey questions and also that political groups sometimes intentionally manipulate polls to produce the appearance that the majority agrees with the group’s position (which makes sense given the influence of public approval). By contrast, a legislative vote may be thought to matter more because the legislators risk their jobs by defying public opinion or simply guessing wrong about it. We may think of legislation therefore as a handy aggregation of the polling data on which the legislators relied, weighted according to their expert opinion of each poll’s reliability. Legislation is particularly good at incorporating and reflecting the intensity of preferences. Stating on an opinion poll that one feels “very strongly” about an issue is cheap talk. Legislators, however, care greatly about not offending individuals who really do care so much about an issue that they will take costly action against the legislator whose vote offends them. In this way, the legislative aggregation of polls is likely to be far better at reflecting the intensity of views held by a minority of voters. Thus, a legislative vote against what the polls reveal is a majority preference is a particularly strong signal that a minority feels very intensely about the topic. In some cases, this is the crucial issue for behavior: not the mild approval or disapproval that a behavior generates, but the risks that it offends individuals sufficiently that they will take costly action against it. Think of individuals who so hate exposure to cigarette smoke or dog waste that they will verbally or even physically accost the smoker or careless dog 146

L EGIS L A T ION AS INFOR M A T ION

owner whose behavior offends them. Or those who will lead a boycott against or vandalize businesses that appear to show hostility to religion, disregard for gun rights or military service, or tolerance of racist jokes or sexual harassment. Finally, polls are not sufficient to prevent large number of individuals from making systematic mistakes about public attitudes. Consider the related phenomena of what economists call “herding” and psychologists term “pluralistic ignorance”17 where most people in a group erroneously think that most other people in the group approve (or disapprove) of some activity, even though they don’t. The problem is that the perception that some activity is socially approved (or disapproved) suppresses the communication that would reveal the perception to be false. Thus, individuals believing that their personal dissent from the prevailing norm is aberrational may never give voice to their dissent and never discover that they are part of a silent majority. For example, psychologists found that some whites in the Jim Crow era south harbored secret disapproval of some segregationist practices, but underestimated how many others, like themselves, felt the same way.18 Many colleges find that their students disapprove of binge drinking, but overestimate the campus-­wide approval levels.19 In sum, individuals frequently have poor information about public attitudes, so that legislation is informative of what those attitudes are. A Brief Model of Attitudinal Signaling Here is a simple Bayesian model of how legislation reveals public attitudes. An individual cares about what level of disapproval he incurs by engaging in behavior X. Take event (A) to be community disapproval of behavior X, specifically, the level of disapproval that would tip the individual in question away from engaging in X. Event (B) is that the legislature enacts a new law prohibiting X. The individual has a prior belief about the probability of (A)—​community disapproval. How does (B)—​a new law—​affect the individual’s beliefs about (A)? More specifically, what is p(A|B), the probability of community disapproval given that the legislature enacted a law against it? Assume the basic correlation argued for previously—​that the legislature is more likely to enact the ban on X when there is community disapproval (at the level of A or greater) than when there is not. Given this assumption—​that p(B|A) > p(B|NOT A)—​then some algebra shows that p(A|B) > p(A), i.e., that the existence of the ban on 147

The Expressive Powers of Law

X increases the individual’s estimate of the probability that the community disapproves of X (to the level of A or greater).20 Informational Consequences: Discontinuous Effects and “Spillovers” By itself, the effect of the attitudinal signal on beliefs and behavior is likely to be marginal, increasing by only some small degree the perceived social disapproval of behavior the law punishes or the perceived social approval of behavior the law subsidizes or mandates. Because these effects tend to reinforce compliance, they may be significant when combined with other expressive consequences of law, such as the focal point effect examined in previous chapters. But for now, consider two special cases: (1) conditions in which the revelation of attitudes produces a sudden and discontinuous shift in behavior and (2) the possibility that attitudinal signaling might cause behavior other than legal compliance. There are several plausible cases where attitudinal signaling produces more than a marginal influence on behavior. One is the idea of a tipping point, a situation where a social norm or custom is on the cusp of a shift to a different equilibrium and requires only a small push to tip into it. In general, such unstable situations may be rare. Yet a new law may be enacted precisely because the existing order is less stable than it once was. If a social movement is already making progress toward disrupting some customary behavior—​as by discouraging drunk driving or sexual harassment or encouraging recycling or child safety seats—​then the legal revelation of public attitudes might prove decisive in further unsettling past patterns of behavior. The second case for a discontinuous effect is the collapse of pluralistic ignorance just described, where most people erroneously think that most other people approve or disapprove of some activity, even though they don’t. The economist Timur Kuran gives some examples, though his term for pluralistic ignorance is “preference falsification.”21 One instance is the abrupt fall of authoritarian regimes of the former Eastern Bloc countries. Kuran says that when censorship failed and people openly criticized the regimes for the first time, masses of individuals who thought they were in the minority in disapproving the regime suddenly discovered that they actually constituted a large majority.22 Collective action then quickly destroyed the regimes. Law can provide the catalyst for the sudden puncturing of pluralistic ignorance. We might even say, in a democracy, that legislation is one 148

L EGIS L A T ION AS INFOR M A T ION

of society’s most general purpose mechanisms (along with elections) for ­correcting pluralistic ignorance. When the perception of public attitudes falls seriously out of line with reality, legislators gain by enacting legislation corresponding to actual attitudes (and actual future votes), which provides a dramatic revelation—​a “wake up call”—​of actual attitudes. The secret ballot offers a constant corrective to pluralistic ignorance because voters can express their views in the polling booth without having to worry about social consequences. Politicians who sense an attitude change before it is generally understood can gain by enacting legislation that reflects and signals the new pattern of attitudes. A state law recognizing the rights of same-­sex couples to adopt or marry might suddenly expose and destroy pluralistic ignorance on the subject, by which many citizens were afraid to express their tolerance of such couples and never previously knew that that tolerance put them in the majority rather than the minority. A third case for a discontinuous effect is the possibility of a bandwagon effect or availability cascade. The correction of pluralistic ignorance suggests a move of beliefs towards the true value, but the public may overcorrect and move to making the opposite error about public attitudes. For example, Kuran claims that support for affirmative action collapsed in the 1990s when the public, which had believed there was a powerful consensus for affirmative action, discovered that opposition was pervasive. If there ever was pluralistic ignorance favoring affirmative action, one might interpret it an overcorrection of the earlier error that public attitudes favored racial segregation much more than they actually did. The abrupt collapse of conventional wisdom may have suppressed expression critical of affirmative action that seemed too connected to that prior Jim Crow convention. Kuran and Cass Sunstein term this sudden and erroneous shifts of opinion as an availability cascade,23 referring to the psychological mechanism in which people estimate the likelihood of an event (such as an individual’s approval or disapproval) by the ease with which its occurrence is called to mind. One mechanism they explore is a reputational cascade, where sudden and erroneous belief change occurs because those who do not share the consensus pretend they do to avoid a reputational loss, thus making everyone think the change is more dramatic than it actually is. Applied to the present context, legal change might chill oppositional expression, leading people to infer the change is even more popular than it really is. 149

The Expressive Powers of Law

All of these theories may be involved in explaining the American success of the regulation of public smoking (in addition to the focal point theory discussed in prior chapters). First, for some time in the 1980s and 1990s, the fall in smoking and increased understanding of its harms may have been undermining the stability of the norms favoring public smoking. Pro-­smoking norms were more resilient to change when there were more smokers and fewer perceived risks of second-­hand smoking. Second, there may have been pluralistic ignorance, i.e., that smokers and nonsmokers were unaware of the magnitude of the long-­term change in attitudes towards disapproval. Thus, the bans on smoking in office buildings and restaurants provided a sudden shock to existing beliefs, a strong signal of previously unappreciated disapproval of the public smoking. As a result, smokers became aware that their exposing others to smoke generated more disapproval than they previously believed, perhaps even more than is actually the case. Nonsmokers became aware that more nonsmokers would approve of efforts to discourage smoking in their presence. Such factors could produce significant rather than marginal behavioral change.24 Finally, although I emphasize the effect of attitudinal signaling on compliance, information has broader spillover effects, what have been termed “expressive externalities.”25 This is because, when law changes beliefs about public attitudes, the new beliefs may motivate behavior not required by the law. If a municipality bans employment discrimination against LGBT workers, the revelation of public disapproval of employment discrimination may cause a person to refrain from other types of discrimination, such as housing or public accommodations, despite being unregulated. New gun laws may signal attitudes in either direction—​more restrictions signal negative attitudes, while expansion of the right to carry concealed weapons signals positive attitudes. Expectations of disapproval could cause some individuals to forego buying or carrying even when legal; expectations of approval could induce some individuals to buy or carry when they otherwise wouldn’t bother. Spillovers or expressive externalities are sometimes unintended consequences (the next section has examples), yet they can be intended. Some of the laws guaranteeing mothers the right to breastfeed their infants in public places were intended to encourage not merely compliance (by those who would otherwise object), but the mother’s decision to breastfeed.26 Sometimes compliance is entirely beside the point. An example is the repeal of unenforceable statutes. After they were ruled unconstitutional, 150

L EGIS L A T ION AS INFOR M A T ION

some southern states retained “on the books” their laws of racial segregation; after Lawrence v. Texas invalidated sodomy statutes, some states have not bothered to repeal them. Yet there are groups in both cases who have sought or still seek the law’s repeal.27 Because the laws have previously been unenforced and unenforceable, it might appear that their repeal can have no behavioral effect. Repeal is, however, plausibly aimed at attitudinal signaling. The point is to show that even though the law retained sufficient popularity not to be repealed before it was ruled unconstitutional, it no longer enjoys that popularity today. Legislative time is scarce, so that placing the repeal bill on the agenda means bumping some other issue off the agenda. There will be legislators who therefore oppose prioritizing the repeal of an unenforceable law. Other legislators oppose repeal because they reject the Supreme Court opinion that declared it unconstitutional. It is the very fact of this opposition that ensures that passage of a repeal bill reveals attitudes. The more that current attitudes oppose the old law, the more likely current legislators are to pass the repeal, and so the more the repeal signals the new attitudes. Repealing the ban on sodomy or interracial marriage can therefore reveal a greater level of acceptance and tolerance of gay and interracial sexual relationships, lowering the expected social sanctions from engaging in them and also from privately expressing views in favor of the liberty to engage in them. Updating beliefs could change behavior in the same way discussed for anti-­discrimination laws: by making people more wary of racial or sexual orientation discrimination even in legally unregulated contexts. Indeed, the revelation of a new consensus in favor of these liberties and revulsion to the invalid law may even drive up the social costs of expressing opposition and intolerance, which strengthens the apparent consensus. Empirical Implications The theory here—​that legislation signals attitudes, which in turn influences behavior—​requires empirical testing. Some existing evidence raises doubts about the theory,28 but more counts in its favor. In particular, consider Patricia Funk’s study finding an expressive effect of the repeal of mandatory voting laws in certain cantons in Switzerland.29 Before repeal, these laws carried only symbolic fines as low as one Swiss Franc (less than $1 in the relevant time periods) for the failure to vote, which for almost 151

The Expressive Powers of Law

anyone is lower than the value they place on their time in voting. After repeal, there was a significant (6–10 percent) drop in voting in those ­cantons where most people were aware of the law being repealed. By comparison, Funk shows that a new system of postal voting did not increase voting despite its dramatically decreasing the transaction costs of voting, making it even more unlikely that people had voted in the past to avoid the symbolic fines. Instead, Funk says the law had an expressive effect on voting. The simplest explanation is the repeal of the law revealed information about a canton-­level change in public attitudes, specifically, that there was less public disapproval of nonvoting than there once was, or that fewer people perceived voting as essential to being a good citizen. If some people voted in part to avoid social disapproval, or guilt from being a bad citizen, then they would now have less incentive to vote.30 The attitudinal signaling model has a variety of testable implications, which, if proven, would be important for policy: • A legislature will have a weaker capacity to generate compliance expressively when more of its constituents believe that the legislature is generally captured by “special interests.” • There will be a stronger expressive effect if the law is enacted over the objection of what the public perceives as “special interests.” • There will be a stronger attitudinal effect the more local the legislature is. Note also an important temporal effect. Even if a statute loses popular support over time, legislative inertia may block its repeal. Knowing this, when one discovers an old statute for the first time, as might a new entrant to a jurisdiction, one must discount its signal by the possibility that it no longer reflects contemporary attitudes. The implication is: • Newer statutes will have stronger attitudinal effects than older ­statutes. Legislators recognize the tendency of a legislative signal to decay over time. The final implication is: • Legislatures seeking to signal attitudes will constantly make minor modifications to old statutes, such as slight and symbolic increases in its scope or penalties. 152

L EGIS L A T ION AS INFOR M A T ION

Risk Signaling Law may also reveal information about risks.31 Instead of there being a correlation between the law and public attitudes, imagine that legislators are more likely to vote for a restriction on some activity if they believe its aggregate social costs outweigh its social benefits. Conversely, suppose legislators are more likely to vote for a mandate or subsidy of some good or service if they believe its aggregate benefits exceed the costs. If so, then law may signal the legislature’s beliefs about costs and benefits. Where attitude signaling derives its power from legislators’ expertise on constituent attitudes, risk signaling works primarily because of the way that legislative voting aggregates the information of individual legislators. The public updates its beliefs about risk in light of the new law. As with attitudes, the public may update beliefs toward the true risks or there might be an availability cascade, in which the public overcorrects and forms exaggerated beliefs about risks. The most likely context for risk signaling is when there is a steady flow of scientific information about the risks and benefits of some activity. Think of the health consequences of smoking, the safety effects of seat belts, child safety seats, or motorcycle helmets, the risks of driving while intoxicated or texting, the effect of carbon consumption on global climate, or the parenting ability of same sex couples. New information on any of these topics may prompt a legislature to change the law regarding them, as by mandating seat belt use or recycling, banning smoking in restaurants, lowering the blood alcohol limit that defines drunk driving, or giving same sex couples the right to adopt children. If so, the new law may signal the legislators’ beliefs about the new information concerning these risks. The revealed information can change public beliefs about the risks or benefits of the behavior and thus change behavior.32 Legislators are mostly not scientific or technical experts on the substantive areas they regulate. The power of the risk signal derives instead from the fact that legislative voting aggregates signals regarding the same fact: the harm or benefit of the regulated activity. The Condorcet Jury Theorem (discussed below) demonstrates that, as long as each individual is, more likely than not, correct, the aggregation of independent votes is more likely to be right than is any one individual. The point is connected to what is sometimes called the “wisdom of crowds.” Even though each legislator 153

The Expressive Powers of Law

has no more expertise than a constituent, the aggregation of legislative votes is a powerful signal of the actual harm or benefit of an activity. For example, a ban on public smoking may signal that legislators interpret new data as revealing that the health costs from exposure to tobacco smoke are higher than previously believed. So the public may in turn believe these costs to be greater. The repeal of a ban on marijuana or handguns may cause individuals to believe those items to pose fewer health or safety problems than previously believed. Legislation recognizing the right of same sex couples to adopt may signal the legislators’ beliefs that such individuals will make good or at least adequate parents. The expansion of the definition of drunk driving may signal that drunk driving causes more accidents than previously understood. Laws favoring the right of mothers to breastfeed infants may signal health benefits of breastfeeding. Of course, some people will strongly resist “learning” from the law’s revelation of information. The “cultural cognition” work of Dan Kahan, Don Braman, and co-­authors shows that people can be essentially immune to all kinds of factual information about risk and regulation, if the information cuts against their fundamental worldviews.33 If conflicting scientific studies entirely fail to move certain people away from their existing beliefs, then legal expression that might be correlated with such studies will undoubtedly fail as well. Kahan and Braman distinguish individuals by their place on two dimensions defined by their polar opposites: (1) egalitarian vs. hierarchical and (2) individualist vs. collectivist. These categories interact to create four main types of individual worldviews—​egalitarian individualists, egalitarian collectivists, hierarchical individualists, and hierarchical collectivists. The worldview explains many beliefs about basic facts related to risk regulation and the resistance to new information. For example, new gun restrictions are not likely to cause individualist hierarchs to believe that guns are any more dangerous to gun owners; new right-­to-­carry laws are not likely to cause the collectivist egalitarians to believe guns are any less dangerous to their owners. Despite this resistance, the law can still, on average, change beliefs about risks. The cultural cognition and other psychological literature mostly shows that people are sometimes stubborn enough to maintain their exist­ ­ing views in the face of unambiguously contrary information. Although it can and does happen on occasion,34 people are not usually moved in the 154

L EGIS L A T ION AS INFOR M A T ION

opposite direction from such information. Thus, the regulation of public smoking may leave individualist hierarchs entirely unmoved but not make them think that tobacco is safer than they previously believed. A law giving women the right to breastfeed children in public may have no effect on the beliefs of some groups as to the health benefits of breastfeeding, but it will not likely make them think that breastfeeding is less healthy than they previously believed. At the same time, other groups will change their minds. Those who favor the new law are likely to update their beliefs as I previously indicated, but cultural cognition implies even greater moves in the direction of facts supporting the new law. Regulation will move the egalitarian collectivist further in the direction of thinking that second-­ hand smoke is risky and that child safety seats are necessary to reduce risk. Moreover, the cultural cognition literature does not claim that every single individual is gripped tightly by a worldview that compels particular beliefs about risk. On any given issue, there are some individuals whose worldview is neutral on the issue, allowing them to be persuaded by the evidence. If those resisting the law retain their same beliefs while those favoring the law, or neutral towards it, update their beliefs, the net effect is still belief change. As an aside, note that I am dividing the models in this chapter by what is signaled—​attitudes or risk—​but I might instead have divided them by the mechanism strengthening the signal—​expertise or aggregation. For attitudinal signaling, I emphasize expertise (because legislators get elected or re-­elected only by being experts on their constituents’ attitudes) and not aggregation (because most legislators represent different districts, their vote is not an aggregation of beliefs about the attitudes of a single group of constituents). Aggregation can still play a role in attitudinal signaling, as where a legislator decides on how to vote by aggregating signals from her staff or where a vote among at-­large representatives aggregates their signal as to the attitudes of the same group of voters. Similarly, for risk signaling I emphasize aggregation because the legislators frequently vote based on their belief about the same fact relevant to risk. I do not emphasize expertise because legislators frequently lack any, but expertise plays a role when legislators with technical knowledge from a prior career serve on the relevant committee or develop expertise by serving on a committee. Returning to risk signaling, the new beliefs about risk will directly change behavior when they concern the costs an individual incurs from his own behavior. A is less likely to engage in drunk driving and more 155

The Expressive Powers of Law

likely to wear a seat belt if A believes the former is more likely to kill him and the latter is more likely to save his life. Even if the ban on public smoking only signals increased risks to nearby nonsmokers who are exposed to second-­hand smoke, a person who is altruistic towards family and friends will want to smoke less in their presence. A pregnant woman who is arranging to give up her newborn to adoption is less likely to reject a same sex couple if she believes such people can be suitable parents. The new beliefs affect behavior indirectly by giving individuals a greater incentive to avoid or control the external costs that others create with their behavior. A nonsmoker who believes there are greater health risks from exposure to second-­hand tobacco smoke will incur greater costs to avoid it, which includes greater efforts to stop a person in his immediate vicinity from smoking. A passenger is more likely to object to a driver talking on a cell phone if the passenger believes the combination of actions poses a greater risk of accident. If there are social norms against the behavior, enforced by third-­party sanctions, then an individual who believes the behavior is more risky may be willing to incur more costs enforcing the norm. Thus, if the expansion of drunk-­driving laws signals greater risk from such activities, then individuals may give more effort to preventing their acquaintances from driving drunk or to shaming them when they do, even though these norm enforcers were not themselves at risk of an accident. To some degree, the risk-­signaling model competes with the attitudinal model. One signal displaces the other. This is not surprising because there is an old political debate about whether legislators should vote just as their constituents want them to vote—​the “delegate” model—​or vote according to their best judgment of what serves their constituents’ interests—​the “trustee” model.35 Legislators who vote according to the delegate model will tend to signal attitudes, while trustees will tend to signal risks. But if there are legislators of both types, we may observe both types of signaling at work. Or an individual legislator may give weight to both factors in his vote (or switch between the factors in different votes), which more weakly signals both attitudinal and risk beliefs. As both signals tend to encourage compliance, attitude and risk signals may reinforce each other. For example, a new ordinance banning smoking in bars may signal lawmaker beliefs that the harms from tobacco smoke are significant and also that community members disapprove of being exposed to second-­hand smoke. Each signal favors compliance. If smokers seek to avoid disapproval, especially embarrassing confrontations with 156

L EGIS L A T ION AS INFOR M A T ION

nonsmokers, then their belief that more people disapprove of their activity raises the perceived costs of public smoking. If smokers believe the risks of second-­hand smoke are greater, and they feel altruistic toward at least some individuals exposed to their cigarette, they will want to cut back on smoking in their presence, including in the places the law regulates. A new law mandating child safety seats provides a similar combination. It potentially signals both legislators’ beliefs that ordinary seatbelts pose significant risks for children and also their beliefs that constituents disapprove of exposing children to these risks. Parents are more likely to comply with the mandate of child safety seats if they believe more strongly (1) in their injury prevention benefits and (2) that community members will otherwise regard them as bad parents. The same may be said for public breastfeeding laws: They signal legislative beliefs in supportive attitudes and health benefits. As with attitude signaling, there may be spillovers to risk signaling, where the belief change prompts behavioral changes not required by the law. If a ban on restaurant smoking causes smokers to believe that second-­hand smoke poses greater health risks, these individuals may be more likely to smoke outside when at home, so as to avoid exposing their loved ones to their smoke. Or they may be more likely to quit altogether. A new law permitting individuals to carry concealed firearms may convince citizens not previously interested in carrying a weapon that doing so will have good consequences, providing them and their community with greater safety. As a final example, consider the work of Dov Fox and Christopher Griffin.36 They argue that the Americans with Disabilities Act had an unintended expressive consequence of increasing abortions of fetuses identified with Down syndrome. They claim that the ADA conveyed to potential parents the information that children with disabilities led a difficult life of constant struggle and discrimination, which caused them to terminate more pregnancies. Whether Fox and Griffin are right as an empirical matter, they identify an important point, the possibility of expressive externalities that are very much contrary to the legislature’s intent in enacting a law. A Model of Risk Signaling A more complete model of risk signaling links the expressive effect of law to the voting decisions of individual legislators. Dhammika Dharmapala and I used the Condorcet Jury Theorem to construct such a model.37 Each 157

The Expressive Powers of Law

legislator has a prior belief π regarding the benefit or harm of some activity X, such as using a seat belt or exposing others to tobacco smoke. Before voting on a bill to mandate or ban X, the legislator seeks information on the activity’s benefit or harm and receives a private signal regarding that fact. For simplicity, we focus on the case where the legislature is considering a ban on X and the signal is either H, that the activity is harmful, or N, that the activity is not harmful. The private signal may derive from their prior experiences and background that give them nonverifiable information pertinent to the issue or that influence how they interpret public information, such as committee testimony. Although legislators make mistakes, assume that the probability p that the signal is correct is greater than one-­half. The risk signaling claim is that legislators tend to vote for the ban if they receive signal H and against the ban if they receive the signal N. Because the ban passes only when more than half the legislators vote for it, one can infer from an enacted ban that more than half of the body received a private signal that the activity X is harmful. As the Condorcet Jury Theorem shows, when signals are independent and each signal is more likely than not correct (about a binary choice), the aggregation of signals is more likely to be correct than any one signal.38 Essentially, when the likelihood that one individual is wrong is some level below 50 percent, the likelihood that a majority of individuals is wrong is even lower. Therefore, the vote aggregation involved in legislation produces a powerful signal of whether the activity is harmful. On average, observers will update their beliefs towards thinking that X is more harmful. A crucial assumption here is that legislators vote their signal. Why would they do that? One motive producing this outcome is that the legislator feels a duty to act in the public interest. A more robust motive is that the legislator believes that voters will judge her vote by the light of later knowledge and punish or reward her accordingly. If there is a steady stream of evidence bearing on the issue of harm or benefit, then future evidence may be more definitive. Assume that if that occurs, the public will be inclined to re-­elect legislators who voted “correctly” by predicting the future consensus, but to punish those who voted incorrectly by this standard. If so, then even though legislators vote strategically, they will tend to vote for a ban when they receive an H signal and against the ban when they receive the N signal. 158

L EGIS L A T ION AS INFOR M A T ION

This more robust assumption, however, can lead to more complex results when we introduce error costs. There are two possible errors: voting “yes” (for the ban) when it later turns out that the activity is not harmful, and voting “no” (against the ban) when it later turns out that the activity is harmful. If probability of error is sufficiently small, legislators will ignore them and vote their signals. But what if the probability of error is significant? However high, if the electoral cost of each type of error is equal, then the error risks offset each other and the legislator should again just vote her signal. But what if the error costs are high and unequal? There are two possibilities. If the legislator perceives greater electoral costs from an erroneous “yes” vote, then she will sometimes vote “no” after receiving an H signal. If the legislator perceives greater costs from an erroneous “no” vote, then she will sometimes vote “yes” after receiving an N signal. The second case, but not the first undermines risk signaling. If the erroneous “yes” votes are more costly (the first case), then the ban will often fail to pass despite the harmfulness of the activity. Yet in these cases, when the legislation does pass, it means that even more than a majority of legislators received an H signal. The legislative signal of “yes” is all the stronger a signal of harm when the error costs push in favor of voting no. By contrast, when erroneous “no” votes are more costly (the second case), legislators may vote for the ban despite receiving an N signal and the legislature enacts the ban despite the fact that most signals show the activity is not harmful. The law does not signal risk. To summarize, the risk signaling story works if the error costs are low, where the error costs are equal, and where the costs of an erroneous “yes” vote exceeds the cost of an erroneous “no” vote. Dharmapala and I claim that legislative lobbying makes the last possibility relatively common, at least for statutory bans and restrictions. Frequently, error costs are high and asymmetric because a bill restricting an activity will trigger opposition by organized interests that benefit from the activity. In a simple public choice framework, small groups with more at stake in a bill have an easier time organizing lobbying pressure than the large groups who stand to incur only diffuse costs or benefits. The error costs are usually greater when one votes against organized interests because they will more effectively exploit a mistaken vote to unseat the incumbent in the next election. Moreover, lobbying also exploits the structure of many legislatures, which 159

The Expressive Powers of Law

gives disproportionate power to committee members, who can block popular legislation. Thus, for legislative bans and restrictions, erroneous “yes” votes are more costly and rare than erroneous no votes. When a legislature enacts a ban despite these costs, there is a signal of harm. For example, when a law restricting public smoking gets enacted over the objections of the tobacco lobby, there is a strong signal of harm. (Later, I consider the interaction of this claim with the attitudinal model). The converse point is that when the public perceives that concentrated industrial or elite interests—​what are commonly called “special” interests— ​are allied in favor of the bill, its passage does not signal risk. A common case may be legislative mandates. For mandates, we may imagine that the legislator receives a private signal of B if the activity being considered creates a social benefit and N if it does not. When well-­organized industrial interests favor passage, the error costs of a “no” vote dominate and legislators vote “yes” despite receiving an N signal. An example may be the use of the child safety seats, where industry gains by requiring purchase of their product. By contrast, however, consider mandatory seat belt laws. The federal government required the installation of seat belts in cars years before states started requiring their use. Because the seat belt suppliers gain from their sale whether or not they are used, it would seem that industrial lobbying would not explain the use mandate, which may still work as a signal of the protective benefits of seat belts. The Problem of Herding Specifying this level of detail now allows us to consider a limitation to the signaling model: the possibility of herd behavior, a belief cascade within the legislative body.39 Herding can arise when decision-­making is sequential and observable, so earlier decisions are signals for later decision-­makers. Legislative voting is usually sequential and open. The first legislative voter has no more information than his own private signal concerning harm (or benefit) and votes accordingly. But the second legislative voter has his own private signal plus the public signal of the first vote, from which he infers the private signal received by the first voter. The third legislative voter has his own private signal plus that of the first two voters. Even if we assume that the private signals each legislator receives are more likely than not correct, it is entirely possible that the first voter or two receives an incorrect signal (i.e., H when the activity is not harmful or N when the activity is harmful). If so, then they can create a cascade of 160

L EGIS L A T ION AS INFOR M A T ION

incorrect votes. Assume the first two legislators receive an incorrect signal H and vote yes, while the third voter receives a correct signal N. Inferring the existence of two H signals from the two yes votes, the third voter reasons that, despite his signal, H is more likely than N. So the third legislator votes yes. All subsequent legislative voters will now reason the same way, voting yes regardless of their own signal. The herd behavior produces legislation that does not signal risk. No doubt this outcome is possible, but there are several reasons to think that the law will still often work as a signal of risk. First, herding depends critically on how much one legislator weighs the signal received by another. One might expect confident legislators to give much less weight to the information revealed by other legislative votes than to their own information. If so, and given that erroneous signals are less common than correct signals, it will be rare that there are sufficient initial votes based on erroneous signals to cause herding. Second, legislative structures may prevent or limit herd behavior. If, prior to any actual vote, each legislator simultaneously casts an anonymous nonbinding vote indicating how they plan to vote—​a straw poll—​then every legislator would see all the signals received by the legislators before the actual vote. To some degree, committee and procedural votes and caucusing may serve this function. Recall that the public can also engage in herding—​what I previously discussed as informational cascades. The important point is that these structures make it less likely that the legislature will herd than will the public. This ­comparative advantage means that legislation is likely to be informative about risk. In any event, where it occurs, herding is still more likely to produce legislation consistent with the majority of signals than legislation contrary to them. If the signals are usually correct, then it is more likely that the first one or two legislators will vote according to the more common signals—​correctly rather than incorrectly. Early voters’ influence on subsequent legislative votes will usually be in the correct direction. Legislative bans will cause individuals to update towards believing the banned activity is more harmful. Empirical Implications The theory here—​that legislation signals risks, which in turn influences behavior—​requires empirical testing. Some existing evidence is ­suggestive. One study found that seat belt laws increased seat belt use “expressively,” 161

The Expressive Powers of Law

that is, beyond the amount that legal enforcement can explain.40 The study didn’t test among expressive theories, but, because the primary beneficiary of the law is the person who complies with it, it is plausible that the mechanism of change is the one just described. Individuals update their beliefs about the benefits of seat belts in light of the mandate; perceiving a greater risk for beltless driving, more people “buckle up.” The risk signaling model has a variety of testable implications. The first three are familiar from the discussion of attitudinal signaling, but the fourth and fifth implications are entirely distinct: • A legislature will have a weaker capacity to generate compliance expressively when more of its constituents believe that the legislature is generally captured by “special interests.” • There will be a stronger expressive effect if the law is enacted over the objection of what the public perceives as “special interests.” This will ordinarily mean that legislative bans affect beliefs more strongly than legislative mandates. • Legislatures seeking to signal risks will constantly make minor modifications to old statutes, such as slight and symbolic increases in its scope or penalties. • There will be a stronger risk signal the larger the legislature is. Given the relative size of Congress, state, and local legislatures, federal statutes provide stronger risk signals than state statutes and most state statutes provide a stronger risk signal than local ordinances. • Legislative structures that impede herding behavior will strengthen the effect of the risk signal.

Violations Signaling and “Crowding Out” Compliance We have so far focused on how changes in the content of legal rules can signal information and change behavior. But now consider the signaling effect of a change in legal sanctions. Legislative changes in sanctions are part of a broader set of signaling possibilities based on enforcement processes, addressed generally in the next chapter. Here, we are focused on legislative signaling, so consider legislative changes in enforcement processes, such as an increase in sanctions. A legislative change in sanctions can signal things other than the legislators’ beliefs about public attitudes and the risks and benefits of the regulated behavior. It might instead signal the legislators’ beliefs about the 162

L EGIS L A T ION AS INFOR M A T ION

violations of existing laws. Suppose a legislature raises the penalty for a behavior, e.g., drunk driving. It is possible to infer that the legislature believes that drunk driving is more harmful than before or subject to greater disapproval. But another, frequently more plausible inference is that the legislature reacted to the discovery that violations are more frequent than previously believed (because they are rising or have previously been underestimated). Certainly when penalties are raised for crimes like drunk driving, the legislators who vote for it will often point to cases of failed deterrence. Signaling greater-­than-­expected violations will, other things equal, weaken compliance with the law. For most legal requirements, it is easier to observe legal sanctions and enforcement actions than to observe violations. Holding constant the resources devoted to detecting violations and the severity of sanctions, the more observed violations, the lower the probability that any given violator is detected and sanctioned. The lower the probability of detection, the lower the expected cost of a legal violation. Thus, holding constant enforcement resources and sanction severity, signaling a higher level of violations weakens deterrence and lowers compliance. If raising the penalty signals a lower probability of detection, it is an empirical question whether the first effect offsets the latter. By signaling a failure of deterrence, increasing penalties could conceivably undermine deterrence. (The conventional claim that increasing penalties increases deterrence remains true, given the contrary assumption that one’s estimate of the probability of detection is unaffected by the change in penalties). The same adverse effect of violations signaling can arise for a reason other than its effect on conventional deterrence. If compliance is based on reciprocity, then the signal of more violations will trigger reciprocal violations. As discussed in Chapter 4, some theorists argue that tax compliance is best explained by reciprocity.41 The claim is that some people—​“reciprocators”—​ pay their taxes because they believe that others pay their taxes. Convince them that many other people cheat, and these reciprocators will cheat. Whether deterrence or reciprocity underlies the effect, note the key conditions for this adverse signal to occur. First, violations signaling is not plausible for most of the examples previously discussed because they concern the initial decision to restrict or mandate conduct or to expand the scope of a prior restriction or mandate. A change in attitudes or perceived risks can fully explain the initial decision to create legal liability for some 163

The Expressive Powers of Law

conduct. Indeed, for many of the examples, e.g., public smoking, seat belt usage, civil rights, everyone assumed that the behavior targeted for legal restriction is common before the ban, but may still expect that the new enforcement measures will suffice. Only when the legislature returns to the issue and raises the sanction does it possibly imply that prior enforcement was inadequate. Second, the adverse inference about violations depends on the public having a noisy estimate of the number of violations. If violations are publicly observable and the public has confidence in its estimate, then it will attribute higher penalties to new legislative assessments of attitudes or risk—​that the prohibited conduct causes more social disapproval or harm than previously believed. For example, one can readily observe public behaviors such as smoking in restaurants, riding a motorcycle without a helmet, and broadcast indecency. One might not observe individual cases of water consumption, the creation of graffiti, or the failure of dog owners to clean up their dog’s waste in public spaces, but the overall rate of such violations is entirely public. By contrast, if the violation occurs in private, the public has low confidence in its estimate. Examples are plagiarism, copyright violations, tax evasion, drunk driving, sexual harassment, and blackmail, each relatively unobservable and subject to noisy estimates. Here, it is rational to infer that the legislature may have decided to punish more because it revised upward its estimate of violations. This signaling account offers an alternative interpretation of data on the “crowding out” of intrinsic motivation by external sanctions.42 These studies identify situations where an increase in formal sanctions produces less compliance. Although there may be several causal mechanisms at play, one possibility is that the rise in sanctions signaled a lower probability of detection than individuals previously believed, and the net effect is to decrease deterrence. A famous example of “crowding out” comes from a study Uri Gneezy and Aldo Rustichini conducted of Israeli daycare centers.43 Initially, the centers had no enforcement of the rule that parents show up by a certain time in the afternoon to pick up their children. When six of the daycare centers gave notice of a monetary fine for late pickups, the number of late pickups rose. And when the centers later abandoned the fine, the late pickups failed to fall back to the initial level. There are several possible explanations for this counter-­intuitive phenomenon. One is violations signaling: increasing the sanctions from zero 164

L EGIS L A T ION AS INFOR M A T ION

signaled that there was a problem with the number of violations, that there were more of them than the parents previously assumed. Note that tardiness is not fully transparent because, if parents picks up their children and leave before the deadline, they do not observe how many children remain after the deadline. Thus, when parents revise upward their estimate of the number of other parents who violate the deadline, the expected informal sanctions for the violation decline. First, the parents expect less disapproval from lateness if many other parents are late. Second, if the parents meet the deadline out of a preference for reciprocity, that preference now permits a certain amount of tardiness. Third, the initial lack of a stated fine did not guarantee that the daycare would not do something more drastic, like forbidding the parents from using the daycare in the future. But this response seems far less likely once one realizes that the daycare centers had tolerated many violations in the past (and have now set an upper bound on punishment—​the fine). All three of these mechanisms are triggered by the initial signal—​from the sanction—​that there are more violations than previously believed. What about the final stage, where the daycare centers abandoned the fine? Violations signaling can also explain why tardiness would not fall back to the initial level. Crucially, while violations are not transparent to those who pick their children up before the deadline, they are transparent to other violators who can observe the number of children remaining when they arrive after the deadline. Thus, if the quantity of violations has been high for a time, then more parents have the occasion to observe other parents being late. Knowing that tardiness is common, a parent does not infer from the abandonment of the fine that tardiness is rare, so the damage of the initial violations signaling is not removed by abandonment of the fine. Violations signaling need not be reversible, so it can explain the behavior in this study usually attributed to the crowding out of intrinsic motivation.

Preference Change and Social Meaning I now comment on two other expressive theories of law. The first is preference change: economics explains behavior as the effort to satisfy preferences subject to one’s beliefs and opportunities. Ordinarily, economics views law as shaping opportunities by raising or lowering the costs of different behaviors. Less conventionally, however, law might change behavior 165

The Expressive Powers of Law

by shaping the individual’s underlying preferences, so that the individual no longer has the desire to act in a way contrary to the law’s command.44 A white person living in a racially segregated society might initially desire this segregation, but lose the desire after living for some time in an integrated society. A person who abstains from wearing seat belts may desire the freedom and convenience of being unbelted, but after wearing a belt for a time, develop a habit for doing so, such that he feels uncomfortable—​ exposed—​when unbelted. In each case, the older person might look back on the younger and not understand how he could have wanted what he wanted then. What should be obvious about such a preference-­shaping theory is that it must be parasitic on some other theory of legal compliance. It is difficult to imagine the racist changing his preferences if the law forbidding segregation has no effect on anyone’s behavior. What makes preference change plausible is the cognitive dissonance that occurs when a person who wants segregation lives in an integrated environment, but if he keeps living in segregated conditions, there is no dissonance and no preference change.45 Thus, the law affects preferences only if, for some other reason, it first changes behavior. The point is even more obvious for habit-­formation. One acquires the habit of wearing a seat belt only if some other mechanism first induces one to wear the seat belt. Similarly, preference-­change may entrench behavioral change that came about initially for some other reason, but unlike the expressive theories I propose, it cannot work expressively on its own. The law first changes behavior by creating a focal point, revealing information, or something else, and then over a period of time, it might also change preferences or habits. Second, law might change the social meaning of an action. Larry Lessig first observed how social meaning could arise from and reinforce a behavioral regularity.46 For example, when the law doesn’t require seat belt use and most people regularly ride in cars unbelted, a passenger’s statistically deviant decision to buckle up demonstrates distrust of the current driver’s ability to avoid an accident. Because passengers wish to avoid giving offense, the social meaning of wearing a seat belt raises its cost, which helps to sustain the regularity of riding unbelted. Even the driver might fear that his use of a seat belt would signal to passengers that he is a bad driver or is unusually fearful. Similarly, when the law permitted dueling and most men accepted challenges, refusing a challenge “meant” one was a coward. The social meaning 166

L EGIS L A T ION AS INFOR M A T ION

of refusing to duel helped to perpetuate dueling. In the Jim Crow South, when law permitted race discrimination and most whites discriminated against blacks, a white person’s act of nondiscrimination conveyed strong dissent from the prevailing racial orthodoxy. Because such dissent would incur social and economic ostracism, the social meaning of nondiscrimination raised its costs. If so, then Lessig and Dan Kahan argue that law affects behavior, in part, by its ability to change social meaning. A new law provides a new motive for behavior and therefore “ambiguates” the existing meaning.47 If seatbelt use rises dramatically after the mandate takes effect, then it is harder to infer from seatbelt use a belief that the particular driver is not safe. After the new anti-­discrimination law, one cannot necessarily infer that Jim Crow era whites reject racial orthodoxy merely because they fail to discriminate; they may simply seek to avoid legal sanctions. In each case, the law changes the social meaning by supplying another inference for the behavior. By compelling some behavior, law lowers its disapproval costs. Thus, law creates an incentive to comply that is distinct from the threat of legal sanctions. Yet this “social meaning” story is, like preference-­shaping, parasitic on some other mechanism of behavioral change. The social meaning changes only if many individuals first comply with the law for reasons other than a social meaning change. If no one initially obeys a new seat belt law and usage remains low, then everyone will still infer from a passenger’s use of the belt that he does not trust the driver; attributing the behavior to the law is implausible if everyone else ignores the law. Similarly, if everyone else accepts challenges to a duel despite the new law, then no one will believe a man’s claim that he refuses a challenge because the law forbids it. In the usual case, if overall behavior remains constant, that behavior’s meaning will remain constant. One might respond that the social mean­ ­ing changes instantly if people merely expect others to obey the law, even before their behavior changes. But even this expectation must precede the social meaning change, which means social meaning change does cause the shift in expectations. People expect the new law to change behavior for some other reason—​sanctions, legitimacy, or focal point—​and that other mechanism then drives the change in social meaning. Thus, when we catalogue the ways that law influences behavior, we should not include preference-­shaping or social meaning change as independent mechanisms. Neither will change behavior on their own. Both 167

The Expressive Powers of Law

remain important as potential multipliers, that is, if some other mechanism (e.g., deterrence or legitimacy) causes compliant behavior, then the new behavior may change preferences or social meaning, which in turn causes even more compliant behavior. Importantly, these multipliers can enhance the importance of expressive theories. If the law influences behavior by providing a coordinating focal point or by signaling information, then the new behavior can shape preferences and change social meaning in a way that multiplies the initial expressive effect. For example, enactment of a seat belt law may signal the personal safety advantages of wearing a seat belt and this information may cause seat belt use to rise from 40 percent to 60 percent. Once the expressive effect tips usage up to a majority, the social meaning may change and cause usage to rise even further. In sum, lawmakers signal information on a variety of matters, some of which produce legal compliance expressively.

168

6 Revelation of Information by Legal Enforcement

L

egal commentators, courts, prosecutors, and defense lawyers often say that some enforcement action, such as a criminal conviction or acquittal, will “send a message.”1 Often these speakers express great concern that the message sent will be the wrong one or possibly the right message long overdue. On examination, sometimes the speaker means nothing more than deterrence—​the imposition of sanctions will send a message that society will punish some kind of transgression; the decision not to punish will send the opposite message. But the “send a message” claimants sometimes see a far more symbolic weightiness to the message the enforcement action implies, as if there is something more at stake. For example, when legal commentators claim that recognition of battered women syndrome, the use of peremptory challenges against potential jurors on the basis of sex, or rape acquittals will reinforce sex-­based stereotypes, they clearly identify a message other than deterrence (or its absence). Sometimes these criticisms merely identify potentially objectionable elements in the logic of the laws, saying essentially, that if you believe in sexual equality (or some other value), you should not accept the rationale for these laws. But expressive claimants sometimes allude to something more consequential, that the message being sent will or at least might have some practical effect on the world, as by strengthening (or weakening) the grip that sexual stereotypes have on the public. In this chapter, I explore the possibility that legal enforcement might affect behavior by sending a message other than deterrence (i.e., that we 169

The Expressive Powers of Law

punish or don’t punish this sort of thing). As in the last chapter, enforcement may convey information and affect beliefs. I use this analysis to support some send a message claims and critique others. As far as I know, no one claims that enforcing the law against murder “sends the wrong message” by suggesting that people are violent or that victims are weak. Not many people worry that the government sends an anti-­individualist message undermining responsibility when it enforces rules of safe driving. To understand when government enforcement sends a consequential message (other than deterrence), we have to understand when enforcement fails to send such a message. To distinguish plausible message-­sending claims from clichéd rhetoric, we need specificity about the causal mechanisms involved. As we shall see, one crucial component is to focus on the audience who, as a result of the enforcement, actually receives a message. It is a cheap criticism to say that a prosecution, jury verdict, or criminal sentence sends the wrong message if the critic of the enforcement decision is free to characterize its meaning in any way. The prior chapter focused on legislative signaling, but the other branches of government also supply information through law. I focus on them here, especially courts and executive officials. I emphasize criminal punishment. As many have noted, punishment has a strong expressive dimension because it denounces or condemns the behavior of the punished. I explore how these executive and judicial actions can reveal information and I seek to identify the conditions when these governmental actions will not reveal information, or not in a way that plausibly affects behavior.

Executive and Judicial Signaling: Generally Consider legal rules announced by the executive or judicial branch through executive orders, administrative rule making, and the common law. Can these rules reveal information about public attitudes, behavioral risks, or anything else that influences behavior? I will identify some limitations to such claims, but let us begin by understanding how the expressive theory of the prior chapter could apply to the legal decisions of executive and judicial officers. These decisions could reveal information about attitudes, risk, or violations. For attitudinal signaling, what is required is a perceived correlation between public attitudes and the rules that chief executives, agency admin­ istrators, or judges articulate. The correlation may easily exist in the ­executive branch. Presidents, governors, mayors, and some other executive 170

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

officials are elected. In the many cases where chief executives appoint the head of an administrative agency, they tend to select individuals who will not endanger the executive’s public standing with popular opinion. Once appointed, chief executives and legislators tend to pressure administrative heads to render popular decisions or at least to avoid unpopular ones. Despite the common view of judges as immune to public pressure, even judicial decisions could be a source of information about popular attitudes. First, many American state judges are elected; for any case that might become publicized, they may gravitate towards decisions they believe are correlated with public sentiment.2 Appointed judges, according to political science models of strategic voting, seek to avoid legislative overruling, which means their decisions are tethered to public opinion on statutory and common law issues.3 Even when federal judges with life tenure decide constitutional matters, they know that their powers depend on some level of public acceptance (as explained in Chapter 3, on the expec­ ­tations that others will follow their orders). Justice Brewer acknowledged the point when he said that “the power of the courts rests on the general support of the people.”4 More recently, empirical studies suggest that even the federal bench, rather than being principally a counter-­majoritarian institution, only rarely issues decisions defying a national majority.5 Judicial signaling of attitudes is consistent with occasional judicial backlash, where an unpopular decision provokes the public into actions that push policy back towards and beyond the status quo ante.6 Judges do occasionally and famously misjudge the level or intensity of public attitudes or their importance as a practical restraint on judicial action.7 Classic examples are Roe v. Wade, prompting a pro-­life movement that has scored decades of political successes, and Furman v. Georgia,8 which prompted thirty-­five states and Congress to re-­enact their death penalty statutes and reinvigorated (for a time) use of the punishment. A more recent example is Kelo v. City of New London, which upheld expansive state power of eminent domain and prompted popular anger and a wave of state laws restricting that power.9 The studies finding that the federal bench is rarely counter-­majoritarian show that backlash is the exception, not the norm. The more typical decision reflects public opinion either by upholding democratically enacted laws against constitutional attack or by striking down state and local laws that are national outliers. Examples of the latter include Loving v. Virginia,10 invalidating the few state laws that still banned interracial marriages, and 171

The Expressive Powers of Law

Lawrence v. Texas,11 invalidating thirteen state laws still prohibiting sodomy. It is the fear of backlash that drives judges to consider and defer to public sentiment or at least seek to “lead” public opinion by only a small amount in the direction of existing trends. This is why Alexis de Tocqueville noted that American federal judges had to be “statesmen,” so they could “confront those obstacles that can be overcome, and to steer out of the current when the tide threatens to carry them away, and with them the sovereignty of the Union and obedience to its laws.”12 Thus, people may generally infer from the statement of a legal rule in an executive order, agency regulation, or even a judicial opinion that public attitudes (at some level) are more in line with the rule—​say, more disapproving of prohibited behavior—​than previously believed. A new ruling that hostile work environment sexual harassment constitutes an unlawful form of sex discrimination may signal that the judges or agency heads who issued the ruling believe there is greater public disapproval of such behavior. A new ruling that the second amendment protects individual rights to gun ownership may signal that the judges believe there is more approval (or less disapproval) of gun ownership. As before, local officials have more influence because they signal local attitudes, which are the attitudes having the greatest affect on behavior. The effect may be marginal but marginal changes are important in a long-­term struggle that might produce a series of changes in the same direction. Also, marginal changes might lead to discontinuous effects for reasons discussed in Chapter 5. The judicial and executive branches may also influence behavior by signaling risk if the public perceives that their legal decisions are correlated with the costs of the regulated behavior. The expressive influence is plausible because the whole point of some administrative agencies is to determine certain facts about risk, such as environmental effects of pollution or the health or safety of food, drugs, and consumer goods. If the structure of the bureaucracy creates incentives to carry out these objectives, then the regulations restricting goods or activities will signal administrator beliefs about their risks, giving individuals stronger reasons to avoid them. A decision to ban a pesticide signals health harms. Conversely, the decision to mandate some medical procedure, such as a vaccine, signals health benefits. Judicial opinions upholding or rejecting agency decisions might send the same signals. The individual updates his beliefs. Many of the beliefs promote compliance, as the individual’s concern for his own 172

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

health makes him more likely to seek the required vaccine and more likely to avoid the prohibited pesticide. But there could be “spillover” effects, independent of compliance. For example, an administrative agency regulation of occupational exposure to workplace chemical hazards might cause individuals to avoid those chemicals in consumer products and a purely prospective safety standard for playground equipment might cause parents to avoid existing playgrounds that do not meet the standard, even though neither the chemical nor the playground are technically noncompliant. Some administrative or judicial rule making may signal attitudes and risks. Previously I said that a pro-­gun rights judicial opinion might signal public attitudes favoring guns, but a decision striking down particular safety regulations, such as gun safes or locks, might also signal the judge’s beliefs about the net benefits of gun possession. A judicial decision permitting same sex couples to adopt children may reveal the judge’s beliefs both about risks—​that being gay or lesbian is consistent with good parenting—​ and attitudes—​that public tolerance of gay parenting is increasing. An administrative agency decision to permit the marketing of a morning-­after birth control pill certainly signals the agency’s belief that the side-­effects are acceptably minor, but may also signal that the agency believes the public will tolerate and use the pill (since there is no reason to incur the wrath of those opposed to it if there is no one who favors it). A judicial decision announcing a necessity defense to the possession of marijuana intended for medicinal use or an agency decision recommending its medical use for certain purposes will certainly signal the judge or agency’s view of the net benefits of marijuana use in certain circumstances and may also imply a growing public acceptance of it.

Enforcement Signaling: Generally Once we set aside an exclusive focus on legislatures, we can move beyond the expressive effects of lawmaking to the expressive effects of legal enforcement. With enforcement, we should consider judges, juries, and a broad set of executive agents. Police issue citations and make arrests; prosecutors indict and try criminal cases; administrative agencies litigate against putative violators; juries render verdicts; judges make preliminary rulings, find facts, issue judgments, and, on appeal, affirm or reverse the decision below. Any enforcement action—​whether or not the action succeeds in imposing sanctions—​may provide information to the public. 173

The Expressive Powers of Law

The most obvious information provided is the existence of the legal rule being enforced. A person who does not know that the local law forbids the private possession of Tasers or the use of a cell phone while driving may learn of these rules by observing a police officer make an arrest or reading media coverage of an indictment or criminal trial. Discovering the legal rule could in turn matter in conventional ways—​triggering the power of deterrence or legitimacy—​or the expressive ways just discussed—​because the law signals attitudes or risks. Enforcement signaling is familiar within the classic theory of deterrence. Sanctions generate deterrence only if potential offenders take past sanctions as information about future sanctions. The damages or criminal sentences the judge imposes at time 1 (T1) deters behavior at time 2 (T2) only because T1 sanctions provide information about the expected T2 sanctions. The same is true about the probability of detection. Arrests and prosecutions for burglary at T1 promote deterrence only if potential burglars take the detection rate for burglaries at T1 to be predictive of the detection rate at T2. Why would present enforcement action correlate with future enforcement action? Increased convictions reveal that prosecutors (and possibly judges) have decided to allocate more enforcement resources to this offense. Increased detection shows that police have either become more efficient in fighting this crime or have decided to commit more resources to it (or both). In sum, present enforcement tends to reveal that enforcers have the knowledge, resources, and motivation to detect and punish to a certain degree. Enforcement decisions may also have more complex signaling effects on deterrence. If one observes more cases where an offense is detected and punished than one expected, then we have considered the possible inference that the enforcers are devoting more resources to its detection because they believe that public attitudes have shifted towards greater disapproval of the behavior. We have considered the inference that the best scientific understanding of the risks shows the behavior to be more harmful than previously understood. Thus, if the public observes an increase in prison sentences for drunk drivers, they may infer that the behavior engenders greater disapproval or poses greater public dangers than they previously understood. Either inference reinforces compliance. Yet there is a very different kind of inference we also saw in the last chapter: There might be more than the expected number of violations detected and punished because there are more violations than one 174

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

expected. The same level of enforcement effort and efficiency will produce more apprehensions if there are more violations. I have referred to this as violations signaling. If violations are higher than expected, it may be because one’s prior assessment of deterrence (the expected costs from committing a violation) was too high. If penalties are observable, the likely source of error was overestimating the probability of detection. Updating for a lower risk of detection, there is less deterrence. Moreover, if violations are higher, it is likely that the disapproval of each violation is lower. Many theorists have noted this inverse relationship between the number of norm violations and the intensity by which norms violations are condemned.13 The result is parallel to the way enforcement may “crowd out” intrinsic motivation. The net effect of new enforcement actions therefore is ambiguous without knowing more about the inferences people will draw from them. As a law professor, I have thought of these effects when discussing whether a school should announce the detection and punishment of student plagiarism (not the name of the offender, which poses separate issues). If students hear nothing about student discipline for plagiarism, will they infer that it rarely occurs and therefore that the probability of detection is high? If so, then announcing the discovery of many cases of plagiarism may cause students to realize that the offense is not well deterred, as they previously believed, meaning that the probability of detection is not so great. Or will students infer from their hearing of no cases that the probability of detection is low? If so, then announcing the discovery of many cases of plagiarism will cause them to infer that the probability of detection has risen and also possibly that the professors are expending greater effort at detection because of greater disapproval of the misconduct. The point is that enforcement can reveal information that produces deterrence and occasionally undermines it. Nonetheless, I have said throughout the book that I am interested in identifying the expressive effects of law on behavior other than deterrence or legitimacy. So I do not refer to the enforcement effects producing deterrence as expressive. Instead, I turn to the other possibilities and discuss how enforcement may reveal extra-­deterrence information. Before we get there, it will help to narrow our focus to a more specific area of law: criminal punishment. I reinterpret the standard claim that criminal punishment is “expressive” as being a claim that criminal punishment reveals information about attitudes and risk. 175

The Expressive Powers of Law

The Expressive Significance of Criminal Punishment It is an understatement to say that the governmental acts bringing about criminal punishment are expressive. Joel Feinberg has famously argued that criminal punishment is defined by the fact that it expresses condemnation of the criminal act.14 Whether or not Feinberg is correct in his definitional claim, it is worth reviewing his account because it illuminates the informational content—​the signaling—​of criminal law enforcement. Feinberg distinguishes “punishment” from other instances where an authority imposes some adverse consequence or “hard treatment” on an individual. For example, the money an individual pays the government to enter a park, to drive on a toll road, or acquire a passport are not “punishments” for using the park, the road, or passport. They are merely prices one pays for these privileges. Basic military training involves hard treatment, but it too is not a punishment for volunteering for military service or submitting to conscription. These easy cases illustrate the simple point that punishment is not just hard treatment an authority imposes, but hard treatment an authority imposes on account of an individual’s “infraction of a rule or command.”15 Feinberg goes one step further. He says that hard treatment does not always qualify as “punishment” even when an authority inflicts it on an individual on account of his rule violation. A government agency might fire one of its employees on account of a rule infraction. Similarly, because of a rule violation, a public school might flunk one of its students and a department of motor vehicles might rescind a driver’s license. But Feinberg seeks to distinguish these categories as mere “penalties” rather than “punishment.” The basis of this distinction is expression. Punishment expresses condemnation of the criminal, while mere penalties do not. He explains: “[P]unishment is a conventional device for the expression of attitudes of resentment and indignation, and of judgments of disapproval and reprobation, on the part either of the punishing authority or of those ‘in whose name’ the punishment is inflicted. Punishment . . . ​has a symbolic significance largely missing from other kinds of penalties.”16 According to this classification, we can have hard treatment that does not symbolize indignation and reprobation (or at least not very much)—​those are mere penalties. And we can have expressions of indignation and reprobation without (additional) hard treatment—​that would be mere denunciation. The two together constitute punishment. 176

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

Again, I am not here interested in whether Feinberg is ultimately correct in his analytic definition of punishment, nor in exploring the ensuing philosophical literature discussing the expressive justification of punishment.17 These are normative theories of expressive law. (I do explore in Chapter 8 whether Feinburg’s definition is useful for constitutional doctrine that determines when the government is punishing). For my positive purposes, what matters is that Feinberg has identified an important and typical characteristic of punishment (whether or not it is necessary or sufficient): its expressive character. His definition is surely plausible and its plausibility reveals the particularly rich expressive dimension of criminal law. When a society wants to express something through law, one of its most powerful conduits is punishment. Labeling a person a “criminal” or “felon” is a more powerful statement of public attitudes—​disapproval, indignation, or revulsion—​than is labeling one a “tortfeasor,” statutory “violator,” or even “lawbreaker.” The point is basic and yet sometimes forgotten, as David Garland observes.18 When New York reinstated the death penalty by statute in 1995, quite a few critics attacked the law on the ground that it was mere political theatre, an empty symbolic gesture or “side-­show” akin to the legislature emotionally acting out. This “irrational-­because-­symbolic” critique was, as Garland observed, fundamentally flawed, being based on the false idea that the only valid purpose of criminal punishment is instrumental—​the control of crime through deterrence or incapacitation. To the contrary, “condemnation is a major end of criminal justice and symbolic communication a major means for achieving this end.”19 Whatever its ultimate (de)merits, the death penalty effectively conveys a strong message, that capital murderers “stand[] condemned of the worst of all crimes” and “do not deserve to live.”20 One may not value the message or think the death penalty sends other less desirable messages, but the concept of expressive condemnation is a coherent and pervasive aspect of criminal punishment. We also see the importance of condemnation in the expressive-­politics literature. In his famous article “What do Alternative Sanctions Mean?,”21 Dan Kahan asked why the public was so unenthusiastic about the use of nontraditional punishments—​fines and community service—​even for minor crimes, when prison is so costly and brutal. He proposed that the public does not perceive fines or community service as being sufficiently expressive of what punishment is supposed to express, what Feinberg called 177

The Expressive Powers of Law

“attitudes of resentment and indignation and judgments of disapproval and reprobation.” (For simplicity, let us call all of these “punitive attitudes”). Fines appear only to be prices, which lack any element of condemnation, while community service involves the kind of public-­spirited work that usually earns admiration and praise. Kahan once proposed that the state use more shaming sanctions, perhaps in combination with fines and community service, to show the public that there are ways to express condemnation effectively without prison.22 But regardless of the much-­debated merits of shaming sanctions, the basic point is that the public demands not just that the state impose harsh treatment on criminals, but that the state imposes harsh treatment that expresses punitive attitudes. I want to suggest a new aspect of the expressive theory of criminal punishment, as part of my causal inquiry into legal influences on behavior. Individual acts of punishment plausibly reveal information to the public. Prosecutors, juries, and judges express their attitude of disapproval (perhaps on behalf of the victim) through the acts they take to bring about the punishment of an individual accused of crime. On rarer occasions, these actors might through punishment express beliefs about the risks—​the social costs—​of the punished behavior. As in the last chapter, the revelation of information may cause individuals to update their beliefs and change their behavior. Indeed, Garland anticipates my claim when he argues that there is no clear distinction between the expressive and instrumental aspects of punishment: “It ought to be clear that symbolic acts have direct effects on attitude and perception and therefore, one might expect, indirect effects on action . . . . ”23 For example, in a case of acquaintance rape, commentators say that the prosecutor’s decision to bring charges or the jury’s decision to convict will send a message that the behavior is not tolerated while the decision to decline prosecution or acquit will send a contrary message. If this refers to something beyond simple deterrence, what could it mean? One possibility is that enforcement decisions reveal information about public attitudes or risk. Assume the rape accusation is particularly controversial because the man claims that he did not believe that the woman’s expressed refusal—​“no”—​was sincere because she had acted in a sexually suggestive and receptive manner during the date and invited him into her apartment for a drink afterwards. Perhaps the media gives great publicity to the case, and it becomes the subject of great discussion within the community. A conviction might then send a message of risk—​that the public 178

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

believes that the facts of the case show the absence of consent and a serious harm—​and attitudes—​that the public is indignant or outraged over the nonconsensual intercourse in the case and considers the violation of sexual autonomy as a grave wrongdoing. An acquittal might send the opposite messages. In the end, these claims are empirical. What I want to do now is to show how the information theory bears on the plausibility of the message-­ sending claims. I first identify the conditions for punishment to “send a message” about attitudes or risk. I then examine some examples, distinguishing cases where the theory makes the expressive claim plausible from cases where it is implausible. The Conditions in Which Enforcement Decisions Signal Attitudes or Risk The theory of law as information from the last chapter identifies the conditions where law can affect the world by sending a message. Here I identify the conditions for executive or judicial enforcement actions to send a message. I conclude that executive and judicial expressions can reveal information so as to induce behavior change, but they are not as likely to do so as legislative action. With legislative signaling, I limited myself to examples where the public could easily infer the facts correlated with the legislative action, such as attitudes and risk. The message that affects beliefs is the one that is received, so if we are to apply signaling theory to enforcement, we must focus on what most people take the enforcement action to imply. Unlike literary criticism, where it may be interesting to identify interpretations of a text that have rarely or never occurred to anyone else and are based on speaker’s meaning or sentence meaning, the only behaviorally significant meaning is audience meaning. Thus, the first condition for a consequential message is that the enforcement action carries some clear audience meaning. When evaluating causal claims about “sending a message,” we are really looking for the message received. The second condition is publicity, that many people receive the message. Recall from the last chapter that legislation reveals information only when the media or other sources give the law sufficient publicity to come to the attention of many members of the public. The same is true of judicial or executive enforcement decisions—​many members of the public must learn of the decision for it to have any expressive consequence. 179

The Expressive Powers of Law

Third, individuals are constantly bombarded by information from sources other than the law: the print media, Internet, social acquaintances, etc. For expression to change beliefs, there must be some factor that makes the legal signal strong enough to stand out against this background. Absent empirical inquiry, there is no a priori way to estimate how strong the signal is, but we can say something about the plausibility of the claim prior to empirical testing. (We can thereby identify the most promising cases to test). In the last chapter I offered two mechanisms that amplify the informational content of the legal signal: expertise (because legislators know more than the public about some issue) and aggregation (where the Condorcet Jury Theorem applies, the majority of a group is more likely to be correct than an individual). I primarily used expertise to explain the influence of attitude signaling and aggregation to explain the influence of risk signaling. So the third condition is that the enforcement message involves aggregation or expertise. Either the expression represents the aggregation of multiple enforcers or the enforcers bring to bear on the issue some special expertise in the risks involved. In sum, there are three conditions for an enforcement decision to change beliefs—​(1) a clear implication; (2) publicity; and (3) expertise or aggregation. Some Implausible Claims About Law’s Expressive Effects As I said in Chapter 1, this book addresses at least two audiences: economists who ignore the nonsanction effects of law and legal commentators who assert expressive claims in a casual and untheorized manner. This section is addressed to the latter group and emphasizes limitations to the information theory. For each condition specified, I will now identify some expressive claims about legal processes (not all of which are criminal) that are implausible because the condition does not hold. The first condition—​a clear audience meaning—​is the one that commentators most frequently ignore. There is a temptation for legal critics to assert an implication the enforcement action has for them without much regard for predicting that others will give the action the same interpretation. Consider, for example, Justice Stevens’ claim in his dissent from Zatco v. California,24 where the Supreme Court for the first time refused to allow two petitioners to proceed in forma pauperis, that is, without paying filing fees, because they had repeatedly brought frivolous claims. Stevens says: “Although the Court may have intended to send a message 180

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

about the need for the orderly administration of justice and respect for the judicial process, the message that it actually conveys is that the Court does not have an overriding concern about equal access to justice for both the rich and the poor.”25 Note that the claim here is what the message “actually” conveys; it is apparently about audience meaning. Yet, viewed as such, the claim is implausible. It is not merely that there is no empirical evidence behind this assertion; empirical evidence is rarely to be found on either side. The problem is that Justice Stevens is really just stating how he interprets the majority’s opinion and then projecting that view onto the public. Given low publicity, the public may have no view of the matter. And there are other interpretations. If it has any view, the public could easily see the majority’s decision as being consistent with the Court having a “concern about equal access to justice” that is powerful but not exclusive; a concern that must be balanced with judicial efficiency, which includes the practical need to prevent frivolous claims from diverting the Court’s attention from meritorious claims, some of which also come from the poor. The latter interpretation seems likely when one considers that the majority’s argument that the nonpoor are deterred from frivolous petitions by the costs that they (but not the poor) must bear and by the fact that the Court denied the waiver of fees to only two of a great many frivolous petitioners and only after the two filed a great many frivolous petitions. Zatco had filed forty-­three frivolous petitions in the prior two years. Given how much Americans fret about frivolous litigation and welfare cheats, it is likely they would regard Zatco as having abused the judicial process and therefore as undeserving of a fee waiver, all of which is consistent with an overriding concern about equal access to justice for those who do not abuse the process. Even if the majority did the balancing incorrectly, and the decision should have come out the other way, Justice Stevens is unconvincing about the message actually conveyed. We might generalize the weakness of Justice Stevens’ expressive claim in Zatco by saying he is using the “egoistic” perspective mentioned in Chap­ ­ter 1. Sometimes a commentator disapproves of a legal enforcement decision for reasons other than its meaning but creates an interpretation that puts the decision in the most negative light. But even a brilliant polemic is a poor basis for identifying audience meaning. Unless the audience possesses the same creative imagination as the polemicist, it will not likely reach the same interpretation. Of course, there are multiple audiences, so 181

The Expressive Powers of Law

one of them might receive the message the polemicist identifies. But then we should be honest about how limited that audience is, rather than assert that it is the message actually being sent. The condition of a “clear audience meaning” also counsels against reliance on esoteric knowledge beyond the ken of the audience. Sometimes the “best” interpretation will not occur to many people in the audience. Here we have a contrast between social science and humanistic methods of interpretation. To illustrate the difference, consider a passage from Jeannie Suk’s book At Home in the Law.26 Suk is immune to the criticism I am making because she is not making a causal claim. For her interpretive project, the use of esoteric knowledge is appropriate. Nonetheless, I will use her work to show why the interpretive method appropriate to one context is not appropriate for another. At one point, Suk interprets Justice Souter’s majority opinion in Georgia v. Randolph,27 where the Court held that the police lack consent to search a house when one adult in the doorway grants permission and the other refuses it. Justice Souter wrote that “a caller standing at the door of shared premises would have no confidence that one occupant’s invitation was a sufficiently good reason to enter when a fellow tenant stood there saying ‘stay out.’   ”28 The word “caller” here might be a mundane category for one who attempts to communicate with residents of a dwelling by coming to an exterior door and announcing oneself (by knocking, ringing the bell, shouting, etc.). But Suk identifies a more interesting meaning by alluding to historic practice: “[T]he effect of the anachronistic term ‘caller’ tends to evoke the social context in which that term was regularly used . . . ​[and] brings to mind the social world of the Gilded Age when the norms of calling upon others in their homes was codified in shared rules of etiquette . . . . ”29 Suk describes those rules, citing a nineteenth century book on etiquette, and concludes: “To invoke this vanished world was to introduce by suggestion a rhetorical figure . . . ​[of] a lady[] in the nineteenth-­ century . . . ​with high-­bourgeois status . . . ​[who was] ‘at home,’ determin­­ing whether to receive or decline visitors—​especially those gentlemen whom the word ‘caller’ even today conjures.”30 The passage usefully illustrates how much audience-­meaning differs from sentence meaning and speaker’s meaning. Most Americans, even most of those who would ever have read a Souter opinion, have never read a nineteenth century etiquette manual and are unaware of the “calling” customs of the high-­bourgeois of the period. Even those members of the 182

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

public who would recognize the term “gentleman caller” from A Streetcar Named Desire are unlikely to associate Souter’s bland use of the term “caller” with this old Southern tradition.31 Thus, when Suk says that Souter’s use of the term caller “brings to mind” and “conjures” the “social world of the Guilded Age,” she means that it brought this image to her mind and perhaps to the minds of others who possess this historic knowledge. She is making a claim about author’s meaning, sentence meaning, or a specialized form of audience meaning limited to people who possess the necessary esoteric knowledge. When people make claims of expressive effects on behavior (unlike Suk), they should attend to these distinctions and not project an interpretation onto the general public unless the public knows the facts necessary to the interpretation. I confess that an exploration of audience meaning—​a matter of social science—​is frequently less stimulating than more literary or humanistic modes of interpretation. Yet the more novel and creative the interpretation is, the less likely it is to predict how the public will see it. Now let us turn to the second condition for an enforcement action to convey information (about something other than deterrence variables): publicity. If there is no publicity, there is no expressive effect. The point may seem obvious, but it is apparently easy to overlook. In J.E.B. v. Alabama, for example, the majority holds that the Equal Protection Clause forbids the state from using peremptory challenges on the basis of sex. In the paternity trial, the state used all but one of its peremptories to strike men from the jury.32 One of the arguments Justice Blackmun advanced for the holding was expressive: that state actors who strike potential jurors on the basis of gender “ratify and reinforce prejudicial views of the relative abilities of men and women . . . . The message it sends to all those in the courtroom, and all those who may later learn of the discriminatory act, is that certain individuals, for no other reason than gender, are presumed unqualified by state actors to decide important questions upon which reasonable people could disagree.”33 Note first that the claim is apparently about audience meaning because a message cannot “reinforce[]” beliefs unless the people potentially holding those beliefs—​the audience to the jury selection—interpret the selection as supporting them. Whatever the ultimate merits of J.E.B., the expressive claim, so understood, is implausible. Part of the reason is the weakness of the condition already discussed—​a clear audience-­meaning implication. A lawyer who strikes men from the jury need not imply that men are 183

The Expressive Powers of Law

g­ enerally “unqualified” for jury duty, or have less ability than women. The lawyer may instead believe that women will, on average, be more sympathetic to her side and, because she knows very little about each potential juror, decides to play the averages. The lawyer might even think that she is striking the jurors who are impartial (and therefore superior from society’s perspective) in favor of jurors who are biased in the preferred direction (and therefore societally inferior). Blackmun would respond that any stereotyping by gender is pernicious, including the generalization that women are more sympathetic to women in paternity suits, and that the challenges reinforce the stereotypes. Yet that argument still concedes that the audience meaning of such strikes against men is not that men are “unqualified” or have less “relative abilit[y].” The more general weakness of the expressive claim in J.E.B. is that there is usually no publicity to the peremptory challenge. Justice Blackmun carefully claims only that the message will be sent to “those in the courtroom and all those who may later learn of the discriminatory act.” But there may be no one outside or even inside the courtroom who ever learns the true basis of a lawyer’s peremptory challenge. Many people do not know what a peremptory challenge is, so they cannot know that a lawyer is exercising it on any particular basis. Even for informed observers, judges may hold the lawyers’ argument for challenges in chambers and return to the courtroom to excuse a group of potential jurors, some who were successfully challenged for cause and others who were subject to peremptory challenge. Because judges frequently don’t state the basis of the dismissal in the courtroom, it is hard for observers to know who was struck by peremptory challenge, much less to know if the cause was the potential juror’s sex. Indeed, even before the judge, the lawyer exercising peremptory challenges does not state the reason for it and therefore does not admit that the reason was the individual’s sex. One might just look at the outcome of jury selection and infer the existence of sex-­based challenges when the jury’s composition is highly skewed in favor of one sex. But there are other inferences one might draw from such a disproportionate outcome: that the venire that day was randomly skewed by sex or that sex happened to correlate that day with non-­sex-­ based reasons for peremptory challenges, such as age, education, class, or occupation. And if the public will inevitably interpret the existence of juries skewed by gender as evidence that peremptory challenges are exercised on the basis of gender stereotypes, that is not a problem the holding 184

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

in J.E.B. solves, since it permits observable gender skewing, just not provably intentional gender skewing.34 Notwithstanding these points, it remains possible that information about the sex-­based exercise of peremptory challenges might leak out and “send the message” that some sex-­based stereotypes are true. But Justice Blackmun doesn’t give any attention to the mechanisms of publicity that determine the plausibility and strength of the expressive effect. And any fair account would recognize that the expressive claim in J.E.B. is quite weak compared to the expressive claims available in other Equal Protection sex-­discrimination cases. Consider, for example, the publicity for a sexist message where the state legislature funds schools that train only one sex for a profession already associated with that sex. Before the Supreme Court ruled the gender discrimination unconstitutional in Mississippi University for Women v. Hogan and United States v. Virginia, Mississippi operated an all-­female nursing school and the Virginia Military Institute (VMI), also public, refused to admit women.35 In the former case the Court said that the nursing school’s “policy of excluding males . . . ​tends to perpetuate the stereotyped view of nursing as an exclusively woman’s job.” The expressive claim here is highly plausible, almost inevitable, as is a similar claim for VMI. Neither institution concealed its sex-­discrimination, thus most adults in either state would have been aware of the policy. If we are to take expressive causal claims seriously, we must pay attention to the vast differences in publicity in these different cases: the expressive whisper in J.E.B. and the deafening shouts in Hogan and Virginia. Considering each of the first two conditions for informational effects—​a clear audience meaning and publicity—​the Supreme Court has been somewhat more careful in Establishment Clause cases (whether or not one agrees with the case outcomes, which I discuss in Chapter 8). One doctrinal test the Court has used to decide whether symbolic government action “establishes” religion is to ask whether the government’s action “endorses” religion.36 Here, the Justices play close attention to the audience meaning of state action, such as the government granting access to religious groups on equal terms as nonreligious groups. Some of the cases examine the possibility that actual observers will not know that the religious expression on government property is actually private speech. For example, in a dissent, Justice Stevens distinguished between “attended” and “unattended” religious speech because it was more likely 185

The Expressive Powers of Law

that onlookers would mistakenly attribute the speech to the government when there was no private individual present.37 In Capital Square v. Pinette, a private group wanted to erect a Latin Cross in a public park, where nonreligious messages were permitted. Stevens notes that prior access cases involved giving access to private individuals, to whom onlookers could attribute the speech, while an unattended cross on government property is more likely to appear to be a government cross. Moreover, he distinguished the “less obtrusive” message that occurs when a religious group meets indoors from the “eye-­catching symbolism” involved in an outdoor display. Whether or not these considerations are persuasive in the specific case, they are the right factors to consider; these details determine whether government action satisfies the first two conditions for a signaling effect. The third condition for an expressive effect is some factor—​expertise or aggregation—​that makes the legal signal strong enough to stand out against the background stream of information on a topic. Consider how this third condition limits the expressive power of enforcement decisions. We are not likely to see much risk signaling through enforcement because prosecutors, judges, and juries usually lack expertise on risk and prosecutors and judges usually act alone and do not aggregate information across groups. For example, that federal DEA agents arrest individuals for medicinal use of marijuana and United States Attorneys prosecute them, is not likely to change citizen beliefs about the health costs and benefits of marijuana, given that the actors are not experts in the subject and do not systematically aggregate information. Police may reveal their enforcement priorities and commitments by the arrests they make, but not much else.38 Of course, juries do aggregate information, if only in a small group, so their verdicts might provide information on risk. But one should not overstate this possibility; the jury’s aggregation is likely to be weaker than the legislature’s. One reason is the obvious fact that the jury is quite a bit smaller than a typical legislature, so the verdict aggregates a smaller amount of information. One might counter that a verdict of guilty must be (in almost all states) unanimous, which seems to make for a stronger signal. But another point complicates that idea. In the last chapter, I noted that legislative votes have potential costs; if a legislator’s vote on a bill later seems “erroneous” given new information, it may damage the legislator’s re-­election chances. So the legislator may seek to get the vote “right” in the first place. Yet a juror incurs no such potential cost in rendering a verdict; one does not lose the 186

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

“job” of being a juror by voting for a verdict that later turns out to be erroneous (and one may not want the job anyway). Thus, the legislator and the juror may have internal reasons—​a sense of civic duty—​for doing their job well, but only the legislator has the external incentive of running for re-­ election. In this context, the unanimity requirement might actually prod unmotivated jurors to just “go along” with others and vote against what they perceive to be the correct outcome in order to reach consensus. That means the votes are likely to be correlated with each other, making more likely what we called “herding.” Thus, despite the name of the Condorcet Jury Theorem, the risk signaling story might apply better to legislatures. Still, one might think there is some aggregation across criminal enforcers in those jurisdictions that require a grand jury indictment before the prosecutor can proceed.39 An adage says that a grand jury will indict a ham sandwich if the prosecutor asks, but that is probably due in part to the fact that elected prosecutors and grand juries often agree about what conduct deserves prosecution. A conviction therefore aggregates the views of the grand jury, prosecutor, and petit jury. In the rape example discussed above, a jury verdict of guilt might send a message about certain facts, such as the serious harm inflicted by the crime, including the psychological trauma of losing one’s sexual autonomy. The “message” makes it more difficult to think of the event as a minor matter, a frat-­boy prank. An acquittal may send the opposite message. But given the lack of expertise and compromised aggregation, the signal is weak and might not do much to change people’s judgments about those facts. Some Cases Where Enforcement Decisions Plausibly Express Attitudes There is more room, however, for prosecutors, judges, and juries to succeed at attitudinal signaling, though again, not as well as legislatures. Some judges and almost all local prosecutors are elected, which makes it plausible that they have expertise in local attitudes.40 Their electoral responsiveness is tempered, however, by professional norms and the fact that judges and prosecutors often act out of the media spotlight. Still, prosecutors value success in jury trials and public opinion generally correlates with how a jury will decide a case. Juries are not composed of attitude experts, but the jury itself is a small sample of the community whose attitudes matter. Prosecutors attempt to use peremptory challenges to strike the most pro-­defense individuals from the jury, and the defense attorney similarly attempts to strike the 187

The Expressive Powers of Law

most pro-­prosecution, so the jury might be roughly representative of the public (though obviously skewed in the way the venire is chosen, as those who do not vote or drive are underrepresented). In addition, the jury is commonly described as the “voice” of the public; its verdict is said to “speak for” the people, so its judgment may be taken as evidence of what the community sentiment is.41 Even if this view is romanticized, the jury will have the power to signal public attitudes if many people believe the romantic view. If so, then perhaps the Second Circuit was correct when it declared: “The public listens with rapt attention to the jury’s pronouncement of guilt or innocence, for in that singular moment the convictions and conscience of the entire community are expressed.”42 The Second Circuit is speaking of the petit jury, but we might say the same about the grand jury, which in most jurisdictions must indict for the case to ­proceed. Thus, if the other conditions hold, an attitudinal signaling claim is entirely plausible for the enforcement actions of elected officials and juries. A conviction reveals significant information about public attitudes when it combines the grand jury’s decision to indict, the elected prosecutor’s decision to charge, and the petit jury’s decision to convict. In the rape example, it is therefore plausible that a conviction effectively signals that public attitudes are hostile towards and outraged by the defendant’s behavior; that whatever the case in the past, the current public regards nonconsensual intercourse as grave wrongdoing even when it occurs between social acquaintances in these circumstances. If public attitudes have shifted on these particular facts, they have probably shifted more generally on the importance of a woman’s sexual autonomy, so that an individual would expect more social disapproval for a wider array of behaviors that are not respectful of that autonomy. And if so, then it is not only true that the conviction sends a message beyond the likelihood of formal sanctions, but it is also true that an acquittal could send a message that public attitudes have not shifted as much as people had thought. That is one reason that groups protest disappointing verdicts—​to present evidence of attitudes contrary to those signaled by the jury verdict. I end this section with two more examples where enforcement (or non-­ enforcement) plausibly sends a message. Consider first an expressive analysis of the now-­famous prosecution and acquittal of George Zimmerman for the murder of Trayvon Martin. Absent expressive considerations, there is something quite puzzling about the nature of the media focus on this 188

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

case, and much of the public reaction, which have pervasively characterized Florida’s “stand your ground” (SYG) law as central to understanding the killing and the legal outcome.43 If one looks at the relevant legal doctrine, it is not at all clear that this aspect of Florida law had anything to do with the outcome of the case.44 First, the SYG law creates a presumption in favor of the right to use deadly defensive force in one’s dwelling or vehicle, but that obviously had no application to this case because the killing occurred outside.45 Second, the law creates serious constraints on civil claims for wrongful death, but they had no effect on the criminal case.46 Third, the law creates a right to have the judge hear the self-­defense claim and dismiss a criminal charge before trial, but Zimmerman waived that right.47 What remains is the eponymous section that alters the substantive standard of self-­defense: “A person who is not engaged in an unlawful activity and who is attacked in any other place where he or she has a right to be has no duty to retreat and has the right to stand his or her ground and meet force with force, including deadly force if he or she reasonably believes it is necessary to do so to prevent death or great bodily harm . . . . ”48 There are two reasons that the media and public emphasis on this language is puzzling. One is that the words “stand his or her ground” appear to add nothing to the operative phrase “no duty to retreat.” One can stand one’s ground only because one has no duty to retreat.49 Yet the duty to retreat was probably never the majority American rule in the twentieth century, even though it was the English common law rule and even though many states followed it.50 Florida had been one of states imposing a duty to retreat before it adopted the SYG statute and put itself in line with what is certainly now the majority of states. Yet from the media coverage and public reaction, one would think that Florida was an extreme outlier in the law on this point. More puzzling still, the absence of a duty to retreat and the right to stand one’s ground had little or nothing to do with the Zimmerman acquittal. Zimmerman’s story is that he shot Martin while he—​ Zimmerman—​was on the ground and Martin was on top of him, beating him.51 The jury is entitled to believe or disbelieve this story. The jury apparently believed Zimmerman or did not find the story false beyond a reasonable doubt. If so, then Zimmerman was physically unable to retreat at the time when he first reasonably believed that shooting Martin was necessary to prevent great bodily harm. In jurisdictions with a duty to 189

The Expressive Powers of Law

retreat, there is no breach of the duty if the defendant lacked the ability to retreat (usually, it is added, to retreat with safety).52 What if the jury had disbelieved Zimmerman? If the jury had believed him to be lying about some facts of the incident, such as the claim that he fired while on his back, then they could have easily decided he was concealing other facts that entailed the basic failure of self-­defense. Perhaps he had started the attack or didn’t fear serious bodily harm; perhaps he could have ended the fight by brandishing but not firing his weapon. If he was lying, the incident might have unfolded in any number of ways that justified conviction, despite his right to “stand his ground.” It was not impos­ ­sible for the jury to have thought that Zimmerman was lying solely for the purpose of concealing that he had an opportunity to retreat, but that doesn’t seem particularly likely given that the law would then have given him a defense. Confusion arises, I think, because some people assume that the duty to retreat would have forbidden Zimmerman from following or approaching Martin. Yet that is almost certainly false. In jurisdictions where there is a duty to retreat, it arises at the time one wants to use otherwise justified deadly force.53 The duty to retreat is not a duty to avoid, as in a duty to stay away from strangers, but a duty to leave their presence once it appears that there is an imminent attack that can only be avoided by force or retreat. Another confusion is the separate issue of “provocation” in this context. One can lose the right to use deadly defensive force if one “provoked” the attack, which raises difficult questions that also vary from state to state.54 But the famous part of the Florida statute doesn’t specify that doctrine, so if the Florida provocation rule is too narrow, it is not because of the SYG rule.55 In sum, as a matter of substantive law, whatever the trial was about, it was not about the lost “duty to retreat” or the new right to sand your ground.56 Yet the matter is puzzling only because I have been ignoring its expressive dimension, which is central. Indeed, my reliance on esoteric legal knowledge is wholly irrelevant to understanding the law’s expressive effects, which depend entirely on audience meaning, unconstrained by expert knowledge of law. The popular focus on the SYG law makes sense when considering the law’s power to signal attitudes to an audience.57 First, there is a clear audience meaning. With the strong words “the right to stand his or her ground and meet force with force,” the statute offers a ringing endorsement of a person’s choosing the option of killing 190

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

over the option of retreating: not only is there no duty to retreat, no limitation on the self-­defense defense against a charge of murder, but also there is an affirmative right to meet force with force. In borderline cases of self-­ defense, the generally commendable notion of “standing up for your rights” now means killing rather than retreating. The legislature reinforced this meaning in the preamble to the SYG bill, stating: “[T]he Legislature finds that it is proper for law-­abiding people to protect themselves, their families, and others from intruders and attackers . . . ​and . . . ​[that] no person or victim of crime should . . . ​be required to needlessly retreat in the face of intrusion or attack.”58 By rejecting the duty to retreat that previously existed, the statutory words imply a recent state-­wide change in attitudes towards greater approval of defensive killings, even when nondeadly alternatives exist, and towards greater disapproval of retreat. One might also expect the public to discern in the statute a statement about honor, even though it does not use the term. When some American courts first rejected the English common law rule of retreat, they did so by emphasizing a distinct American masculinity: the “true man,” who would not flee from a fight.59 As Dan Kahan has noted, the honor norms underlying this notion of masculinity also supported the dueling and lynching culture of the aristocratic South.60 One lacking this particular historical knowledge might still easily grasp from the statute the idea that it is cowardly and dishonorable to “cut and run” from a fight, but honorable to stand one’s ground. The other conditions for an expressive influence are satisfied. Media attention to the passage of the new law and then to various killings, prosecutions, and acquittals, most prominently the Zimmerman trial, gave intense publicity to these legal expressions. The legislature has expertise over public attitudes and jury verdicts of acquittal may be thought to aggregate evidence of those attitudes. In contrast, compare Florida to any of the many states that never adopted the duty to retreat and thus have always had the same no-­retreat rule as a matter of judicial precedent that Florida now has a matter of statute. The Florida no-­retreat rule may be substantively identical, but is expressively different. First, a new law has more expressive significance than an old law. Because of legislative inertia, old statutes may remain on the books despite the fact that public attitudes no longer support it. So an older law is weaker evidence of today’s attitudes than a newer law. Second, a statute is more significant than a judicial precedent, because people expect the 191

The Expressive Powers of Law

legislature to be more expert in and concerned about public opinion. Third, most people didn’t know how uncommon the duty to retreat was in the United States, so it appeared to be a very big change in the law, signaling a major change in attitudes. The result is an expressive “externality.” When the law permits the use of force, there is compliance whether or not one uses force. But the signal that public attitudes favor killing over retreating might lower the perceived public disapproval from borderline defensive killings and raise the perceived disapproval of retreating. Either way, we might get a greater willingness to use deadly defensive force in cases where there is an alternative. So it is understandable to react to Zimmerman’s shooting of Martin by wondering, not only whether Zimmerman was guilty of murder, but also whether the law’s expression, in this case or others, contributed to an avoidable shooting. The concern is not the formal definition of the duty to retreat, but the law’s exuberant endorsement of defensive deadly force. One could have changed the law and eliminated the duty to retreat without creating this expressive effect, merely by avoiding the language for which the law became known, “stand your ground.” The normative implication is that we should attend to such expressive externalities, but it is of course possible that the groups pushing for SYG laws affirmatively desired this expressive effect, to encourage greater use of arguably defensive deadly force. As a final example of enforcement signaling, consider Kenworthey Bilz’s theory of “delegated revenge.” She claims that legal expression condemning the criminal makes crime victims less likely to seek private vengeance.61 The key to this effect is that punishment not only expressively condemns the perpetrator but also implicitly expresses the victim’s worth. As Bilz explains, there is substantial psychological evidence that victims and observers perceive a crime as lowering the social status of the victim, while authoritatively imposed punishment for that crime restores it. By contrast, the victim’s private vengeance does not express the public’s disapproval of the crime nor the public’s recognition of the victim’s worth. As Bilz notes, economics otherwise has a difficulty explaining why crime victims are usually willing and even eager to delegate to the state the job of punishing the criminal. If the victim views efforts to punish as a cost, she should want to free-­ride on the punishment efforts of others and not even bother to cooperate with the state in pursuing the perpetrator 192

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

(unless she thought the probability of being victimized again by the same person made the benefits of cooperating exceed the costs). If the victim views efforts to punish as a benefit, satisfying an intrinsic preference for revenge, then she should want to carry out the punishment herself, perhaps with the assistance of family and friends, whether or not the state also punished (though obviously not if the criminal is unknown or too powerful). Yet we don’t observe that victims are indifferent to state punishment nor engage in self-­help as a default. Instead, victims usually refrain from self-­help and demand that the state enforcement apparatus find and punish their perpetrators. Bilz discusses the point in terms of legitimacy—​that only politically legitimate authorities can expressly restore the victim’s social standing. But one can also explain the claim using the information theory. The victim’s private vengeance, even if publicized, does not reveal information about public attitudes of concern for the victim’s suffering, nor of outrage against the perpetrator. But a public conviction and punishment plausibly suggest that the grand jury, prosecutor, judge, and absent a guilty plea, the jury, were motivated enough by the victim’s social standing to put forth the effort to punish the perpetrator. One might object that those actors are motivated only to achieve future deterrence of crime or incapacitation of this criminal. Yet police and prosecutors are usually overwhelmed by the amount of crime and must choose what cases to pursue. So for many crimes their action tends to show particular solicitude for the victim and/ or particular indignation at the perpetrator. Because the decisions of juries and elected prosecutors usually correspond to public attitudes, the implication of punishment is that the victim “counts” in this community, which is collectively outraged by the perpetrator’s wrong. Thus, the attitudinal signal of state punishment offers the victim something that private punishment lacks.

Executive and Judicial Signaling in Civil Cases This chapter’s focus on criminal cases should not obscure the fact that legal enforcement of noncriminal law can also signal attitudes and risk. Here, I offer a few examples. First, take a simple governmental lawsuit against a business. Suppose the media gives substantial coverage to a state attorney general’s decision to sue a payday lender over excessive interest rates or inadequate 193

The Expressive Powers of Law

­ isclosure. The public plausibly views the Attorney General (AG) as a d legal expert. We would then expect that the lawsuit would cause people to believe that the defendants were more likely than other lenders to be breaking the law, which makes it appear to be riskier to borrow from them.62 To isolate the expressive effect, let us assume that the usury law the state AG seeks to enforce has entirely nondeterring remedies. Thus, the lenders do not fear the lawsuit because of the legal sanctions it may produce. But they may fear the lawsuit because of its reputational consequences. The suit signals the AG’s beliefs about the company’s shady practices, which may drive away potential customers. Thus, the lawsuit has expressive consequences. The lawsuit’s expressive consequences depend on media coverage. In theory, the consequence of the AG’s lawsuit might seem to be the equivalent of the AG holding a press conference or issuing a press release accusing the payday lender of some unlawful practices. But the lawsuit is likely to be a stronger signal because it is costlier than the press release. Given a budget for litigation, there may be no reason not to spend the money, but the lawsuit represents an opportunity cost of the alternative cases the AG does not bring. The press release may crowd out other press releases, but press releases are cheaper than lawsuits so the opportunity cost is lower. For this reason or others, the media is more likely to cover the press release about a lawsuit than a press release about legal claims not backed by a lawsuit. In the latter case, the absence of action creates doubt whether the AG actually believes the claim he is making. After a trial, a publicized judicial decision against the defendant may be additionally informative. Even if the available sanctions are themselves nondeterring, the court’s judgment of liability further diminishes the defendant’s reputation beyond the AG’s decision to initiate suit—​more potential borrowers believe it is risky to deal with the defendant firm. What is true of mundane commercial litigation is more obviously true when government litigates over a “hot button” social issue, like affirmative action, abortion, LGBT rights, gun rights, or religion, though the information conveyed is more likely about attitudes than risk. If the public believes the actions of the elected state AG are correlated with public opinion, his litigation position signals what the majority, or at least a highly motivated minority, want. If the public believes the judgments of its judiciary are correlated with public opinion, we get the same result. Of course, 194

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

if the court rules against the AG’s position, the two signals cut against each other and there may be no net effect. But where the courts rule in favor of the AG, the signals reinforce one another. Consider the lawsuit Oren Adar and Mickey Ray Smith brought in 2007 in Louisiana, seeking to force the Louisiana State Registrar to issue a new birth certificate for their infant adopted under New York law.63 As generally permitted under Louisiana law, they sought to have themselves listed as the child’s parents and sued when the Registrar refused. The federal district court granted their requested injunction, and the Louisiana AG’s office appealed to the Fifth Circuit, which affirmed. As is often the case, the court did not deal directly with the hot-­button issue but technical matters, like the meaning of the full faith and credit clause of the federal constitution. But everyone understood that the overwhelming majority of those opposed to adoption by same-­sex couples supported the Registrar and that the overwhelming majority of those who supported same-­sex adoption opposed the Registrar. The state AG’s decision to litigate the issue, even asking for a rehearing en banc,64 signaled which side the AG believed was more politically powerful. The Fifth Circuit’s decision probably did nothing to blunt the signal of local attitudes from the Louisiana enforcers (the Registrar and AG), though the decision is obviously important for its substantive effect, and perhaps for the expressive effect of making Louisianans realize that other Americans were recognizing the legal parenthood of same-­sex couples. As a final example, consider an executive order. Many states have “English-­only” statutes.65 President Bill Clinton and NYC Mayor Michael Bloomberg issued executive orders establishing that government agencies shall make their services available in multiple languages.66 Such an order may signal public attitudes of tolerance toward non-­English speakers. This is not very plausible in the Clinton example because he issued the order at the end of his second term, when his costs for flouting public opinion were minimal, but quite plausible in Bloomberg’s case where he could expect to suffer politically if he misread public opinion. Inferring that fellow New Yorkers, even English speakers, agree with Bloomberg, city bureaucrats are more likely to expect public disapproval for failing to offer multilingual access, which creates a stronger incentive for bureaucrats to comply with the order quickly. The “spillover” is that those not subject to the order—​private employers, for example—​might see greater benefit to making their institutions accessible to non-­English speakers. 195

The Expressive Powers of Law

The Interaction of the Information and Focal Point Theories This book does not present an expressive effects theory of law. It presents two expressive effects theories of law. Having now described the information theory, it is appropriate to reflect on the connections between that theory and the focal point theory presented in earlier chapters. First, there are contexts in which only one of these expressive effects theories is plausible. Start with the focal point theory. When I discussed law as a coordinating focal point, quite a few of my examples offered no plausible informational effect. A new state might have no reason to favor driving on the left or driving on the right; the underlying attitudes and risks might be in equipoise. So the choice of driving one way or the other does not signal the lawmakers’ beliefs about risks or attitudes. But it still works as a coordinating focal point, affecting expectations about how the other drivers will behave. Similarly, some parts of constitutional law merely make a particular choice where most people would think any of a variety of choices would serve equally well. Setting the length of a Senatorial term at six years doesn’t expressively foster compliance by revealing information; it works by creating a coordinating focal point, after which everyone expects Senators to insist on at least that amount of time and everyone else to insist on no more. Even when the choice of rule matters, it may expressively influence behavior only as a focal point, not as information. Suppose we put a “yield” or “one-­way” sign on a street rather than another because, by relieving traffic congestion, it is efficient and fair. These signs or other priority rules might therefore reveal information about what behavior fairly minimizes the social costs of driving. But these updated beliefs do not give the driver a selfish reason to comply. What matters is the fact that the signs or priority rules create a focal point for the other drivers that, if ignored, will cause a collision. The same is probably true of the way property law allocates power between the “owner” and everyone else. There is probably informational content to the rules—​a nuisance rule might be efficient or fair. But the expressive reason for people to defer to it, to the extent they do, is that it provides a coordinating focal point for resolving their dispute without a mutually destructive fight. International treaties usually work the same way. There are also situations where the information theory is plausible but the focal point theory is not. Recall from Chapter 4 the broad discussion 196

REVE L A T ION OF INFOR M A T ION B Y L EGA L ENFOR C E M EN T

of the Assurance Game. It would be possible to describe almost any legal compliance as an equilibrium in an Assurance Game where one repays the legal obedience of other citizens or enforcers with one’s own obedience. But I said in Chapter 4, this analysis is merely a translation of legitimacy into the terms of game theory. If we set aside legitimacy and focus only on the narrower focal point theory (everything that came before this section in Chapter 4), we will see that it does not explain compliance where noncompliance of one person does not risk a roughly contemporaneous private sanction from another (as traffic noncompliance risks an accident or smoking noncompliance risks an altercation). Take rules mandating the use of safety devices. When you fail to wear a seatbelt or motorcycle helmet, or fail to put your child in a safety seat, other drivers have no selfish incentive to insist on your “buckling up,” wearing a helmet, or securing your child in a safety seat. So information is the best explanation for any expressive effects of these laws. The same is true when the law bans the use of a cell phone while driving—​other drivers are more likely to avoid than to confront such a person, especially given the difficulties of communicating with a distracted driver. Or consider the repeal of anti-­sodomy laws or other statutes already declared unconstitutional. The only plausible effects here are informational, that the new law signals risk or attitudes. There are also situations in which both theories apply and reinforce each other. I have already extensively discussed smoking laws. Public bans signal the lawmakers’ beliefs about anti-­smoking attitudes and/or second-­ hand health risks. At the same time, smoking bans create a coordinating focal point in which nonsmokers insist on getting their way and smokers defer. Given this overlap, perhaps it is not a coincidence that, while many laws generate modest or low compliance, these laws generate substantial compliance in many countries, notwithstanding the meager use of sanctions. The same might be true of, say, a new mandate that mothers be allowed to breastfeed their infants, which conveys public approval and healthiness of breastfeeding and “waves on” an individual mother to insist on her rights. A new mandate that owners clean up after their dogs conveys both disapproval of (previously socially acceptable) behavior and “waves on” others to insist that their neighbor control his dog in the ways the law specifies. Finally, there are situations where neither theory applies. Take the crime of theft. In the normal situation, where there is no ambiguity, theft statutes 197

The Expressive Powers of Law

convey no new information about attitudes or risk. People grow up knowing that their society disapproves of simple theft. Moreover, the risks the law reveals are not the ones suffered by the thief (except for sanctions, but that is a deterrence story, not an expressive one). At the same time, the law cannot work here as a coordinating focal point. Where theft occurs, the thief is typically not worried that his victim will “insist” on his property rights nor that mutual insistence will produce a fight. Instead, the thief plans either to use stealth to avoid the fight or to initiate the use of force to win it. The thief is not coordinating and does not need a focal point. Thus, despite two expressive theories that may produce legal compliance, the domain for expressive effects is limited. Sometimes, we require sanctions and/or legitimacy.

Conclusion A great many claims of legal symbolism and “message-­sending,” if they have any content, rely on the idea of information transmission. The prior chapter contended that legislatures could signal information about public attitudes and behavioral risks. This chapter argued more tentatively that the other branches of government could supply information, change beliefs, and influence behavior. The conditions are more restrictive but there are cases where the expressive claim is plausible. In particular, criminal punishment expresses the community’s condemnation of the criminal and recognition of the victim, which affects the behavior of both. As I said in Chapter 1, this book addresses both economists who ignore the nonsanction effects of law and legal commentators who assert expressive claims in a casual and sometimes limitless manner. This chapter addressed both groups. I extended the informational theory, but continued to pay attention to cases where expressive claims are implausible. I compared the informational theory and the focal point theory from earlier chapters. These theories can be overlapping and reinforcing, but because the domain of each theory is limited, there are contexts where only one theory applies and contexts in which neither theory applies. This analysis identifies when sanctions or legitimacy are strictly necessary for compliance because the expressive effects do not exist.

198

7 The Power of Arbitral Expression

P

rior chapters have primarily explored the prospective power of legal expression. The legislature, court, or executive official announces or enforces a legal rule and thereafter members of the public either use the rule to coordinate or update their beliefs (in light of the focal point or information the legal action, respectively, creates or reveals). Even in Chap­­ter 3, when I said that law supplies a focal point to resolve a potential “dispute” (broadly conceived), I described expressive effects where the expression precedes the potential dispute. The idea there was that the disputants look back to custom already articulated and clarified by law and the person disfavored by the expressed rule is more likely to concede to the person favored by it (if the parties share an interest in coordinating to avoid the costs of continued conflict). In this chapter, however, I explore how legal expression works retrospectively, where the expression occurs only after the dispute has arisen. When two individuals find themselves in a dispute, which the current legal rule fails to resolve, they may then seek a third-­party arbiter to “decide” their “case.” The arbiter (my general term for any third party consulted for this purpose including an arbitrator, mediator, judge, etc.) announces a result and this post-­dispute expression can influence the disputant’s behavior.1 My claim in this chapter is that an arbiter can resolve disputes between parties even without the threat of sanctions or the possession of legitimacy. Independent of sanctions and legitimacy, the arbiter can expressively influence the parties, causing the declared loser of the dispute to concede to the declared winner. This expressive power of dispute 199

The Expressive Powers of Law

r­ esolution arises from both of the mechanisms discussed in prior chapters: the arbiter’s ability (1) to create a coordinating focal point and (2) to reveal information. Start with coordination. The arbiter’s coordinating power is at least as great as the coordinating power of a coin flip that we sometimes see parties use to settle a dispute. A randomizing device works in a coordination situation in which the expected returns from a 50 percent chance of prevailing (and of losing) is higher than the expected returns from continuing the dispute, especially if the aggressive insistence of both parties produces the worst outcome for each. As I will explain, the randomizing device only works in these cases because it makes focal a particular resolution; with or without randomization, an arbiter can create the same focal effect. Now consider the information mechanism. Depending on her perceived ability and motivation, the arbiter’s expression may also reveal information. Where the parties’ divergent beliefs create the dispute, the arbiter’s statement can cause the disputants to update in the same direction, narrowing the gap in beliefs and potentially resolving the dispute. I explain this claim and also discuss a reputation mechanism, where the arbiter assists the parties in maintaining their reputation for strength by giving one of them a reason to capitulate other than being weak. Thus, adjudication has expressive power; an arbiter need not in every case wield the power of sanctions or legitimacy to generate compliance with her announced resolution. Before exploring these theoretical claims, however, this chapter begins with examples of successful adjudication or arbitration in the absence of sanctions. For the economist, there is a puzzle in the cases in which a third party resolves a conflict without the threat of sanctions.

The Puzzling Strength of Dispute Resolution Third-­party dispute resolution is ubiquitous. In his book on courts, Martin Shapiro explains: “Cutting quite across cultural lines, it appears that whenever two persons come into a conflict that they cannot themselves solve, one solution appealing to common sense is to call upon a third for assistance in achieving a resolution. So universal across both time and space is this simple social invention of triads that we can discover almost no society that fails to employ it.”2 Shapiro concludes that “the triad for purposes of conflict resolution is the basic social logic of courts, a logic so compelling that courts have become a universal political phenomenon.”3 200

T HE POWER OF AR B I T RA L EXPRESSION

The conventional economic account identifies sanctions as the source of judicial power: Courts can impose costs on parties that do not comply with its demands. But if sanctions underlie the “logic of the triad,” it is puzzling that third parties frequently resolve disputes without wielding sanctions. This section reviews many such examples. Note that, for the most part, we cannot count modern mediation and nonbinding arbitration as sanctionless. Though these arbiters threaten no sanctions, they may influence disputant behavior by revealing their estimate of the outcome of a subsequent trial, where that trial court would be able to enforce the outcome by sanctions. Bargaining in the shadow of the law is bargaining in the shadow of sanctions. Yet we have already seen some important examples where tribunals function without the power of sanctions, not even the indirect threat of some other judicial body imposing them. In Chapter 3, I discussed how constitutional and international law work without the threat of sanctions. At the time, I stressed the courts’ power to prospectively influence behavior, but now we can consider the same power retrospectively, that is, a court’s power to induce the party it rules against to back down and comply with its order. Thus, in the domestic context, we observe national legislatures and executives concede the legal invalidity of their actions in the face of a judicial ruling against them, even though the courts depend on the legislature for funding and the executive for enforcement. The strong defer to the weak. For example, President Nixon famously backed down when the Supreme Court ordered him to release audiotapes from Oval Office conversations.4 President Bush abandoned plans for military tribunals when the Court declared them invalid, and Presidential candidate Al Gore, the winner of the national popular vote in 2000, backed down when the Supreme Court, in Bush v. Gore, halted the Florida recount on which he pinned his hopes for winning the Electoral College.5 Congress routinely backs down when the Court declares its statutes unconstitutional. Similarly, in the international context, we see that nations frequently comply with the orders of international courts that cannot credibly threaten to enforce their orders with sanctions. As previously noted, Tom Ginsburg and I reviewed all the decisions by the ICJ, and we conservatively estimated a compliance rate of 68 percent.6 Other scholars find high compliance with other international tribunals.7 For a realist, it seems puzzling that states would concede to so weak a force as a body of lawyers with no military. 201

The Expressive Powers of Law

Perhaps the most famous historic example of sanctionless adjudication is medieval Iceland. William Miller observed that for about three centuries Iceland’s legal system thrived “in the absence of any coercive state institutions,” that is, with “no state apparatus to pretend to monopolize the legitimate use of force.”8 There was a court, but “there was no sheriff to issue a summons to a hostile party, to keep the peace in the court, or to execute the judgment.” Instead, “[i]t was up to the litigants to serve process on their opponents, maintain order in court, and enforce court judgments in their favor.”9 Ultimately, the sanction behind legal judgment and arbitrated settlement was self-­help, most often appearing in the guise of the bloodfeud.10 Courts did declare a sanction, ranging from a simple fine to “full outlawry (skóggangr)—​which meant a loss of all juridical status and property, privileging anyone to kill the outlaw and indeed obliging the prosecutor to do so.”11 But the immunity granted to the killer of an outlaw was merely the right to be free from adverse judgments issued by a court that had no power to enforce its judgments. Yet “there must have been more to getting an outlawry judgment than simply being put back into the same problem one had before going to law: killing the enemy.”12 Thus, the puzzle: “[W]hy people made as much use of the law as they did,” given that courts lacked the power of sanctions? At the end of the chapter, I return to this question.13 There are many such historic examples. In the seminal Governing the Commons, Elinor Ostrom identified long-­enduring common pool resource (CPR) regimes, nongovernmental organizations that successfully prevent the “tragedy of the commons” for common resources like water supplies, fisheries, and pastures. Ostrom noted that one of the successful design principles is “rapid access to low-­cost local areas to resolve conflicts among appropriators or between appropriators and officials.”14 All of the six “robust” CPR regimes she discussed had this feature, while three of the five failed regimes did not.15 Ostrom concluded that “it is difficult to imagine how any complex system of rules could be maintained over time without such mechanisms,”16 yet a “quite informal” mechanism is sufficient. She did not separately discuss how these informal arbiters worked, given the lack of centralized enforcement, leaving the same puzzle as the other examples. Perhaps Ostrom’s CPR case studies are not surprising because the long­lived regimes might contain an unusual set of cooperative individuals. But 202

T HE POWER OF AR B I T RA L EXPRESSION

consider disputes between individuals who are known to be selfish, indeed for being unusual in their willingness to sacrifice social values for the sake of personal fortune. Andrea McDowell describes successful dispute resolution between mostly armed men during the nineteenth century Californian gold rush, where early courts had only a weak or nonexistent threat of legal sanctions.17 Peter Leeson observes that eighteenth century pirates resolved disputes among themselves by consulting with their quartermaster.18 The quartermaster did not use his power as a ship officer to coerce the disputants to accept his proposed resolution; each party retained and sometimes used the power to force the dispute into a duel with the other party. Similar to medieval Iceland, the puzzle is why the mere suggestion could sometimes bring about a peaceful resolution otherwise beyond the disputants’ reach. These examples show that third-­party arbiters can get one side to back down, to concede to the arbiter’s declared outcome, without the threat of sanctions. Yet I must admit that legitimacy remains a potential explanation for this success. It is possible that eighteenth century pirates, nineteenth century gold rushers, and twentieth century nation states, despite being fiercely selfish and even (at times) ruthless, defer to sanctionless dispute resolution because, in each case, they perceive the arbiter as a legitimate authority. The modest goal of this chapter is merely to demonstrate the plausibility of a rational choice alternative to legitimacy theory, the mechanisms by which arbitral expression changes the incentives facing selfish disputants. I leave for the future the empirical job of testing the expressive theory against the legitimacy theory. Note, however, two reasons for giving particular attention to the rational choice approach in this context. First, in other ways, rational self-­interest is obviously important to the behavior of the disputants in these cases. Those leaving their home and family to mine for California gold were, quite obviously, highly motivated by material rewards, the desire to “strike it rich.” Pirates sought gold as well, using violent means that repudiated the most basic and widely shared laws of human societies (against murder and theft) that would seem to be among the most legitimate. History is full of the self-­interested behavior of national leaders who seek material gains through military adventurism despite a horrific toll of military and civilian casualties. In each case, there is obviously a lot to be said for explaining these activities as the selfish and rational pursuit of material gains. If so, there is something to be said for using the same framework to explain the 203

The Expressive Powers of Law

willingness of these selfish actors to defer to arbiters who lacked the power of sanctions, to constrain their selfish pursuit of material gains. That is what this chapter does. Second, as I said in Chapter 4, legitimacy seems most plausible after an institution has functioned successfully for a time, building up credibility, because wholly ineffective institutions are not likely to be perceived as legitimate. Thus, there is a need to explain how an arbitral institution that now enjoys the power of legitimacy initially succeeded at generating compliance. One possibility is sanctions, if an arbitral institution is initially able to impose costs on those who failed to comply with its orders. But the puzzle for legitimacy theories is how institutions generated compliance at the beginning when they could not impose sanctions and had not yet earned legitimacy. Here, I argue that the expressive power may create the first increment—​the origin—​of compliance that permits the institution to gain credibility and, in the long run, legitimacy. In this sense, the expressive theory is a complement to the legitimacy theory, rather than a rival. The power of arbitral expression arises from the two mechanisms discussed in prior chapters: The arbiter’s declaration provides a coordinating focal point (Chapters 2, 3, and 4) and information (Chapters 5 and 6). Arbitral expression gains power from each and, indeed, I propose that there are important synergies between the two mechanisms. To isolate this expressive power from the sanctions and legitimacy that courts usually wield, I engage in the thought experiment of imagining adjudication without these two typical characteristics. I therefore devote much of the chapter to describing an expressive influence that any third-­party might have over two parties in a dispute. Legal actors sometimes have advantages over nonlegal actors, but a certain part of the power of adjudicators comes from the power shared by any arbiter that two disputants agree to consult. Thus, the chapter addresses the expressive power of mediation or nonbinding arbitration. Except as noted , a nation’s courts should retain at least as much expressive power in their decisions.

Resolving Disputes with Focal Points From the discussion in Chapters 2, 3 and 4, it should be obvious that there are situations in which a third party could resolve a dispute between two individuals simply by announcing that one of them is to defer to the other. A third party’s cheap talk could create a coordinating focal point for resolving the dispute in the way the third party suggests. The essential 204

T HE POWER OF AR B I T RA L EXPRESSION

conditions are, of course, that the situation has more than one equilibrium and that the disputants have some common interest in avoiding certain outcomes. If so, they will seek some basis for coordinating, one option for which is the expression of third parties, as the experimental data reviewed in Chapters 2 and 3 confirms. Here, I explain the function and limits of arbitral focal points. Arbitral Focal Points To begin, why would disputants agree to select an arbiter? Recall the point from Chapter 2 that, in coordination games, the creation of a focal point can improve the ex ante payoffs of two players. This is the mutual benefit of coordination. Disputants may lose more from the uncoordinated resolution of their dispute than from giving in to the other side. For example, perhaps two individuals are dividing property that, as a practical matter, cannot be shared or cut in half, as two siblings dividing family heirlooms or separate dwellings. Suppose that, if they cannot agree how to allocate the property, the executor of their parents’ estate will sell the property and give them each half of the proceeds, which will destroy sentimental value, producing what each regards as the worst possible outcome.19 At the simplest level, we could model such a dispute with the BOS Game in Figure 7.1.20 George and Cindy each choose between strategy N (taking the “nice” property) and strategy S (taking the “second-­best” property) where the two sets of opposite strategies (NS or SN) means resolving the dispute in a coordinated fashion so as to keep the property from being sold. If they each insist on the same property (NN or SS), they fail to coordinate to resolve the dispute and each suffers. Taking the second-­best property (payoff of 2) is better for either than liquidating the property and taking the cash (payoff of 1), but each prefers the better property (payoff

George

Cindy

F i g u r e 7.1  

S

N

N

3,2

1,1

S

1,1

2,3

Another Battle of the Sexes Game 205

The Expressive Powers of Law

of 3). So there are two paths for resolving the dispute,21 but George and Cindy each strongly prefer a different path. Thus, this BOS game is somewhat like the HD game: The disputants lose by failing to agree, but each may purposely risk impasse in negotiations to get his or her preferred ­outcome. One familiar way to resolve this kind of dispute is to flip a coin.22 Let’s examine why that is. We first need to determine how people in this situation would behave in the absence of a coin flip. The logical baseline is what game theory calls the mixed strategy equilibrium, as discussed in Chapter 2. A mixed strategy—​playing each action with some probabilities that sum to one—​makes sense for the following reason: Neither player wants to choose N with certainty, given the possibility that the other would choose N, resulting in the waste of an NN outcome; neither wants to choose S with certainty, given the possibility that the other would choose S, resulting in SS. Instead, if the players must simultaneously decide on their final demand, game theory predicts that each player will demand N with probability 2/3 and S with probability 1/3; at this point, the returns from demanding N and demanding S are equal. This mixed strategy equilibrium produces an expected return of 5/3 for each player.23 Now suppose that Cindy flips a coin while announcing “Heads, I get N; tails, you get N.” Let us assume that this statement is cheap talk; it carries no enforcement mechanism, not even an internalized sense of guilt that will plague Cindy or George for failing to follow through on her statement. The payoff structure for the game therefore remains exactly the same as it did before the announcement and coin flip. Nonetheless, the randomizing action creates a new strategy for each player, which is to condition action on the coin as Cindy suggested. Cindy can play the strategy “If heads, N, and if tails, S,” while George plays the strategy “If heads, S, and if tails, N.” Following the expressed strategy is better in expectation for both players. The coin flip gives each player a 50 percent chance of the best payoff (3), a 50 percent chance of the second best payoff (2), and a 0 percent chance of the noncoordinated outcome both wish to avoid (1). The expected value is now 2.5, which beats the baseline of 12/3. A third party’s announcement can work in a manner similar to the coin flip. Assume a sequential game where the third stage is the BOS of Figure 7.1. At stage one, the disputants communicate about whether to consult a third party—​an arbiter. If they don’t both agree to do so, they move immediately to stage three and play the BOS. If they do agree, then in the 206

T HE POWER OF AR B I T RA L EXPRESSION

second stage they consult an arbiter, who provides an announcement or declaration favoring a particular equilibrium: either “Cindy take N and George take S” or “Cindy take S and George take N.” They can now ­correlate their strategy with the announcement by doing what the arbiter says. If they each expect a 50 percent probability of receiving a favorable announcement, they are in the same position as they were with the coin flip. Yet the benefit to Cindy and George is, by itself, not quite sufficient for the coin flip or arbiter to have expressive influence; there still needs to be a focal point. Without a focal point, given that there are only two (pure strategy) equilibria, there is a logical problem.24 Even though the expected value of the following the arbiter’s expression exceeds the expected value of playing the game without expression, the expected value of doing the opposite of the arbiter’s expression is just as good. Suppose Cindy and George each follow a strategy of defying what the arbiter says. For example, if the arbiter announces: “Cindy take N and George take S,” then Cindy takes S and George takes N. If the odds of each announcement are 50 percent, then doing the opposite also gives Cindy and George a 50 per­ ­cent chance of the best outcome, a 50 percent chance of the second best outcome, and no chance of the worst outcome. If following or defying the arbiter pay equally well, then the expected value is not a sufficient reason for an individual to perform the announced action (nor to believe that the other will do so). The expected payoff advantage of following the arbiter disappears. Nonetheless, the announced action is more likely than the omitted action because the announced action is focal. When there are only two equilibria, there is only one named equilibrium and one unnamed equilibrium, so either is in a logical sense “singled out” by the announcement. Yet as an empirical matter, announcing a strategy makes it more salient than the unannounced strategy and each player expects it to be more salient to the other. Recall my experiment with Nadler, discussed in Chapter 2, where a spinner that pointed to one equilibrium made people more likely to play that equilibrium, even though there was only one other equilibrium that was not pointed to. As a fact of human (or perhaps western) psychology, the arrow makes salient the thing it points to. Similarly, what makes the coin flip work is that its landing on heads makes salient the strategic outcome (Cindy: N, George: S) that the cheap talk previously associated with heads—​and not the opposite outcome. 207

The Expressive Powers of Law

Thus, in disputes of this kind, it pays to correlate one’s behavior with arbitral expression by following that expression. While the focal point explains the special power of the arbiter’s message, the ability of an arbiter to increase expected payoffs (to achieve the mutual advantage of coordination) explains why the parties will seek out the message in the first place. Indeed, looking back, we can see the logic of a correlated equilibrium in our initial example of the Bystander who steps into the intersection to direct traffic when the traffic light breaks (Chapter 2). He wields no sanctions and lacks the ordinary elements of legitimacy, but he still influences how drivers proceed. His cheap talk gestures create the new strategy: “If he gestures me to stop, stop; if he gestures me to go, go.” If both drivers play the same strategy, they eliminate the possibility of a Hawk/Hawk collision and therefore expect higher payoffs than their mixed strategy equilibrium. Of course there is also the new strategy: “If he gestures me to stop, go; if he gestures me to go, stop.” But, as an empirical matter, in most situations, the nonperverse strategy of conforming to the announcement is more obvious to the parties than the opposite strategy. Focal Competition: Third-­Parties and Losers The focal-­point idea matters in another way. Even if we foreclose the problem of “doing the opposite,” there is another problem—​competing cheap talk. Talk being cheap, there may be too much of it; parties other than the arbiter may offer an alternative basis for coordinating. Because the game has stages, and it takes time to consult the arbiter, other third parties—​interlopers—​may also communicate messages favoring Player 1 or Player 2. Moreover, after the arbiter speaks, the disappointed disputant—​ the loser—​might speak. The arbitral loser might propose sampling a larger number of arbiters—​“Best two out of three!”—or declare that she has now decided to reject external assistance and return to her pre-­arbiter strategy (probably a mixed strategy). So there are two problems: (1) the interloper’s influence (ex ante or ex post) and (2) the loser’s (ex post) resistance, either of which may disrupt the arbiter’s ability to resolve the conflict. Consider first the interloper. The disputants or the state may structure the arbiter’s message to ensure its salience compared to alternative messages. If the adjudication is consensual and the parties agree to designate an individual as their arbiter, their agreement creates a common belief that everyone will pay attention to what this arbiter says. The common belief will not exist for interlopers. Even if there were the special ­designation 208

T HE POWER OF AR B I T RA L EXPRESSION

of a single individual as the arbiter tends to make her message unique and therefore salient. Thus, two parties can give a third a particular power of influencing their behavior by cheap talk by agreeing (via cheap talk) to designate that individual as their arbiter. The rituals of adjudication work further towards the creation of a focal point.25 The disputants’ joint attendance at the proceeding, their standing together and facing the arbiter during the formal pronouncement of the judgment, and the arbiter’s use of customary attire (e.g., black robes), accoutrements (e.g., a gavel), and formulaic language, all work to draw attention to and create mutual salience (or common knowledge) of the declared resolution. And the history of an individual or tribunal succeeding at dispute resolution may make that arbiter’s expression stand out amongst competitors. Competing messages may also have ritual, but will usually not. Thus, the consent of the parties and/or the rituals of a successful arbiter solve the problems of competing expression. Now consider the second problem: the resistant loser. The issue is not well posed by the particular BOS game in Figure 7.1. Recall that the expected payoff of the game without an arbiter (the mixed strategy equilibrium) is 5/3. But with an arbiter, even the losing player is better off with the arbitral equilibrium because the losing payoff is 2. That fact renders noncredible his threat to return to his mixed strategy after the arbitral declaration. But this particular feature is not general. In other situations, the losing player would be better off returning to the baseline (mixed) strategy equilibrium than playing the strategy the arbiter announces. That is, the loser would be better off if both parties ignored the arbitrator. Consider the HD game in Figure 7.2. In previous chapters, I used HD to model disputes where a party reaches his best outcome by insisting while the other party Player 2 Player 1

Dove

Hawk

Dove

2, 2

0, 4

Hawk

4, 0

−1,−1

F i g u r e 7. 2  

A Final Hawk-­Dove Game 209

The Expressive Powers of Law

defers, but where beyond some point mutually insisting produces a fight that each regards at the worst outcome. For example, if two neighbors disagree about the location of their property boundary, Hawk is to insist on the disputed land and Dove as the strategy of conceding it. Dove/Dove might be to split the disputed land equally, while Hawk/Hawk represents a fight. With these payoffs, the mixed strategy equilibrium occurs when each plays Dove with probability 1/3 and Hawk with probability 2/3. For each player, the expected payoff at this equilibrium is .67.26 By contrast, if the two players designate a third party to randomly select one of the two equilibria, and each player follows this recommendation, then the expected outcome for each is now 2 (because there is a 50 percent chance of 4 and a 50 percent chance of 0). The arbiter creates a substantial expected gain for each player because he never recommends that both players select Hawk or that both select Dave. But now we can understand precisely the ex post problem: Once the arbiter announces an equilibrium outcome, the losing player expects 0, which is worse than the expected outcome (.67) of the mixed strategy equilibrium. No longer wanting the third party to resolve the dispute, he asks for another arbiter or says he abandons any desire for third-­party resolution. The general answer to this problem is that, even after one is disappointed by the arbiter’s announced outcome, it continues to be focal. A focal point does not depend on consent, but only on its ability to tilt expectations about how the players will proceed.27 The loser is better off going back to his mixed strategy only if the winner also returns to her mixed strategy, yet the focal point makes that unlikely. The analysis and experiments reviewed in Chapters 2 and 3 demonstrate this proposition by showing that the focal point influence operates in “mixed motive” games involving conflict, where the loser prefers to resist. We might say that the loser would like to return to the world without an arbiter, but after the arbiter speaks, that world no longer exists. For example, in the Bystander example, we might raise the same objection: that the driver who looks ahead to see the Bystander’s gesture for him to stop will reason that he is better off playing his mixed strategy than heeding the Bystander. Yet the Bystander is still influential because the losing driver worries that the Bystander’s gestures, by making salient the outcome where the other driver proceeds, will increase the likelihood that 210

T HE POWER OF AR B I T RA L EXPRESSION

the other driver will proceed. The expectations the focal point creates remain self-­fulfilling. My experiment with Nadler, discussed in Chapter 2, found the focal influ­ ­ence in a HD game, which we selected precisely to show that a focal point works even though a “losing” party—​the one disfavored by expression—​ would want to resist the expression and return to his mixed strategy.28 One wants to return to the mixed strategy only if the other player will do so as well, but the focal point makes that unlikely. Adjudication may seem to differ from these examples because the parties have the opportunity to communicate. But the disputants cannot make an outcome focal by their own cheap talk declarations because their predictable expressions offset one another. Before adjudication, Player 1 says he will continue to insist no matter what and Player 2 says that she will continue to insist no matter what. In the ordinary dispute, nothing breaks the symmetry of these offsetting declarations. The failure of the exchange to resolve the conflict explains why the disputants seek an arbiter in the first place. After the arbiter’s declaration, the losing party’s expression is just as powerless. The losing party says he will resist and continue to insist; the winning party says she will comply and continue to insist. The only communicative factor that differs is the arbiter’s declaration, provided by the consent of both parties. That “tie-­breaker” tips expectations towards the declared outcome. To return to coin flipping, if an individual agrees to a coin flip and then calls “heads,” the result of “tails” will create a pull of inevitability to his giving in to the other, despite his expressed desire not to give in. In sum, there are certain kinds of disputes in which the disputants will want to seek an arbiter because she will improve their expected outcome by creating a focal point resolution. That focal point can obviously influence their behavior in the case where the “losing” party is better off following the expression than not having any expression to follow. But even if, ex post, one party is worse off following the expression than not having any expression to follow, the arbiter can still create a sense of inevitability for his or her proposed resolution. By now it has occurred to readers that if the third-­party arbiter only performs the function of a coin flip, there is no reason for disputants to seek arbiters. They can flip the coin themselves. Also, arbiters usually don’t just randomly decide between the two disputing parties, but listen to evidence 211

The Expressive Powers of Law

and arguments and perhaps even investigate facts. Rather than address these concerns now, let us first consider the other way that arbiters influence disputants expressively—​by revealing information.

Adjudication as Information Revelation A second expressive theory is that an arbiter’s decision provides information to disputants. I explored information revelation in Chapters 5 and 6. According to this idea, those receiving the signal will update their belief about the feature of the world being signaled, and this change in beliefs can change behavior. For adjudication, the key is that, at least sometimes, beliefs determine disputes. If the dispute arises because the parties have divergent beliefs, the arbiter might resolve the dispute by persuading the parties to a common belief. Consider two examples. First, in an iterated PD, one’s best strategy may depend on whether one’s counterpart “cooperated” or “defected” in the last round, and this may depend on matters of observed fact. If the underlying situation is “noisy,” it is not always obvious whether an individual has cooperated or defected, so divergent beliefs could cause a breakdown in cooperation.29 Second, in a HD game of property, A demands property from B and takes costly action to get it because, under what A believes to be the customary rule and the facts, A is entitled to the property and expects B to defer. The property custom might be first possession, in which individuals aggressively insist on resources they first possessed and defer to the claims of others who first possessed a resource. If so, A’s behavior depends on her belief about the facts of first possession, that is, the relevant history of the resource in dispute. The arbiter can influence the disputant’s beliefs on these critical facts by revealing his or her own beliefs.30 As an analogy, consider the expression of a professional critic or reviewer (of film, restaurants, stereo equipment, etc.). A review is not cheap talk if the reviewer values her reputation for providing reliable information for a consumer’s purchasing decision. If the reviewer values her reputation, deriving monetary or other satisfaction from influencing these decisions, then her public statements are costly because they affect her reputation. Positively reviewing a film or wine that most recipients find revolting will damage one’s reputation; leading people to some undiscovered gem will enhance the reviewer’s reputation. Given reputational effects, a reviewer can cause readers to change their beliefs about the reviewed item, which in turns changes their behavior. 212

T HE POWER OF AR B I T RA L EXPRESSION

Another analogy, closer to the arbiter, is the informal sports referee, who lacks any power to sanction players for disobeying his rulings. Nonetheless, athletic competitors sometimes solicit individuals to serve as informal referees and often obey their rulings. Among other reasons is information revelation. If the players believe the referee cares about preserving his reputation for accurate rulings, which is plausible in many cases, then they will view his rulings as signals of his beliefs and update their beliefs accordingly. If the informal referee signals a goal in a football game, when the issue is a close one, members of the team ostensibly scored against would assign greater probability to the fact that the ball crossed the goal line, and expect the team that apparently scored also to assign greater probability to that fact. As a result, the former team is more likely to concede the goal. Like the reviewer, the informal referee has persuasive, rather than coercive, influence. To cause individuals to update their beliefs about the state of the world, a reviewer, referee, or arbiter must have a private signal of the relevant information and an incentive to reveal it. The private information could be based on independent observation or the arbiter might review evidence provided by the parties and evaluate it with her unique set of experiences and judgment. The incentive to disclose arises from the reputational concern just discussed. Individuals gain something of value from being selected as an arbiter, the price the arbitrator charges the parties, the value of influence, or something else, which means they will seek to establish and protect their reputation for the accuracy disputants seek. In most cases, the best way to appear accurate is to, as the cliché goes, “call it as you see it,” that is, for the arbiter to reveal his own judgment as to what happened, his own private signal. For example, in a property dispute arising under the custom of first possession, if the arbiter concludes (receives a private signal that) “A is the first possessor,” he reveals that belief by declaring “A is the first possessor,” which may cause B to update her beliefs and give in to A’s property claim. If, in an iterated PD, the arbiter reveals his belief that, in effect, “A cooperated in round n,” then B may update his beliefs sufficiently that he perceives it in his interest to cooperate (rather than to engage in a mistaken effort of retaliatory defection). In general, the losing disputant will comply with the arbitral expression, so long as the signal is strong enough to cause him to update his beliefs sufficiently. Thus, adjudicative signaling about facts can generally 213

The Expressive Powers of Law

influence disputant’s behavior, generating compliance independent of sanctions. Yet there are also limits to this information theory of adjudication, as we shall see next.

The Interaction of Arbitral Focal Points and Information In sum, both of the theories discussed in prior chapters—​focal points and information—​apply to retroactive third-­party expression directed at dispute resolution. Both theories suggest that arbitral expression can influence the behavior of disputants towards compliance. Now I wish to discuss the interaction between the two expressive mechanisms. Possible Limits to the Power of Arbitral Expression Each arbitral theory has problems. I already noted that the focal point theory does not explain the need for a third party. Why don’t disputants flip coins themselves instead of soliciting arbiters? Consider a second problem. The effort to solve disputes by randomly generating focal points (including arbiters if they decided matters randomly) would be subject to strategic exploitation. In the previous discussion, there was no threat of exploitation because I assumed that the dispute was exogenous—​if you simply find yourself in a dispute of the structure of Figure 7.1 or 7.2, you realize that randomization will help. But if we consider disputes to be endogenous to the method of dispute resolution, then it is obvious that randomization will attract exploitation, by rewarding those who, in bad faith, invent sham disputes.31 For example, if A is known to use randomization to resolve all property disputes, then he will encourage others to assert an ownership interest in his property in order to gain a 50 percent chance of acquiring it by randomization. People will line up to claim some the property in A’s possession (even the shirt on his back), asserting implausibly “This is mine.” Eventually, someone will win a coin toss for each item and A will wind up with nothing. We know people do sometimes randomize, but they must have a way of screening out bad faith disputes. Note the solution. The other expressive mechanism—​information revelation—​is not subject to these problems. First, if the disputants need more information to resolve their dispute, then they need a third party—​a human being—​rather than a randomizing device. Second, if the arbiter is sufficiently accurate, then she will discover and reveal the cases where one 214

T HE POWER OF AR B I T RA L EXPRESSION

of the parties is acting strategically, inventing a dispute in bad faith. Agreeing to resolve disputes after third-­party information revelation is not subject to the same strategic vulnerability. I return to these points in the next section. But note now that there are also problems in the information theory of expressive adjudication. First, information revelation will frequently not cause enough updating in beliefs to change the losing party’s behavior. The difficulty is that, in many cases, disputants are initially too confident in the rightness of their position. For example, A might treat B’s behavior as defection, and act accordingly to punish her, if he believes it to be at least 50 percent probable that she defected in the prior round. But if A initially believes it is 90 percent likely that B has defected, then even after a contrary signal from the arbiter, A may still (rationally) believe it is at least 50 percent likely that B defected. People don’t have to be particularly stubborn to think that the arbiter has simply made a mistake. There is a second problem. If two disputants recognize the possible advantage of receiving information from others, so as to avoid conflict caused by ambiguous facts, then they would presumably prefer to seek signals of a great many third parties, who collectively have the potential for producing greater belief change and greater accuracy than any one individual. Recall the Condorcet Jury Theorem and the power of aggregation (the “wisdom of crowds”) discussed in Chapter 5. This implication provides a very poor fit with informal dispute resolution—​Martin Shapiro’s “logic of the triad”—​where two disputants seek a single third party to help them resolve their dispute. Crowdsourcing is also a poor fit with formal adjudication, which is typically centralized in a single arbiter or tribunal (perhaps with limited appeals) who provides the dispute resolution services. Disputants may choose between many different fora and may appeal from one decision maker to a higher-­ranking one, but they do not just gather a large number of opinions and aggregate them as the information theory seems to imply. Yet the focal-­point theory does not suffer from these defects in the information theory. First, focal points can work without persuading the losing party of any set of contested facts. In the Bystander case, the driver told to “wait” need not be made to believe that the pertinent facts favor the other driver, only that the driver is likely to proceed given the Bystander’s hand gesture. Second, focal point creation is best achieved by a single third 215

The Expressive Powers of Law

party. It is a disadvantage to have a multitude of parties declaring how the dispute should be resolved because, with divergent recommendations, no one message would be focal. Thus, the punch line: Adjudication works expressively by combining the focal point and information mechanisms. The Synergy of Arbitral Focal Points and Information My claim is that an arbiter will wield an expressive influence on disputants if she draws on both the power of focal-­point creation and information revelation, doing whatever the situation requires. The synergy between these mechanisms of expressive power is that each one addresses weaknesses in the other. Thus, in common circumstances, adjudicative expression can itself induce even selfishly motivated disputants to comply. These points require elaboration. Fact-­Finding Works Against Bad Faith Disputants. ​Suppose the third-­ party arbiter studies the evidence presented by the parties or acquired by a separate investigation and reaches a conclusion about the disputed facts. To make matters simple, let us say that the arbiter can reach only one of three possible conclusions: (1) the facts favor party A, (2) the facts favor party B, or (3) the facts do not favor either side because the evidence is closely divided. Assume that the arbiter reveals her actual beliefs to the parties. If the arbiter reveals belief (1) or (2), then they will update their beliefs, which might end the dispute by causing the losing party to agree with the prevailing party. By contrast, revealing belief (3) will have very little effect on the beliefs of either party, and is not at all likely to cause the disputants’ beliefs to align with one another. Revealing belief (3) seems like a failure that ensures the continuation of the dispute. But the arbitral message (3)—​“the facts do not favor either side because the evidence is closely divided”—​is in other respects the best signal for resolving the dispute. When the facts are closely divided, the dispute is likely to be genuine, rather than strategic. It is not likely to be the kind of fraudulent dispute where a person appears out of nowhere and implausibly claims to own the property that has been in the hands of another for a long time. If message (3) erases doubt about the genuineness of the dispute, then the logic of randomization will apply and each party will benefit from a 50 percent chance of prevailing. In other words, one should refuse to randomize as a general mechanism for resolving all disputes, for fear of 216

T HE POWER OF AR B I T RA L EXPRESSION

strategic exploitation, but an individual can gain from randomizing in just those cases where a credible arbiter has announced that the facts do not clearly favor either side. Information revelation screens out those cases where the parties would be exploited by using randomization and leaves those where they would not. Put differently, to address the problem of strategic disputes, we need a mechanism that tends to resolve disputes against the party who raises a dispute strategically. Randomization does not do this but fact-­based adjudication does. Thus, although one party may not be able to tell whether the other party’s claim of dispute is genuine or strategic (because he is confident in his position in either case), he will still expect to win the adjudication far more often when the other side’s claim is strategic. And a party expects to lose disputes when he decides to raise a dispute strategically. With these assurances in place, the parties may then agree to randomize in the cases where the arbiter believes the facts are unclear. Of course, an arbiter will err on occasion, so this solution is not perfect. A party will still expect to win some small number of strategic disputes. If adjudication were costless, then any positive error rate would prevent expressive adjudication from working because the creation of a strategic dispute would have a possible upside (the favorable result when the arbiter errs) and no possible downside. If it is costless to invest in bogus disputes, then it doesn’t matter if the fraudulent disputer has a 50 percent chance of prevailing (as with randomization) or a 0.5 percent chance (as with an arbiter attempting to identify invented disputes). Either mechanism will attract an endless number of strategic disputes. However, dispute resolution, even by arbitral expression, is not free and, that cost is a part of its effectiveness. One cost is any fee the arbiter charges. Another is the time and resources adjudication requires. A third is reputational: If strategic disputes are sometimes exposed as such, then a reputation for inventing disputes will make one more likely to lose future adjudications, even those one is entitled to win. These costs constitute disincentives to fabricating disputes because parties are certain to incur costs and yet unlikely to gain the benefits of winning the adjudication. If A makes up a fantastical story to assert an ownership interest in some property that has, by all appearances, always belonged to B, then A and B will expect the arbiter to rule against A with a very high probability. If A must incur the costs of bringing 100 or 100,000 such strategic disputes in order to win one of them, then the expected costs may easily swamp the 217

The Expressive Powers of Law

expected benefits. The point is not that the system will necessarily, or even probably, deter all strategic disputes but that it could deter enough of them to make the parties want to randomize in disputes the arbiter says are “close.” Subsequently, the outcome selected at random is now focal, which creates self-­fulfilling expectations this outcome will occur. Of course, the party entitled to prevail must also incur the costs of adjudication. There is a literature on negative expected value suits that shows how a plaintiff can, under some circumstances, exploit the defendant’s litigation costs to extract a settlement even when the plaintiff’s claim costs more to litigate than it is worth (in expectation).32 This theoretical result, however, is not general, but limited to certain conditions, such as separate stages of the litigation in which costs are incurred. One implicit condition of importance for this discussion is that the plaintiff has the power to force the defendant to defend itself or face a default judgment. Yet my assumption in this chapter, as stated initially, is that the arbiter lacks the power of sanctions over the disputants, which means there are no legal sanctions to enforce a default judgment. Instead, I seek to explain the examples of sanctionless adjudication described at the beginning this chapter, where the parties must consent before the arbiter can “decide” the case. This situation robs the strategic disputant of essential leverage. (For this reason, when courts acquire the power of sanctions and compulsory personal jurisdiction, the prospect of negative expected value suits may undermine the power of arbitral expression). Thus, if the arbiter’s revelation of information is sufficiently accurate, it overcomes the two weaknesses of pure randomization. First, fact-­finding requires a third party. Second, by screening out most strategic disputes, it minimizes the risk of strategic exploitation. Because the parties do not know in advance whether they need fact-­finding to screen out a strategic dispute or randomization (a coin flip) to resolve a genuine dispute, they require a third party in every dispute. The next step is to address the objections I previously raised to the information theory of adjudication. Among Good Faith Disputants, the Arbiter Creates a Focal Point. ​Now let’s return to the other two messages the arbiter may send: (1) the facts favor party A, and (2) the facts favor party B. If an accurate arbiter sends these signals, the losing party will sometimes update her beliefs sufficiently to align them with the prevailing party, thereby resolving the dispute. But, as noted before, frequently this will not happen. If a party’s prior belief 218

T HE POWER OF AR B I T RA L EXPRESSION

about the disputed facts is sufficiently strong, the new information will not cause enough updating to change the behavior of the party the arbiter rules against. Experience suggests that this is the normal situation, where disputants still believe in the merits of their positions even after the arbiter rules for the other party. The crucial point, however, is that the parties don’t need to rely entirely on information revelation. Once they sufficiently minimize the risk of strategic disputing, the efficiency of randomization provides an incentive to accept it. Recall the mutual benefit of coordination: Correlating strategies with a random event gives each party an equal chance of reaching one’s preferred payoff, while at the same time reducing to zero the chance of uncoordinated outcome—​continued conflict—​that harms them both. Even for genuine disputes, each party knows that there is some chance that the arbiter will err in her decision about the facts. But if the arbiter is unbiased, each side expects to be the beneficiary of error as often as it is harmed by error. Because (or when) failing to resolve the dispute is costly to both sides, both parties are better off ex ante accepting the equal chance of prevailing, which means correlating their strategies with the arbiter’s message. Note how the psychology literature on the optimism bias applies here.33 Optimism favors the use of an arbiter. Even people with optimism bias in situations of ambiguity might recognize that their odds of winning the coin flip really are only 50 percent. But they both might still believe that their chances of prevailing before an unbiased arbiter are greater than 50 percent. Of course, the optimism bias makes it even less likely that the arbiter’s message causes the loser to switch his belief to one favoring the other party. But the randomization/focal point analysis still gives him a reason to comply. I also raised a second objection to the information story: It implies that disputants would seek the input of a multitude of third parties, to gain “the wisdom of crowds,” rather than just one arbiter or tribunal (perhaps subject to appeal). But correlating strategies with a focal point does require identifying in advance the one and only signal that will serve as the focal basis for coordinating, which explains why the parties seek a centralized and hierarchical set of adjudicative signals. As explained, the focality of the agreed-­to arbitral expression is necessary to produce ex post compliance with the endorsed outcome rather than its opposite and rather than with some competing expression (from an interloper or the loser). That 219

The Expressive Powers of Law

the expression has to be focal explains why adjudication tends to be centralized in a single actor or institution. The Arbiter Masks Factual Uncertainty. ​Up to this point, as I have explained the power of adjudicative expression, the process does not entirely map onto real-­world adjudication because actual arbiters do not send signals like the third one described that fail to favor either side. Arbitrators and courts do not say “too close to call.” Here, I explain how and why the arbiter’s uncertainty is concealed. Recall our assumption: An arbiter assesses the competing claims of the disputants in one of three ways: (1) the facts favor party A, (2) the facts favor party B, or (3) the facts do not favor either side because the evidence is too closely divided. But now suppose that the arbiter adopts a strategy of only sending either of two messages: that A prevails or that B prevails. The arbiter maps his private signal and resulting beliefs to the public message he sends as follows: “If I believe (1), I signal ‘A prevails.’ If I believe (2), I signal ‘B prevails.’ If I believe (3), I randomly choose between delivering the message ‘A prevails’ and the message ‘B prevails.’   ” In other words, the arbiter no longer differentiates in her message between the cases where she perceives the evidence to be clear and those where she perceives the evidence to be unclear. Instead of publicly identifying the close cases where the disputants could gain by randomizing, the arbiter herself randomizes in those cases and then sends one of the messages she would send when the evidence actually favored a particular party. From the parties’ perspective, there is no difference in following a strategy that randomizes in certain cases the arbiter identifies and following a strategy that permits the arbiter to randomize in those same cases. If A and B are planning on correlating strategies with message (1) and (2) and randomizing after message (3), then they would also be willing to correlate strategies with messages (1) and (2) and to allow the arbiter to randomize between them in case she believes (3). For the parties, nothing of importance changes in moving from the fully candid to the more discreet arbiter. At this point we reach a process that looks more like arbitration or adjudication, where the arbiter purports to decide every case and resists full disclosure of his own uncertainties about the dispute. The question arises, however, why arbitrators and judges rarely reveal their own indecision. If arbiters occasionally reach belief (3), why do we observe a near universal refusal to signal this indecisiveness? There are 220

T HE POWER OF AR B I T RA L EXPRESSION

many possible reasons. One is that individuals are more likely to comply with decisions they regard as legitimate and individuals seem far less likely to perceive as legitimate a decision where the arbiter admits uncertainty and randomizes.34 But there is a self-­interested reason I want to identify if only because the idea of concealing uncertainty cuts against my claim that individuals who care about their reputation for accuracy will signal their actual beliefs. If that is true, why not signal one’s actual indecision? The answer is that the arbiter is concerned about revealing fallibility. Disputants seek accurate arbiters. The arbiter wants to project competence and accuracy in order to obtain future arbitral opportunities. That goal creates an incentive to conceal the failure to form a determinative belief, that is, to conceal the fact that one had to randomize. In a sporting event, for example, a referee who has to resort to randomization ten times a game will seem less competent than a referee who has to resort to randomization only five times a game, other things equal. Even though everyone knows there are some calls that are too close for anyone to determine in real time, those competing to be referees will understandably engage in a race to “the bottom,” where the bottom means the referee claims never to require randomization because, in every case, he has actually determined the relevant facts. To summarize: disputants may comply with the declared outcome in a process of dispute resolution not only because the arbiter threatens sanctions or possesses legitimacy, but also because the arbiter’s expression reveals information and creates a focal point. In many circumstances, disputants have an incentive to gather information on the facts underlying their dispute and to find a way to coordinate, in both cases by seeking out and complying with arbitral expression. These two functions work together. Expressive adjudication can generate compliance because disputants benefit, ex ante, by seeking an arbiter whose information revelation minimizes the problem of strategic disputing, which in turn makes it attractive for the disputants to correlate their behavior on the arbiter’s declared resolution, the focal point. An Extension: The Arbiter Resolves Disputes Over Rules. ​So far, I described the analysis only for disputes of fact. But arbitral expression can also resolve disputes over customary rules, again, independently of sanctions or legitimacy. One possibility is that one party is clearly wrong and that the existing 221

The Expressive Powers of Law

custom really does resolve the dispute in favor of the other party. Here, the arbiter signals the actual existence of the customary rule, that it clearly favors one party over the other. This is similar to the case where the facts of the dispute clearly favor one party. For the reasons explained, if the arbiter is unbiased, the parties benefit ex ante by deferring to her signal of the rule. The other possibility is that there really is no rule governing the dispute because the relevant custom is ambiguous and doesn’t clearly favor either party. Recall from chapter 4 that when the custom is ambiguous, clarifying expression can influence behavior by resolving the ambiguity. Chapter 4 discussed the clarifying mechanisms of codification and soft law, but much of the actual clarification of custom is achieved by courts. For an individual dispute, an ambiguous customary rule is the equivalent to the factual situation where the facts do not favor either side because the evidence is too closely divided. In this situation, the parties can do no better than to resolve their dispute by randomization, which the arbiter can do. The ICJ sometimes resolves disputes over customary international law (CIL), clarifying its requirements. For example, in 1993, the ICJ decided a Maritime Delimitation Case between Greenland (formally, part of Denmark) and the island of Jan Mayen, part of Norway.35 In deciding the boundary of the fishery zone and continental shelf, the Court considered “relevant circumstances” for deviating from the line halfway between the two territories, a standard part of CIL (and oddly reminiscent of the blockhead fish experiment from Chapter 3). The ICJ specifically rejected Denmark’s argument that population and economic differences in the territories were relevant. (Norway’s island was sparsely populated). The case generated compliance by the parties with the line it drew, but the theory also predicts an influence on future negotiations between nations over maritime boundaries.36 Recall the experimental evidence finding that focal points affect bargaining.37 One expects that the ICJ’s rejection of the relevance of socioeconomic factors to the boundary strengthens the position of future bargainers taking the same position, especially because the parties may expect that bargaining impasse will ultimately lead to ICJ litigation and an opinion taking that position. As previously discussed, it is difficult to avoid complying with the ICJ when both sides have an interest in coordinating to avoid escalation of their dispute. Of course, if the arbiter is creating precedent, we expect her to decide the case by identifying the fair or efficient rule, rather than flipping a coin. 222

T HE POWER OF AR B I T RA L EXPRESSION

As long as such a selection process, ex ante, favors neither party, however, they will both benefit ex ante by giving the arbiter this authority for resolving their dispute. Obviously, confidence that the arbiter will select a fair or efficient rule may also give the rule legitimacy, which can provide another reason to comply.

A Reinforcing Role for Reputation Until now, I have ignored the effect of reputation on the analysis of arbitral expression, but let us briefly consider how it reinforces the expressive power. Return to the example of disputes that have the structure of a HD game. As before, I am not going to model the game, but use it informally to capture situations where the parties to a dispute regard the worst possible outcome as the unconstrained conflict where neither backs down (Hawk/Hawk). Assume that there is heterogeneity among individuals in their payoffs from unconstrained conflict (e.g., an altercation or bargaining impasse), either because some individuals are more likely than others to win or lose the conflict or are likely to suffer harm that is greater or lesser than average. Thus, we can categorize individuals by type, as by having higher-­or lower-­ than-­average expected costs for the Hawk/Hawk outcome. Those with lower costs are the “strong” type; “weak” types have higher costs from fighting. Now suppose the existence of this asymmetric information: Individuals know their precise payoffs, but do not know the exact payoffs of other individuals. So there is uncertainty in the population whether any particular individual is strong or weak. But individuals have some knowledge of another individual’s past interactions, from which one might judge whether the other is strong or weak. The estimate is noisy, especially because the parameters can change over time. These assumptions create a role for signaling and reputation, where an individual can invest in the Hawk strategy, insisting on her demands, despite the short-­term losses from Hawk/Hawk outcomes, so that others will expect her to play Hawk in the future. I am still assuming that Hawk/ Hawk outcomes are still sufficiently bad that few players will ever intentionally reach that outcome. The role for reputation is also limited by the fact that a Hawk/Hawk outcome entails fighting and the results of a fight may themselves reveal that a weak player is weak. This environment may produce an equilibrium like the following: When individuals of starkly different reputations dispute, the weaker party defers 223

The Expressive Powers of Law

to the stronger party, but when the difference in reputations—​the apparent “power gap”—​is below some threshold, the parties resolve their dispute by following a customary rule. For example, given a property custom of first possession (the first possessor insists and all others defer), the strong claimant might defer to the weak first possessor, unless the difference is too great, in which case the weaker first possessor defers to the strong claimant. Reputation may affect behavior in several ways. First, when two parties dispute and their power gap is actually just below the threshold, a stronger party might decide to incur the cost of playing Hawk so as to signal to others that it is stronger than it actually is, i.e., that the gap in power is above the threshold. The stronger party will expect to prevail in these Hawk/Hawk outcomes (but the expected costs would have exceeded the expected gain, but for the reputational gain). Second, just above the threshold, a weaker party might play Hawk for the same reason, to signal that the power gap is smaller than it actually is. The weaker party expects to lose the fight, but the net effect on reputation may be positive as everyone expected this fighting outcome, but are surprised at the decision to fight. Third, equilibrium outcomes—​where one plays Hawk and the other Dove—​need not but might damage the reputation of the party playing Dove. There is no damage when one behaves as expected, but the stronger party loses reputation by deferring to the custom when observers thought the strength gap was above the threshold, while the weaker party loses reputation by deferring to the stronger party when observers thought the power gap was below the threshold. How does this analysis change the non-­reputational story of expressive adjudication that constitutes the main argument of this chapter? For some cases, the reputational issues might cause parties to decline to seek an arbiter in the hopes that this too might signal strength. For other cases, nothing changes. It is sometimes obvious to everyone that two disputants are of near-­equal power, so a Hawk/Hawk outcome will seem too much like one side is bluffing. There will not be enough reputational gain to be worth the mutually bad outcome, so the disputants will be in the same situation as before. They will seek an arbiter to resolve the factual or customary ambiguity causing their dispute, thus avoiding Hawk/Hawk. Yet reputation creates one interesting new effect: It makes the Dove/ Dove outcome worse. I have neglected much discussion of this outcome; what mostly drove the previous analysis was the parties’ desire to avoid 224

T HE POWER OF AR B I T RA L EXPRESSION

Hawk/Hawk. Indeed, without reputation, Dove/Dove might even be an efficient outcome, despite not being an equilibrium. Yet in the present circumstance, Dove/Dove will damage the reputation of at least one party because at least one party behaved in a surprisingly weak manner. Perhaps the observers expected the party favored by the custom to play Hawk, but that party apparently perceived the power gap as too large, signaling its private information of its weakness. Alternatively, the observers expected the party disfavored by the custom to play Hawk, because the strength gap exceeded the threshold, but that party perceived the gap as too small, signaling its private information about its lack of strength. Possibly both parties will look weaker, if different observers had different expectations, as where they expected the custom to control, but the custom is ambiguous. Thus, in some cases, the parties have an additional incentive to seek an arbiter—​not only to avoid Hawk/Hawk, as before, but also to avoid Dove/ Dove. The underlying intuition here is that a party cannot afford to have a reputation for failing to “stand up for his rights,” as by the failure to be aggressive in the defense of one’s property. In this context, there is a new role for arbitral expression, which is to minimize reputational damage to the disputants. First, when the parties seek the assistance of a third party who declares that one disputant is entitled to prevail, the declaration effectively eliminates the possibility that neither party will play Hawk (because the party declared as the winner will always insist). Second, when there is an ambiguity in what the custom requires (because of uncertainty about facts or the customary rule), the arbiter’s declaration that the custom requires one party to defer diminishes the competing inference that the custom favored her but she deferred out of weakness. I don’t say that it eliminates the inference. A party might show strength by defying the arbiter (doing so shows a willingness to insist even knowing that the other party will insist). But if most parties comply with the arbiter’s declaration to defer, for reasons already stated (to avoid Hawk/Hawk outcomes), then it weakens the inference that those who obey the arbiter’s declaration to defer are weak. Note that the reputation described here is quite different from the reputational account some theorists give for why, say, nation-­states would comply with international court judgments despite the absence of credible judicial sanctions. In Andrew Guzman’s theory, for example, states comply with rules to signal that they are a “good” type,38 the sort committed to complying with international law. In Guzman’s account, a stronger state 225

The Expressive Powers of Law

backs down in the face of an adverse ruling by an international court because it will signal itself as being more reliable by following the court. In the alternative reputational account given here, a state seeks to signal not that it is a reliable follower of international law (a good type) but that it is a reliable defender of its international rights (a strong type). Thus, when favored by custom, a state refuses to back down to prevent other states from challenging its customary entitlements; knowing this, the other state will back down to avoid the cost of unconstrained conflict. Up to a point, even a stronger state will back down to a weaker state if the former knows that the latter will not back down. One of the difficulties for Guzman’s account, for which he has been criticized,39 is explaining why reputation in one issue domain affects reputation in another issue domain. If a nation violates human rights, why will that weaken its reputation for complying with the rules of international trade? This is an important question because the cost a nation incurs from failing to adhere to the international law of human rights may be negligible unless it adversely affects its reputation in other domains. Guzman wants to say that a nation that shows a weak commitment to international law in one domain will have a weak commitment in other domains. But, as his critics observe, this does not follow. A nation’s leaders might face domestic pressures from powerful political groups to honor its trade obligations (because exporters want to avoid trade sanctions from other nations), but lack domestic pressure to honor its human rights commitments. An observer can then infer that its failure in one domain does not predict failure in another domain. This criticism does not apply to the reputational story given here. When a disputant nation plays Dove in the presence of a custom entitling it to play Hawk, the only rational inference is that the nation bears unusually high costs from conflict. Those costs might be high because the nation’s leaders assess themselves as being unlikely to prevail or as being likely to incur high costs from the attempt. Either weakness transcends the particular issue domain that the custom governs. If a nation, for example, doesn’t insist on what the custom defines as its territorial integrity, but offers compromise, then the implication is that it is militarily weak. That military weakness will apply across domains. This theory leads to a competing hypothesis about international law: Where Guzman would expect a nation’s failure to comply in one legal domain to be correlated with failure in other domains, the reputational story I am telling directs our attention 226

T HE POWER OF AR B I T RA L EXPRESSION

instead to the victims of noncompliance, predicting that a nations’ sufferance of an international law violation in one domain will be correlated with its sufferance in other domains. In sum, where the interactions take the form of a HD game or something similar, there is a strong reputational concern for playing Hawk and not compromising when the custom creates that expectation. When an arbiter publicly clarifies the meaning of the custom in a particular dispute, the party declared as the winner therefore has a powerful reputational reason to follow through by insisting on its rights, giving the other party an incentive to back down.

The Limits of Expression and the Need for Impartiality Now for some caveats. As explained in prior chapters, there are limits to the power of focal points and information revelation. The expressive theory does not apply to disputes outside its domain and does not predict perfect compliance even in its domain. Thus, there are certainly some conditions in which arbitral compliance depends strictly on the presence of sanctions, legitimacy, or some other mechanism. The first condition for expressive power—​as a coordinating focal point—​ is a situation of multiple equilibria where both parties have some common interest in avoiding unconstrained conflict and neither party can achieve a first-­mover advantage. I have represented this situation by using the BOS and HD games. These situations commonly arise where parties of roughly equal power both regard the continuation or escalation of the dispute—​as by mutual resort to self-­help—​as the worst possible outcome. By contrast, some games have just one equilibrium, in which case the payoffs alone determine how the players will behave, and there is no room for focal influence. Severe inequalities of power may create situations where the only equilibrium is that the stronger party wins the dispute. The United States did not particularly fear conflict with Nicaragua in the 1980s when it mined a harbor, and so there was no incentive to back down when the ICJ ruled against it.40 Another single-­equilibrium situation exists, even between two individuals of roughly equal power, if the object in dispute involves high stakes relative to the cost of conflict. If a violent dictator is locked in a political struggle for power with opposition leaders, neither side is likely to pay attention to judicial expression when the winner stands to gain the political control of the country and the loser will face exile or death. 227

The Expressive Powers of Law

The second condition is the need for an arbiter whom the disputants regard as sufficiently accurate to screen out strategic disputes. If the perceived bias or error rate is sufficiently high, there is too great an opportunity for strategic disputing. This condition points to an important positive implication of the expressive theory of adjudication: the importance of impartiality. The perception of arbitral impartiality matters for reasons independent of a concern for legitimacy. Instead, impartiality matters to both the analysis of focal points and of information revelation. Where arbitral expression can resolve a dispute by creating a focal point, I previously noted that the key condition is that each side has a sufficient probability of winning the message it prefers. As an example, reconsider the HD payoffs of Figure 7.2. Each party expects a payoff of .67 from the mixed strategy equilibrium. With randomization, where each party has exactly a 50 percent chance of winning, the expected payoff is 2. The point I made is that randomization is not essential; the parties will each prefer arbitration as long as each expects to prevail with a probability sufficiently close to 50 percent or greater. But now we can see how much bias the parties can tolerate. Suppose that a disputant in Figure 7.2 perceives that the arbiter is biased in favor of the other party. Let us quantify the bias by saying that one party believes that in the event of a “genuine dispute” (i.e., the arbiter believes the factual or conceptual issue is “too close to call”; for simplicity, let’s assume in this example that anyone raising a sham dispute will lose with certainty), the arbiter is only 10 percent likely to rule in his favor and 90 per­­ cent likely to rule in favor of the other party. Now the expected returns for this party after arbitration are only .4 (90 percent chance of receiving zero and 10 percent chance of receiving four), which is lower than he expected without arbitration (.67). As a result, this individual will seek to avoid adjudication with this particular arbiter. Instead, given these payoffs, each party needs to believe it is at least 16.75 percent probable that it will win in this scenario (a genuine dispute) in order to consent to the arbiter, which leaves room for allowing considerable bias, if there is no other choice but the mixed strategy equilibrium. A better choice, of course, is a different arbiter, without bias, but to a degree, the biased arbiter is better than none. Information revelation also requires a degree of impartiality. A party who believes the arbiter is biased against him will be less persuadable; given an arbitral signal contradicting his current beliefs, he will update his 228

T HE POWER OF AR B I T RA L EXPRESSION

beliefs more when he thinks the arbiter is neutral than when he thinks the arbiter is biased against him. Moreover, only a sufficiently accurate signal will screen out strategic disputes; the party perceiving arbitral bias against him will still fear strategic exploitation if the arbiter is too likely to rule in favor of a sham disputant. If so, then it will not pay an individual to correlate his strategy with the arbitral expression. As a result of these rational considerations of self-­interest, a party will not consent to even purely expressive arbitration (i.e., without sanctions) if the perceived bias is too great. When a party refuses consent for this reason, what will happen if the adjudication proceeds anyway (because of, say, compulsory jurisdiction) and the arbiter (as expected) rules against that party? The threat of sanctions may still generate compliance, but the expressive effect is weakened or eliminated. The party perceiving arbitral bias against him will fear strategic exploitation. He might take costly action to self-­commit to ignoring the arbitration, seeking to tie his hands against compliance (as by resorting more quickly to violent self-­help). As a positive matter, impartiality matters for reasons beyond legitimacy. If we observe that parties seek out and obey arbiters they perceive as impartial to a greater degree than arbiters they perceive as biased, this is not by itself evidence for the legitimacy theory of legal compliance. For those who are skeptical of legitimacy theories, or for contexts in which one is skeptical of such theories, there is still a reason to predict that perceived bias would affect compliance.

Explaining the Puzzle of Courts without Sanctions To return to the observation of Martin Shapiro that opened this chapter, it is no surprise that individuals “across cultural lines” seek third parties to help them resolve disputes. Third-­party arbiters have even more power to resolve disputes than has been previously understood. Independent of sanctions or legitimacy, arbiters can sometimes resolve disputes expressively, by the revelation of information and the creation of focal points. By revealing information about the facts and customs in dispute, third parties can screen out insincere disputes. By a mutually observed declaration, third parties make salient one way for the parties to resolve the dispute. When the dispute is one requiring coordination (because each side wants to avoid the outcome of unconstrained conflict), the parties benefit ex ante by seeking arbitral expression and, ex post, the arbitral expression creates self-­fulfilling expectations of compliance. Finally, arbitral expression can 229

The Expressive Powers of Law

also invoke reputational concerns because the disputant the arbiter declares as entitled to prevail in the dispute does not wish to back down for fear of signaling weakness to the world. The theory explains compliance with the arbiters discussed at the beginning of this chapter. Regarding medieval Icelandic courts, one can certainly read William Miller’s account to support a legitimacy theory, but the theory of focal points is quite consistent with the evidence, given the background threat of private violence. Individual litigants could enforce or resist a judgment only by gathering the support of their kin. “[Power] meant having others think one had the ability to muster bodies to assist in the various procedures that made up a legal action.”41 Translated into game theory, the situation is one with multiple equilibria similar to a HD game where the Hawk/Hawk outcome is violence. The expected costs of the violence were usually higher than the expected costs of giving in to the other’s demand, yet each preferred to have the other give in to him. In this setting, a court could influence the behavior of the parties by providing a focal point. Once the court announced a winner, it raised the expectation that the winner would not back down. Once it appears the winner will certainly fight, this expectation made it more difficult for the loser to gather or retain kin to fight on his behalf. The expressive theory can also explain why eighteenth-­century pirates consulted their quartermaster for mediation of disputes, even though each party retained the power to force the dispute into a duel.42 The disputants could probably rely on the quartermaster to identify frivolous disputes and to randomize in close cases. If he announced a recommended resolution, each side could be certain that the other side would accept nothing less. Thus, after the mediation, the choice would be clear: accept what the quartermaster suggested or possibly die in a duel. In many cases, the proposed split would be better for each than the expected outcome of a duel. Besides the effect of focal points, we can easily imagine the strong reputational incentive to insist on anything the quartermaster suggested one should have, which in turn creates an incentive to defer to anything the quartermaster suggested the other party should have. International adjudication follows the same pattern. As previously discussed, Garrett and Weingast were the first to offer a focal-­point explanation of a court, specifically, the European Court of Justice.43 They posit that treaties are self-­enforcing agreements for cooperation in an iterated 230

T HE POWER OF AR B I T RA L EXPRESSION

PD game, but because contracts are incomplete, unforeseen contingencies create situations where the parties disagree on what behavior counts as “cooperation.” In the absence of natural focal points defining cooperation, “an institution” such as an international court “can construct one by devising the required set of specifications (as to the nature of the agreement, and hence as to what constitutes cooperation and defection) and by making them known to the community.”44 Garrett and Weingast omit the problem of strategic disputing and thus the need for an informational component to adjudication, but that analysis supports their general claim. Similarly, Ginsburg and I used these theories to explain the high compliance rate (at least 68 percent) we measured in the decisions of the ICJ.45 The compliance rate with the ICJ is high, we claim, because nations tend to bring to the ICJ those disputes for which the factors identified are sufficient to generate compliance (as discussed in Chapter 5). We find a large part of the ICJ’s success occurs in territorial disputes where both parties desire the adjudication to resolve ambiguities in facts or conventional categories defining possession. We believe there is a strong convention respecting possession, but that factual and conceptual ambiguities produce conflict. The ICJ examines the facts and announces a ruling that clarifies the convention. The tribunal works like the Icelandic court—​ by granting one side the disputed territory, it seems to ensure that this side will not back down. If not, the declared loser is best off ceding the territory. Finally, consider the more complex dynamic that explains the success of dispute resolution during the California gold rush, despite the weakness of courts at the time.46 Andrea McDowell describes the compliance with adjudicative declarations as arising from the threat of private third-­party enforcement, those miners other than the one immediately victimized by “claim-­jumping.” This is different from the previous examples, which involve second-­party enforcement or self-­help. But we discussed this situation in Chapter 4, when I used the focal point power to show how sanctions arise endogenously. We may imagine that the miners understood the mutual value of some form of order, which required that they coordinate on which party to threaten with informal sanctions. In other words, when two miners asserted ownership of the same mining claim, the worst outcome was for half of the miners in the area to side with one disputant 231

The Expressive Powers of Law

against the other; the result would be a violent chaos. The arbitrators’ expression created a focal point that facilitated the third-­party sanctioning, so that it focused solely on the party declared to be the loser. The arbiter conducts third-­party sanctioners like an orchestra conductor coordinates musicians. Arbiters will still generate more compliance if they wield the additional powers of sanctions and legitimacy. Legal compliance is difficult enough to achieve that there is no reason to rely only on the power of arbitral expression. But sanctions and legitimacy are not the arbiter’s only power. The logic of the triad is expressive.

232

8 Normative Implications

A

s I said in Chapter 1, this book primarily concerns the expressive effects of law, i.e., causal theories about law’s expressive influence on beliefs and behavior. Because compliance is of paramount concern to legal scholars—​we often want more compliance than we have—​we should seek a full understanding of the mechanisms that produce it. This book describes two mechanisms that have been largely neglected: law’s power to affect behavior by creating a coordinating focal point and by revealing information. In this concluding chapter, I want to move beyond the positive theory and identify some of the normative implications of the law’s expressive powers. Much of the existing normative theory of expressive law or legal expressivism is nonconsequentialist. So the question arises: Do the causal theories of legal focal points and information signaling have anything to contribute to the normative analysis of law? As much of the positive theory in this book remains to be tested, many of these normative implications are tentative, and I will identify them only briefly. In the following three sections, I discuss the efficient use of law’s expressive powers, critique the use of conventional meaning in legal doctrine, and offer the law’s information power in support of the endorsement test of the Establishment Clause.

The Costs of Enforcement: Using the Law’s Expressive Powers Efficiently We should want to use the law’s powers to produce compliance in an efficient manner. Starting with Gary Becker’s early theoretical work, this kind 233

The Expressive Powers of Law

of input-­output efficiency is a standard idea for the use of legal sanctions.1 One adjusts the probability and severity of sanctions to minimize the total costs of legal violations and legal enforcement. For expressive powers, we should have the same goal, though we may have less control over the variables determining expressive success for a particular law. We also should conserve sanctions for domains in which the law lacks expressive force. Consider three normative implications of the focal point theory (ignoring the information theory for the moment). First, optimal enforcement should be sensitive to the difference between domains in which coordination produces some compliance and domains in which coordination produces no compliance. Where there is no focal-­point power, optimal enforcement can remain focused only on the conventional variables. But where the law successfully provides a coordinating focal point, society does not require as much enforcement through sanctions. Optimal enforcement thus puts fewer enforcement resources into domains in which law is a focal point and consequently more resources into domains where law is not a focal point. For example, compared to the world without law’s coordinating power, society need not allocate as many enforcement resources to the rules requiring individuals to drive on the right side of the road and to stop at red lights. That is not to say that we should cut conventional enforcement to zero. As explained, “outliers” will sometimes want to disobey even these rules of the road because they gain an unusually large advantage from the violation, are extremely risk preferring, and/or optimistically believe that they can miraculously avoid an accident. Another relevant factor for a complete normative analysis is the unusual severity of the harm risked by the particular violations of driving on the wrong side of the road or running red lights. Nonetheless, the bulk of compliance with these rules likely arises without formal legal sanctions because, when there are other vehicles on the road (when compliance has social value), drivers have an incentive to stay on their side of the road and stop at red lights. The level of enforcement can be lower than what would be required in the absence of law’s coordinating power. By contrast, there is usually no element of coordination in conforming to the law against theft. When individuals have or create the opportunity to take the property of another by force or stealth, their dominant strategy is often to take it. The threat of self-­help that normally enforces property rights is not available here because the stealth or force gives the thief a 234

NOR M A T IVE I M P L I C A T IONS

first-­mover advantage. Or consider speeding. There is an element of coordination here: A driver wants to roughly match the speed of traffic. But that kind of coordination may (and frequently does) produce more speeding rather than less. Thus, compared to the world without law’s coordinating power, it makes sense to focus more enforcement resources on the control of crimes like theft and speeding. The second implication concerns the abstract tradeoffs between rules and standards. There is a substantial theoretical literature on when it is best to embody the law into a rule and when it is best instead to use a standard.2 Rules are specific in identifying the legal effect of a certain set of concrete facts. Standards identify some more general concern and leave for later consideration how the exact facts bear on the issue. The requirement that drivers limit their speed to forty-­five miles per hour is a rule, while the requirement that drivers limit their speed to what is “reasonable and proper” under the conditions of the road at the time is a standard.3 The requirement that one refrain from smoking inside a restaurant is a rule, while the requirement that the state give citizens due process of law before depriving them of liberty is a standard. The categories are not rigid, but points along a continuum. The literature addresses the conditions in which rules are better than standards and vice versa. The focal-­point theory identifies a new virtue of rules, or a new cost of standards. Rules are potentially better at harnessing law’s focal-­point power and thus better at generating compliance in this manner. The point applies only when the situation is one requiring coordination (there being more than one way for individuals to “match” their behavior with each other for mutual benefit) among a large, heterogeneous population. With such an audience, the law can provide a focal point only if it is sufficiently clear. It is difficult to see how a standard could be clear enough to align the expectations of everyone in a large, heterogeneous audience. By creating a focal point, the rule might “buy” some compliance that a standard would not. For example, when a driver in fast-­moving traffic is trying to execute a left turn across the path of an oncoming car, the two drivers prefer a legal requirement that leaves no uncertainty over which one of them has priority. Thus, the rule that “a turning driver gives way to a driver proceeding straight” is better than a standard of reasonableness or courteousness or anything else. When two neighbors wonder whether one of them has the right to cut branches overhanging his property, even though the trunk of the tree is on the other’s land, a simple rule is more likely to generate 235

The Expressive Powers of Law

compliance (not that it always will) than a complicated rule or a standard. This is not to say that a rule is always better than a standard when coordination is as stake, but only that, in these circumstances, the focal-­point function is one advantage of a rule to be balanced with everything else. Given the need to align expectations, the focal point may require not only rules, but rules that are simpler than is otherwise optimal. By contrast with a large, heterogeneous audience, the law sometimes addresses a smaller, specialized audience, with more background knowledge about the situation.4 Contract law can be an example, if the only parties affected by the contract are the two parties bound by it. Commercial contractors might also share substantial knowledge about the industry and industrial customs of which they are a part. In these cases, there might be a shared understanding of how competing standards apply, so the legal endorsement of one standard over another (whether mandatory or merely a default) might align expectations just as well as a rule. Yet as the prior paragraph indicated, there is a substantial domain where law addresses the general public and the rule’s focal point trumps that of the standard. The third implication concerns the costs of publicity. Any theory of compliance requires that people know the law, but the focal-­point theory requires a deeper knowledge. One needs to know or, at least, to have reason to believe, that others have the same legal knowledge. Thus, legal rules might work as focal points only if the state does something to give them publicity. Think of the amount of private publicity required each year to get everyone to switch to and from daylight savings time, an important form of coordination. Newspapers and websites offer much of that for free, but if one needed to duplicate the effort by paying for a publicity campaign, the costs would be considerable. For law, one option is to run public service announcements about new legislation or regulation. For traffic rules, the state requires licenses that are conditioned on knowing the rules of the road. For traffic rules, park regulations, and no-­smoking zones, among many other examples, the state sometimes posts the rules on a sign erected by the state. Sometimes the state forces others to bear the costs of publicizing the law, as when it requires restaurants or bars to post “smoking prohibited” signs. For some laws, the threat of legal liability creates corporate bureaucracies focused on legal compliance, some overseeing programs training employees in the relevant law of sexual harassment, customer privacy, antitrust, etc. 236

NOR M A T IVE I M P L I C A T IONS

All of these methods of publicity require resources that might be used in other ways. Thus, the third implication is the need to account for the costs of publicity in determining optimal enforcement. Other things equal, publicizing law will have greater returns in coordination situations that harness the law’s focal power than in noncoordination situations. In sum, the focal-­point power is not free. It may require forgoing the use of a standard or a more complex rule that would, in the circumstance, otherwise be optimal. Its use may require costly publicity. Yet the incremental costs of maximizing the focal-­point power may be worthwhile if it is less costly than the increment of conventional enforcement (through police or courts) that it replaces. Optimal enforcement requires weighing these costs and benefits in a particular context. What about the information theory? To a large extent, one cannot manipulate an individual law’s power to signal information. In a particular case, the law may not surprise citizens because they have already noticed the change in attitudes or assessments of risk. Or they may be particularly resistant to belief change that challenges their fundamental worldviews. Or people may change their beliefs, but the new beliefs may not motivate much compliance, if the costs of compliance are high relative to the disapproval or risk the compliance avoids. Recall from Chapter 5 that the informational effect is likely to be marginal in most cases, but to have discontinuous effects in some instances. Yet it will be difficult even to predict those discontinuous effects, much less to harness them. The information theory does, nonetheless, offer three general implications for enforcement policy. First, if a new law will signal attitudes, then it weighs in favor of making law at the local level. Recall from Chapter 5 that the attitudes that matter most to behavior are the local ones. There I claimed that local law is more strongly correlated than state or national law with the attitudes of the strangers and acquaintances who, given ­geographical proximity, are most likely to approve or disapprove of an ­individual’s behavior. Other things equal, local law has more power to influence behavior by signaling attitudes. The more powerful the informational effect, the less one must spend on conventional enforcement to achieve a given level of compliance. Thus, when deciding on the level at which to regulate, the greater attitudinal signaling power of local law is one factor (among many to be considered) that favors local regulation. For example, the fact that local smoking laws in the United States mostly 237

The Expressive Powers of Law

worked without the threat of legal sanctions does not mean that federal smoking laws would have worked expressively as well. Given the savings in enforcement costs, it may have been preferable to regulate smokers through a series of local laws. Second, recall also the Chapter 5 implication that the strength of the informational signal—​the amount of updating in beliefs the law causes—​is affected by how much the public believes the legislature producing the law is captured by “special interests.” That is, the more the public believes that the legislature represents the general public, the stronger the inference that new legislation reflects new social attitudes. Conversely, the more the public believes that lobbying groups or elite insiders run the legislature, the weaker the signal (unless the concentrated groups were uniformly against enactment of the new law, in which case its signal is stronger). To the extent that any policy—​e.g., term limits, campaign finance reform, restricting lobbying by former legislators—​can favorably affect public perceptions of legislatures, we should list as one of the benefits of such a policy a general increase in the expressive power of law. Interestingly, this normative point looks much like a recommendation of legitimacy theory—​the greater the perception of legitimacy, the greater the deference to the law. But the implication arises simply from the logic of signaling attitudes. One might make the same point about risk signaling and legislative expertise. To the extent that the public perceives legislators as having average or below average understanding of or concern for scientific issues, the signal of risk carried in legislation will be weaker. To the extent they perceive the technical committees of the legislature as being populated by those with genuine expertise, they will perceive the law as being more reflective of the state of scientific knowledge and will update their beliefs to a greater degree. Finally, the information theory identifies a certain kind of unintended consequence, what Dov Fox and Christopher Griffin call an “expressive externality.”5 As reported in Chapter 5, they claim that the Americans with Disabilities Act caused an increase in the abortion of fetuses identified with Down syndrome, by conveying to potential parents that children with disabilities lead a difficult life of struggle and discrimination. The information conveyed is thus partly about attitudes (of intolerance) and partly about risks (of Down syndrome). Whatever the merits of Fox and Griffin’s empirical claim, the point is that law may signal risks in ways that encourage behaviors other than simple compliance. 238

NOR M A T IVE I M P L I C A T IONS

Unwanted informational signaling is a key worry of drug legalization or decriminalization. The legislators who move in this direction may be motivated by the costs and failure of the drug war, not a newfound belief in the safety or public approval of addictive recreational drugs. Nonetheless, many members of the public may interpret legalization as revealing the latter message. If drugs appear safer and more popular, drug use will rise more than is explained simply by the fact that the users no longer face the risk of prison. One would like the option of ramping down the drug war without seeming to endorse drug use. One implication here is the importance of structuring any legal change to minimize the unwanted message. There are a great many policy choices available for ramping down the drug war, but it might actually matter to their effect if everyone labels them “legalization,” “decriminalization,” or something else. Portugal decriminalized the possession of small quantities of all drugs several years ago, transforming the offense into an administrative violation. At the same time, Portugal created the “dissuasion commission” with the power to inflict noncriminal sanctions.6 A Commission for the Dissuasion of Drug Addiction—​consisting of a psychiatrist, a social worker, and an attorney—​will interview an individual found in possession of personal use quantities of the drugs and may choose not to impose civil penalties if the violator seeks drug counseling or meets other conditions. There is some evidence that these commissions rarely impose any actual administrative penalty even on those who fail to get counseling.7 Whatever is the actual degree of sanctioning, one could imagine a purely expressive purpose for creating the appearance of compelled counseling, even if it is not commonly used, which is to blunt the unwanted infor­ ­mational message from decriminalization—​that drugs are safe. Government officials could just announce “drugs are dangerous” at the time of decriminalization, but the dissuasion commissions are likely to be more expressively effective: that they are costly for the government to staff makes the message more credible, a stronger signal of legislative beliefs; that they exist over time resends the message, making the audience more likely to receive it; and that they meet with individual drug possessors targets the message to the most relevant population. Now I turn to a different normative issue, not the efficient use of law’s expressive powers, but the wisdom of using conventional meaning in legal doctrine. 239

The Expressive Powers of Law

Expressive Legal Doctrine: A Critique of Conventional Meaning In Chapters 1 and 6, I identified three prominent legal doctrines that incorporate an expressive element: the Establishment Clause concern with government actions endorsing religion; the Equal Protection Clause scrutiny of racial classifications in part for the messages they send; and the need to distinguish criminal punishment from civil penalties under the Ex Post Facto clause, in part by whether the government action expresses punitive attitudes. None of these doctrines is entirely reducible to an expressive matter, but all have an important domain of application that turns on the meaning of a government action. Commentators debate whether courts should expand, contract, or eliminate these expressive domains. The coordination and information theories do not offer definitive resolutions of these doctrinal issues. One should not expect as much of a causal theory; doctrinal choice is normative. Nonetheless, the analysis of the prior chapters can shed some light on the issues, rendering some normative claims more plausible and others less so. In this section, I discuss the distinction in the sources of meaning and argue against the doctrinal relevance of conventional meaning. Sources of Meaning in Three Legal Doctrines In Chapter 1, I argued for the need to distinguish explicitly between different perspectives or sources of meaning, which I simplified as speaker’s meaning (first party/intended meaning), audience meaning (second party/ perceived meaning), and sentence meaning (third party/conventional meaning). I have so far emphasized the second party perspective, judging meaning by the uptake of the audience. This perspective inevitably follows from my focus on behavior. The mechanisms of behavioral change—​ coordination and information—​depend on what the law means to the people who seek coordination or information, not on intended or conventional meaning, if those differ. These distinctions in perspective matter greatly to normative analysis. Deborah Hellman makes this point as explicitly as anyone. She identifies the wrong of discrimination (private and governmental) from the meaning of the discriminatory action.8 According to her theory, discrimination is wrongful when it demeans an individual or group, when it expresses denigration in a way that has the power “to put the other down.”9 But how is this meaning to be judged? What perspective is determinative? 240

NOR M A T IVE I M P L I C A T IONS

Hellman identifies the same choices I have discussed: “Meaning can come from three sources: the intent of the speaker, the perception or understanding of the listener, and the context in which the ‘utterance’ is made.”10 She argues that meaning should be judged not by the intent of the discriminator, nor by the uptake of anyone who perceives the discrimination, including the target.11 She thus rejects current legal doctrine that judges whether government action receives the strictest scrutiny under the Equal Protection Clause of the Fourteenth Amendment by whether the government actor (roughly speaking) intended to discriminate (on the basis of race or some other impermissible factor).12 And she rejects the standard doctrinal alternative in the legal literature, which asks whether the governmental action (roughly speaking) burdens a historically disadvantaged racial group.13 Instead, Hellman proposes to determine whether an action is “objectively” demeaning by using “context,” or what she more commonly calls convention. One constructs conventional meaning through “a complex inter­­pretive judgment” under “ideal epistemic conditions.”14 Such an interpretation is sensitive to the way that the history of a practice shapes its meaning, looks for appropriate analogies, and does not “reduce objective truth to mere majoritarianism.” Indeed, Hellman says her approach at constructing meaning “allow[s] for the possibility that everyone is wrong (about whether a given practice demeans).”15 From this description of expert practices of interpretation, I think it is fair to say that Hellman would allow, if not demand, the kind of esoteric knowledge that I have argued is not appropriate for ascertaining audience meaning (at least when the audience is large). In searching for the best analogy between a challenged action and a prior practice in the history of the same culture, it would not appear to matter, for example, how many people were aware of the prior practice. In this respect, Hellman’s approach is similar to Joel Feinberg’s expressive theory of punishment, discussed in Chapter 6. Important procedural rules turn on whether a government action constitutes punishment. A recurring issue is the interpretation of the Ex Post Facto clause of the federal constitution, which prohibits legislation from retroactively authorizing criminal punishment but places no such limits on retroactive civil regulation or penalties.16 For example, Feinberg refers to the 1960 Supreme Court case, Flemming v. Nestor,17 where the Court held that depriving someone of social security benefits was not punishment, so the deprivation could be 241

The Expressive Powers of Law

imposed retroactively. The Supreme Court has similarly decided that it is not punishment to deny an ex-­felon the right to practice medicine, nor to deport an immigrant, nor to require sex offenders to register their presence with local police.18 Feinberg defines punishment as a subset of the hard treatment imposed by the state on account of a rule violation: only such hard treatment that expresses “attitudes of resentment and indignation and judgments of disapproval and reprobation,” more briefly, “punitive attitudes.”19 Punishment thus requires public, symbolic condemnation. Again, the question arises: From whose perspective are we to judge whether government action expresses punitive attitudes and condemns? Feinberg is less clear than Hellman in distinguishing the alternatives, but he appears to give the same answer. He draws the civil-­criminal line in cases of this kind by asking whether the government action “is a conventional device for the expression of” the punitive attitudes.20 On this account, the denial of the right to practice medicine is not a conventional means of expressing condemnation because the government has not generally used this denial to condemn (but merely to screen out those lacking medical competence). Using an example of hard treatment self-­imposed for psychological or religious reasons, Feinberg says at one point: “Even floggings and imposed fastings do not constitute punishment . . . ​where social conventions are such that they do not express public censure.”21 Feinberg’s formulation might refer to audience meaning, if one imagines that the social conventions matter only as empirical evidence of audience uptake. But one could read him as following Hellman in focusing on conventional meaning, including the expert interpretation. He never explicitly prefers audience (or speaker’s) meaning. This issue has been discussed repeatedly in the context of the Estab­ lishment Clause, where Justice O’Connor and others have applied a doctrinal test forbidding the government from “endorsing” religion.22 Asking whether a crèche on government property, for example, endorses religion, requires that one specify meaning in the usual ways: the speaker’s intended meaning, the audience’s perceived meaning, or the constructed conventional meaning. (The context helpfully reminds us that audience meaning need not be monolithic because, when the government speaks to a large population, different groups will take different meanings). The endorsement cases are not particularly consistent in the selection of a source of meaning. Academics have dissected statements in the 242

NOR M A T IVE I M P L I C A T IONS

Establishment Clause cases that point to a concern for the perceptions of actual human beings—​audience meaning—​and a concern for the “objective” or “reasonable” perceptions of meaning.23 Justice Kennedy stated in a recent case that the relevant “test requires the hypothetical construct of an objective observer who knows all of the pertinent facts and circumstances surrounding the symbol and its placement.”24 That certainly sounds like it might be stating a third-­party perspective, constructed from convention, though it might be that the Court has other substantive concerns, but is grappling to come up with a workable test to achieve them.25 Across the three doctrines—​Equal Protection, Ex Post Facto, and Establishment—​is repeated the same normative issue: once we decide that a legal distinction depends on what the government is expressing, should we focus on what the government intended to express, what some audience (presumably, a sizeable group, though not necessarily the entire population) actually understood to be expressed, or what would hypothetically be expressed if we rigorously (with expert knowledge) constructed meaning form the relevant conventions? Or some conjunctive or disjunctive combination of two or three of these? A Critique of Conventional Meaning I seek to intervene in these debates in a highly circumscribed manner. I defer the threshold normative question of whether we should adopt an expressive element in each of these domains (until the next section in the case of one doctrine). I do not contend that there is a right answer for all contexts; the right source of meaning for defining punishment need not be the right one for the Establishment Clause. Nor do I offer to resolve the issue definitively. My point is only this critique: Whatever the choice between speaker and audience meanings and other issues, it is an error to use conventional meaning as the substantive normative concern of these doctrines, as Hellman, Feinberg, and others propose. I distinguish the legal doctrine’s substantive normative concern from its merely evidentiary concern. One might concede that the actual normative concern is tied to speaker’s meaning or audience meaning, but prefer for pragmatic reasons to have courts focus on conventional meaning. Perhaps conventional meaning is more tractable; a direct focus on audience meaning might encourage bad-­faith assertions of idiosyncratic perceptions, while the instrumental use of conventional meaning might ordinarily capture audience meaning. Yet there is nothing of this kind of 243

The Expressive Powers of Law

argument in Hellman or Feinberg. They advance the conventional meaning as the “right answer” to the problem, not a second-­best solution in a world of limited information.26 We are entitled to test their claims by assuming we have all the information we need to determine the meanings the different sources provide. Of course, the selection of a source of meaning is entirely unnecessary if the meaning is identical under all three sources, as it frequently is. A intends to express proposition X, uses the conventional means of expressing X, and the audience understands A to be expressing X. The frequency of this correlation is what allows us to use one meaning as evidence of another meaning. The selection of a source matters, however, because these meanings can diverge; different types of bounded rationality or local dialects allow for slippage. A intends to express proposition X, ineptly or idiosyncratically uses the conventional means of expressing proposition Y, and the audience ineptly or idiosyncratically understands A to be expressing Z. If A is the government and we are deciding whether the government has, by some action, expressed punitive attitudes, the inferiority of a racial group, or the endorsement of religion, the question is which source of meaning governs when they diverge. We are entitled to test Hellman and Feinberg’s claims based on this situation of divergence. So consider the case where the conventional perspective diverges from both speaker’s meaning and audience meaning, which coincide with each other. If the speaker intended to convey X and the audience interprets the message as conveying X, why would any legal doctrine be ultimately concerned with a divergent conventional message Y? There are plausible moral theories for why individuals or government should not be allowed to act for certain purposes, which would rule out symbolic action with certain intended expressions. There are plausible consequential theories that are ultimately concerned with the effects of expressions, which would depend on audience meaning. When conventional meaning overlaps with these illicit speaker’s or audience meaning, they will be objectionable just because of the overlap. But should we ever focus on conventional meaning that is disconnected from actual speakers and actual audiences? My answer is no. To illustrate my concern, suppose two non-­native and deficient speakers of English, A and B, are groping to use the language. Via private email, A accidentally calls B a “fiend,” with the intent of expressing “friend,” which B happily misinterprets as meaning “friend.” From the conventional perspective, A has insulted B, but it is mysterious 244

NOR M A T IVE I M P L I C A T IONS

why anyone would draw any moral conclusions from that fact, given that the intent and effect are not insulting. By accident, A and B have communicated effectively and unobjectionably, just unconventionally. Consider the critique in the context of one of Hellman’s examples, in which the Food and Drug Administration (FDA) approves a pharmaceutical specifically for use by African-­Americans.27 Is this government racial classification wrong? Hellman argues it does not, as a matter of convention, carry the meaning of denigration and is therefore permissible, but she acknowledges the reasonableness of contrary arguments, based on the implied message that African-­Americans are genetically different from Caucasian-­Americans, which in light of history could make the FDA decision denigrating. As Hellman maintains, we don’t have to accept her analysis of conventional meaning in a particular case in order to accept the centrality of conventional meaning for moral analysis. For purposes of my argument, assume the conventional meaning of the FDA approval might come out either way. Now imagine these two cases: (1) the conventional meaning is unobjectionable (Hellman’s specific argument is right), but both the FDA’s intended meaning and the widespread perception are that the FDA distinction denigrates African-­Americans; (2) the conventional meaning is one of racial denigration (Hellman’s specific argument is wrong), but both the FDA’s intended meaning and the widespread perception are not one of racial insult but respect. Suppose that the meanings diverge in each case because of expert knowledge of the history of race and medicine in the United States. Recall from Chapter 6 the discussion of Jeannie Suk’s interpretation of a Justice Souter opinion. Suk brings to bear a nineteenth century etiquette manual, knowledge of which opened up interpretations of Souter that would not otherwise be available. Similarly, suppose there were some esoteric aspects of FDA history that supported a conventional meaning that diverged in either direction from speaker’s or audience mean­ ­ing. Given Hellman’s preference for conventional meaning that is untethered from majoritarian understanding, the best conventional meaning could be one that depends on esoteric information of this sort. Yet the implications are not attractive. If Equal Protection doctrine focused ultimately on conventional meaning, then in case (1), there should be no objection to a racial classification that government actors selected for the purpose of denigrating African-­Americans even though it successfully communicated denigration. Because ordinary people sometimes 245

The Expressive Powers of Law

­ redictably interpret things unconventionally, the doctrine would essenp tially create a loophole in which government actors were free to send messages with racially denigrating intent and effects so long as they “coded” them in an innovative and unconventional way. Conversely, if conventional meaning is the touchstone, then in case (2) there would be an objection to racial classification government officials selected without intending to insult and that produced no perception of insult. We would presumably not allow the FDA to approve drugs along racial lines even if there was no intended or perceived expressive harm. Hellman supplies no convincing argument for these odd results. She makes the plausible deontological claim that an action may be wrong regardless of consequences, so discrimination can be wrong regardless of audience uptake. She makes the plausible observation that expressive differentiation can demean despite the benign intent of the speaker, so discrimination can be wrong regardless of speaker’s meaning. While each of these arguments is individually reasonable, the combination—​that neither intent nor effects matter—​is not. The intuition behind the wrong of demeaning action arises entirely from either of these sources. When there is no effect because no one perceives the denigration, we intuitively grasp that a person can still be wrong for what they intend (and attempt) to cause. When there is no intent to demean, we intuitively grasp that that an expression can still be wrong in the sense that it causes harm. With neither harm nor intent, the remaining concern, however, is a bit surreal, as if we should expend government resources to detect communicative “wrongs” that must be elaborately constructed to be perceived, and that would otherwise go unnoticed by everyone. It might seem that I am being unfair because courts are often interested in “objective” meaning, as what a hypothetical reasonable person would take to be the meaning. But judicial doctrine may be justified pragmatically. Conventional meaning may be good evidence of speaker’s meaning or audience uptake, possibly more available to the court than direct evidence of the latter two. In particular, when there is more than one relevant audience to a government action, there may be a concern that one audience has an incentive to exaggerate or distort the meaning it actually perceives, as a means of strengthening its argument against the government action (say, it endorses a religion or denigrates a race). An inquiry into conventional meaning might be a useful way to constrain extravagant and untestable claims of audience meaning. But as I said before, Hellman is 246

NOR M A T IVE I M P L I C A T IONS

not making a pragmatic claim. She argues for conventional meaning as the right answer for deciding these matters, even if we had no evidentiary concerns, but my critique shows that to be a mistake. The same is true of Feinberg, but here there is a separate concern. I took for granted that Hellman’s account of the morality of discrimination could serve as the basis for Equal Protection doctrine (or bans on private discrimination). Although there are counterarguments, one might plausibly want to craft the doctrine to prohibit government action that expresses things about race (or other characteristics or groups) that are wrong to express, an affront to equality. Yet when Feinberg addresses the Ex Post Facto clause, his account suffers from the fact that he offers no political theory of the permissibility of retroactive regulation. He seems to assume that, if any legal doctrine contains a civil-­criminal line, his analysis of criminal punishment applies. More likely, I would think, one should start with a theory of the political dangers of retroactive law and then derive from the theory a line between permissible and impermissible retroactivity, which might or might not track the civil-­criminal distinction. The political dangers might also identify the type of meaning we should use for judging whether the government is engaged in punishment, or whether we should care about meaning at all. Let us put that issue aside and simply accept, arguendo, that the Ex Post Facto clause should apply if and only if the government is retroactively punishing according to Feinberg’s definition—​inflicting hard treatment and condemnation on account of a rule violation. Now we must decide whether to determine the existence of condemnation—​the expression of punitive attitudes—​by (1) the government’s intent to express punitive attitudes, (2) some audience’s perception that government has expressed punitive attitudes, or (3) the conventions for expressing punitive attitudes. Again, on its face, (3) is implausible or at least it has nothing obvious going for it. Suppose that the government requires sex offenders to register with local police and have their photo and crimes publicly listed online. Suppose we have it on good authority that, in requiring registration, the government (1) acted with the intent to express punitive attitudes of the sex offenders and (2) caused the public to perceive the government to have expressed such attitudes. If it turns out that registration is not a conventional means of expressing the punitive attitudes (plausible when sex registration was new because registration is also used for noncriminal regulatory matters), why would the conventional meaning control? The 247

The Expressive Powers of Law

result again looks like a loophole. If the government can successfully innovate, finding a coded language that everyone understands as expressing punishment, just as intended, then it can avoid the ban on Ex Post Facto legislation. Or suppose the reverse: The government does not intend to condemn an alien it deports for illegally entering the United States, and the public does not perceive condemnation, but the deportation is a conventional means of expressing punitive attitudes. The latter might be true because the conventional view uses expert knowledge and that might include the fact that deportation is so similar to banishment (at least if applied to an alien who lived for a long time in the place from which he or she is deported), which was, certainly at one time, a conventional way to express punitive attitudes.28 Again, whatever the purpose of prohibiting retroactive punishment, if punishment requires condemnation, it seems odd to count government action as punishment when it lacks the intent to condemn and fails to create the actual perception of condemnation. Without rehearsing the arguments, I would make the same point about the Establishment Clause’s endorsement test: the government’s intended meaning and the audience uptake seem worthy of consideration, but the conventional meaning should not be a substantive doctrinal concern. In all of these cases, I have offered only a critique, a doctrinal path to avoid. I leave open the many remaining questions, including whether to have an expressive element to the doctrine in the first place (addressed below for one doctrine) and, if so, whether to make speaker’s meaning or audience meaning the ultimate doctrinal concern or some (conjunctive or disjunctive) combination of the two. I also leave open the question of which audience to count when there is more than one, an issue particularly common in Establishment Clause cases. There is not likely to be any general answer to that question; it must be addressed based on the reasons for having an expressive element to the doctrine.

The Costs of Symbolic Conflict: Implications for Constitutional Law As noted in Chapter 1, the information theory predicts more “symbolic” struggle in politics. Even in the absence of sanctions and legitimacy, the law’s expressive power allows it to affect behavior. The more likely it is that law influences behavior, the more that is at stake in the law. 248

NOR M A T IVE I M P L I C A T IONS

For example, legislation restricting abortion signals stronger attitudes against abortion.29 As people update their beliefs, they perceive higher social costs to abortion. An individual expects that more people in her community will ostracize her if they discover that she had an abortion or even if she supports abortion rights. Expecting greater costs, she is slightly less likely to get an abortion, less likely to identify herself as pro-­choice, perhaps even more likely to rationalize her true views to conform to the social consensus she observes. If she and others merely speak out less often, then their silence strengthens the perception of a consensus against abortion, further raising the perceived costs of the procedure or of speaking in favor of such rights. As a result, the political stakes for abortion regulation are higher than they would be if the law lacked expressive power. The law might appear to be wholly symbolic, as might a ban on a method of abortion that does not prevent a single abortion because other methods remain available; yet the law’s ability to affect behavior means it is rarely just symbolic.30 I do not suggest that these behavioral effects are necessary for the existence of political controversy over legislation regarded as symbolic. Cass Sunstein once observed the irony in the political and symbolic struggle over the attempt to prohibit flag burning as a form of desecration: The law was more likely to increase than decrease the behavior it condemned.31 First, there was and is very little flag burning even though it is legal (protected as free speech), so it is difficult to decrease the frequency of a behavior that is already extremely rare. Second, the prohibition would have made the issue of flag burning a more salient means of protest and attracted violators who sought to engage in an act of civil disobedience to bring attention to their protest.32 Yet many of those who favored the constitutional amendment banning flag burning would probably not relent even if they accepted the truth of Sunstein’s prediction. Proponents want the law to express their opposition to flag burning, and thereby to express special reverence for the flag, regardless of the actual consequences. An expressive-­politics theory of law predicts this kind of outcome: the law gets support because of its patriotic meaning for those who support it. That law expresses that the patriots who revere the flag are in the right and the disloyal protesters who dishonor it are in the wrong. Thus, I want to concede the possibility that some conflicts really are entirely symbolic. Behavior is unaffected. My point is that it is more 249

The Expressive Powers of Law

common that the symbolic struggle coincides with a struggle over the informational influence on behavior. Expressive effects intensify political conflict, even if the conflict would continue without them. Even in Sunstein’s example, behavioral influences are plausibly at work. A constitutional amendment and legislative ban on flag burning would communicate public attitudes of reverence for the American flag and for the patriotism that its display represents. At least for those not inclined to burn the flag in protest, which is presumably the vast majority of Americans, the effect is to change beliefs about the informal social costs one will bear from flag desecration and the informal social benefits one will gain from the patriotic display of the flag. After updating beliefs, given that most people value approval, most will be more inclined to engage in patriotic behaviors, to show respect to public flags, to fly the flag at home, or perhaps to wear a flag lapel pin. As more people display the flag, the choice not to display it seems more unpatriotic, and thus socially costly. At the extreme, nationalist conformity reaches a point where the world consists of dissidents willing to burn the flag and everyone else, who patriotically display it.33 This is not a likely result of a flag-­burning amendment, but the possibility does show that even Sunstein’s example need not be purely symbolic. Proponents could contemplate that the amendment would push modal behavior towards greater reverence for the flag’s patriotic display, even if it did not decrease flag burning. None of this is itself normative, but the fact that law’s expressive power can create behavioral stakes to ostensibly symbolic political conflict does have an important normative implication. We should be concerned to limit the problem of expressive rent-seeking. Ordinary, material rent-­seeking occurs in legislation where the law does not create greater wealth but simply redistributes it in a way that is not justified by a distributional theory. Most commonly, an interest group with few members finds it easy to organize to extract wealth from the rest of the public who find it difficult to organize. So we get corn subsidies. And we get the resources wasted in the process of creating (or attempting to prevent) rent-­seeking. Of course, in any given case, there may be disagreement over whether rent-­ seeking exists based on disagreement over arguments that the redistribution promotes social welfare.34 By expressive rent-­seeking, I refer to the effort of social groups to extract some expressive benefit from the government that is not justified by a theory of social welfare or distributive justice. The “benefit” to one group 250

NOR M A T IVE I M P L I C A T IONS

may be the expressive harm of a competing social group. Again, if the expression is not justified, the resources committed to bringing it about (or attempting to prevent it) are, from a social perspective, wasted. The problem is that the law’s expressive effect on behavior makes the expressive benefit all the more valuable, attracting even more political resources into sterile legal expression. As with material rent-­seeking, it may be difficult to identify expressive rent-­seeking because there are always potential normative arguments in favor of the legal change. Yet a social group’s use of resources in a competition for expressive dominance is, when identified as such, socially undesirable. So is the diversion of scarce resources from the creation of more substantive legislation. For example, in American death penalty law, it is necessary to specify “aggravating circumstances” of a murder, the existence of which permits the imposition of the sentence of death. Identifying the worst murder, “the worst of the worst” crime, is a notoriously difficult task. Yet one political reality complicating the task was the status competition created by such a list.35 Once the killing of a police officer or judge was an aggravating factor, something that could justify imposing society’s most severe sanction, other groups sought to validate their importance by getting themselves listed as well. In some states, killing a senior citizen or a teacher is an aggravating circumstance. “By singling out these occupational and social groups for special protection, the state accords them a measure of social esteem that is not accorded to other victims.”36 Ideally, one would want to end expressive rent-­seeking. It is, however, quite difficult to constrain competition for the allocation of expressive harms and benefits. Substantive policies will expressively favor some groups over others, as Prohibition favored the older Protestant immigrants to the United States over the newer Catholic ones. As the work on cultural cognition shows, the regulation of guns, abortion, and carbon emissions, all favor some groups over others.37 It is impractical to shackle the government from making substantive law just because the law has expressive effects, even if those effects cause political groups to invest heavily in lobbying and counter-­lobbying for legislative victory. One might try to focus on and limit “purely” symbolic lawmaking. Yet this too is impractical. Governments name public streets, parks, schools, post offices, and military bases. Governments fly flags, recognize official holidays, and make resolutions and proclamations. Naming places and holidays to honor people and events will, inevitably, create a symbol 251

The Expressive Powers of Law

favoring some groups over others. If this is true of Cesar Chavez Day and National Cowboy Day, consider how much more potent are the symbols of Confederate Memorial Day, celebrated in several southern states, or federal military bases named after Southern slave-­owning Confederate generals.38 Yet it seems too costly to prevent the competition for this symbolic rent-­seeking. It would be difficult to construct a standard of neutrality to judge such matters, and it would require enormous judicial resources to enforce. Nonetheless, we might read the United States Constitution as attempting to limit expressive competition in certain categories of government action that are not only likely to constitute expressive rent-­seeking, but are also, in the light of history, most likely to be destructive. This is a reasonable interpretation of some of the Supreme Court’s doctrine enforcing the Establishment Clause of the first amendment, as well as the Equal Pro­­ tection Clause of the fourteenth. Neither of these clauses is only about expression. The Establishment Clause puts some limits on governmental funding of religious schools, which does not require a finding that the funding express support for religion.39 Similarly, Equal Protection doctrine has a strong anti-­classification principle—​that legal classifications based on race must be given “strict scrutiny,” which is usually fatal.40 Michael Dorf notes that one might justify the aversion to racial classifications based on what they express—​the importance of race.41 But classifications retain an independent importance; they are suspect regardless of what they express in a particular case. Instead, we might see that these constitutional provisions have multiple purposes, where one purpose is to regulate expressive competition. What is difficult, however, is to draw the line between objectionable and nonobjectionable expressions. Once we say there is any expressive component to the Establishment and Equal Protection clauses, why aren’t many more expressions rendered unconstitutional? This is the challenging question Dorf raises. He frames the matter this way: The baseline for government expression is a presumption in favor of the government expressing whatever it wants. Across several constitutional law doctrines, “American law sometimes recognizes expressive harms as real harms but more commonly it does not.”42 We often don’t notice the baseline cases where government is permitted to harm individuals expressively, as a means of advancing some governmental end. Dorf gives the example of smokers, who may 252

NOR M A T IVE I M P L I C A T IONS

reasonably feel insulted or stigmatized by anti-­smoking messages emphasizing their negative qualities. The same is obviously true of criminal punishment, as by stigmatizing users of certain recreational drugs but not others, but also many other government expressions. Teaching the theory of biological evolution but not creationism insults creationists. “Stay in school” ads denigrate high school dropouts. Taxes and elections expresses that political anarchists are marginalized outsiders. That there are good reasons justifying these expressions does not deny their harmfulness to certain groups. As Dorf puts it, “[b]eing incidentally insulted or otherwise harmed by government speech . . . ​is just a part of the price each of us potentially pays for having an effective government.”43 Put differently, even though many governmental expressions are insulting and, indeed, even if the political action that produced them was aimed at raising the status of some social groups by lowering the status of others, we cannot afford to apply significant scrutiny to most government expressions. In a democracy, such judicial regulation would excessively constrain legislative and executive action (not to mention the demand on judicial resources to police government expression comprehensively). Against this baseline, there are some isolated instances where the courts recognize an individual’s claim that the government has harmed their dignitary interests by expression. Which ones should we recognize? I briefly argue that the expressive effects of law can shed a light on what is special about some of the cases. Constitutional theory is dominated by nonconsequential concerns, which includes much of the discussion of expressive harms (based on conventional meaning), but here I proffer a consequentialist explanation (based on audience meaning). The most obvious illustration would be the concern of the Equal Protec­ ­tion Clause with government actions that express racial hierarchy and stigmatization, a subject well-­mined in a literature interpreting the clause as stating an anti-­caste principle.44 The information theory supports this literature in showing how the expression of racial inequality could exacerbate racial status competition and stratification.45 I will instead illustrate the kind of consequentialist normative analysis I have in mind by focusing on the Establishment Clause, which might also be interpreted to further the anti-­caste principle.46 What follows is a purely functionalist rather than an originalist account of the expressive element of this provision. I do 253

The Expressive Powers of Law

not claim that this expressive concern motivated the Founders to adopt the Establishment Clause. In her Lynch v. Donnelly concurrence, Justice O’Connor originated what came to be known as the “endorsement test,” which invalidates government endorsement of religion, though its current status in the Supreme Court is uncertain.47 Her argument against endorsement (first quoted in Chapter 6) is often repeated by courts: “Endorsement [of religion] sends a message to nonadherents that they are outsiders, not full members of the political community, and an accompanying message to adherents that they are insiders, favored members of the political community. Disapproval sends the opposite message.”48 The claim is that endorsement of religion is harmful. What exactly is the harm? One interpretation of this passage is that the harm of endorsement is the negative feeling it creates in nonadherents. When the government endorses a religion (or religion generally), individuals who do not belong to the endorsed religion (or any religion) are made to feel badly about themselves and their place in the political community. Recall from Chapter 1 my categories of expressive literature about law. Although I said I would focus on behavior, the expressive consequences of law also include changes in beliefs and emotions. The problem of governmental endorsement of religion might be only these latter consequences—​beliefs in outsider status and the negative emotions they create. From my observation, this is a common understanding of O’Conner’s famous passage. If so, it is unpersuasive. The world is full of reminders that voters who are persistently in the minority are political “outsiders.” In a reliably blue (red) electoral district, where only liberal (conservative) politicians are elected, almost every governmental action is a “message” to conservatives (liberals) that they are outsiders. As stated, the government intentionally expresses the outsider status of smokers, creationists, and anarchists. Those who want to return to the gold standard or legalize heroin cannot help but infer from any election that they are outsiders whose views are not even taken seriously by the majority. Pacifists must look at government war memorials, war-­related national holidays, and government advertisements for military enlistments as reminders that they are outsiders. And these are just the messages government sends. Pacifists surely feel like outsiders because they live in a culture that glorifies war and treats violence as entertainment. Many religious minorities are, despite the endorsement test, intensely aware of their outsider status. The one non-­Christian in a small town does 254

NOR M A T IVE I M P L I C A T IONS

not doubt that the government officials and employees in the town are Christian. If there are Christmas decorations on every house and private business, the fact that there are none on the courthouse is not a very convincing refutation of one’s outside status, especially if one knows that it was removed only at the command of judicial elites outside of one’s community. Perhaps the forced absence of Christian symbols on government property uniquely demonstrates that these religious minorities “count” in the polity. Yet, as an empirical matter, this symbolic regulation might not be necessary—​if minorities already feel that (among other things) the right to vote and freely practice their religion demonstrates that they “count”—​ or sufficient—​if the fact that all of one’s elected representatives are practicing Christians makes non-­Christians feel that they don’t “count.” It seems unlikely that the endorsement test is both necessary and sufficient to prevent religious minorities from feeling like outsiders. Put differently, if the expressive component of the Establishment Clause is designed to shield people from the unhappy news that they are political outsiders, in order to make them feel better, the implication for the clause seems radical. Why do we accept the motto “In God We Trust” on the currency, the existence of a chaplain for the House of Representatives, the governmental declaration of a National Day of Prayer, or the fact that Christmas but no other sacred religious day is a national holiday? Why do we accept that federal politicians of a Christian denomination can gather for an annual prayer breakfast in the nation’s Capitol? These are notoriously difficult questions for many theories of the Establishment Clause, so perhaps the answer is that we are imperfectly committed to the clause, whatever its purpose, so Justice O’Conner’s justification could still be correct.49 Still, if we interpret the function of the clause as one of preserving religious minorities from the mere feeling that they are outsiders, the problem is not just that we don’t perfectly achieve that result but that we don’t come anywhere close to it, and we probably couldn’t come anywhere close to it without radically reorganizing our society. Predictably, my claim is that there is a second and better interpretation of O’Conner’s rationale, one that focuses on behavior. The point is not simply that religious minorities feel badly, but that the message of their outsider status has the capacity to change behavior towards an unhealthy religious conformity, one that threatens creating a kind of religious caste. The message of insider/outsider difference threatens, as Christopher Eisgruber puts it, to “create or reinforce the division of American society 255

The Expressive Powers of Law

along lines defined by religious differences.”50 The argument here is akin to the argument just stated about flag burning and patriotism. In brief, the argument is this: Governmental endorsement of religions signals attitudes. Attitudinal signaling changes beliefs about the informal social costs and benefits of public religious practices. Those changed beliefs can produce greater religious conformity. As John Stuart Mill contended, social conformity can be as stifling of liberty as “the tyranny of the magistrate,” or even more so, since private social sanctions can reach into more details of life.51 Historically, religious liberty has been particularly vulnerable to factionalization and social pressure to conform to the religious norms of their community. The Establishment Clause cannot guarantee the social freedom of religion, but it can at least block one mechanism for creating the zealous norms that threaten social freedom, which is a majority using government to signal its religious attitudes. To illustrate, let’s say that a city puts a cross on its letterhead, along with the motto “Jesus saves,” so that anyone receiving a letter from the city sees that the local government is unambiguously endorsing Christianity. The endorsement signals stronger attitudes of approval of Christian beliefs and disapproval of non-­Christian beliefs, a change in the numbers or intensity of those attitudes. People update their beliefs. Now the social costs of appearing to be non-­Christian in this community are a bit higher, so a few more Christian believers who heretofore were private about their religion decide to wear a cross and put a Christian fish symbol on their car. A few more non-­Christians decide to conceal their nonconforming religions, by not wearing a yarmulke or hijab, or even to pretend to be Christian. Now that there are fewer people who display no sign of Christianity, the pressure on the remaining nonconformers increases. As a result, another round of conformity occurs, more people display Christian symbols, and the nonconformers become an even smaller minority. What I said of the flag pin could be true of the cross or any other outward marker of religiosity: Social pressure to conform could produce an extreme outcome where the failure to wear a cross in public becomes proof that one is not a Christian. At this point, when even atheists sometimes wear crosses and the ones that don’t are treated as pariahs, we have reached Mill’s tyranny of the social custom. To be clear, the probability that the single governmental endorsement leads to a cascade of this sort is very low. But if one endorsement is acceptable, then so is a second and a third. The courts won’t likely know in 256

NOR M A T IVE I M P L I C A T IONS

advance whether they are dealing with the rare case where the norm of religious tolerance and private observance is on the brink of unraveling. So the logic of the endorsement prong of the Establishment Clause is to ban most of them (or at least the ones courts can reach fairly easily). The theory reveals why the clause is so controversial. We already tolerate a fair amount of government endorsement of religion and quite possibly we could have substantially more without tipping into a norm of zealous conformity, so any judicial intervention looks both like too little and too much: too little because the courts allow the many endorsements mentioned and too much because it is unlikely that any single additional endorsement in a given place would change the social norms of religiosity. Yet if courts do not know enough to predict such things, then they may be justified in groping towards some limit to religious endorsements, rather than no limit, and rather than attempting the politically impossible task of eliminating them all. The inevitable political controversy of a prophylactic restriction on religious endorsements justifies the need for a constitutionally entrenched rule against them. The information signaling theory also addresses some subtopics of controversy in this area. One is the judicial distinction between old and new endorsements. Why should courts more freely permit old governmental actions like putting “In God We Trust” on the money? The expressive-­ effects theory gives us a reason to care less about historic traditions of this sort. Recall from Chapter 5 the implication that the expressive significance of a law decays over time; it works as a signal of the attitudes at the time that the law passed. Old expressions have already been incorporated into the beliefs of adults who have resided in the United States for a long time. And even for new immigrants, an old tradition communicates less about current attitudes than a new governmental endorsement. One could point to the failure to change the tradition, but legislative inertia obviously makes it costly to change the status quo, which is why all kinds of laws can survive even after they lose majority support. As a result, there is a theoretical reason for the courts to be less concerned about religious endorsements that have existed for a long time than about new endorsements. Second, consider the accommodationist complaint that American governments today favor the values and philosophy of secularism. The argument is that removing religion from the public square while leaving the government free to endorse nonreligious values is akin to establishing a religion—​a fundamental worldview—​of humanism. 257

The Expressive Powers of Law

One should concede that government actions do routinely express the value of secularism in a way that some people experience as an insult. The issue is not limited to the teaching of Darwinian evolution. Federal and state governments subsidize scientists and universities that explore and teach theories on the physical origin of the universe, the chemical origins of life, the biology of human sexuality and sex differences, the geology and climatology of natural disasters, and neural explanations of so-­called near­death experiences. Funding decisions clearly elevate the scientific method of inquiry over the teachings of religious traditions. That divinity degrees or statements of belief are insufficient to get an individual funded by the National Science Foundation or hired by the National Weather Bureau or the Centers for Disease Control and Prevention expresses the inferiority of religious modes of inquiry for a great many important domains. The response to this point about secularism, however, is, as Dorf states, that such expressive harms are an inevitable consequence of having an effective government. The majority of people want the advantages of a scientifically oriented government, at least some of the time, and it would be quite impractical to forbid government from taking actions that, on occasion, endorse scientific values over religion. The endorsement of secularism will become parallel to religious endorsement only if and when secularist norms are so strong that they threaten the kind of social tyranny Mill warned of. Alarmists might say that the United States has already reached that point, but in most settings that is clearly not true. Not only are practicing believers generally not afraid to be identified as such, but also the culture demands the appearance of belief and piety in its politicians. All of this could one day change, of course, in which case this consequentialist analysis would make different recommendations. As a final point, let us consider the part of O’Conner’s passage I have ignored up to now. Besides the message to non-­adherents that they are outsiders, there is “an accompanying message to adherents that they are insiders, favored members of the political community.” Again, if we take this claim to be about beliefs and emotions, and to be a separate argument from the one just discussed, it is puzzling. Why would it be bad that a group felt like insiders? Why is an emotional gain—​a sense of pride—​a reason not to permit the government from making an endorsement? The information-­signaling model gives an explanation here too, though I do not claim it is what O’Conner had in mind. The risk of behavioral conformity arises from both parts of the signaled status hierarchy: not just 258

NOR M A T IVE I M P L I C A T IONS

that religious minorities are made to feel low, but also that the majority is made to feel high. The social cascade towards displays of the dominant religion is not just triggered because some members of the minority conceal their faith in public, but also because members of the majority start to display their membership publicly for the first time or to a greater degree. To return to Dorf’s difficult question: if government is presumptively allowed to express what it wants, because “[b]eing incidentally insulted” is “part of the price” of effective government, what explains the exceptions, where certain expressive harms overcome this presumption? The account I have sketched is by no means a complete answer because it merely identifies the relevant empirical inquiry at a fairly high level of generality: Courts should not try to limit all forms of “expressive rent-­seeking,” but should limit government expression endorsing a religious (or racial) hierarchy because such expression risks triggering the zealous behavioral conformity that Mill described as social tyranny. Given the history of religion and race, in the United States and elsewhere, those domains are far more credible candidates for judicial intervention than, say, government expression insulting smokers or the obese. One might raise the same concern about nationalism, as I described, but it is difficult to imagine a judiciary intervening to deny government the power to express support for patriot­ ­ism. Government expression in the service of public morale is generally useful in a way that government expression in the service of religious (or racial) differences is not. For a long time, the normative implications of legal expression have been almost the exclusive domain of deontological theorists whose focus excludes consequences. My hope is that the sketch of this chapter will interest people in taking seriously expressive consequences as a standard part of the normative analysis of law.

259

Conclusion Law’s Expressive Powers

A

s stated in the introduction, this book focuses on the causal theory (rather than normative theory) explaining how law expressively influences behavior (rather than explaining how expression or meaning influences the content of law). If we care about legal compliance, we need to know in detail the possible mechanisms by which law can influence behavior. I have identified two neglected paths of compliance. First, law is perhaps society’s most general-­purpose tool for creating focal points and achieving coordination. Coordinated behavior requires concordant expectations, and the law creates those expectations by the dictates it expresses. Constitutions coordinate expectations and behavior on the most basic framework for governing. Treaties coordinate expectations and behavior between nations. Ordinary domestic law coordinates expectations among citizens, who otherwise might find themselves in a conflict that is worse for each. Even legal sanctions exist only because of coordination on the person whose sanctioning orders will be obeyed. In a more dynamic setting, we can see that law helps the actors in a social movement coordinate their efforts at creating social change. And law continuously clarifies and refines informal rules (social norms, conventions, customs) at the boundaries, aligning expectations about what those rules demand, so as to avoid conflict. Law is also informative, revealing lawmakers’ beliefs about attitudes or risks associated with the behavior the law regulates. Making the proper inferences, people observing the law will update their beliefs about social attitudes or behavioral risks. Because people seek to avoid disapproval, the 260

C ON C L USION

new beliefs about attitudes change their behavior. Because people seek to avoid risk for themselves and family members, and to avoid harming others, new beliefs about risk will also change behavior. Both effects raise the expected costs of noncompliance and increase the level of compliance. By contrast, legal change regarding sanctions can signal that there are more legal violations than were previously understood, which can weaken deterrence and lower compliance. The bottom line is that legal economists should not assume that the effect of law on behavior is limited to the effect of legal sanctions, nor that those sanctions alone justify extensive theoretical inquiry. Sociologists and psychologists should not be content to assume that any nondeterrence (non-­incapacitation) effects are legitimacy effects. Indeed, the focal point and information effects interact with each other and the other mechanisms of compliance, particularly legitimacy. There remains much need for additional empirical confirmation, but we must start by identifying the precise mechanisms to be tested. My hope is that greater theoretical pluralism about the sources of law’s influence will spark greater interest in empiricism on the law’s power to coordinate and inform. At the same time, I have tried to emphasize throughout that there are limited conditions under which the focal point effect and the information effect are plausible. Political actors and law professors sometimes engage in expressive overclaiming, exuberantly asserting that a law or legal action “sends a message” in whatever way helps them make their rhetorical point. Those seriously making such claims should accept the burden of arguing that the conditions exist for a legal focal point or for the law to reveal information, or they should articulate another theory of law’s expressive effects, or they should be clear that their claims are not about behavior. There is more to compliance than deterrence and legitimacy, but there remain some contexts in which these are the only powers of law. With a more complete grasp of law’s effect on behavior, we have better tools for making normative judgments about law, a better understanding of the full range of law’s costs and benefits. The expressive effects of law matter to the economic analysis of optimal enforcement and the doctrinal analysis of topics as diverse as equal protection, ex post facto punishment, and the establishment of religion. I offer no distinctively expressive normative theory of law, but expressive consequences, like all consequences, should inform an evaluation of law. 261

Notes

Introduction 1. See, e.g., Steven D. Levitt and Thomas Miles, “Economic Analyses of Deterrence: Empirical,” in The New Palgrave Dictionary of Economics, 2nd ed., ed. Steven Durlauf and Lawrence Blume, (Palgrave MacMillan, 2008); Raymond Paternoster, “How Much Do We Really Know about Criminal Deterrence?,” Journal of Criminal Law and Criminology 100 (2010): 765–823. 2. See, e.g., Mats Persson and Claes-­Henric Siven, “Incentive and Incarceration Effects in a General Equilibrium Model of Crime,” Journal of Economic Behavior and Organization 59, no. 2 (2006): 214–229; Avinash Bhati, “Estimating the Number of Crimes Averted by Incapacitation: An Information Theoretic Approach,” Journal of Quantitative Criminology 23, no. 4 (2007): 355–375; Steven D. Levitt, “Why Do Increased Arrest Rates Appear to Reduce Crime: Deterrence, Incapacitation, or Measurement Error?,” Economic Inquiry 36, no. 3 (1998): 353–372. 3. Max Weber, Economy and Society: An Outline of Interpretive Sociology, vol. 1, ed. Guenther Roth and Claus Wittich (Berkeley, CA: University of California Press, 1978), 31. 4. Ibid. 5. Ibid., 37. 6. For some cleverly designed experimental support, see, e.g., Yuval Feldman and Doron Teichman, “Are All Legal Probabilities Created Equal?,” New York University Law Review 84 (2009): 980–1022; Yuval Feldman and Doron Teichman, “Are All Contractual Obligations Created Equal?,” Georgetown Law Journal 100 (2011): 5–52. 7. Kenworthey Bilz and Janice Nadler, “Law, Psychology, and Morality,” in The Psychology of Learning and Motivation: Moral Judgment and Decision Making, 263

notes to pages 3 – 4

vol. 50, ed. Daniel M. Bartels, Christopher W. Bauman, Linda J. Stitka, and Douglas L. Medin (San Diego, CA: Academic Press, 2009), 101, 117 (citing Tom R. Tyler, Why People Obey the Law: Procedural Justice, Legitimacy, and Compliance (New Haven, CT: Yale University Press, 1990)). 8. Bilz and Nadler, “Law, Psychology, and Morality,” at 107, 109. 9. See Tyler, Why People Obey; Tom R. Tyler, “Procedural Fairness and Compliance with the Law,” Swiss Journal of Economics and Statistics 133, no. 2 (1997): 219–240; Tom R. Tyler, “Psychological Perspectives on Legitimacy and Legitimation,” Annual Review of Social Psychology 57 (2006): 375–400. 10. See Paul Robinson and John M. Darley, Justice, Liability and Blame: Community Views and the Criminal Law (Boulder, CO: Westview Press, 1995); Paul H. Robinson and John M. Darley, “The Utility of Desert,” Northwestern University Law Review 91 (1997): 453–499, 473–477; John M. Darley, “Citizens’ Sense of Justice and the Legal System,” Current Directions in Psychological Science 10, no. 1 (2001): 10–13; Elizabeth Mullen and Janice Nadler, “Moral Spillovers: The Effect of Moral Mandate Violations on Deviant Behavior,” Journal of Experimental Social Psychology 44, no. 5 (2008): 1239–1245; Janice Nadler, “Flouting the Law,” Texas Law Review 83 (2005): 1399–1441, 1404–1410; William Stuntz, “Self-­Defeating Crimes,” Virginia Law Review 86 (2000): 1871–1899. For a review of competing psychological theories of justice, see Jason A. Colquitt, Donald E. Conlon, Michael J. Wesson, Christopher O. L. H. Porter, and K. Yee Ng, “Justice at the Millennium: A Meta-­A nalytic Review of 25 years of Organizational Justice Research,” Journal of Applied Psychology 86, no. 3 (2001): 425–445. 11. Weber defined law by the existence of an enforcement bureaucracy, which provides a reason to comply, but noted: “the motives for obedience may rather [than coercive sanctions enforced by state] be of many different kinds. In the majority of cases, they are predominantly utilitarian or ethical or subjectively conventional, i.e., consisting of the fear of disapproval by the environment.” Weber, Economy and Society, 314. 12. See, e.g., Martina Hartner, Silvia Rechberger, Erich Kirchler, and Alfred Schabmann, “Procedural Fairness and Tax Compliance,” Economic Analysis and Policy 38, no. 1 (2008): 137–152; Chung Kweon Kim, “Does Fairness Matter in Tax Reporting Behavior?,” Journal of Economic Psychology 23, no. 6 (2002): 771–785. 13. Raymond Fisman and Edward Miguel, “Corruption, Norms and Legal Enforcement: Evidence from Diplomatic Parking Tickets,” Journal of Political Economy 115 (2007): 1020–1048. 14. See, e.g., Feldman and Teichman, “Are All Legal Probabilities Created Equal?”; Tom R. Tyler, Stephen Schulhofer, and Aziz Z. Huq, “Legitimacy and Deterrence Effects in Counterterrorism Policing: A Study of Muslim 264

n o t e s t o p a g e s 4 – 11

Americans,” Law & Society Review 44, no. 2, 365–402 (2010); Peter J. May, “Compliance Motivations: Affirmative and Negative Bases,” Law & Society Review 38, no. 1, 41–68 (2004); Herbert Jacob, “Deterrent Effects of Formal and Informal Sanctions,” Law and Policy 2, no. 1 (1980): 61–80. 15. David Garland calls for this kind of pluralism in criminal theory: “Instead of assuming that an action can be categorized as instrumental or symbolic, crime-­control or penal, we ought to think in terms of a balance of multiple aims or dimensions, and differentiate actions accordingly . . .” David Garland, “Rethinking the Symbolic-­Instrumental Distinction: Meanings and Motives in American Capital Punishment,” in Governance and Regulation in Social Life: Essays in Honour of W. G. Carson, ed. Augustine Brannigan and George Pavlich, (Abingdon, UK: Routledge-­Cavendish, 2007): 178, 185. 1. Expressive Claims about Law 1. See, e.g., I. Bennett Capers, “Real Women, Real Rape,” UCLA Law Review 60 (2013): 826–882; David A. Dana, “The Law and Expressive Meaning of Condemning the Poor after Kelo,” Northwestern University Law Review 101 (2007): 365–382; Timothy R. Holbrook, “The Expressive Impact of Patents,” Washington University Law Review 84 (2006): 573–622; Sandeep Gopalan, “Say on Pay and the SEC Disclosure Rules: Expressive Law and CEO Compensation,” Pepperdine Law Review 35 (2008): 207–246; Jonathan C. Lipson, “The Expressive Function of Directors’ Duties to Creditors,” Stanford Journal of Law, Business and Finance 12 (2007): 224–288; Deborah Hellman, “Judging by Appearances: Professional Ethics, Expressive Government, and the Moral Significance of How Things Seem,” Maryland Law Review 60 (2001): 653–687; Note, “Expressive Harms and Standing,” Harvard Law Review 112 (1999): 1313–1330; Danielle Keats Citron, “Law’s Expressive Value in Combating Cyber Gender Harassment,” Michigan Law Review 108 (2009): 373–415; Andrew E. Taslitz, “The Expressive Fourth Amendment: Rethinking the Good Faith Exception to the Exclusionary Rule,” Mississippi Law Journal 76 (2006): 483–580; Alex Geisinger and Michael Ashley Stein, “A Theory of Expressive International Law,” Vanderbilt Law Review 60 (2007): 77–131; Onnig H. Dombalagian, “The Expressive Synergies of the Volcker Rule,” Boston College Law Review, 54 (2013): 469–533. 2. My work on this point begins with Richard H. McAdams, “The Origin, Development, and Regulation of Norms,” Michigan Law Review 96 (1997): 338–433, 397–408. My other early work included Richard H. McAdams, “A Focal Point Theory of Expressive Law,” Virginia Law Review 86 (2000): 1649– 1729; Richard H. McAdams, “An Attitudinal Theory of Expressive Law,” Oregon Law Review 79 (2000): 339–390; and Dhammika Dharmapala and Richard H. McAdams, “The Condorcet Jury Theorem and the Expressive 265

n o t e s t o p a g e s 11 – 1 4

Function of Law: A Theory of Informative Law,” American Law and Economics Review 5, no. 1 (2003): 1–31. Many others make this type of claim. See, e.g., Robert D. Cooter, “Expressive Law and Economics,” Journal of Legal Studies 27, no. S2 (1998): 585–607; Cass Sunstein, “On the Expressive Function of Law,” University of Pennsylvania Law Review 144 (1996): 2021–2053; Alex Geisinger, “A Belief Change Theory of Expressive Law,” Iowa Law Review 88 (2002): 35–73. 3. Lawrence Lessig, “The Regulation of Social Meaning,” University of Chicago Law Review 62 (1995): 943–1045, 968–972; Dan M. Kahan, “Social Influence, Social Meaning, and Deterrence,” Virginia Law Review 83 (1997): 349–395. 4. See Hope Clinic v. Ryan, 195 F.3d 857, 876, 879–81 (7th Cir. 1999) (Posner, J., dissenting), cert. granted and judgment vacated, 530 U.S. 1271 (2000). Commenting that there could be no policy or moral difference between the permitted and prohibited abortion techniques, Judge Posner concluded that the law was enacted “merely for [its] symbolic or aspirational effect” and that its proponents were “concerned with making a statement in an ongoing war for public opinion, though an incidental effect may be to discourage some late-­ term abortions.” 5. See Margaret Jane Radin, Contested Commodities (Cambridge, MA: Harvard University Press, 2001). 6. Rick Pildes and Elizabeth Anderson, “Expressive Theories of Law: A General Restatement,” University of Pennsylvania Law Review 148 (2000): 1503–1575; Deborah Hellman, “The Expressive Dimension of Equal Protection,” Minnesota Law Review 85 (2000): 1–69. 7. Shaw v. Reno, 509 U.S. 630, 631 (1993); Mississippi University for Women v. Hogan, 458 U.S. 718 (1982); J.E.B. v. Alabama ex rel. T.B., 511 U.S. 127 (1994). 8. I follow Matthew D. Adler, “Expressive Theories of Law: A Skeptical Overview,” University of Pennsylvania Law Review 148 (2000): 1363–1500, in distinguishing the entirely different subject of expressivism in metaethics. Roughly speaking, the expressivist view is that moral claims are mere expressions of emotive states, in contrast to a cognitive view that there is substantive content, which can be true or false, in moral claims. 9. Joseph Gusfield, Symbolic Crusade: Status Politics and the American Temperance Movement, 2nd ed. (Urbana, IL: University of Illinois Press, 1986). 10. See also Murray Edelman, The Symbolic Uses of Politics (Urbana, IL: University of Illinois Press, 1985); W. G. Carson, “Symbolic and Instrumental Dimensions of Early Factory Legislation: A Case Study in the Social Origins of Criminal Law,” in Crime, Criminology and Public Policy: Essays in Honour of Sir Leon Radnowicz, ed. Roger Hood (London: Heinemann, 1974). 11. Gusfield, Symbolic Crusade, 11.

266

no tes t o pag es 14 –19

12. Dan M. Kahan, “Ideology, Motivated Reasoning, and Cognitive Reflection,” Judgment and Decision Making 8, no. 4 (2013): 407–424; Dan M. Kahan and Donald Braman, “Cultural Cognition and Public Policy,” Yale Law and Policy Review 24 (2006): 149–172; Dan M. Kahan and Donald Braman, “More Statistics, Less Persuasion: A Cultural Theory of Gun-­Risk Perceptions,” University of Pennsylvania Law Review 151 (2003): 1291–1327; Dan M. Kahan, Donald Braman, Geoffrey L. Cohen, John Gastil, and Paul Slovic, “Who Fears the HPV Vaccine, Who Doesn’t, and Why? An Experimental Study of the Mechanisms of Cultural Cognition,” Law and Human Behavior 34, no. 6 (2010): 501–516; Dan M. Kahan, “The Secret Ambition of Deterrence,” Harvard Law Review 113 (1999): 413–500. 13. Pildes and Anderson, “Expressive Theories,” 1531–1532. 14. Matt Adler critiqued these claims in “Expressive Theories of Law: A Skeptical Overview,” 1428–1447; see also Matthew D. Adler, “Linguistic Meaning, Nonlinguistic ‘Expression’ and the Multiple Variants of Expressivism: A Reply to Professors Anderson and Pildes,” University of Pennsylvania Law Review 148 (2000): 1577–1594. An example of the confusion between the different literatures is that this article is often cited as a criticism of all expressive claims, not just the particular normative claims it actually attacks. 15. Joel Feinberg, “The Expressive Function of Punishment,” in Doing and Deserving: Essays in the Theory of Responsibility (Princeton: Princeton University Press, 1970), 95–118. 16. Deborah Hellman, When is Discrimination Wrong? (Cambridge, MA: Harvard University Press, 2008). 17. Kahan, “The Secret Ambition,” 462–476. For a more consequentialist approach, see Dhammika Dharmapala, Nuno Garoupa, and Richard H. McAdams, “Belief in a Just World, Blaming the Victim, and Hate Crime Statutes,” Review of Law and Economics 5, no. 1 (2009): 311–345. 18. See David Garland, “Rethinking the Symbolic-­Instrumental Distinction: Meanings and Motives in American Capital Punishment,” in Governance and Regulation in Social Life: Essays in Honour of W. G. Carson, ed. Augustine Brannigan and George Pavlich (Abingdon, UK: Routledge-­Cavendish, 2007). 19. See Deborah Hellman, When is Discrimination Wrong?, (Cambridge MA: Harvard University Press, 2008), 25; Heidi M. Hurd and Michael S. Moore, “Punishing Hatred and Prejudice,” Stanford Law Review 56 (2004): 1081–1146, 1102. 20. Adler, “Expressive Theories,” 1387. 21. Hurd and Moore, “Punishing,” 1102 (citing J. L. Austin, How to Do Things with Words, 2nd ed., ed. J. O. Ursmon and Marina Sbisà (Cambridge, MA: Harvard University Press, 1975), 117).

267

no tes t o pag es 19 – 23

22. See H. P. Grice, “Utterer’s Meaning, Sentence-­Meaning, and Word-­Meaning,” Foundations of Language 4 (1968): 225–242. 23. Within sentence meaning, Matt Adler observes the difference between prescriptive meaning and further meaning, which seems to be something like the difference between text and subtext. Adler, “Expressive Theories,” 1396–1401. 24. For example, Grice distinguished speaker’s meaning and sentence meaning as types of linguistic meaning, meanings that arise from the symbols of language. But I am going to extend these perspectives also to discuss non-­linguistic or “natural” meaning. Adler, “Expressive Theories,” 1384–1385; Hurd and Moore, “Punishing,” 1106. Natural meaning refers to the fact that an “action, event, or state of affairs . . . ​provides evidence of something,” as in Grice’s example, “Those spots mean (meant) measles.” Adler, “Expressive Theories,” 1384. In this example, “mean” is synonymous with “evidences.” Indeed, in Chapters 5 and 6, I claim that law influences behavior in exactly this way, by providing evidence or information. 25. See Henry E. Smith, “The Language of Property: Form, Context, and Audience,” Stanford Law Review 55 (2003): 1105–1191. 26. The perspective has other implications. For example, if your theory focuses on speaker’s meaning, you will have to address certain well known problems in defining the intent of a collective body, like a legislature. For a critique of collective intent, see Adler, “Expressive Theories,” 1389–1393. For recent defenses, see Christian List and Philip Pettit, Group Agency: The Possibility, Status, and Design of Corporate Agents, (New York: Oxford University Press, 2011); Richard Ekins, The Nature of Legislative Intent (New York: Oxford University Press, 2012). 27. See Yolanda Woodlee, “D.C. Mayor Acted ‘Hastily,’ Will Rehire Aide,” Washington Post, sec. A, February 4, 1999, http://www.washingtonpost.com /wp-­srv/local/longterm/williams/williams020499.htm. 2. The Focal Point Power of Expression 1. See Thomas C. Schelling, The Strategy of Conflict (New York: Oxford University Press, 1963), 144. 2. I originally discussed some of the ideas in this chapter in Richard H. McAdams, “A Focal Point Theory of Expressive Law,” Virginia Law Review 86 (2000): 1649–1729. 3. Andrew M. Colman, “Thomas C. Schelling’s Psychological Decision Theory: Introduction to a Special Issue,” Journal of Economic Psychology 27, no. 5 (2006): 603–608, 604. Some credit David Hume with the first recognition of the idea of focal points, though without using that term. See, e.g., Peter T. Leeson, Christopher J. Coyne, and Peter J. Boettke, “Converting Social Conflict: Focal Points and the Evolution of Cooperation,” Review of Austrian 268

notes to pages 23 – 27

Economics 19, no. 2-3 (2006): 137–147 (citing David Hume, A Treatise of Human Nature (Oxford: Oxford University Press, 2000), 315 (Hume’s example two rowers)). Hume is more widely credited with recognizing the importance of strategic action to convention and social order. See Peter Vanderschraaf, “The Informal Game Theory in Hume’s Account of Convention,” Economics and Philosophy 14, no. 2 (1998): 215–247. 4. Robert Sugden and Ignacio E. Zamarrón, “Finding the Key: The Riddle of Focal Points,” Journal of Economic Psychology 27, no. 5 (2006): 609–621, 610. 5. Roger B. Myerson, “Justice, Institutions, and Multiple Equilibria,” Chicago Journal of International Law 5 (2004): 91–107, 92. 6. Schelling, The Strategy of Conflict, 144 (emphasis added). He continues: “Similarly, the participants of a square dance may all be thoroughly dissatisfied with the particular dances being called, but as long as the caller has the microphone, nobody can dance anything else.” 7. See Damien Cave, “The Rhapsody of Port-­au-­Prince’s Streets,” New York Times, June 3, 2010, http://www.nytimes.com/2010/06/04/world/americas /04haiti.html 8. See John Geanakoplos, “Common Knowledge,” in Handbook of Game Theory with Economic Applications, vol. 2, ed. Robert J. Aumann and Sergiu Hart (North Holland: Elsevier Science, 1994); Peter Vanderschraaf and Giacomo Sillari, “Common Knowledge,” Stanford Encyclopedia of Philosophy (2007), http://plato.stanford.edu/entries/common-­knowledge/#5. 9. Michael Chwe, Rational Ritual: Culture, Coordination, and Common Knowledge (Princeton: Princeton University Press, 2001). 10. See Brian Skyrms, Evolution of the Social Contract (New York: Cambridge University Press, 1996), 84; Kenneth Binmore, “Do Conventions Need to Be Common Knowledge?,” Topoi 27, no. 1-2 (2008): 17–27, 23. Both doubt that any of the required common knowledge can arise in a large population, which casts doubt that legal rules could produce a convention. They propose instead the tools of evolutionary game theory, which do not require common knowledge or even rational agents. For attempts to reconstruct the common knowledge concept to make it less demanding, so as to answer this concern, see Robin P. Cubitt and Robert Sugden, “Common Knowledge, Salience and Convention: A Reconstruction of David Lewis’ Game Theory,” Economics and Philosophy 19 (2003): 175–210 and Peter Vanderschraaf, “Knowledge, Equilibrium and Convention,” Erkenntnis 49, no. 3 (1998): 337–369. 11. Tom Vanderbilt, Traffic: Why We Drive the Way We Do (and What It Says About Us) (New York: Random House, 2008), 211–243. 12. Some of the early statements of this idea come in Geoffrey Garrett and Barry R. Weingast, “Ideas, Interests, and Institutions: Constructing the European Community’s Internal Market,” in Ideas and Foreign Policy: Beliefs, 269

notes to pages 27– 29

Institutions, and Political Change, ed. Judith Goldstein and Robert O. Keohane (Ithaca, NY: Cornell University Press, 1993), 176; Russell Hardin, “Why a Constitution?,” in The Federalist Papers and the New Institutionalism, ed. Bernard Grofman and Donald Wittman (New York: Agathon Press, 1989), 113; Robert Cooter, “Expressive Law and Economics,” Journal of Legal Studies 27, no. 52 (1998): 585–608; and McAdams, “A Focal Point Theory,” 1651. See also Richard H. McAdams and Janice Nadler, “Testing the Focal Point Theory of Legal Compliance: The Effect of Third-­Party Expression in an Experimen­ ­tal Hawk/Dove Game,” Journal of Empirical Legal Studies 2, no. 1 (2005): 87–123; Gillian K. Hadfield and Barry R. Weingast, “What Is Law? A Coordination Model of the Characteristics of Legal Order,” Journal of Legal Analysis 4, no. 2 (2012): 471–514. 13. The seminal work on the role of coordination in language is David Lewis, Conven­ ­tion: A Philosophical Study (Cambridge, MA: Harvard University Press, 1969). 14. More recently, however, some claim that it would be easier to coordinate global business meetings and transactions if we had fewer time zones, perhaps even just one. See “A Brief History of Time Zones,” http://www.timeanddate .com/time/time-­zones-­history.html; Steve H. Hanke and Richard Conn Henry, “Changing Times,” Jan. 2012, http://www.cato.org/publications/commentary /changing-­times. See also Daniel S. Hamermesh, Caitlin Knowles Myers and Mark L. Pocock, “Cues for Timing and Coordination: Latitude, Letterman, and Longitude,” Journal of Labor Economics, 26, no. 2 (2008): 223–246. 15. See Robert B. Ahdieh, “The Visible Hand: Coordination Functions of the Regulatory State,” Minnesota Law Review 95 (2010): 578–649, 579–581 (describing generally how “[w]e live in a coordination economy”). 16. This point is in contrast to Andrei Marmor, “On Convention,” Synthese 107, no. 3 (1996): 349–371, who denies the coordination element in games. His argument might be correct for games of one, like solitaire, but seems implausible for games of two or more, where the rules of the game solve a coordination in the manner described in the text. 17. See Richard H. McAdams, “Conventions and Norms: Philosophical Aspects,” in International Encyclopedia of the Social and Behavioral Sciences, Vol. 4, ed., N. J. Smelser and P. B. Baltes, (Oxford: Elsevier Science, 2001), 2735–41. 18. I draw here on Richard H. McAdams, “Beyond the Prisoners’ Dilemma: Coordination, Game Theory, and Law,” Southern California Law Review 82 (2009): 209–258. Figures 2.1, 2.3, and 2.6 appear courtesy of Southern California Law Review. 19. The numbers represent years in prison, stated negatively to reflect that longer terms are worse. Game theory actually focuses not on concrete outcomes, but on the utility or disutility generated by outcomes. The differences are not 270

notes to pages 29 – 34

important in this one-­shot example so long as the players rank the outcomes in the way the numbers describe: best is to defect when the other player cooperates, second best is to cooperate when the other cooperates, third best is to defect when the other defects, and worst is to cooperate when the other defects. 20. Myerson, “Justice,” 92. 21. Douglas G. Baird, Robert H. Gertner, and Randal C. Picker, Game Theory and the Law (Cambridge, MA: Harvard University Press, 1996), 310. A “Nash equilibrium” is “based on the principle that the combination of strategies that players are likely to choose is one in which no player could do better by choosing a different strategy given the ones the others choose. A pair of strategies will form a Nash equilibrium if each strategy is one that cannot be improved upon given the other strategy. We establish whether a particular strategy combination forms a Nash equilibrium by asking if either player has an incentive to deviate from it.” Ibid. (emphasis deleted). 22. See Mark E. Van Der Weide and Satish M. Kini, “Subordinated Debt: A Capital Markets Approach to Bank Regulation,” Boston College Law Review 41 (2000): 195–264, 204. See also Daniel R. Fischel, Andrew M. Rosenfield and Robert S. Stillman, “The Regulation of Banks and Bank Holding Companies,” Virginia Law Review 73 (1987): 301–338, 307–308; Jonathan R. Macey, “The Business of Banking: Before and after Gramm-­Leach-­Bliley,” The Journal of Corporation Law 25 (2000): 691–722, 696. 23. See Macey, “The Business of Banking,” 696. 24. Along these lines, there is a famous story, possibly an urban legend, of four college students facing an alibi problem. They had explained their lateness for an exam by the excuse of a flat tire. The professor appeared to accept their story, but the revised exam prominently included the question “which tire?” See http://urbanlegends.about.com/od/college/a/flat_tire.htm. 25. A “pure strategy” is one that selects (in a given circumstance) a certain “move” or action with certainty. By contrast, a “mixed strategy” involves selecting between at least two actions with some probabilities that sum to one. Concordantly, in a pure strategy equilibrium, “each player adopts a particular strategy with certainty,” whereas in a mixed strategy equilibrium “one or more of the players adopts a strategy that randomizes among a number of pure strategies.” See Baird, Gertner and Picker, Game Theory and the Law, 310. To keep things simple, I focus on pure strategy equilibria. 26. For discussions of the game, see, e.g., Brian Skyrms, The Stag Hunt and the Evolution of Social Structure (Cambridge: Cambridge University Press, 2004), 3; Dennis Chong, Collective Action and the Civil Rights Movement (Chicago: The University of Chicago Press, 1991), 103–140; Gary Bornstein and Zohar Gilula, “Between-­Group Communication and Conflict Resolution in Assurance 271

notes to pages 34 – 36

and Chicken Games,” The Journal of Conflict Resolution 47, no. 3 (2003): 326–339. For technical differences in “Assurance” and “Stag Hunt,” see Baird, Gertner and Picker, Game Theory and the Law, 301, 315. 27. The name “Stag Hunt” comes from Rousseau’s illustration of the choice between hunting stag and hunting hare, where one succeeds in hunting stag only if the other hunter also hunts stag, where sharing a stag with the other hunter is the best outcome, but where hunting hare is safer because one can succeed on one’s own. Jean-­Jacques Rousseau, A Discourse on Inequality, trans. Maurice Cranston (Harmondsworth, Middlesex, England: Penguin Books, 1984), 111. 28. One might challenge the example by saying that a prosecutor would always choose to make the game a PD by offering zero years if a defendant is the only one who confesses (instead of one). In that case, it seems weakly dominant to confess, which is better for the prosecutor. Yet the reality is that, regardless of the prosecutor’s framing, most defendants will never regard an outcome of confessing and avoiding criminal sanctions as being as good as not confessing and avoiding criminal sanctions. Confessing may force the confessor to stop engaging in profitable illegal activities, suffer the reputational cost of being a snitch, and harm confederates he would rather not harm. 29. See Christopher R. Leslie, “Trust, Distrust, and Antitrust,” Texas Law Review 82 (2004): 515–680, 638–639. More evidence of the plausibility of the “Prisoners Assurance game” comes from the fact that, in a number of cases, scholars attempting to describe the PD erroneously state that the best outcome is mutual silence, which actually creates this Assurance game. See McAdams, “Beyond the Prisoners’ Dilemma,” 217 n.31. 30. For two recent counter-­examples, where legal scholars use the Assurance/Stag Hunt game, see Eric A. Posner, Kathryn E. Spier, and Adrian Vermeule, “Divide and Conquer,” Journal of Legal Analysis 2, no. 2 (2010): 417–472, 441–446; Julia Y. Lee, “Gaining Assurances,” Wisconsin Law Review 2012 (2012): 1137–1175. 31. The game gets its name from the original illustration: a husband and wife each prefer spending the evening together to spending it apart, but each prefer attending a different event. For discussions, see Colin Camerer, Behavioral Game Theory: Experiments in Strategic Interaction (Princeton, NJ: Princeton University Press, 2003), 353–367; Jennifer Gerarda Brown and Ian Ayres, “Economic Rationales for Mediation,” Virginia Law Review 80 (1994): 323–95, 373–77. 32. This kind of HD or Chicken game is sometimes called the “Volunteers’ Dilemma,” but it is not a PD (because it is not a dominant strategy to avoid volunteering). See William Poundstone, Prisoner’s Dilemma: John Von Neumann, Game Theory, and the Puzzle of the Bomb (New York: Doubleday, 1992), 201–204. 272

notes to pages 36 – 40

33. For discussions, see Robert Sugden, The Economics of Rights, Cooperation, and Welfare (New York: B. Blackwell, 1986), 55–103; Gary Bornstein, David Budescu and Shmuel Zamir, “Cooperation in Intergroup, N-­Person, and Two­Person Games of Chicken,” Journal of Conflict Resolution 41, no. 3 (1997): 384–406; Hugh Ward, “The Risks of a Reputation for Toughness: Strategy in Public Goods Provision Problems Modelled by Chicken Supergames,” British Journal of Political Science 17, no. 1 (1987): 23–52. 34. The other non-­equilibrium outcome is Dove/Dove. In Figure 2.5, the combined prison years at this outcome (−1 and −1 = −2) (which for purposes of illustration I am treating as equivalent to the utility payoffs) are the same as the combined years at the two equilibria (−2 and 0 = −2). This is not necessary; the joint Dove/Dove payoffs could be less or more than the joint equilibria payoffs, as long as the rank ordering of the outcomes remain the same. If the Dove/Dove payoffs are better, then we have another situation, like a PD, where the efficient outcome is not an equilibrium. This possibility represents another difference from BOS, where both non-­equilibrium outcomes are inferior to the two equilibria. 35. See Anatol Rapoport and Melvin Guyer, “A Taxonomy of 2 X 2 Games,” General Systems: A Yearbook of the Society for Advancement of General Systems Theory 11 (1966): 203–214, 204. See also Anatol Rapoport, Melvin J. Guyer, and David G. Gordon, The 2X2 Game (Ann Arbor: University of Michigan Press, 1976), 17. Each player can ordinally rank the four outcomes in the two-­by-­t wo setting in twenty-­four ways, which means the two players can rank the outcomes in 24 X 24 = 576 ways. But “[t]he game matrices are strategically equivalent whenever only the rows, the columns, both rows and columns, or, in symmetric games, the players are interchanged.” Katharina Holzinger, “The Problems of Collective Action: A New Approach,” Preprints aus der Max-­ Planck-­Projektgruppe Recht der Gemeinschaftsguter 5 (2003), available at http://www.coll.mpg.de/pdf_dat/2003_02online.pdf. 36. See Katherina Holzinger, “The Problems of Collective Action,” available at http://www.coll.mpg.de/pdf_dat/2003_02online.pdf. According to Holzinger, there are three close variants to the PD, five distinct cases of the Assurance (Stag Hunt) game or close variants, one case of the Hawk-­Dove (Chicken) game, and five distinct cases of the Battle of the Sexes game. 37. There are other ways of creating Assurance than these inequalities. My point is just to show the smallest change in inequalities that transforms the game from a PD to Assurance. 38. If each player were playing a tit-­for-­tat strategy, which reciprocates what the other player did in the prior round, then we might have perfect alternation of cooperation and defection. 39. See Garrett and Weingast, “Ideas, Interests, and Institutions,” 179–181. 273

notes to pages 42 – 46

40. Schelling, The Strategy of Conflict, 55 n.1, 56. 41. Ibid. 42. Psychological research finds that people from different cultures have different perceptual styles, rendering different objects “focal” in their visual field and visual recall. See Takahiko Masuda and Richard E. Nisbett, “Attending Holistically Versus Analytically: Comparing the Context Sensitivity of Japanese and Americans,” Journal of Personality and Social Psychology 81, no. 5 (2001): 922–934; Richard E. Nisbett, Kaiping Peng, Incheol Choi, and Ara Norenzayan, “Culture and Systems of Thought: Holistic Versus Analytic Cognition,” Psychological Review 108, no. 2 (2001): 291–310. 43. Schelling, The Strategy of Conflict, 96. 44. Ibid., 57. 45. Ibid., 98 (emphasis in original). 46. See Adrian Furnham and Hua Chu Boo, “A Literature Review of the Anchoring Effect,” Journal of Socio-­Economics 40, no. 1 (2011): 35–42; Thomas Mussweiler and Birte Englich, “Subliminal Anchoring: Judgmental Conse­ quences and Underlying Mechanisms,” Organizational Behavior and Human Decision Processes, 98, no. 2 (2005): 133–143. 47. Amos Tversky and Daniel Kahneman, “Judgment under Uncertainty: Heuristics and Biases,” Science 185, no. 4157 (1974): 1124–1131. 48. Judith Mehta, Chris Starmer and Robert Sugden, “The Nature of Salience: An Experimental Investigation of Pure Coordination Games,” The American Economic Review 84, no. 3 (1994): 658–673, 667. 49. Keynes famously analogized financial investment to a newspaper beauty competition that asks one to predict from an array of photographs the face that will be most frequently identified as the “prettiest.” A competitor does not select the face he believes to be the prettiest, but the one “he thinks likeliest to catch the fancy of the other competitors, all of whom are looking at the problem from the same point of view.” See John Maynard Keynes, “The State of Long­term Expectation,” chap. 12 in The General Theory of Employment, Interest, and Money (Palgrave Macmillian, 1936). For an informal experiment validating this theory, see David Kestenbaum, “Ranking Cute Animals: A Stock Market Experiment, NPR Planet Money, January 14, 2011, at http://www.npr .org/blogs/money/2011/01/14/132906135/ranking-­cute-­animals-­a-­stock-­market -­experiment. 50. A more formal description of the game would therefore show it to be sequential, where if both players choose stop, the game continues another round. The result might be a game like the “war of attrition” or something else. I simplify here by using a reduced form expected payoff for the Wait/Wait ­outcome.

274

notes to pages 48 – 50

51. Take p to be the probability the other driver waits. The expected value of Wait is the probability that the other driver waits (p) times the outcome for Wait/ Wait (−2) plus the probability that the other driver proceeds (1 − p) times the outcome for Wait/Proceed (−1). The expected value of Proceed is the probability that the other driver waits (p) times the outcome for Proceed/Wait (0) plus the probability that the other driver proceeds (1 − p) times the outcome for Proceed/Proceed (−9). Game theory predicts a mixed strategy equilibrium where the payoffs of Wait and Proceed are equal. If we set the payoffs equal and solve for p, the result is p = 0.8. With each player selecting Wait with probability .8 and Proceed with probability .2, the expected value is −1.8. 52. Schelling, The Strategy of Conflict, 66. 53. Ibid., 144. 54. Ibid. 55. See, e.g., Michael Bacharach and Michele Bernasconi, “The Variable Frame Theory of Focal Points: An Experimental Study,” Games and Economic Behav­ ­ior 19, no. 1 (1997): 1–45; Antoni Bosch-­Domènech and Nicolaas J. Vriend, “On the Role of Non-­equilibrium Focal Points as Coordination Devices” (Feb. 2008). Department of Economics and Business, Universitat Pompeu Fabra Working Paper No. 1064. Available at http://www.econ.upf.edu/docs/papers/downloads /1064.pdf; Mehta, Starmer, and Sugden, “The Nature of Salience,” 658–673. Judith Mehta, Chris Starmer, and Robert Sugden, “Focal Points in Pure Coor­ dination Games: An Experimental Investigation,” Theory and Decision 36, no. 2 (1994): 163–185; Judith Mehta, Chris Starmer, and Robert Sugden “An Experimental Investigation of Focal Points in Coord­ination and Bargain­ing: Some Preliminary Results,” in Decision Making Under Risk and Uncertainty: New Models and Empirical Findings, ed. John Geweke (Dordrecht: Kluwer Academic Publishers, 1992), 211–219. 56. See Vincent Crawford, “A Survey of Experiments on Communication via Cheap Talk,” Journal of Economic Theory 78, no. 2 (1998): 286–298; Kathleen Valley, Leigh Thompson, Robert Gibbons and Max H. Bazerman, “How Communication Improves Efficiency in Bargaining Games,” Games and Economic Behavior 38, no. 1 (2002): 127–155. 57. See Jordi Brandts and W. Bentley MacLeod, “Equilibrium Selection in Experimental Games with Recommended Play,” Games and Economic Behavior 11, no. 1 (1995): 36–63; Rachel Croson and Melanie Marks, “The Effect of Recommended Contributions in the Voluntary Provision of Public Goods,” Economic Inquiry 39, no. 2 (2001): 238–249; McAdams and Nadler, “Testing the Focal Point Theory,” 87–123; Richard H. McAdams and Janice Nadler, “Coordinating in the Shadow of the Law: Two Contextualized Tests of the Focal Point Theory of Legal Compliance,” Law and Society Review 42, no.

275

notes to pages 50 – 54

4 (2008): 865–898; Andrew Schotter and Barry Sopher, “Social Learning and Coordination Conventions in Intergenerational Games: An Experimental Study,” Journal of Political Economy 111, no. 3 (2003): 498–529; Rick K. Wilson and Carl M. Rhodes, “Leadership and Credibility in N-­Person Coordination Games,” Journal of Conflict Resolution 41, no. 6 (1997): 767–791. 58. Vincent P. Crawford, Uri Gneezy, and Yuval Rottenstreich, “The Power of Focal Points Is Limited: Even Minute Payoff Asymmetry May Yield Large Coordina­ tion Failures,” American Economic Review 98, no. 4 (2008): 1443–1458. 59. Ibid., 1455. 60. Ibid., 1454. 61. Subhasish Dugar and Quazi Shahriar, “Focal Points and Economic Efficiency: The Role of Relative Label Salience,” Southern Economic Journal 78, no. 3 (2012): 954–975. 62. Mehta, Starmer, and Sugden, “The Nature of Salience,” 658–673; Mehta, Starmer, and Sugden, “Focal Points in Pure Coordination Games,” 163–185. 63. Richard H. McAdams and Janice Nadler, “The Power of Focal Points is Pervasive: Experimental Studies of Game Labels in Disparate Settings” (work in progress, 2011). 64. For example, in one condition we gave the four possible strategies the labels A, B, C, and D. In another, we used the labels New York, Tallahassee, Wichita, and Bakersfield. The labels “A” and “New York” were influential in the symmetric coordination game and remained so in the asymmetric game. 65. Bosch-­Domènech and Vriend, “Non-­Equilibrium Focal Points,” 6. In their two-­person pure coordination game, without a focal point, subjects rarely reach one of the thirty Nash Equilibria, all with payoffs of (100, 100), and thus usually wind up with payoffs of (0, 0). When the researchers “shaved” the payoffs of a single equilibrium down to (87, 87), it became focal because it was the only outcome whose payoffs are neither 0 nor 100. The effect was to significantly increase the selection of the strategies associated with the unique payoffs (87, 87). Oddly, the 87, 87 outcome is not a Nash equilibrium. Given the strategies that produce this outcome, either player could switch to a strategy that produces the 100, 100 outcome. The problem for the players is that if both switch, they will reach one of the 0, 0 outcomes. In this setting, focal points are so strong, they induce the players to select a non-­equilibrium outcome (or to transform our understanding of what an equilibrium is). 66. See Andrew Isoni, Anders Poulsen, Robert Sugden, and Kei Tsutsui, “Focal Points in Tacit Bargaining Problems: Experimental Evidence,” European Economic Review 59 (2013): 167–188. 67. The authors focus on differences in disk selection, efficiency (the amount of coordination), and distribution. The focal point loses significance for some of 276

notes to pages 54 – 55

these measures as the inequality increases, but the authors summarize their findings: “The relative strength of the efficiency and distributional effects of labelling cues varies between the games we have studied, but in all the games in which there are cues . . . ​at least one of those effects is significant.” Ibid. at 181. 68. Erik W. de Kwaadsteniet and Eric van Dijk, “Social status as a cue for tacit coordination,” Journal of Experimental Social Psychology 46, no. 3 (2010): 515–524. For a review of such literature, see Erik W. de Kwaadsteiniet and Eric van Dijk, “A Social-­Psychological Perspective on Tacit Coordination: How it Works, When it Works, (and When It Does Not),” European Review of Social Psychology 23, no. 1 (2012): 187–223, 206–213. 69. See Hakan J. Holm, “Gender-­Based Focal Points,” Games and Economic Behavior 32, no. 2 (2000): 292. Two experiments were conducted in Sweden (306 subjects) and one in the United States (164 subjects). 70. In the first Swedish experiment (of two reported), subjects selected the more aggressive strategy 67.6 percent of the time when matched against a woman, but only 47.9 percent of the time when matched against a man. Ibid., 299. For the American study, the numbers were 50 percent and 37.5 percent. Ibid., 304–05. In the first Swedish experiment, men earned 27 percent more than the women. In the American study, male subjects earned 28 percent more than female subjects. Ibid., 305. I discuss the long term effect of this sort of result in Richard H. McAdams, “Conformity to Inegalitarian Conventions and Norms: The Contribution of Coordination and Esteem,” The Monist 88, no. 2 (2005): 238–59. 71. McAdams and Nadler, “Testing the Focal Point Theory,” 87–123. 72. Ibid., at 106 (“If you wish, you may consider the result of the spin in your decision, in whatever manner you choose. You are also entirely free to ignore it.”). 73. If p is the probability that player two selects C1, then, for player one, the expected value of R1 is the probability that player two selects C1 (p) times the outcome for R1/C1 (1) plus the probability that player two selects C2 (1 − p) times the outcome for R1/C2 (0). The expected value of R2 for player one is the probability that player two selects C1 (p) times the outcome for R2/C1 (2) plus the probability that player two selects C2 (1 − p) times the outcome for R2/C2 (−1). Setting the payoffs equal, the unique solution is p = 0.5. Because the payoffs are symmetric, the same applies to player two. With each player selecting each action with probability .5, the payoffs create an expected value of .5. 74. The learning experiments include David Bodoff, “When Learning Meets Salience,” Theory and Decision 74, no. 2 (2013): 241–266; Andreas Blume and Uri Gneezy, “An Experimental Investigation of Optimal Learning in Coordi­ nation Games,” Journal of Economic Theory 90, no. 1 (2000): 161–172. For CGR’s observation that learning might change their results, see Crawford, Gneezy, and Rottenstreich, “The Power of Focal Points Is Limited,” 1444. 277

notes to pages 56 – 60

75. Richard H. McAdams and Janice Nadler, “The Power of Focal Points Is Pervasive: Experimental Studies of Game Labels in Disparate Settings” (work in progress, 2011). 3. Law as Focal Point 1. See N. Gregory Mankiw & Laurence Ball, Macroeconomics and the Financial System (New York: Worth Publishers, 2011), 81 (“[I]n the end the use of money in exchange is a social convention; everyone values fiat money because they expect everyone else to value it.”). 2. See Walter W. Heering, “Money and Reciprocity in the Extended Order—​A n Essay on the Evolution and Cultural Function of Money,” in Entrepreneurship, Monday and Coordination: Hayek’s Theory of Cultural Evolution (Cheltenham, UK: Edward Elgar, 2005), 156, 170–171 (describing acceptance of money as a coordination game but limited access (not counterfeiting) as a prisoners’ dilemma). 3. Or, as a David Mamet character says in the film Heist (2001), “Everybody needs money. That’s why they call it money.” 4. See Worcester v. Georgia, 31 U.S. 515 (1832); Paul F. Boller and John H. George, They Never Said It: A Book of False Quotes, Misquotes, & False Attributions (New York: Oxford University Press, 1989), 53. Apparently, Jackson actually said: The decision was “stillborn” and “it cannot coerce Georgia to yield to its mandate.” See Robert V. Remini, Andrew Jackson and the Course of American Freedom, 1822–1832, vol. 2 (New York: Harper & Row, 1981), 276–77. 5. See Stoddard v. Donahoe, 224 Ariz. 152 (Ariz. Ct. App. 2010). 6. J. J. Hensley, “Maricopa County officer Stoddard starts jail term for taking documents,” The Arizona Republic, December 2, 2009, http://www.azcentral.com /community/phoenix/articles/2009/12/02/20091202doincustody1202-­ON.html; Michael Kiefer, “Contempt-­of-­court ruling against Arpaio deputy upheld,” The Arizona Republic, April 7, 2010, http://www.azcentral.com/news/articles /2010/04/06/20100406joe-­arpaio-­deputy-­contempt-­of-­court.html; Michael Kiefer, “Superior Court Judge Gary Donahoe leaves post,” The Arizona Republic, March 30, 2010, http://www.azcentral.com/ news/articles/2010/03 /29/20100329superior-­court-­judge-­leaves-­post.html. 7. George Mailath, Stephen Morris and Andrew Postlewaite, “Laws and Author­ ­ity” (Apr. 2001). Working Paper, 3, http://www.princeton.edu/∼smorris/pdfs /laws&authority.pdf. 8. Thomas C. Schelling, The Strategy of Conflict (New York: Oxford University Press, 1963), 91. 9. So do more spontaneous examples, such as a riot or wildcat strike. See David D. Haddock and Daniel Polsby, “Understanding Riots,” Cato Journal 14, no. 1 (1994): 147–157 (discussing the role of focal points and leadership in creating 278

notes to pages 60 – 66

riots); Dennis Chong, Collective Action and the Civil Rights Movement (Chicago: University of Chicago Press, 1991), 118–120 (discussing the element of coordination and assurance in a wildcat strike). 10. Roger Myerson proposes a useful thought experiment: A group of players, all wearing name tags, are sitting in a circle. Each must independently write one player’s name on a piece of paper. If they all write the same name then they each get $100, except that the person named gets $200. Otherwise they all get $0. The players have never met each other before. But just before they play, someone walks in, puts a big shiny crown on one player’s head, and walks away.

Roger B. Myerson, “Justice, Institutions, and Multiple Equilibria,” The Chicago Journal of International Law 5 (2004): 91–107, 93. We imagine that everyone will name the crowned person merely because they expect everyone else to do so, even though the crown is awarded arbitrarily, not fairly. 11. Mailath, Morris, and Postlewaite, “Laws and Authority,” 23. 12. Richard H. McAdams and Janice Nadler, “Coordinating in the Shadow of the Law: Two Contextualized Tests of the Focal Point Theory of Legal Compliance,” Law and Society Review 42, no. 4 (2008): 865–898. 13. See, e.g., Roberto Galbiati and Pietro Vertova, “Obligations and cooperative behaviour in public good games,” Games and Economic Behavior 64, no. 1 (2008): 146–170 (experimental instructions state that certain conduct was “required”); Iris Bohnet and Robert D. Cooter, “Expressive Law: Framing or Equilibrium Selection?”, November 2003, KSG Working Paper No. RWP03–046 and UC Berkeley Public Law Research Paper No. 138, http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=452420 (experimental instructions state that certain behavior will be “punished”). 14. McAdams and Nadler, “Coordinating in the Shadow of the Law,” 875–895. 15. Ibid., 877–878. In all vignettes, one claimant was the original owner who lost the cat to a thief and the other was a good faith purchaser who innocently bought the cat from the thief. The participant was always assigned to the second role, while the old precedent said that the cat belonged to the original owner. 16. Ibid., 883. 17. Ibid., 891. 18. The finding is consistent with the finding that default rules are “sticky.” See Stewart J. Schwab, “A Coasean Experiment on Contract Presumptions,” Journal of Legal Studies 17, no. 2 (1988): 237–68, 254–56; Russell Korobkin, “Inertia and Preference in Contract Negotiation: The Psychological Power of Default Rules and Form Terms,” Vanderbilt Law Review 51 (1998): 1583–1651, 1626–27. The psychological explanation is the endowment effect or status quo 279

no tes t o pag es 66 –71

bias, but the economic alternative is the power of focal points (given that bargaining involves coordination). A more recent experiment suggests that people intensify their moral judgments about the wrongfulness of contractual breach in light of the legal remedy of specific performance. See Ben Depoorter and Stephan Tontrup, “How Law Frames Moral Intuitions: The Expressive Effect of Specific Performance,” Arizona Law Review 54 (2012): 673–717. 19. Schelling, Strategy of Conflict. See also Maarten C. W. Janssen, “On the strategic use of focal points in bargaining situations,” Journal of Economic Psychology 27, no. 5 (2006): 622–634. 20. Schelling, Strategy of Conflict, at 62–63. 21. Geoffrey Garrett and Barry R. Weingast, “Ideas, Interests, and Institutions: Constructing the European Community’s Internal Market,” in Ideas and Foreign Policy, ed. Judith Goldstein and Robert O. Keohane (Ithaca, NY: Cornell University Press, 1993), 178–181. 22. See Daniel W. Drezner, All Politics Is Global: Explaining International Regulatory Regimes (Princeton: Princeton University Press, 2007), 11–12. 23. Hector Vera, “Decimal Time: Misadventures of a Revolutionary Idea, 1793– 2008,” KronoScope 9, no. 1–2 (2009): 29–48. 24. Thomas Hobbes, Leviathan (London: Printed for Andrew Cooke, 1651), XIII.9. 25. But not on all accounts. See, e.g., Boaz Moselle and Benjamin Polak, “A Model of a Predatory State,” Journal of Law, Economics and Organization 17, no. 1 (2001): 1–33. 26. “There appears to be virtually unanimous agreement that the State of Nature should be represented as a Prisoner’s Dilemma.” Andrew Alexandra, “Should Hobbes’s State of Nature Be Represented as a Prisoner’s Dilemma?,” Southern Journal of Philosophy 30, no. 2 (1992): 1–16. Alexandra argues instead for understanding the State of Nature as an Assurance Game. 27. See Alexandra, “Should Hobbes’s State of Nature Be Represented as a Prisoners’ Dilemma?,” 1–16; Michael Moehler, “Why Hobbes’ State of Nature is Best Modeled by an Assurance Game,” Utilitas 21, no. 3 (2009): 297–326. 28. See Nicholas Almendares and Dimitri Landa, “Strategic Coordination and the Law,” Law and Philosophy 26, no. 5 (2007): 501–529, 520. 29. Russell Hardin, “Why a Constitution?,” in The Federalist Papers and the New Institutionalism, ed. Bernard Grofman and Donald Wittman (New York: Agathon Press, 1989), 102–108; See also Russell Hardin, Liberalism, Constitutionalism, and Democracy (New York: Oxford University Press, 2003), 82–140; Russell Hardin, “Constitutionalism,” in The Oxford Handbook of Political Economy, ed. Barry R. Weingast and Donald A. Wittman (New York: Oxford University Press, 2006), 289, 291. 30. Others make a similar point using something like an iterated PD game. See Matthew Stephenson, “   ‘When the Devil Turns . . .’: The Political 280

n o tes t o pag es 71–74

Foundations of Independent Judicial Review,” Journal of Legal Studies 32, no. 1 (2003): 59–89; Jon Elster, “Unwritten Constitutional Norms,” (unpublished manuscript, cited by permission, 2010), 39–40. The idea is that political order arises as an agreement between powerful political parties who expect to alternate c­ ontrol of the state. Cooperation takes the form of one side’s adhering to certain restraints of power when in office in exchange for the other side’s restraint when it is in office. The element of coordination is that, to cooperate, the parties must agree on the restraints on power when there is more than one way to define them. 31. See Daryl J. Levinson, “Parchment and Politics: The Positive Puzzle of Constitutional Commitment,” Harvard Law Review 124 (2011): 657-746; Peter C. Ordeshook, “Are ‘Western’ Constitutions Relevant to Anything Other Than the Countries They Serve?,” Constitutional Political Economy 13, no. 1 (2002): 3–24; Eric A. Posner, “Strategies of Constitutional Scholarship,” Law and Social Inquiry 26, no. 2 (2001): 529–546; Adrian Vermeule and Eric A. Posner, “Constitutional Showdowns,” University of Pennsylvania Law Review 156 (2008): 991–1048, 1022. 32. For a wide-­ranging discussion of the abandonment of constitutional powers, see Adrian Vermeule, “The Atrophy of Constitutional Powers,” Oxford Journal of Legal Studies 32, no. 3 (2012): 421–444. 33. See, e.g., Russell Hardin, “Contractarianism: Wistful Thinking,” Constitutional Political Economy 1, no. 2 (1990): 35–52. 34. See Ex parte Merryman, 17 F. Cas. 144, 147 (C.C.D. Md. 1861) and the discussion in Saikrishna Bangalore Prakash, “The Great Suspender’s Unconstitutional Suspension of the Great Writ,” Albany Government Law Review 3 (2010): 575–614. 35. See James Madison “Federalist, No. 48,” in The Federalist Papers, ed. Isaac Kramnick (Harmondsworth: Penguin Books, 1987), 309. 36. Stephenson, “   ‘When the Devil Turns . . .’,” 60. 37. Barry R. Weingast, “The Political Foundations of Democracy and the Rule of Law,” American Political Science Review 91, no. 2 (1997): 245–263, 245–246. 38. Ibid., 251 (“Policing the sovereign requires that citizens coordinate their reactions, which requires constructing a coordination device,” such as a written constitution). 39. See Eric A. Posner, Kathryn E. Spier, and Adrian Vermeule, “Divide and Conquer,” Journal of Legal Analysis 2, no. 2 (2010): 417–472, 441–446. See also James D. Fearon, “Self-­Enforcing Democracy,” Quarterly Journal of Economics 126, no. 4 (2011): 1661–1708. 40. Tiberiu Dragu and Mattias Polborn, “The Administrative Foundation of the Rule of Law,” Journal of Politics 75, no. 4 (2013): 1038–1050. 41. Dragu and Polborn, “The Administrative Foundations of the Rule of Law.” 281

notes to pages 76 –78

42. David Strauss observes that the focal point theory directs courts to give serious weight to constitutional text. See David A. Strauss, “Common Law Constitutional Interpretation,” University of Chicago Law Review 63 (1996): 877–935, 910–911; David A. Strauss, “Common Law, Common Ground, and Jefferson’s Principle,” Yale Law Journal 112 (2003): 1717–1755, 1733–1736. 43. Regarding pirate constitutions, also written, see Peter T. Leeson, “An-­arrgh­chy: The Law and Economics of Pirate Organization,” Journal of Political Economy 115, no. 6 (2007): 1049–1094; Peter T. Leeson, “The Calculus of Piratical Consent: The Myth of the Myth of Social Contract,” Public Choice 139, no. 3–4 (2009): 443–459. 44. See Richard H. McAdams, “The Expressive Power of Adjudication,” University of Illinois Law Review 2005 (2005): 1043–1122, 1080; Richard H. McAdams and Janice Nadler, “Testing the Focal Point Theory of Legal Compliance: The Effect of Third-­Party Expression in an Experimental Hawk/Dove Game,” Journal of Empirical Legal Studies 2, no. 1 (2005): 87–123, 92–97; McAdams and Nadler, “Coordinating in the Shadow of the Law,” 874–875. 45. See The World Health Organization, Global Status Report on Road Safety 2013: Supporting a Decade of Action, at 1, http://www.who.int/violence_injury _prevention/road_safety_status/2013/en/. 46. Possibly, but not necessarily, to the point where the yield sign is common knowledge between them. 47. The state can lower speed limits, but the element of coordination in driving speeds does not necessarily support compliance; a driver wants to roughly match the speed of traffic and if others maintain their previous high speed, then one wants to maintain that speed as well. 48. See Tom Vanderbilt, Traffic: Why We Drive the Way We Do (and What It Says About Us) (New York: Random House, 2008), 186–204. If signage and lane markings have unintended consequences, so do their absence: See Rob Imrie, “Shared Space and the Post-­politics of Environmental Change,” Urban Studies 50, no. 16 (2013): 3446-­3462 (studying the political failure to consider the needs of the visually impaired in creating Shared Space in the UK); “   ‘Shared street’ Problem for Blind,” BBC News, May 20 2009, http://news.bbc.co.uk/2/hi/uk _news/wales/8058604.stm. 49. See Vanderbilt, Traffic, 229 (quoting Evans). 50. Vanderbilt offers to explain the puzzle as follows. First, “[p]erhaps in speeding the driver feels as if he’s in control, while going through a red light, even carefully, puts one at the risk of others.” Yet Vanderbilt does not explain why the driver feels more in control in the riskier situation when the safety of both situations depends on other drivers. Second, Vanderbilt says: “He may also speed because most other people do (whereas if everyone decided to cross through 282

notes to pages 78 – 86

red lights, anarchy would ensue).” That is essentially the point of the text: the interest in coordination pushes against running red lights but need not push against and sometimes works in favor of speeding. Ibid. 51. See Richard F. Beaubien, “Controlling Speeds on Residential Streets,” ITE Journal 59, no. 4 (1989): 37–39; W. Martin Bretherton Jr., “Multi-­way Stops—​ The Research Shows the MUTCD is Correct!,” http://www.ite.org/traffic /documents/aha99b49.pdf. 52. See 2B.04, Right of Way at Intersections, Manual on Uniform Traffic Control Devices (2009 edition), at p.50 (“YIELD or STOP signs should not be used for speed control.”), at http://mutcd.fhwa.dot.gov/pdfs/2009r1r2/part2b.pdf. 53. See, e.g., Sally Engle Merry, Urban Danger: Life in a Neighborhood of Strangers (Philadelphia: Temple University Press, 1981), 175–186; Richard E. Nisbett and Dov Cohen, Culture of Honor: The Psychology of Violence in the South (Boulder: Westview Press, 1996), xv–xvi, 25–39; Donald Black, “Crime as Social Control,” American Sociological Review 48, no. 1 (1983): 34–45. 54. See James M. Acheson, The Lobster Gangs of Maine (Durham, NH: University Press of New England, 1988), 74 (reporting on lobstermen who enforce customary territorial claims by destroying “offending” lobster traps); Robert Ellickson, Order without Law: How Neighbors Settle Disputes (Cambridge: Harvard University Press, 1991), 215–219 (reporting that the informal remedy of last resort for cattle trespass is the forceful destruction or seizure of property); Anna-­Maria Marshall, “Idle Rights: Employees’ Rights Consciousness and the Construction of Sexual Harassment Policies,” Law and Society Review 39, no. 1 (2005): 83–124, 111 (quoting a harassment victim: “   ‘I promptly slugged him and almost threw him on the conveyer belt in the [animal] cage wash area . . . .’ She admitted that violence was ‘not a good idea, but it seems the only action that works.’   ”). 55. From a broader view, emotions like this might serve a rational purpose; by committing the angry person to retaliatory action, the emotion gives others a reason not to make the person angry. See Robert Frank, Passions within Reason: The Strategic Role of the Emotions (New York: W. W. Norton, 1988). 56. Compare Michael McCann, Rights at Work: Pay Equity Reform and the Politics of Legal Mobilization (Chicago: University of Chicago Press, 1994). 57. Steve Rhodes, “Is Daley Calling Dibs or Not?,” NBC Chicago, January 23, 2009, http://www.nbcchicago.com/news/local/Daleys-­Dibs.html; Tom Vanderbilt, “The Ethics of Winter Dibs Parking,” Slate, February 25, 2011, http://www.slate.com/articles/life/transport/2011/02/the_ethics_of_winter _dibs_parking.html. 58. Informal property rights exist in a variety of settings around the world with far more significant effects. Any such property rights are likely to benefit from a 283

notes to pages 86 – 90

focal point definition. See, e.g., Shitong Qiao, “Small Property, Big Market: A Focal Point Explanation,” American Journal of Comparative Law, 63, no. 1 (forthcoming 2015), http://ssrn.com/abstract=2399675. 59. The National Conference of State Legislatures has collected statutes protecting mothers’ right to breastfeed publicly. See http://www.ncsl.org /issues-­research/health/breastfeeding-­state-­laws.aspx#State. 60. State laws differ widely. Some only exempt breastfeeding from the reach of criminal prohibitions against public nudity or indecency. E.g., Mich. Comp. Laws §§ 41.181, 67.1aa and 117.4i. These are not likely to create a focal point. Other laws, however, refer (at least in the title) to a “right to breastfeed” and provide that a mother may breastfeed in any public location (e.g., Va. Code § 2.2–1147.1), or public and private locations (e.g., 740 I.L.C.S. 137/10 and Ohio Rev. Code Ann. § 3781.55), if the mother is otherwise authorized to be present there. Some statutes authorize injunctions against non-­compliant owners who do not permit breastfeeding on their premises. E.g., 740 I.L.C.S. 137/15 (exempting private residences and houses of worship). 61. See Ellickson, Order without Law, 52–64; Elizabeth A. Hoffman, Co-­ Operative Workplace Dispute Resolution: Organizational Structure, Ownership, and Ideology (Surrey, England: Gower Publishing, 2012); Marshall, “Idle Rights,” 83–124. 62. Michael Mesterton-­Gibbons and Eldridge S. Adams, “Landmarks in Territory Partitioning: A Strategically Stable Convention?,” The American Naturalist 161, no. 5 (2003): 685–697. 63. Ibid. 64. Ibid., 695. 65. See Derek C. Penn & Daniel J. Povinelli, “On the Lack of Evidence that Non­human Animals Possess Anything Remotely Resembling a ‘Theory of Mind,’   ” Philosophical Transactions: Biological Sciences 362, no. 1480 (2007): 731–744; N. J. Emery & N. S. Clayton, “Comparative Social Cognition,” Annual Review of Psychology 60 (2009): 87–113; Derek C. Penn, Keith J. Holyoak and Daniel J. Povinelli, “Darwin’s mistake: Explaining the discontinuity between human and nonhuman minds,” Behavioral and Brain Sciences 31, no. 2 (2008): 109–130. 66. See Justin R. LaManna and Perri K. Eason, “Effects of Landmarks on Territorial Establishment,” Animal Behaviour 65, no. 3 (2003): 471–478. 67. Ibid., 477. 68. See also David Ley and Roman Cybriwsky, “Urban Graffiti as Territorial Markers,” Annals of the Association of American Geographers, 64, no. 4 (1974): 491–505. 69. See Henry E. Smith, “The Language of Property: Form, Context, and Audience,” Stanford Law Review 55 (2003): 1105–1191, 1116. 284

notes to pages 90 – 97

70. Not that the boundary is entirely irrelevant to nuisance actions. See Henry E. Smith, “Exclusion and Property Rules in the Law of Nuisance,” Virginia Law Review 90 (2004): 965–1049. 71. See Henry E. Smith, “Exclusion Versus Governance: Two Strategies for Delineating Property Rights,” Journal of Legal Studies 31, no. S2 (2002): 453–488. 72. Tom Ginsburg and Richard H. McAdams, “Adjudicating in Anarchy: An Expressive Theory of International Dispute Resolution,” William and Mary Law Review 45 (2004): 1229–1339. 4. Law’s Focal Power in Dynamic Perspective 1. Thomas C. Schelling, The Strategy of Conflict (Cambridge, MA: Harvard University Press, 1980): 98. 2. Ibid., 57. 3. See André Casajus, Focal Points in Framed Games: Breaking the Symmetry (Berlin: Springer-­Verlag, 2001); Michael Bacharach, “The Variable Frame Theory of Focal Points: An Experimental Study,” Games and Economic Behavior 19, no. 1 (1997): 1–45; Maarten Janssen, “Rationalising Focal Points,” Theory and Decision, 50, no. 2 (2001): 119–148; Robert Sugden, “A Theory of Focal Points,” Economic Journal 105, no. 433 (1995): 533–550. For a critique of modeling focal points (even his own theory, it would seem), see Robert Sugden, “The Role of Inductive Reasoning in the Evolution of Conventions,” Law & Philosophy 17, no. 4 (1998): 377–410. 4. See, e.g., H. Peyton Young, Individual Strategy and Social Structure: An Evolutionary Theory of Institutions (Princeton: Princeton University Press, 1998); Drew Fudenberg and David K. Levine, The Theory of Learning in Games (Cambridge, MA: MIT Press, 2008); Brian Skyrms, Evolution of the Social Contract (New York: Cambridge University Press, 1996). 5. Skyrms, Evolution of the Social Contract, 102–103. 6. John Maynard Keynes, “The Theory of Money and the Foreign Exchanges,” in A Tract on Monetary Reform (London: MacMillan, 1924). 7. Robert Sugden, “The Evolutionary Turn in Game Theory,” Journal of Economic Methodology 8, no. 1 (2001): 113–130. 8. Paul Mahoney and Chris William Sanchirico, “Norms, Repeated Games, and the Role of Law,” California Law Review 91 (2003): 1281–1329; Ken Binmore and Larry Samuelson, “The Evolution of Focal Points,” Games and Economic Behavior 55, no. 1 (2006): 21–42. 9. See Schelling, The Strategy of Conflict, 67–68. 10. Gerry Mackie, “Female Genital Cutting: The Beginning of the End,” in Female Circumcision: Multidisciplinary Perspectives, ed. Bettina Shell-­Duncan and Ylva Hernlund (Boulder, CO: Lynne Reinner Publishers, 2000), 245–282. 285

notes to pages 97–103

See generally Gerry Mackie and John LeJeune, “Social Dynamics of Abandonment of Harmful Practices: A New Look at the Theory,” Innocenti Working Paper No. 2009–06, Florence, UNICEF Innocenti Research Centre; Gerry Mackie, “Ending Footbinding and Infibulation: A Convention Account,” American Sociological Review 61, no. 6 (1996): 999–1017; Michael Chwe, Rational Ritual: Culture, Coordination, and Common Knowledge (Princeton: Princeton University Press, 2001). 11. See, e.g., City of Daytona Beach v. Tona-­R ama, Inc., 294 So. 2d 73 (Fla. 1974); In re Ashford, 440 P. 2d 76 (Haw. 1968); State ex rel. Thornton v. Hay, 462 P. 2d 671 (Or. 1969); Matcha v. Mattox, 711 S.W.2d 95 (Tex. App. 1986). 12. Jon Elster, Unwritten Constitutional Norms (2010) (unpublished manuscript, cited by permission), http://www.ucl.ac.uk/laws/jurisprudence/docs/2010 /Elster24Feb2010.pdf. 13. See H. Peyton Young, “The Economics of Convention,” Journal of Economic Perspectives 10, no. 2 (1996): 105–122, 106; Mick Hamer, “Left is Right on the Road: The History of Road Traffic Regulations,” New Scientist 112, 1540–41 (1986): 16–18. 14. See Ilana R. Azulay Chertok and Melanie L. Hoover, “Breastfeeding Legislation in States with Relatively Low Breastfeeding Rates Compared to Breastfeeding Legislation of Other States,” Journal of Nursing Law 13, no. 2 (2009): 45–53, 50; See also Li et al, “Public Beliefs about Breastfeeding Policies in Various Settings,” Journal of the American Dietetic Association 104, no. 7 (2004): 1162–1168. 15. See http://www.nycclash.com/. 16. See Jordan Raphael, “The Calabasas Smoking Ban: A Local Ordinance Points the Way for the Future of Environmental Tobacco Smoke Regulation,” Southern California Law Review 80 (2007): 393–424, 400 (Berkeley City Council enacted such a law in 1977); William E. Lambert, Jonathan M. Samet, and John D. Spengler, “Environmental Tobacco Smoke Concentrations in No-­Smoking and Smoking Sections of Restaurants,” American Journal of Public Health 83, no. 9 (1993): 1339–1341; “Restaurants Adjusting to Law That Segregates, Bans Smokers,” New Hampshire Union Leader, July 6, 1993. 17. For general explorations of incrementalism and “divide and conquer” techniques, see Eric A. Posner, Kathryn E. Spier, and Adrian Vermeule, “Divide and Conquer,” Journal of Legal Analysis 2, no. 2 (2010): 417–472; Saul Levmore, “Interest Groups and the Problem with Incrementalism,” University of Pennsylvania Law Review 158 (2010): 815–858. 18. See, e.g., Patrick Kabat, “   ‘Till Naught but Ash is Left to See’: Statewide Smoking Bans, Ballot Initiatives, and the Public Sphere,” Yale Journal of

286

notes to pages 103 –109

Health Policy, Law, and Ethics 9 (2009): 128–200; Marot Williamson, “When One Person’s Habit Becomes Everyone’s Problem: The Battle over Smoking Bans in Bars and Restaurants,” Villanova Sports and Entertainment Law Journal 14 (2007): 161–190; Luis R. Rivero et al, “Towards the World-­wide Ban of Indoor Cigarette Smoking in Public Places,” International Journal of Hygiene and Environmental Health 209, no. 1 (2005): 1–14; 19. See, e.g., Tim Bartley, “Transnational Governance as the Layering of Rules: Intersections of Public and Private Standards,” Theoretical Inquiries in Law 12, no. 2 (2011): 517–542; Oren Perez, “Private Environmental Governance as Ensemble Regulation: A Critical Exploration of Sustainability Indexes and the New Ensemble Politics,” Theoretical Inquiries in Law 12, no. 2 (2011): 543–580. 20. See Richard R. W. Brooks and Carol M. Rose, Saving the Neighborhood: Racially Restrictive Covenants, Law, and Social Norms (Cambridge, MA: Harvard University Press, 2013), 13. 21. Shelley v. Kraemer, 334 U.S. 1 (1948). 22. See Richard W. Brooks, “Covenants without Courts: Enforcing Residential Segregation with Legally Unenforceable Agreements,” American Economic Review 101, no. 3 (2011): 360–365. 23. On convention, see Richard H. McAdams, “Conventions and Norms: Philosophical Aspects,” in ed. N. J. Smelser and P. B. Baltes, International Encyclopedia of the Social and Behavioral Sciences, Vol. 4, (Oxford: Elsevier Science 2001), 2735–41. 24. See Henry E. Smith, “Community and Custom in Property,” Theoretical Inquiries in Law 10, no. 1 (2009): 5–41; Henry E. Smith, “Custom in American Property Law: A Vanishing Act,” Texas International Law Journal 48 (2013): 507–522; Frederick Schauer, “The Jurisprudence of Custom,” Texas International Law Journal 48 (2013): 523–534. 25. David Lewis, Convention: A Philosophical Study (Cambridge, MA: Harvard University Press, 1969). See also Brian Skyrms, Signals: Evolution, Learning, and Information (Oxford: Oxford University Press, 2011). 26. Of course, there has long been a conflict among lexicographers as to whether a dictionary should be descriptive or prescriptive. See, e.g., Jack Lynch, The Lexicographer’s Dilemma (New York: Walker and Company, 2009); Rickie Sonpal, Note, “Old Dictionaries and New Textualists,” Fordham Law Review 71 (2003): 2177–2226. 27. See, e.g., Eric A. Posner, Social Norms and the Law (Cambridge: Harvard University Press, 2000); Richard H. McAdams and Eric Rasmusen, “Norms and the Law,” in The Handbook of Law and Economics, ed. A. Mitchell Polinsky & Steve M. Shavell (North Holland: Elsevier Science, 2007), 1573– 1618; Gerald J. Postema, “Custom, Normative Practice, and the Law,” Duke

287

n o t e s t o p a g e s 1 0 9 – 111

Law Journal 62 (2012–2013): 707–738; Lewis, Convention; Richard H. McAdams, “Conventions and Norms.” 28. Gerald J. Postema, “Coordination and Convention at the Foundations of Law,” Journal of Legal Studies 11, no. 1 (1982): 165–203, 178. 29. Smith, “Community and Custom in Property,” 8. 30. Ibid., 8–12. 31. David Hume, A Treatise of Human Nature (Oxford, Oxford University Press, 2000). 32. See, e.g., Jack Hirshleifer, Economic Behavior in Adversity (Chicago, IL: University of Chicago Press, 1987), 223–234; James E. Krier, “Evolutionary Theory and the Origin of Property Rights,” 95 Cornell Law Review 95 (2009): 139–159; Robert Sugden, The Economics of Rights, Co-­operation and Welfare (Oxford: Basil Blackwell, 1986), 55–103; Kenton K. Yee, “Ownership and Trade from Evolutionary Games,” International Review of Law and Economics 23, no. 2 (2003): 183–97. 33. Carol Rose rejects the HD model because it does not attend to the differences in possession and property. See Carol Rose, “Psychologies of Property (and Why Property is not a Hawk-­Dove Game),” in Philosophical Foundations of Property Law ed., James Penner and Henry Smith, (Oxford: Oxford University Press, 2014), 272–288. Yet the criticism is overstated. She is trying to explain what property is now, given centuries of recognition. The model offers to explain the micro-­foundations of the origin of property, the logical first step for which was the respect for possession, even though humans eventually respect a more complex convention. 34. See Robert C. Ellickson, “Property in Land,” Yale Law Journal 102 (1993): 1315– 1400, 1328–30; Carol M. Rose, “Possession as the Origin of Property,” University of Chicago Law Review 52 (1985): 73–88, 81–82. 35. Henry E. Smith, “The Language of Property: Form, Context, and Audience,” Stanford Law Review 55 (2003): 1105–1191, 1115. 36. Ibid. 37. See Rose, “Possession as the Origin,” 82–88. 38. See George J. Klir and Tina A. Folger, Fuzzy Sets, Uncertainty, and Information (Englewood Cliffs, NJ: Prentice Hall, 1988); Lotti Zadeh, “Fuzzy Sets,” Information & Control 8, no. 3 (1965): 338–353. 39. Pierson v. Post, 3 Cai. R. 175 (N.Y. Sup. Ct. 1805). 40. The court ruled for Pierson, holding that one must bring the animal within “certain control” before one possesses and therefore owns it. 41. See Robert C. Ellickson, Order Without Law: How Neighbors Settle Disputes (Cambridge, MA: Harvard University Press, 1991), 191–206 (whaling); Rose, “Possession as the Origin,” 77 n.21 (oil and gas); Smith, “The Language of Property,” 1117–1125 (foxes, whales, and manure). 288

n o t e s t o p a g e s 111 – 11 4

42. See Tom Ginsburg and Richard H. McAdams, “Adjudicating in Anarchy: An Expressive Theory of International Dispute Resolution,” William and Mary Law Review 45 (2004): 1229–1339, 1258–1259. 43. See Richard Craswell, “Do Trade Customs Exist?” in The Jurisprudential Foundations of Corporate and Commercial Law, ed. Jody S. Krauss and Steven D. Walt (Cambridge: Cambridge University Press, 2000), 118–148. 44. See Rose, “Possession as the Origin,” 79. See discussion of international examples in Ginsburg and McAdams, “Adjudicating in Anarchy,” 1260–1261. 45. Lawyers are familiar with the idea that the facts of two cases are never exactly alike. See Sugden, “The Role of Inductive Reasoning in the Evolution of Conventions.” 46. See Binmore and Samuelson, “Evolution of Focal Points,” 21–42. 47. Geoffrey Garrett and Barry R. Weingast, “Ideas, Interests, and Institutions: Constructing the European Community’s Internal Market,” in Ideas and Foreign Policy: Beliefs, Institutions, and Political Change, ed. Judith Goldstein and Robert O. Keohane (Ithaca, NY: Cornell University Press, 1993). 48. William B. Hall, Jr., “The Evolution of Whist,” The Sewanee Review 3 (1985): 457–467, 460–461. 49. See World Chess Federation, Laws of Chess, http://www.fide.com/component /handbook/?id=124&view=article. Making a connection to legal reasoning, the second sentence of the rules states: “Where cases are not precisely regulated by an Article of the Laws, it should be possible to reach a correct decision by studying analogous situations which are discussed in the Laws.” Ibid. 50. Don H. Doyle, “Rules of Order: Henry Martyn Robert and the Popularization of American Parliamentary Law,” American Quarterly 32, no. 1 (1980): 3–18. 51. See Jacob E. Gersen and Eric A. Posner, “Soft Law: Lessons from Congressional Practice,” Stanford Law Review 61 (2008): 573–627. 52. 4 U.S.C. § 5 (emphasis added). 53. 4 U.S.C. §§ 6–9. 54. For example, the code expresses a preference for disposing of a worn flag through burning (not burial), 4 U.S.C. § 8(k); states that when hung in a building lobby to which there is more than one main entrance, the stars should face east, 4 U.S.C. § 7(o); and specifies the number of days for which the flag should be flown at half-­mast after the death of certain federal officials, 4 U.S.C. § 7(m). 55. Compare Thomas W. Merrill and Henry E. Smith, “Optimal Standardization in the Law of Property: The Numerus Clausus Principle,” Yale Law Journal 110 (2000): 1–70. 56. For example, an early California statute adopted mining customs. See Morton v. Solambo Copper Mining Co., 26 Cal. 527 (Cal. 1864) (citing Section 621 of the Practice Act). 289

n o t e s t o p a g e s 11 4 – 1 2 2

57. Richard A. Epstein, “Possession as the Root of Title,” Georgia Law Review 13 (1979): 1221–1243. 58. See, e.g., Pierre-­Marie Dupuy, “Soft Law and the International Law of the Environment,” Michigan Journal of International Law 12 (1991): 420–435. 59. See, e.g., The Paquete Habana, 175 US 677, 701 (1900) (consulting “some leading French treatises on international law,” including Pistoye and Duverdy’s Treatise on Maritime Prizes, 1855, to determine the international law of prizes constituted “by the general consent of civilized nations.”); United States v Smith, 18 US 153, 163, n. h (1820) (citing domestic and foreign treatises and other scholarly works to determine the international law of piracy). 60. See Ginsburg and McAdams, “Adjudicating in Anarchy,” 1255–1256. 61. See John Fabian Witt, Lincoln’s Code: The Laws of War in American History (New York: Free Press, 2012); The Lieber Code, “General Orders No. 100, Instructions for the Government of Armies of the United States in the Field” (1863), http://avalon.law.yale.edu/19th_century/lieber.asp. 62. See Robert Axelrod, “More Effective Choice in the Prisoner’s Dilemma,” Journal of Conflict Resolution 24, no. 3 (1980): 379–403. 63. For a review, see McAdams and Rasmusen, “Norms and the Law,” 1573–1618. 64. Robert C. Ellickson distinguishes second-­and third-­party enforcement in Order Without Law, 126–127, where he also supplies evidence of both types of enforcement in property disputes between neighbors. For game theory models of third-­party enforcement, see Mahoney and Sanchirico, “Norms, Repeated Games, and the Role of Law,” 1281–1329; Robert Boyd, Herbert Gintis, and Samuel Bowles, “Coordinated Punishment of Defectors Sustains Cooperation and Can Proliferate When Rare,” Science 328, no. 5978 (2010): 617–620. 65. Gillian K. Hadfield and Barry R. Weingast, “What Is Law? A Coordination Model of the Characteristics of Legal Order,” Journal of Legal Analysis 4, no. 2 (2012): 471–514. 66. See Ole Spiermann, International Legal Argument in the Permanent Court of International Justice: The Rise of the International Judiciary (Cambridge: Cambridge University Press, 2005). 67. See Attila Tanzil, “Problems of Enforcement of Decisions of the International Court of Justice and the Law of the United Nations,” European Journal of International Law 6, no. 1 (1995): 539. 68. Ginsburg and McAdams, “Adjudicating in Anarchy,” 1229–1339. 69. Empirical studies of legitimacy tend to miss this point when they study robust domestic institutions, like American law. 70. See ICJ Advisory Opinion on the Legality of the Threat or Use of Nuclear Weapons, 8th July, 1996, http://www.icj-­cij.org/docket/index.php?sum=498 &code=unan&p1=3&p2=4&case=95&k=e1&p3=5; Jasjit Singh, “Re-­examining 290

notes to pages 122 –129

the 1996 ICJ Advisory Opinion: Concerning the Legality of Nuclear Weapons,” Cadmus 1, no. 5 (2012): 158–165. 71. Schelling, “The Strategy of Conflict,” 66. 72. Jean-­Robert Tyran and Lars P. Feld, “Achieving Compliance When Legal Sanctions Are Non-­Deterrent,” Scandinavian Journal of Economics 108, no. 1 (2006): 135–156. 73. See Oliver Wendell Holmes, “The Path of the Law,” Harvard Law Review 10 (1897): 457, 461. 74. See Paul H. Robinson and John M. Darley, Justice, Liability and Blame: Community Views and the Criminal Law (Boulder, CO: Westview Press, 1995); Janice Nadler, “Flouting the Law,” Texas Law Review 83 (2005): 1399–1441; Elizabeth Mullen and Janice Nadler, “Moral Spillovers: The Effect of Moral Mandate Violations on Deviant Behavior,” Journal of Experimental Social Psychology 44, no. 5 (2008): 1239–1245. 75. See, for example, Jason Sunshine and Tom R. Tyler, “The Role of Procedural Justice and Legitimacy in Shaping Public Support for Policing,” Law & Society Review 37, no. 3 (2003): 513–548. 76. See, e.g., Marco Achetti and István Scheuring, “Review: Game Theory of Public Goods in One-­shot Social Dilemmas without Assortment,” Journal of Theoretical Biology 299 (April 2012): 9–20; John Cadigan, Patrick T. Wayland, Pamela Schmitt, and Kurtis Swope, “An Experimental Dynamic Public Goods Game with Carryover,” Journal of Economic Behavior and Organization 80, no. 3 (2011): 523–531. 77. The externality does not necessarily lead to inefficiency, as it is possible for an individual to incur net private benefits up to the point where the efficient level of the public good is produced. The point of “irrelevant externalities” is usually discussed for the case of negative externalities, but the logic applies to ­positive externalities. See Lee Anne Fennell, “The Problem of Resource Access,” Harvard Law Review 126 (2013): 1471–1531, 1521–22; David D. Haddock, “Irrelevant Externality Angst,” Journal of Interdisciplinary Economics 19, no. 1 (2007): 3–18. 78. See, e.g., John O. Ledyard, “Public Goods: A Survey of Experimental Research,” in The Handbook of Experimental Economics, ed. John H. Kagel and Alvin E. Roth (Princeton: Princeton University Press, 1995); Tibor Neugebauer, Javier Perote, Ulrich Schmidt, and Malte Loos, “Selfish-­biased Conditional Cooperation: On the Decline of Contributions in Repeated Public Goods Experiments,” Journal of Economic Psychology, 30, no. 1 (2009): 52–60. 79. Dan M. Kahan, “The Logic of Reciprocity: Trust, Collective Action, and Law,” Michigan Law Review 102 (2003): 71–103. 80. See Richard H. McAdams, “The Origin, Development, and Regulation of Norms,” Michigan Law Review 96 (1997): 338–433; Philip Pettit and Geoffrey 291

notes to pages 129 –137

Brennan, “The Hidden Economy of Esteem,” Economics and Philosophy 16, no. 1 (2000): 77–98; Tyler Cowen, “The Esteem Theory of Norms,” Public Choice 113, no. 1–2 (2002): 211–224. For a classical discussion, see See Adam Smith, The Theory of Moral Sentiments, ed. D. D. Raphael & A. L. Macfie (Oxford Univ. Press 1976) (1759), 62 (“To deserve, to acquire, and to enjoy the respect and admiration of mankind, are the great objects of ambition and emulation.”). 81. Samuel Bowles and Herbert Gintis, “Behavioural Science: Homo Reciprocans,” Nature 415 (2002): 125–128; Samuel Bowles and Herbert Gintis, “The Evolution of Strong Reciprocity: Cooperation in Heterogeneous Populations,” Theoretical Population Biology 65, no. 1 (2004): 17–28; Armin Falk and Urs Fischbacher, “A Theory of Reciprocity,” Games and Economic Behavior 54, no. 2 (2006): 293–315; Simon Gaechter and Ernst Fehr, “Fairness and Retaliation: The Economics of Reciprocity,” Journal of Economic Perspectives 14, no. 3 (2000): 159–181. 82. Roberto Galbiati & Pietro Vertova, “Obligations and Cooperative Behaviour in Public Good Games,” Games and Economic Behavior 64, no. 1 (2008): 146–170. 83. There was a decline in contributions from round one to round ten, but it was not statistically significant. Ibid., 156–157. See also Tyran and Feld, “Achieving Compliance When Legal Sanctions Are Non-­Deterrent,” 147–150; Iris Bohnet and Robert D. Cooter, “Expressive Law: Framing or Equilibrium Selection?” (Nov. 2003) KSG Working Paper No. RWP03–046 and UC Berkeley Public Law Research Paper No. 138, http://papers.ssrn.com/sol3/papers.cfm?abstract _id=452420. 84. Dennis Chong, Collective Action and the Civil Rights Movement (Chicago: University of Chicago Press, 1991). 85. Kahan, “The Logic of Reciprocity,” 80–85. 5. Legislation as Information 1. See, e.g., Robert C. Ellickson, “Of Coase and Cattle: Dispute Resolution among Neighbors in Shasta County,” Stanford Law Review 38 (1986): 623–687, 668–671 (documenting limited knowledge of trespass law among laymen property owners and professionals); Paul H. Robinson and John M. Darley, “Does the Criminal Law Deter? A Social Science Investigation,” Oxford Journal of Legal Studies 24, no. 2 (2004): 173–205 (reviewing evidence that people lack knowledge of criminal law). 2. See Michael Spence, “Job Market Signaling,” Quarterly Journal of Economics, 87, no. 3(1973): 355–374. A potential employee may signal her type by her willingness to invest years in education. Thus, signaling may solve a problem of adverse selection.

292

notes to pages 138 –139

3. See Hope Clinic v. Ryan, 195 F.3d 857, 876, 879–81 (7th Cir. 1999) (Posner, J., dissenting), cert. granted and judgment vacated, 530 U.S. 1271 (2000) (concluding that the proponents of a “partial-­birth abortion” ban were “concerned with making a statement in an ongoing war for public opinion, though an incidental effect may be to discourage some late-­term abortions.”). 4. I discussed an earlier version of some of the ideas of this chapter in three ­articles: Richard H. McAdams, “The Origin, Development, and Regulation of Norms,” Michigan Law Review, 96 (1997): 338–443, 397–408; Richard H. McAdams, “An Attitudinal Theory of Expressive Law,” 79 Oregon Law Review 79 (2000): 339–390; and Dhammika Dharmapala and Richard H. McAdams, “The Condorcet Jury Theorem and the Expressive Function of Law: A Theory of Informative Law,” American Law & Economic Review 5, no. 1 (2003): 1–31. Other contributions to this literature include Cass R. Sunstein, “On the Expressive Function of Law,” University of Pennsylvania Law Review 144 (1996): 2021–2053; Alex Geisinger, “A Belief-­Change Theory of Expressive Law,” Iowa Law Review 88 (2002): 35–73; Steven Shavell, “When Is Compli­ance With the Law Socially Desirable?,” Journal of Legal Studies 41 (Jan. 2012): 1–36; Jacob E. Gersen and Eric A. Posner, “Soft Law: Lessons from Congres­sional Practice,” Stanford Law Review 61 (2008): 573–627; Roland Benabou and Jean Tirole, “Law and Norms,” NBER Work­ ­ing Paper 17579 (2011). Dharmapala and I also argued that private (hate) speech might convey information about attitudes in ways that affect behav­ ­ior. See Dhammika Dharmapala and Richard H. McAdams, “Words That Kill? An Economic Model of the Influence of Speech on Behavior (with Particular Reference to Hate Speech),” Journal of Legal Studies 34, no. 1 (2005): 93–136. 5. See McAdams, “An Attitudinal Theory,” 339–390. 6. See, e.g., Ivo Bischoff and Henrik Egbert, “Social Information and Bandwagon Behavior in Voting: An Economic Experiment,” Journal of Economic Psychology 34 (Feb. 2013): 270–284. 7. See Geoffrey Brennan and Philip Pettit, The Economy of Esteem (Oxford: Oxford University Press, 2004); Tyler Cowen, “The Esteem Theory of Norms,” Public Choice 113, no. 1–2 (2002): 211–224; Richard H. McAdams, “The Origin, Development, and Regulation of Norms,” Michigan Law Review 96 (1997): 338–433. See also James S. Coleman, Foundations of Social Theory (1990): 130 (although it can bring various benefits, “[social] status, or recognition from others, has long been regarded by psychologists as a primary source of satisfaction to the self.”) 8. See, e.g., Tyler Cowen, What Price Fame? (Cambridge, MA: Harvard University Press, 2002); William J. Goode, The Celebration of Heroes: Prestige

293

n o t e s t o pag e s 139 –141

as a Social Control System (Berkeley, CA: University of California Press, 1978), 344–345; Richard E. Nisbett and Dov Cohen, Culture of Honor: The Psychology of Violence in the South (Boulder, CO: Westview Press, 1996); Richard H. McAdams, “Relative Preferences,” Yale Law Journal 102 (2002): 1–104. 9. See, e.g., Martha C. Nussbaum, Hiding from Humanity: Disgust, Shame, and the Law (Princeton: Princeton University Press, 2004); John Braithwaite, Crime, Shame and Reintegration (Cambridge: Cambridge University Press, 1989). 10. See, e.g., Mark Casson, The Economics of Business Culture: Game Theory, Transaction Costs, and Economic Performance (Oxford: Oxford University Press, 1991), 29–52; Peter H. Huang and Ho-­Mou Wu, “More Order Without More Law: A Theory of Social Norms and Organizational Cultures,” Journal of Law, Economics, and Organization 10, no. 2 (1994): 390–406; Robert D. Cooter, “The Theory of Market Modernization of Law,” International Review of Law and Economics 16, no. 2 (1996): 141–172. 11. See Eric A. Posner, Law and Social Norms (Cambridge, MA: Harvard Uni­ versity Press, 2000), 28–29 (describing the commitment model, in which an individual shows loyalty to members of a outsider group “by ostentatiously violating the norms of a dominant group”); Martin A. Leroch, “Punishment as Defiance: Deterrence and Perverse Effects in the Case of Expressive Crime,” CESifo Economic Studies doi: 10.1093/cesifo/ift009 (first published online, July 9, 2013). 12. For research finding politics to respond to the median voter, see, e.g., Robert S. Erikson, Michael B. MacKuen, and James A. Stimson, The Macro Polity (New York: Cambridge University Press, 2002); Robert S. Erikson, Gerald C. Wright, and John P. McIver, Statehouse Democracy: Public Opinion and Policy in the American States (New York: Cambridge University Press, 1993); Alan D. Monroe, “Consistency between Public Preferences and National Policy Decisions,” American Politics Quarterly 7, no. 1 (1979): 3–18; Alan D. Monroe, “Public Opinion and Public Policy 1980–1993.” Public Opinion Quarterly 68, no. 1 (1998): 6–28. By contrast, some recent research finds instead that American elite opinion dominates public policy. See Martin Gilens, Affluence and Influence: Economic Inequality and Political Power in America, (New York: Russell Sage Foundation and Princeton University Press, 2012); Larry M. Bartels, Unequal Democracy: The Political Economy of the New Gilded Age, (New York: Russell Sage Foundation and Princeton University Press, 2008); Martin Gilens & Benjamin I. Page, “Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens,” forthcoming Perspectives and Politics, March 7, 2014 draft, http://www.polisci.northwestern.edu 294

n o t e s t o pag e s 141–14 8

/people/documents/TestingTheoriesOfAmericanPoliticsFINALforProduction6 March2014.pdf. 13. See M. Dane Waters, Initiative and Referendum Almanac 12 (Chapel Hill, NC: Carolina Academic Press, 2003) (more than twenty states have a popular process to create or repeal legislation). 14. For example, if one knows 500 people locally, that would be 1 percent of the population of a town of 50,000 people, but if one knows an additional 1000 people nationally, that is less than one-­third of one-­thousandth of 1 percent of the national population. 15. See, e.g., Geoffrey Brennan and Loren Lomasky, Democracy and Decision: The Pure Theory of Electoral Preference (Cambridge: Cambridge University Press, 1993); Daniel A. Farber and Philip P. Frickey, Law and Public Choice (Chicago: The University of Chicago Press, 1991), 21–33. 16. See, e.g., Stanton A. Glantz, John Slade, Lisa A. Bero, Peter Hanauer, and Deborah E. Barnes, The Cigarette Papers (Berkeley: University of California Press, 1998), 431 (tobacco industry lost a San Francisco ballot initiative in 1983 despite spending $1.25 million, then a national record for a local ­measure). 17. See, e.g., Deborah A. Prentice and Dale T. Miller, “Pluralistic Ignorance and the Perpetuation of Social Norms by Unwitting Actors,” Advances in Experimental Social Psychology 28 (1996): 161–210; Dale T. Miller and Deborah A. Prentice, “Collective Errors and Errors About the Collective,” Journal of Personality and Social Psychology Bulletin 20, no. 5 (1994): 541–550; James M. Fields and Howard Schuman, “Public Beliefs About the Beliefs of the Public,” Public Opinion Quarterly 40, no. 4 (1976): 427–448; Hubert J. O’Gorman and Stephen L. Garry, “Pluralistic Ignorance—​A Replication and Extension,” Public Opinion Quarterly 40, no. 4 (1976): 449–458. 18. See Hubert J. O’Gorman, “Pluralistic Ignorance and White Estimates of White Support for Racial Segregation,” Public Opinion Quarterly 39, no. 3 (1975): 313–330. 19. See Prentice and Miller, “Pluralistic Ignorance;” Deborah A. Prentice and Dale T. Miller, “Pluralistic Ignorance and Alcohol Use on Campus: Some Consequences of Misperceiving the Social Norm,” Journal of Personality and Social Psychology 64, no. 2 (1993): 243–256. 20. See McAdams, “An Attitudinal Theory,” 367 n.66. 21. Timur Kuran, Private Truths, Public Lies: The Social Consequences of Preference Falsification (Cambridge, MA: Harvard University Press, 1995). 22. Ibid., 118–127. Kuran’s other main examples are the persistence of India’s caste system, 128–136, and what he describes as “the unwanted spread of affirmative action,” 137–54. 295

n o t e s t o pag e s 149 –151

23. See Timur Kuran and Cass R. Sunstein, “Availability Cascades and Risk Regulation,” Stanford Law Review 51 (1999): 683–768. 24. See Robert A. Kagan and Jerome H. Skolnick, “Banning Smoking: Compliance Without Enforcement,” in Smoking Policy: Law, Politics, and Culture, ed. Robert L. Rabin and Stephen D. Sugarman (New York: Oxford University Press, 1993), 69, 71–76. They state: [F]ormal rules serve an important communication function, overcoming the familiar problem of pluralistic ignorance and inaction. Even if, in 1983, most nonsmokers (and many smokers) favored restrictions on smoking in the workplace . . . ​individual nonsmokers may not have realized the extent of support for such rules, and hence may not have felt emboldened to complain directly to smokers. The enactment of ordinances and workplace rules told nonsmokers that they had a right to breathe air that was free of smoke.

Ibid., 86. 25. See, e.g., Dov Fox and Christopher Griffin, Jr., “Disability-­Selective Abortion and the ADA,” Utah Law Review 2009 (2009): 845–906. 26. For example, the Texas law begins with this “legislative finding”: “The legislature finds that breast-­feeding a baby is an important and basic act of nurture that must be encouraged in the interests of maternal and child health and family values.” VTCA 156.001. 27. See Stephanie Condon, “After 148 years, Mississippi finally ratifies 13th Amend­ment, which banned slavery,” CBS News, February 18, 2013, http:// www.cbsnews.com/8301-­250_162-­57569880/after-­1 48-­years-­mississippi-­finally -­ratifies-­13th-­amendment-­which-­banned-­slavery/; “How to repeal archaic sodomy laws?,” Contemporary Sexuality 34, no. 11 (2000): 1–7; Gabriel J. Chin et al., “Still on the Books: Jim Crow and Segregation Laws Fifty Years After Brown v. Board of Education,” Michigan State Law Review 2006 (2006): 457–476. 28. One experimental survey finds no evidence that knowledge of law influences the perceived disapproval from sharing trade secrets. See Yuval Feldman, “The Expressive Function of Trade Secret Law: Legality, Cost, Intrinsic Motivation, and Consensus,” Journal of Empirical Legal Studies 6, no. 1 (2009): 177–212. However, the study tests the law’s information effect just where the theory would predict it to be at its weakest, asking Silicon Valley employees about the approval of their Silicon Valley co-­workers, while making law salient by telling respondents about the “Uniform Trade Secrets Act” (UTSA). Given the size of the state, a California statute is extraordinarily weak evidence of the attitudes of one’s Silicon Valley co-­workers, especially when the statute was enacted almost twenty years before the survey was taken; it is even possible that 296

n o t e s t o pag e s 151–15 4

r­ espondents assumed the UTSA was federal law. See also Yuval Feldman and Janice Nadler, “The Law and Norms of File Sharing,” San Diego Law Review 43 (2006): 577–615 (mixed evidence for expressive effects of copyright law on college students, but the relevant law is federal). 29. Patricia Funk, “Is There an Expressive Function of Law?,” American Law and Economics Review 9, no. 1 (2007): 135–159. 30. Other supporting evidence comes from experimental surveys. See Adam S. Chilton, “The Influence of International Human Rights Agreements on Public Opinion: An Experimental Study,” forthcoming Chicago Journal of International Law, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2390183, (American respondents more likely to disapprove of solitary confinements after being told it might violate United States treaty obligations); Paul M. Sniderman and Thomas Piazza, The Scar of Race (Cambridge, MA: Harvard University Press, 1993): 131–133, 193 (respondents expressed significantly more support for minority contractor set-­aside programs if they were first told that Congress had enacted such a law). Compare Harold G. Grasmick, et al., “Shame and embarrassment as deterrents to noncompliance with the law: The case of an anti-­littering campaign,” Environment and Behavior, 23, no. 2 (1991): 233–251 (finding that an antilittering campaign that sought to evoke shame and embarrassment increased reported compliance with antilittering laws); Robert B. Cialdini et al., “A Focus Theory of Normative Conduct,” Advances in Experimental Social Psychology 24 (1991): 201–234 (merely reminding an individual of an antilittering norm significantly decreased the amount of littering). 31. See Dhammika Dharmapala & Richard H. McAdams, “A Theory of Informative Law,” 1–31. 32. A political science literature emphasizes this kind of retrospective judgment. See, e.g., V.O. Key, Jr., Public Opinion and American Democracy (New York: Knopf, 1961): 472–476; Morris Fiorina, Retrospective Voting in American National Elections (New Haven: Yale University Press, 1981); John Zaller, “Coming to Terms with V.O. Key’s Concept of Latent Opinion,” in Electoral Democracy, ed. Michael B. MacKuen and George Rabinowitz. (Ann Arbor: University of Michigan Press, 2003). 33. See Dan M. Kahan and Donald Braman, “Cultural Cognition and Public Policy,” Yale Law and Policy Review 24 (2006): 149–172; Dan M. Kahan and Donald Braman, “More Statistics, Less Persuasion: A Cultural Theory of Gun­Risk Perceptions,” University of Pennsylvania Law Review 151 (2003): 1291–1327; Dan M. Kahan, Donald Braman, Geoffrey L. Cohen, John Gastil, and Paul Slovic, “Who Fears the HPV Vaccine, Who Doesn’t, and Why? An Experimental Study of the Mechanisms of Cultural Cognition,” Law and Human Behavior 34, no. 6 (2010): 501–516; Dan M. Kahan, “The Secret Ambition of Deterrence,” Harvard Law Review 113 (1999): 413–500. 297

notes to pag es 154 –164

34. See; Brendan Nyhan, Jason Reifler, and Peter Ubel, “The hazards of correcting myths about health care reform,” Medical Care 51, no. 2 (2013) 127–32; Brendan Nyhan and Jason Reifler, “When Corrections Fail: The Persistence of Political Misperceptions,” Political Behavior, 32, no. 2 (2010): pp. 303–330; Dan Kahan, “Amazingly Cool and Important Article on Virulence of Ideologically Motivated Reasoning,” http://www.culturalcognition.net /blog/2013/1/11/amazingly-­cool-­important-­article-­on-­virulence-­of-­ideological .html; Edward Glaeser and Cass R. Sunstein, “Does More Speech Correct Falsehoods?,” Journal of Legal Studies, 43, no. 1 (2014): 65–93. 35. Edmund Burke articulated the distinction in advocating for his role as trustee. See Susan Dovi, Political Representation (revised 2011), (section 1.1 on “delegate vs. trustee”), http://plato.stanford.edu/entries/political-­representation /#DelVsTru. 36. Fox and Griffin, “Disability-­Selective Abortion,” 845–906. 37. See Dharmapala &. McAdams, “A Theory of Informative Law,” 1–31. For contributions to the political science literature on the Theorem, see Jason Brennan, “Condorcet’s Jury Theorem and the Optimum Number of Voters,” Politics 31, no. 2 (2011): 55–62; Bezalel Peleg and Shmuel Zahir, “Extending the Condorcet Jury Theorem to a General Dependent Jury,” Social Choice and Welfare 39, no. 1 (2012): 91–125; Sergeui Kaniovski, “Aggregation of Correlated Votes and Condorcet’s Jury Theorem,” Theory and Decision 69, no. 3 (2010): 453–468. 38. See Dharmpala and McAdams, “A Theory of Informative Law,” 6. 39. See, e.g., Sushil Bikhchandani, David Hirshleifer, and Ivo Welch, “Learning from the Behavior of Others: Conformity, Fads, and Informational Cascades,” Journal of Economic Perspectives 12, no. 3 (1998): 151–170; Sushil Bikhchandani, David Hirshleifer, and Ivo Welch, “A Theory of Fads, Fashion, Custom, and Cultural Change as Informational Cascades,” Journal of Political Economy 100, no. 5 (1992): 992–1026; A. V. Banerjee, “A Simple Model of Herd Behavior,” Quarterly Journal of Economics 107, no. 3 (1992): 797–817. 40. See Maggie Wittlin, “Buckling Under Pressure, An Empirical Test of the Expressive Effects of Law,” Yale Journal on Regulation, 28 (2011): 419–469. 41. See Dan M. Kahan, “The Logic of Reciprocity: Trust, Collective Action, and Law,” Michigan Law Review 102 (2003): 71–103, and sources cited therein. 42. See Yuval Feldman, “The Complexity of Disentangling Intrinsic and Extrinsic Compliance Motivations: Theoretical and Empirical Insights from the Behavioral Analysis of Law,” 35 Washington University Journal of Law and Policy 35 (2011): 11–51; Bruno S. Frey and Reto Jegen, “Motivation Crowding Theory,” Journal of Economic Surveys 15, no. 5 (2001): 589–611. 43. Uri Gneezy and Aldo Rustichini, “A Fine Is a Price,” The Journal of Legal Studies 29, no. 1 (2000): 1–17.

298

n o t e s t o pag e s 166 –171

44. Robert Cooter, “Models of Morality in Law and Economics: Self-­Control and Self-­Improvement for the ‘Bad Man’ of Holmes,” Boston University Law Review 78 (1998): 903–930; Kenneth G. Dau-­Schmidt, “An Economic Analysis of the Criminal Law as a Preference-­Shaping Policy,” Duke Law Journal 1990 (1990): 1–38. 45. Leon Festinger, “Cognitive Dissonance,” Scientific American 207, no. 4 (1962): 93–102. 46. See Dan M. Kahan, “Social Influence, Social Meaning, and Deterrence,” Virginia Law Review 83 (1997): 349–395; Lawrence Lessig, “Social Meaning and Social Norms,” University of Pennsylvania Law Review 144 (1996): 2181– 2189, 2187; Lawrence Lessig, “The Regulation of Social Meaning,” University of Chicago Law Review 62 (1995): 943–1045, 947. 47. Lessig, “Social Meaning,” 943–1045. 6. Revelation of Information by Legal Enforcement 1. See, e.g., Caldwell v. Mississippi, 472 U.S. 320, 331 (1985) (referring to a jury seeking to send a message “of extreme disapproval for the defendant’s acts” by imposing the death penalty); Brendan Brosh, “2 dogfighting convictions ‘send a message,’   ” New York Daily News, March, 9, 2010, http://www.nydailynews .com/new-­york/bronx/2-­dogfighting-­convictions-­send-­messagearticle-­1.174119 (message about dogfighting). 2. See Larry V. Starcher, “Choosing West Virginia’s Judges,” West Virginia Lawyer 12 (1998): 18–20, 19 n. 10 (stating that judges in thirty-­eight states are “subject to popular elections”). 3. See, e.g., Lee Epstein and Jack Knight, The Choices Justices Make (Washington, D.C.: Congressional Quarterly, Inc., 1998). 4. See In re Debs, 158 U.S. 564, 597 (1895). Justice Brewer referenced the dependence of courts on public opinion while explaining that it would have been “puerile and ridiculous” for Lincoln to have sought an end to the Civil War by suing the Southern States in federal court. 5. See, e.g., Thomas R. Marshall, Public Opinion and the Supreme Court (Boston, MA: Unwin Hyman, 1989); Robert A. Dahl, “Decision-­Making in a Democracy: The Supreme Court as a National Policy-­Maker,” Journal of Public Law 6 (1957): 279–295, 285; Roy B. Flemming and B. Dan Wood, “The Public and the Supreme Court: Individual Justice Responsiveness to American Policy Moods,” American Journal of Political Science 41, no. 2 (1997): 468–498; Mark A. Graber, “The Nonmajoritarian Difficulty: Legislative Deference to the Judiciary,” Studies in American Political Development 7, no. 1 (1993): 35–73; Michael J. Klarman, “Rethinking the Civil Rights and Civil Liberties Revolutions,” Virginia Law Review 82 (1996): 1–68.

299

n o t e s t o pag e s 171–177

6. See David Fontana and Donald Braman, “Judicial Backlash or Just Backlash? Evidence from a National Experiment,” Columbia Law Review, 112 (2012): 731–799; Dan Kahan, “Gentle Nudges vs. Hard Shoves: Solving the Sticky Norm Problem,” University of Chicago Law Review 67 (2000): 607–645; Emanuela Carbonara, Francesco Parisi, and Georg von Wangenheim, “Unjust Laws and Illegal Norms,” International Review of Law and Economics 32, no. 3 (2012): 285–299. 7. See Michael J. Klarman, From the Closet to the Altar: Courts, Backlash, and the Struggle for Same-­Sex Marriage (New York: Oxford University Press 2012), 165–192; Daryl J. Levinson, “Parchment and Politics: The Positive Puzzle of Constitutional Commitment,” Harvard Law Review 124 (2011): 657–746, 744. 8. See Roe v. Wade, 410 U.S. 113 (1973); Furman v. Georgia, 408 U.S. 238 (1972). 9. See Kelo v. City of New London, 545 U.S. 2655 (2005); Janice Nadler, Shari Seidman Diamond and Matthew M. Patton, “Government Takings of Private Property: Kelo and the Perfect Storm” in Public Opinion and Constitutional Controversy, ed. Nathaniel Persily, Jack Citrin and Patrick J. Egan (Oxford: Oxford University Press, 2008). 10. Loving v. Virginia, 388 U.S. 1 (1967). 11. Lawrence v. Texas, 539 U.S. 558 (2003). 12. Alexis De Tocqueville, Democracy in America, ed. J. P. Mayer, trans. George Lawrence (New York: Doubleday, 1969) (1835): 150–151. 13. See Richard H. McAdams, “The Origin, Development, and Regulation of Norms, Michigan Law Review 96 (1997): 338–433, 366–369; Eric Rasmusen, “Stigma and Self-­Fulfilling Expectations of Criminality,” Journal of Law and Economics 39, no. 2 (1996): 519–543. 14. Joel Feinberg, “The Expressive Function of Punishment,” in Doing & Deserving: Essays in the Theory of Responsibility (Princeton: Princeton University Press, 1970), 95–118. 15. Ibid., 95. 16. Ibid., 98. 17. Feinberg raises the need to justify the condemnation as well as the hard treatment, ibid., 115, and complicates the idea of proportionality (because the state might seek to make either the hard treatment or the condemnation “fit” the crime, ibid., 118). For other efforts to engage these normative questions, see Christopher Bennett, The Apology Ritual: A Philosophical Theory of Punishment (Cambridge: Cambridge University Press, 2008); Antony Duff, Punishment, Communication, and Community (Oxford: Oxford University Press, 2001); Jean Hampton, “The Moral Education Theory of Punishment,” Philosophy and Public Affairs 13, no. 3 (1984): 208–38; Igor Primoratz, “Punishment as Language,” Philosophy 64, no. 248 (1989): 187–205. For 300

n o t e s t o pag e s 177–18 5

c­ riticisms of expressive approaches to the justification of punishment, see David Boonin, The Problem of Punishment (New York: Cambridge University Press, 2008), 171–180; Nathan Hanna, “Say What? A Critique of Expressive Retributivism,” Law and Philosophy 27, no. 2 (2008): 123–150; A. J. Skillen, “How to Say Things with Walls,” Philosophy 55, no. 214 (1980): 509–23. 18. See David Garland, “Rethinking the Symbolic-­instrumental Distinction: Meanings and Motives in American Capital Punishment,” in Governance and Regulation in Social Life: Essays in Honour of W. G. Carson, ed. Augustine Brannigan and George Pavlich (Abingdon, UK: Routledge-­Cavendish, 2007), 109–130 19. Ibid., 124. 20. Ibid., 117. 21. Dan M. Kahan, “What Do Alternative Sanctions Mean?,” University of Chicago Law Review 63 (1996): 591–653. 22. He has since recanted this proposal. Relying on his cultural cognition work, Kahan now contends that shaming sanctions will never have widespread appeal. See Dan M. Kahan, “What’s Really Wrong with Shaming Sanctions?,” Texas Law Review 84 (2006): 2075–2095. See also Michael S. Kirsch, “Alternative Sanctions and the Federal Tax Law: Symbols, Shaming, and Social Norm Management as a Substitute for Effective Tax Policy,” Iowa Law Review 89 (2004): 863–939. 23. Garland, “Rethinking the Symbolic-­Instrumental Distinction,” 125. 24. Zatco v. California, 502 U.S. 16, 19–20 (1991). 25. Ibid. (emphasis added). 26. Jeannie Suk, At Home in the Law: How the Domestic Violence Revolution is Transforming Privacy, (New Haven, CT: Yale University Press, 2009). 27. Georgia v. Randolph, 547 U.S. 103, 112–113 (2006). 28. Ibid., 113. 29. Suk, At Home in the Law, 114–115. 30. Ibid. 31. Tennessee Williams, A Streetcar Named Desire (Sewanee: The University of the South, 1947). 32. See J.E.B. v. Alabama, 511 U.S. 127, 140–142 (1994). J.E.B. extended Batson v. Kentucky, 476 U.S. 79 (1986), which had invalidated the use of race-­based peremptory challenges, and Edmonson v. Leesville Concrete Co., Inc., 500 U.S. 614, 628 (1991), which applied Batson to parties in civil as well as criminal cases. 33. Ibid., 140, 142 (emphasis added). 34. J.E.B. also leaves parties free to select their lawyers or expert witnesses on the basis of gender, in order to pander to the jury’s gender stereotypes, which is additionally inconsistent with the expressive rationale. 301

notes to pages 188 –189

35. See United States v. Virginia, 518 U.S. 515 (1996) (invalidating state support of all-­male military academy); Mississippi University for Women v. Hogan, 458 U.S. 718 (1982) (invalidating state support of all-­female nursing school). 36. See, e.g., Lynch v. Donnelly, 465 U.S. 668 (1984) (O’Connor, J., concurring); County of Allegheny v. American Civil Liberties Union, Greater Pittsburgh Chapter, 492 U.S. 573, 592–594 (1989); McCreary County KY v. American Civil Liberties Union of Ky., 545 U.S. 844, 859–866 (2005). The endorsement test may no longer command the support of a majority of the justices. Although not deciding the issue, Town of Greece v. Galloway, 134 S. Ct. 1811 (2014), casts doubt on the test’s continued validity. 37. Capitol Square Review and Advisory Board v. Pinette, 515 U.S. 753, 811 (1995). 38. See “Trenton PD’s ‘Send a Message’ Campaign Nets 41 Arrests in 1 Weekend,” New Jersey Criminal Defense Lawyer, October 30, 2012, http://www.mynew jerseydefenselawyer.com/trenton-­pd%E2%80%99s-­%E2%80%9Csend-­a -­message%E2%80%9D-­campaign-­nets-­41-­arrests-­in-­1-­weekend/; Tom Gjelten, “FBI Tries To Send Message With Hacker Arrests,” National Public Radio, July 20, 2011, http://www.npr.org/2011/07/20/138555799/fbi-­arrests-­alleged -­anonymous-­hackers. 39. See John F. Decker, “Legislating New Federalism: The Call for Grand Jury Reform in the States,” Oklahoma Law Review 58 (2005): 341–395, 354 (“About . . . ​half of the states require a grand jury indictment for certain ­categories of crimes.”). 40. Jason Miles Levien and Stacie L. Fatka, “Cleaning up Judicial Elections: Examining the First Amendment Limitations on Judicial Campaign Regulation,” Michigan Law & Policy Review 2 (1997): 71, 74 (a majority of judges in thirty-­eight states are subject to popular elections); Steven W. Perry, “Prosecutors in State Courts, 2005,” Bureau of Justice Statistics Bulletin (July 2006), at 2, http://www.bjs.gov/content/pub/pdf/psc05.pdf (reporting that all states elect their local chief prosecutors except for Alaska, Connecticut, the District of Columbia, and New Jersey). 41. See Sparf v. United States, 156 U.S. 51, 143 (1895)(quoting from John Adams’ discussion of juries: “   ‘no man can be condemned of life or limb or property or reputation without the concurrence of the voice of the people.’   ”); Witherspoon v. Illinois, 391 U.S. 510, 519–520 (1968)(“[I]n a nation less than half of whose people believe in the death penalty, a jury composed exclusively of such people cannot speak for the community . . . . [S]uch a jury can speak only for a distinct and dwindling minority.”); United States v. Gilliam, 994 F.2d 97, 101 (2d Cir. 1997) (“Our constitution guarantees . . . ​trial by a jury . . . ​primarily in order to ensure that the accused is judged by prevailing community mores . . . . ​Without full knowledge of the nature of the crime, the jury cannot speak for the people or exert their authority.”). All emphases added. Compare 302

notes to pages 185 –188

Hicks v. Collins, 384 F.3d 204 (6th Cir. 2004) (upholding the prosecution’s argument that “it is time you sent a message to the community”). 42. United States v. Gilliam, 994 F.2d 97, 101 (2d Cir. 1997). 43. Pete Williams and Tracy Connor, “Holder Speaks Out against ‘Stand Your Ground’ Laws after Zimmerman Verdict,” NBC News, July 16, 2013, http: //usnews.nbcnews.com/_news/2013/07/16/19506188-­holder-­speaks-­out-­against -­stand-­your-­ground-­laws-­after-­zimmerman-­verdict?lite; Ta-­Nehisi Coates, “How Stand Your Ground Relates to George Zimmerman,” The Atlantic, July 16, 2013, http://www.theatlantic.com/national/archive/2013/07/how-­stand -­your-­ground-­relates-­to-­george-­zimmerman/277829/. 44. For a similar argument, see Louis N. Schulze, Jr., “Of Trayvon Martin, George Zimmerman, and Legal Expressivism: Why Massachusetts Should Stand Its Ground on ‘Stand Your Ground,’   ” New England Law Review On Remand 47 (2012): 34–41. 45. West’s F.S.A. § 776.013(1)(a). 46. West’s F.S.A. §§ 776.032, 776.085. 47. See West’s F.S.A. § 776.032; Erin Donaghue, “George Zimmerman Waives Right to ‘Stand Your Ground’ Hearing,” CBS News, April 30, 2013, http://www .cbsnews.com/8301-­504083_162-­57582067-­504083/trayvon-­martin-­case-­george -­zimmerman-­waives-­right-­to-­stand-­your-­ground-­hearing/. The law may have contributed to the delay of the arrest of Zimmerman, but that hardly explains the focus on SYG at the time of trial and after the acquittal. 48. West’s F.S.A. § 776.013(3). 49. American Courts have been using the entitlement to “stand one’s ground” as synonymous with the absence of a “duty to retreat” since at least Beard v. United States, 158 U.S. 550, 562–563 (1895), where the court quoted legal authorities as stating (emphasis added): “In the case of justifiable self-­defense, the injured party may repel force with force . . . ​[Because t]he defendant was where he had the right to be, . . . ​he was not obliged to retreat, nor to consider whether he could safely retreat, but was entitled to stand his ground.” See also People v. Lewis, 117 Cal. 186, 191–192, 48 P. 1088, 1089 (Cal. 1897): “[W]hile, at common law, there was a contrariety of opinion upon the part of the writers as to the duty of retreat . . . ​this state has upheld a defendant’s right to stand his ground, and meet by force a sudden and violent attack.” 50. Before the recent spate of “stand your ground” laws, the Florida Supreme Court noted that, while it required retreat, “a majority of jurisdictions do not impose a duty to retreat before a defendant may resort to deadly force when threatened with death or great bodily harm.” Weiand v. State, 732 So.2d 1044, 1049 (Fla.1999) (citing Wayne R. LaFave and Austin W. Scott, Jr., Substantive Criminal Law § 5.7(f) (West, 2d ed.1986)). See also Gillis v. United States, 400 A.2d 311, 312 (D.C. 1979)(“[P]robably the majority [of states] have adopted the 303

no tes t o pag es 189 –190

rule that one is not required to retreat but may stand his ground and defend himself. This has been called the American rule and in at least two cases the Supreme Court has indicated approval of it.”). 51. Graham Winch and Amanda Sloane, “Zimmerman: ‘I couldn’t see. I couldn’t breathe,’   ” HLN TV, July 2, 2013, http://www.hlntv.com/article /2013/07/01/george-­zimmerman-­trial-­trayvon-­martin-­day-­6. 52. See Wayne R. LaFave, Substantive Criminal Law § 10.4(f), 2nd ed. (St. Paul, MN: West, 2003) (“[E]ven in those jurisdiction which require retreat, the defendant need not retreat unless he knows he can do so in complete safety.”). The American Law Institute’s Model Penal Code requires retreat, but only if one knows one can retreat with complete safety. See MPC § 3.04(2)(b)(ii). 53. See, e.g., Davis v. Strack, 270 F.3d 111 (2d Cir. 2001) (“Under the terms of [New York law], the duty to retreat does not arise until the defendant forms a reasonable belief that the other person ‘is using or about to use deadly physical force.’   ”); Henderson v. State, 906 S.W.2d 589 (Tex. App. 1995) (“A defendant’s duty to retreat under [Texas Law] does not arise until the defendant believes deadly force was immediately necessary to protect the defendant against the other’s use or attempted use of unlawful deadly force.”). 54. See Margaret Raymond, “Looking for Trouble: Framing and the Dignitary Interest in the Law of Self-­Defense,” Ohio State Law Journal 71 (2010): 287–339, 293–295. 55. The Florida statute on provocation, West’s F.S.A. § 776.041, predates the adoption of the “stand your ground” law, as does one of the major cases defining the concept. See Gibbs v. State, 789 So.2d 443 (Fla. App. 4 Dist., 2001) (holding that, to defeat a self-­defense claim, the provocation must be by force or threat of force). 56. In a post-­trial interview, one Zimmerman juror mentioned “stand your ground” as a basis for the decision (“Well, because of the heat of the moment and the stand your ground. He had a right to defend himself. If he felt threatened that his life was going to be taken away from him or he was going to have bodily harm, he had a right.”). See “Exclusive Interview with Juror B-­37,” Anderson Cooper 360 Degrees, July 15, 2013, http://transcripts.cnn.com /TRANSCRIPTS/1307/15/acd.01.html. Some interpret this statement as evidence that the SYG rule mattered to the case. First, the juror appears to be using “stand your ground” as synonymous with the entire self-­defense defense. The statement is not evidence that the case would have come out differently had the retreat rule—​the subset of the doctrine SYG changes—​been different. Second, if the textual analysis is correct, the juror had to be confused to think that the retreat rule affected the judgment she was asked to make. Nonetheless, the juror’s statement is evidence of the importance of the law’s expressive effect, discussed next in the text. 304

no tes t o pag es 190 – 201

57. See Schulze, Jr., “Of Trayvon Martin, George Zimmerman, and Legal Expressivism.” 58. West’s F.S.A. § 776.013 (historical note, preamble). 59. See Erwin v. State, 29 Ohio St. 186, 199–200 (1876). 60. Dan M. Kahan, “The Secret Ambition of Deterrence,” Harvard Law Review 113 (1999): 413–500, 431. See also Dan M. Kahan and Donald Braman, “The Self-­Defensive Cognition of Self-­Defense,” American Criminal Law Review 45 (2008): 1–65. 61. Kenworthey Bilz, “The Puzzle of Delegated Revenge,” Boston University Law Review 87 (2007): 1059–1112. 62. Jonathon M. Karpoff and John R. Lott, Jr., “The Reputational Penalty Firms Bear from Committing Criminal Fraud,” in The Economics of Organised Crime, ed. Gianluca Fiorentini and Sam Peltzman (Cambridge: Cambridge University Press, 1995); Eliezer M. Fich and Anil Shivdasani, “Financial fraud, director reputation, and shareholder wealth,” Journal of Financial Economics 86, no. 2 (2007): 306–336. 63. See Adar v. Smith, 597 F.3d. 697 (5th Cir. 2010). 64. Bill Barrow, “State to appeal gay adoption ruling,” The Times-­Picayune, February 22, 2010, http://www.nola.com/crime/index.ssf/2010/02/state_to _appeal_gay_adoption_r.html. 65. See The Nation’s Leading English Language Advocates, Official English Map, http://www.proenglish.org/official-­english/state-­profiles.html. 66. See Presidential Executive Order No. 13166, 65 Federal Register 50121, (Aug. 11, 2000); City of New York, Office of the Mayor, Executive Order No. 120 (July 22, 2008). See also Kevin Johnston, “Nashville Mayor Takes Stand Against English Only Ordinance,” ImmigrationProf Blog, February 19, 2007, http://lawprofessors .typepad.com/immigration/2007/02/nashville_mayor.html. 7. The Power of Arbitral Expression 1. See Richard H. McAdams, “The Expressive Power of Adjudication,” University of Illinois Law Review 2005 (2005): 1043–1122. 2. Martin Shapiro, Courts, A Comparative and Political Analysis (Chicago: The University of Chicago Press, 1981), 1. 3. Ibid. Even nonhuman primates have third parties intervene to manage violent conflict. See, e.g., Jessica C. Flack, Michelle Girvan, Frans B. M. de Waal, and David C. Krakauer, “Policing Stabilizes Construction of Social Niches in Primates,” Nature 439, no. 7075 (2006): 426–429. 4. See, e.g., Sean Wilentz, The Age of Reagan: A History 1974–2008 (New York: HarperCollins, 2001), 18; Yanek Mieczkowski, Gerald Ford and the Challenges of the 1970s (Lexington, KY: The University Press of Kentucky, 2005), 18. 305

notes to pages 201– 203

5. See, e.g., Cass R. Sunstein, “Introduction: Of Law and Politics” in The Vote: Bush, Gore and The Supreme Court, ed. Cass R. Sunstein and Richard A. Epstein (Chicago: The University of Chicago Press, 2001), 4. 6. Tom Ginsburg and Richard H. McAdams, “Adjudicating in Anarchy: An Expressive Theory of International Dispute Resolution,” William and Mary Law Review 45 (2004): 1229–1340, 1310–1311. 7. See, e.g., Robert E. Hudec, Enforcing International Trade Law: The Evolution of the Modern GATT Legal System (Salem, NH: Butterworth Legal Publishers, 1993), 286; M. K. Bulterman and M. Kuijer, eds., Compliance with Judgments of International Courts (The Hague: Kluwer Law International, 1996), 35 (most decisions complied with); John G. Collier and Vaughan Lowe, eds., The Settlement of Disputes in International Law: Institutions and Procedures (New York: Oxford University Press, 1998), 178 (“all decisions were, sooner or later, complied with”); Warren F. Schwartz and Alan O. Sykes, “The Economic Structure of Renegotiation and Dispute Resolution in the World Trade Organization,” Journal of Legal Studies 31, no. 1 (2002): S179–S204, S200 (“[T]he level of compliance with trade commitments is quite high.”). 8. See William I. Miller, Bloodtaking and Peacemaking: Feud, Law, and Society in Saga Iceland (Chicago: The University of Chicago Press, 1990), 228. See also David Friedman, Law’s Order: What Economics Has to Do with Law and Why It Matters, (Princeton: Princeton University Press, 2000), 263–267; Richard A. Posner, “Medieval Iceland and Modern Legal Scholarship,” Michigan Law Review 90 (1992): 1495–1511 (reviewing Miller’s book). 9. Miller, Bloodtaking and Peacemaking, 232. 10. Ibid., 20–21. 11. Ibid., 234. 12. Ibid., 236. 13. Ibid. 14. Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge: Cambridge University Press, 1990), 100–101. 15. Ibid., 180. 16. Ibid., 101. 17. Andrea McDowell, “Real Property, Spontaneous Order, and Norms in the Gold Mines,” Law and Social Inquiry 29, no. 4 (2004): 771–818. See also Andrea McDowell, “From Commons to Claims: Property Rights in the California Gold Rush,” Yale Journal of Law and the Humanities 14 (2002): 1–72. Regarding the dispute resolution that occurred during the long overland trip to the California mining camps, see generally John Phillip Reid, Law for the Elephant: Property and Social Behavior on the Overland Trail (San Marino, CA: Huntington Library, 1980).

306

notes to pag es 203 – 210

18. Peter T. Leeson, “An-­arrgh-­chy: The Law and Economics of Pirate Organization,” Journal of Political Economy 115, no. 6 (2007): 1049–1094, 1065. 19. Or the figure might model a hiring decision, where two business partners disagree over which individual to hire, but each prefers their second choice to the failure to hire anyone. 20. Adapted from Jennifer Gerarda Brown and Ian Ayres, “Economic Rationales for Mediation,” Virginia Law Review 80 (1994): 323–402. 21. At NS, either player would be worse off being the only one to switch strategies because the results NN and SS are worse for each. The same is true of SN. Note that the example assumes that the players cannot make side payments, as where they might face liquidity constraints or dislike the idea of “selling” their heirlooms even to each other. 22. The theory here bears a similarity to Robert Aumann’s concept of a correlated equilibrium. See Robert J. Aumann, “Correlated Equilibrium as an Expression of Bayesian Rationality,” Econometrica 55, no. 1 (1987): 1–18; Robert J. Aumann, “Subjectivity and Correlation in Randomized Strategies,” Journal of Mathematical Economics 1, no. 1 (1974): 67–96. The concept was first used to explain arbitral dispute resolution by Roger B. Myerson, Game Theory: Analysis of Conflict (Cambridge, MA: Harvard University Press, 1991), 244–258. See also Brown and Ayres, “Economic Rationales for Mediation,” 323. 23. Assume p = probability that the other player plays S. Cindy’s expected value from playing N is p(3) + (1 − p)(1) and her expected value from playing S is p(1) + (1 − p)(2). Setting these equations equal to one another (because the equilibrium is reached only at a probability in which neither strategy is better than the other), p = 1/3. Cindy is indifferent between her strategies when George plays S with probability 1/3 and N with probability 2/3. Given that the payoffs for each player are symmetrical, George is indifferent when Cindy plays the strategies with the same probabilities. At this equilibrium, the expected value of either strategy is 5/3. 24. This logical problem does not exist when there are three or more possible equilibria. With three equilibria, for example, announcing one omits the other two. If there is nothing to distinguish between the two omitted strategies, the announcements singles out only the strategy it endorses. 25. Michael Suk-­Young Chwe, Rational Ritual: Culture, Coordination, and Common Knowledge (Princeton: Princeton University Press, 2003). 26. If p is the probability that the other player plays Dove, Player 1’s expected value from playing Dove is p(2) + (1 − p)(0) and her expected value from playing Hawk is p(4) + (1 − p)(–1). The same is true for Player 2. Setting these equations equal to one another (because the equilibrium is reached only at a probability

307

notes to pag es 210 – 222

in which neither action is better than the other), we get p = 1/3. At this equilibrium, the expected value of either strategy is 2/3. 27. One possibility I will not explore here is that the arbiter fashions an outcome that gives the loser slightly more than his mixed strategy equilibrium. 28. McAdams and Nadler, “Testing the Focal Point Theory,” 102–103. 29. See, e.g., Barbara Sainty, “Achieving Greater Cooperation in a Noisy Prisoner’s Dilemma: An Experimental Investigation,” Journal of Economic Behavior and Organization 39, no. 4 (1999): 421–435; Jianzhong Wu and Robert Axelrod, “How to Cope with Noise in the Iterated Prisoners’ Dilemma,” Journal of Conflict Resolution 39, no. 1 (1995): 183–189. 30. Compare Dhammika Dharmapala and Richard H. McAdams, “Words That Kill? An Economic Model of the Influence of Speech on Behavior (with Particular Reference to Hate Speech),” Journal of Legal Studies 34 (2005): 93–136. 31. Brown and Ayres explicitly assume, for this part of their analysis, that the parties have no private information, which means that the dispute is genuine. Brown and Ayres, “Economic Rationales for Mediation,” 372. 32. See, e.g., Lucian A. Bebchuk and Alon Klement, “Negative Expected Value Suits,” in Procedural Law and Economics, vol. 8, 2nd ed., ed. Chris W. Sanchirico (Cheltenham: Edward Elgar Publishing Limited, 2012). 33. See Heather C. Lench and Shane W. Bench, “Automatic Optimism: Why People Assume Their Futures Will Be Bright,” Social and Personality Psychology Compass 6, no. 4 (2012): 347–360; Ying Zhang, Ayelet Fishbach, and Ravi Dhar, “When Thinking Beats Doing: The Role of Optimistic Expectations on Goal-­Based Choice,” Advances in Consumer Research 33 (2006): 57–58; Joan Costa-­Font, Elias Mossialos, and Caroline Rudisill, “Optimism and the Perceptions of New Risks,” Journal of Risk Research 12, no. 1 (2009): 27–41. 34. Legal theorists struggle to understand the popular objection to resolving disputes by randomization, which seems tolerable in a few contexts (e.g., draft lotteries and assignment of judges), but usually unthinkable. See Jon Elster, “Solomonic Judgments: Against the Best Interests of the Child,” University of Chicago Law Review, 54 (1987): 1–45, (defending a role for randomization in child custody decisions, but acknowledging that “[m]any people seem to think the proposal is inhuman, frivolous, or both”); Adam M. Samaha, “Randomization in Adjudication,” William & Mary Law Review 51 (2009): 1–86. 35. International Court of Justice, Maritime Delimitation in the Area between Greenland and Jan Mayen (Denmark v. Norway), http://www.icj-­cij.org/docket /index.php?sum=401&code=gjm&p1=3&p2=3&case=78&k=e0&p3=5 36. Ibid. 308

notes to pages 222 – 239

37. Andrea Isoni, Anders Poulsen, Robert Sugden, and Kei Tsutsui, “Focal Points in Tacit Bargaining Problems: Experimental Evidence,” European Economic Review 59 (April 2013): 167–188. 38. Andrew T. Guzman, “A Compliance-­Based Theory of International Law,” California Law Review 90 (2002): 1823–1887, 1847–1850. 39. See Rachel Brewster, “The Limits of Reputation on Compliance,” International Theory 1, no. 2 (2009): 323–333; G. W. Downs and M. A. Jones, “Reputation, Compliance, and International Law,” Journal of Legal Studies 31, no. 1 (2002): S95–S114. 40. Military and Paramilitary Activities in and against Nicaragua (Nicaragua v. United States of America), Jurisdiction and Admissibility, Judgment, I.C.J. Reports 1984, p. 392 (1986): 54. 41. Miller, Bloodtaking and Peacemaking, 245. 42. Leeson, “An-­arrgh-­chy,” 1074. 43. Geoffrey Garrett and Barry R. Weingast, “Ideas, Interests, and Institutions: Constructing the European Community’s Internal Market,” in Ideas and Foreign Policy: Beliefs, Institutions, and Political Change, ed. Judith Goldstein and Robert O. Keohane (Ithaca, NY: Cornell University Press, 1993), 179–186. 44. Ibid., 183. 45. Ginsburg and McAdams, “Adjudicating in Anarchy,” 1303–1326. 46. McDowell, “Real Property, Spontaneous Order, and Norms in the Gold Mines,” 778–803. See also McDowell, “Property Rights in the California Gold Rush,” 1–48. 8. Normative Implications 1. See Gary S. Becker, “Crime and Punishment: An Economic Approach,” Journal of Political Economy 76, no. 2 (1968): 169–217. 2. Louis Kaplow, “Rules Versus Standards: An Economic Analysis,” Duke Law Journal 42 (1992): 557–629; Kathleen M. Sullivan, “The Justices of Rules and Standards,” Harvard Law Review 106 (1992): 22–123. 3. See Stanko v. State, 974 P. 2d 1132 (Mont. 1998) (striking down the statutory requirement of “reasonable and proper” speed as unconstitutionally vague). 4. See Henry E. Smith, “The Language of Property: Form, Context, and Audience,” Stanford Law Review 55 (2003): 1105–1191, 1110–1111, 1182–1183. 5. Dov Fox and Christopher Griffin, Jr., “Disability-­Selective Abortion and the ADA,” Utah Law Review 2009 (2009): 845–906. 6. See Jordan Blair Woods, “A Decade After Drug Decriminalization: What Can the United States Learn From the Portugal Model?,” University of the District of Columbia Law Review 15 (2011): 1–31, 16–17; “Portugal’s Drug Policy: Treating, Not Punishing,” Economist 392, no. 8646 (2009): 43–44. 7. Woods, “A Decade After Drug Decriminalization,” 24–25. 309

n o t e s t o pag e s 2 4 0 – 2 41

8. Deborah Hellman, When is Discrimination Wrong?, (Cambridge MA: Harvard University Press, 2008); Deborah Hellman, “The Expressive Dimension of Equal Protection,” Minnesota Law Review 85 (2000), 1–69. 9. Hellman, When is Discrimination Wrong?, 8. More precisely: “Demeaning action thus requires (1) an expression of the unequal humanity of the other and (2) that the speaker occupy a position of status such that this expression is one that can put the other down.” Ibid., 38. 10. Ibid., 25. 11. Ibid., 166 (“[T]he intention of the person differentiating among people is irrelevant to whether the differentiation distinguishes on the basis of any particular trait and . . . ​is irrelevant to the moral assessment of this action.”); Ibid., 81 (“[T]his account of wrongful discrimination grounds the moral wrong in whether a practice is objectively demeaning, not in whether an individual or group of people feels demeaned or stigmatized.”). More generally, she rejects the importance of intent at p. 138–168 and she rejects the importance of effects on p. 21–27. 12. See Washington v. Davis, 426 U.S. 229, 240 (1976); Julia Kobick, “Discriminatory Intent Reconsidered: Folk Concepts of Intentionality and Equal Protection Jurisprudence,” Harvard Civil Rights-­Civil Liberties Law Review 45 (2010): 517–562, 522 (“If a plaintiff can prove that the government intended to discriminate on the basis of race despite the facially neutral action, the Court will apply a strict scrutiny analysis . . .”). 13. See, e.g., Kenneth L. Karst, Belonging to America: Equal Citizenship and the Constitution (New Haven, CT: Yale University Press, 1989), 13; Reva Siegel, “Equality Talk: Antisubordination and Anticlassification Values in Consti­ tutional Struggles over Brown,” Harvard Law Review 117 (2004): 1470–1547; Andrew Koppelman, Antidiscrimination Law and Social Equality (New Haven, CT: Yale University Press, 1998), 57–114. Federal law prohibiting racial discrimination in employment has a version of each of the alternatives: the doctrine of disparate treatment condemns intentional discrimination while disparate impact reaches employment practices that have an adverse and unjustified effect on racial groups. See, e.g., Ricci v. DeStefano, 557 U.S. 557, 577 (2009). 14. Hellman, When is Discrimination Wrong?, 41, 75–79 15. Ibid., 41, 77–78, 83. At an earlier time, Hellman alluded to the “fair conditions” specified by Jurgen Habermas, but she does not raise that standard in her book. See Hellman, “The Expressive Dimension of Equal Protection,” 23 (citing Jurgen Habermas, Moral Consciousness and Communicative Action (Christian Lenhardt & Shierry Weber Nicholsen translation) (Cambridge, MA: MIT Press 1990) (1983)). 16. See U.S. Const., Art. I, § 9, cl. 3 (“No Bill of Attainder or ex post facto Law shall be passed.”). As another example, Chief Justice Roberts decided that the 310

n o t e s t o pag e s 2 41– 2 4 3

levy the Affordable Care Act imposes on those who do not buy health insurance is a mere tax, as distinguished from a penalty or punishment, and therefore could be justified by the federal government’s power to tax. National Federation of Independent Business v. Sebelius, 132 S. Ct. 2566, 2600, 2608 (2012). Taxes are not punishments, even when the government would like to discourage the activity taxed, such as tobacco consumption. 17. Flemming v. Nestor, 363 U.S. 603 (1960). 18. See Hawker v. New York, 170 U.S. 189 (1898) (stating that forbidding an ex-­ felon from obtaining a license is not punishment because “the conviction of felony is evidence of the unfitness of such persons as a class”); Hudson v. United States, 522 U.S. 93 (1997) (finding that the monetary and occupational sanctions for misapplication of bank funds did not constitute criminal punishment); Fong Yue Ting v. United States, 149 U.S. 698, 730 (1893)(“The order of deportation is not a punishment for crime. It is not a banishment . . .”); Smith v. Doe, 538 U.S. 84 (2003) (holding that the Alaska Sex Offender Registration Act was not punitive and therefore did not violate the ex post facto clause). But in Commonwealth v. Cory, 454 Mass. 559 (Mass. 2009), the Massachusetts Supreme Court ruled 4–3 that it was punishment to require a sex offender to wear a GPS device during probationary period. 19. Joel Feinberg, “The Expressive Function of Punishment,” in Doing and Deserving: Essays in the Theory of Responsibility (Princeton: Princeton University Press, 1970), 95–118. 20. Ibid., 98 (emphasis added). 21. Ibid., 114. 22. See, e.g., County of Allegheny v. ACLU, 492 U.S. 573, 594–97 (1989) (the Court must “ascertain whether ‘the challenged governmental action is sufficiently likely to be perceived by adherents of the controlling denominations as an endorsement, and by the nonadherents as a disapproval, of their individual religious choices’   ” (quoting School Dist. of Grand Rapids v. Ball, 473 U.S. 373, 390 (1985)). See also Lynch v. Donnelly, 465 U.S. 668 (1984) (O’Connor, J., concurring). The Court treated endorsement as part of its Establishment Clause doctrine in cases such as School District of City of Grand Rapids v. Ball, 473 U.S. 373 (1985); Capital Square Review and Advisory Board v. Pinette, 515 U.S. 753 (1995); Santa Fe Independent School Dist. v. Doe, 530 U.S. 290 (2000); and McCreary County, Ky. v. American Civil Liberties Union of Ky., 545 U.S. 844 (2005). Since Justice O’Connor’s departure from the Court, however, the endorsement test may no longer command the support of a majority of the justices. Although not deciding the issue, Town of Greece v. Galloway, 134 S. Ct. 1811 (2014), casts doubt on the test’s continued validity. 23. See, e.g., Samuel J. Levine, “A Look at the Establishment Clause through the Prism of Religious Perspectives: Religious Majorities, Religious 311

notes to pages 243 – 250

Minorities, and Nonbelievers,” Chicago-­Kent Law Review 87 (2012): 775– 809; Alan Brownstein, “Continuing the Constitutional Dialogue: A Dis­ cussion of Justice Stevens’s Establishment Clause and Free Exercise Jurisprudence,” Northwestern University Law Review 106 (2012): 605–656, 638–643. 24. Salazar v. Buono, 559 U.S. 700, 721 (2010). 25. There is a distinct inquiry into whether the government intended to endorse religion, but there is also a suggestion that this matters only because one expects an actual audience to pick up on the intended meaning. Lynch v. Donnelly, 465 U.S. 668, 690 (1984) (O’Connor, J., concurring). 26. See Hellman, When is Discrimination Wrong?, 84: (“If objectivity is possi­ ­ble but poor judgment likely, we would have reason to substitute other tests for political or judicial decision makers . . . . ​But . . . ​to adopt this ­viewpoint, one must have a prior conception of what counts as the right ­judgment.”). 27. Ibid., 63–68. 28. See Lawrence Friedman, A History of American Law (New York: Simon and Schuster, 1985). 29. See Hope Clinic v. Ryan, 195 F.3d 857, 876, 879–81 (7th Cir. 1999) (Posner, J., dissenting), cert. granted and judgment vacated, 530 U.S. 1271 (2000). 30. See David Garland, “Rethinking the Symbolic-­Instrumental Distinction: Meanings and Motives in American Capital Punishment,” in Governance and Regulation in Social Life: Essays in Honour of W. G. Carson, ed., Augustine Brannigan and George Pavlich, (Abington, UK: Routledge-­ Cavendish, 2007). 31. Cass Sunstein, “On the Expressive Function of Law,” University of Pennsylvania Law Review 144 (1996): 2021–2053. 32. See Martin A. Leroch, “Punishment as Defiance: Deterrence and Perverse Effects in the Case of Expressive Crime,” CESifo Economic Studies (July 9, 2013): doi:10.1093/cesifo/ift009. 33. For discussions of this kind of cascade of intolerance, see Eric A. Posner, Law and Social Norms (Cambridge, MA: Harvard University Press, 2000), 133–147; Timur Kuran, Private Truths, Public Lies: The Social Consequences of Prefer­ ence Falsification (Cambridge, MA: Harvard University Press, 1995); B. Douglas Bernheim, “A Theory of Conformity,” Journal of Political Economy 102 (1994); 841–877. See also Richard H. McAdams, “Conformity to Inegalitarian Con­ ventions and Norms: The Contribution of Coordination and Esteem,” The Monist 88 (2005): 238–59. 34. See Daniel D. Polsby, “What if This Is as Good as It Gets?,” Green Bag 2 (1998): 115–123, 122 (reviewing Richard A. Epstein, Principles for a Free Society:

312

notes to pages 250 – 253

Reconciling Individual Liberty with the Common Good (Reading, MA: Perseus Books, 1998)). 35. See Jonathan Simon and Christina Spaulding, “Tokens of Our Esteem: Aggravating Factors in the Era of Deregulated Death Penalties,” in The Killing State: Capital Punishment in Law, Politics, and Culture, ed., Austin Sarat (New York: Oxford University Press, 1999), 81. 36. See Garland, “Rethinking the Symbolic-­Instrumental Distinction: Meanings and Motives in American Capital Punishment,” 166. 37. See Dan M. Kahan, “Ideology, Motivated Reasoning, and Cognitive Reflection,” Judgment and Decision Making 8, no. 4 (2013): 407–424; Dan M. Kahan and Donald Braman, “Cultural Cognition and Public Policy,” Yale Law and Policy Review 24 (2006): 149–172; Dan M. Kahan and Donald Braman, “More Statistics, Less Persuasion: A Cultural Theory of Gun-­Risk Perceptions,” University of Pennsylvania Law Review 151 (2003): 1291–1327. 38. See Presidential Proclamation—​Cesar Chavez Day, http://m.whitehouse.gov /the-­press-­office/2011/03/30/presidential-­proclamation-­cesar-­chavez-­day; National Day of the American Cowboy, http://www.govtrack.us/congress /bills/113/sres191; Confederate Memorial Day in the United States, http://www .timeanddate.com/holidays/us/confederate-­memorial-­day; Jamie Malanowski, “Misplaced Honor,” New York Times, May 25, 2013, http://www.nytimes.com /2013/05/26/opinion/sunday/misplaced-­honor.html?_r=0. 39. See, e.g., Committee for Public Ed. and Religious Liberty v. Nyquist, 413 U.S. 756 (1973); Levitt v. Committee for Public Ed. and Religious Liberty, 413 U.S. 472 (1973). It might seem that funding would inevitably endorse religion, but the issues are too complex to make such a general claim. The funding of math education in parochial schools might work to subsidize religion, even if the funding were generally understood expressively to favor math, not religion. 40. Sometimes it is said that strict scrutiny is “fatal in fact.” See, e.g., Missouri v. Jenkins, 515 U.S. 70, 121 (1995) (Thomas, J., concurring) (“[W]e must subject all racial classifications to the strictest scrutiny, which [excepting two war-­time cases] . . . ​has proven automatically fatal.”); City of Richmond v. J.A. Croson Co., 488 U.S. 469, 552 (1989) (Marshall, J., concurring) (“[S]trict scrutiny . . . ​is strict in theory, but fatal in fact.”). Yet this phrasing is contested. See Grutter v Bollinger 539 U.S. 306, 326 (2003) (“Strict scrutiny is not “strict in theory, but fatal in fact.”); Adarand Constructors, Inc. v. Peña, 515 U.S. 200, 237 (1995) (same). 41. Michael C. Dorf, “Same-­Sex Marriage, Second Class Citizenship, and Law’s Social Meanings,” Virginia Law Review 97 (2011): 1267–1346. 42. Ibid., 1279. 43. Ibid., 1286.

313

notes to pages 253 – 256

44. See, e.g., Kenneth L. Karst, “The Supreme Court, 1976 Term—​Foreword: Equal Citizenship under the Fourteenth Amendment,” Harvard Law Review 91 (1977): 1–68, 48–53; Owen Fiss, “Groups and the Equal Protection Clause,” Philosophy & Public Affairs 5, no. 2 (1976): 107; Garrett Epps, “Of Constitutional Seances and Color-­Blind Ghosts,” North Carolina Law Review 72 (1994): 401–451; Cass R. Sunstein, “The Anticaste Principle,” Michigan Law Review 92 (1994): 2410–2455. 45. See Richard H. McAdams, “Cooperation and Conflict: The Economics of Group Status Production and Race Discrimination,” Harvard Law Review 108 (1995): 1003–1084. 46. See Christopher L. Eisgruber, “Political Unity and the Powers of Government,” UCLA Law Review 41 (1994): 1297–1336, 1304–1306 (“The Establishment Clause turns out to have a surprising affinity to the Titles of Nobility Clauses,” which prohibits Congress from granting aristocratic titles to American citizens). 47. See Lynch v. Donnelly, 465 U.S. 668 (1984) (O’Connor, J., concurring). See also County of Allegheny v. ACLU, 492 U.S. 573, 594–97 (1989). 48. Lynch, 465 U.S. at 688 (O’Connor, J., concurring). 49. See, e.g., Richard C. Schragger, “The Relative Irrelevance of the Establishment Clause,” Texas Law Review 89 (2011), 583–651 (emphasizing the extent to which the Establishment Clause is underenforced). 50. Eisgruber, “Political Unity,” 1305. 51. Ibid.

314

Ac k n o w l e d g m e n t s

I have been discussing the ideas in this book for so long with so many astute scholars that I could not possibly remember all those to whom I owe a debt of gratitude. But I will try. I am particularly indebted to my gifted and generous co-­authors on projects related to the power of legal expression: Dhammika Dharmapala, Tom Ginsburg, and Janice Nadler. One could not hope for better collaborators in economics, political science, and psychology. I learned a great deal from each of them, without which the book could not have been written. I owe a special thanks to the wise and wonderful Tom Ulen, who commented on many prior related articles and then on every draft chapter. The book and I would be noticeably inferior without his input. I thank Jack Knight for, long ago, encouraging me to put these ideas in book form. Substantial portions of the faculties at the University of Chicago Law School and the University of Illinois College of Law have contributed to this project, with invaluable comments on prior papers and draft chapters, as did individuals at various other universities. In particular, I must thank Bruce Ackerman, Matt Adler, Ian Ayres, Hugh Baxter, Stefan Bechtold, Omri Ben-­Shahar, Douglas Baird, Bob Cooter, Jacob Corré, Rachel Croson, Bob Ellickson, Yuval Feldman, David Garland, Mitu Gulati, Gillian Hadfield, Gerard Hertig, Aziz Huq, Dan Kahan, Kim Krawiec, Andy Leipold, Saul Levmore, Anna Marshall, Jonathan Masur, Martha Nussbaum, Philip Pettit, Randy Picker, Eric Posner, Neil Richards, Arden Rowell, Rich Schragger, Henry Smith, Larry Solum, Deron Teichman, Jim Whitman, and two anonymous referees. Over the years, I received insightful commentary when presenting these ideas at various venues, including participants of the law school faculty workshops at Arizona State University, Boston University, Case-­Western, University of Connecticut, Duke, Florida State, Northwestern, Seton Hall, UCLA, the University of North Carolina, University of Toronto, University of Virginia, Washington University, and Yale University, as well as participants at the Hoffinger Criminal Justice Colloquium at NYU; the American Law and Economics 315

A C KNOW L E D G M EN T S

Association Meetings at Toronto and Georgetown; a Seminar on the Expressive Function of Law, ETH Zurich; the first Meeting of the Spanish Association of Law and Economics; the Center for Law, Economics and Organization at USC; a Law & Social Norms conference at Fordham; a Conference on New and Critical Approaches to Law & Economics at the University of Oregon; and an Olin Conference on The Legal Construction of Norms at the University of Virginia. I wish to acknowledge the generous research support of the Robert B. Roesing Faculty Research Fund. The empirical research with Janice Nadler was supported by the National Science Foundation (#0351530) and the American Bar Foundation. I received exemplary research assistance from Elpitha Betondo, Kayla Gamin, Lauren Jaffe, Caroline Malone, Peggy Olsen, David Ogles, Bill Watson, Michael Zhu, and Kevin Zimmerman. As with all things, I wrote this book powered by the love and support of my family and friends. Deepest gratitude to my parents Rick McAdams and Anne Holt McAdams, and to my wife Anna Marshall.

316

Index

abortion regulation, 12, 16, 251; and attitudinal signaling, 100, 138, 145, 238, 249 According to Hoyle, 112 ambiguity: in custom, 9, 107–109, 109–112, 118, 222, 224–225; legal resolution of, 112–117; in statutes, 197 Anderson, Elizabeth, 15 animal territoriality, 87–90 arbiters, 9, 91, 199, 200–232; and impartiality, 50, 227–229 arbitral expression: as an alternative to legitimacy theory, 199–204; and focal points, 199–200, 204–212, 214–221, 227–229, 229– 232; and information revelation, 200, 204, 212–214, 214–223, 228–229, 229–232; and the synergy of focal points and information revelation, 9, 214–223; and reputation, 223–227, 230 Arpaio, Joe, 58 Assurance game, 42, 52, 105, 127, 197; compared to a Prisoners’ Dilemma game, 31–32, 38–39; definition of, 32–34; modeling constitutional law, 71; modeling public goods, 128–132, 133– 135; Prisoners’ Assurance game, 33 attitudinal signaling, 137–138, 139–152, 155–156, 160, 237, 256; and executive or judicial signaling, 170–173, 187–188, 193–195 audience meaning: and choice of meaning in normative issues, 19–20, 240–243, 243–248,

253; and the irrelevance of esoteric knowledge, 21, 182–183, 190–191, 241, 245; as a requirement for legal expression to affect behavior, 19–21, 170, 179, 180–185, 235–236 availability cascade, 149, 153 Azor, Levy, 23–24 Battle of the Sexes game (BOS), 37, 38, 43, 108, 118, 227; definition of, 34–36; embedded in a Prisoners’ Dilemma game, 40–42, 69; in the experimental literature, 46, 51–56, 60; modeling arbitration, 205–208, 209; modeling constitutional law, 71; modeling a contract dispute, 65–66; modeling international law, 67–70; Prisoners’ Battle of the Sexes game, 35 Bayesian updating, 147–148 Bilz, Kenworthey, 192–193 Blackmun, Harry, 183–185 Braman, Don, 14, 154 breastfeeding laws, 86, 100, 145, 150, 154, 155, 157, 197 Brooks, Rick, 105–106 Bystander Example, 45, 48, 61, 96, 99, 123; as a Battle of the Sexes game, 46–48; and cheap talk, 44, 208; as a Hawk–Dove game, 37, 85–86; initial description of, 23–26; and mutual salience, 25–26, 46, 62, 77, 210, 215

317

index

cheap talk, 44, 50, 54–56, 118, 131, 146; and arbi­ tral focal points, 204–208, 211, 212; and com­ peting focal points, 208–209; after an existing equilibrium, 96–97; in treaty or constitution making, 68, 71, 73 84 Chicken game. See Hawk–Dove game (HD) child safety seat, 140–141, 146, 148, 153, 155, 157, 160, 197 Chong, Dennis, 132 codification, 8, 109–115, 222 cognitive bias. See availability cascade; opti­ mism bias common knowledge, 25–26, 45, 55, 75, 96, 97–99, 20–209 Condorcet Jury Theorem, 153, 157–158, 180, 187, 215, 219 constitutional law, 11, 12, 171, 185, 196; amend­ ments, 99; focal power of, 67, 71–76, 84, 99, 106; as within the range of legal expression, 8, 22, 27, 57, 67; and unconstitutional stat­ utes, 150–151, 197, 201; and symbolic conflict, 248–59. See also Equal Protection Clause; Establishment Clause conventions, 83, 149, 231, 260; coordination through, 29, 46, 58, 87, 126–127; as context, 241–243, 245, 247; as focal points competing with law, 9, 94, 95, 97–100, 100–106, 106–117, 135. See also customs, norms conventional meaning, 19–21, 233, 239, 240–248, 253 Cooter, Robert, 266, 270, 279, 292, 294, 299 Crawford, Vincent, 51–56 criminal law, 9, 11, 127, 133–134, 145, 177. See also punishment crowding out, 138, 164–165 cultural cognition theory, 14, 154–155, 251 customs, 79, 86, 182, 199, 236, 260; and arbitra­ tion, 212–213, 221–222, 224–227, 229; enforce­ ment of, 22, 27; legal clarifications of, 106–117; as focal points competing with the law, 9, 27, 94, 95, 97–100, 106–117, 123, 135, 148; legal codification of, 8, 109–115, 222; tyranny of, 256. See also conventions, norms Darley, John, 3, 127, 133 deterrence: 20, 27, 169–170, 177–178, 183, 193; as an account of legal compliance, 2–4, 92;

and enforcement signaling, 174–175; as an insufficient explanation for legal compliance, 7, 9, 16–17, 79, 106, 175; interaction with legal expression, 7, 124, 168; and violations sig­ naling, 163–165, 175, 261. See also sanctions Dharmapala, Dhammika, 157, 159 dictionaries, 107–109, 112 discontinuous effects, 148–150, 172, 237 discrimination, 8, 100, 145; based on disability, 157, 238; based on gender, 54, 172, 183–185; based on race, 16, 105–106, 151, 167, 171; based on sexual orientation, 137–138, 150–151, 153, 172–173, 195, 197; normative theories of, 15–16, 240–241, 246–247. See also Equal Protection dispute resolution. See arbiters; arbitral ­expression Dorf, Michael, 252–253, 258–259 endorsement test, 185, 233, 242–243, 248, 254–259. See also Establishment Clause enforcement signaling, 173–195. See also ­violations signaling Equal Protection Clause: and normative claims, 12, 15, 241–243, 245, 247; and race ­discrimination, 240–243, 245, 252, 253; and sex discrimination, 183–185. See also ­discrimination equilibrium, definition of, 30 Establishment Clause, 10, 15, 185, 233, 240, 242, 243, 248, 252–259. See also endorsement test ex post facto laws, 240, 241–243, 247–248, 261 expressive externalities. See spillovers expressive rent seeking, 250–252, 259 expressive theories. See expressive–politics theory of law; expressive theory of law’s effects; normative theory of expressive con­ duct; normative theory of expressive law expressive theory of law’s effects, 13, 14, 18, 20, 136, 198. See also focal point; information revelation expressive–politics theory of law, 13, 70, 177, 249 Feinberg, Joel, 15, 176–177, 241–244, 247 flag burning, 249–250, 256 focal points, 8–9, 22–27, 57, 97, 137, 148, 233, 260–261; and animal territoriality, 87–90; and arbitration, 199–200, 204–212, 214–223,

318

index

227–229, 229–232; and the clarification of cus­ ­tom, 106–117; and conditions of focal point theory of law’s effects, 62–63; and constitu­ tional law, 67, 71–76, 84, 99, 106; distin­ guished from legitimacy, 48–50, 63–67, 68, 75, 77, 79, 94, 96, 103–106; experimental lit­ erature, 42–43, 45, 50–56, 63–67; interaction with information theory, 196–198, 214–223; interaction with preference–shaping or social meaning change, 165–168; and international law, 67–70, 76, 84, 91–92, 112, 114–117, 120– 122, 196; interaction with legitimacy, 117, 119–127, 127–135; and law generally, 62–63, 82–87, 90–91, 196–198; limits of, 81, 92–93, 197–198, 227–228; and normative implica­ tions, 234–237, 240; and objections to the focal point theory of legal expression, 94, 95–96, 96–100; and the origins of legal sanc­ tions, 60–61, 75, 117–119; power of, 42–50; and smoking regulation, 85, 102–105, 122, 125, 150, 197; and social movements, 102– 106; and traffic regulation, 5–6, 26, 76–82, 136, 196. See also Schelling, Thomas Folk Theorem, 30, 40–41 Fourteenth Amendment. See Equal Protection Clause Fox, Dov, 157, 238 Funk, Patricia, 151–152

Hawk–Dove game (HD), 38–39, 42, 111, 117, 120; definition of, 36–37; effect of reputation on, 223–227; experiments involving, 54–55, 64–65; and law’s focal influence in everyday disputes, 76–92, 110; modeling arbitration, 206, 211, 212, 223–227, 227–228, 230; modeling smoking regulation, 100–102; Prisoner’s Hawk– Dove game, 36; and shouting matches, 84–86, 101; and violence, 82–86, 91, 230 Hellman, Deborah, 15, 241–242, 243–247 helmet laws, 17, 138, 145, 153, 164, 197 herding, 147, 160–162, 187 Hobbes, Thomas, 71

Galbiati, Roberto, 131, 132 game theory, 50, 61, 97, 115, 206, 230; experi­ mental literature of, 50–57; explaining focal points, 5, 8, 22, 30, 42–43, 47; evolu­ tionary, 95–96, 106; and legitimacy, 127–128, 197 Garland, David, 177, 178 Garrett, Geoffrey, 40–41, 69, 112, 230–231 Ginsburg, Tom, 91, 120–121, 201, 231 Gneezy, Uri, 51, 164 Griffin, Christopher, 157, 238 gun regulation, 14, 145, 150, 154, 172–173, 251 Gusfield, Joseph, 13–14 Guzman, Andrew, 225–226

information revelation, 9, 19–20, 136, 179, 212– 213, 231, 233, 260–261; and anti–discrimina­ tion laws or judgments, 137–138, 145, 150–151, 153, 167, 172–173, 183–185, 195, 197; and arbi­ tration, 200, 204, 212–214, 214–223, 228–229, 229–232; and attitudinal signaling, 137–138, 139–152, 170–173, 187–188, 193–195, 237, 256; and Bayesian updating, 147–148; and execu­ tive or judicial signaling, 169–196; and exper­­ tise versus aggregation, 155; interaction with focal point theory, 196–198, 214–223; interac­ tion with preference–shaping or social mean­ ­ing change, 165–168; limits of, 179–187, 197–198, 227–229; and normative implications of, 237– 239, 240, 253, 255–259; and risk signaling, 138, 153–162, 172–173; and smoking regulation, 143, 145, 146, 150, 153–157, 160, 164, 197; and traffic regulation, 6–8, 136, 138, 156–157, 196– 197; and violations signaling, 138–139, 162–165, 174–175 International Court of Justice (ICJ), 68, 91–92, 120–122, 201, 222, 227, 231 international law, 112, 120–122, 201, 221–223, 225–227, 230–231; customary, 114–115; focal power of 67–70, 76, 84, 91–92, 196; of war, 68, 115–116; as within the range of legal expression, 8, 11, 22, 27, 57 Isoni, Andrew, 53, 67, 90 iterated Prisoners’ Dilemma, 40, 67–70, 112, 115, 212–213, 230–231. See also Prisoners’ Dilemma

habit, 1–3, 36, 78, 166 Hadfield, Gillian, 117

juries: and information revelation, 173, 186–187, 187–193; selection of, 12, 169, 183–185 319

index

and signaling, 148, 150, 156, 256–257; strength­­ ened by the law, 16, 150, 156, 260. See also customs, conventions normative theory of expressive conduct, 13, 15 normative theory of expressive law, 13, 15, 16, 21, 177

Kahan, Dan, 16, 154, 167, 177–178, 191 Keynes, John Maynard, 45, 96 Kuran, Timur, 148–149 leadership, 59–61, 75–76, 98–99, 104, 118, 123; in experiments to test expressive effects, 54–55 legislators, 61, 72, 171, 180, 186–187; and attitu­ dinal signaling, 137–138, 139–152; and legal legitimacy, 133–135; and risk signaling, 138, 153–162, 238–239; and violations signaling, 162–163 legitimacy, 61, 63, 82, 85, 144, 238; as an account of legal compliance, 3–4, 22, 193; and the cod­­ification of custom, 108, 112, 113, 115; as distinct from expressive power, 48–50, 63–67, 75, 77, 79, 94, 96, 103–106; as an insufficient explanation for legal or arbitral compliance, 6–7, 9, 16–17, 20–27, 87–89, 137–138, 199–200, 200–204, 208, 221–223, 229–232, 248, 261; interaction with legal expression, 117, 118, 119–135, 174–175, 197– 198, 227–229; and social meaning change, 167–168. See also procedural legitimacy; ­substantive legitimacy Lessig, Larry, 166–167 Lieber Code, 115–116

O’Connor, Sandra Day, 242, 254 optimism bias, 219 Ostrom, Elinor, 202

Martin, Trayvon, 188–192 McDowell, Andrea, 203, 231 meaning. See audience meaning; conventional meaning; social meaning; speaker’s meaning Mill, John Stuart, 256, 258, 259 Miller, William, 202, 230 mixed strategy, 47–48, 55, 206, 208, 209–211, 228 mutual salience, 22, 25–26, 45–46, 55, 77, 85, 89, 97–99, 123, 127, 209 Nadler, Janice: experiments to test expressive effects, 52, 54–55, 62, 63–65, 77, 115, 207, 211; on legal legitimacy, 3, 127 norms, 19, 78, 83, 109, 121, 187; and audience meaning, 182, 191; enforcement of, 117–118, 139, 140, 147, 175; as focal points competing with the law, 9, 27, 62, 97, 100–103, 123, 135;

Pildes, Richard, 15 pluralistic ignorance, 147–150 Posner, Eric, 272, 281, 286, 287, 289, 293, 294, 312 Postema, Gerald, 109, 111 preference change, 165–166, 168 presidential inauguration, 98 Prisoner’s Alibi game, 31–32, 33 Prisoners’ Dilemma (PD), 105, 111, 116; absence of coordination, 42, 62, 64; definition of, 24, 29–31; real–life frequency as compared to other games, 32–34, 35, 38–42, 128–132, 134; experiments involving, 64–65, 65–66; and the State of Nature, 71. See also iterated Prisoners’ Dilemma procedural legitimacy, 3, 27, 120–122, 124–125, 127, 133–134 property disputes: and arbitral focal points, 205–208, 210, 212–217, 224–225; beach access, 86, 98–99; experimental design involving, 64–65; and legal focal points, 86–87, 90–91, 93, 110–111, 235; and Pierson v. Post, 110–111, 114; as within the range of legal expression, 8, 11, 22, 27, 57 public choice theory, 144–145, 159 public goods, 128–132, 133 punishment, 2, 8, 9, 63; definition of, 15, 16; and ex post facto laws, 240, 241–243, 247–248, 261; expressive effects of, 169–170, 173–175, 176–193, 198, 253 racism. See discrimination, based on race; Equal Protection rape, 11, 145, 169, 178–179, 187, 188 rational choice theory, 7, 17, 109, 128, 134, 203

320

index

religion. See endorsement test; Establishment Clause reputation, 83–84, 139–140, 149, 194; and arbitration, 50, 200, 212–213, 217, 221, 223–227, 230 risk signaling, 138, 153–162, 172–173, 180, 186, 187, 238 ritual, 26, 98–99, 209 Robert’s Rules of Order, 113 Robinson, Paul, 3, 127, 133, 134 Rose, Carol, 105, 106 rule of law, 74, 75, 127, 133–135 rules versus standards, 9–10, 235–237 sanctions, 2–9, 112–115, 188, 194, 248, 256; informal, 16, 44, 108–109, 151, 156, 256; as an insufficient explanation for legal or arbitral compliance, 5–7, 26, 42, 62–63, 63–67, 67–76, 76–93, 102–104, 120–122, 167, 199–200, 200– 204, 213–214, 218, 221, 225, 229–232, 248; interaction with legal expression, 7, 27, 61, 117– 119, 124–125, 135; when necessary for law to affect behavior, 22, 197–198, 227; and optimal enforce­­ment, 233–239; puzzle of, 57–61; shaming, 178; and signaling, 138–139, 162–165, 169, 173–174, 188, 261. See also ­deterrence Schelling, Thomas, 45, 50, 53, 60, 95, 122; Bystander Example of, 23, 37, 44, 62, 77, 85, 96, 99; department store example of, 48–49; informal experiments of, 42–44, 67, 88, 90; road lines example of, 49, 107; on tacit bargaining, 67. See also focal points seat belt laws, 11, 145, 160, 164, 166–168; and risk signaling, 138, 153, 156, 161–162 self–fulfilling expectations, 44, 131, 211; as a result of arbitral expression, 48, 218, 229; as a result of legal expression, 22, 56, 61, 62, 69–70, 71, 85–87 sexism. See discrimination, based on gender; Equal Protection Shapiro, Martin, 200, 215, 229 signaling. See attitudinal signaling; enforcement signaling; risk signaling; violations ­signaling Smith, Henry, 20, 110

smoking regulation, 20, 63, 81–82, 235–238, 253; and legal focal points, 85–86, 91, 99–100, 122, 125, 197; as within the range of legal expression, 8, 11, 22, 27, 57, 93; and information revelation, 143, 145, 146, 150, 153–157, 160, 164, 197; and social movements, 100–104, 123 social meaning, 11, 165–168 social movements, 8, 9, 63, 100–106, 135, 148, 260; and public goods, 132; and smoking regulation, 100–104, 123 soft law, 105, 112–115, 222 Souter, David, 182–183, 245 sentence meaning. See conventional meaning speaker’s meaning, 19–21, 179, 182, 240–242, 243–248 special interest groups, 152, 160, 162, 238 spillovers, 150–151, 157, 173, 195 Stag Hunt game. See Assurance game stand your ground laws, 189–192 standard setting, 28, 36, 57, 69–70, 93, 118 Stevens, John Paul, 180–181, 185–186 stochastic shocks, 95–96 substantive legitimacy, 3, 49, 124–125, 134 Sugden, Robert, 23, 53, 96 Suk, Jeannie, 182–183, 245 Sunstein, Cass, 149, 249, 250 Supreme Court of the United States, 12, 58, 98, 151, 201; in Bush v. Gore, 201; in Capital Square v. Pinette, 186; on the endorsement test, 185, 252, 254; in Fleming v. Nestor, 241– 242; in Furman v. Georgia, 171; in Georgia v. Randolph, 182–183; in J.E.B. v. Alabama, 183; in Kelo v. City of New London, 171; in Lawrence v. Texas, 151, 172; in Loving v. Virginia, 171; in Lynch v. Donnelly, 254; in Mississippi University for Women v. Hogan, 185; in Roe v. Wade, 100, 171; in Shelley v. Kraemer, 106; in United States v. Virginia, 185; in Worcester v. Georgia, 58; in Zatco v. California, 180. See also Blackmun, Harry; O’Connor, Sandra Day; Souter, David; Stevens, John Paul symbolic conflict, 14, 248–260 tax law, 4, 133, 144–145, 163, 253 testable implications, 9, 152, 162, 226–227

321

index

theoretical pluralism, 4, 7, 261 traffic regulation, 1–8; center line, 1–2, 4, 5–6, 49, 107, 136; and the clarification of custom, 107; and focal points, 5–8, 26–27, 35, 49, 76–82, 126, 196; and focal points in dynamic perspective, 94, 97, 99; and information revelation, 6–8, 136, 138, 156–157, 196–197; and normative implications, 234–236; and “Shared Spaces,” 78; stop sign, 5, 78–79, 80–81; yield sign, 1–2, 4, 5–6, 77, 78, 80, 85, 136; as within the range of legal expression, 8, 22, 57, 93 Tyler, Tom, 3, 127, 133, 134

Vertova, Pietro, 131–132 violations signaling, 138–139, 162–165, 174–175 voting regulation, 8, 133, 151–152 Weber, Max, 3 Weingast, Barry, 40–41, 69, 73–75, 112, 117, 230–231 wisdom of crowds. See Condorcet Jury Theorem Zimmerman, George, 188–192

322