\documentclass[12pt,reqno,twoside,usenames,dvipsnames]{amsart} \usepackage{amssymb} \usepackage{graphicx} \usepackage{amsmath} \usepackage{hyperref} \usepackage{verbatim} \hypersetup{breaklinks=true, pagecolor=white, colorlinks=true, linkcolor= blue, hyperfootnotes= true, urlcolor=blue } \urlstyle{rm} \newcommand{\margincomment}[1] {\mbox{}\marginpar{\tiny\hspace{0pt}#1}} \newcommand{\comments}[1]{} \renewcommand{\baselinestretch}{1.2} \parindent 24pt \parskip 10pt \begin{document} \titlepage \vspace*{12pt} \begin{center} {\large {\bf Law, Coercion, and Expression: A Review Essay on Frederick Schauer's {\it The Force of Law} and Richard McAdams's {\it The Expressive Powers of Law} }} October 17, 2015 \bigskip Eric Rasmusen {\it Abstract} \end{center} \begin{small} {\it What is law and why do people obey it? This question from jurisprudence has recently been tackled using the tools of economics. The field of law-and-economics has for many years studied how fines and imprisonment affect behavior. Nobody believes, however, that all compliance is motivated by penalties and it is questionable whether that is even the typical motivation. Two books published in 2015, Frederick Schauer's {\rm The Force of Law} and Richard McAdams's {\rm The Expressive Powers of Law}, consider alternative motivations, Schauer skeptically and McAdams more sympathetically. While coercion, either directly or in its support of internalized norms, seems to dominate law {\rm qua} law (and not as a mere expression of morality), a considerable portion of law serves other uses such as coordination, information provision, expression, and reduction of transaction costs. (H11, K00, K40) } \noindent Rasmusen: Dan R. and Catherine M. Dalton Professor, Department of Business Economics and Public Policy, Kelley School of Business, Indiana University. BU 438, 1309 E. 10th Street, Bloomington, Indiana, 47405-1701. (812) 855-9219. Fax: 812-855-3354. \href{mailto:erasmuse@indiana.edu}{ erasmuse@indiana.edu}, \url{http://www.rasmusen.org}. {\small \noindent This paper: \url{http://www.rasmusen.org/papers/expressive-law-rasmusen.pdf}. } I thank Michael Alexeev and participants in the Ostrom Workshop, Indiana University for their comments. \end{small} \newpage \begin{center} {\it 1. Introduction} \end{center} It would not be unfair to say that economics relies on two big ideas, incentives and markets, which Marshall's supply and demand curves beautifully combine. Buyers and sellers respond to the incentive of price, and the price in turn is formed by their responses. Thus it was natural when economics began to be applied to law that the same model was applied to crime. The seminal article of Becker (1968) uses marginal cost and marginal revenue diagrams, and it has become common to speak of the supply and demand for crime, confusing though the metaphor may be when demand is for a ``good'' with negative utility. Once we think of crime this way, we have a useful division between the incentives of those who supply criminal acts and those of the ``demanders'' who provide sellers with opportunities, how incentives differ on average and at the margin, how the price and quantity change, entry and exit, changes in technology, and so forth. Crime is a special product, though, in that a large component of its price is a tax--- the criminal penalty--- and the incidence of this tax falls on the sellers alone because of the good's negative utility to the buyer. This tax is is set by the state through the medium of law. Law, too, can be studied in terms of its supply and demand, but in this essay we will consider its properties rather than its creation. What is law and how does it work? Two recent books that have addressed the subject are Frederick Schauer's {\it The Force of Law} and Richard McAdams's {\it The Expressive Powers of Law}. Schauer and McAdams are senior professors from prominent law schools (Virginia and Chicago), but their approaches are different. Professor's Schauer's perspective is that of traditional jurisprudence and Professor McAdams's is that of law and economics. Interestingly enough, each emphasizes the opposite of what one might expect. Schauer emphasizes coercion in law, while McAdams emphasizes indirect incentives such as coordination and information. Schauer is tackling the old and big question in jurisprudence of ``What is law?'' We in economics are skeptical of spending energy on definitions, but definitions do have their place. Underneath definitions are concepts. We all agree that understanding concepts is important, and if that is true then so are definitions, because it is hard to think straight while mentally readjusting a definition to fit the underlying concept. It is like trying to do arithmetic in a system where the number $x$ denotes $3x$. Thus, even in economics it is worthwhile to spend time pondering what we mean by ``externality", ``transaction cost", and ``the firm". Clarity's importance was recognized long ago by Confucius, who argued for its importance to public policy in one of his major doctrine, ``The Rectification of Names". \begin{small} ``A superior man, in regard to what he does not know, shows a cautious reserve. If names be not correct, language is not in accordance with the truth of things. If language be not in accordance with the truth of things, affairs cannot be carried on to success. When affairs cannot be carried on to success, proprieties and music do not flourish. When proprieties and music do not flourish, punishments will not be properly awarded. When punishments are not properly awarded, the people do not know how to move hand or foot." Confucius, {\it Analects}, Book XIII, Chapter 3, verses 4-7, translated by James Legge. \end{small} This passage, as it happens, alludes to the two aspects of law that separate Schauer and McAdams: ``punishment" and ``proprieties". Is the essence of law coercion, or should we look elsewhere? One place we might start is with a dictionary. The Merriam-Webster online dictionary defines law as ``a binding custom or practice of a community: a rule of conduct or action prescribed or formally recognized as binding or enforced by a controlling authority." (\url{http://www.merriam-webster.com/dictionary/law}) That is a bit of cheat, though, as dictionary definitions so often are. Is the rule really still a law if it is prescribed by a controlling authority but not recognized as binding, and not enforced? Or if it is recognized as binding, but not prescribed and not enforced? Or if it is enforced but not prescribed or formally recognized as binding? And what is a ruling authority? In the end, a thoroughly satisfactory definition will elude us, and perhaps Confucius is right that this is related to why society is disordered. Nonetheless, we can still learn something from the search for the concept's meaning. A good part of that search involves why laws are obeyed, and in particular the question of whether law needs to be defined as a rule enforced by the coercive power of the state. Coercion is central to the ``prediction'' or ``bad man'' theory of law offered by Oliver Wendell Holmes, Jr. (1897). He says that to understand law, one must put aside thoughts of morality. Law is not for the good man, who will do what is right whether there is a law about it or no. Rather, law is set up for the bad man, who wants to misbehave and who worries only about how and whether he will be punished. In this, Holmes follows in the tradition that leads from Machiavelli to Madison, who said, ``If men were angels, no government would be necessary" (as cited in Schauer, p. 97). Thus, \begin{small} ``If you want to know the law and nothing else, you must look at it as a bad man, who cares only for the material consequences which such knowledge enables him to predict, not as a good one, who finds his reasons for conduct, whether inside the law or outside of it, in the vaguer sanctions of conscience.... if we take the view of our friend the bad man we shall find that he does not care two straws for the axioms or deductions, but that he does want to know what the Massachusetts or English courts are likely to do in fact. I am much of this mind. The prophecies of what the courts will do in fact, and nothing more pretentious, are what I mean by the law." \end{small} This concept of law seems clear: law boils down to the penalty the state will inflict on you if you break a rule. Useful as the concept is, though, especially for practising lawyers, it really does not fit with how we think about law. Even the bad man does not think about law that way when he is thinking about public policy rather than his own embezzling, fraud, or tax filing. A court can surely violate the law in its rulings, even if a higher court does not overrule it; indeed, the problem in a corrupt legal system is that it has the rule of men, not rule of law. And there are rules we call laws that cannot be enforced in court. The law says that Eric Rasmusen must pay income tax at a certain rate, but if the Secretary of the Treasury chooses to let Rasmusen get off scot free, nobody can dispute that in court. First, of course, the Secretary and Rasmusen may keep the nonpayment secret. Even if it is boldly proclaimed in public, however, the matter can be brought to court only by someone with ``standing"--- someone whose rights have been violated clearly enough that the law allows him to be the one to bring the case before a judge. In this case, only the Secretary and Rasmusen would have standing and neither would go to court, as we explain in Ramseyer and Rasmusen (2011) in the context of the 2009 TARP bailouts. Courts have repeatedly ruled against ``taxpayer standing'', the idea that some other taxpayer can sue because Rasmusen's nonpayment is bad for every other taxpayer. Thus, as a ``bad man'', the Secretary would feel safe in his favoritism; he will not be punished or reversed. Yet we would not say that the law exempts Rasmusen from taxation. Schauer's effort is to try to sort out what makes laws work. One might base a theory of law on deterrence or on legitimacy. Deterrence is the foundation for Holmes and Becker, and for Jeremy Bentham (1782) and John Austin (1832) before them. Law works because it commands people to obey the rules or else pay a price. Legitimacy might be based on morality, as in natural law theory, where a law that is immoral is no true law. There is also a legal positivist view of legitimacy, however, where it is a recognition by people in general that a rule has been properly formulated, whether it be good or bad. H. L. A. H. L. A. Hart argues for this in his 1961 book, {\it The Concept of Law}, the starting point for modern jurisprudence. He argues that penalties and morality are secondary elements of law. Laws are not like commands, because they apply to the issuer as well as to others, and some laws confer powers rather than imposing duties. Nor do all laws arise as commands of the sovereign, and they persist even when the sovereign changes. On the other hand, the use of coercion is not a sufficient condition for a command to be a law; coercion is what the robber uses to compel obedience, yet we do not regard his demand for your wallet as law. The alternative is to ask whether a law is legitimate, whether it has been enacted according to generally accepted principles and thus satisfies what Hart calls ``the rule of recognition''. One may take the idea a step further to argue that at a deeper level people obey the law not so much because of penalties as because they believe it has been promulgated by a rightful authority. Max Weber said, ``The most common form of legitimacy is the belief in legality, the compliance with enactments which are formally correct and which have been made in the accustomed manner" (as quoted by McAdams, p. 3). Here we have the difference between Holmes's bad man--- who cares only about deterrence--- and Holmes's good man--- who cares about legitimacy and morality. In economic language, we have the difference between the incentive of an external price and the motivation of an internal taste. Schauer sides with deterrence. ``Law makes us do things we do not want to do," is the first sentence of his book, and the title, {\it The Force of Law}, is no accident. But he recognizes that deterrence theory has problems. There are constitutive rules of law, for example, as well are regulative (to use the terminology of philosopher John Searle (1969)). Regulative rules are what we think of first. They restrain and regulate behavior that would happen even under anarchy. A regulative rule makes killing someone into murder--- unless you do it in self-defense, and then it is allowed. It makes dumping toxic waste into a violation of the Environmental Protection Act--- unless you do it according the rules. Constitutive rules, on the other hand, create new opportunities that would not be possible under anarchy. The law of wills and testaments, for example, allows you to leave your house to your nephew with the assurance that if your son objects, the coercive power of the state will block his objections. Contract law is the example par excellence. You can make promises without law, and regulative law can constrain them--- for example, it is unlawful for you to promise to deliver marijuana--- but contract law allows us to make promises into binding agreements. Coercion is an element of constitutive law, to be sure, but it is coercion voluntarily accepted. The buyer accepts his obligation under the contract because that is how he can impose an obligation on the seller. Contract law also shows the power of law without coercion, however. A business's biggest loss if it violates the law of contracts in dealing with its supplier is that the supplier will stop dealing with it and other suppliers will be reluctant to fill the gap without a price premium. Much of Schauer's position can be explained by his willingness to broaden the definition of coercion to include such things as shaming, reputation loss, and expulsion from cooperative relations (pp. 133-135), indirect penalties that others of us would contrast with fines, imprisonment, and corporal punishment. Even this broadening, though, would not include obedience to the law independent of consequences, obedience of whose practical importance Schauer is skeptical. Thus, Schauer dismisses the person whom Hart calls the ``the puzzled man", the man who wants to do what is right, but who wants the law to tell him right from wrong. The puzzled man does not need to be coerced, only informed. The idea brings to mind ``the three uses of the law" in John Calvin's 1536 {\it Institutes of the Christian Religion}. Calvin's first use of the law is to maintain order, to control the bad man. The second is to convict men of sin--- that is, to challenge them by revealing their inability to deal with the evil within them--- and the third is to educate, to provide a guide to how someone desiring to do good ought to behave. While admitting that the puzzled man may exist, Schauer is skeptical of his practical importance. Many of us are puzzled about what is moral, but how many of us look to the law for what is moral, instead of deciding what is moral first and then seeing if the law is close enough that if we do the right thing the law won't punish us? This, indeed, is a problem for Holmes's idea of the ``good man''; the good man, like the bad man, might look at the law with only an eye to what it will punish, having already decided what he wants to do based on what is moral. Little scope is then left for the puzzled man, for whom law per se determines what is moral. Schauer admits that coercion may not be absolutely essential to law, but he argues that it is more useful to look for what is typical, and the puzzled man is the exception that proves the rule. In general, laws are imposed to make people change their behavior. And most people will not change their behavior without the threat of coercion. \begin{center} {\it 2. Jurisprudence Games} \end{center} McAdams, on the other hand, does recognize that coercion is an important feature of law, and even the dominant feature if norms are independent of law, but he thinks special cases are important if we are to understand how laws affect behavior and why are they are enacted. He looks at two categories: coordination and information. Simple games give insight into why laws can be useful, especially in the case of ``expressive law'', which does not rely on coercion. The first game relevant to law is the prisoner's dilemma, by now well recognized even in legal academia. In an earlier article, McAdams (2009) notes that the prisoner's dilemma has been mentioned in over 3,000 law review articles compared to 246 mentions of the three coordination games we will look at below. The story is a familiar one, but I will teach it yet again. Two prisoners are being held on suspicion of having committed a felony and a misdemeanor together. If they both deny having committed the felony, they will both convicted of the lesser misdemeanor and sentenced to two years in prison. If both confess, they will each be sentenced to six years. If one confesses and the other denies, the prisoner who confesses will serve one year and the one who denies will serve twenty years. Table 1 shows the payoffs. \begin{center} \begin{tabular}{cccc} & & \multicolumn{2}{c} { Prisoner 2}\\ & & Deny & Confess\\ \cline{3-4} % & &\multicolumn{1}{|c}{}&\\ & Deny &\multicolumn{1}{|c}{ -2,-2} & -20, -1\\ Prisoner 1 & & \multicolumn{1}{|c}{}& \\ & Confess &\multicolumn{1}{|c}{ -1, -20} & -6, -6\\ \end{tabular} {\sc Figure 1: The Prisoners' Dilemma} \end{center} Making their decisions independently, both prisoner choose to confess because that is not just the Nash equilibrium--- the best response to the other player's equilibrium action--- but dominant: whether the other player chooses to confess or to deny, to confess is the best response. The puzzle is that both players' payoffs would rise if they could bind themselves both to deny. The prisoner's dilemma is not a model of expressive law, but it is helpful in understanding the need for both regulative and constitutive law. For regulative law, it represents the social contract. We all would like to pay rather than steal goods from each other, because stealing leads to inefficient allocation and rent-seeking. Pay/steal parallels the payoff matrix's deny/confess. The solution of law is to introduce a third party, the state, which punishes deviation from the jointly optimal action. If you and I can vote for what the state does, we vote for coercive laws. If the state is a dictatorship, it chooses the same coercive laws so as to maximize social wealth. We can take the idea further and think about laws either in a dictatorship or a democracy that are themselves for rent-seeking, more akin to steal than to pay (a law granting one firm a monopoly, for example), and then about binding ourselves not to enact rent-seeking laws. The prisoner's dilemma also applies to constitutive law, the law that enlarges the sphere of opportunity rather than diminishing it. Two parties to a deal can make a promise even in the absence of law and each has the choice to perform their obligation or to breach. The dominant strategy will be to breach, regardless of what the other party does, in the absence of morality, reputation, and other private incentives. What law creates is the choice to make a contract instead of just a promise, the payoffs changing because the state adds a penalty, enforced by coercion, that makes breach more costly. Like the law against theft, this imposes a penalty, but it is one of Holmes's main examples for his bad man theory because retribution is a poor explanation for the particular penalty. The state's remedy for breach is not to require performance, or that the breacher disgorge the profits he makes by breaching, but rather that he pay damages sufficient to make the injured party whole. The law says that it is a breach of a duty, but what it means is not that breach is immoral, or even that the state forbids it, but that the breacher must pay compensation. Breach damages are a price, not a penalty, for breaching. How a price differs from a penalty and how a penalty differs from a tax are interesting questions, of course, and important to the law. {\it National Federation of Independent Business v. Sebelius} 56 U.S. \_\_ (2012), the most central of the Supreme Court's health care mandate cases, is a prime example. The court ruled that without going beyond its constitutional powers the federal government could not use fines as penalties to force someone to buy health insurance, but that it could impose a tax on people who failed to do so. The distinction sounds humorous, and one might argue (and the lawyers did!) over whether the monetary payment really was a tax , but we all know there is a difference between taxes and penalties. I recall George Stigler suggesting to me back in 1990 that a negative incentive is the price for an action if one is allowed to repeat it as many times as one wishes so long as payment is made, but a penalty if recidivism is punished more and the state tries to restrict you from repeating the offence. Thus, parking tickets and penalties for not buying health insurance would be prices, but speeding tickets and felony sentences would be penalties. Perhaps this is what the Supreme Court was getting at. Schauer's focus is on regulative law as a payoff-changing solution to the prisoner's dilemma. He notes that constitutive law is also coercive, in a certain sense. Once we have contract law, it becomes more difficult to use promises. In the absence of courts, the parties might be better able to trust each other's promises, because social norms for promises are stronger. Once contract law is introduced, though, the parties are more likely to excuse their behavior by saying that they complied with the words of the contract, and in contract law words trump spirit. Thus, they are coerced by the state into abandoning their social norm for the new legal rule. On the other hand, legal rules may bring liberation from undesired social norms. In a world without courts, the bad man cannot find partners for his deals, but though the good man can find partners, the details of his agreement are constrained by whatever the social norm may be. Or, in the case of either a bad or a good man who can find partners because he has a good reputation and wishes to keep it, he is constrained by whatever behavior is required to maintain his reputation. It may be, for example, that the norm is that you keep your promises, however costly they may have become to keep. In that case, the party who wants to breach might get far more benefit from breaching than the injured party gets cost, but he fears for his soul or his reputation if he breaks his promise. Or, there may be a mutually beneficial modification to the contract, but the party desiring to breach can be ``held up" by the other party and made to pay exorbitantly. In this case, the good man might breathe a sigh of relief in having his norms overridden by law, because the law would permit breach if the breacher pays reasonable damages, freeing him from the higher claims of the promisor's duty. Indeed, law permits parties to a contract to implicitly include in their agreement a whole set of default rules established by others' wisdom and experience with the contingencies that arise. Of course, it can also happen that the legal rule is less efficient than the norm; Bernstein (1992) tells us that in the New York diamond industry the norm is never to breach and that such a rule is efficient in the particular context because of the importance of prompt payment for cash flow. \bigskip \noindent {\it 2.1 Pure Coordination} The prisoner's dilemma is central to why law is desirable, but McAdams brings it up to distinguish it from coordination games. Using other stories of prisoners he illustrates the paradigmatic three types of coordination games: pure coordination, the assurance game, and the battle of the sexes/hawk-dove. Pure coordination is the simplest of the three. Imagine now that the police have no evidence of the misdemeanor, but they will have sufficient proof of a felony with a five-year sentence if the prisoners have inconsistent alibis. That story gives the game of Table 2, in which the payoffs from both choosing alibi A or both choosing B are (0,0), but the payoffs from one choosing A and the other B are (-5,-5). There are two Nash equilibria in pure strategies, one for each alibi. There is also a mixed-strategy equilibrium in which each player chooses alibi A and alibi B with equal probability. \begin{center} \begin{tabular}{cccc} & & \multicolumn{2}{c} { Prisoner 2}\\ & & Alibi A & Alibi B\\ \cline{3-4} % & &\multicolumn{1}{|c}{}&\\ &Alibi A &\multicolumn{1}{|c}{0,0} & -5,-5\\ Prisoner 1 & & \multicolumn{1}{|c}{}& \\ &Alibi B &\multicolumn{1}{|c}{-5,-5} & 0,0\\ \end{tabular} {\sc Figure 2: The Alibi Game of Pure Coordination} \end{center} I would supplement McAdams's pure coordination game with ``ranked coordination''. In a ranked coordination game, discoordination yields the worst payoff, but different coordinated actions have different payoffs. Suppose alibi A is that the prisoners were playing air hockey together at the other end of town, for a payoff of (0,0) if they both choose it, but Alibi B is that they were committing shoplifting together, for a payoff of (-1,-1). There are still two Nash equilibria. If prisoner 1 expects prisoner 2 to choose alibi A, he will choose it too. But if prisoner 1 expects prisoner 2 to choose alibi B, his payoff from also choosing it is -1, compared to -5 from choosing alibi A. Thus, (B, B) is also an equilibrium. Pure and ranked coordination are the games for standard setting and conventions. Not all conventions are rankable; some fit the category of pure coordination. The rule of driving on the right side of the road is an example, or was before we in America became accustomed to it. There are, of course, coercive penalties for driving on the left, but everyone would choose to drive on the right anyway to avoid an accident. Law works by suggesting an equibrium. Law provides one of the Schelling (1961)'s ``focal points'', a reason for players to think other players will choose to play out a particular equilibrium. The coercive penalty has some influence, both directly and because it may be focal to choose the equilibrium which is unpunished, but the dominant force is expectations; if we knew we might receive a ticket for driving on the left but we thought other drivers would kepp left, we would prefer the ticket to driving differently from everybody else. Some standards, however, need both focal points and careful consideration of which standard is best, and so are better considered ranked coordination games. The U.S. federal government defines what it means for lettuce to be ``organic'' so different kinds of sellers can communicate to organic-loving and organic-indifferent consumers what good is being sold. ``Organic'' has a variety of possible meanings, which would yield different amounts of social surplus depending on their cost and the value placed on them by consumers. The government picks one from among those meanings. This illustrates a second feature of ranked coordination: the possibility of an information function for law. In the alibi game it is obvious which alibi is best. In lettuce labelling, it is not. The citizens know coordination is desirable to make labels meaningful, but they do not know which definition is best. The law thus serves a purpose beyond coordination: to identify the best definition. There is reason, however, that this might be classified under McAdams's coordination theory rather than his information theory. The new information about the payoff of the action is helpful, but it is not why citizens use the government’s definition. Rather, they use it firstbecause of the underlying law against fraud (the solution to a prisoner's dilemma), and second because they know other citizens will use it and would do so even if another definition were better. Merrill and Smith (2000) apply the idea of coordination on definitions in a sophisticated way to property law's {\it numerus clausus} (closed number), the principle that land property can only be enforced as falling into one of a small number of legal forms. A rental agreement that says a tenancy will last ``for the duration of the war'' will not be enforced. Instead, the courts will try to fit it into one of the four recognized forms of tenancy. Most courts have treated it as ``tenancy at will'', lasting only so long as both parties agree, or as ``periodic tenancy'', if the agreement provides for payments at, say, monthly intervals. This contrasts with contract law, which is extremely flexible as to which terms can be inserted into contracts. It is possible, for example, to write a contract in which I sell you my house with the provision that you never bring oranges into it on pain of \$5,000 damages. That will be enforced as a contract between the two of us, but not as part of the house as property; if you resell the house the new owner is not bound, and would not be bound even if the contract contained a clause specifically saying that the orange clause applies even after resale. Why the limitation for property law? One possibility is the worry that property could result in fragmentation of interests that would create waste after the purpose of the novel form had disappeared. If such waste were likely, though, the original seller would put a term of years on the novelty to raise the selling price. Merrill and Smith dismiss fragmentation in favor of a different explanation: that land interests frequently involve third parties who would have to incur transaction costs in determining who exactly had which interests in the property. Even if almost all most property were held in one of the conventional forms, a third party wishing to buy the property, base a mortgage upon it, or rent it would have to check for oddities in its ownership. Thus, law can help with coordination. Schauer, however, would object at this point that the coordination function of law is easy to overstate. What if the government did not tell us which side of the road to drive on, or what lettuce to label organic? Customs, would develop for driving, social norms for behavior. Industry associations would decide on standards and issue certificates of organic compliance. When force is not necessary to implement a rule, private actors can do it. The government has an advantage only because it is powerful and focal, because we expect other people to follow the standard the government suggests. This objection ccould be answered by saying that power and expectations are essential features of government. Thus, the same objection could be made to the idea of the central bank as lender of last resort, since a large enough private bank could do the same thing and resolving a crisis by lending to sound institutions is profitable. One of the most useful features of a government is that it is big, loud, and respected enough that people listen to it and expect other people to listen too. That's part of legitimacy: it's not just that people feel they ought to obey the government if it is against their interest, but that they think it in their self-interest to obey it because other people will too. In addition, a government can begin with coercion and then continue without it. First, the government imposes punishments that make the desired behavior an equilibrium. Later it can eliminate the penalties without disturbing the behavior, if continuation of past behavior is focal in a coordination game. What starts by command becomes coordination. \noindent {\it 2.2 The Assurance Game} Legitimacy itself is modelled using our next game, the assurance game. The assurance game in Figure 3 changes just one of the payoff combinations of the prisoner's dilemma. Now, instead of (deny, deny) resulting in two years of prison each, the prosecutor has very little evidence and would have to release both prisoners. As a result, there are two Nash equilibria. The prisoner's dilemma outcome of (confess, confess) is still a Nash equilibrium, but confessing is no longer a dominant strategy. Instead, if prisoner 1 expects prisoner 2 to deny, he should pick deny too, since (deny, deny) has a payoff of 0 compared to -1 from (confess, deny). \begin{center} \begin{tabular}{cccc} & & \multicolumn{2}{c} { Prisoner 2}\\ & & Deny & Confess\\ \cline{3-4} % & &\multicolumn{1}{|c}{}&\\ & Deny &\multicolumn{1}{|c}{0,0} & -20, -1\\ Prisoner 1 & & \multicolumn{1}{|c}{}& \\ & Confess &\multicolumn{1}{|c}{ -1, -20} & -6, -6\\ \end{tabular} {\sc Figure 3: The Prisoner's Assurance Game} \end{center} The assurance game is like ranked coordination in having two Nash equilibria ranked equally by both players. The difference is that the assurance game retains the prisoner's dilemma feature that if one player deviates from equilibrium in the pareto-superior action combination, the other player's payoff is particularly low-- twenty years in prison. If we add a probability $\alpha$ that a player chooses the out-of-equilibrium strategy by mistake that would not affect the player's choices in the prisoners' dilemma, pure coordination, or ranked coordination. Here, though, if that probability were greater than $\alpha = 2/5$ the (deny, deny) equilibrium would disappear. A player needs assurance that the other player will play to their mutual benefit. McAdams uses the assurance game to think about a constitution. In the conventional view, a constitution is used to solve a prisoner's dilemma. In the state of nature, everyone chooses to plunder rather than create because plunder is individually the dominant strategy. To escape, they agree to form a government that punishes plunder. McAdams suggests this is more like an assurance game. Each player has the choice to refrain from plunder and support the rule of law, on the one hand, or to launch a pre-emptive attack on the other. It is not a prisoner's dilemma, because if player 1 expects other players to support the rule of law, doing so is also to his private advantage. If, however, expectations become pessimistic, each player seeks to protect himself as best he can, and the rule of law dies. As McAdams notes, this is the same idea as a repeated contribution game, a repeated prisoner's dilemma in which each player's choice is to contribute to a public good or to be selfish. Infinitely repeated games have multiple equilibria, but in the two extreme equilibria the actions are the same in each repetition: nobody contributes, or everyone contributes. Everyone contributing is is an equlibrium because if one player deviates he will receive a higher current payoff than the other players, but by causing them to stop contributing in the future he reduces his own overall payoff. Condensing this to a one-shot game gives the assurance game, a high payoff for a player if he and everyone else contributes, and a low payoff if he contributes in isolation. This idea of equilibrium as legitimacy can be found also in Weingast (1992), in verbal form. Hadfield and Weingast (2012) formally model it using repeated games in ``What is Law: A Coordination Model of Legal Order.'' The idea builds on the well-known use of repeated games to explain cooperation in general (see Fudenberg and Maskin (1986)). I have used it myself in the context of modelling judicial legitimacy and adherence to precedent (Rasmusen [1994]). What government adds is a player who declares what is to be equilibrium behavior and what is to be deviation. This is a form of ``cheap talk''--- a move in a game that has no direct impact on payoffs but on which strategies can be based. In some games, cheap talk expands the scope for desirable outcomes (Farrell and Rabin [1996] is a survey). Cheap talk does not solve the problem of multiple equilibria, since in equilibria players can always use strategies that ignore it--- but it makes the plausibility of desired outcomes as focal points even more compelling. Coordination in repeated games gives us a theory of legitimacy in general contexts. Legitimate behavior is equilibrium behavior enforced by the threat of low payoffs that would result from deviation that sends the players into a subgame in which legitimacy is lost. The idea is very much like the Baron and Bowen (2015) explanation for the stability of coalitions against deviation by subcoalitions that might prefer a different allocation of power, but are deterred from trying to form a new coalition by the possibility that failure might reshuffle things so much they would be worse off than ever. The threat of general breakdown is different from another, equally interesting, way in which deviation can be punished: by an equilibrium specifying that other players must punish a deviator or be punished themselves. This, too can support a variety of equilibria in repeated games, and it is the standard way to construct them (Maskin and Fudenberg (1986). Perhaps such punishments can be described as the upholding of legitimate rules too, but it is based on the existence of a second level of punishment, and so is closer to the idea of coercion. Schauer would classify both threats as forms of government coercion (p. 135). The rival of legitimacy as repeated games is legitimacy as internalized principle. If legitimacy is a principle, people are educated to believe that the government is legitimate and that disobedience is wrong, generating the disutility of guilt (mental pain felt even aside from the existence of other people) or shame (mental pain from other people seeing what you have done, or you imagining them see it). No doubt both legitimacy as coordination and legitimacy as principle contain truth, but the mechanics and implications are different. Legitimacy via principle is a solid legitimacy that outlasts the sovereign's public sway. An example is the legitimacy French royalists and foreign governments granted Louis XVIII in his twenty years of exile while France was ruled by republic, directory, and Bonaparte. Legitimacy via coordination is a brittle legitimacy, which lasts in citizen 1 only while citizen 2 acknowledges it. This legitimacy is the gameskeepers' loyalty to the crown in Shakespeare's {\it King Henry VI-Part 3} (III-1). Poetry sometimes conveys ideas better than prose. The gameskeepers are chided by King Henry when they capture him fleeing from his rival, Edward IV, and propose to turn him in for the reward: \begin{small} \noindent {\it King Henry VI}: But did you never swear, and break an oath?\\ \noindent {\it Second Keeper} : No, never such an oath; nor will not now.\\ \noindent {\it King Henry VI}: Where did you dwell when I was King of England?\\ \noindent {\it Second Keeper} : Here in this country, where we now remain.\\ \noindent {\it King Henry VI}:I was anointed king at nine months old;\\ My father and my grandfather were kings,\\ And you were sworn true subjects unto me:\\ And tell me, then, have you not broke your oaths?\\ \noindent {\it First Keeper:} No; For we were subjects but while you were king. \end{small} The gameskeepers obey the sovereign, but only the sovereign that everyone else obeys (which, indeed, is a theme throughout Shakespeare's history plays). Sovereignty is partly history--- Henry's grandfather had captured the throne from a weak rival--- and partly self-interest--- Henry himself was a weak king and his nobles saw a chance for self-aggrandizement. Thus, rival kings jockeyed to capture the expectations of the nobility and the people. Expectations like that are not without their own internal morality of taste. Taste and coordination are hard to distinguish. Consider the internalized belief that one should obey the sovereign everyone else is obeying. That is the oath the gameskeepers thought they were swearing. And in practice a common citizen in a large community does not obey the law because he fears his disobedience will society will crumble. That may well be the reason Republican senators respect the minority's right to filibuster when the minority are Democrats, why members of a family behave respectably for each others' sake, or why factions long respected the peace in Northern Ireland and Yugoslavia, but for the individual in a large group the free rider problem is too great. Rather, obedience to the law, written or unwritten, is supported by guilt, shame, or disapproval --- feelings that may depend on what other people do, but create a different coordination problem. If we again return to Schauer's cautions about the coercion behind expressive law, they will apply here too. Force creates a coordination game of its own. If everyone obeys the law, a single deviator is easy to force into compliance. If everyone disobeys, the state is helpless. Much depends on the expectations with which we start. I think I wouldn't steal even if the police did not exist to stop me, but that thought is easy to maintain in the absence of true temptation. Maybe my internally motivated respect of property is fragile, controlling my behavior when crime doesn't pay, but eroding once it does. So it is with legitimacy generally; its support from internal tastes in ordinary times may vanish once tastes are not constrained by force. \noindent {\it 2.3 The Battle of the Sexes} McAdams's third coordination game is the battle of the sexes, with the payoffs in Figure 4. Unlike the games described so far, the battle of the sexes has asymmetric payoffs. Here, instead of McAdams's story, I will use a modification of the alibi pure coordination game that I will call the battle of the alibis. Suppose that instead of an alibi for innocence, the only alibi available is a story that both prisoners were committing a lesser crime, but that one of the prisoners was the culprit and the other merely a witness. Alibi A is the story with prisoner 1 as the witness and payoffs of (0,-3); alibi B has prisoner 2 as the witness and payoffs of (-3,0). Again, both prisoners choosing alibi A and both prisoners choosing alibi B are Nash equilibria, but now the two prisoners have opposite preferences between them. \begin{center} \begin{tabular}{cccc} & & \multicolumn{2}{c} { Prisoner 2}\\ & & Alibi A & Alibi B\\ \cline{3-4} % & &\multicolumn{1}{|c}{}&\\ &Alibi A &\multicolumn{1}{|c}{0,-3} & -5,-5\\ Prisoner 1 & & \multicolumn{1}{|c}{}& \\ &Alibi B &\multicolumn{1}{|c}{-5,-5} & -3,0\\ \end{tabular} {\sc Figure 4: The Battle of the Alibis} \end{center} This game is known as the battle of the sexes when the coordinating actions are the same for each player, as in the battle of the alibis, and as hawk-dove when the players wish to coordinate on a different, complementary, action for each. The difference from the other coordination games is that the players have different preferences over equilibria. Prisoner 1 wants alibi A, but prisoner 2 wants alibi B. The various tricks of Schelling's {\it The Strategy of Conflict} can come into play as players strive to control expectations. Prisoner 1 would want to move first, blabbing out alibi A as soon as the two prisoners are caught, and in the presence of prisoner 2. Or, he might shout out "Alibi A!" to prisoner 2 as he is dragged away. Or, he might lie when the prisoners are put in the same cell later and tell prisoner 2 he had already told alibi A to the police. The principle that different players have different preferences certainly holds true for laws, and explains the maneuvering over their formation. When a desirable law is to be passed, each player tries to make his version of the law the focal point, using announcements to the media, control of the first draft, or confident announcements of victory. Our earlier example of labelling organic lettuce could be seen as a battle of the sexes, since all producers would benefit from standards, but each producer would want to make his personal standard the standard for everyone rather than have to adapt to something new. \begin{center} {\it 3. Law as Information Provision} \end{center} Let us now turn to McAdams's second theory of law: the information theory. McAdams says that law conveys three kinds of information to the public. The first kind is facts about the physical world, such as whether anti-lock brakes are really worth having on a car (``risk signalling''). The second is information about other people's opinions (``attitudinal signalling''). The third is information about the level of enforcement (``violations signalling"). An example of risk signalling is the Food and Drug Administration's requirement that pharmaceutical products be proven safe and effective. The main motivation for the seller to obey the law is coercion. For the buyer, however, the main effect of the law is informational. He knows that any product the FDA allows to be sold has passed a high threshold of safety. In the case of prescription drugs, the informational effect extends not only to the patient but to the doctor, who has the FDA's approval as a first cut for which drugs to consider for a particular malady. If the FDA allowed all drugs to be sold, but put a seal of approval on drugs proved safe and effective, the law would still be largely effective. Attitudinal signalling conveys different information. The successful passage of a law is an indication that somebody wanted it passed. Thus, from the law the citizen can deduce something about what other citizens think. Moreover, he learns not about a random sample of other citizens, but about a group powerful enough to enact a law. Depending on the context, this group may be a majority, or it may be a minority with strong enough beliefs that it can use log-rolling or lobbying to obtain its desired law. Social legislation concerning such things as sodomy, marijuana, flag burning, same-sex marriage, abortion, and gun control all are attititudinal signalling. While bills on these matters usually have coercive effects, the bitterness of the fights over them point to something else at stake. Victory communicates political power. Being able to convince a legislator that voting for your side will be best for his career, or begin able to unseat legislators who will not cooperate, is a sign that your side has more people, resources or talent. The less the vote corresponds with the legislator's personal beliefs or past allegiances, the better it is for showing your group's power. The extreme comes in totalitarian societies. Theodore Dalrymple says (in Glazov [2005]), \begin{small} In my study of communist societies, I came to the conclusion that the purpose of communist propaganda was not to persuade or convince, nor to inform, but to humiliate; and therefore, the less it corresponded to reality the better. When people are forced to remain silent when they are being told the most obvious lies, or even worse when they are forced to repeat the lies themselves, they lose once and for all their sense of probity. To assent to obvious lies is to co-operate with evil, and in some small way to become evil oneself. One's standing to resist anything is thus eroded, and even destroyed. A society of emasculated liars is easy to control. \end{small} The idea generalizes to less malevolent forms of political expression. A law is a public expression of the society's will. When one state passes a bill in favor of same-sex marriage, that says something about the beliefs of the majority in that state, or of those who care the most, are most able, or have the most money. Whatever the currency of power, the law represents every citizen of the state, whether in favor or opposed. A state passing a bill against same-sex marriage is an even clearer example. Such a law has no coercive effect at all if it merely restates the status quo. It is, rather, a declaration of where political power is strongest. Why would such a declaration be useful? A law against same-sex marriage is a warning shot fired at the judiciary. A judge might create a new law himself if he thinks the citizens don't care or would be on his side, but his creativity might flag if he fears public disapproval. Of course, the judge may himself engage in attitudinal signalling when he creates new law. In that case it is the political balance within the judiciary that is conveyed as information. In the 2015 same-sex marriage case, {\it Obergefell v. Hodges}, Justice Scalia criticized his fellow-judges for ruling based on personal beliefs, given that the Court was composed entirely of graduates of Harvard and Yale Law School. Just because a political body is unrepresentative, however, doesn't mean its actions don't convey information. In this case, it conveyed the information that the legal elite's view of same-sex marriage had changed, which in turn implied that the elite view generally had changed. That information might have even more impact than the information that the median American voter had changed his view. In other cases the legislature may be signalling the executive branch. In February 2014, a law was passed requiring the President to notify Congress of prisoner trades thirty days prior to the actual exchange of bodies. In May 2014 the President traded five Taliban prisoners for an American prisoner, Bowe Bergdahl, but he gave Congress less than a day's notice rather than thirty days. The law had no enforcement provisions, however, so he suffered no punishment. Why, then, enact the law? What the law did was to communicate to the President that a majority in Congress wanted to be notified of prisoner exchanges in advance, so Congressmen could raise possible objections. The President was willing to sign the bill, which would seem to indicate his agreement to obey it, but he apparently decided that notifying Congress 30 days before this particular trade would create more bad publicity than ignoring the law and revealing Congress's impotence. Nor is the prisoner-exchange law anomalous; there are many laws that lack enforcement provisions. Let's return to the effect of attitudinal signalling on the ordinary citizen. Citizens would find signalling useful because information on what other people believe should and does affect one's own beliefs. This can be either rational or taste-based. Blind conformity is irrational, of course, but Bayesian updating requires that one update towards the beliefs of other people unless one understands why they went wrong. Most people are politically uninvolved, and hence have weak priors, so going with the majority is rational. For them, even if the judiciary is unrepresentative, its declaration is evidence that intelligent people hold a particular belief. Or, it might just be that people have a taste for agreeing with the majority, or like to back a winner. This returns us to coordination, but it is a two-step coordination game. In the ordinary coordination games, the law creates a focal point and shifts everyone's behavior at once. Attitudinal signalling is better modelled as a sequential game. First, some of the players engage in a public battle--- political, judicial, or in the media--- and which side wins becomes public information. Second, the other players decide which side to join. The simple fact that position X has won the first round can make X focal, or, if the winner is determined by which side has the majority, the fact that it won a majority of first-round players can make it focal. Or, it could be that the second-round players actually have a taste for adopting the position that won the first round, regardless of whether it is adopted by a majority of players once the second round is over. This is one explanation for the phenomenon, well-known in political science, that the winner of an election gets a bigger majority from respondents in surveys after the election than he did in the actual vote count (another is that people forget who they voted for and give the only name they can think of). For example, 59\% of respondents reported having voted for the party nominee in Democratic primaries from 1972 to 1992, but the actual vote percentage was 44\% (Atkeson [1998, p. 199]; see also Wright [1993]). Not all players have to have the taste for conformity for attitudinal signalling to be effective. To use the terminology of Haltiwanger and Waldman (1991), some can be ``responders'' who change their move to back the winner, while others are ``nonresponders'' whose attitudes are independent of popularity or power. How attitudinal signalling might work is hard to pin down. That Congress passes a bill does not tell us that a majority of citizens support the bill, but that a majority of political heft is behind it. It is easier to imagine why citizens would decide to conform their own opinions to those of the majority than to those of the powerful, at least in a country where the powerful cannot punish dissidence. This is particularly true in the context of one law, where ``power'' may mean simply the predominant influence of a special interest that stakes its entire political capital on one issue. Also, there is ample scope for ulterior motives when a law is not majority-based. Citizens know that the stated reason for a congressional bill is often not the real reason. This is a problem even with risk signalling. The chicken industry wants the government to tell people that cholesterol is unhealthy and the beef industry wants the opposite. If people think that what the government announces is based purely on interest-group politics, laws will have no information content. The government has a limited amount of credibility, which is used up whenever one of its assertions is discovered to be false. If they believe the government is lying 30\% of the time but don't know which 30\%, their willingness to obey laws will be correspondingly reduced. This applies even more to attitudinal signalling. Those who pass a law with the motive of attitudinal signalling are intending to convey information about other people's opinions. Their own bias will lead them to overestimate how popular their opinions are, and even aside from that, they may think it ethical to exaggerate how many people share their opinion, especially if they think their exaggeration will be self-confirming by shifting public opinion. Other citizens know this, of course, so just as they take the claims of commercial advertisers with a grain of salt, so too will the effect of attitudinal signalling be weakened by skepticism. Recall Hart's puzzled man, who wonders how to behave. It is perhaps under the category of attitudinal signalling that we would place law's educative purpose. Whether the puzzled man is really puzzled or is just ignorant, if he violates a criminal law he will discover how society wants him to behave. Even in the absence of penalties this would happen--- think of the traffic cop who lets you off with a caution--- but penalties drive home the message. This is the theme of Dau-Schmidt (1990), whose message of which is clear from the title: ``An Economic Analysis of the Criminal Law as Preference-Shaping Policy.'' Indeed, rehabilitation is one of the functions of criminal punishment that is routinely cited, whatever difficulties it has in practice. We make use of the legal status of actions in private life too. We tell children ``Stealing's against the law'', even if what we really mean is ``Stealing is sinful'' or ``Stealing is against our people's moral code'' or ``If you steal, you'll get a bad reputation." The illegality of the act is shorthand, an example of how law reduces transaction costs. In this way, law works by making it easier to teach norms. Plato's {\it Laws} includes 214 mentions of the word ``education'': for him, education and law were subjects that had to be considered jointly. Although it may be true that people are really doing what they think is right, an internalized norm, when they obey a law, and they would not change their behavior even if the law disappeared, that does not imply that the law has no effect on them. The existence of the law made it easier for parents and others to instill them with the norm. Of course, under this story the educative power of law does depend on how closely it tracks social norms. The more that law is morally arbitrary or offensive, the less it is respected and useful in teaching children. McAdams's third category of signalling is violation signalling. This is different from the first two categories because it consists of involuntary transmission of information. Sometimes the passage of a law signals the existence of a problem. A law increasing the penalty for a crime is passed because the current penalty is insufficient. Discovering this, citizens may learn that criminal acts are more profitable than they thought. Violation signalling is an attractive idea with its paradoxical conclusion that increased penalties lead to more crime, but examples are hard to come by. McAdams suggests the study of Israeli daycare centers by Uri Gneezy and Aldo Rustichini (2000). Initially, the day care centers imposed no fine on parents who picked up their children late. After a fine was imposed in six of the daycare centers, late pick-ups rose rather than fell. One explanation is that this changed the sanction from a shame penalty, which would increase for repeat offenders, to a price, which would not. Another explanation is violation signalling. Imposing the fine conveyed to parents that other parents were picking up their children late, and they responded to that information by being late themselves. McAdams discusses risk, attitudinal, and violation signalling, but he does not discuss a fourth kind of signalling: people who use their obedience to the law to signal information about themselves to other people. The first three categories of signalling are not ``signalling'' in the technical sense in economics: a player with desirable but hidden characteristics intentionally engaging in behavior that communicates his type because the behavior is more costly for an undesirable player. Making statements, communicating power by winning political battles, and involuntarily conveying information are not that kind of signalling, though they are ``signals'' in the everday sense of the word. To refrain from crime, however, is easiest for the person with greater self control, a characteristic the person may wish to communicate to others. Thus, there could be an equilibrium in which everyone except those with very little self control obey the law for fear of being thought reckless. Such an equilibrium would require no penalty except the state's public declaration that the offender was guilty. I model this ``stigmatization'' in Rasmusen (1996) as an adverse selection model; for a search model version, see Harel and Clement (2007). Stigmatization is a cheap form of deterrent, and can even have negative cost because it allows those who obey the law to credibly communicate their value (though some, e.g. Funk [2004], argue that poorer information is better since stigmatization reduces the non-criminal opportunities of those convicted). Because of stigmatization--- or, the other side of the coin, because of validation of one's desirable type, someone looking for an employee or a spouse is able to separate types with low and high self-control and make a more appropriate match, shooting neither too low nor too high. In such situations, however, there is a second equilibrium, in which stigmatization fails and everyone disobeys the law. If everybody commits crime, including people with high self-control, the stigma disappears. A conviction conveys no information, because everybody knows that in equilibrium anyone without a conviction is merely a criminal who got away. A unilateral deviation to noncriminal behavior by someone will not convey any information. The effect of stigma and which equilibrium is played out can be strongly affected by the probability and size of formal punishment. Signalling via stigma avoidance is an explanation for why people obey even unjust or arbitrary laws, because the motive for obeying the law is no longer to do what is right or to avoid guilt, but to be considered a law-abiding person. Indeed, the more arbitrary the law, the better it serves this purpose, because the less can signalling be confused with other motivations. Moreover, signalling separates obedience to the law from any direct benefit. It is not like coordination, where a deviation from convention immediately reduces one's payoff, or informative law, which teaches that obedience has a higher immediate payoff than disobedience. Schauer wonders whether stigmatization is even a function of law at all. ``When and where this is so, the role of law {\it qua} law may be less than is often thought. The state, after all, could simply publicize the names of those who engaged in widely scorned activities without making them illegal and would thus achieve its content-based goals without having to resort to the law at all" (p. 134). The decision to publicize, however, is a decision to use law. In doing so, the state labels an act as special, puts the question of whether it happened through a verification process, and ends up harming the person who took the action. The presence or absence of a direct penalty seems unimportant to whether we should call process a legal one. Indeed, as I will discuss later, Friedman (1979) has given us the example of medieval Iceland, where the government acted as a court but left enforcement to private citizens. \begin{center} {\it 4. Alternatives to Coordination and Information: Expression and Transaction Costs} \end{center} Largely missing from McAdams and Schauer are two other alternatives to law as coercion: law as expression {\it per se}, and law as the reduction of transaction costs. By expression {\it per se} I mean expressive law as expression, as declaring an opinion to the world. Sometimes a law is passed just for the sake of passing a law, independent of whether it can be enforced. The law is not meant to coerce people into acting differently, or to coordinate their actions, or to provide them with information. It is meant to make a statement, to put the feelings of its proponents on record. When a crisis arises, people want to do something. One thing they can do is to announce their opinion. Action may be better than words, but words are more satisfying than nothing at all. Part of the reason for such an announcement may be to persuade other people, as in McAdams's information theory, but another part is to salve one's own frustration. In economic terms, expression enters directly into the utility function. We all know this from committee meetings. There is always someone talking just to hear himself speak. His point may already have been made, the decision may already be clear, nobody wants to listen any more, but he wants to voice his opinion. We resent speakers like that, but their speeches are not a complete social waste, despite the externality. The speaker himself gets utility from speaking, and it is perfectly rational for him to speak even when it will have no effect on the rest of the committee. The direct benefit is sufficient. So it is with laws: they may be purely expressive. Or, laws may mix expression with other motives The segregation law at issue in {\it Brown vs. Board of Education}, 347 U.S. 483, is an example. The issue was whether segregation by race was permissible if the schools were of equal quality, since the separation would treat both races symmetrically. The court's rationale for striking down segregation was that the separation was not really symmetric, because asymmetry in expression harmed the black children: ``The policy of separating the races is usually interpreted as denoting the inferiority of the negro group. A sense of inferiority affects the motivation of a child to learn.'' The Supreme Court is still alert for state or federal expression of forbidden viewpoints; in particular, for support of religion. Despite not addressing the direct utility of expression, McAdams devotes an entire chapter, ``Normative Implications,'' to the difficulties of legal doctrine in deciding which government expressions are legitimate and which illegitimate. On a more routine level, expression is part of the function of imprisonment and corporal punishment (including capital punishment) in criminal law. As Feinberg (1965) says, ``Punishment is a conventional device for the expression of attitudes of resentment and indignation, and of judgments of disapproval and reprobration, on the part either of the punishing authority himself or of those in whose name' the punishment is inflicted." This is not the same as retribution, which with deterrence, incapacitation and rehabilitation is one of the standard four purposes of punishment. Retribution is the infliction of disutility on an offender to satisfy moral principles or desire for revenge, independent of any future effects, but such infliction of disutility can satisfy even if the fact of the punishment is not public. Public punishment announces the citizens', and especially the victims', satisfaction that justice has been done. Kahan (1996, 1998) makes the same point in his influential work. One practical implication he draws out is that fines, despite their welfare cost being low, may be undesirable if they fail to make express disapproval as well as imprisonment. If less costly punishment is desired, we should look to alternatives that still shame the criminal and condemn his action. Transaction cost reduction is another role of law that does not require coercion. We have already talked about constitutive law, coordination on definitions, the benefit of everyone using the same definition, and how government can try to choose the best available equilibrium. A distinct motivation for an individual is to use default provisions of the law instead of having to craft provisions himself. Consider a contract. It is, of course, good to know that the Uniform Commercial Code, the basis for the state law of sales in the United States, standardizes the definitions used in contracts, and that both parties can be in agreement as to the terms. Contracts will often specify the state law by which they are to be governed, e.g. ``This agreement shall be governed by the laws of the State of Indiana.'' But although it is important to be in agreement with the other party to forestall disputes, an equally important concern is to write an efficient contract. Party 1 wants sensible treatment of unforeseen contingencies whether or not party 2 cares or not. By agreeing to be governed by the law provided by the state, Party 1 avoids having to write a contract covering all possible contingencies. Instead, by saying nothing, the contract is implicitly agreeing to use the default provisions legislators enacted in statutes and the case law courts have developed to fill the statutes's gaps. The parties are free to override most provisions of state contract law. They may, for example, say that instead of UCC \S2-308's default provision that ``the place for delivery of goods is the seller's place of business or if he has none his residence,'' delivery will be to certain geographic coordinates. If they don't mention the place of delivery, though, the law provides a sensible default. Use of the law's defaults also reduces another transaction cost: what I call ``contract reading costs'' in Rasmusen (2001). Hard as it is to write the drafting party to write a long contract, it is can be even more difficult for the other party to read it carefully enough to find what advantages the drafting party inserted for himself. It's hard to put a landmine in a meadow, but even harder to detect it afterwards. A short contract leaves the missing terms to state law, a neutral party. Searle's idea of constitutive law, law that creates opportunities, is close to the idea of law as reduction in transaction costs. The only difference is that putting it in terms of transaction costs emphasizes that what is going on is not really the creation of new opportunities but the reduction in cost of transactions that could still take place without law. A detailed contract can substitute for default rules; it's just more expensive. The parties can even create almost all of the features of a corporation--- governance, share ownership, and transferrability of ownership--- by crafting a partnership agreement specifying all of those things. The only feature that cannot be inserted without statutory permission is limited liability, since that concerns third parties, who are not in a position to agree or disagree, but even limited liability's benefits can be largely obtained by using a limited partnership (in which only the general partner is liable) or by adding a clause requiring tort insurance. Even in our example of making a will, it is possible to write a trust agreement transferring enough characteristics of ownership while the testator is still alive to make it difficult for the son to displace the favored nephew. A trust does this by giving control over the property to trustees, who survive the testator. In effect, this happens in monarchies and dictatorship when the sovereign is of doubtful legitimacy and can't rely on law. Instead, the tyrant uses partial transfer of power during his lifetime to make the succession more certain. Think of Kim Il-sung succeeding Kim Jong-il in Korea in 1994, or the succession of Augustus by Tiberius in Rome in A.D. 14. The advantage of law is that it the testator and dictator do not have to be so clever in organizing their affairs. To be sure, custom and standard-form contracts are also means of reducing transaction costs, and private law and private courts could replace government law and government courts, in theory. We will address this topic of competing sources of law next. \begin{center} {\it 5. Is the State Necessary for Law?} \end{center} Without the state, would law disappear? Even when the state exists, it is not the only source of rules. Universities, corporations, football leagues, churches, condominium associations, and families all have rules. Often, they even call them ``laws''. All these groups except the family are voluntary associations, which can enforce their rules by expulsion, or by lesser penalties with the threat of expulsion for contumacy. Indeed, a contract between two parties can itself be thought of as a voluntary association, with money penalties--- though if it is a short-term relationship, the state's law is still needed. In long-term relationship, even in contracts expulsion (that is, refusal to deal further) is the common penaltys, as evidenced in the detailed investigations by Bernstein (2015) and Bozovic and Hadfield (2015). Even if the state exists, it does not have to retain its coercive powers to enact laws enforced by coercion. That sounds paradoxical, but paradox is resolved when we realize that the state can let private actors enforce the law. Friedman (1979) tells us that this actually was the case in medieval Iceland. Iceland had a weak government---if one can even call it government--- with courts whose function was to identify what happened, but not to inflict punishment. In the case of killings, for example, the court would determine whether the killer was liable for damages paid to the victim's family. If the killer was found liable and did not pay, or if he attempted to conceal the death instead of announcing it publicly, he was declared an outlaw, whom anyone could kill without having to pay a penalty. An example closer to home is the right to self-defense. If Smith tries to murder Jones, Jones has the right to fight back, and even to kill Smith if that much coercion is required to stop the crime. Max Weber (1919) defined the state as having a monopoly on the legitimate use of force. Iceland and self-defense are examples of a qualification he had to add: this has to include state permission for private citizens to use force when appropriate. To me, it seems the exception undoes the definition. Would we say the state has a monopoly on farming because it permits people to use land for farming under appropriate circumstances? l In any case, we see that it is possible to separate the party that makes a law from the party that enforces it. Private laws enforced by nonviolent means have been the subject of much scholarship. Macaulay (1961) is the standard cite for the point that businesses generally work out disputes themselves rather than going to court. Some industries even publish formal laws. Lisa Bernstein (1992, 2001) has studied such laws in the diamond and cotton industries. At the time of her article, 80\% of the rough diamonds in America passed through the hands of the 2,000 members of the New York Diamond Dealers Club. They had their own set of formal rules, e.g., ``Any oral offer is binding among dealers, when agreement is expressed by the accepted words `Mazel and Broche' or any other words expressing the words of accord." Bernstein discusses the problems of enforcing contracts in government courts (e.g., delay) and how reputation and social ostracism work to enforce club rules, which include arbitration to resolve disputes. Similarly, the merchant-to-mill cotton trade in the United States has almost entirely opted out of the formal legal system, creating its own commercial law administered by the American Cotton Shippers Association and the American Textile Manufacturers Institute. They use arbitrators with rules special to the industry, who rely on documents and not oral hearings. Sometimes the arbitrators give reasons for their results, sometimes not, but either way, the opinions are circulated to members. Failure to comply with the decision is grounds for expulsion, and expulsions are publicized. Such failure is rare, except when a business is in severe financial distress. Trade associations are dealing with disputes which are in essence based on cooperation--- contractual disputes--- so the threat of expulsion is particularly potent. Not all private rules are formalized. The subject of social norms has generated a large literature in law-and-economics Cooter (1998), Ostrom (2000), and Posner (2009) are three of the most cited works. McAdams himself has published on this topic (notably McAdams [1997]) and has surveyed the field with me (McAdams and Rasmusen (2007)). ``Norms'' is as hard to define as ``law'', but the term is ordinarily used to represent unwritten rules of behavior--- not just common patterns of behavior, but rules which people expect others to obey and whose violation creates disapproval. Disapproval might just mean unhappiness, but some authors define norms as requiring moral disapproval (Cooter 1996; Ellickson 1991; Kaplow and Shavell 2002), McAdams (1997, 2001). Other authors would not, and use norms to refer to a morally neutral coordinated equilibrium (Picker (1997); Mahoney and Sanchirico (2001), E. Posner (2000)). Norms, like the formal rules of private organizations, often replace government law. The best-known empirical study of the relation between norms and law is Ellickson (1986) (later expanded to a 1991 book). Ellickson sets out to explore the famous example in Coase (1960) of how in the presence of clearly established legal rules, the ranchers running cattle and the farmers raising crops would negotiate efficient outcomes for the disputes caused by wandering cattle eating crops. He found that in Shasta County, California, the county government was authorized to determine what the trespass laws would be in differents parts of the county. Some were ``open range'', where the owner of cattle was not liable for grazing damage, and some were ``closed range'', where he was strictly liable. He found that the formal law was unimportant. Rather, neighbors resolved disputes by gossip, negotiation based on norms, and physical reprisal. Curiously enough, even insurance adjusters paid little attention to who was formally liable. It seems formal law was too slow, too costly, and, perhaps most important, too disruptive of social relationships. Social norms are enforced by the same means as law, prison excepted, including monetary payments backed up by incentives such as ostracism rather than by imprisonment. Guilt, shame, fear of disapproval, coordination, signalling that one is a good person, and information conveyal are all used. Even violence is used in some contexts, though more in states which do not have the ability to enforce law very well. Ellickson notes that cattle-killing did occur even in Shasta County, if rarely, and castration of a straying bull in one case. What, then, makes private law or social norms different from government law? Not much. If laws are formal rules of general applicability, based on legitimacy or coercion, there is no reason why several sets of laws cannot co-exist in one locality, used by different people in disputes of different sizes. If we require legitimacy, there is of course a problem when laws conflict, but that is also true of morals, as in the famous Trolley Problem where you must decide between flipping a switch to divert a trolley so it kills ten people or not flipping and letting it kill twenty other people (Foot [1967]). When someone says ``law'' we think of government law, a useful default meaning, but there is no contradiction in terms if we say ``private law''. Social norms sometimes have the difference that they are not formally stated by a universally recognized authority, but they work the same way and are often more powerful. The relationship between norms and law runs both ways, too: norms can support law, and law can support norms--- or they can undermine each other. Common law was originally based on norms, and even today judges look to custom to help with such questions as whether behavior is negligent. Much of law is enforced by social norms--- that is how one might classify stigma, for example, as well as morality. That is why why Schauer wishes to emphasize coercion as why law as law is obeyed. If a law is obeyed because it is aligned with a social norm, then it is not obeyed as law at all; the action is coming from the social norm and its formalization in law is unimportant. Norms and laws usually interwine in support each other, though, as with the aid law gives stigma by publicizing someone's violation of a rule and confirming (or disconfirming) the violation by an unbiased and careful trial. \begin{center} {\it 6. Conclusion} \end{center} Schauer and McAdams show that economics has become a standard tool in law, reaching even subjects as philosophical as jurisprudence. To the question of why people obey laws, economics contributes its expertise in how people respond to incentives. The incentive of fines and prison is the most obvious, but one of the big questions in jurisprudence has been whether direct penalties are an essential feature of law. Can economic reasoning help? It certainly can help by contributing to the understanding of the incentives to follow equilibrium strategies in coordination games, and to respond to information revealed by the passage of laws and the breaking of laws. Judging the importance of these indirect incentives to obey the law is profoundly difficult. When Schauer says that law makes us do things that we do not want to do, an economist might say he is violating a tautology. Doesn't someone always make the choice he thinks will maximize his utility? What he wants to do depends on both costs and benefits. He always would like to have a different choice with higher benefits and lower costs. But an alternative phrasing of Schauer's idea is that law imposes extra costs on certain decisions. As McAdams shows, however, although a law saying people must drive on the right does result in a higher cost for those who want to drive on the left, it also results in a higher benefit for those who drive on the right. Similarly, a law requiring seat belts results in higher prices for cars but also in higher utility for those it binds--- or at least most of them because it informs them that seat belts are a feature worth the cost. In the end, Schauer is probably right that coercion is the feature of law that matters the most in ensuring compliance, but McAdams also is right that in many cases coordination and information incentives matter and are worthy of analysis. Much of law lies outside both paradigms, though, and exists to reduce transaction costs. We could think of that as coordination on definitions or information about what contract clauses are most useful, but the main purpose is to avoid having to create one's own institutions for each transaction. In the end, though, the hardest to measure and most important rivals to coercion as an explanation for compliance with the law are morality and stigma, not coordination or information. How would we even quantify which is more important, coercion, stigma, or morality? Each of the three supports the other two, and only coercion changes rapidly enough to be amenable to technical analysis. We all realize that higher direct penalties reduce law-breaking, and we can make some effort to measure that using regression analysis, but if only we knew how to improve the workings of morality and stigma, might we not achieve better results at lower cost? Perhaps that is like saying that if only we had fusion power, electricity would be cheaper than using natural gas--- true but hypothetical. But it may be that Plato is right, and in the long run the best replacement for prisons would be moral education, a long-term investment in virtue. \begin{center} {\sc References} \end{center} Atkeson, Lonna Rae. 1999. `` `Sure, I Voted for the Winner!' Overreport of the Primary Vote for the Party Nominee in the National Election Studies.'' {\it Political Behavior} 21 (3): 197-215. Austin, John. 1832. {\it The Province of Jurisprudence Determined.} \url{http://www.koeblergerhard.de/Fontes/AustinJohnTheprovinceofjurisprudencedetermined1832.pdf}. Baron, David P. and T. Renee Bowen. 2015. ``Dynamic Coalitions.'' Stanford working paper. Becker, Gary S. 1968. ``Crime and Punishment: An Economic Approach.'' {\it Journal of Political Economy} 76: 169–217. Bentham, Jeremy. 1782/2010. {\it Of the Limits of the Penal Branch of Jurisprudence}. Ed. Philip Schofield. Oxford: Oxford University Press. Bernstein, Lisa. 1992. ``Opting Out of the Legal System: Extralegal Contractual Relations in the Diamond Industry.'' {\it Journal of Legal Studies} 21: 115-157. Bernstein, Lisa. 2001. ``Private Commercial Law in the Cotton Industry: Creating Cooperation through Rules, Norms, and Institutions.'' {\it Michigan Law Review} 99 (7): 1724-1788. Bernstein, Lisa. 2015. ``Private Ordering, Social Capital, and Network Governance in Procurement Contracts: A Preliminary Exploration," U. of Chicago Law School working paper. Bozovic, Iva and Gillian K. Hadfield. 2015. ``Scaffolding: Using Formal Contracts To Build Informal Relations in Support of Innovation,'' University of Southern California Law School working paper (February 25, 2015). Calvin, John. 1536. {\it Institutes of the Christian Religion}. \url{http://www.ccel.org/ccel/calvin/institutes.html.} Confucius, {\it Analects}, Book XIII, Chapter 3, verses 4-7, translated by James Legge. \url{https://ebooks.adelaide.edu.au/c/confucius/c748a/}. Cooter, Robert D. 1996. ``Decentralized Law for a Complex Economy: The Structural Approach to Adjudicating the New Law Merchant.'' {\it University of Pennsylvania Law Review} 144: 1643-1696. Cooter, Robert D. 1998. ``Expressive Law and Economics.'' {\it Journal of Legal Studies} 27: 585–608. Dau-Schmidt, Kenneth G. 1990. ``An Economic Analysis of the Criminal Law as a Preference-Shaping Policy.'' {\it Duke Law Journal} 1990: 1-38. Ellickson, Robert C. 1986. ``Of Coase and Cattle: Dispute Resolution among Neighbors in Shasta County.'' {\it Stanford Law Review} 38 (3): 623-687. Ellickson, Robert C. 1991. {\it Order without Law: How Neighbors Settle Disputes.} Cambridge, MA: Harvard University Press. Farrell, Joseph and Matthew Rabin. 1996. {\it The Journal of Economic Perspectives} 10 (3): 103-111. Feinberg, Joel. 1965. ``The Expressive Function of Punishment.'' {\it The Monist} 49 (3): 397-423. Foot, Philippa. 1967. ``The Problem of Abortion and the Doctrine of the Double Effect.” {\it Oxford Review} 5: 5–15. Friedman, David. 1979. ``Private Creation and Enforcement of Law: A Historical Case.'' {\it The Journal of Legal Studies} 8 (2): 399-415. Fudenberg, Drew and Eric Maskin (1986) ``The Folk Theorem in Repeated Games with Discounting or Incomplete Information.'' {\it Econometrica} 54 (3): 533-554. Funk, Patricia. 2004. ``On the Effective Use of Stigma as a Crime-Deterrent.'' {\it European Economic Review} 48: 715-728. Glazov, Jamie. 2005. ``Our Culture, What’s Left of It.'' {\it FrontPageMagazine.com}. August 31, 2005. \url{http://archive.frontpagemag.com/readArticle.aspx?ARTID=7445}. Gneezy, Uri and Aldo Rustichini. 2000. ``A Fine is a Price.'' {\it The Journal of Legal Studies} 29 (1): 1-17. Haltiwanger, John and Michael Waldman. 1991. ``Responders Versus Non-Responders: A New Perspective on Heterogeneity.'' {\it The Economic Journal}, 101 (48): 1085-1102. Harel, Alon and Alon Klement. 2007. ``The Economics of Stigma: Why More Detection of Crime May Result in Less Stigmatization.'' {\it Journal of Legal Studies} 36: 355-377. Hart, H. L. A. 1961/1997. {\it The Concept of Law. } 2nd edition. New York: Oxford University Press. Holmes, Oliver Wendell Jr. 1897. ``The Path of the Law.'' {\it Harvard Law Review } 10 (8): 457-478. Kahan, Dan M. ``What Do Alternative Sanctions Mean?'' {\it University of Chicago Law Review} 63 (2): 591-653. Kahan, Dan M. 1998. ``Social Meaning and the Economic Analysis of Crime.'' {\it The Journal of Legal Studies} 27 (S2): 609-622. Kaplow, Louis and Steven Shavell. 2002. {\it Fairness versus Welfare} Cambridge, MA: Harvard University Press. Macaulay, Stewart. 1963. ``Non-Contractual Relations in Business: A Preliminary Study." {\it American Sociological Review} 28 (1): 55-67. Mahoney, Paul G. and Christopher Sanchirico. 2001. ``Competing Norms and Social Evolution: Is the Fittest Norm Efficient?” {\it University of Pennsylvania Law Review} 149: 2027-2062. McAdams, Richard H. 2009.``Beyond the Prisoners' Dilemma: Coordination, Game Theory, and Law.'' 82 {\it Southern California Law Review} 82: 209-258. McAdams, Richard H. 1997. ``The Origin, Development and Regulation of Norms.” {\it Michigan Law Review} 96 (2): 338-433. McAdams, Richard. 2015. {\it The Expressive Powers of Law.} Cambridge, MA: Harvard University Press. McAdams, Richard and Eric Rasmusen. 2007. ``Norms and the Law.'' {\it Handbook of Law and Economics, Vol 2}, eds. A. Mitchell Polinsky and Steven Shavell, chapter 20, pp. 1573-1618. Amsterdam, Elsevier. Merrill, Thomas W. and Henry E. Smith. 2000. ``Optimal Standardization in the Law of Property: The Numerus Clausus Principle.'' {\it The Yale Law Journal} 110 (2): 1-70. Picker, Randal C. 1997. ``Simple Games in a Complex World: A Generative Approach to the Adoption of Norms.'' {\it University of Chicago Law Review} 64: 1225-1287. Plato {\it The Laws}, tr. Benjamin Jowett, Project Gutenberg, \url{http://www.gutenberg.org/files/1750/1750-h/1750-h.htm#link2H_4_0015}. Posner, Eric A. (2000) {\it Law and Social Norms} Cambridge, Mass: Harvard University Press. Ramseyer, J. Mark and Eric Rasmusen. 2011. ``Can the Treasury Exempt Companies It Owns from Taxes? The \$45 Billion General Motors Loss Carryforward Rule." {\it The Cato Papers on Public Policy} 1 (1): 1-54. Rasmusen, Eric. 1994. ``Judicial Legitimacy as a Repeated Game.'' {\it Journal of Law, Economics, and Organization} 10: 68-83. Rasmusen, Eric. 1996. ``Stigma and Self-Fulfilling Expectations of Criminality.'' {\it Journal of Law and Economics} 39 (2): 519-543. Rasmusen, Eric. 2001. ``Explaining Incomplete Contracts as the Result of Contract-Reading Costs.'' {\it BE Press: Advances in Economic Analysis and Policy} 1 (1): Article 2. \url{http://www.bepress.com/bejeap/advances/vol1/iss1/art2}. Schauer, Frederick. 2015. {\it The Force of Law}. Cambridge, MA: Harvard University Press. Schelling, Thomas C. 1981. {\it The Strategy of Conflict.} Cambridge, MA: Harvard University Press. Searle, John R. (1969) {\it Speech Acts: An Essay in the Philosophy of Language} Cambridge: Cambridge University Press. Shakepeare, William. 1591. \url{http://shakespeare.mit.edu/3henryvi/full.html}. Weber, Max. 1919. {\it Politics as a Vocation.} \url{http://anthropos-lab.net/wp/wp-content/uploads/2011/12/Weber-Politics-as-a-Vocation.pdf}. Weingast, Barry R. 1997. ``The Political Foundations of Democracy and the Rule of the Law.'' {\it American Political Science Review} 91 (2): 245-263. Wright, Gerald C. 1993. ``Errors in Measuring Vote Choice in the National Election Studies, 1952-88.'' {\it American Journal of Political Science} 37 (1): 291-316. \end{document} 11959 words, July 30.