To the modern ear, “law enforcement” implies police, but police are only one of several mechanisms for enforcing legal rules. If someone breaks your arm you call a cop; if he breaks your window you are more likely to call a lawyer. Criminal law and tort law are mildly different mechanisms for doing the same work. Someone does something bad to someone, the legal system is called in, something bad happens to him, and that is a reason not to do bad things to people. In one case the offender is found and prosecuted by government employees, police and public prosecutors, in the other by private actors, the victim and his agents, but the basic logic is in both cases the same. Sometimes either or both approaches can be used; O.J. Simpson was first acquitted of the crime of killing his wife and then convicted of the tort of killing his wife.
Lots and Lots of Ways of Doing it
Since we are looking at the legal systems of a wide variety of societies, I will spend this chapter listing all the enforcement mechanisms I know of and considering the problems with each. As you will see, it is sometimes hard to draw sharp lines between them. As a first cut at classification it is useful to think of the task to be done as having three parts, corresponding, in the criminal case, to the jobs done by:
1. police and public prosecutors—finding the offender and proving his guilt
2. judge and jury—deciding guilt
3. police, prisons, etc.—enforcing the punishment.
Under criminal law, all three jobs are done by professionals employed by a government to do them. Tort law is criminal law with one change—the police and public prosecutors are replaced by the victim and his agents. It is their job to figure out who committed the tort, gather evidence, and convince the court. The police and public prosecutors are professionals who do it because that is the job they are hired to do. The tort victim does it in order to collect damages; what the tortfeasor pays as penalty the successful victim receives as a reward. The victim may also have the additional incentive of private deterrence, establishing the reputation of a bad person to commit torts against, as discussed in Chapter XXX.
Privately prosecuted criminal law is intermediate between the first two. Prosecution is by a private party who need not be the victim; in both the English and Athenian cases, any adult male citizen could prosecute any crime. In Athens the immediate incentive was a reward, a share of the fine that the criminal defendant paid if convicted. Prosecution could also be motivated by personal hostility or political rivalry. In England, during most of the 18th century, there was no reward for a successful prosecution. The incentives were private deterrence and the possibility of extorting an out of court settlement from a defendant who, if convicted, was at risk of being hanged or transported.
All three of these approaches to enforcement—criminal, tort, and privately prosecuted criminal—are broadly similar in their approach. Someone violates a rule, someone—police officer, tort victim, or private prosecutor—convinces a court that a rule has been violated and shows by whom, the court delivers a verdict and the state enforces it.
For a very different approach to achieving the same objective consider a private system such as that of the Rominchal. Jerry violates a rule, harming Larry. Larry threatens Jerry with violence if he does not compensate Larry for the harm. The three jobs of criminal law are all done by the victim. He is, simultaneously, policeman, prosecutor, judge, jury, and executioner.
At first glance, such a feud system sounds more likely to produce bloodshed than law enforcement, but in practice that need not be the case. Jerry's ability to use force against Larry is limited by the existence of Cary, Gary, Harry, … potential allies of both parties. If they believe that Larry''s demand is unreasonable and Jerry justified in refusing it, they will be unwilling to support Larry and likely to support Jerry, making the attempt by Larry to carry out his threat hazardous in the extreme. Thus a feud system can successfully enforce a set of rules understood and accepted by the participants. Enforcement depends on the threat of violence, but that is true of other mechanisms for enforcing law as well. The difference is that violence by one private party against another is constrained by the symmetry of the situation; unlike the state, the initiator of a feud does not have access to enormously greater resources than the defendant.
Saga period Iceland had a legal system intermediate between tort and feud. As in our first three cases, there was an explicit body of law and a system of courts to judge whether it had been violated and, if so, with what legal consequences. But the final step of enforcing the verdict was done privately by the tort victim and his allies. If the defendant lost his case and refused to pay the resulting fine, the plaintiff could have him outlawed and, if he failed to leave Iceland in the next two weeks, be free to hunt him down and kill him. And a victim who chose not to go to court had the option of demanding compensation from the offender as in a feud system and, if it was not forthcoming, using violence against him—at the risk of provoking retaliation in the form of either return violence or a law suit.
All of the ways of enforcing law that I have so far described depend on force or the threat of force. For a very different mechanism, imagine that you have bought a jacket from a department store that guarantees to refund your money if you are not satisfied with what you bought. If, when you discover that the jacket you bought is the wrong size and your wife points out that purple is not really your color, the store refuses to give you a refund, you are unlikely to sue them—the amount at stake is not enough to make it worth the time and trouble. Nonetheless, almost all stores in that situation will, at least in my experience, take the product back—because they want the reputation, with you and with others, of living up to their promises.
A more elaborate example of the same approach is provided by the New York diamond industry as described in a classic article by Lisa Bernstein. At one point, somewhat before the time she studied it, the industry had been mostly in the hands of orthodox Jews, forbidden by their religious beliefs from suing each other. They settled disputes instead by a system of trusted arbitrators and reputational sanctions. If one party to a dispute refused to accept the arbitrator’s verdict the information would be rapidly spread through the community, with the result that he would no longer be able to function in that industry. The system of reputational enforcement survived after membership in the industry became more diverse, with organizations such as the New York Diamond Dealer’s Club providing both trusted arbitration and information spreading.
The reason the department store or diamond merchant is concerned about his reputation is not fear of being disliked but of losing business. The reason your friend will shop at another store if you tell him that this one refused to take your jacket back is not that he wishes to punish the store for cheating you but that he does not himself want to be cheated. Reputational enforcement works by spreading true information about bad behavior, information that makes it in the interest of some who receive it to modify their actions in a way which imposes costs on the person who has behaved badly. Deterrence is not anyone’s objective, merely a side effect of interested third parties acting in their own interest.
Next consider the same sort of enforcement, this time used as a deliberate punishment. A community finds one of its members guilty of violating its rules and pronounces a sentence of ostracism—in the context of the Vlach Rom described in Chapter XXX, a sentence of marimé imposed by a kris romani. Other members of the community refuse to associate with the convicted defendant. If, as in the case of a gypsy community, ostracism imposes very large costs on its target the threat provides a reason not to violate the rules of the community.
There is, however, a problem with this enforcement mechanism. The reason not to associate with the convicted defendant is in order to enforce the rules of the community. That is what economists call a public good; an individual who goes along with it bears all of the cost of giving up any mutually profitable dealing he might otherwise have with the defendant, while the benefit is shared with all members of the community. That makes it unlikely that an individual's share of the benefit will be larger than his cost.
One solution is the following rule:
You shall not associate with anyone who has violated a communal rule—including this one.
To see how this works, start at a time when everyone accepts and follows all of the communal rules. Jerry violates one of them and is sentenced to ostracism. A second member of the community, Larry, has to decide whether to help enforce the sentence by refusing to deal with him. The cost/benefit calculation described above suggests that doing so is probably not in Larry's interest, but there is now an additional cost to be considered—continuing to associate with Jerry will itself be a rules violation and will subject Larry in turn to ostracism. That cost makes in in Larry's interest to refuse to deal with Jerry. The members of the community (no longer including Jerry) are in what is described in game theory as a Nash equilibrium; each of them is following the best strategy for himself (refusing to associate with rules violators), given how the others are acting. Rewind our film a little to before Larry violated our rule and the whole community, Larry now included, is in a Nash equilibrium; given how everyone else is acting, it is in Larry's interest not to violate communal rules, and the same is true for all the others.
So far I have assumed a setting with a formal mechanism for imposing ostracism. Informal norm enforcement is a more familiar version of the same logic. There is no legal rule that forbids me from teaching classes stripped to the waist, but doing so would be imprudent. Many of my colleagues would conclude that I was not the sort of person they wished to associate with—in part because if they continued associating with me, their colleagues might reach a similar conclusion about them.
The Nash equilibrium provides one way in which ostracism can be a workable mechanism for enforcing rules. Before offering a second solution to the public good problem, I need to introduce one other approach to rules enforcement: Divine.
Consider an orthodox Jew who keeps kosher even when he is sure nobody is watching, or a Muslim convert who suppresses his remembered taste for pork. Either can be seen as an example of rule enforcement by the threat of divine sanctions, whether a stroke of lightning or refusal of entry to heaven. Alternatively, either can be seen as obedience to a command not from fear of punishment but because obedience is seen as right, what one ought to do. There is a whole continuum of variants, with the believer in fear of divine punishment at one end, Mencken's atheist brought up to believe that gentlemen do not cheat at cards at the other; for simplicity I lump all of it into the category of divine enforcement, even though some examples depend on neither the existence of a divinity nor the threat of punishment. What they have in common is that they depend on either immaterial costs, such as guilt from behaving as you believe you ought not to, or the belief in material costs, such as being struck by lightening, due to supernatural causes.
Divine enforcement provides not only an independent mechanism for rules enforcement but also a way of facilitating other mechanisms. The point of swearing to tell the truth, the whole truth, and nothing but the truth, so help me God, is that the oath makes it more likely that you will actually do it, for fear of divine punishment or out of belief in your obligation to act as God wishes. The reason to allow the defendant in a suit under Jewish law to “swear and be quit,” defeating the suit (Chapter XXX), or the plaintiff, in a case where the evidence is stronger, to “swear and take,” is that plaintiff and defendant, as believing Jews, will be very reluctant to swear falsely. Both the criminal and civil mechanisms for rules enforcement depend on a court being able to judge guilt or innocence, doing that is easier if the court is provided with a lie detector, and the practice of swearing oaths by believers reluctant to swear falsely for fear of divine punishment provides one.
Divine intervention can also substitute completely for a court trial. If God supports the right, trial by combat or ordeal vindicates the innocent and convicts the guilty. Even if divine intervention is only mythical,, the belief in it gives an advantage to the party who knows he is innocent over the party who knows he is guilty.
Divine enforcement could also provide another solution to the public good problem faced by enforcement via the threat of ostracism. One feature of marimé, from the point of view of a traditional gypsy, is that it is contagious. Another is that being polluted is not merely a violation of communal norms, it also has serious real world effects, bad luck and related risks. That provides members of the community a reason not to associate with a fellow gypsy under sentence of marimé for failing to conform to the verdict of a kris Romane.
For another example of the same pattern, consider the mechanism by which the Cheyenne indians capped the level of physical violence by one Cheyenne to another. Fighting, even quite violent fighting, was allowable. But a Cheyenne who killed another member of the tribe, for whatever reason, was banished from the tribe—not because he had been wicked but because he now smelled of death, and the smell, and its consequences in ill fortune, were contagious. After a period of years he could petition for readmission, and the petition might be granted, especially if the circumstances of the killing made it appear excusable. But for the rest of his life, nobody else would drink from his bowl or share his pipe, because they, and he, still smelled of death, even if time had diluted the effect enough to make his presence in the tribe again tolerable.
And What is Wrong With Each of Them
All of these methods for enforcing rules have been used by real world societies; they all work. Sometimes—all have also, at various times and places, failed to work or worked very imperfectly. Why? What are the problems with each of the alternative approaches? I start with criminal law.
In an ideal system of criminal law enforcement, everyone involved in the project, from prison guards and police officers up to the politicians ultimately responsible for selecting and controlling everyone else, will act in the general interest of the population they serve, doing their best to convict the guilty and acquit the innocent while minimizing unnecessary costs of the legal process.
That is a desirable outcome, but a hard one to achieve. Politicians, like the rest of us, are more interested in serving their own objectives than those of other people. If, in a high profile case, the real criminal can not be found, there may be much to be said for finding someone else who can be convicted; what the voters don't know won't hurt them. If campaign contributions are needed, organized groups of government employees, such as cops and prison guards, are one place to get them, which gives the politician an incentive to follow the policies those those interest groups support. In California, the prison guard union routinely, and often successfully, lobbies against policies that would reduce the number of prisoners and thus the demand for their services.
Some costs of enforcing criminal law are paid by the law enforcement agency that incurs them, giving it an incentive to hold them down.Other costs are imposed on other people without compensation and so may, whenever those people have insufficient political influence to make their complaints matter, safely be ignored. Criminal law in imperial China provides multiple examples. Torturing a witness or an innocent defendant imposes a cost on him but not on those who impose the torture, so there is little reason for them to take account of that cost in deciding when to employ torture and when to refrain from doing so.
The same logic applies in modern systems. Holding a suspect in jail imposes a cost on him. Shutting up a restaurant where a crime has occurred until all relevant evidence has been collected imposes a cost on the restaurant owner and his customers and employees. Seizing computers containing all the copies of important information—the only draft of a book or a doctoral thesis, say (real cases)—and holding them until the police are certain that they no longer contains evidence of interest to them, imposes costs, possibly very large costs, on the computers' owner, but costs nothing to the police. Not only do they have the option of ignoring costs they impose on others, they have the option of deliberately imposing such costs as a way of punishing people whose legal behavior they disapprove of. If the police smash down my door, shoot my dog, scare my children half to death, and only them discover that they have come to the wrong address, they have no obligation to compensate me for the damage they have done, hence little reason to take extra care to avoid such costly (to me, not them) errors.
Focussing less on police and more on the whole system, it is far from clear how to make it take account of one of the most serious costs it imposes on other people, the cost to defendants who end up convicted of and punished for crimes they did not commit. For a discussion of that problem, in the context of criminal law and alternative enforcement systems, see the next chapter.
A different sort of incentive problem was pointed out in an old article by George Stigler and Gary Becker. Imagine that I am a cop, you are a criminal, and I have the goods on you, the evidence that will get you convicted. The punishment that will result is, from your standpoint, equivalent to a hundred thousand dollar fine. If I bring in you and the evidence, I will be rewarded with a gold star on my report card, an improvement in my reputation with my employer that will result in a twenty thousand dollar increase in my lifetime income.
From the standpoint of Dragnet, the rest of the story is obvious: I turn the evidence over to the D.A., you are arrested, tried and convicted. From the standpoint of economics, the rest of the story is also obvious. I have something that is worth twenty thousand dollars to me and a hundred thousand dollars to you. Resources move to their highest valued use; you pay me fifty thousand dollars, I give you the evidence, you burn it. In order for the criminal model of law enforcement to work, something has to prevent this obvious and mutually profitable transaction, which means that we need a second cop watching me, which may substantially raise the cost of law enforcement and may or may not entirely eliminate the problem.
Becker and Stigler proposed a solution the problem they described; stop paying salaries to police officers and instead compensate them with a reward equal to the penalty paid by those they convict. Now the only bribe I will accept is one of at least a hundred thousand dollars, in which case you have paid your fine and I have received my pay. We have simply cut out the middleman.
Richard Posner and William Landes responded to Becker and Stigler with an article of their own, pointing out a number of problems with the proposal, among them the problem of allocating the right to catch a criminal; I would not want to spend lots of time and effort collecting the evidence only to have a competitor arrest the criminal first and collect the fine. One simple solution is to give the right to the victim and let him sell it to whomever offers him the highest price. Work through a few organizational details and we have privatized law enforcement, replacing police departments with firms that buy offenses from the victims, catch and convict offenders, and support themselves on the resulting fines.
Becker and Stigler had, without realizing it, reinvented tort law, although not in precisely the current form. The victim of a tort owns the right to collect the fine—called a damage payment—paid by the convicted tortfeasor. He can transfer part of that right to a private enforcement agency—we call it a law firm—that is willing to take the case on a contingency basis. If the law firm convicts the defendant some of what he pays goes to the firm to pay for its efforts, some to the victim as the price at which he has sold his claim. Which brings us to our next topic.
Tort law, as the arguments of Becker and Stigler suggest, in part privatizes the job of enforcing legal rules; by doing so it solves some, but not all, of the problems raised by criminal law. A tort plaintiff can impose some costs on the defendant without compensation, such as the cost of paying for a lawyer to defend against the suit or of responding to a discovery demand, but, not being protected by the principle of sovereign immunity, he is less able to do so than a police department. The lawyer representing the plaintiff might, like the corrupt policeman, sell out to the other side, but there is already someone in whose interest it is to prevent that from happening—the plaintiff. Many of the arguments that suggest that a competitive private market usually does a better job of producing goods and services than a government monopoly apply here, where the service being produced is that of detecting and proving violations of legal rules. We do not usually think of government enforcement of criminal law as a form of socialism but, economically speaking, that is what it is—government ownership and control of the means of production—and the usual arguments for the inefficiency of socialism apply.
Tort law brings with it, however, a new set of problems. Consider first tort law as it now exists, in which the damage payment owed by the convicted tortfeasor is “enough to make his victim whole,”to compensate him for the damage that the tort did to him. The damage payment is both the penalty to the convicted tortfeasor and the reward to the plaintiff and his lawyer for convicting him, and there is no good reason to expect it to be the proper sum for either purpose.
Consider, for instance, a tort which is difficult to detect and prove, with the result that half the tortfeasors escape unpunished. The damage done is (say) ten thousand dollars, so that is what convicted tortfeasors pay, making the average penalty five thousand. If deliberately committing the tort benefits me or precautions to prevent it cost me more than five thousand dollars, I am better off committing the tort and (sometimes) paying for it, even though what saves me (say) six thousand dollars is costing you ten, and so on net makes us worse off. So some torts that should be deterred will not be.
The problem is worse than that, because we must consider the incentives of the victim as well as those of the potential tortfeasor. If spending six thousand dollars on a law suit buys me only a fifty percent chance of collecting a ten thousand dollar damage payment, suing you makes me poorer, not richer, so I don't. And if I am not going to sue you, the penalty you pay for committing a tort is not five thousand dollars but zero.
One solution to this problem might be a probability multiplier for damage awards. If damage done is ten thousand but the probability of successful suit is only fifty percent, set the award at twenty thousand. Now the average punishment is equal to the damage done, giving potential tortfeasors an incentive to prevent any tort whose prevention costs less than the damage done by not preventing it. Some scholars, including Landes and Posner, have suggested that punitive damages, damage awards greater than damage done, serve just that purpose.
While this might be an attractive solution in some cases, it too has problems. For one thing, the damage award is limited to what the tortfeasor can pay; in the case of a serious offense with only a low probability of successful prosecution, that may be less, perhaps much less, than the damage done scaled up to allow for the low probability of conviction.
A second problem is that a probability multiplier may make it profitable to create fictitious torts. Late at night, as your car comes around the bend, I shove mine into the road and hastily stand back. When the dust has cleared and you have gotten free of the wreckage—I considerately staged my fake accident on a slow road, so that you would survive to be sued—I commence a legal action, claiming five times the value of my car on the grounds that it was only by great good luck that you did not succeed in escaping after smashing my car and that four friends of mine just happened to be lurking in the underbrush to witness the accident and testify against you. In order to frame you I had to create a real accident, so it was made profitable only by the existence of a probability multiplier.
So far, I have assumed that the tort victim only sues in order to collect damages, hence will not sue if damages come to less than the cost of suit. If that were always the case, then privately prosecuted criminal law without rewards, the criminal law of England through most of the 18th century, could not have worked, since there was no damage payment to compensate for even modest costs of prosecution. Yet that system, described in Chapter XXX, existed and functioned.
Part of the explanation was that plaintiff and defendant together could convert a criminal punishment, such as hanging, into a tort punishment by a covert out of court settlement. Another part is that potential victims could precommit to prosecute and inform potential felons of that commitment, making deterrence a private good, hence one the potential victims were willing to pay for.
Another assumption I have been making is that the tort victim has sufficient resources to prosecute if he chooses to. That may not always be the case. If we shift the story from 21st century American to tenth century Iceland, the man whose brother has been killed be old, weak, and with no remaining relatives willing to help him press his suit against offenders perfectly willing, if they can get away with it, to beat him up when he tries to go through the legal procedures required for a law suit.
The Icelanders had a simple solution to that problem; tort claims were transferable. I may not have the resources to press my claim, but one of my neighbors, a wealthy farmer with four strong sons who went a-viking in their youth and lots of relatives and allies, does. Successfully pressing the claim results in collecting several hundred ounces of silver as wergeld, tort damages for the killing. I transfer my claim to my neighbor, he prosecutes it successfully, keeps whatever share of the profits we have agreed to and gives me the rest.
The same solution could be applied in a modern context. A tort victim in our legal system has the option of finding a lawyer willing to take the case on a contingency basis, but he has no way of knowing which lawyer will do a good job of winning a hefty award and which will sell him out for a concealed side payment from the other side. If tort claims were transferrable, he could sell his to the highest bidder, at which point he no longer cares how well it is prosecuted.
Transferable tort claims could solve another problem in our system as well. Consider a tort that does a small amount of damage to each of a large number of victims, small enough so that no individual victim or small group will find it worth the trouble of suing. The current solution is a class action; an enterprising lawyer gets himself named as attorney for the class of all victims, sues on their behalf, and collects damages or accepts an out of court settlement. One problem with that solution is that there is nothing much to keep the attorney from acting in his own interest instead of that of his imaginary clients, settling on terms that give him a substantial sum in real cash and them compensation in the form of discounts on their hypothetical future dealings with the defendant.
Transferable claims make possible a better solution. The lawyer purchases a large number of small claims, perhaps with the assistance of middlemen, then sues on his own behalf as their owner. In this respect, at least, our legal system is a mere eleven hundred years behind the cutting edge of legal technology.
So far I have explored problems with the two legal mechanisms central to modern legal systems, hence the two most familiar to my readers, especially any of them who happen to be lawyers. The third distinct system, feud, is less familiar; inasmuch as moderns have opinions on it they are highly colored and almost entirely negative. Yet it too is a system that can function and has functioned for enforcing legal rules. What problems limit its ability to do so?
One is that it depends, to a considerable extent, on the availability of commitment mechanisms. The benefit of it being known that if I am injured I will revenge myself, whatever the cost, may be very great, since it sharply reduces the risk both of being injured and of having to bear the costs of revenge. But after I have suffered a serious injury from a formidable opponent, the cost of revenging myself, assuming he refuses to offer adequate compensation, may be more than the benefit is worth—better to be a live wimp than a dead hero. The solution, as in the parallel case with private prosecution of criminal law, is to find some way of committing myself in advance. In 18th century England it was done by prepayment to a prosecution association. In a less developed society, it depends more on some combination of social norms and internal incentives.
Seen from this standpoint, the human trait of vengefulness is not an irrational passion but a hardwired commitment strategy, a psychological pattern whose result appears irrational after the fact, when I have to do my best to hunt down the killer of my kinsman at considerable risk to myself, but rational ex ante, when the knowledge that I will hunt down anyone who kills my kin is one reason my kin do not get killed, and the knowledge that they will hunt down anyone who kills me one reason I don't.
Part of the modern hostility to feud is the belief that its violence is unbounded in both time and magnitude, with a minor conflict generating seas of blood for a century or more. Part of the basis for this view, at least among Americans, is the legendary Hatfield/McCoy feud—legendary not only because it is famous but also because it is mostly fictional. In the appendix to this chapter, I sketch the real history. The parts that fit the feud stereotype of revenge killing total one exchange in which three members of one family killed one member of the other and were themselves killed in revenge, and a second, five years later, in which a renewal of the conflict set off by the intervention of the governor of Kentucky resulted in one death on one side, revenged by two on the other. The rest of the famous feud exists in movies and newspaper stories but not in the work of careful historians, illustrating my rule of thumb for evaluating historical accounts: View with suspicion any anecdote that makes a good enough story to have survived on its literary merits.
Moderns are less familiar with the much more extensive record from societies such as saga period Iceland. A careless reading of the sagas makes it sound like a violent society, until you notice that two successive chapters, recounting two violent clashes, are separated by enough time for a character not yet born at the time of the first to participate in the second as an adult. A more careful examination suggests that most feuds terminated either at the first round, when the offender agreed to pay the penalty set by court or arbitrator, or at the second, after the victim of the first clash had reversed roles for the second. Only a few continued beyond that for multiple exchanges—and so, with the boring sections removed, provided the material for sagas.
If I honestly believe that the compensation you are offering me for the damage you and your allies did to me is absurdly low while you regard it as generous, we may have a hard time finding any settlement that does not make at least one of us feel as though he has abandoned his own commitment strategy, making him fair game for any future aggressor. So a well functioning feud system may require, in addition to mechanisms for commitment, enough commonality among the participants to make it possible for the two sides to agree on settlements, whether negotiated between them or arranged by a third party arbitrator. An alternative may be a mechanism whose authority is commonly recognized, such as the Icelandic court system. Even if I believe that what the court awarded me was unreasonably little I can still accept it without signaling my unwillingness to defend my rights against future aggressors, since they will have no reason to expect the court in the cases spawned by future clashes to err in their favor.
Reputational enforcement works well for parties engaged in repeat voluntary transactions; once you have obtained a reputation forfailing to fulfill your contractual obligations, other people become unwilling to deal with you. Anticipating that, you fulfill your obligations. It does not work for a one time transaction, a con-man who intends to cheat his victim out of a fortune and then retire. Nor does it work for involuntary transactions; a mugger does not require the consent of his victims. And even in the case of repeat voluntary transactions, there is a risk of end game betrayal, the trader who, having built up a reputation for trustworthyness by twenty years of honest dealing, spends the final year before his planned but unannounced retirement borrowing money he has no intention of paying back and taking payment for orders he does not plan to fill.
These problems may be ameliorated in several ways. Even if I am only an occasional participant in any particular market, I routinely engage in other transactions, social, economic, political. If I sell you a car which I know, and you do not know, is about to fall apart, I suffer a reputational penalty even if I never plan to sell another car. The fact that I cheated you over a car will be seen as evidence of bad character, implying that I am likely to mistreat other people in other ways. One way of making sense of our moral judgements is that we put two different acts into the same moral category precisely because we believe that the willingness to do one correlates positively with the willingness to do the other. Once enough people have heard your story and inspected your car, I will find landlords unwilling to rent me a room, employers reluctant to hire me, fathers refusing to allow me to date their daughters.
Mugging is an involuntary transactions, but even muggers depend on voluntary transactions for much that they value. In a society where people know each other well enough for reputational enforcement to work well, the mugger, even if he has never been convicted, even if his society has no courts to convict him in, may find that his reputation costs him in that part of his life that does not consist of mugging people.
The argument can be expanded beyond the loss of trading opportunities. Humans are social animals; most of us want to be liked, want to be admired, want to be seen as high status. Being viewed as someone who cannot be trusted costs in all those dimensions, which may be a reason, even for someone planning no repeat transactions that depend on trust, to choose to be trustworthy. Human beings reveal their feelings, values, thoughts in facial expressions, voice tones, gestures. That makes it difficult, although not necessarily impossible, to maintain the reputation of being an honest, trustworthy, generous person while awaiting a good opportunity to swindle all and sundry. Good con men exist, but they are rare. Reputational enforcement becomes a little harder in the case of firms, which have no faces to show expressions and which can discard a bad reputation by filing a new set of incorporation papers under a new name.
There is a further problem with reputational enforcement, one which becomes more serious the larger the society. When you cheat me by failing to fulfill your side of a contract and I complain, you respond by explaining to any who ask that you only pulled out of the contract because I had failed to fulfill my side of the bargain. An interested third party with ne easy way to tell which of us is lying may conclude that it would be prudent not to trust either of us. Anticipating that response I conclude that public complaint about your default will only make a bad situation worse—better to take my losses and keep my mouth shut. This raises a serious problem of incentive compatibility for a system which depends on action taken, not by government employees hired to enforce the law, but by private individuals acting in their own self interest.
So one requirement for reputational enforcement to work is that the information cost to interested third parties of figuring out who was at fault be low enough to make it worth their while to do so. One way of fulfilling this condition has already been mentioned—employ reputational enforcement in a small society where everyone knows everyone else and has a pretty good idea whose account can or cannot be trusted. Another way is the use of arbitration, ideally prearranged. If you and I have publicly agreed in advance that any dispute in our gemstone dealings will be settled by the New York Diamond Dealers' Association, or by the local rabbi, all a third party has to know is the arbitrator's verdict, which takes a lot less effort than investigating the details of the controversy himself.
A third way is demonstrated by the dealings of Chinese merchants on Taiwan, in an environment almost entirely without contract law hence largely dependent on reputational enforcement, as described in Chapter XXX. Structure your contracts in ways that make it as easy as possible, when there is a disagreement, for third parties to tell who was at fault. If my goods are being stored in your warehouse and I complain that you have let them be damaged, or that some of them are missing, it will be hard for a third party to judge the dispute, so follow the simple rule that ownership goes with physical possession. Make a buyer responsible for inspecting goods before he accepts them and forbidding him, save in the most extreme cases, from demanding compensation if he later discovers that their quality is less than he was led to expect—the rule of caveat emptor. It is easier to demonstrate whether or not I delivered the goods than what happened to them afterwards, or what condition they were delivered in.
The effectiveness of reputational enforcement is different for different people, strong for repeat players, weaker for someone who only expects to participate in such transactions rarely. So design contracts to put the temptation to default on those who can be trusted not to yield to it. If you can be trusted and I cannot, then I pay in advance; you could take my money and fail to deliver what it bought, but you won't.
There are two major problems with ostracism as a way of enforcing rules. The first is that someone has to decide who to ostracize, which requires some decision making procedure whose results practically everyone accepts. Absent that, an attempt at ostracism may divide the community between those who accept the verdict and go along and those who do neither. The second problem is that the collection of people participating in the ostracism has to include a large enough proportion of those the target would like to interact with so that their refusal to interact imposes a serious cost on him.
In a large society this is hard to arrange, since only a small proportion of its members need defect to provide the target with an adequate number of trading partners. In a small community embedded in a larger society, the gypsy (and Amish and Mormon and …) case, it works only if the barriers to a member of the community interacting outside it are high. That condition can be met, as in the case of the Vlach Rom, but at a considerable cost, and it may cease to be met due to changes that make either the community or the society less intolerant of the other. If the barriers are not high, then the threat of ostracism is only a mild deterrent, giving the community only a very limited ability to enforce its rules on its members—which may be why the Vlach Rom in America found it necessary to modify their divorce rules to make them more favorable to wives than before, to avoid the risk that unhappy wives would look for a divorce from American courts rather than the gypsy Kris.
That suggests that ostracism will be most effective for a small community either entirely isolated or set in a very unfriendly larger society. Outside of one of those situations it can still be used to enforce rules, but only rules that nobody has strong reasons to want to violate.
Divine Enforcement: In Foro Interno
The pure form of divine enforcement depends on everyone believing in the divinity and his willingness to enforce or their obligation to obey. That belief may be shaken if violators of the rules are seen to escape unpunished. One way to avoid that is to start with beliefs sufficiently strong to make violations, or at least detected violations, rare. Another is to incorporate in the belief system forms of divine punishment that are unobservable from the outside, such as punishment post-mortem, or at least difficult to observe, such as bad luck.
Even if belief is not sufficiently strong and uniform to make divine enforcement a substitute for more mundane alternatives, it may still make those alternatives more workable. To use oaths as a lie detector it is not necessary that everyone be a believer, only that there is some reasonably reliable way of knowing who is not and refusing to accept their oaths. One way is for there to be religious practices which are costly and not enforced, hence engaged in only by the believers; as I suggested in Chapter XXX, one way of interpreting kosher rules and similar restrictions is that they provide a way of recognizing who believes, hence whose oaths are to be trusted. An extreme version of this is a situation where being a member of the group is itself costly, so costly that nobody would choose to join unless he really believed in the religion. Arguably that describes those who participated in the early years of the Mormon church, when the faithful were fleeing most of the rest of mankind to set up their own refuge in what would become Utah, and equally describes the core of believers who accepted expulsion for Mecca in order to join Mohammed in Medina.
Which suggests that religions such as the LDS or Islam, at least in their early stages, have a significant advantage over more conventions and less controversial rivals; they know who can be trusted, because nobody who can't will join. Arguably an analogous situation exists for political movements. There are undoubtedly advantages to being in power, but also a significant disadvantage—since identifying with the party in power pays, there is no easy way of distinguishing dedicated believers who can be trusted from rice Christians who cannot.
Appendix: Hatfields and McCoys, The Feud that Mostly Wasn't
According to a popular tale, the Hatfields and McCoys, two families of Appalachian hillbillies living on opposite sides of the Tug Fork, the border between Kentucky and West Virginia, started killing each other before the end of the Civil War and continued to do so through a long series of revenge killings, each inspiring the next, for some decades thereafter.
The real history is shorter and less dramatic. Asa McCoy volunteered for the Union Army, was invalided out, returned to the Tug valley, and was killed, a murder often claimed to have been the first in the feud. There is, however, no evidence that his death had anything to do with a family feud, and who killed him was never discovered. Practically everyone in the Tug Valley other than Asa, both Hatfields and McCoys, was pro-Confederate, providing an obvious motive for his killing.
The first definite conflict between Hatfield and McCoy occurred 13 years later, when Old Ranel, the patriarch of the McCoy clan, suspected that Floyd Hatfield had stolen one of his hogs. Instead of getting down his rifle and telling his boys to fetch theirs, however, Ranel took his suspicions to the nearest Justice of the Peace. Floyd Hatfield was tried and acquitted. A year and a half later, two of Ranel's nephews got in a fight with Bill Staton, a relative of both families who had been a witness for Floyd Hatfield's side in the trial, and killed him. They were arrested, tried, and acquitted on grounds of self-defense—on the Hatfield side of the river by a Hatfield judge. No revenge killings followed.
The first exchange of violence fitting the pattern of a proper family feud occurred in 1882. In the course of an election day fight, Ellison Hatfield was badly wounded by three of the McCoys. The three were arrested and on their way to the Pikeville, Kentucky, jail when their party was intercepted by a larger group led by Anse Hatfield and the prisoners seized. The captors, uncertain whether Ellison's wounds were mortal, waited until he died and then killed the three McCoys.
That should, of course, have brought forth return violence from the McCoys. What actually happened was that they took the matter to a Kentucky court, which issued warrants for twenty-one of those involved in the killing but did nothing else. Five years passed with no further violence.
At which point a new governor of Kentucky was elected who happened to be a friend of Perry Cline, a lawyer, friend and distant relative of Ranel McCoy. Cline persuaded the governor to announce rewards for Anse Hatfield and his fellow killers and begin extradition proceedings. Kentucky posses crossed the state border into West Virginia with no authorization from its authorities, killed Anse's uncle, captured nine of those charged with the earlier killing and brought them back to Kentucky, where they were eventually tried and convicted; one was hanged, the other eight given life sentences.
That set off the second and final instance of retaliatory killing; a group of Hatfield supporters attacked Ranel's home, burned it, and in the process killed two of his children. That was the point at which the conflict began to draw national attention, with newspapers describing it in dramatic, and wildly exaggerated, terms. Also the point at which the whole conflict was over, at least for the two families.
Not, however, for the lawyers. The governor of West Virginia demanded the return of the nine citizens of his state who had been kidnapped at the bidding of the governor of Kentucky, carried off, and tried in a Kentucky court. The case eventually reached the Supreme Court which held that although the kidnapping was illegal, there was no recourse; once the nine were in Kentucky, the Kentucky court could try them.
Two things are worth noting about the story. The first is that at most seven killings are clearly due to conflict between the families, of which five were revenge killings, two the killings that inspired the revenge. The other is that, so far as we can tell, the “feud” would have ended with a total of only four killings—for five years had ended—had not the state government of Kentucky chosen to revive it.
Becker, Gary S. & Stigler, George J. , "Law Enforcement, Malfeasance, and Compensation of Enforcers," 3 J. Legal Stud. 1 (1974).
Friedman David, "Private Creation and Enforcement of Law -- A Historical Case." Journal of Legal Studies, (March 1979), pp. 399-415.
"Efficient Institutions for the Private Enforcement of Law." Journal of Legal Studies, June (1984).
Hidden Order: The Economics of Everyday Life: HarperBusiness 1996.
Law's Order: What Economics has to do with Law and Why It Matters, Princeton University Press, 2000.
"Contracts in Cyberspace," 6 Journal of Internet Law 12 (Dec. 2002).
From Imperial China to Cyberspace: Contracting Without the State. Journal of Law, Economics and Policy, July 2005.
Landes, William M. & Posner, Richard A., "The Private Enforcement of Law," 4 J. Legal Stud. 1 (1975)
Llewellyn, Karl N. and Hoebel, E. Adamson, The Cheyenne Way, U. of Oklahoma Press (1983).
MacDowell, Douglas M., The Law in Classical Athens, Cornell University Press 1978.
Sterling, Bruce, The Hacker Crackdown. Law and Disorder on the Electronic Frontier. Bantam (1993). Available online at: http://www.mit.edu/hacker/hacker.html
1. As when Aiskhines sued Ktesiphon for having proposed a decree conferring a gold crown on Demosthenes “in honor of his merit and virtue and 'because he continues saying and doing what is best for the people.' Part of Aiskhines' argument was that it was illegal to include a false statement in a decree, and it was false to say that Demosthenes' speeches and policies were good for Athens. (MacDowell 1978)
3. One might interpret the Fair Courts of medieval Europe as using a mix of this and the previous approach. If a merchant refused to accept a verdict against him others would refuse to deal with him in part for fear of being cheated themselves, in part in order to help enforce the rules of the trade fair.
5. “I would rather play poker with an atheist brought up to believe that gentlemen do not cheat at cards than with a moral philosopher raised by card sharks.” (Not verbatim—I haven't found the source of the quote yet)
11. Landes and Posner (1975). In Friedman (1984) I showed how, by slightly altering the institutional rules they assumed, the problem they pointed out could be solved. Readers interested in a more detailed explanation of their argument and my rebuttal can find it there; the article is webbed at:
12. At this point we are considering both criminal law and tort law as alternative approaches to preventing offenses, not limited to the particular offenses that are currently classified as crimes or torts. So we can include as possible subjects for tort law the sort of deliberate offenses presently classified as criminal.
13. I am not trying here to offer anything approaching a full theory of optimal punishment or optimal enforcement; readers interested in the subject will find it in Chapter 15 of Friedman (2000). The simple rule “set average punishment equal to damage done so as to deter any offense that benefits the offender by less than it costs the victim” should be sufficient for our purposes although, as I demonstrate there, it is not strictly correct. A similar approach applied to how rules are enforced suggests avoiding actions that impose costs greater than the benefits they provide.
16. Territorial behavior by animals, including many species of birds and fishes, provides another example of such a hardwired commitment strategy, in that case a strategy of fighting to the death against a trespasser of the same species. (add reference)
20. Some readers may associate arbitration primarily with arbitrators selected only after the dispute arises. The mechanism I am describing works better if arbitrators are chosen in advance by parties when they sign a contract that might lead to future disputes, since then third parties do not have to have independent information about the honesty or competence of the arbitrator. It is sufficient to know thate one party agreed to abide by the arbitrator's verdict and then reneged on that agreement.