What We Have Been Doing for the Past 19 Chapters


A Rough Sketch of an Elephant


The world has limited space and resources and is occupied by people with differing beliefs and objectives. From those simple facts comes the potential for conflict. I want to hunt a deer across the field where you are trying to grow wheat. You want to go swimming in the stream where I am trying to catch fish.

The simple and obvious solution is the direct use of physical force. You plant a thorn hedge around your wheat field to persuade me to hunt deer somewhere else. I hit you over the head with a tree branch to convince you to swim somewhere else.

This is not a very satisfactory solution to the problem; it is expensive in time, effort, scratches and bruises, and frequently fails of its objective. Indeed, the direct use of force is so poor a solution to the problem of limited resources and diverse ends that it is rarely employed save by small children and great nations.

A more sophisticated version of that solution, substituting threats for violence, works considerably better—well enough to be observed in a considerable number of historical societies. You and I somehow work out a division of our claims to resources. I know that the field is yours—meaning that if I trample it, you will do things I don't like, such as hitting me over the head or swimming in my pool. You know that the pool is mine, meaning that if you swim in it I will do things you don't like, such as hitting you over the head or trampling your wheat field.

With only two people, there is an obvious temptation to solve the problem once and for all by sneaking up on you when you are asleep and hitting you over the head very hard. With a society of more realistic size, that is prevented by my knowledge that if I kill you, your friends and relatives will kill me in revenge. The result is a feud society (not to be confused with a feudal society), in which legal rules are privately enforced by the threat of private retaliation against those who violate them.

Consider Greece at the time of the Homeric epics. When Agammemnon comes home victorious from the Trojan war he is promptly murdered by his wife and her lover. Nobody calls the police and has him arrested, because there are no police to call. Murder, even murder of the King, is a private affair. It is up to his kin to avenge him.

His kin, in this case, means his son. The problem is that one of the murderers against whom he owes vengeance is his mother. A similar conflict about two thousand years later generates much of the plot of Gislisaga. Gisli takes vengeance against Thorgrim, one of the killers of his blood brother (and brother in law) Vestein, but leaves the other, Thorkell, untouched—because Thorkell is Gisli's brother. Orestes chose the other horn of the dilemma, killing his mother and providing the plot for plays by, among others, Aeschylus and Euripides.

Feud systems have functioned tolerably well in a variety of historical societies. To some extent they still do. The privately enforced norms of Shasta county are, in essence, a feud system, although one in which the range of retaliation is considerably more restricted than in Homeric Greece or saga period Iceland. And many other human interactions can be understood as systems of private rules privately enforced by the threat of one party doing things the other party will not like.

One can imagine a modern society built on a more elaborate and formalized version of such a system of decentralized law and law enforcement. Thirty years ago, I devoted about a third of my first book to sketching out how such a society might work, with private firms enforcing the rights of their customers and bargaining among themselves to establish private courts to settle disputes among them. That is not, however, the society we live in. Hence, despite digressions on Iceland and Shasta County, this book is dedicated to making sense of a different solution to the problem of human conflict.

[book link]

Direct private force is the obvious decentralized solution. The obvious centralized solution is dictatorship. Someone appoints himself ruler and gives orders. I am not allowed to trample the wheat; you are not allowed to scare the fish. If anyone does something he is not allowed to do, the dictator or his helpers beat him up.

This solution too exists in a variety of contexts in the modern world. The patriarchal family is one example, the firm another, government a third. While it can work tolerably well for small groups and simple problems, it works badly, often catastrophically badly, for large groups of people doing complicated things.

It works badly for two reasons. The first is that individuals control themselves in their own interest; the dictatorial solution requires some mechanism to force them to do what the dictator wants instead of what they want. The second is that information is dispersed; each of us has a lot of specific knowledge about his own abilities, circumstances and desires. In order for the dictatorial method to work well, to produce attractive outcomes for the people it is applied to, that information has to be somehow passed up a hierarchy of command from us to the dictator, then crunched by the dictator in order to figure out what all of us should do, then passed back down to us as orders. In practice, most of it gets lost in the process—which is one reason why large, centrally controlled organizations usually work very badly.

For large societies, it turns out that the best solutions are decentralized ones, built on the institutions of private property and trade. Each individual owns himself; land and things are mostly owned by particular individuals. If I want your services or the use of your property, I make you an offer you are willing to accept; if getting what I want is worth more to me than giving it costs you, there should be some such offer that I am willing to make.

This sounds simple, but implementing that solution raises a considerable variety of problems. Legal rules, defining what rights each person has and how disagreements or deliberate violations of rights are to be dealt with, are one approach to dealing with those problems. For example … .


Property, Contract, Tort and Crime in Three Paragraphs


In a system of private property we need some way of defining what the boundaries of my property are, not only in physical space but in rights space as well—what uses of my neighbor's property violate my rights in mine and vice versa. We need some way of determining who owns a particular piece of property and establishing property rights over previously unowned property. We need mechanisms for enforcing these rules and for settling disputes over them. And all of this must be generalized from the special case of real property—property in land—to the more general case that includes property in things and intellectual property. Hence property law, including intellectual property.

One of things I can do with my property is to sell it to you. Part of the reason the decentralized solution works is that it permits things, and rights in things, to move via trade to whatever person values them most. In order for that mechanism to work beyond the simplest cases (I give you an apple in exchange for your orange), we need contracts, agreements specifying the mutual obligations of the parties to more complicated exchanges. Contracts can lead to disagreements in interpretation, honest or otherwise, hence require some body of rules to determine when a contract exists and how disputes are to be resolved. Even in the simplest case of exchange, we need enough legal rules to determine whether I was trading my apple for your orange, in which case it is now my orange and your apple, or merely letting you hold my apple while I inspected your orange. Hence contract law.

To prevent me from violating a legal rule, there must be some mechanism to make something I don't like happen to me when I do. That mechanism, whatever it is, will also have to deal with cases where one party believes a rule was violated and another does not. Hence we require tort law and criminal law to specify the relevant rights and the mechanisms by which they are enforced and we require procedural rules and a court system to settle disputes.




This book is about a particular approach to understanding legal rules, developed by economists and by legal scholars influenced by economics. The central question around which that approach has been organized is a simple one: What set of rules and institutions maximize the size of the pie—result, so far as is possible, in everyone achieving his objectives. What legal rules are economically efficient?

There are at least three reasons why that is the question we ask. The first is that while economic efficiency—roughly speaking, maximizing total human happiness—is not the only thing that matters to human beings, it is something that matters quite a lot to most human beings. That is true both for selfish reasons—all else being equal, the larger the pie the larger I can expect my slice to be—and for unselfish reasons. Since the objective is important to almost everyone, it makes sense to think about what rules best achieve it.

A second reason is that there is evidence that considerable parts of the legal system we live in can be explained as tools to generate efficient outcomes. If so, the economic approach may provide a correct description of why law, or at least much law, has its present form, and thus a way of understanding it. It is a lot easier to make sense out of a tool if you know what it is designed to do.

A final reason is that figuring out what rules lead to more or less efficient outcomes is one of the things economists know how to do—and when you have a hammer, everything looks like a nail.


Efficient Legal Rules and How to Get Them


The first approximation to a system of efficient rules is simple—private property and trade. I belong to me, everything belongs to someone, and if you want the use of me or mine, you make me an offer I am willing to accept. Through trade all goods and services move to their highest valued use, producing an efficient outcome.

This description implicitly assumes that the real world can be divided up into unrelated pieces, each belonging to one person—or, to put it differently, it assumes that how I use my property does not affect you and how you use your property does not affect me. Sometimes that is true. Often it is not, giving us the problem of externalities; a standard example in both economics and law is air pollution. If I do not bear all of the costs of my action I may take it even though the net effect, including the cost I am imposing on others, is negative. If so, my action is inefficient—it makes the size of the pie smaller even if it increases the size of my slice.

There are at least four different ways in which such problems may be dealt with. The first and perhaps most obvious is direct regulation—someone decides what people should do and requires them to do it. This works if the regulator is benevolent, omnipotent, and omniscient. In practice, none of those conditions is likely to be met. It is the dictatorial solution to social order, applied within the decentralized solution to resolve some of its problems.

A second alternative is a liability rule requiring me to pay someone, the state or my downwind neighbor, the cost of the damage I do. Since I am now bearing the full cost of my action I will take it only if the net effect is positive, in which case the action is efficient. This works if we have some mechanism that accurately measures external costs and forces the responsible party to pay for them.

There is a problem hidden in the idea of a "responsible party;" the cost I am being made to bear is not entirely due to me. My air pollution would do no damage if nobody happened to be living downwind. What we call an external costs is, in almost every case, due to the actions of multiple people. Choosing one of them and making him responsible may result in a costly solution by him instead of a less costly solution by someone else. This suggests a generalization of the approach: put the incentive where it does the most good. Try to design legal rules so that the party in the best position to reduce a cost is the one who must bear it.

A third solution is to apply the same approach to rights that we are already applying to goods and services: private property and trade. The right to pollute belongs to someone; if that someone is you, the polluter, and the right is worth more to me, your downwind neighbor, then I buy it from you. This works well in contexts where the costs of such private transactions are very low, not so well when they are high, and not at all when they are very high—air pollution in Southern California, for example.

A final solution, and a very common one, is to do nothing—to permit inefficiencies due to externalities because eliminating them costs more than it is worth.

My example was air pollution but the analysis applies across a wide range of legal issues. When I breach my contract with you I impose a cost on you—and might choose to breach even though the net effect, taking account of your costs as well as my benefits, is negative. One solution is administrative—a court decides whether my breach is on net efficient, permits it if it is, forbids it if it is not. A second is a rule of expectation damages: I am permitted to breach but must reimburse you for the resulting costs. A third is private negotiation: I am permitted to breach only if I first buy your permission. A fourth is to do nothing at all—in which case what we have is not a legally enforceable contract but an unenforceable agreement: When you fail to show up for my dinner party, despite having accepted my invitation, I can complain but have no grounds to sue. These alternatives apply to the control of externalities in a variety of contexts—risk allocation in insurance or contract design, product liability law and tort law more generally, and elsewhere through the legal system.

The control of externalities is not the only common thread that runs through our analysis, although it may be the most central. In a system of private property there must be some way of defining what rights ownership implies and who they belong to. If I own a piece of land, can I forbid you from digging a hole on your side of the property line into which my house slides? From digging a coal mine on your property and mining coal under mine? From playing loud music at three A.M. on your front porch, twenty feet from my bedroom window? If we live on a river bank, what happens to our property rights when the river shifts? What if I build a house on what both I and my neighbor think is my land, only to discover, when one of us actually checks the survey, that part of it is on his?

Another set of issues concern the problem of how to enforce whatever legal rules we have: the choice between ex post and ex ante punishment, the problem of giving enforcers the right incentives, costs of apprehending, convicting, and punishing offenders, and the implication of those costs for optimal enforcement.

What comes out of nineteen chapters of this is a general approach to designing an efficient legal system:


Choose an appropriate combination of property and liability rules taking account of the costs of both.


Define and bundle property rights in a way that minimizes costs associated with defining and defending them and transacting over them.


Enforce the whole by some mix of public and private action under a suitable set of rules of proof and liability.


Throughout the project consider the incentives of all parties, including the enforcers, transaction costs, and the problem of using dispersed and imperfect information. Where possible, create institutions that make it in someone's interest to use such information to generate the appropriate rules.


I have described one direction in which the analysis can be applied—figuring out what legal rules would be efficient. Running the project in the opposite direction, what comes out is a way of understanding the legal system we have—on the conjecture that it, or at least large parts of it, can be explained as a system of efficient rules.

An important thing to remember in either version of the project is that utopia is not an option. As Coase argued, in the same essay that gave us the Coasian approach to externalities, in the real world all solutions are imperfect. Our project is not to eliminate costs but to minimize them, to find the least bad system of legal rules. Even that we can do only imperfectly; we do not know enough to reduce legal design to mathematical calculation. What we can do is to understand the advantages and disadvantages of alternative rules as a first step toward choosing among them. For example:

Strict liability in tort law compensates the victim and so gives him no incentive to take precautions to reduce the cost or probability of accidents. Negligence solves that problem at the cost of eliminating the incentive for the tortfeasor to take those precautions that the court cannot observe or cannot judge. Hence strict liability is most attractive where victim precautions are unimportant and unobservable precautions by the tortfeasor important, negligence under the opposite circumstances.

A rule of caveat venditor in product liability eliminates the consumer's incentive to minimize risk and cost and produces litigation and associated costs; a rule of caveat emptor eliminates the litigation, restores the consumer's incentive, but eliminates the producer's incentive to control risks that customers are unaware of. Hence the choice between the rules hinges on who can best control the risk (putting the incentive where it does the most good), on how costly and accurate litigation is, and on how well informed consumers are with regard to at least the average level of risk of a particular brand. Freedom of contract provides a private mechanism for choosing among alternative rules, provided that consumers are sufficiently well informed, not about the risks of particular products, but about the value to them of guarantees.

An ex ante punishment such as a speeding ticket provides incentives based on the information available to the enforcement apparatus but not the (usually much more detailed and less costly) information available to the actors. An ex post punishment such as tort liability for automobile accidents gives the actor an incentive to use his private information but requires large punishments imposed with low probability, hence typically results in high punishment costs. So the choice between the two (or a mixture of the two) depends on the relative quality of the information available to enforcers and to actors concerning what the actor is and should be doing and on the size and resulting cost of ex post punishment. One conclusion is that where even ex post punishments are small enough to raise little problem of risk aversion or judgement proof defendants, they are unambiguously superior to ex ante, unless for some reason the enforcement apparatus has better information about what the actor is and should be doing than the actor does. Another is that a pure ex ante system is almost never efficient, since it can be improved by adding an ex post punishment low enough not to impose significant punishment costs.

A legal system might choose to treat the same act as tort, crime, both or neither. Treating an offense as a tort raises potential problems with judgement proof defendants, especially in contexts where deterrence cannot easily be made into a private good. Where the probability of catching and convicting an offender is low, it raises a further problem of inadequate punishment if damages awarded are not scaled up to compensate for the low probability and a risk of fraudulent prosecution if they are. Treating an offense as a crime solves the problem of judgement proof defendants but raises incentive problems both for victims, who may have an inadequate incentive to report crimes and assist in prosecuting them, and for enforcers. More generally, it has the problems usually associated with centralized forms of production. And where deterrence can be produced as a private good, the privately enforced solution has the advantage of a built in incentive to convict the right person.

Antitrust law has the potential to reduce both rent seeking costs associated with the attempt of firms to get monopoly profits via price fixing agreements or merger and monopoly dead weight costs due to inefficiently high prices when such attempts succeed. But it raises the risk of hindering productive transactions made in the shadow of antitrust law, such as mergers designed to reduce costs, and contractual terms, such as retail price maintanance, that are best explained as means of increasing the value that firms produce, not merely the price of what they sell. In the worst case—the history of the Interstate Commerce Commission provides an instructive example—antitrust may be used to impede competition and promote monopoly.


The Modesty of our Circumstances


As these examples suggest, we rarely if ever know enough to be able to provide a precise solution to the question of what legal rules are efficient—to calculate that the optimal punishment for robbery is three years two months and a day in jail or that conviction should require at least a .932 probability of guilt. Most of the time, the best we can hope for is to understand enough of the logic of the problem to yield an informed guess about what a reasonably good solution would look like. That is the process that I worked through to exhaustion—probably both yours and mine—in chapter 5, analyzing a simple legal problem involving railroads, farmers, and sparks.

What I am offering in this book are not answers but ways of finding them. That is not because I do not have opinions as to what the law ought to be but because, in most cases, I do not believe that I have arguments for those conclusions that are strong enough so that any reasonable person who understands the arguments must agree with me. One reason for that view is the existence of other economists, including some of whose ability I have a high regard, who understand the arguments but, on one point or another, have reached different conclusions.

My own view of antitrust law, for example, is that, with the exception of the refusal of courts to enforce contracts in restraint of trade, we would be better off without it. But what I want you to end up with is not necessarily that conclusion—offered by an economist with no expert knowledge of the field—but the analysis which led me to it and might lead someone else to a somewhat different conclusion. Similarly, my own view is that our legal system would be improved by a substantial shift in the direction of tenth century Iceland—a redefinition of the boundaries of criminal and tort law moving us toward greater reliance on private enforcement, with appropriate adjustments in the relevant law. But what I want you to end up with is not that conclusion but an understanding of how one can usefully think about the choice between private and public enforcement and the appropriate legal institutions for each. Similarly with such other issues as freedom of contract, the marketability of organs and parental rights, and much else.

Economics is neither a set of questions nor a set of answers; it is an approach to understanding behavior. What comes out of that approach depends not only on the economic theory but on the facts of the real world to which the theory is applied. I know a great deal about the real world—as do you, and every other sane human past the age of one—but there is a great deal more I do not know. And while I have thought through substantial parts of the logic of law, doing so in one mind and a reasonable length of time has often required me to ignore complications that might turn out to be important.

So I will end with a brief list of some of the more relevant parts of my ignorance. I do not have:

An adequate theory of transaction costs—although fragments of such a theory can be found in chapter 5 and elsewhere.

An adequate theory of the incentives of public employees such as police and judges. Such analysis as I provide depends mostly on the very weak assumption that they must have some incentive to do what they are doing, else they would not be doing it, plus the assumption that they desire the same sorts of things as other human beings.

An adequate theory of the mechanisms that generate legal rules narrowly defined—statute and common law. I do offer at least a brief sketch of the mechanisms that generate private norms in chapter 17, and elsewhere I have offered an analysis of the mechanisms that would generate legal rules in a system where law and law enforcement were produced as private goods.

[book link]

In addition to these major gaps, the theoretical analysis I offer ignores many of the relevant complications of the real world to which legal rules apply. For example:

I have implicitly assumed that all offenders face the same probability of being apprehended, which is obviously not true. The analysis of efficient punishment could and should be redone with that assumption dropped.

I have made no serious effort to analyse the interaction between public enforcement of law and private enforcement via reputation and private norms—beyond a very brief discussion of the way in which stigma imposes costs on someone convicted of a crime, costs due to private parties taking into account the information generated by his conviction.

While information costs have occasionally played a role in the analysis, I have for the most part used the rationality assumption of economics in its simplest form—individuals take the right action—rather than assuming that individuals make the right choice about gathering costly information and then take action based on the imperfect information they have found it worth obtaining. Thus, for example, I have assumed, explicitly or implicitly, that everyone knows the legal rules that apply to his actions—surely false.

I could go on, but I will leave you to list additional omissions—and, if sufficiently ambitious, attempt to repair some of them.

Table of Contents

My Academic Pages

My Home Page

Email to me