[This is from my hard disk, and may differ in detail from the published version. It is Copyright Cambridge University Press 1995. Reprinted with the permission of Cambridge University Press. Any reproduction, copying downloading or use of any kind of this material is a violation of copyright and such uses must first receive the written consent of Cambridge University Press, 40 West 20th Street, NY,NY 10011.]


Rational Criminals and Profit-Maximizing Police:

Gary Becker's Contribution to the Economic Analysis of Law and Law Enforcement


" ... how many resources and how much punishment should be used to enforce different kinds of legislation? Put equivalently, although more strangely, how many offenses should be permitted and how many offenders should go unpunished." Gary Becker (1968).

The economic analysis of crime starts with one simple assumption: Criminals are rational. A mugger is a mugger for the same reason I am a professor-because that profession makes him better off, by his own standards, than any other alternative available to him. Here, as elsewhere in economics, the assumption of rationality does not imply that muggers (or economics professors) calculate the costs and benefits of available alternatives to seventeen decimal places-merely that they tend to choose the one that best achieves their objectives.

If muggers are rational, we do not have to make mugging impossible in order to prevent it, merely unprofitable. If the benefits of a profession decrease or its costs increase, fewer people will enter it-whether the profession is plumbing or burglary. If little old ladies start carrying pistols in their purses, so that one mugging in ten puts the mugger in the hospital or the morgue, the number of muggers will decrease drastically-not because they have all been shot but because most will have switched to safer ways of making a living. If mugging becomes sufficiently unprofitable, nobody will do it.

The assumption of rationality applies not only to criminals but to everyone included in the analysis. Judges, policemen, legislators and potential victims are all, like criminals, rational individuals pursuing their own ends as best they can. Economic theory gives us no reason to assume that criminals are less rational than judges, or judges less self-interested than criminals.

When the analysis moves from descriptive to prescriptive, it seems natural to apply a similar approach to ourselves. In designing institutions to control crime, our concern is not sin but cost. Eliminating all murders, even all muggings, would no doubt be a fine thing-but whether we ought to try to do it depends on how much it costs. If reducing the annual murder rate from ten to zero requires us to turn half the population into police, judges, and prison guards, it is probably not worth doing.


Part I: The Theory of Optimal Enforcement


What I have said suggests a simple question: How can we structure laws and law enforcement so as to minimize the net costs of crime and crime prevention? The modern attempt to answer that question begins with Gary Becker's classic article "Crime and Punishment: An Economic Approach." In trying to explain the theory as it now exists, I will not distinguish Becker's own work from what has been added later by others; interested readers may wish to look at the articles cited at the end of this chapter.

It is useful to begin our analysis of optimal law enforcement with a simple, although entirely unrealistic, case. Suppose we can catch all criminals and collect fines from them, and that it costs nothing to do so. In this fortunate world, how should we set the fines?

Our first guess might be that the fine should be high enough to deter all crimes. But there may be some crimes we do not want to deter. Consider a hunter, lost and starving in the woods, who comes across a locked cabin containing food and a telephone. The benefit to him of breaking in and calling for rescue is much larger than the cost to the cabin's owner; we will, on net, be better off if that particular offense is not deterred. For a less exotic example, consider the driver who occasionally exceeds the speed limit when he is in a hurry. We could deter all or almost all speeding if we routinely confiscated the cars of convicted speeders-but many of us would regard that as more deterrence than we want.

If a crime produces a net benefit, if the gain to the speeder or the lost hunter is more than the loss to the rest of us, we are better off not deterring it. In many cases our legal system permits such "efficient crimes" by not classifying them as crimes. Every time I breath out I exhale carbon dioxide-regarded, in some contexts, as a pollutant. Most of us are confident that this particular offense is an efficient one-we are better off if we put up with a certain amount of extra carbon dioxide than if we all stop breathing. So exhaling, unlike some other forms of pollution, is not illegal. Similarly, the lost hunter of my example would probably be excused from criminal liability under the doctrine of necessity.

But there are other cases where a sharp line between efficient and inefficient crimes is hard to draw. By driving my car at eighty miles an hour I impose a cost, a risk of accidents, on other drivers. Often that is inefficient-if I had to take into account the risk I impose on them as well as the risk I impose on myself, I would not drive so fast. Occasionally, however, driving fast is efficient-the benefit of getting where I am going half an hour earlier more than balances the associated cost. How, in such situations, can we deter inefficient speeding while permitting efficient speeding?

The answer is simple: Set the expected punishment equal to the damage done. If exceeding the speed limit imposes, on other drivers, a cost in increased risk of $100, then a driver who speeds will have to pay a $100 fine. If, to make our assumptions a little more realistic, only one speeder in ten is caught, we set the fine at $1000, making his average cost of speeding, his "expected punishment," again equal to $100. If speeding is worth more than $100 to him, he will speed; if it is not, he will obey the limit.

Under this rule, criminal punishment functions as a probabilistic price system. Each crime has a price-a certain probability of a certain fine-set so that the average fine equals the damage done. Someone willing to pay that price will commit the crime-and, from the standpoint of economic efficiency, should. In just the same way, on an ordinary market, each good has a price. That a buyer is willing to pay that price shows that the good is worth more to him than to the seller, so the transaction produces a net gain.

While this may be a reasonable way of looking at speeding, it seems less appropriate for more serious offenses. If we are really setting our punishments in the way I have described, it follows that the offenses which still occur are ones that produce net benefits. Do we really have a legal system where the reason we do not raise the punishment for murder is the fear that we would then have too few murders?

The answer is no. So far I have been describing a world where crime control costs nothing and we therefore buy as much of it as we want. In such a world we would set the punishment for murder so high that only efficient murders occurred. While some of us may be able to think of one or two people who would improve the world by their absence, it is unlikely we would agree on who they were-so the consensus would probably be for an efficient murder rate of zero.

In the real world, that is not a practical option. In order to catch and convict criminals we must pay police, hold trials, train lawyers-pay a variety of costs which I will refer to, for convenience, as apprehension costs. Once the criminals are convicted, we must punish them-and that too is costly.

What is the net cost of punishment? In the simple case of a fine, it is zero-the criminal pays $100 and the court system collects $100. In the case of imprisonment or execution, the cost is higher. The cost of imprisonment includes not only the wages of guards but also the time of jail inmates-unlike the case of a fine, nobody gets what the prisoner loses. An execution costs more than the price of a bullet or a hangman's rope-one life more.

In controlling crime, we can get a given amount of deterrence in many different ways. We might catch half the criminals and fine them each $500, we might catch one criminal in twenty and put him in jail for a year, or we might catch one in five hundred and hang him. Suppose that all of these alternatives are equivalent from the standpoint of the criminal and thus have the same deterrent effect, just as a $100 fine was equivalent, in the earlier example, to one chance in ten of a $1000 fine. Generalizing our earlier term, we may say that all of these alternative combinations of probability and punishment represent the same expected punishment.

In choosing among them, we are trading off one kind of cost against another. It takes fewer policemen to catch five criminals out of a hundred than fifty and fewer judges to try them, so lowering the number we catch while raising their punishment saves on apprehension costs. But, since there is a limit to how large a fine a convicted criminal can pay, raising the punishment typically means shifting from fines to imprisonment or execution-which increases punishment cost. An efficient law enforcement system would produce any given level of deterrence with whatever combination of probability and punishment minimized the sum of the two sorts of cost.

This raises an interesting puzzle-why do we imprison anyone? One answer is that we imprison people because they are judgement proof--they do not have enough money to pay a fine high enough to represent an adequate punishment. But while that explains why we cannot fine some people, it does not explain why we imprison them, since imprisonment appears to be not only less efficient than a fine but also less efficient than execution. Bullets and rope are cheap, so the total cost of an execution consists almost entirely of the cost paid by the criminal-his life. Jails and jail guards are expensive, so the total cost of imprisonment is substantially more than the cost to the criminal.

Suppose we have determined that the cost to a criminal of life imprisonment is exactly half the cost of execution-meaning that a criminal is indifferent between a certainty of a life sentence and a fifty percent chance of execution. We then make a small modification to present law. Whenever a criminal receives a life sentence, we flip a coin. Heads he goes free; tails we hang him.

On the face of it, this seems an unambiguous improvement. The criminal is no worse off, on average, since he consider a fifty percent chance of execution as equivalent to a certainty of life in prison. Potential victims are no worse off-since the two punishments are equivalent from the standpoint of criminals, they will have the same deterrent effect. The only significant effect of the change is that we can now replace a large number of prison guards with a small number of hangmen-for a substantial net savings.

Nothing in the logic of the argument limits it to criminals who have been sentenced to life terms. The same argument would apply to anyone sentenced to serve time, however short-the only difference is in the appropriate odds. If my sentence is one year, the court uses a roulette wheel instead of a coin-on a double zero they hang me, on any other number I go free.

What is wrong with the grisly conclusion of this simple and straightforward argument? One possibility is that there is nothing wrong with it-I have described an efficient system of law enforcement, and the reason our system does not work that way is that it is not efficient, either because efficiency is not our objective or because we have done a bad job of getting it.

There are at least two other possible explanations of our preference for expensive imprisonment over cheap execution. One is that execution is not really cheap-indeed is more costly than imprisonment, if the calculation is done correctly. One respect in which it is costly is that execution is irreversible; if the real murderer confesses a week after we hang someone else, there is no way of unhanging him. Irreversible error also involves costs, although ones we have so far ignored. Another is that execution offends the sensibilities of many citizens in ways in which imprisonment does not. If many of us are made unhappy by the knowledge that the legal system we support and pay for kills people in cold blood, that too is a cost. In addition, in the U.S. legal system as it now exists, execution is very expensive because it involves a lengthy process of appeals-although one should perhaps regard that less as a reason for our reluctance to execute criminals than a result of it.

A second explanation is that we are reluctant to make too much use of execution precisely because it is efficient. So far I have applied the assumption of rationality only to criminals. I have treated courts and police as merely servants of the public, with no will of their own. That might be a dangerous mistake.

Suppose we suspect that the people running the enforcement system may sometimes, under some circumstances, have an incentive to carry their job substantially too far. An ambitious mayor, or governor, or president might decide that there is political profit to be made from a reputation as a crime fighter-enough so that it is worth ignoring the costs to those he convicts, rightly or wrongly, and punishes. If, for most crimes, punishment means imprisonment, he will be constrained by the reluctance of taxpayers to pay for an unlimited number of prisons. If he is free to execute criminals even for relatively minor offenses, that constraint vanishes. So one possible explanation of our reluctance to make more use of execution is that we wish to limit our officials to punishments that impose costs on the enforcement system as well as on the convicted criminal, as a way of limiting the risk of overpunishment in a system in which we have no guarantee that political decisionmakers will take all costs into account in making their decisions.

This may seem, to some readers, like a purely theoretical argument, and an implausible one at that. It becomes more plausible if we consider a real world case in which we have provided the police with the power to impose efficient punishments-and may well be worse off as a result. The example I have in mind is civil forfeiture for drug crimes.

Under present law, police can, in a wide range of cases, seize property that they claim was used in connection with illegal drugs-without having to convict the owner of anything. It is up to the owner to go to court and prove his innocence in order to get his property back. Critics of the system argue that it results in widespread abuse. There is no reason, after all, to expect police, or the local governments that employ them, to be less rational than criminals, or less interested in serving their own objectives. It may sometimes be prudent to ban punishments that are so efficient that they produce a profit for the punisher at the expense of the punishee-at least in situations where it is the punisher who gets to determine guilt and set the punishment.

Having offered some possible solutions to the puzzle of why we prefer inefficient imprisonment to (apparently) efficient execution, let us now return to the main topic-what an efficient system of enforcement and punishment would look like.

In deciding how much deterrence we want, we must balance the costs and benefits of increasing expected punishment. It is inefficient for me to steal a television set that is worth five hundred dollars to you and only four hundred dollars to me. But it is still more inefficient to prevent me from stealing the set if the cost of doing so is two hundred dollars additional expenditure on police, courts, and prisons. The rule "prevent all inefficient offenses and only inefficient offenses" is correct only if doing so is costless. The more generally correct rule is to prevent an offense if and only if the net cost from the offense occurring is greater than the cost of preventing it.

To provide more deterrence for a particular crime, we must either catch a larger fraction of offenders or impose a more severe punishment-which means increasing either apprehension or punishment cost per offense. Increasing cost per offense, however, does not necessarily increase total cost-it depends what happens to the number of offenses. Suppose that, for some particular offense, an increase in the expected punishment from $100 to $110 reduces the number of offenses from 100 to 10. The cost per offense of catching and punishing offenders has increased, but the total cost has almost certainly decreased, since there are many fewer offenders to be apprehended and punished.

This is a situation where, in the language of economics, the demand for offenses is highly elastic; when we increase the price of committing an offense the number committed falls sharply. The result is that the marginal cost of deterrence, the extra cost of deterring one more offense, is negative-increasing deterrence lowers cost. If, on the other hand, the demand were highly inelastic, if, for example, the increased punishment only reduced the number of offenses from 100 to 99, total cost would almost certainly increase. In such a situation, increasing deterrence raises cost, so the marginal cost of deterrence is positive.

The difference between crimes for which marginal cost is positive and crimes for which it is negative is important when it comes to setting the optimal punishment. If deterrence is costly, then it is only worth deterring offenses that impose substantial net costs. If a particular offense costs the victim $100 and benefits the criminal by $99, deterring it produces a net benefit of only $1. If the marginal cost of deterrence is positive, crimes that are only slightly inefficient are not worth deterring.

Which crimes we deter is determined by how high an expected punishment we set. If, in the previous example, we set the expected punishment at $99 for a crime that does $100 worth of damage, then any criminal whose benefit is less than $99 will be deterred while any whose benefit is more than $99 will find it worth committing the crime. We will be deterring all offenses whose net harm is more than $1 and permitting, although still punishing, offenses that produce a net harm of less than one dollar, and that it is therefore worth less than a dollar to deter.

Generalizing the example gives us a simple rule. The expected punishment for a crime should equal the damage done minus the marginal cost of deterrence. A potential offender whose gain is less than that will be deterred-and should be, since the net damage of his crime (victim's loss minus criminal's gain) is more than the cost of deterring it.

This probably describes the situation for serious crimes such as murder. Catching a large enough fraction of the criminals and punishing them severely enough to make the expected punishment equal to the damage done, if possible at all, would be very expensive and provide only a moderate increase in deterrence over a lower but still substantial expected punishment. The marginal cost of punishment is positive, so we impose a punishment well below the damage done.

What about crimes for which the demand for offenses is elastic enough to make the marginal cost of deterrence negative? The same analysis applies; only the sign has changed. If raising expected punishment by enough to deter one more crime decreases apprehension and enforcement cost by $10, the efficient rule is to set punishment equal to damage plus $10. We will be deterring some slightly efficient offenses, but it is worth deterring them in order to save the cost of catching and punishing the offenders.

Our rule for punishment combines elements of two different intuitions: punishment equal to damage done and enough punishment to deter. If catching and punishing criminals is inexpensive, the optimum is about equal to damage done-enforcement and punishment costs are unimportant, so we simply design our system to deter all inefficient and only inefficient offenses. If, on the other hand, catching and punishing criminals is costly and the supply of offenses is highly elastic at some particular level of punishment, so that below that level there are many offenses and above it very few, then we set the punishment at the point where any further increase would have very little deterrent effect to balance its cost-just enough punishment to deter most offenses.

One implication is that it may sometimes be worth treating different sorts of criminals differently. An example is the insanity defense. If insane people cannot be deterred by the threat of punishment then their demand for offenses is perfectly inelastic; the marginal cost of deterring them is infinite. If they can be deterred, but a large punishment produces only a very small amount of deterrence, the marginal cost of deterrence is so high that it is probably not worth doing. We may still wish to confine insane criminals in order to prevent them from committing more crimes, or in the hope of curing them, but not as a punishment intended to deter.

A less extreme example is the imposition of a higher punishment for a premeditated murder than for a murder committed in a blind rage. Presumably the offender who plans out his murder in advance is more likely to consider consequences-including punishment-and is therefore more easily deterred. The marginal cost of deterring such murderers is lower, so the optimal punishment is higher.

Another example of the same logic is raised by a question that my students sometimes ask: should rich and poor criminals pay the same fine? A hundred dollars may be a crushing fine to a pauper but a trivial punishment to a wealthy man. In a legal system where all criminals fact the same expected punishment, are we not, in effect, overpunishing the poor and underpunishing the rich?

There are two answers to this question. The first is that, in defining economic efficiency, we measure costs and benefits by willingness to pay-a criterion which weights the desires of rich people more highly than the desires of poor people. A legal system in which rich people "buy" more illegal driving than poor people may or may not be just, but it is efficient for the same reason that a system in which rich people buy more television sets than poor people is efficient. A full discussion of the reasons why I (and many other economists) do not regard that as an adequate reason to reject efficiency as a useful criterion would carry us far beyond the bounds of this chapter (Friedman 1990, Chapter 15).

A second answer, however, is that it is not clear that an efficient system will always impose the same fines on rich people and poor people. As we have seen, in a world where law enforcement is costly the correct rule for determining punishments is more complicated than simply setting expected punishment equal to the damage done. We must also take account of how much additional deterrence we can get for an additional dollar of enforcement and punishment cost. For some crimes, the deterrent effect of punishment depends on the income of the criminal.

Stealing a hundred dollars provides the same dollar benefit for rich and poor, so the same dollar punishment will deter it. The same is not true for speeding. On average, a rich person is willing to spend more dollars to save half an hour of his time than a poor person, so deterring him from speeding requires a higher fine. The same is true for a variety of other crimes. A richer person, on average, is willing to pay more dollars for the pleasure of slugging someone he is mad at, or murdering someone he is very mad at. We may generalize these examples by saying that for crimes where the payoff is in money, deterring rich people requires about the same punishment as deterring poor people, while for crimes where the payoff is in time or pleasure, it typically requires a larger fine to deter a richer victim.

Since one element in setting the punishment for a crime is how much punishment is required to deter, it will sometimes, for some crimes, be efficient to impose more severe punishments on richer criminals. For other crimes, however, an efficient system might actually set lower punishments for richer criminals-because they are so hard to deter that deterring them, like deterring lunatics, is not worth doing.

There is a second difference between rich and poor that leads to a less ambiguous conclusion. One constraint on punishment is its cost. Fines are a more efficient punishment than imprisonment, and richer people can pay larger fines. If deterring poor criminals is expensive and deterring rich criminals is cheap, an efficient system will prefer to deter rich criminals-and will do so by imposing higher punishments on them.

Before ending this part of the chapter, I would like to raise one further point that may have struck some readers. In my analysis of costs and benefits I have made no distinction between criminals, victims, and taxpayers. The cost of punishment to criminals and the benefit of crime to criminals receive the same weight as any other costs and benefits. To many non-economists, even to a few economists, this seems wrong. Why should we care about criminals-have they not, by their acts, put themselves beyond the pale, forfeited their claim to have us take account of their welfare in our policies?

I think not-for two reasons. The first is that our analysis is intended to cover a broad range of illegal activity-speeding and illegal parking as well as murder and rape. With regard to at least the milder sorts of crimes we are all potential criminals as well as potential victims; it seems an odd form of economic accounting to include our costs and benefits in the latter role but not in the former.

A second reason is that economics does not start with assumptions about right and wrong-merely with the idea that individuals have objectives and choose means to achieve them. Economic efficiency is an attempt to define, as objectively as practical, the degree to which each of us succeeds in achieving his own objectives-without making any ethical judgement about the objectives themselves. Part of what is exciting about the economic analysis of law is its ability to reason from the goal of economic efficiency to what look like ethical rules-"thou shalt not steal," for example. If we start out by assuming that certain acts are wicked and that benefits to people who commit wicked acts should not count in our social calculus, we are assuming our conclusions where we ought to be deriving them.

We now have a sketch of an economically efficient legal system system for controlling crime. Of course, no court and no legislature has the information necessary to follow fully the program I have described-we do not know enough about how much different levels of enforcement and punishment cost, about how high a level of crime will be associated with any given level of deterrence, or about how much damage is done by each offense. The purpose of such analysis is not to provide a precise formula to be entered into a computer and used to generate a police budget and a schedule of punishments. It is rather to figure out what we would be doing if we only knew enough to do it, in order that we can use that knowledge in trying to come as close as our actual knowledge permits to constructing an efficient system of laws and law enforcement.

So far I have not distinguished between efficient deterrence as a prescriptive rule and efficient deterrence as a descriptive rule. I have not said whether I am describing what our legal system does or what it ought to do. Here as elsewhere in the economic analysis of law, economic efficiency plays a dual role. There are arguments for why creating an economically efficient system of law is desirable (Friedman 1990, chapter 15). There are also arguments for why real world legal institutions might be expected to be economically efficient (Posner 1992, pp. 12-16). If one accepts the first sort of argument, the analysis I have just given is a description of what we should have; if one accepts the second, it is an explanation of what we do have.

I have also not said much about evidence in favor of the underlying assumptions of the economic approach to crime, in particular the much debated question of whether and to what degree punishment deters. Interested readers may want to look at the summary of the evidence in Nagin 1978. The author reports on more than twenty empirical studies done between 1968 and 1975; only one failed to find "some significant inverse associations between crime and sanctions." He argues, however, that the evidence for deterrence is not conclusive, since there are other possible explanations for such an association.


Part II: Should We Abolish the Criminal Law?

In the first part of this chapter, I applied the assumption of rationality to criminals but not, save for one digression, to other participants in the criminal justice system. I will now try to correct that omission. What does the assumption that policemen, like criminals, are rational imply for how we ought to enforce the law?

The modern discussion of this issue owes a good deal to an article by Gary Becker and George Stigler, entitled "Law Enforcement, Malfeasance and Compensation of Enforcers." The relevant part of their argument may be summarized quite simply:

I am a policeman, you are a criminal, and I have the evidence that will convict you. The cost to you of being convicted is $50,000-either a fine or an equivalent jail sentence. The benefit to me of convicting you is a commendation from my superior officer and a small increase in my prospects for promotion-worth, say, $10,000 in future income.

Seen from the standpoint of Dragnet, the rest is obvious. I deliver both you and the evidence to the D.A., and the story is over. Seen from the standpoint of an economically rational police officer, it is equally obvious. I sell you the evidence for something between ten and fifty thousand dollars, and we both go home.

Of course, this is not the whole story-if it were, criminals would almost never be convicted. Real world legal systems spend considerable time and effort trying to prevent such transactions and punishing those who engage in them. But the need to do so is a substantial cost-it means that police officers must spend part of their time watching each other instead of watching criminals. And sometimes, when one police officer eludes the vigilance of his fellows or when a whole department succeeds in conspiring together in their own interest and against the interest of the taxpayers who employ them, the economics text is a better description of the real world than the television program.

Becker and Stigler suggested a simple and radical solution to this problem-privatize the catching of criminals. Instead of paying the policeman a salary, pay him the fines collected from the criminals he brings in. If the convicted criminal will owe a $50,000 fine to the policeman, the lowest bribe the policeman will accept to let him off is $50,000. If the criminal offers that much, in order to avoid the expense of defending a hopeless case in court, there is no reason we should object-the criminal has paid his fine, the policeman has received his salary, and the taxpayers have been saved the cost of a trial.

Such a system of private enforcement raises a new issue: how to allocate crimes. Since policemen are now private bounty hunters, how do we decide which one is entitled to catch a particular criminal and collect his fine? One solution is to make the crime the property of the victim. He sells the right to solve it to a (private) policeman. This process allocates crimes to enforcers-efficiently, since the enforcer best able to catch the criminal will be willing to pay the highest price. Victims receive some reimbursement for their loss, and we need no longer worry about keeping policemen from accepting bribes.

By following out the line of argument begun by Becker and Stigler, we have reinvented civil law-the law of torts as distinguished from the law of crimes. It is the victim of a tort, not the state, who has a claim against the tortfeasor. While modern American law does not permit him to sell the entire claim to the lawyer who will go to court and collect it, he can sell part of the claim by hiring the lawyer on a contingency basis. Some earlier forms of civil law permitted the outright sale of civil claims, and some modern writers have argued that we should do the same. (Friedman 1984, Shukaitis 1987).

What we call bribery in the criminal context is called an out of court settlement in civil law-and is how most civil claims are collected. Since the payment is made to the person who would have collected the fine (called a "damage payment" in the civil system), an out of court settlement achieves the same result as a trial and at lower cost.

This suggests an interesting question: Should we abolish the criminal law? Would we be better off if we turned all crimes into torts, replacing enforcement by the state with enforcement by private police selected by the victims? To put the question differently, is there any logic to our present system, where if someone assaults me I call the police but if he reneges on a contract I call my lawyer?

The civil system has some obvious advantages. It replaces a centralized, government run system for catching criminals with a decentralized market. In many other contexts, from schools to postal delivery, there seems to be evidence that markets produce a better product than governments at a lower cost. In addition, a civil system provides reimbursement to the victim. This not only appeals to our sense of justice, it also provides an incentive for victims to report crimes, even when doing so involves a certain amount of risk or inconvenience-something notably lacking in a criminal system. And it eliminates the conflict of interest between the enforcer and his employers that appears as bribery, and precautions to prevent bribery, under current institutions.

There are, however, some problems with a pure civil system. One is that under a civil system, at least our present civil system, the victim is limited to collecting a damage payment that will "make him whole." The fine, in other words, is supposed to be set equal to damage done. That may be close to the efficient level if almost all tortfeasors are identified and convicted. But suppose only one burglar in ten is caught. Nine times out of ten he steals a hundred dollars from me and keeps it, one time out of ten he is caught and gives it back-making burglary a profitable profession.

If we replace criminal law by civil law, we should probably modify the damage rule, scaling up the amount collected to compensate for the uncertainty of collection. We might, for instance, require the burglar who is caught one time in ten to pay a thousand dollars to his victim. On average the burglar pays as much in fines as he steals. He receives no net compensation for his time and effort, so finds some more attractive profession. In a more sophisticated version of such a legal system, the analysis of Part I of this chapter could be used to calculate the optimum punishment-now a damage payment rather than a fine.

This, however, raises a further problem. The larger the damage payment, the less likely it is that the burglar can pay it. A criminal system can punish offenders unable to pay fines by putting them in jail. But a private law enforcer, whether lawyer or bounty hunter, receives no salary from the taxpayer. Hunting down a criminal who is judgment proof, unable to pay the damages the court awards, is a waste of time and effort, so if most criminals are judgment proof there will be little incentive for a civil system to catch them. For this reason, Becker has suggested that the essential distinction between civil and criminal offenses is that criminal acts are, or at least should be, those for which the offender cannot pay the appropriate fine (Becker ).

One solution to this problem would be to replace damage payments by bounties in the case of judgement proof criminals. The state would pay the convicted criminal's fine to the enforcer who got him convicted, then punish the criminal by putting him in jail. Criminals able to pay their own fines would do so. Such a system is more costly than a pure civil system, but not more costly than our current system, since fines paid to enforcers would replace salaries now paid to police.

An alternative approach would be to make criminals less judgement proof, by eliminating some current restrictions on what assets can be seized to satisfy a civil judgement. In an extreme version of such a system, convicted criminals might be required to work off their fines, perhaps with private prisons competing with each other on how quickly they would pay off a fine in exchange for an inmate's labor, just as private employers competed, two or three hundred years ago, on how short a period of service they would accept in exchange for paying the travel costs of immigrants who came to this country as indentured servants (Friedman 1979, .

A second problem with scaling up a damage payment to compensate for the chance that the criminal will not pay it is that it gives "victims" an incentive to forge crimes. Under present law, there is very little point in my arranging for your car to run into mine under circumstances in which you will appear to be to blame, since all I will get if I win my suit is enough money to fix the car. But if the law permits the victim to recover ten times the damage, on the theory that ninety percent of those responsible for such offenses are never apprehended, arranging a fake accident may be profitable. This is the same problem of overenforcement that I discussed in the previous section in the context of charging people with drug crimes in order to confiscate their property, this time in the context of private rather than public law enforcement.

It would take more space than I have here to discuss fully arguments for, and problems with, a pure civil system, in which crime is controlled by damage suits instead of criminal indictments. Readers interested in the subject may wish to read some of the more technical articles listed at the end of the chapter; they may also be interested in Friedman (1979), which describes the workings of a real world system in which killing was a civil offense, prosecuted by relatives of the victim. In fairness, I should say that while I believe a good case can be made for a pure civil system, with appropriate modifications to take account of special problems posed by the offenses that we now call crimes, that is almost certainly a minority opinion among those who have written on the subject.


Part III: What I have Left Out

In this chapter I have tried to sketch out two areas of the economic analysis of law based in part on early work by Gary Becker: the theory of optimal punishment and the analysis of the choice between civil and criminal law. In doing so, I have covered only a small part of the relevant territory, ignoring applications of the analysis outside the fields of tort and crime. One example would be patent law, which, Becker argued, could be viewed as a sort of negative punishment-a system for rewarding those who provide benefits to the rest of us. Another would be the application of the arguments that Becker and Stigler made with regard to the compensation of enforcers to the compensation of other agents facing similar conflicts between their interest and the interest of their employers.

I have also said very little about ways in which the theory ignores real world complications, some of them important for the analysis. I have not, for example, discussed the issue of marginal deterrence, the problem of deterring someone out of one crime into another. If we make the penalty for armed robbery as high as the penalty for murder, robbers have no incentive not to kill their victims (Friedman and Sjostrom 1993), since there is no additional punishment for doing so. I have not discussed the implications for the theory of the fact that some criminals are abler than others, and thus harder to catch, nor of the fact that courts sometimes make mistakes. Nor have I considered the issue of incapacitation-imprisoning a criminal not only to punish him for the crime but also in order to prevent him, at least for the term of his sentence, from repeating it. And I have said very little about empirical questions associated with statistical analysis of crime-as exemplified in the work of Isaac Ehrlich (1972), who found that the death penalty had a substantial effect in deterring murder, and his critics (Blumstein et. al., 1978).

I hope, however, that I have given the reader some feel for how economic analysis can be applied to the problem of controlling behavior, whether labelled criminal or civil, by which one person imposes costs on others, and of how sharply our understanding of such issues is changed by the simple assumption that human beings, whether criminals or police, are rational.

David Friedman




Becker, Gary S., "Crime and Punishment: An Economic Approach," Journal of Political Economy 76 (1968) pp. 169-217.

Becker, Gary S. and George J. Stigler, Law Enforcement, Malfeasance, and Compensation of Enforcers, 3 Journal of Legal Studies 1 (1974).

Blumstein, Alfred, Jacqueline Cohen, and Daniel Nagin, editors, Deterrence and Incapacitations: Estimating the Effects of Criminal Sanctions on Crime Rates, National Academy of Sciences, Washington, D.C. 1978.

Ehrlich, Isaac, The Deterrent Effect of Criminal law Enforcement , 1 J. Leg. Stud. 259 (1972).

Friedman, David, "Private Creation and Enforcement of Law -- A Historical Case." Journal of Legal Studies, March 1979.

____"Reflections on Optimal Punishment or Should the Rich Pay Higher Fines?," Research in Law and Economics, 1981.

____ "Efficient Institutions for the Private Enforcement of Law." Journal of Legal Studies, June 1984.

____Price Theory: An Intermediate Text, South-western 1990.

____ and William Sjostrom , "Hanged for a Sheep-The Economics of Marginal Deterrence," forthcoming in Journal of Legal Studies.

Landes, William M. and Richard A. Posner, "The Private Enforcement of Law," 4 Journal of Legal Studies 1 (1975).

Nagin, Daniel, "General Deterrence: A Review of the Empirical Evidence," in Blumstein et. al. (1978).

Polinsky & Shavell, The optimal tradeoff between the probability and magnitude of fines, 69 AER 880 (1979).

Posner, Richard A., Economic Analysis of Law, Little, Brown and Company, Boston 1992.

Shukaitis, Marc J., "A Market in Personal Injury Tort Claims," Journal of Legal Studies XVI (2) June 1987, pp. 329-349.

Back to the
list of articles.

Back to my home page.