1For purposes of simplicity, I ignore problems of risk aversion here and throughout; all actors are assumed to be risk neutral.

2Judge Posner's famous example of such an "efficient crime" is the lost man who breaks into a locked cabin in the woods in order to get food and summon help. A more mundane example is the driver who exceeds the speed limit because he is in enough of a hurry so that it is worth doing so, even after allowing for the costs that his speed imposes on both himself and others .

3This can be generalized from the binary case of take/don't take precaution to situations where level of precaution is a continuous variable. The rule then becomes "take that level of precaution for which marginal cost of additional precaution equals marginal expected benefit."

4Under this approach (and, I believe, under existing law), the starving hunter who breaks into the cabin has not committed a crime. Similarly, the driver who is speeding to get his wife to the hospital before she finishes giving birth is unlikely to actually receive a ticket. In the one case the offense is excused under the doctrine of necessity, in the other permitted by an exercise of discretion on the part of a police officer, but the effect is in both cases the same.

5One argument is that ordinary damages fail to compensate the victim for the cost to him of litigating the case. Another is that some victims will not sue, or will lose their cases due to court error, and that the tortfeasor therefore pays damages with a probability substantially less than one. A third is that ordinary damages are, or at least traditionally have been, limited to those damages easily measured, and therefore omit important non-pecuniary costs. A number of these points are discussed in other papers in this volume and in Ellis, Fairness and Efficiency in the Law of Punitive Damages, 56 So. Cal. L. REV. 1 (1982).

6>One possible answer to this question, suggested by Landes and Posner, is that deliberate torts are more likely to be concealed, hence require a larger punishment when detected. One would think, however, that just the opposite would be the case for reckless torts. Another answer suggested by the same authors is that, if occurrences of a particular sort of tort are very unlikely to be efficient, punishment above damage done has no disadvantages, and protects against accidental underdeterrence due to court errors (underestimating damage done, for instance). Both of these arguments are made in William M. Landes and Richard A. Posner, The Economic Structure of Tort Law, (Cambridge: Harvard University Press, 1987) p. 160-162.

7 Ellis (1982)

8As Dean Ellis pointed out in his paper, the idea of illegitimate satisfactions has been used before (in other contexts), by Gordon Tullock (The Logic of the Law) and George Stigler (The Optimum Enforcement of Laws, 78 J. Pol. Econ. 526 (1970) at 529).

9For a discussion of that justification, see David Friedman, Price Theory, an Intermediate Text, 2nd edn. Chapter 15. Other arguments for the desirability of economic efficiency, sometimes described as wealth maximization, are given in Richard Posner, Wealth Maximization Revisited, 2 Notre Dame Journal of Law, Ethics, & Public Policy 85 (1985), Landes and Posner (1987) pp. 16-19, and Lloyd Cohen, A Justification of Social Wealth Maximization as a Rights-based Ethical Theory, 10 Harvard Journal of Law & Public Policy 411 (1987).

10An alternative explanation is that there is some deep connection between the prescriptions of utilitarianism and those of some other ethical principle upon which our judgements are in large part based.

11As other papers in this volume make clear, a variety of terms were used, but most of them fit into one or the other of those categories.

12The latter situation exists if the two effects happen to exactly cancel; while this is possible, there is no reason to expect it.

13It is occasionally said that the efficient punishment is the damage done including the cost of litigation. This is correct if "including" means "modified to take account of." It is incorrect if, as a reader would normally assume, "including the cost of litigation" means "adding in the cost of litigation," since in that case the efficient punishment would always be at least as great as the damage done.

14The same analysis can also be found, at greater length, in David Friedman, Reflections on Optimal Punishment or Should the Rich Pay Higher Fines?, Research in Law and Economics, 1981. An earlier discussion of the effect of elasticity on optimal punishment is in Gary Becker, , 76 JPE 169 (1968).

15Throughout this discussion, I will assume that guilty tortfeasors are always sued and always lose. The analysis can be modified to include uncertainty by replacing the punishment with its expected value.

16For simplicity, I am assuming here that the only cost of running the system is litigation cost, and that it is simply proportional to the amount paid out in damages. A fuller version of the analysis would take account of additional costs, such as those associated with incorrect court decisions, and would allow for the possibility that costs might be related to damages in a more complicated fashion, as discussed in the appendix.

17As I show in the appendix, the argument can be made more precise. If litigation costs are proportional to damage payments, as I assume in my examples, then the efficient level of damages is equal to the harm done if elasticity of supply (at the efficient level of damages) is one, greater if elasticity is greater than one , and less if it is less than one.

18At this point, I am still assuming that all occurrences of torts result in successful suits. Dropping that assumption would lower expected cost in the case of both the accidental and the intentional tort.

19Such a system might also involve payments by the government to victims of torts for which the efficient damage payment was low.

20Note that this is a qualitative, not a quantitative, argument. I have not shown that the efficient solution requires the same amount paid by the tortfeasor and received by the victim, only that both should be larger the more elastic the supply of the offense.

21This seems to imply that there should be a relation between how good the mechanisms of a particular legal system are at solving problems and how hard the problems are that they are asked to solve, with specialized courts allowing more complicated legal rules. This point was suggested to me by Judge Posner.

22This point is relevant to many issues other than the one I have been discussing. One example occurs in Professor David Owen's contribution to this volume. He justifies the idea that all humans are entitled to equal respect, and thus have equal rights, in part on the grounds that all humans have the capacity for rational thought and possess free will. This strikes me as a singularly unconvincing argument, tenable only because of the reluctance of those who hear or read it to come out against equal rights. Rationality, after all, is something that we all believe is possessed in unequal amounts by different people, although we may sometimes disagree about which of us are the rational ones. If rights are founded on rationality, that would seem to imply unequal rights. So far as free will is concerned, the evidence available to me that my cats have free will is the same as the evidence that my colleagues do, yet I am confident that neither group would concede equal rights to the other.


A more plausible way to justify equal (legal) rights, I think, is to start with the fact that although we differ in our degree of rationality, there is no simple, objective way of measuring that difference. If courts are instructed to give different rights to people according to how rational they are, the result will be a great deal of very expensive litigation, with everyone trying to prove that he belongs at the top of the hierarchy of rationality, and hence of rights. Whether or not someone is a human being, on the other hand, is fairly easy to determine. We therefore instruct our courts to give the same rights to all humans, with a few exceptions for differences that are fairly easy to measure, such as age and the more extreme forms of mental incapacity.

233Very likely, more than eliminate--the cost to the victim is at least the gain to me (since I could always offer to pay him to do what I want) and may be more.

24Note that the argument does not apply to reckless torts, unless we are willing to claim that recklessness is evidence the tort was really deliberate.

25If, as suggested before, about half of what losing tortfeasors pay is consumed in litigation expenses, then damages sufficient to make the victim whole, after allowing for his litigation expenses, imposed with a probability of more than 50%, would result in the expected cost to the tortfeasor being greater than the damage done, and thus deter all torts worth less to the tortfeasor than they cost the victim.

26For simplicity, I ignore other litigation costs, such as the value of the judge's time; including them would not affect my results.

27Landes and Posner (1987).