The Economic Structure of Tort Law [Reprint 2014 ed.] 9780674864030, 9780674864023

Citation preview

The Economic Structure of Tort Law

T H E E C O N O M I C S T R U C T U R E OF

TORT LAW William M. Landes and Richard A. Posner

HARVARD UNIVERSITY

PRESS

Cambridge, Massachusetts, and London, England

1987

Copyright © 1987 by the President and Fellows of Harvard College All rights reserved Printed in the United States of America 10 9 8 7 6 5 4 3 2 1 This book is printed on acid-free paper, and its binding materials have been chosen for strength and durability. Library of Congress Cataloging

in Publication

Data

Landes, William M. The economic structure of tort law. Includes indexes. 1. Torts—Economic aspects—United States. I. Posner, Richard Α. II. Title. KF1251.L36 1987 346.7303 86-18450 ISBN 0-674-23051-5 (alk. paper) 347.3063

For Lisa and Charterte

Preface

the early 1970s a number of studies have appeared that apply economic theory to the common law—the body of English and American judge-made rules, many of great antiquity, governing torts (civil wrongs that result in personal injury or property damage), contracts, property, crimes, and many other fields of private conduct. Surprisingly, many of these studies find that common law rules can best be explained as if they were designed to increase economic efficiency. This literature is summarized in Richard A. Posner, Economic Analysis of Law, pt. 2 (3d ed. 1986). This book tests the efficiency theory of the common law by examining the rules of tort law. It is the first book-length study that attempts to apply the theory to a single field of law, as well as the first book-length study of the economics of tort law. Simple economic models of alternative liability rules are constructed and applied. Because the book is addressed to both economists and lawyers, we present both formal economic models and verbal interpretations. The models yield predictions regarding the specific rules of tort law that would be adopted to promote efficiency, and we compare the predictions of the models with the actual rules. We of course do not attempt to discuss all of the hundreds of thousands of reported tort cases, but we do consider a representative sample and touch on most of the important common law rules of tort law. The qualification "common law" is important, however: statutory and constitutional torts are not discussed. Even within the common law area, torts that arise mainly out of problems in the market for information—notably deceit (misrepresentation), unfair competition (including trademark infringement), defamation, and invasion of privacy— either are not discussed or are considered only briefly, as in the treatment of defamation and misrepresentation in Chapter 6. They have been anaSINCE

viii · Preface lyzed from an economic standpoint more extensively elsewhere. See William Bishop, "Negligent Misrepresentation through Economists' Eyes," 96 L.Q. Rev. 360 (1980); Ellen R. Jordan and Paul H. Rubin, "An Economic Analysis of the Law of False Advertising," 8 J. Legal Stud. 527 (1979); William M. Landes and Richard A. Posner, "Trademark Law: An Economic Perspective" (forthcoming in }. Law & Ecort.)·, Richard A. Posner, The Economics of Justice, pt. 3 (1981). Although our emphasis is on the exposition and testing of a positive theory of tort law, we have also ventured to make occasional suggestions for law reform. The reader's attention is directed particularly to our proposal regarding the measurement of damages for loss of life in Chapter 6; and, in Chapter 9, to our recommendation regarding the determination of liability and damages in cases where an illness or injury (as from low-level nuclear radiation) is not experienced until many years after the allegedly tortious act and it is impossible to determine who among the exposed population would have escaped illness except for that act. The book was completed, however, before the recent explosion of concern with the effect of tort law on the price and availability of liability insurance. See, for example, Report of the Tort Policy Working Group on the Causes, Extent, and Policy Implications of the Current Crisis in Insurance Availability and Affordability (GPO Feb. 1986). Also, the new sixvolume revision of the Harper and James tort treatise appeared too late for us to make use of it. See Fowler V. Harper, Fleming James, Jr., and Oscar S. Gray, The Law of Torts (2d ed. 1986). Our book also has a pedagogic dimension: we believe that our attempt to express the essential features of tort law in a handful of simple models provides a key to understanding what is at first glance an incomprehensible array of unrelated rules and doctrines. If we are right, tort law is a much simpler, more elegant, and more consistent body of law than most tort scholars have suspected. We hope that readers who are not fully convinced by our theory will nevertheless agree that the economic approach provides an illuminating perspective from which to understand better the basic principles of tort law and their interrelation. The book draws heavily, although with many additions and revisions, on several of our articles: "Salvors, Finders, Good Samaritans, and Other Rescuers: An Economic Study of Law and Altruism," 7 ]. Legal Stud. 83 (1978), copyright 1978 by the University of Chicago; "Joint and Multiple Tortfeasors: An Economic Analysis," 9 J. Legal Stud. 517 (1980), copyright 1980 by the University of Chicago; "An Economic Theory of Intentional Torts," 1 Intl. Rev. Law & Econ. 127 (1981), copyright © 1981 Butterworths; "The Positive Economic Theory of Tort Law," 15 Ga. L. Rev. 851 (1981),

Preface

· ix

copyright © 1981 Georgia Law Review Association, Inc.; "Causation in Tort Law: An Economic Approach," 12 J. Legal Stud. 109 (1983), copyright 1983 by the University of Chicago; "Tort Law as a Regulatory Regime for Catastrophic Personal Injuries," 13 ]. Legal Stud. 417 (1984), copyright 1984 by the University of Chicago; "A Positive Economic Analysis of Products Liability," 14 J. Legal Stud. 535 (1985), copyright 1986 by the University of Chicago. Brief passages from Richard A. Posner, "Can Lawyers Solve the Problems of the Tort System?" 73 Calif. L. Rev. 747 (1985), copyright © 1985 California Law Review, Inc., are reprinted in Chapter 1. We are grateful to the National Bureau of Economic Research, the Law and Economics Program of the University of Chicago Law School, and the Center for the Study of the Economy and the State at the University of Chicago for research support; to Linz Audain, Gregory Barton, Carole Cooke, Gregory Kabel, Andrew Miller, Dwight Miller, Randal Picker, Helene Serota, Susan Stukenberg, Douglas Weinfield, and Michael Williams for research assistance; to Steven Shavell and an anonymous reader for Harvard University Press for their comments on the entire manuscript; to Michael Aronson of the Press for his encouragement and support; and to Charlene Posner for editorial assistance. W.M.L R.A.P

Contents

1

The Positive Economic Theory of Tort Law

1

2

Property Rights versus Liability Rules

29

3

Strict Liability versus Negligence

54

4

Basic Principles of Accident Law

85

5

Additional Principles of Accident Law

123

6

Intentional Torts and Damages

149

7

Joint and Multiple Torts

190

8

Causation

228

9

Catastrophic Personal Injuries

256

10

Products Liability and Industrial-Accident Law

273

11

A Brief Summary, with Suggestions for Further Research

312

Case Index

317

Author Index

321

Subject Index

325

The Economic Structure of Tort Law

. 1 . The Positive Economic Theory of Tort Law explores the hypothesis that the common law of torts is best explained as if the judges who created the law through decisions operating as precedents in later cases were trying to promote efficient resource allocation. We call this hypothesis the positive economic theory of tort law because no rival positive economic theory of tort law has been proposed. The objective of this chapter is to introduce the nonlawyer to tort law. We trace the history of the positive economic theory of tort law and relate it to other tort scholarship, legal and economic, discuss the principal criticisms of the positive economic theory, and explain the structure of the book. T H I S BOOK

Tort Law, in Brief The word tort comes (via Old French) from the Latin word tortus, which means "twisted" or "crooked." A tort is any wrongful act, other than a breach of contract, for which a civil lawsuit may be brought by a private person. Striking a person in anger (battery), running down a pedestrian carelessly (negligence), entering another's land without authorization (trespass), and slandering someone (defamation) are examples of tortious conduct. Tort law is older than criminal law; it predates the creation of the state. In most ancient and primitive societies, if A hits Β and injures him, A has committed a wrong against Β for which Β is entitled to compensation from A. Indeed, antecedents of many of the modern torts are found in ancient and primitive legal systems. 1 1. On the tort law of primitive and ancient societies see Richard A. Posner, " A Theory

2

·

The Economic Structure of Tort Law

Tort law was part of the customary or unwritten law of England at the time of the Norman Conquest,2 but the earliest cases that became recognized as sources of common law tort rules were those decided by the royal courts established by William the Conqueror's successors. To bring an action in one of these courts the individual had first to obtain a document known as a writ from the Lord Chancellor's office, directing the local sheriff to bring the defendant before the court to answer the plaintiff's charges. The writ of trespass vi et armis (by force and arms) was at first the principal vehicle by which tort cases were brought before the royal courts. As early as the twelfth century courts were awarding damages in battery cases commenced by this writ. By the end of the fourteenth century the writ was being used in cases of assault (a threatening gesture not involving an actual contact with the plantiff's person); and soon the writ of "trespass on the case" was devised to allow various indirect injuries to be brought before the courts, such as an injury inflicted on the plaintiff by the defendant's servant. 3 By the latter part of the eighteenth century, when Blackstone wrote his treatise on the English common law, 4 several of the modern principles of tort law were already part of the common (unwritten, judge-made) law of England;5 and they were adopted by the state and federal courts of the United States after the American Revolution. Before the coming of the railroad, however, tort law was an unimportant field because few reported cases involved accidents, as distinct from intentional wrongs such as assault and battery, which were rarely litigated. There was considerable uncertainty whether accident cases were governed by a principle of strict liability (an injurer is liable to the victim for the consequences of the injury whether or not the injurer was careless or otherwise at fault) or by some other principle such as neg-

of Primitive Society, with Special Reference to L a w , " 23 /. Law & Econ. 1, 4 2 - 5 2 (1980), reprinted with some changes in Richard A. Posner, The Economics of Justice 192-203 (1981), and references cited therein. See also Roscoe Pound, "The End of Law as Developed in Legal Rules and Doctrines," 27 Harv. L. Rev. 195 (1914). 2. Anglo-Saxon tort law followed the primitive pattern: "injuries and assaults to the person were dealt with by a minute scale of fixed compensations." 1 Frederick Pollock and Frederic William Maitland, The History of English Law 53 (2d ed. 1899). 3. On the history of English tort law see J. H. Baker, An Introduction to English Legal History, ch. 4 (2d ed. 1979); M. J. Prichard, "Trespass, Case and the Rule in Williams v. Holland," 22 Camb. L. }. 234 (1964). 4. William Blackstone, Commentaries on the Laws of England (1765-1769). 5. See 3 id. at 120-27, 2 1 6 - 2 2 (1768).

The Positive Economic Theory

·

3

ligence (liability only for injuries caused by carelessness). 6 The explosion of accident cases consequent upon the development of the railroad resulted in a considerable amplification and refinement of the tort principles governing accidents. By the middle of the nineteenth century it was settled in both England and America that the normal standard of liability in accident cases was negligence; only exceptionally was it strict liability. In the same period the forms of action (trespass, trespass on the case, and the like) were abolished. By the end of the nineteenth century tort law had assumed essentially its modern shape. This century has seen a contraction in the common law of torts in some respects, mainly in connection with industrial accidents. In the nineteenth century the right of a worker to obtain compensation from his employer because of an injury suffered on the job was governed by the law of negligence. The workmen's compensation movement swept the land in the early decades of this century and replaced negligence liability with a form of strict liability administered by commissions rather than by courts. The movement did not succeed in all industries, however; in particular, railroad and maritime workers, by the Federal Employers Liability Act and the Jones Act, respectively, were placed under an extremely liberal negligence standard that eliminated important common-law defenses to liability.7 More than offsetting the displacement of tort liability by the workmen's compensation movement has been the expansion of that liability as a result of the elimination of all sorts of immunities, such as the tort immunities of governmental and charitable agencies; the elimination of many defenses to tort liability, notably the privity defense in products liability and, in many states, contributory negligence (replaced by comparative negligence) and assumption of risk; and the enormous expansion of constitutional tort law (not examined in this book). The domain of tort law is greater today than ever before; whether the expansion is consistent with efficiency will be considered in due course. 8 6. The question whether and to what extent accidents were subject to a rule of strict liability rather than negligence before the nineteenth century has been debated extensively. For good discussions of the issue see the articles cited in note 3 above; Charles O. Gregory, "Trespass to Negligence to Absolute Liability," 37 Va. L. Rev. 359 (1951); Robert L. Rabin, "The Historical Development of the Fault Principle: A Reinterpretation," 15 Ga. L. Rev. 925 (1981). 7. See 45 U.S.C. §§ 51-60; 46 U.S.C. § 688(a). Industrial-accident law is discussed very briefly in note 56 below and more extensively in Chapter 10. 8. Particularly in Chapters 9 and 10.

4

·

The Economic Structure of Tort Law

A n t e c e d e n t s of the Positive E c o n o m i c Theory of Tort L a w The scholarly tradition from which the positive economic theory of tort law emerged begins with Jeremy Bentham, who first applied economics to laws regulating nonmarket behavior, or perhaps even Adam Smith. But the first even protoeconomic scholarly discussion of tort law itself that we have found is the analysis of tort law in Oliver Wendell Holmes's book The Common Law, published in 1881. With the benefit of hindsight it is possible to find in Holmes's chapters on trespass and negligence, and in somewhat later articles on tort law by James Barr Ames and Henry Terry, prefigurings of the modern economic approach. Holmes suggested that the only difference between negligence and strict liability as tort standards was that the latter provided a form of accident insurance. 9 As we shall see, this is not the whole of the difference from an economic standpoint, but it is an important part of it. Ames wrote that the law was "utilitarian," 10 although he did not explain what he meant by this term, and Terry described the negligence standard in terms of a balance of utilities.11 Yet absent from these early writings is any awareness that tort law might establish standards of conduct designed to promote efficient resource allocation. Commentators recognized the deterrent effect of tort law 12 but did not connect the idea that tort principles 9. Addressing the question why a man is not liable for an injury that he could not have foreseen inflicting, Holmes wrote: "The general principle of our law is that loss from accident must lie where it falls, and this principle is not affected by the fact that a human being is the instrument of misfortune. But relatively to a given human being anything is accident which he could not fairly have been expected to contemplate as possible, and therefore to avoid... " A man need not, it is true, do this or that act,—the term act implies a choice,—but he must act somehow. Furthermore, the public generally profits by individual activity. As action cannot be avoided, and tends to the public good, there is obviously no policy in throwing the hazard of what is at once desirable and inevitable upon the actor. "The state might conceivably make itself a mutual insurance company against accidents, and distribute the burden of its citizens' mishaps among all its members.. .The state does none of these things, however, and the prevailing view is that its cumbrous and expensive machinery ought not to be set in motion unless some clear benefit is to be derived from disturbing the status quo. State interference is an evil, where it cannot be shown to be a good. Universal insurance, if desired, can be better and more cheaply accomplished by private enterprise." Oliver Wendell Holmes, Jr., The Common Law 94-96 (1881). 10. James Barr Ames, "Law and Morals," 22 Harv. L. Rev. 97, 110 (1908). 11. Henry T. Terry, "Negligence," 29 Harv. L. Rev. 40 (1915). 12. See, for instance, William Schofield, "Davies v. Mann: Theory of Contributory Negligence," 3 Harv. L. Rev. 263, 269 (1890): "In an action for negligence it is of no consequence to the law whether the particular defendant shall be compelled to pay damages, or whether the loss shall be allowed to lie where it fell. The really important matter

The Positive Economic Theory

·

5

are based on utilitarian values to the idea that liability deters conduct that is not justifiable on utilitarian grounds. 13 The next burst of tort scholarship came in the heyday of the legalrealist movement, the 1920s and 1930s. Legal realists such as Fleming James and Charles Gregory argued that it was naive to talk of fault in connection with accidents or to expect that the law could have much effect on the accident level; legal doctrine, they believed, had little effect on how cases were actually decided (this was their "realism"). They saw the only attainable function of tort law as the provision of social insurance. On this basis they recommended placing liability, regardless of fault, on injurers having "deep pockets" and advocated the abolition of defenses such as contributory negligence and assumption of risk that reduced the scope of liability.14 Among the modern descendants of the legal realists can be found both the advocates of "no-fault" compensation (in effect, compulsory accident insurance coupled with abolition or curtailment of tort liability) and many advocates of strict liability. The latter turn Holmes on his head. He had opposed strict liability on the ground that the state had no business providing insurance. Modern realists, to the contrary, urge either strict liability or no liability plus compulsory accident insurance is to adjust the dispute between the parties by a rule of conduct which shall do justice if possible in the particular case, but which shall also be suitable to the needs of the community, and tend to prevent like accidents from happening in the future." See also note 14 below. 13. An article by a French legal scholar, published in 1921, contains an amazingly prescient discussion of accident law in terms of externalities. See Rene Demogue, "Fault, Risk, and Apportionment of Loss in Responsibility," 15 III. L. Rev. 369, 373, 380 (1921). Demogue may have been familiar with Pigou's work (see note 18 below and accompanying text), although he does not cite it. 14. Some examples of this literature are William O. Douglas, "Vicarious Liability and the Administration of Risk I , " 38 Yale L. J. 584 (1929); Gregory, note 6 above; Fleming James, Jr., "Accident Liability Reconsidered: The Impact of Liability Insurance," 57 Yale L.J. 549 (1948); Fleming James, Jr., and John J. Dickinson, "Accident Proneness and Accident Law," 63 Harv. L. Rev. 769 (1950). One distinguished realist, however, Clarence Morris, stressed the deterrent effect of tort liability and even proposed a deterrent theory of respondeat superior (the liability of an employer for the torts of his employees committed in the course of their employment) that is very similar to the one we set forth in Chapters 4 and 7. See Clarence Morris, "The Torts of an Independent Contractor," 29 III. L. Rev. 339, 341 (1934) ("Servants, who cannot be reached by the law directly [because they lack the financial ability to respond to damage judgments], are deterred from committing torts through threatened and actual punishment by their masters," provided the master is liable in damages for the servant's tort) (footnote omitted). (Master and servant here mean employer and employee.) The point had been made even earlier. See H. J. Laski, "The Basis of Vicarious Liability," 26 Yale L.J. I l l , 113-14 (1916).

6

·

The Economic Structure of Tort Law

on the ground that the only function of tort law should be to provide insurance coverage to those who are not already insured against the risk of being injured in an accident.15 Our present interest, however, is not in the normative analysis of the legal realists but in their premise that the legal system in practice diverges fundamentally from its formal functioning in accordance with legal doctrine. The third wave of tort scholarship relevant to this book, the explicitly economic analysis of torts, began in 1961 with the publication of Ronald Coase's seminal article on social cost and Guido Calabresi's first article on tort law.16 The germ of the analysis is Bentham's proposition that people maximize utility in all areas of life.17 Although this implies that liability rules can be used to affect the level of accidents, Bentham himself did not comment on this implication. A more direct antecedent of the modern economic approach to torts is the concept of social cost, or external cost, notably as articulated by Pigou. Using among other examples that of the locomotive that emits sparks which damage the crops of farmers located along the railroad right-of-way, Pigou noted the potential divergence of social from private cost. 18 The crop damage, he argued, is not a private cost to the railroad; and only private costs determine behavior. But it is a social cost because the farmer is a member of society, as is the railroad. Therefore, unless some means were found to force the railroad to internalize the cost, there would be too much or too careless railroading. Maybe because Pigou himself thought that the proper way to force the cost to be internalized was by means of a tax, he did not discuss tort liability. Social-cost analysis was not applied to tort law until the Coase and Calabresi articles. Coase's article is best known for its criticism of Pigou's position. Coase showed that if transaction costs were zero, the railroad and farmer in Pigou's example of locomotive sparks would 15. The leading advocate of the position that all accidents should be governed by either strict liability or no fault (that is, no liability) is Jeffrey O'Connell. See, for example, Ending Insult to Injury: No-Fault Insurance for Products and Services (1975); see also Robert E. Keeton and Jeffrey O'Connell, Basic Protection for the Traffic Victim: A Blueprint for Reforming Automobile Insurance (1965). 16. See R. H. Coase, "The Problem of Social Cost," 3 J. Law & Ε con. 1 (1960); Guido Calabresi, "Some Thoughts on Risk Distribution and the Law of Torts," 70 Yale L.J. 499 (1961). Although it bears a date of 1960, the issue of the Journal of Law and Economics in which Coase's article appeared was actually published in 1961, and Calabresi's paper was written without knowledge of Coase's. 17. See Jeremy Bentham, A Fragment on Government and an Introduction to the Principles of Morals and Legislation 125, 298 (W. Harrison ed. 1948 [1789]). 18. A. C. Pigou, The Economics of Welfare 134, 192 (4th ed. 1932).

The Positive Economic Theory

·

7

bargain to an efficient allocation of resources whether or not the railroad was liable for the crop damage. This important theoretical insight, which has come to be known as the Coase theorem, is discussed in the next chapter. An initially neglected aspect of Coase's article was his examination of actual English nuisance cases and his suggestion that the English courts had displayed a better intuitive grasp of the economics of the problem than had economists in the Pigovian tradition.19 Calabresi was not primarily interested in how (or whether) the courts tried to use tort principles to internalize the costs of accidents. Like Bentham in regard to criminal law, Calabresi was interested in constructing an efficient system of accident law from first principles rather than in appraising the existing system of accident law, the tort system— although in later work he criticized that system for failing to conform to the requirements of economic efficiency.20 Calabresi has made important contributions to positive analysis, however, particularly in relation to the distinction between property rights and liability rules and the principles of causation in tort law. These are discussed in subsequent chapters of this book. After a ten-year hiatus following the publication of the Coase and Calabresi articles, economic scholarship on torts erupted in a sustained flow that continues to this day along two paths which can, in a rough way, be traced back to the Coase and Calabresi articles. Calabresi had sketched a model of efficient accident law that Peter Diamond and other theoretical economists proceeded to formalize.21 Coase had suggested that the common law was a mechanism for internalizing social costs, and Demsetz, sketchily,22 and one of us, more comprehensively, pro19. See Coase, note 16 above, at 17-28, 38. 20. See, for instance, Guido Calabresi, The Costs of Accidents: A Legal and Economic Analysis, pt. 4 (1970). 21. The formal literature on liability rules is cited and discussed in Chapter 3. 22. See Harold Demsetz, "Issues in Automobile Accidents and Reparations from the Viewpoint of Economics" (June 1968), in Charles O. Gregory and Harry Kalven, Jr., Cases and Materials on Torts 870 (2d ed. 1969). Demsetz devotes only two pages to this question. See id. at 873-74. Many economists in this period probably agreed with Calabresi that the tort system, especially in accident cases, was an inefficient system of accident regulation. See William Vickrey, "Automobile Accidents, Tort Law, Externalities, and Insurance: An Economist's Critique," 33 Law & Contemp. Prob. 464 (1968). Most economists, however, simply had no interest in the matter. A 1964 article by Blum and Kalven (law professors) on automobile compensation plans (later published as a book) contained unmistakable hints that the negligence principle might be explained in economic terms, although the authors' own view was that both the actual and the appropriate concern of tort law was to promote equity rather than efficiency. See Walter J. Blum and Harry Kalven, Jr., Public Law Perspectives on a Private Law Problem: Auto Compensation Plans 63, 65 (1965). Finally,

8

·

The Economic Structure of Tort Law

ceeded to develop this insight. In a 1972 article, Posner, on the basis of a study of more than fifteen hundred late nineteenth-century and early twentieth-century tort cases, argued that the negligence standard itself and a number of related doctrines (contributory negligence, last clear chance, assumption of risk, and others) were methods of bringing about an efficient allocation of resources to safety and care. 23 An article published the next year as well as the first edition of his Economic Analysis of Law, published in 1973, extended the analysis to strict liability.24 In the same year John Brown produced his formal analysis of liability rules.25 Since then, we and other economic analysts of law, Steven Shavell in particular, have analyzed various areas of tort law with a view to testing their efficiency. 26 Although this literature, of which the present book is of course an illustration, is in the realist tradition in treating law as a manifestation of social policy, it does not have the skeptical cast of the older legal realism; it accepts the existence, validity and importance of legal doctrine, although it seeks to explain it in economic terms. In ending our review of the scholarly literature here, we do not want to leave the impression that the economic analysis of law has swept all other forms of tort scholarship from the board. That is far from true, and we shall cite the other forms throughout this book. Our purpose, however, has been to sketch the history of the economic approach. Moreover, the noneconomic literature does not provide an alternative positive theory of tort law to the economic theory expounded in this book. Conventional legal scholarship in torts as in other areas is overwhelmingly normative in its emphasis; this is also true, perhaps particularly true, of the newer scholarship that is influenced by moral philosophy (such as the Fletcher piece, cited earlier). This is not to say that conGeorge P. Fletcher, "Fairness and Utility in Tort Theory," 85 Haw. I. Rev. 537, 5 4 2 - 4 3 , 5 5 7 - 6 0 , 5 6 3 - 6 4 (1972), sketched a cost-benefit model of negligence law but, like Blum and Kalven, rejected it. 23. See Richard A. Posner, " A Theory of Negligence," 1 J. Legal Stud. 29 (1972). Fletcher, note 22 above, at 5 4 2 - 4 3 , like Posner, noted the implicit economic character of the Hand formula—of which more in Chapter 4. 24. See Richard A. Posner, "Strict Liability: A C o m m e n t , " 2 /. Legal Stud. 205 (1973). Richard A. Posner, Economic Analysis of Law 92-95 (1973). 25. See John Prather Brown, "Toward an Economic Theory of Liability," 2 }. Legal Stud. 323 (1973). Brown's work is perhaps better classified with that of Diamond and the other formalizers of ideal as distinct from actual liability rules. Although Brown modeled concepts of negligence, contributory negligence, comparative negligence, and strict liability, he did not attempt to give these terms their legal meanings. 26. This literature is cited and discussed throughout the book. See also the economic notes scattered throughout Richard A. Posner, Tort Law: Cases and Economic Analysis (1982), which in part draw on, and in part are drawn on by, the present book.

The Positive Economic Theory

·

9

ventional scholars or philosophical commentators have conceded the validity of the economic theory as a positive theory of tort law; most lawyers and law professors still believe, with Blum and Kalven, that the actual as well as the ideal function of tort law is to achieve fairness rather than efficiency. But in the absence of a more precise specification of fairness we find no necessary incompatibility between a positive theory that stresses fairness and one that stresses efficiency. Maybe in this area fairness equals efficiency; so indeed we shall argue in the next part of this chapter. In the last part of Chapter 8 we shall assess one of the rare attempts of conventional torts scholarship to give meaning to fairness in a positive theory of tort law.

Criticisms of the Theory The economic analysis of tort law has been criticized on a variety of grounds. Many of the criticisms are not, however, criticisms of the positive theory. For example, critics like Ronald Dworkin have argued that a system of law designed to promote efficiency is immoral.27 However forceful a criticism this might be of a normative economic theory of tort law, it is largely misplaced when directed at the positive theory. We are interested in explaining, rather than defending, the common law of torts. Its moral inadequacy, if it is morally inadequate, is relevant only insofar as that might cast doubt on the plausibility of attributing such an approach to the judges who made the common law of torts—a criticism addressed below. Its Behavioral

Assumptions

A recurrent objection to the positive economic theory of tort law is that the theory rests on unrealistic behavioral assumptions,28 with the consequence that tort doctrines, however well calculated in the abstract to promote efficient resource allocation, do not actually affect human behavior. Most people, it is argued, do not even know those doctrines, and in any case behavior in the face of danger is dominated by a concern for personal safety rather than for the legal and hence financial conse27. See Ronald M. Dworkin, "Is Wealth a Value?" 9 }. Legal Stud. 191 (1980). 28. A position argued forcefully in G. Edward White, Τort Law in America: An Intellectual History, 220-23, 230 (1980). See also Gary T. Schwartz, "Contributory and Comparative Negligence: A Reappraisal," 87 Yale L.j. 697 (1978); text accompanying note 35 below.

10

·

The Economic Structure of Tort Law

quences of behavior—especially on the part of those who carry liability insurance. And (the argument continues) people lack sufficient information about accident probabilities to make carefully calibrated judgments concerning accident avoidance. These points, which imply that tort law is not an effective method of deterring inefficient behavior, draw support from a deep vein of skepticism concerning the deterrent effect of law generally. We are not persuaded, on either empirical or theoretical grounds. The same points (except with regard to liability insurance) could be and have been made to show that the criminal law does not deter violent crime, yet a large body of statistical evidence demonstrates that both the severity and the certainty of criminal punishment have a substantial deterrent effect on such crime.29 Moreover, although there has been little systematic study of the deterrent effect of tort law, what empirical evidence there is indicates that tort law likewise deters, even where, notably in the area of automobile accidents, liability insurance is widespread (although its economic significance is misunderstood, as we shall see) and personal safety might be expected to be of greater concern than the potential financial consequences of an accident.30 One way it deters is 29. For a recent review of this literature see David J. Pyle, The Economics of Crime and Law Enforcement, ch. 3 (1983). 30. See Richard W. Grayston, "Deterrence in Automobile Liability Insurance" (1971) (Ph.D. thesis, Univ. of Chi., Grad. Sch. Bus.); Elisabeth M. Landes, "Insurance, Liability, and Accidents: A Theoretical and Empirical Investigation of the Effect of No-Fault Accidents," 25 /. Law & Econ. 49 (1982); Peter L. Swan, "The Economics of Law: Economic Imperialism in Negligence Law, No-Fault Insurance, Occupational Licensing and Criminology," Australian Econ. Rev., 3d qu. 1984, at p. 92; R. Ian McEwin, "No Fault Insurance and Automobile Accidents: Some Australian Evidence" (mimeo., Univ. of South Wales, 1985). Also relevant is Peltzman's study showing that drivers respond to a compulsory seat-belt law by driving less carefully. See Sam Peltzman, "The Effects of Automobile Safety Regulation," 83 J. Pol. Econ. 677 (1975); but see Bjorn Lindgren and Charles Stuart, "The Effects of Traffic Safety Regulation in Sweden," 88 J. Pol. Econ. 412 (1980); Robert W. Crandall and John D. Graham, "Auto Safety Regulation and Offsetting Behavior: Some New Empirical Estimates," 74 Am. Econ. Rev. Papers & Proceedings 328 (May 1984). Many other economic studies also find that people behave rationally in response to dangers to health and safety. See, as examples, Richard J. Butler, "Wage and Injury Rate Responses to Shifting Levels of Workers' Compensation," in Safety and the Work Force: Incentives and Disincentives in Workers' Compensation 61 (John D. Worrall ed. 1983); Rachel Dardis, "The Value of a Life: New Evidence from the Marketplace," 70 Am. Econ. Rev. 1077 (1980); Craig A. Olson, "An Analysis of Wage Differentials Received by Workers on Dangerous Jobs," 16 J. Human Resources 167 (1981); Richard Thaler and Sherwin Rosen, "The Value of Saving a Life: Evidence from the Labor Market," in Household Production and Consumption 265 (Nestor E. Terleckyj ed. 1975); W. Kip Viscusi, Risk by Choice: Regulating Health and Safety in the Work Place, ch. 3 (1983). Some economic evidence, however, suggests that people underestimate the benefits of safety devices—specifically, seat belts. See Richard J. Ar-

The Positive Economic Theory

·

11

by forcing up liability insurance rates to the point where people are priced out of driving—notably young people whose parents are unwilling to pay the cost of insurance.31 In addition, a study by Elisabeth Landes finds that in some states the adoption of no-fault laws may have increased automobile accident deaths by as much as 15 percent.32 This finding may seem incredible, because no-fault plans preserve tort liability for fatal and other serious accidents. But care in driving is stochastic (that is, probabilistic); if the incentive to take care is reduced because the scope of liability is reduced, people will be less careful, and the infrequent but cumulatively significant result will be more fatal accidents. A recent survey of the empirical literature relating to the effects of tort law on automobile accidents contains additional evidence that tort law leads to lower accident rates.33 Moreover, automobile accidents are not the only type of conduct regulated by tort law; and there is widespread agreement that the imposition of tort liability on professionals (for example, in the form of medical malpractice), and on business and other enterprises, does affect behavior,34 does deter—some think too much! (This is a standard criticism of medical malpractice.) The case against the deterrent effect of tort liability has been restated recently by Howard Latin, who writes: The fact that some decisionmaking undoubtedly occurs on the unconscious level in no way suggests that people can achieve optimal unconscious choices. Unless the unconscious mind is omniscient, possesses nould and Henry Grabowski, "Auto Safety Regulation: An Analysis of Market Failure," 12 Bell ]. Eccm. 27 (1981). 31. This is the thesis of Grayston's study. See note 30 above. 32. See Landes, note 30 above, at 50. Similarly Swan's study of New Zealand's no-fault law, note 30 above, at 103, found that the law had caused a 20 percent increase in automobile accident deaths. 33. See Christopher J. Bruce, "The Deterrent Effects of Automobile Insurance and Tort Law: A Survey of the Empirical Literature," 6 Law b Policy 67, 84-90 (1984). 34. On the effects of products liability on manufacturers' safety decisions see George Eads and Peter Reuter, Designing Safer Products xvii (R-3022-ICJ, Rand Corp. 1983). And for empirical evidence of the effect of tort liability on behavior of physicians see Peter A. Bell, "Legislative Intrusions into the Common Law of Medical Malpractice: Thoughts about the Deterrent Effect of Tort Liability," 35 Syracuse L. Rev. 939, 965-73 (1984); Daniel J. Givelber, William J. Bowers, and Carolyn L. Blitch, "Tarasoff, Myth and Reality: An Empirical Study of Private Law in Action," 1984 Wis. L. Rev. 443, 477-83; Jerry Wiley, "The Impact of Judicial Decisions on Professional Conduct: An Empirical Study," 55 So. Calif. L. Rev. 345, 383-84 (1981); Note, "Where the Public Peril Begins: A Survey of Psychotherapists to Determine the Effects of Tarasoff," 31 Stan. L. Rev. 165 (1978). Bell concludes from a survey of studies that "medical malpractice liability probably is resulting in safer medical behavior." 35 Syracuse L. Rev. at 973.

12 · The Economic Structure of Tort Law infinite computational and attentional capacities, and is free from cognitive biases, unconscious decisionmaking could not possibly achieve maximizing results in all cases. To put it another way, there is no evidence and no reason to presume that the unconscious mind is superhuman. 35

This passage reveals a fundamental misconception about the economic approach to human behavior: that its validity depends on the assumption that people have superhuman mental qualities. This misconception is common among lawyers and therefore worth taking a moment to try to dispel. Economic models frequently are built on unrealistic assumptions. For example, the conventional model of perfect competition assumes a market that consists of an extremely large number of small firms making the identical product. It assumes that no firm can affect the market price by altering its own output and that neither buyers nor sellers have any search costs. These assumptions commonly are false. Nevertheless, the model is quite good at predicting things like the response of the price of cigarettes or gasoline to the imposition of an excise tax. More refined models may yield even better predictions but the simple, unrealistic model does quite well.36 This is possible for two reasons. The first of course is that excluded factors may cancel each other out. The second is that the builders of economic models, for the sake of simplicity or logical rigor, often make more demands on reality than they have to. It cannot be proved mathematically that a firm will be a price-taker unless the number of firms in the market is infinite. As a matter of fact, though, a firm may be a price-taker when there are only ten firms in the market— maybe only two, or if entry into the market is sufficiently easy, one. Similarly, economic studies of behavior that implicitly assume, unrealistically, that the consumer "possesses infinite computational and attentional capacities" may still predict correctly how consumers will respond to incentives created by a seat-belt law, no-fault automobile compensation plans, or liability insurance premiums. We do not deny that people frequently are inattentive, clumsy, ignorant about the law and about accident probabilities, and so forth, and that these deficiencies blunt the effectiveness of tort law as a deterrent to careless behavior. As a matter of fact, these points play an important role in the positive economic theory of tort law. But granted that tort law is a less than adequate system of social control in many of the areas that Professor Latin is 35. Howard A. Latin, "Problem-Solving Behavior and Theories of Tort Liability," 73 Calif. L. Rev. 677, 685 n. 46 (1985) (emphasis in original). 36. For an interesting illustration of this point, see Jack Hirshleifer, Price Theory and Applications 4 1 9 - 2 0 (3d ed. 1984).

The Positive Economic Theory

· 13

concerned with,37 the empirical studies we have cited suggest that it is far from totally inefficacious. A persistent fallacy is that liability insurance externalizes the costs of accidents and hence reduces the deterrent effect of tort law.38 It does reduce deterrence, but it does not necessarily create an externality and thus need not reduce the efficiency of the tort law as a method of social control—always bearing in mind that the economic function of tort law is to optimize rather than minimize the number of accidents. If injurers are fully liable for accidents owing to their negligence and if they persuade others (liability insurers) to bear a part of the burden of this tort liability, there is no externality. Victims are by definition fully compensated for any extra accidents that occur, and liability insurers are fully compensated ex ante for the payments they make to the injurers whom they insure. There may be more accidents, but the situation with liability insurance is Pareto superior to the situation without it (provided, we emphasize, that victims indeed are fully compensated); victims are no worse off, and both injurers and insurers are better off—otherwise they would not contract for insurance. (We say there "may be," not there "must be," more accidents, because insurers will have an incentive to take steps to reduce the accident rate of their insureds.)39-Of course, this assumes that potential injurers have enough insurance or other assets to be fully responsible financially for the accident costs they impose; if not, there is a potential externality,40 just as there is if damage awards in tort cases underestimate victims' losses. Even if tort law does not have a significant effect on behavior, the theory advanced in this book is not refuted. Ours is a theory of the rules of tort law rather than of the consequences of those rules for behavior. It might seem that if the rules had no effect on behavior, an 37. For additional evidence of this see Patricia M. Danzon, "Tort Reform and the Role of Government in Private Insurance Markets," 13 }. Legal Stud. 517 (1984); Patricia M. Danzon, Medical Malpractice: Theory, Evidence, and Public Policy (1985). An important issue is the extent to which juries honestly apply legal doctrines. To a substantial extent but not completely is the answer suggested by Donald Wittman, "The Price of Negligence under Differing Liability Rules," 29 J. Law & Ε con. 151 (1986). 38. Unmasked in Steven Shavell, " O n Liability and Insurance," 13 Bell J. Econ 120 (1982). 39. As emphasized in James, "Accident Liability Reconsidered," note 14 above, at 5 5 9 62. James advocated converting tort law into a scheme for social insurance and therefore was at pains to argue that insurance does not necessarily increase the accident rate. His article also contains an excellent discussion of the deterrent effects of tort liability. See id. at 5 5 7 - 6 3 . 40. See William R. Keeton and Evan Kwerel, "Externalities in Automobile Insurance and the Underinsured Driver Problem," 27 ]. Law & Econ. 149 (1984). W e come back to this problem briefly in Chapter 9.

14 · The Economic Structure of Tort Law efficient set of rules would be one that minimized the costs of using the legal system and would thus consist solely in a rule of no liability applied in all cases. But this conclusion does not necessarily follow. Assume with Aristotle that the purpose of tort law is to do "corrective justice," that is, to restore to a person what has been wrongfully taken from him rather than to improve the allocation of resources.41 It would still be necessary to inquire into the source of the norms on the basis of which certain conduct is deemed wrongful. The source might be economic. Efforts have been made to explain ethical concepts, including the sense of being wronged, in economic (or closely related biological) terms.42 It would be consistent with these efforts to find that the tort concept of fault has an economic rationale also. An avoidable injury—implying social waste—might be perceived as wrongful and therefore arouse indignation and desire for retribution for which tort remedies are a surrogate. This would illustrate a merging of fairness and efficiency, retributive and deterrent, tort theories. The doctrinal structure would still be economic even if the social function of tort law was to assuage feelings of indignation and avert breaches of the peace rather than to promote an efficient allocation of resources to safety.

The Absence of a Causal Mechanism A second criticism of the positive economic theory of tort law is that its proponents have not explained the mechanism by which efficient tort rules might have emerged in the common law system. Twenty years ago this omission would not have seemed troublesome. In that period, one of general optimism concerning governmental intervention in the economic system, the dominant positive theory of the state was that the state supplied public goods.43 For obvious reasons the provision of public goods by a free market is a problem, and until recently it was thought 41. See Aristotle, The Nicomachean Ethics 1 1 4 - 1 7 (David Ross trans., rev. ed. 1980); Richard A. Posner, "The Concept of Corrective Justice in Recent Theories of Tort L a w , " 10 J. Legal Stud. 187 (1981). Of course, these need not be inconsistent objectives. 42. See, with special reference to indignation, Robert L. Trivers, "The Evolution of Reciprocal Altruism," 46 Q. Rev. Biology 35, 49 (1971); J. Hirshleifer, "Natural Economy Versus Political E c o n o m y , " 1 J. Soc. ö Biological Structures 319, 332, 334 (1978); Richard A. Posner, The Economics of Justice, ch. 7 (1981). 43. See, for instance, William J. Baumol, Welfare Economics and the Theory of the State (1952). Public goods are goods that can be consumed without paying for them. An example is national defense. Someone who does not contribute to the cost of national defense nonetheless enjoys whatever deterrent or other benefits it creates.

The Positive Economic Theory

· 15

that the actual as well as the ideal function of the state was to provide such goods and thereby correct a failure of the private market. Today, however, the dominant theory of the state among economists is redistributive: interest groups struggle for a place at the public trough; government intervenes in the economy to redistribute wealth from politically less powerful to politically more powerful groups rather than to provide public goods. 44 But at least in the simple form in which we have just described it, this view obviously is overstated. There are the armed forces, police, courts, and other (relatively) uncontroversial public goods provided by government, and it strains credulity to argue that these goods are produced simply as the accidental by-products of redistributive policies. Tort law, we suggest, is a public good. This is all the more plausible because it deals with activities (mainly accidents) that do not lend themselves well to redistribution in favor of politically influential interest groups. The probability of being involved in an accident, either as victim or as injurer, does not vary in a regular or consistent manner across the compact and readily identifiable interest groups that bulk so large as recipients of government largesse in the redistributive theories of government. Consequently, it does not seem plausible to suppose either that an interest group would organize to seek redistribution through the accident-law system (a group consisting of the accident prone or of drunken drivers, for example) or that some existing group—the poor, blacks, retail druggists—would place accident law high on the agenda for legislative action. We do not wish to overstate our point: the nuclear industry obtained a limitation on liability from Congress; 45 well-organized groups of the personal-injury bar lobby vigorously against no-fault automobile compensation plans; the railroad workers' unions fight modification of the Federal Employers Liability Act. However, the scope for systematic wealth redistribution ; n regard to accidents (and, even more clearly, in regard to intentional injuries such as assault and battery) does seem much more limited than in regard to tariffs, price regulation, taxation, wages, the development of public lands, and other familiar areas of special interest legislation. Where systematic redistribution is difficult to achieve, an interest group's best strategy is to support policies that will increase the wealth of the 44. See, for example, Gary S. Becker, "Pressure Groups and Political Behavior," in Capitalism and Democracy: Schumpeter Revisited 120 (R. D. Coe and C. K. Wilber eds. 1985); Sam Peltzman, "The Growth of Government," 23 J. Law & Econ. 209 (1980); George J. Stigler, "The Theory of Economic Regulation," 2 Bell J. Econ. & Mgmt. Sei. 3 (1971). 45. In the Price-Anderson Act, 42 U.S.C. § 2210. See Duke Power Co. v. Carolina Environmental Study Group, Inc., 438 U.S. 59 (1978).

16 ·

The Economic Structure of Tort Law

society as a whole, because the members of the group can be expected to share in that increase. Hence it is consistent that the AFL-CIO should support both the minimum wage and a strong national defense. The former is a redistributive policy that favors its members; the latter is a nonredistributive policy that also benefits its members, although no more than other members of society. Tort law is the same kind of policy. Some believe that although some kind of tort law may be a public good, a tort law whose objective is to promote efficiency cannot be a public good because efficiency or wealth maximization is not a defensible objective of public policy.46 This confuses two senses of the term public good. In economics a public good need not be a good affected with a public interest; it just means a good whose benefits enure to others besides those who have a contractual relationship, direct or indirect, with the producer of the good. A law of torts that increases social wealth is a public good in this sense; it might be demanded by and supplied to powerful interest groups because it would increase the size of the economic pie and therefore the absolute size of their slice of the pie, even if it did not redistribute wealth to them and thus increase the size of their slice relative to the shares of other groups. But ignore this point and assume that an efficient tort law could not be a public good unless it was good as well as public. Some definitions will help to focus the question whether tort law that promotes efficiency is good in some moral sense. We use efficiency throughout this book in the Kaldor-Hicks (or potential Pareto superiority) sense, in which a policy change is said to be efficient if the winners from the change could compensate the losers, that is, if the winners gain more from the change than the losers lose, whether or not there is actual compensation. Another way of stating the Kaldor-Hicks criterion is in terms of wealth maximization. A change is wealth maximizing if the dollar value of the gains to the winners is greater than the dollar cost of the losses to the losers. The positive economic theory of tort law holds that tort rules are efficient in the sense of wealth maximizing. A change is said to be Pareto preferred or superior (or efficient in the Pareto sense) if at least one person is made better off by the change and 46. See Dworkin, note 27 above. There is lively debate about wealth maximization as an ethical principle. See Jules L. Coleman, "Economics and the Law: A Critical Review of the Foundations of the Economic Approach to L a w , " 94 Ethics 649 (1984); Ronald M. Dworkin, A Matter of Principle, ch. 12 (1985); Posner, note 42 above, chs. 3 - 4 ; Richard A. Posner, Economic Analysis of Law, §§ 1.2, 2.3 (3d ed. 1986); Joseph M. Steiner, "Economics, Morality, and the Law of Torts," 26 U. Toronto L.J. 227 (1976); "Symposium on Efficiency as a Legal Concern," 8 Hofstra L. Rev. 485, 811 (1980).

The Positive Economic Theory

· 17

no one is made worse off. To satisfy this criterion requires in practice that all losers must be compensated. We do not argue that tort law is efficient in the strict Pareto-superior sense. For example, as we interpret the negligence standard, an accident will not give rise to liability when the cost of accident avoidance is greater than the expected accident cost. The result is efficient from the Kaldor-Hicks or wealth-maximization perspective because the "winner" (injurer) could compensate the "loser" (victim); the expected cost to the victim is less than the avoidance cost (or benefit of the accident) to the injurer. But the result is not efficient in the Pareto sense, because the victim is worse off. Various criticisms have been made of wealth maximization as a guide to public policy. We shall not discuss them in detail, because they are more relevant to normative than to positive analysis, but shall merely make the following observations. 47 1. Common law rules that are efficient in the Kaldor-Hicks sense may well be efficient in the Pareto-superiority sense on an ex ante (before the fact) basis. Suppose for example that a negligence system is cheaper overall than a strict liability system, say for automobile accidents. Then although motorists will pay more for accident (first-party) insurance and less for liability (third-party) insurance under a negligence than under a strict liability system—because under the former they have a greater chance of not being compensated if they are insured in an automobile accident but a smaller chance of being held liable should they injure another in such an accident—the sum of their insurance costs will be less under negligence. So they will be better off under negligence and will therefore prefer it to strict liability. 2. The use of efficiency in the Kaldor-Hicks sense is an ancient and honorable guide to social policy—it is the basis on which Adam Smith urged repeal of the Corn Laws, and it underlies the concept of consumer surplus introduced by Alfred Marshall as well as such important economic policies as the condemnation of monopolies and cartels. 3. The objection to the use of the Kaldor-Hicks criterion is based on distributive concerns (implicitly, as we have just seen, the criterion treats a dollar as worth the same to everybody), which were not paramount in the ideology of the formative period of the common law. 4. A tort law based on the Kaldor-Hicks criterion will be Pareto superior if the redistributive branches of government stand ready to correct any 47. These points are elaborated by Posner in the works cited in note 46 above and in Richard A. Posner, "Wealth Maximization Revisited," 2 Notre Dame }. Law, Ethics & Pub. Polict/ 85 (1985).

18 ·

The Economic Structure of Tort Law

maldistribution attributable to the application of the Kaldor-Hicks criterion in tort cases. Such a division of responsibilities is preferable on a variety of practical grounds to requiring judges and juries in tort cases to balance Kaldor-Hicks efficiency against distributive considerations.48 5. Accepting this point about the superior efficiency of the legislature as a redistributive mechanism, Kaldor-Hicks efficiency equates to utilitarianism. The objection to the Kaldor-Hicks concept is that it disregards the different utility that people derive from the same number of dollars. In principle the redistributive branch of government can redistribute wealth until the marginal utility of a dollar is the same to everybody, and then there is no ground for worry that efficient rules will reduce utility. Of course the legislature cannot do this in practice, even if it wants to. Nevertheless, a judicial system dedicated to Kaldor-Hicks efficiency could be a rational component of an overall governmental system trying as best it could to pomote the greatest-happiness principle. Although utilitarianism is not at the moment fashionable among philosophers, it was the dominant social philosophy of the nineteenth century, which was the formative period of modern tort law, and it continues to be the dominant philosophy of judges and most other people in this country who are practically concerned with public policy. Almost all judicial rationalizations of tort rules are stated in utilitarian terms. 6. The institutional characteristics of courts make it likely that courtmade substantive rules would be guided by efficiency considerations. Efficiency has long commanded broad support as an important social value, the debate being about how important it is relative to other values. Redistributive policies are highly controversial, and no single such policy commands a social consensus. It is both natural and appropriate therefore for courts, as the least overtly political branch of government, to focus on the uncontroversial value and eschew or at least downplay the controversial ones. 49 This is made still more appropriate by the fact that courts do not have good tools for achieving redistributive goals. From the redistributive standpoint all a court can do with a rule of liability is to impose an excise tax on the activity in which those engaged are made liable. Excise taxes are poorly suited for achieving "equitable" 48. See Steven Shavell, "A Note on Efficiency vs. Distributional Equity in Legal Rulemaking: Should Distributional Equity Matter Given Optimal Income Taxation?" 71 Am. Ε con. Rev. Papers & Proceedings 414 (May 1981); A. Mitchell Polinsky, An Introduction to Law and Economics 1 1 0 - 1 3 (1983). 49. A similar argument has been made for making efficiency in the Kaldor-Hicks sense the sole goal of antitrust policy. See Robert H. Bork, The Antitrust Paradox: A Policy at ]Nar with Itself, ch. 3 (1978).

The Positive Economic Theory

· 19

wealth redistributions unless the activity taxed is in the nature of a luxury good, a point with no obvious applications to tort law. The fact that the tort system is a poor method of redistributing wealth not only deters interest groups from seeking to "capture" it (a previous point) but should also reduce the incentive of redistribution-minded judges to deflect the system from the goal of efficiency. 7. Despite all the criticism of the theory that tort law is grounded in wealth maximization, there is remarkably little agreement about an alternative grounding. No theory of tort law drawn from moral philosophy commands wide support; and although, as we noted earlier, most tort lawyers, judges, and scholars would if asked say that the basis of tort law was the notion of fairness, they are unable to agree on what fairness requires. Perhaps the best summation would be that wealth maximization is one plausible normative theory of tort law among many— sufficiently plausible, we believe, to refute the argument that the judges who have fashioned the common law of torts could not possibly have found wealth maximization to be as plausible a guide as our positive theory assumes. Even if we have now persuaded the reader both that government creates some public goods notwithstanding the pressure of interest groups and that a Kaldor-Hicks (wealth-maximizing) common law of torts may be one of them, we have not explained the incentive of judges to cooperate in the production of this good. To regard judges as simply the agents of legislators who have decided to provide an efficient law of torts as they have decided to provide for the national defense ignores the fact that the judicial office is hedged about with various safeguards designed to make judges independent of legislative preferences. But of course judges are not completely independent; and persons are not likely to be elected or appointed as judges if they do not share the basic values of the dominant political groupings in society. We shall not attempt in this book to develop a theory of judicial incentives and relate it to the positive economic theory of tort law, but we feel on fairly safe ground in assuming that judges are good enough agents of society's dominant groups that if an efficient system of tort law is demanded judges will supply it—although less consistently than if they were perfect agents. Methodological

Criticisms

The positive economic theory of tort law is also criticized on the ground that the evidence for it is unsatisfactory. This may seem rather an odd

20 ·

The Economic Structure of Tort Law

point to take up here. Almost the whole point of this book is to marshal evidence pro and con the theory, and in the course of doing so we shall reply to specific criticisms of its evidentiary basis.50 What makes the issue methodological as well as evidentiary is that because ours is a theory about rules, the principal evidence for and against it consists of interpretations of rules, and several problems are encountered with this mode of empirical verification. First, rules are difficult to quantify for purposes of sampling and hypothesis testing. What would it mean, for example, to say that 82 percent of the rules of tort law are efficient? Would it mean 82 percent of someone's catalogue of the rules of tort law? Whose catalogue? Moreover, rules differ vastly in their importance. The "rule," for instance, that each dog is entitled to one bite (really an oversimplification of the principle that liability for damage inflicted by domestic animals requires proving the owner knew or had reason to know the animal was dangerous) is not nearly so important as the rule that employers are liable for the torts of their employees committed in furtherance of the employer's business (respondeat superior). No recognized method exists for weighting different rules of law by their importance. We attempt to meet this objection by examining a sufficiently large fraction of the rules of tort law that no tort lawyer can accuse us of biasing our sample. The problem can be overcome in other ways. First, appellate decisions can be sampled randomly.51 (A difficulty here, however, is that rules of no liability are for obvious reasons underrepresented in such a sampling procedure.) Second, leading cases can be identified by the number of citations to them. 52 Third, casebooks, treatises, and the Restatement of Torts can be used as a source of authoritative cases and doctrines. Given these methods (the third being the one used in this book), we do not consider the sample-bias objection to our methodology for testing the positive theory to be disabling. There is still the problem of evaluating the efficiency of each rule or decision examined. Rarely will there be enough information about the 50. See also Richard A. Posner, " A Reply to Some Recent Criticisms of the Efficiency Theory of the Common L a w , " 9 Hofstra L. Rev. 7 7 5 , 7 8 0 - 8 5 (1981). Criticisms of the evidence for the broader positive economic theory of the common law, of which the theory set forth in this book is a component, are illustrated by Note, "The Inefficient Common L a w , " 92 Yale L.f. 862 (1983), and by Cento G. Veljanovski, "The Role of Economics in the Common L a w , " 7 Research in Law & Ε con. 41 (1985). 51. See Posner, note 23 above, at 3 4 - 3 6 , 41. 52. On the use of citations to identify and weight precedents see William M. Landes and Richard A. Posner, "Legal Precedent: A Theoretical and Empirical Analysis," 19 J. Law & Ε con. 249 (1976).

The Positive Economic Theory

· 21

costs and benefits of alternative safety measures to enable a confident judgment that the court's solution is the efficient one. This problem leads to another: our methodology may not enable us to distinguish between a tort system that is efficient in some interesting sense and a system that is not totally indifferent to efficiency. We will rarely be able to say of a rule or a decision more than that it indicates a sensitivity to the economics of the situation before the court. Yet one would not want to describe the system of resource allocation in the Soviet Union as efficient because it allows people to marry whom they choose and pays people on a piece-rate basis. Merely because tort law makes some crude economic distinctions it does not follow that efficiency is its lifeblood. We have several comments on this criticism: 1. Economic analysis has identified subtle as well as gross economic distinctions made by tort law, as we shall note throughout this book. This by the way answers another common criticism of economic analysis of law—that it is merely the relabeling of commonsense intuitions in the jargon of economics—that it is, in a word, obvious; to which the critic may add, without felt inconsistency, obviously wrong. 2. It may be possible to go beyond qualitative assessments by correlating rule changes with changes in relevant quantitative variables; we do this in Chapter 10 in an effort to explain the pattern by which states have abandoned rules that once made it very difficult for plaintiffs to bring products liability suits against manufacturers. 3. It is not true that rules are a completely refractory form of scientific data. The science of linguistics is to a large extent the scientific study of rules (of sound change, grammar, syntax, and so on). 53 4. One is not persuaded that the Soviet Union allocates resources efficiently by isolated evidence of sensitivity to economic considerations because one knows that other principles besides efficiency guide resource allocation in the Soviet system; but there is no well-developed theory of what motivates or explains tort law besides efficiency. It is not as if efficiency entered as a constraint (as apparently it does in the Soviet system) on allocative choices made on different grounds. Another objection to our empirical procedure is that it is rationalization rather than prediction. We already know the rules of tort law, it is argued, and knowing them we can always put forward a plausible explanation based on economic theory. If a rule seems allocatively ineffi53. See, for instance, Theodora Bynon, Historical Linguistics ciolinguistic Patterns (1972).

(1977); William Labov, So-

22 ·

The Economic Structure of Tort Law

dent, considerations of administrative cost are available to save the theory from contradiction. Or if allocative and administrative costs appear to dictate a different rule from the actual rule, we may be able to appeal to risk aversion to bail us out. This procedure is quite different, it is said, from formulating a theory and then testing it on unknown bodies of data. Much of science in general, however, and of economic theory in particular, is directed toward explaining known data. This book is an example of applying a theory originally developed and tested on one body of data (the late nineteenth-century and early twentieth-century tort cases discussed in Posner's 1972 article) to a different body of data. The danger to which the criticism points, however, is a real one. One can complicate a theory to the point where any empirical observation is consistent with it. When that point is reached the theory cannot be refuted, and a theory that cannot be refuted cannot be confirmed either. Concern with this problem is one reason for our proposing an extremely simple economic model of tort law—a model in which, for example, risk aversion and therefore insurance play no role.54 Another possible objection to our empirical procedure is that we use as evidence to test our theory only two features of appellate opinions in tort cases—the outcome of the case and the rule that can be extracted from the opinion—and ignore the language of the opinion. Thus, it is argued, in explaining judicial opinions in economic terms we ignore the terms in which the authors themselves thought and wrote. A partial reply is that none of the actors whose behavior the economist seeks to explain uses an explicit economic vocabulary to describe his actions: not the consumer, not the worker, not the criminal, not even the businessman. But the consumer, the worker, the criminal, and the businessman are not engaged in a self-conscious, expressive activity comparable to drafting a judicial opinion. They are not required by their occupations to articulate reasons for their actions, but the judge is. The judge's reasoning ought therefore not be brushed aside as an irrelevance. We agree that judicial reasoning is relevant to the economic analysis of law, because it is essential for determining the actual meaning of the rule that can be extracted from the judicial opinion. Our quarrel is with the narrower proposition that the refusal of judges to speak in the lan54. On the methodology and scientific character of positive economics see further Milton Friedman, "The Methodology of Positive Economics," in his Essays in Positive Economics 3 (1953); Bruce J. Caldwell, Beyond Positivism: Economic Methodology in the Twentieth Century (1982); Alexander Rosenberg, Microeconomic Laws: A Philosophical Analysis (1976); William Letwin, The Origins of Scientific Economics, ch. 8 (1964).

The Positive Economic Theory

· 23

guage of economics shows that the rules they formulate are not based on efficiency. People can apply the principles of economics intuitively— and thus "do" economics without knowing they are doing it. We think that economic principles are encoded in the ethical vocabulary that is a staple of judicial language, and that the language of justice and equity that dominates judicial opinions is to a large extent the translation of economic principles into ethical language. People who say that judges are not economists are sometimes confused about the meaning of economics. If economics were limited to explicitly economic phenomena such as monopoly and inflation, it would indeed be odd to describe a judge deciding an accident case as engaged in economic reasoning. But if economics is defined as the science of rational choice55 or (equivalently) as the attempt to get the most from scarce resources, it becomes more natural to conceive of a judge in an accident case as trying to ascertain whether the injurer and the victim were behaving carefully in the sense of trying to minimize the sum of expected-accident and accident-avoidance costs. Of course the judge will not use these words to describe (even to himself) what he is doing; but the vocabulary of economics is designed for the use of scholars, not judges. Another criticism is that by drawing precedents indiscriminately from the fourteenth through the twentieth centuries we conceal significant trends in the common law—including a recent and pronounced trend away from efficiency. Far from denying that the common law has changed over the years, we regard changes in the law as important tests of the positive economic theory. If a change occurs in a relevant social circumstance (such as the relative costs of buyer and seller in determining whether a product is defective), we predict that the law will change. A related criticism is that the efficiency theory of tort law is poorly defined. Does the theory hold that all tort rules and outcomes are efficient or just most of them? If the latter, how can it be refuted? Is it a theory equally applicable to all common law jurisdictions, or does it exclude, for example, California, which in recent years has seemed to be moving away from efficiency as a governing principle of tort law? These are legitimate questions. Our reply is that throughout the history of the common law efficiency has been the dominant value embodied in tort law, but not the only value. Distributive and other nonefficiency con55. An increasingly common definition. See Gary S. Becker, The Economic Approach to Human Behavior (1976); Jack Hirshleifer, "The Expanding Domain of Economics," 75 Am. Econ. Rev. Anniversary Issue 53 (Dec. 1985).

24 · The Economic Structure of Tort Law cerns, which have shaped legislative interventions in safety questions from workmen's compensation through no-fault automobile accident compensation, have also influenced the courts in common law cases. Thus if a rule of tort law cannot be explained on efficiency grounds, this is not a contradiction of the theory, because it is consistent with the proposition that most rather than all tort doctrines are efficient. But this point does not show that our theory cannot be refuted. A theory that most rules are efficient can be refuted by evidence that most are inefficient or that a nonefficiency theory explains more tort rules than an efficiency theory. We may seem to be bobbing and weaving in a thoroughly evasive manner in refusing to hypothesize that all rather than most common law rules are efficient. But because judges both are insulated from many of the incentives that motivate other economic actors and are a part (although often a somewhat isolated part) of the political process, it is improbable that they would pursue efficient ends with the same energy and single-mindedness that actors in conventional economic markets display—and even in those markets it is the rare economist who would be so bold as to assert that all action can be explained by a particular economic model. We could discuss the methodological criticisms of the positive economic theory of tort law at even greater length, but the discussion would be inconclusive—most debates about methodology are. The reader who is unconvinced by our rebuttals should suspend judgment until the end of the book. He (or, it goes without saying, she) then will be better able to evaluate what we consider the most troubling criticism of our approach: the difficulty of forming confident judgments about the efficiency of specific case outcomes without knowing more than the opinions reveal about the facts. Here we merely note that other scholars besides ourselves have found significant support for the positive economic theory of tort law—provided that the theory is not conceived as asserting a perfect congruence between law and efficiency.56 56. The principal examples (besides—a very major exception—Shavell's work, which is referred to throughout this book) are William Bishop, "Economic Loss in Tort," 2 Oxford J. Legal Stud. 1 (1982), and "The Contract-Tort Boundary and the Economics of Insurance," 12 J. Legal Stud. 241 (1983); Christopher J. Bruce, "Testing the Hypothesis of Common Law Efficiency: The Doctrine of Informed Consent," 6 Research in Law & Ε con. 227 (1984); James L. Croyle, "Industrial Accident Liability Policy of the Early Twentieth Century," 7 /. Legal Stud. 279 (1978); William D. Manson, " A Reexamination of Nuisance L a w , " 8 Haw. J. Law & Pub. Policy 185 (1985); Mario J. Rizzo, " A Theory of Economic Loss in the Law of Torts," 11 /. Legal Stud. 281 (1982); David Partlett, "Economic Analysis in the Law of Torts," in Law and Economics 59, 7 9 - 8 5 (Ross Cranston and Anne Schick eds. 1982); Paul

The Positive Economic Theory

· 25

Organization of the Book It remains to describe, very briefly, the organization of the book, noting a few highlights of subsequent chapters. Chapter 2 focuses on the differences between property rights and liability rules as methods of internalizing costs. We show that property rights are efficient when transaction costs are low, and liability rules are efficient when such costs are high. Through an analysis of trespass law and nuisance law, we show that the law has followed this pattern in choosing between property rights and liability rules as methods of internalizing costs. We introduce our discussion of liability rules by comparing two types of liability, strict liability and negligence liability, under the simplifying assumption that harm occurs with certainty rather than probabilistically. We use this simple model of liability to explain the choice within nuisance law between a strict liability approach and a reasonableness of harm approach. Chapter 3 brings in the concept of uncertainty of harm, thus enabling us to write a full formal model of negligence liability. It also introduces into the analysis administrative costs, particularly the costs of acquiring information. With all the economic considerations relevant to the choices between strict liability, no liability, and negligence liability having thus been introduced, we proceed to model the choice among these liability rules. In Chapter 4 the model of liability rules developed in Chapter 3 is H. Rubin, Business Firms and the Common Law: The Evolution of Efficient Rules, ch. 6 (1983); Gary T. Schwartz, "Tort Law and the Economy in Nineteenth-Century America: A Reinterpretation/' 90 Yale L.J. 1717 (1981); Swan, note 30 above, at 9 4 - 9 5 ; Note, "The Case of the Disappearing Defendant: An Economic Analysis," 132 U. Pa. L. Rev. 145 (1983); Note, " A n Efficiency Analysis of Vicarious Liability under the Law of Agency," 91 Yale L.J. 168 (1981). For contrary evidence see James R. Chelius, "Liability for Industrial Accidents: A Comparison of Negligence and Strict Liability Systems," 5 /. Legal Stud. 293 (1976), which suggests that the substitution of workmen's compensation (a form of strict liability, without a defense of contributory negligence) led to a reduction in the industrial-accident rate. For critical analysis of Chelius's study see Linda Darling-Hammond and Thomas J. Kniesner, The Law and Economics of Workers' Compensation 5 6 - 6 2 (Rand Ind. for Civ. Justice 1980). Veljanovski, who has written very critically about the positive economic theory of the common law in general and of tort law in particular—see Veljanovski, note 50 above, and C. G. Veljanovski, The New Law-and-Economics: A Research Review 1 2 6 - 3 6 (1983)—has in his most recent writings displayed markedly greater sympathy for the theory. See Cento G. Veljanovski, "Economic Theorising about Tort," 1985 Current Legal Problems 117; Cento Veljanovski, "Legal Theory, Economic Analysis and the Law of Torts," in The Common Law and Legal Theory 215 (W. Twining ed. 1986); Donald Harris and Cento Veljanovski, "Liability for Economic Loss in Tort," in The Law of Tort, ch. 3 (Michael Furmstron ed. 1986).

26 ·

The Economic Structure of Tort Law

compared with the basic pattern of the tort law of accidents. First we show the fundamental conformity of the Hand formula—a compact epitome of the legal approach to negligence—to the economic model, and then we analyze the application of the negligence standard in various situations. Chapter 4 also deals with contributory and comparative negligence, some defenses to contributory negligence (such as "last clear chance"), and the areas in which the standard of tort law is strict liability rather than negligence, notably liability for ultrahazardous activities and for the torts of employees (respondeat superior). Chapter 5 continues the discussion of accident doctrines by examining the "reasonable man" standard, the role of custom, the assumption of risk defense, and the duty limitation. We show that these important tort doctrines are consistent with the economic model developed in Chapter 3. Chapter 6 models the intentional tort. We explain on economic grounds why some torts are classified as intentional, discuss briefly two important "information torts," defamation and deceit, and also consider various special doctrines such as consent and defense of property that have evolved in the intentional-tort area. Because damage principles are so important in distinguishing the legal treatment of negligent and intentional wrongdoing, we include in this chapter a discussion of the basic damage rules of tort law. In Chapter 7 we show how the rules governing the liability of joint and multiple tortfeasors, including such rules as no contribution and indemnity, are explainable on economic grounds, and we discuss the tort of interference with contract. This chapter leads to the issue of causation; Chapter 8 systematizes our analysis of the principles of causation, both factual and legal, in tort law and touches on a variety of special doctrines, such as the "eggshell skull" rule and liability for "economic loss." Chapters 9 and 10 deal with recent challenges to and developments in tort law. In Chapter 9 we examine the proposition that tort law cannot cope effectively with the catastrophic injury or mass-accident tort, typified by the asbestosis, dioxin, and diethylstilbestrol (DES) disasters. In Chapter 10 we describe the growth of products liability law—perhaps the area of tort law that provides the greatest challenge to the positive economic analyst of tort law, because it has been subjected to severe criticism on economic grounds—and contrast it with the evolution of industrial-accident law. The juxtaposition of these two fields, conventionally deemed unrelated, reflects the fact that both involve torts in a contractual setting. Chapter 11 is a very brief summary of the book, with suggestions for further research.

The Positive Economic Theory

· 27

Although products liability is the area where the reader might expect us to express the most concern with the efficiency of common-law rules, our analysis suggests that the rules of products liability make a good deal of economic sense. In making this assertion we part company with much conservative thinking on tort law, which regards modern products liability law as an unmitigated disaster. Of course the actual case outcomes may be quite different and support the conservative criticisms. This is a general limitation of our analysis. We can (or we think we can) assess the efficiency of the substantive rules of tort law and illustrate those rules with many cases that make economic sense. But we make no attempt in this book to evaluate the overall efficiency of any field of tort law, which depends on the actual administration of the law by judges, juries, lawyers, insurance claims adjusters, and others. We find the greatest divergence between efficient and actual rules of tort law in the following areas: 1. Custom and disclaimers. Our theory implies that compliance with industry custom should be a complete defense to a tort suit in situations where the costs of voluntary transactions between potential injurers and potential victims are low. But as we note in Chapter 5, the law has rejected this position (although with some exceptions) along with the cognate proposition that disclaimers of negligence liability should be enforced where transaction costs are low. 2. Damages in death cases. The common law seems never to have identified correct principles for assessing such damages (Chapter 6). 3. Victim negligence. The common law rule of contributory negligence was allocatively efficient and cheap to administer. The modern trend is toward replacing contributory with comparative negligence. Comparative negligence is allocatively efficient, too, but no more so than contributory negligence, and it costs more to administer. The additional administrative costs generate no social benefit (Chapter 4). The trend, therefore, is inefficient. 4. Joint tortfeasors. The common law rule of no contribution among joint tortfeasors was efficient and the modern trend toward allowing contribution, like the trend toward comparative negligence, appears to have as its only effect the imposition of socially sterile administrative costs (Chapter 7). That not all of the substantive rules of tort law are efficient is hardly surprising when one considers the weak incentives and limited capacity of judges to create efficient rules. Some critics of the economic approach will, we are sure, maintain that this observation means that our theory

28

·

The Economic Structure of Tort Law

has been refuted. We offer two anticipatory responses. The first is that the existence of anomalies is a challenge to further research. Why is compliance with custom rejected as a defense even when transaction costs are low? Why does the common law make no attempt to assess damages for the loss of utility experienced by the victim of a wrongful death? Why is there a trend toward comparative negligence and toward contribution among joint tortfeasors? We offer very tentative answers to the first and third questions, but research in these areas is still at the threshold. Perhaps further research will offer explanations for the rules that will enable us or other economic analysts to embed the positive economic theory of tort law (and of the common law generally) in a broader theory of law, politics, and judicial behavior. Perhaps it will even disclose as yet unsuspected efficiency properties of these seemingly inefficient rules and thereby dispel the anomalies even within the framework of our narrower theory. The refusal of the common law to impose liability for failure to rescue where the net expected benefits of rescue seem positive was once thought to be a dramatic refutation of the positive economic theory of tort law; we show in Chapter 5 that this is not the case. Maybe someday a similar discovery will be made about the anomalies that our analysis has found. Our second response is to repeat that the theory does not pretend to explain every doctrine and decision of tort law. The incentives of judges to fashion efficient doctrines are weak, although not nonexistent; what is surprising is not that judges sometimes fail to achieve efficient rules but how much of tort law can be explained on the simple hypothesis that it is indeed a system for bringing about an efficient allocation of resources to safety.

.

2

.

Property Rights versus Liability Rules

Theoretical Analysis The Difference between Property Rights and Liability Rules This chapter focuses on two important distinctions in the economic theory of tort law, the first between property rights and liability rules, the second between the two fundamental liability rules—strict liability and negligence. We discuss these rules under the simplifying assumption that there is a certainty rather than a probability of harm from a given activity (in the next chapter this assumption is relaxed). In distinguishing between property rights and liability rules, we shall introduce the Coase theorem, which plays a fundamental role in the economic analysis of tort law. A property right is an exclusive right to the use, control, and enjoyment of some resource—that is, a right to exclude anyone else in the world from using the resource without the consent of the owner of the right, irrespective of any argument that the general welfare, whether defined in economic or any other terms, would be increased by transferring the right to someone else. Thus, if A has a property right to a car, Β cannot take the car, with or without compensation, by convincing a court that the car is actually worth more to him (or to society) than it is to A. He must persuade A to sell it to him on whatever terms A sets.1 Property rights in law, however, are never so absolute as the foregoing example suggests. A's right to his car may be absolute against attempted takings by private individuals, but it is qualified against the government, which under the eminent-domain power can take his car for any public 1. See Guido Calabresi and A. Douglas Melamed, "Property Rules, Liability Rules, and Inalienability: One View of the Cathedral," 85 Harv. L. Rev. 1089 (1972); Richard A. Posner, Economic Analysis of Law § 3.1 (3d ed. 1986).

30 · The Economic Structure of Tort Law use, subject only to having to pay A the car's market value (which may be less than the car's actual value to A). Nevertheless, the conventional property rights—to land, personal property, intellectual property protected by copyrights and patents, and certain valuable resources—confer on the owner of the right a substantial power to exclude others from the use and enjoyment of such property. In contrast to a property right, a liability rule creates a right not to exclude another from using a resource but only to claim damages for certain injuries to the resource. For example, a pedestrian has no right to prevent a driver from accidentally running him down but he may be able to claim damages from the driver for the injury resulting from the accident. The difference is not between strict liability and negligence but between property rights and liability. Even if running-down accidents were governed by a rule of strict liability with no defenses, meaning that the injurer always had to pay the victim's damages, the victim would not have a property right because he would have no right to prevent the injury from occurring. Often property rights and liability rules coexist in the same resource. This is dramatically true of the human body: a person has a property right against someone who strikes him deliberately and without justification, but he enjoys only the protection of a liability rule against someone who strikes him accidentally. It is true of land as well, as we shall see later in this chapter. The practical way to distinguish between a property right and a liability rule is by reference to the legal remedy available to the prospective or actual victim of an injury. 2 One who has a right to enjoin the injurycausing conduct or to obtain punitive damages has a property right, because either the injunction or the threat of punitive damages should deter a potential injurer from seeking to appropriate the right without negotiating with the owner. But if the owner has a right just to compensatory damages, then he is protected only by a liability rule, because anyone who is prepared to pay the cost to him of an injury will not be deterred from inflicting it. In effect, a liability rule gives a potential injurer a right to inflict injury as long as he is willing to pay the victim's cost (or, under a negligence or reasonable-use rule, as long as the gain to the injurer exceeds the victim's cost—in which event the injurer will not have to compensate the victim). Thus a liability rule allows resources 2. See Frank I. Michelman, "Book Review: Pollution as a Tort: A Non-Accidental Perspective on Calabresi's Costs," 80 Yale L.J. 647 (1971); Calabresi and Melamed, note 1 above; Posner, note 1 above, § 3.8.

Property Rights versus Liability Rules

· 31

to be transferred from the victim to the injurer without the victim's consent, whereas a property right and its implementing remedies require the victim's consent to the transfer, thus preventing one person from taking something belonging to another even upon a showing that the thing is worth more to him than to the owner. The property-right approach channels transactions into the market; the liability-rule approach allows transactions to be made via the legal system, bypassing the market. When the costs of voluntary market transactions are low, the property approach is economically preferable to the liability aproach because the market is a more reliable register of values than the legal system. But when the costs of voluntary market transactions are high, the property approach is inferior because it will prevent resources from being shifted to their most valuable uses. A major prediction of the economic theory of the common law is that property rights will be assigned in settings of low transaction costs but that liability rules will be used in settings of high transaction costs. The Coase Theorem This fundamental distinction can be elucidated further through an exposition of the Coase theorem. 3 The theorem holds that the efficiency with which resources will be employed is unaffected by the initial assignment of rights, provided that transaction costs are zero. Suppose locomotives emit sparks that damage crops planted along the railroad's right of way, and there is no method of reducing the damage except by running fewer trains per day. Let R(x) be the railroad's profits as a function of the number of trains per day (x) and Rx the marginal profits (the derivative of R with respect to x), or, less formally, the increment in profits, from adding one more train. We assume that marginal profits are initially positive but declining (the first train is more profitable than the second, and so on) and that they become zero at some output and negative thereafter. This is illustrated in Table 2.1 using a profit function of the form R(x) = 4x — ('Λ)*2 (where Rz = 4 - x). Notice that profits are maximized by running four trains per day, because until that level is reached, running an additional train adds to profits.4 3. See R. H. Coase, "The Problem of Social C o s t , " 3 J. Law & Econ. 1 (1960). Our exposition borrows from the review article by A. Mitchell Polinsky, "Controlling Externalities and Protecting Entitlements: Property Right, Liability Rule, and Tax-Subsidy Appoaches," 8 /. Legal Stud. 1 (1979). 4. We assume that both the railroad and the farmer are competitive firms (that is, the

32 ·

The Economic Structure of Tort Law Table 2.1.

Railroad's profits and farmer's crop damage

χ

R(x) ($)

Rx ($)

D(x) ($)

Dx ($)

1

3.50

3

0.50

1

2

6.00

2

2.00

3 4 5

7.50 8.00 7.50

1 0 -1

4.50 8.00 12.50

2 3 4 5

6

6.00

-2

18.00

6

R - D ($) 3 4 3

0 -5

-12

Let D(x) equal the total damage to the farmer's crops and Dx equal the marginal damage. We assume that total damage increases with the number of trains and that marginal damage also increases with the number of trains (the second train causes more damage than the first, and so on). To complete the numerical example in Table 2.1, we assume that D(x) can be written as (V?)x2; hence marginal damage, Dx (the derivative of D with respect to x), equals x. The marginal profit and marginal damage schedules are illustrated in Figure 2.1. Consider first the solution where the railroad and farm have a single owner, who, being rational, naturally wants to maximize the combined profits of the two activities. Because crop damage is an explicit cost of the joint enterprise, there is no externality from the railroad's operation. If we assume provisionally that the farm does not alter the amount of crops it plants (which is implicit in the assumption that the only way to reduce crop damage is by reducing the number of trains), the joint owner will maximize ir(x) = R(x) - D(x)

(2.1)

which yields R* = DX

(2.2)

or the familiar equality between marginal benefits and costs—here marginal profits and marginal damages. In terms of our explicit functions, prices of the railroad's services and of the farmer's crop are independent of the number of trains run per day and the farmer's output respectively). Both firms may earn positive profits or rents, but this is attributable to lower costs, not monopoly power. The reason for this assumption is to avoid complicating the analysis by having to discuss potential monopoly problems at the same time that we are analyzing the effects of property rights and liability rules.

Property Rights versus Liability Rules

· 33

setting 4 - χ ( = Rx) equal to χ ( = Dx) yields χ = 2 and a maximum value of $4 for R(x) - D(x) (see Figure 2.1 and Table 2.1). The intuitive explanation for this result is straightforward: if the railroad runs more than two trains per day, the gain in profits is less than the added damages; if fewer than two trains, the reduction in profits exceeds the savings in damages. This solution assumes that it is efficient to have the railroad and the farm operating next to each other. Suppose the farm is earning gross profits of $10. At two trains per day the combined enterprise earns $14 in profits ($6 in railroad profits minus $2 in crop damages plus $10 in farm profits). If the railroad shuts down, profits fall to $10. An even worse alternative is to plant no crops. Railroad profits would rise to $8 (because four trains are being run), but $10 in farm profits would be lost. But now suppose that farm profits are only $3. Then it will be optimal not to plant any crops. If crops are planted, joint profits will be $7 when the railroad operates two trains ($6 in railroad profits minus $2 in crop damage plus $3 in farm profits) and lower if a different number of trains is run—and thus always lower than the $8 profit that would be earned if the railroad ran four trains and the farm ceased operation. As noted in Chapter 1, until Coase wrote his article on social cost economists thought that unless railroad and farm were under common ownership, giving the railroad a right to emit sparks would create an externality. Coase showed, however, that if the costs of transacting between the railroad and the farmer were zero, the railroad would run only two trains a day (we are assuming that the farm earns gross profits of $10, so that it is more efficient for the farm to operate than to shut down) even if it had a right to emit all the sparks it wanted. If initially the railroad intended to run four trains per day, the farmer would offer the railroad up to $6 to reduce the number to two, because that would lower the amount of crop damage from $8 to $2. As both parties would be better off by eliminating two trains and the costs of reaching an agreement are assumed to be zero, the railroad would agree to reduce

Figure 2.1

34 ·

The Economic Structure of Tort Law

the number of trains by two and would receive in compensation between $2 and $6, the precise amount depending on the relative bargaining skills of the two parties. 5 As should be obvious, the result is unchanged if instead of the railroad's having a right to emit sparks, the farmer has a right to enjoin the railroad from damaging his crops. The railroad will have an incentive to offer, and the farmer to accept, compensation for the crop damage caused by running more than two trains, because until that level is reached a payment of any amount between the damage to the crops and the profits to the railroad from the additional train will make both parties better off. Again the precise amount of compensation will depend on the relative bargaining skills of the parties. Of course much is concealed in the words "bargaining skills." One way to become an effective bargainer is to develop a reputation for being a hard bargainer. 6 Sometimes this may require such hard bargaining that no bargain is struck—especially if the other party to the negotiation is a hard bargainer too—even though in principle there is a price that, as in our numerical example, would make both parties better off. The problem is that there is never a single price; it is always a range, and each party is eager to engross as much as possible of the range for himself. This is an example of bilateral monopoly—an important source of transaction costs—and shows that the Coase theorem's assumption of zero (or negligible) transaction costs may not be fulfilled even in the apparently simple two-party rancher-farmer case. We shall consider in due course the common law's ingenious efforts to reduce the severity of bilateral-monopoly problems, but for now it is enough to assume, not unreasonably, that those problems, although present in the situations discussed by Coase and by us in this chapter, normally will not prevent value-maximizing transactions, though it will make them more costly. Although the number of trains is independent of the assignment of property rights, the distribution of income is not. At two trains per day 5. It may be possible for the railroad to extract more than $6 from the farmer—for example, by threatening to run five trains per day, which would generate $12.50 in crop damage (see Table 2.1) and thus cause the farmer to cease planting. If the railroad actually carried out this threat, however, its profits would be only $7.50, which is less than it would receive if it made an agreement with the farmer. So maybe the threat would not be credible. On the possibility of extortion and its relationship to the Coase theorem, see Harold Demsetz, " W h e n Does the Rule of Liability Matter?" 1 J, Legal Stud. 13, 22 (1972). 6. Even irrationally hard, as in note 5 above, where, if the railroad could credibly threaten to run five trains even though its profits would be lower, it might get more than the upper limit of the (rational) bargaining range.

Property Rights versus Liability Rules

· 35

the combined profits are, under our original assumption, $14 ($6 in railroad profits plus $10 in farm profits minus $2 in crop damage). The party with the property right will be able to obtain at least the profits he would have obtained in the absence of the other party's conflicting land use ($8 for the railroad and $10 for the farmer), and he may be able to extract part or all of the other party's profit as well. Therefore we do not say that the allocation of resources is unaffected by whom the property right is assigned to; such assignment will have wealth effects that could feed back into prices and so affect resource allocation. But efficiency as defined in Chapter 1 will not be affected. Suppose the farm in our analysis is only a marginally profitable activity: its gross profit before crop damage is not $10, but only $3. Again the single-ownership solution provides a guide to efficient resource use. That solution, as noted earlier, involves shutting down the farm. This results in a joint profit of $8 (all, of course, from the railroad's running four trains per day), compared with only $7 ($6 in railroad profit plus $3 in gross farm profit minus $2 in crop damage) if the railroad runs two trains and the farmer continues to operate. The Coase theorem shows that, under separate ownership and assuming zero transaction costs, the farm will shut down regardless of the initial assignment of rights. If the railroad has the right to damage the farmer's crop, it will run four trains per day, because a $2 payment by the farmer, which is the minimum amount the railroad would accept to reduce the number of trains to two, would leave the farmer with an income of - $1 ($3 in gross profit minus $2 in damages minus $2 payment) and hence will not be offered. Alternatively, if the farmer has the property right, he will sell it to the railroad and shut the farm down. For suppose that the railroad runs two trains per day and pays the farmer up to $6 for the right to do so. The farm will continue to operate because it will earn a net profit of $1 ($3 in gross profit minus $2 damages), but this is not an equilibrium solution. There is a potential gain of $1 if the railroad runs four trains per day and the farm shuts down. Because the railroad can offer up to $2 more to increase the number of trains to four and the farmer will find attractive an offer of at least $1 in addition to the sum it is already getting from the railroad, the parties will reach an agreement to run four trains and the farm will shut down. If we drop the assumption that the only methods of reducing crop damage are reducing the number of trains or closing down the farm, efficiency is still unaffected by who is assigned the property right. Suppose the farmer can reduce crop damage to zero either by leaving a firebreak along the railroad's right of way or by switching to a fire-

36 · The Economic Structure of Tort Law resistant crop; the cheaper of the two measures costs $2; and again the farmer's gross profits are $10. Under single ownership—and therefore also by contract if transaction costs are zero—the cheaper measure will be taken because it will yield a profit of $16 for the joint enterprise (the railroad runs four trains per day and earns $8 in profits, and the farm earns a gross profit of $10 minus $2 to eliminate crop damage), compared with a maximum profit of $14 when the railroad runs two trains per day but the $2 measure is not taken. By showing that the initial assignment of property rights is unimportant to efficiency whenever transaction costs are zero, Coase implicitly showed that property rights are preferable to liability rules whenever transaction costs are low (they are never zero in the real world)—more precisely, whenever they are lower than the gains from transacting. This assumes that property rights are cheaper to administer than liability rules (or direct public regulation) where voluntary transactions are feasible, but the assumption is reasonable, because the market registers values more accurately than the legal system. Notice that so long as transaction costs are low, the assignment of a property right to the railroad (or to the farmer) will not create an externality, even though the farmer has no preexisting contractual relationship with the railroad, unlike the railroad's customers and suppliers. An externality will arise only if transaction costs exceed the cost savings from internalizing the externality. Suppose that transaction costs are so high as to preclude a market solution. Property rights will still be feasible, although perhaps not optimal, if either the optimal level of damage is zero or the optimal solution involves no limitation of damage at all. In the first case the farmer can be given a property right against spark damage and in the second case the railroad can be given a property right to emit sparks at will. Even in these cases property rights may be objectionable as being too inflexible. It may be that at first the optimal level of damage is zero but that with changes in technology the optimal level becomes positive. In principle the farmer's property rights could be redefined as no longer including the right to be free from spark damage, but a more flexible alternative would be a rule that made the railroad pay only the damage it caused: a liability rule. The strongest argument for using liability rules rather than property rights can be made when the optimal solution to a problem of conflicting uses requires some but not total restriction of the potential injurer's use. In such a case a property-rights approach will be inefficient if, as we are

Property Rights versus Liability Rules

· 37

now assuming, transaction costs are prohibitive, because if the injurer has the property right, there will be too much damage, and if the victim has it, there will be too little. To see this, refer to the initial example in Table 2.1 where the efficient solution is to reduce the number of trains to two per day, but now assume that the cost of a voluntary transaction between the railroad and the farmer is prohibitive. In that case, if the railroad has a property right that includes emitting sparks, it will run four trains a day, an inefficient number; if the farmer has a propety right that includes preventing sparks from coming on his land, the railroad will run no trains, also an inefficient number. If instead the railroad is merely made liable for damage to the crops, the costs of that damage will be internalized to the railroad's decision making and the railroad will have an incentive to run two trains a day, the efficient number. Under the liability approach, Dx in Figure 2.1 becomes the railroad's marginal legal-damages cost of destroying crops along the right of way. The railroad will equate this marginal cost to its marginal profit from adding trains, and will thus be led to run two trains a day, the efficient result. 7 Suppose the efficient solution is not for the railroad to run fewer trains but either for the farmer to shut down, because his profit before taking account of crop damage is only $3 (and hence, with two trains per day plus crops, the combined profits are $7, compared with $8 with four trains per day and no farming) or for the farmer to eliminate all crop damage at a cost of $2 by either leaving a firebreak along the railroad's right of way or switching to a fire-resistant crop. The correct rule would be no liability. The farmer would have an incentive to adopt the most efficient method of reducing crop damage because the cost of crop damage would be a cost to him. But notice that in this case no liability is much the same as the railroad's having a property right to emit sparks at will. The only difference is that the law may be more reluctant to alter property rights than to alter liability rules as conditions change. Property rights, being designed as they are for settings of low transaction costs, are not supposed to be casually redefined by courts; they are meant to be traded in the market. If market transactions are infeasible, the more flexible regime of liability rules may be preferable even if, at the moment, 7. If, instead, the most efficient solution is to eliminate all crop damage by having the railroad install spark-arresting equipment at a cost of $1, a rule making the railroad liable for damages will lead the railroad to choose this course rather than running two trains per day and paying $2 for crop damage. The parties' joint profits (now $17) will again be maximized.

38 • The Economic Structure of Tort Law it seems that the optimal solution is the kind of extreme solution (either no damage or no damage limitation) that a regime of property rights will achieve. Thus, a simple rule of strict liability is optimal when the optimal solution is for the injurer to take measures to avoid damage and for the victim to do nothing, and no liability is optimal when the optimal solution is for the injurer to do nothing and the victim to take measures to avoid damage—always supposing that the setting is one of high transaction costs: if they were low, the efficient approach would be to assign a property right to either the injurer or the victim. But what about the case where efficiently limiting harm requires action by both parties— where it is inefficient for one party to take all the precautions and the other to do nothing—and transaction costs are high? Can an efficient solution still be achieved through liability rules, and if so which ones?

Extending the Analysis: Alternative Liability Rules with High Transaction Costs and Joint Care Table 2.2 modifies Table 2.1 by adding a row for running two and onehalf trains per day (for example, two trains on the first day, three on the second day, two on the third day, and so forth) and by adding a column based on the assumption that the farmer can create a firebreak at a cost of $0.50 that will reduce his damages from D(x) = (Ά)* 2 to (V2)(x - l) 2 . This new damage schedule is shown under the column labeled D*(x). The efficient solution is for the railroad to run two and one-half trains per day and for the farmer to leave a firebreak. This combination yields aggregate profits of $15.25 ($6.88 in railroading profits plus $10 in gross farming profits minus $1.13 in crop damage minus $0.50 in firebreak

Table 2.2.

Railroad's profits and farmer's damages, allowing for a firebreak

X

R(x) ($)

D(x) ($)

R(x) - D(x) ($)

1 2 2.5 3 4

3.50 6.00 6.88 7.50 8.00

0.50 2.00 3.13 4.50 8.00

3.00 4.00 3.75 3.00

Note: R(x)

= 4x

-

( 0 and b > 0. Notice that U is positive for all levels of income and increases at a constant rate equal to b, the marginal utility of income. Because a person's expenditure on care will be the same for all values of a and b, it is convenient to assume that a = 0 and b = 1 and therefore that U = I. If, consistently with our basic assumption of wealth maximization, we let social welfare equal the sum of all utilities, the assumption that U = I has the computational advantage of allowing us to measure social welfare as simply the sum of all incomes. It does not mean that social welfare actually is or should be simply the sum of all incomes or even a function of incomes—or even that there is such a thing as a social welfare function. The purpose of assuming a particular social welfare function is to enable us to test the hypothesis that the rules governing tort liability are best explained as efforts to minimize 9. In cases growing out of automobile accidents, the most common type of tort case, liability insurance costs $2.07 for every $1 in benefits to tort victims. Computed from U.S. Dept. of Transportation, Motor Vehicle Crash Losses and Their Compensation in the United States (1971). For similar estimates, but including products liability and medical malpractice cases as well as automobile cases, see Patricia Munch, Costs and Benefits of the Tort System If Viewed as a Compensation System (P-5921, Rand Corp. 1977); and references cited in Gary T. Schwartz, "Directions in Contemporary Products Liability Scholarship," 14 J. Legal Stud. 763, 766 (1985). 10. See James S. Kakalik et al., Costs of Asbestos Litigation (Rand Inst. Civ. Justice 1983).

Strict Liability versus Negligence

· 59

social costs. The same social welfare function was, of course, implicit in our analysis of trespass and nuisances in Chapter 2. Assume that an accident involves two people, A and B, where A is the victim (plaintiff) and Β the injurer (defendant). Defining χ as A's inputs of care and y as B's inputs of care, we can express p, the probability of an accident between A and B, as a function of χ and y—that is, V = V(X'V)·

(3-4)

We assume that px and py, the marginal products of care, are both negative and diminishing, meaning that a small increase in A's or B's inputs of care—in χ or in y—will reduce p, but at a decreasing rate.11 Diminishing marginal product is denoted by pxx > 0 and pyy > 0. Letting D denote the dollar equivalent of the injury to A if the accident occurs and A{x) and B(y) the costs of care to A and B, respectively, we can write the expected utilities (equal to expected incomes under the assumption of risk neutrality) of A and Β as LP = p[I° - D - Λ(χ)] + (1 - ρ) [1° - A{x)]

(3.5)

= Ρ - pD - A(x) and

LP = p[P - B(y)] + (1 -

ρ) [P - B(y)]

(3.6)

= Ρ - B(y), and thus the sum of their expected utilities as LP + LP = Ρ + Ρ - pD - A(x) - B(y).

(3.7)

If we assume that everyone else's utility is independent of A's and B's, social welfare is maximized when LP + LP is maximized. Because Ρ and Ρ are fixed, this is equivalent to minimizing the social costs of accidents, defined as the sum of the expected accident losses and the costs of care (or avoidance). Those social costs (which we denote by L), expressed as a function of the parties' levels of care, are given by L(x,y) = p(x,y)D + A{x) + B(y).

(3.8)

Let x* and y* be the values that minimize L(x,y)—that is, the "due care" levels—and assume that the parties' marginal costs of care, Ax and By, are positive and nondecreasing. Then x* and y* can be found by taking the first partial derivatives of L with respect to χ and y and setting 11. The first partial derivatives of ρ with respect to χ and y are px and py, and p„ and Pj,y are the second partial derivatives. Throughout the book we denote derivatives by subscripts unless otherwise indicated.

60 ·

The Economic Structure of Tort Law

the resulting expressions equal to zero. 12 This requires that x* and y* satisfy the following conditions:

and

A = "P,D

(3.9)

ßy = -pyD.

(3.10)

The results are intuitive. A should keep adding inputs of care until the reduction in expected damages ( - pxD) is equal to the marginal cost of the last unit of care. Before that point is reached, an additional input of care would confer a greater social benefit in reducing A's expected damages than it would cost in additional expenditures on care; beyond that point an additional input of care would cost more than it was worth in damage avoidance. The analysis of B's optimal care is parallel. The optimal or due care levels are shown graphically in Figures 3.2 and 3.3. In both diagrams the demand (actually marginal social product) curves show the expected reduction in the victim's damages from additional units of care, and the supply curves show the marginal costs of taking care. Demand curves are shown downward sloping, indicating decreasing marginal products of care; supply curves are shown upward sloping, indicating increasing marginal costs of care. The due care levels, x* and y*, are found at the intersections of the demand and supply curves. 13 When both x* and y* are positive, meaning that it is optimal for both A and Β to take some care, we can speak of the situation as one of "joint care." Where it is optimal for either A or Β to take care and for the other

Figure 3.2

Figure 3.3

12. Our assumptions of diminishing marginal products (p„ > 0 and pyy > 0) and increasing or constant marginal costs of care (A„ > 0 and Byy > 0) assure that x* and y* yield the minimum and not the maximum value of L. 13. Generally, the marginal product of A's care will also depend on the level of y, and the marginal product of B's care on the level of * (that is, piy Φ 0). In terms of Figure 3.2, therefore, a different demand curve exists for each level of y and in Figure 3.3 a different demand curve for each level of x. To simplify the diagrams, the demand curve in Figure 3.2 is drawn assuming y = y* and in Figure 3.3 assuming χ = χ*.

Strict Liability versus Negligence

· 61

party not to take care, we speak of "alternative care." 14 Another important distinction arises from the fact that inputs of care, χ and y, can take two distinct forms. One is to perform an activity more carefully; the other is to do less of the activity.15 One way of reducing crop damage from locomotive sparks is to reduce the level of the activity, either railroading or farming; another is to conduct the activity more carefully. The railroad could run the same number of trains but install sparkarresting equipment on each locomotive, or the farmer could plant the same amount of crops as close to the railroad tracks as before but spray the crops with fire-retardant chemical. Although unimportant at this point in our formal analysis, the distinction between more care and less activity will become important when we compare the legal standards of negligence and strict liability as methods of creating incentives for avoiding accidents. We assume that the parties' inputs of care affect only the probability of an accident and not the cost of the accident (D) if it occurs. The model could easily, indeed trivially, be expanded to allow χ and y to affect D as well as p, in order to deal, for example, with the case where wearing seat belts will reduce the cost of an accident if it occurs while not affecting the probability that it will occur; we shall discuss in Chapter 7 some doctrines by which the common law creates incentives for victims to mitigate the severity of an accident. We also assume that people can be sorted into potential injurers and potential victims rather than that everyone is simultaneously both potential injurer and potential victim. Other economic analysts have assumed that injurers and victims are drawn from an identical pool and that when they encounter each other in an accident it is a matter of pure chance who is victim and who is injurer. 16 Although our model is simpler, there are no important differences in substantive implications, as we shall see later in this chapter. Finally, we assume that A and Β are strangers and that the costs of a 14. Alternative care, for example where x* = 0, would be depicted in Figure 3.2 by a demand curve that was everywhere below the supply curve. Joint and alternative care were alluded to briefly in Chapter 2, in our discussion of the locomotive sparks case. 15. See Richard A. Posner, Economic Analysis of Law 139-41 (2d ed. 1977); Steven Shavell, "Strict Liability versus Negligence," 9 J. Legal Stud. 1 (1980). 16. See Peter A. Diamond, "Single Activity Accidents," 3 /. Legal Stud. 107 (1974); Peter A. Diamond, "Accident Law and Resource Allocation," 5 Bell J. Ε con. & Mgmt. Sei. 366 (1974); Peter A. Diamond and James A. Mirrlees, " O n the Assignment of Liability: The Uniform C a s e , " 6 Bell J. Econ. & Mgmt. Sei. 487 (1975); Elisabeth M. Landes, "Insurance, Liability, and Accidents: A Theoretical and Empirical Investigation of the Effect of NoFault on Accidents," 25 ]. Law & Econ. 49 (1982). We present a simplified version of Diamond's model later in this chapter to show that it yields the same results as our own.

62 ·

The Economic Structure of Tort Law

voluntary negotiation between them are prohibitive. (Our formal analysis could easily be extended to torts that arise out of a contractual setting—for example, where A buys a product or service from Β—and will be in Chapter 10, where we deal with products liability.) Without this assumption liability rules might not affect the level of care. The Coase theorem implies that the parties would, if they could, agree to use x* and y* inputs of care, because these are the levels of care that maximize their joint incomes. The assumption that a transaction is infeasible gives rise, as noted in Chapter 2, to a problem of externality. Why should Β expend any resources on care when the benefits of that expenditure will inure to A? And if Β takes no care, will not A (the potential victim) be led to take care that is excessive from an overall social standpoint? The question we turn to next is how liability rules can serve to internalize the costs of accidents and thereby induce A and Β to invest x* and y* respectively.

A Model of Liability Rules No Liability "No liability" is a liability rule in a sense—an economic sense—because it affects the incentives of the parties. Indeed, it is sometimes the optimal rule, as noted in Chapter 2. A will have an incentive to invest in care up to the point where - pxD = Ax, given y = 0. This point will be optimal if y* = 0—that is, in the alternative care case where A, the potential victim, is the more efficient accident avoider. If the case is one of joint care, however, so that x* and y* are both positive, no liability is inefficient. If the inputs of injurers and victims on care are partly substitutable, 17 the victim may take more care under no liability than x*. For example, if it is optimal for drivers to take some care, a pedestrian will probably take more care than x* if drivers are not liable for running down pedestrians. But he might take less than x*. Measures of accident avoidance by the potential victim that were feasible when drivers took some care against pedestrian accidents may become infeasible (unproductive) when drivers take no care against such accidents, and no other pedestrian-care measures may be feasible either. 17. By substitutable we mean that a reduction in y raises the marginal product (px) of χ—that is, pxy > 0. Because px is negative, a higher marginal product as y declines means a larger negative number (say, - 6 instead of —4) for px and therefore a positive value for Pxy

Strict Liability versus Negligence

· 63

Strict Liability Strict liability is symmetrical with no liability. Under strict liability (with no defense of contributory negligence), the potential victim, A, has no incentive to take care, because he is fully compensated for his injury, whereas Β has an incentive to invest in care up to the point where - pyD = By, given χ = 0, because under strict liability D is a cost to Β if an accident occurs. Just as no liability is optimal in the case of alternative care and y* = 0, strict liability is efficient in the case of alternative care with x* = 0—that is, where there is no reasonable (cost-justified) measure of care that a potential victim can take to avert the accident or reduce the probability that it will occur. However, an important asymmetry between no liability and strict liability appears once we relax the implicit assumption that there are no costs of operating the legal system. No liability is cheaper to administer—is in fact costless to administer. Strict liability is costly to administer because (ignoring possible defenses to strict liability, which presumably are no more extensive than the corresponding defenses to negligence) every accident gives rise to a legal claim for compensation and thus a possible lawsuit.

Negligence Under a negligence rule (and ignoring for the moment contributory negligence) an injurer is liable for his victim's damages only if the accident resulted from the injurer's failure to take due care, which for now we will assume (an assumption to be relaxed later) means a failure to use y* inputs of care (y < y*). So defined, a negligence rule gives Β an incentive to use the optimal amount of care, y*. If we put to one side the problem of contributory negligence by assuming that x* = 0, L is minimized when L(0,y*) = p(0 ,y*) + B(y*).

(3.11)

Hence we can state the negligence rule as follows: if an accident occurs, Β must pay A's damages (D) if B's inputs of care are lower than y*, the due care level; conversely, Β is not liable if y > y*. Let y0 be any level of y (including 0) less than y*. Β will choose to be negligent or nonnegligent depending on whether p(0,yo)D + B(y0) § B(y*). We know, however, that

(3.12)

64 ·

The Economic Structure of Tort Law L(0,y o ) = p{0,y 0 )D + B(y) > L(0,y*)

(3.13)

because, by definition, y* minimizes L(0,y). Since B(y*) < L(0,y*), it is also less than L(0,y 0 ); hence a negligence rule creates an incentive for a potential injurer to use due care. Strict Liability and Negligence

Compared

Many legal analysts think that strict liability would induce potential injurers to be more careful than they would be under a negligence standard. 18 Economic analysis suggests that this view is superficial. Assuming that χ* = 0, Β will choose y*, the due care level, whether he is strictly liable or liable only if negligent. Even if strictly liable, meaning that he must pay D whether he was at the due care level or not, Β will not exceed y*. By definition, at any level of care greater than y* the cost of the marginal unit of care to Β is greater than the expected reduction in his liability for A's damages. 1 9 Similarly, by definition, if Β uses less than y*, another unit of care will cost him less than the reduction in pD, which is an expected cost to him as well as a social cost. Therefore, B's expected net income, lb - pD - B(y), is greater at y* than at any other level of y. Β will not use care greater than y* under a negligence rule, because he will not be liable for any accident that occurs once the due care level, 18. Here are some examples: "Allowing injured plaintiffs to proceed on a theory of manufacturer's liability, without the necessity of proving negligence will cause manufacturers to take cautionary steps to prevent the marketing of dangerously defective products." First Nat'l Bank in Albuquerque v. Nor-Am Agricultural Products, Inc., 88 N.M. 74, 87, 537 P.2d 682, 695 (Ct. App. 1975). "Even if there is no negligence, however, public policy demands that responsibility be fixed wherever it will most effectively reduce the hazards to life and health inherent in defective products that reach the market. It is evident that the manufacturer can anticipate some hazards and guard against the recurrence of others, as the public cannot." Escola v. Coca-Cola Bottling Co., 24 Cal. 2d 453, 461-63, 150 P.2d 436,440-41 (1944) (Traynor, J., concurring). See also Phillips v. Kimwood Machine Co., 525 P.2d 1033, 1041-42 (Ore. 1974); William Prosser, "The Assault upon the Citadel (Strict Liability to the Consumer)," 69 Yale L.J. 1099, 1119 (1960); W. Page Keeton et al„ Prosser and Keeton on the Law of Torts 693 (5th ed. 1984). These authorities are not necessarily wrong; s'trict liability can result in fewer accidents, as we shall see. But the point is not self-evident, as the cited authorities seem to believe; and in Chapter 9 we show how in some cases negligence liability will induce potential injurers to take more care than if they were strictly liable. 19. The idea that strict liability will not cause a potential injurer to take more than due care even if he is made strictly liable may have been glimpsed by the court in Guinard v. Knapp-Stout & Co., 95 Wis. 482, 487, 70 N.W. 671, 672 (1897), when it said, in explaining why negligence is an adequate standard of liability, "The law aims to be practical, and to favor what is practicable."

Strict Liability versus Negligence

· 65

y*, is reached. Notice also that the victim will take zero care under both strict liability and negligence (assuming a case of alternative care where x* = 0). Under strict liability A has no incentive to take care, because he is compensated for his injury should an accident occur.20 Under negligence A does not expect to be compensated for his injury, but he still has no incentive to take care, because the costs of care are greater than the resulting reduction in expected damages. Strict liability and negligence are not, however, economically identical standards. They differ in three major respects. The first involves the costs of administering the liability rule. These costs are of two types, which we shall call information costs and claim costs. The former we define as the costs to the court system of ascertaining whether B's actual level of care, y, was equal to y*. They are zero under strict liability because due care is not an issue in a case decided under a strict liability standard, but they may be substantial in a negligence case. The second type of administrative cost, claim costs, comprises the costs (other than information costs as narrowly defined above) of processing and collecting a legal claim—that is, of determining damages, causation, and other issues not involving the level of care of the defendant and of transferring money from injurer to victim if the injurer is held liable. These costs are higher under strict liability than under negligence because there are more claims; and the gap is greater the larger the fraction of accidents unavoidable by due care. Under strict liability a claim arises every time there is an accident caused by someone worth suing; under negligence there is a claim only if the victim thinks that he can show that the defendant failed to use due care. Defenses to liability may reduce the scope of liability under both negligence and strict liability but presumably not more so under the latter; thus the relative administrative costs are unaffected. Because information costs are higher under a negligence rule and claim costs higher under strict liability, we would expect that, other things being equal, a fall in information costs would result in a shift away from strict liability and toward negligence. Looking broadly at the history of liability rules and at the differences between liability rules in primitive and in modern societies, we find that the secular decline in the costs of. information associated with growing literacy and knowledge of how the 20. Of course it is artificial to suppose that tort compensation is full compensation when the accident causes death or a very serious bodily injury; we shall explain the economics of this common-sense intuition in Chapter 6. Therefore the prospect of compensation will not always eliminate incentives by potential accident victims (or potential injurers who themselves are endangered) to take care, but presumably it will reduce them.

66 · The Economic Structure of Tort Law physical world operates has been accompanied by a movement away from strict liability and toward negligence as the dominant rule of liability.21 A second and related respect in which strict liability and negligence differ is that strict liability has a larger insurance component. Negligence liability "insures" potential accident victims against the consequences of risky events by promising to compensate them when those consequences come to pass, but the insurance is limited to the subset of accidents in which the injurer is negligent. Strict liability, by insuring against "unavoidable" (by due care) accidents as well, broadens the insurance component. The cheaper that market insurance or insurance through the family is, the less reason will the broader insurance component of strict liability provide for preferring strict liability to negligence. The third and most interesting respect in which negligence and strict liability differ concerns the incentive to avoid accidents by reducing the level of an activity rather than by increasing the care with which the activity is conducted. We mentioned earlier that y* might involve running fewer trains as well as installing spark-arresting equipment on each locomotive. Under strict liability a potential injurer will interpret y* broadly. He is interested in any measure that would reduce his expected damage by more than the cost of the measure; it is a matter of indifference to him whether the cost results from purchasing some safety input or from forgoing the profits from a higher level of productive activity. However, when a court determines y*, as it must when the rule is negligence rather than strict liability, it may take a narrower view of the potential injurer's options than the potential injurer would. The court will consider how the defendant might conduct his activity more carefully but may not consider the benefits and costs of avoiding accidents by the defendant7 s reducing his activity. For example, a court in an automobile accident case will consider whether the defendant was driving carefully when the accident occurred but will not consider whether the trip was really necessary in the sense that the benefits of the trip exceeded all of its costs, including expected accident costs.22 21. These trends, on which see Posner, note 7 above, chs. 6 - 7 , support our theory that tort law has been shaped by economic considerations. We do not want to put too much weight on this factor, though. Knowledge is growing—but the world is also becoming more complicated. Indeed, the growing complexity of products may suggest an opposite movement from negligence to strict liability for injuries caused by product defects. See Chapter 10. Admittedly we are painting with a very broad brush in speculating on these secular changes. 22. Seavey many years ago gave a striking example of this point. He said that if you

Strict Liability versus Negligence

· 67

This is essentially a point about information costs. One way of economizing on the costs of determining y* is for the court to look at only a subset of care inputs—those that relate to the safety with which the activity is carried out, as distinct from the amount of the activity. If courts in negligence suits constrict their vision in this way—and evidence discussed in the next chapter suggests that they do—then strict liability may (subject to all sorts of qualifications, however) result in fewer accidents than negligence, not by inducing the defendant to be more careful but by inducing him to reduce the level of his activity and with it the expected number of accidents. The effects of negligence and strict liability on activity levels can be explained more fully with the help of the following example. Suppose A's income equals I", before deducting expected accident costs and that B's income equals if he maintains his activity level but J? if he reduces it or switches to another activity. B's original activity might involve transporting goods via railroad, and reducing the level of activity might mean switching to barges. By assumption, > If because otherwise Β would not have chosen the higher activity level in the first place. Assume that if Β maintains this activity level, his optimal care to avoid injuring A is Β(y*), so that A's expected accident damages are p(0,y*)D; whereas if Β reduces his activity level A's expected damages are zero. 23 Provisionally, assume also that A's activity level is fixed. Rows 1 and 2 of Table 3.1 illustrate the income of the parties for different activity levels and liability rules. If we limit our focus to rows 1 and 2, the joint maximizing solution requires Β to maintain his activity level and take y* care or to alter his activity level, depending on whether Is + Ibo - p(0,y*)D - B(y*) § J§ +

(3.14)

or, equivalently, whether P0 - 7? § p(0,y*)D + B(y*).

(3.15)

In words, if the reduction in B's income from reducing his activity is kidnap a child and while driving carefully away with it run down a pedestrian, you will not be liable in negligence for the pedestrian's injury. Warren A. Seavey, "Mr. Justice Cardozo and the Law of Torts," 29 Colum. L. Rev. 20, 34 (1939), 52 Harv. L. Rev. 372, 386 (1939), 48 Yale L.J. 390, 404 (1939). 23. Notice that we have included in y (and hence in y*) only inputs of care, on the assumption that Β maintains his activity level. B's cost of changing activity levels is measured by the reduction in B's income. We would have reached the same results by defining two types of care: y0, which is care given B's activity level, and y„ which is care to alter the activity. The cost of y^ would equal B(y,) where the latter is identical to /J - /?.

68 · The Economic Structure of Tort Law Table 3.1.

Activity level and care Negligence

Maintain activity Β alters activity A alters activity

Strict liability

A

Β

A

Β

Ii - p(0,y*)D Is I"

P0 - B(y*) ii η

Ii Ii

IS - p(0,y*)D - B(y*) It IS

Iί

greater than the sum of the expected damages to A and the costs of care to B, we do not want Β to reduce his activity; if it is less, we want him to reduce it. In the second case the reduction generates a net saving in social cost, measured by the difference between the loss in income to Β (a cost)24 and the gain in reduced damage costs to A and reduced costs of care to Β (a benefit). Strict liability, by making the expected damage costs of A the expected legal-judgment costs of B, assures that Β will make the socially correct decision about whether to reduce his activity, because, with all costs thus internalized, B's choice between taking care and altering activity will depend on whether I$ - p(0,y*)D - B(y*) § l\, which is identical to the socially efficient choice in Eq. (3.15) above. Under negligence, Β can avoid liability just by taking due care, so his choice between maintaining or reducing his activity level will depend simply on whether Ig - B(y*) § /?. Β will therefore maintain his preexisting activity level when the more efficient solution is for him to alter it, which will be when (J£ - /') - B(y*) is positive but less than p(0,y)D. More generally, if B's gain in income from the higher activity level exceeds his costs of care—(/£ - /f) > B(y*)—a negligence rule is more likely to yield an inefficient outcome the greater the expected accident damages to A and the smaller the loss in B's income from reducing his activity. That loss will be smaller the greater the availability of good substitutes (not involving potential accidents) for B's initial activity. Moreover, even if the fall in B's income from reducing his level of activity (perhaps to zero) is enormous, the social cost may be much less and outweighed by the social benefits of reduced expected accident damage to A. Much of B's loss of income may be offset by gains to persons who provide substitutes for B's activity. Β might be a railroad that because of heavy sunk costs would incur financial disaster from scaling down its activities, but 24. But we shall see that B's loss of income may actually exaggerate the social costs of the reduction in the level of activity.

Strict Liability versus Negligence

· 69

maybe its losses would largely be made up by the additional profits of a barge or trucking company that would expand to take up the slack left by B's contraction. This point is obscured in our formal model because A's and B's incomes are assumed to be independent of all other incomes, but the assumption can be relaxed without undermining our analysis. Before concluding from all this that strict liability is allocatively more efficient than negligence, 25 we must consider a heretofore neglected factor, the victim's activity level. We place to one side any issue of the victim's being careless in a negligence sense by assuming that there is a defense of contributory negligence to strict liability,26 and we also assume that contributory negligence resembles negligence in that the victim's activity level normally is not considered in determining whether the victim used due care. Then strict liability and negligence have opposite and potentially offsetting effects on injurer and victim activity. Strict liability gives the potential injurer an incentive to reduce his activity when such an adjustment is less costly than his taking care (or letting the accident occur and paying the victim's damages), because he bears the costs of an accident that could be avoided by such an adjustment. But the victim has no incentive to adjust his activity; he is compensated for his injury. Under negligence the opposite is true: the injurer has no incentive to adjust his activity but the victim has because he is not compensated for any injury that occurs unless it could have been avoided by the injurer's being more careful rather than by the injurer's reducing his activity. To illustrate, if in Table 3.1 Jg = l\ = $100, I{ = $95, 7? = $80, p(0,y*)D = $10, and B(y*) = $5, it would be optimal for A to reduce his activity, because the parties' joint income would then be $195 ($95 for A and $100 for B, who maintains his initial activity level), compared with $180 when Β reduces his activity and $185 when neither party reduces activity but Β takes care equal to y*. The optimal rule would be either negligence or no liability. Under either, if A maintains his activity level, his expected income will be at most $90 (assuming that Β takes care), compared with $95 if he changes his activity level. Under strict liability A will have no incentive to change his activity level, because he will be fully compensated for any injury, so his income will be $100, which is $5 more than if he reduces his activity. 25. When - If) - B(y*) is positive but less than p(0,y*)D. By the allocative dimension of a rule, we mean, of course, its effect in internalizing external costs; by its administrative dimension we mean the costs of administering the rule. 26. The legal situation is more complicated, as we shall see in the next chapter.

70 ·

The Economic Structure of Tort Law

Thus the relative effects of strict liability and negligence on activity provide a reason for preferring strict liability only when an adjustment in the defendant's activity, but not in the plaintiff's, would be an efficient method of accident avoidance. Strict liability will reduce the defendant's activity level but it will increase the plaintiff's, and the number of accidents may be greater, fewer, or the same as under a negligence standard. This analysis yields predictions concerning the pattern of strict liability and negligence to be observed in the law if the positive economic theory of tort law is correct. If a change in the defendant's but not in the plaintiff's activity level is an efficient method of accident avoidance, strict liability is attractive and will be the rule chosen. It will deter many accidents that are not socially cost justified, and this not only will have good allocative effects but will also reduce what we have called the claim costs of strict liability (the information costs of strict liability are already low, as will be recalled). But where greater care rather than less activity is the optimal method of accident avoidance by potential injurers, the case for negligence is strengthened, because now the claim costs of strict liability are likely to be high relative to its allocative benefits. The activity point is not only relevant to the choice between strict liability and negligence. Activity was an explicit factor in the analysis of nuisance cases in the previous chapter. We saw that in the Restatement's original (now alternative) formulation of the nuisance standard the utility of the plaintiff's and of the defendant's activities is compared in deciding whether nuisance exists. Often a change in activity levels is a feasible means of avoiding damage in a nuisance case where greater care would not be. Although the reduction in neighborhood property values caused by a funeral parlor may not be eliminated by conducting funerals more discreetly or carefully, it could be eliminated by relocating the funeral parlor from a residential to a nonresidential area—a change in activity rather than care. 27 The nuisance standard illustrates the common law's awareness that when the place, kind, or amount of activity is an important factor in bringing about the right level of damage, it should be part of the legal standard. Activity level is occasionally considered in negligence cases as well. The utility of the plaintiff's activity was one of the factors singled out 27. See, for example, Williams v. Montgomver, 184 Miss. 547,186 So. 302 (1939). Keeton et al., note 18 above, at 630-33, 634-46 (5th ed. 1984), stresses the importance of relocating either the defendant's or the plaintiff's operations as a method of avoiding damage in nuisance cases.

Strict Liability versus Negligence

· 71

by Henry Terry in his formulation of the negligence standard. 28 Terry was discussing rescue cases, one of the principal areas where activity is considered in negligence cases. Suppose that A rushes into a burning building to save his hat and is burned. Even if he did everything possible to minimize the hazard to himself—short of not entering the building in the first place—he would be deemed contributorily negligent, because the cost of curtailing the activity is less than the expected accident cost. 29 The costs to the factfinder of determining the optimal activity level are low in this case, so the concept of due care is expanded to take in the level of activity as well as the care with which it is undertaken. When it is difficult to ascertain the optimal level of activity, a court applying a negligence standard will not try to do so. It will not consider, for example, whether the best way to avoid accidents at railroad crossings is for railroads to run fewer trains or for travelers using the crossing to travel less. There is a tradeoff between the information costs of considering the injurer's activity level as an aspect of due care and the allocative costs of ignoring the activity level. And although, as noted earlier, strict liability eliminates the problem of having to determine whether particular changes in the injurer's activity level would have been cost justified, it does so at the price of relaxing the victim's incentive to avoid the accident through a change in his own activity level. The impossibility of pronouncing strict liability economically superior to negligence across the board casts doubt on the argument that the movement from strict liability to negligence in nineteenth-century tort law (a movement, by the way, that has been exaggerated by some legal historians)30 amounted to a covert subsidy of productive but dangerous activities.31 It is true that, compared with negligence, strict liability increases injurers' private costs and thus diminishes their wealth; but if imposing accident costs on them is not the best method of accident control, not imposing those costs on them cannot be described as a subsidy. The repeal of an unjustified tax is not a subsidy to the former taxpayer. 28. See Henry T. Terry, "Negligence," 29 Harv. L. Rev. 40, 4 3 - 4 4 (1915), cited in Chapter 1. 29. See discussion of cases in Chapter 4. 30. As shown in Gary T. Schwartz, "Tort Law and the Economy in Nineteenth-Century America: A Reinterpretation," 90 Yale L.J. 1717, 1 7 2 2 - 3 0 (1981). 31. See Morton J. Horwitz, The Transformation of American Law, 1780-1860, at 9 9 - 1 0 0 (1977). For economic criticism of Horwitz's argument, see (besides Schwartz's article in note 30 above) Herbert Hovenkamp, "The Economics of Legal History," 67 Minn. L. Rev. 645 (1983); Stephen Williams, "Transforming American Law: Doubtful Economics Makes Doubtful History," 25 U.C.L.A. L. Rev. 1187 (1977).

72 • The Economic Structure of Tort Law One important question remains to be considered in our comparison of strict liability and negligence: why are there any negligence cases? It is easy to see why there will be cases under strict liability even if the threat of liability induces all potential injurers to take due care (y*). Although the probability of an accident will be lower than if injurers take less or no care (assuming that the plaintiff's optimal level of care is zero or negligible), the number of accidents will still be positive unless strict liability reduces the activity level of potential injurers to zero; and injurers will be liable for the damages resulting from these accidents. Under negligence, however, it might seem that the number of lawsuits should be zero. If the injurer maintains the due care level, y*, he is not liable for any accidents that occur; and he will not set a lower level of care, because he minimizes his private costs by adhering to that level. Knowing this, victims will not bring suits, because their prospects for recovery will be zero in all cases; the only victims will be victims of "unavoidable accidents," for which there is no liability under a negligence system. There are several reasons for nevertheless expecting a positive number of negligence cases even if all parties are rational economic maximizers: 1. Either the court or the injurer may make a mistake in applying the legal standard to the facts, with the result that either a nonnegligent injurer is held liable or an injurer fails to prevent an accident that could have been prevented for less than the expected accident cost. Victims may make mistakes too, but their mistakes would result in cases where negligence was alleged rather than cases where it was found. 2. Care often has a stochastic element. 32 A driver can try not to stray across the center line, but because he does not have complete control over his attentiveness, the car, road conditions, and the like, the best he can do is to drive so as to make it highly improbable that he will cross the line. In other words, a potential injurer may try to adhere to a level of care y*, but his realized care will be y = y* + e, where e is a random error term with a zero mean. Although E(y), the injurer's expected care, is y*, occasionally e will be negative and y will fall below y*. If an injury occurs when y L(x*,y*) - B(y*) > A(x*). Since Β expects A always to use due care, Β will expect to be strictly liable and will select a level of y that minimizes his expected loss p(x*,y)D + B(y). This is the due care level, y*. Thus strict liability with a defense of contributory negligence yields the same levels of due care for potential injurers and potential victims as negligence with or without a defense of contributory negligence. Even without such a defense, strict liability would yield optimal results in the alternative care case where x* = 0, that is, where it is not optimal for the victim to take any care. The problem with strict liability without a defense of contributory negligence in the joint care case is simply that the potential victim knows that even if he takes no care, the injurer will be liable if an accident occurs. There is no way that the injurer can escape liability except by preventing the accident, unlike the situation in a negligence system; there the potential injurer can escape liability by taking due care, even if an accident occurs. Knowing all this, in a regime of strict liability the potential victim has no incentive to spend anything on preventing the accident. Of course, realistically speaking, he may have some incentive, because tort compensation is not always full compensation; but his incentive would be greater if there were a defense of contributory negligence. Hence contributory negligence serves an essential allocative purpose in a regime of strict liability, whereas in a negligence system its economic function (other than in the sequentialcare case) is to reduce the costs of administering the system. If, however, victim care is unimportant, there will be less allocative need for a defense of contributory negligence to strict liability. We shall give examples of this point in the next chapter.

Comparative

Negligence

We can define comparative negligence 40 as follows: 1. If y = y*, then regardless of the victim's care the injurer is not negligent and not liable for any damages. 2. If y < y* and χ = χ*, the victim is not contributorily negligent but the injurer is negligent and pays full damages. 40. David Haddock and Christopher Curran, in " A n Economic Theory of Comparative Negligence," 14 }. Legal Stud. 49 (1985), proceed in apparent unawareness of our analysis, first published in William M. Landes and Richard A. Posner, "Joint and Multiple Tortfeasors: An Economic Analysis," 9 J. Legal Stud. 517, 5 3 7 - 3 9 (1980). Their central finding, that comparative negligence yields as a first approximation the same allocative results as contributory negligence (see 14 /. Legal Stud, at 59-63), duplicates ours.

Strict Liability versus Negligence

· 81

3. If y < y* and χ < χ*, so that both parties are negligent, the victim's damages are apportioned between A and B, s" being the share of damages that falls on the victim and sb the share that falls on the injurer. Observe that sa + sb = 1 and that under a negligence rule (with a defense of contributory negligence) sfl = 1 and sh = 0. With comparative negligence the victim's expected losses at different levels of care for injurer and victim are as shown in Table 3.6. Suppose that Β chooses y*. From the definition of due care L(x0,y*) - B(y*) > L(x*,y*) - B(y*), so A would choose x*. If Β chose y0 < y*, A would choose χ or x* depending on whether s°p(x0,y0)D + A(x0) § A(x*).

(3.23)

If s" is relatively small and A(x*) - A(x) relatively large, A would behave negligently. This suggests that we do not have a unique solution under comparative negligence, because if Β chooses y*, A will use χ*, whereas if Β is negligent, it is unclear whether A will choose x0 or x*. Now consider B's choices in reaction to A's (Table 3.7). If A chooses χ*, Β will choose y* because L(x*,y0) - A(x*) > L(x*,y*) - A(x*) > B(y*). But if A chooses x0, B's choice of y0 or y* will depend on whether sYx0,y0)D

+ B(y0) § B(y*).

(3.24)

In sum, it seems that if either party chooses due care, the other will do likewise, but if one party chooses less than due care the other may choose less than due care. Table 3.6.

A's expected losses A's care

B's care

x0 < x*

y0 < y* y*

sy(x0,y0)O + A(x0) p(Xo,y*)D + Λ(*0)

Table 3.7.

A(x*) p(x*,y*)D + A{x*)

B's expected losses B's care

A's care x0Ax, because then the railroad would have "bought" Eckert's services if transacting had been feasible. 40 If the child's life and Eckert's life are assumed to have the same value, the question whether Eckert was negligent is reduced to whether the probability of his rescuing the child was less than the probability of his being killed. If it was less, then -pxD (the probability of rescuing the child times the value of the child's life) would be greater than Ax (the probability of Eckert's losing his own life times the value of that life). In asking whether Eckert "could probably save the child without serious injury to himself," the court was comparing these probabilities; if Eckert probably could have saved the child without serious injury to himself, this implies that the probability of a successful rescue was greater than the probability of his being hit by the train himself.41 39. Id. 40. An alternative but equivalent way to model Eckert's behavior is to define the expected cost of the rescue to Eckert, As, as the injury to him discounted by the probability that he would be hit by the train. To determine the benefits of care, -pxD, we must ask what Eckert would have had to give up if he had forgone the rescue and thereby eliminated the expected accident cost to himself. The answer is the value of the child's life discounted by the probability of a successful rescue. Eckert would not be negligent if - p , D > A,. 41. Two other questions raised by this discussion are postponed to the next chapter: Why did Eckert try to save the child? Would he have been liable if he had failed to try to rescue the child, assuming that the expected benefits of the rescue exceeded the expected costs?

102 ' The Economic Structure of Tort Law If the object of the rescue had not been a child but instead property of slight value, - pxD would have been much lower because D, the value of the object rescued, would have been much lower. This much the court made clear, but presumably it did not mean to imply that it is never prudent to risk life for property. In a subsequent case, Liming v. Illinois Cent. R. Co., 42 the plaintiff, in an effort to rescue some horses, ran into a building that was on fire because of the defendant's negligence, and he was burned. The court stated: "A person would not be justified in exposing himself to as great danger in saving property as he would in saving human life, and whether the person injured acted with reasonable prudence would, in most cases, be a question of fact depending upon the circumstances under which the act was done." 4 3 Thus the court refused to hold the plaintiff contributorily negligent per se. Notice, however, that its discussion implies that changes in activity level will, at least sometimes, enter into determinations of negligence. 44 The discussion of cases has been designed to establish several points: 1. The Hand formula appears to be an accurate description of the negligence standard (including contributory negligence) in diverse periods and jurisdictions. 2. Courts apply the formula in marginal rather than total terms, and thus it coincides with the economic model of negligence. This is clearest in Blyth. 3. In none of the cases is it likely that optimal accident avoidance required a change in the level of activity of either plaintiff or defendant. 45 For example, in Adams v. Bullock, eliminating trolley lines would not have been an efficient way of preventing the accident to the plaintiff. This suggests that cases where activity is an important factor in achieving an efficient level of accidents are not decided under a negligence standard; further evidence of this point is presented later. 4. In none of the cases was the sequential application of the Hand formula (first to negligence and then to contributory negligence) a source of economic difficulties. In Hendricks, an alternative care case where the defendant was probably the lower-cost accident avoider, the defense of 42. 81 la. 246, 47 N . W . 66 (1890). 43. Id. at 254, 47 N . W . at 68. 44. See also note 14 above and Harding v. Philadelphia Rapid Transit Co., 217 Pa. 69, 66 Atl. 151 (1907). 45. Except possibly in Nussbaum, where the golfer-defendant was a trespasser on the golf course, and Liming. However, in Nussbaum the defendant's trespass was probably not a cause of the accident when cause is interpreted in legal and economic terms, but we defer explanation of this point to Chapter 8.

Basic Principles of Accident Law · 103 contributory negligence was precluded by the attractive nuisance doctrine. In Eckert, another alternative care case, the cost of accident avoidance clearly was lower to the railroad than to the victim, who was allowed to recover damages. To all this it may be replied that we have discussed only a tiny and perhaps deliberately selected sample of cases applying the negligence standard. However, one of us once examined a sample of more than fifteen hundred tort cases decided between 1875 and 1905 and found considerable congruence between the economic model and the legal practice. 46 The cases discussed here should be regarded as an addition to that large and randomly selected sample. Furthermore, most of the cases we have discussed in this chapter are leading cases on negligence. Certainly Eckert, Blyth, Stepney, Bolton, and Carroll Towing are in this category. These are not legal "sports" that happen to coincide with economic analysis. The leading treatise on tort law concludes that the Hand formula summarizes the "usual" approach used by courts to determine negligence. 47 In an effort to construct a sample that is immune from any charge of bias, we now examine all of the cases reprinted in the current edition of Gregory, Kalven, and Epstein's torts casebook that deal with the meaning of negligence. 48 Those authors cannot be accused of undue partiality to the economic theory of tort law. Charles Gregory was an advocate of converting tort law into a system of social insurance, 49 Harry Kalven believed that the proper concern of tort law was with equity rather than efficiency, 50 and Richard Epstein has long been critical of the 46. See Richard A. Posner, " A Theory of Negligence," 1 ]. Legal Stud. 29 (1972). A study that reaches similar conclusions except with regard to industrial (workplace) accidents is Gary T. Schwartz, "Tort Law and the Economy in Nineteenth Century America: A Reinterpretation," 90 Yale L.J. 1717 (1981). Cf. Gary T. Schwartz, "The Vitality of Negligence and the Ethics of Strict Liability," 15 Ga. L. Rev. 963 (1981). 47. Keeton et al., note 4 above, at 173 and n. 46. 48. See Charles O. Gregory, Harry Kalven, Jr., and Richard A. Epstein, Cases and Materials on Torts 1 0 2 - 3 0 (3d ed. 1977). This section is entitled "Calculus of Risk" and is described as an exploration of "the judicial efforts to fashion and apply a standard of reasonable care." Id. at 102. After we completed our study, the fourth edition of the book appeared: Richard A. Epstein, Charles O. Gregory, and Harry Kalven, Jr., Cases and Materials on Torts (4th ed. 1984). The "Calculus of Risk" section (pp. 138-67) contains the same cases as the third edition except that the Oppen case (see notes 5 5 - 5 7 below and accompanying text) has been omitted. 49. See Charles O. Gregory, "Trespass to Nuisance to Absolute Liability," 37 Va. L. Rev. 359, 3 8 3 - 8 4 (1951). 50. See article cited in note 22 in Chapter 1.

104 · The Economic Structure of Tort Law positive economic theory of tort law.51 Nevertheless, every one of the cases included by these authors to illustrate the legal meaning of negligence is consistent with the Hand formula interpreted in economic terms. Three of the cases reprinted in the relevant portion of the casebook have been discussed already: Blyth, Eckert, and Carroll Towing. But we discussed only the statement of the Hand formula in Carroll Towing and not the facts of the case. The defendant's bargee had left his barge for twenty-one hours, during which time the barge broke away from its mooring and caused damage. No good reason for the bargee's protracted absence was offered. The harbor was a busy one (it was the height of World War II), the days were short, and the danger of collision was manifest. Hence - pyD was substantial, whereas the lack of a satisfactory explanation for the bargee's absence implied that By was very low. The court properly found negligence. The casebook reprints the portion of the opinion in Osborne v. Montgomery 52 that deals with the negligence standard. The discussion is consistent with the economic approach. For example, the court states: Circumstances may require the driver of a fire truck to take his truck through a thickly populated district at a high rate of speed, but if he exercises that degree of care which such drivers ordinarily exercise under the same or similar circumstances, society, weighing the benefits against the probabilities of damage, in spite of the fact that as a reasonably prudent and intelligent man he should foresee that harm may result, justifies the risk and holds him not liable.53

In Cooley v. Public Service Co., 54 another case reprinted in the casebook to illustrate the meaning of negligence, a power line owned by the defendant electric company broke during a storm and hit a telephone cable that was strung below it, causing a loud noise on the telephone line that the plaintiff was using. She fainted. The evidence showed that the only devices that the power company might have used to prevent the accident would have increased the risk to people on the street below of electrocution from a falling power line. Balancing the risk of electrocution against the expected reduction in the injury from noise, the court held that the power company had not been negligent in failing to adopt either device suggested by the plaintiff. By was higher than -p y D. 51. See Richard A. Epstein, " A Theory of Strict Liability," 2 J. Legal Stud. 151 (1973). 52. 203 Wis. 223, 234 N.W. 372 (1931). 53. Id. at 232-33, 234 N.W. at 376. The court did not discuss the application of the standard to the facts of the case. 54. 90 N.H. 460, 10 A.2d 673 (1940).

Basic Principles of Accident Law · 105 Union Oil Co. v. Oppen 55 upheld the right of commercial fishermen whose livelihood had been impaired by an oil spill in the Santa Barbara Channel to recover damages from the oil companies that had caused the spill. The defendants' negligence was stipulated, and the only issue was whether the plaintiffs could recover for a purely "economic" loss (the fish that were destroyed were not their property). The court held that they could. The opinion contains a somewhat confused discussion of the economic approach to tort questions, 56 but the result is correct: giving the plaintiffs a tort right was a way of making the defendant internalize some of the external costs of the oil spill.57 Next in the casebook come a pair of railroad accident cases, Hauser v. Chicago, R.I. & P. Ry. 58 and McDowall v. Great Western Ry. 59 In Hauser, the plaintiff, a passenger on the defendant's railroad, fainted in the toilet compartment and was severely burned when her face came into contact with a hot steam pipe that ran along the wall of the compartment beneath the sink. The court held that the railroad had not been negligent in the design or construction of the toilet compartment, because the probability that a passenger would fall and wedge her face or another exposed part of the body against the hot steam pipe was too remote to require the railroad to have relocated or shielded the pipe, assuming that this would have been practicable. In economic terms - pyD was lower than By, so the railroad was not negligent in not having taken greater care. In McDowall, some boys broke into a railroad car parked on an incline and decoupled and unbraked it. The car rolled down the incline and caused damage. The plaintiff argued that the railroad should have placed the car on the other side of a "catch-point" which would have prevented it from rolling down. In rejecting this argument and exonerating the railroad from the charge of negligence, the court made two points. First, although the boys previously had broken into railroad cars to steal apples and the like, they had never before tried to decouple and unbrake a car, and the probability that they would do so was slight. Second, because opening the catch-point would have been a simpler operation for the boys than decoupling and un55. 501 F.2d 558 (9th Cir. 1974). 56. See Richard A. Posner, "Some Uses and Abuses of Economics in L a w , " 46 U. Chi. L. Rev. 281, 297-301 (1979). 57. See Richard A. Posner, "Epstein's Tort Theory: A Critique," 8 ]. Legal Stud. 457, 4 6 5 - 6 8 (1979). We return to the "economic loss" cases in Chapter 7 of this book. 58. 205 la. 940, 219 N.W. 60 (1928). 59. [1903] 2 K.B. 331 (C.A.), reversing [1902] 1 K.B. 618. Both the lower-court and the court of appeal decisions are reprinted in the casebook; since the court of appeal is more authoritative, we discuss its opinion.

106 · The Economic Structure of Tort Law braking the car, installing such a catch-point would not have been a significant obstacle to their plan. Thus, — pyD was low and (from the second point) probably negative, suggesting that it would have been inefficient to make even a negligible expenditure to move the car. The last two cases, Quintal v. Laurel Grove Hospital60 and Lucy Webb Hayes National Training School v. Perotti,61 are medical malpractice cases. In Quintal, the plaintiff, while being operated on by an ophthalmologist for an eye condition, suffered cardiac arrest, resulting in permanent brain damage. One issue was whether the ophthalmologist should have known how to perform open heart massage and if not whether a surgeon competent to perform such a procedure should have been present. There was evidence that the standard of care in the locality where the eye operation was performed required that a surgeon capable of doing open heart massage be present at every operation involving general anesthesia. Cardiac arrest is a risk in every such operation and can have terrible consequences—death or severe brain damage—if not corrected within three minutes, which is too short a time in which to get a surgeon from another part of the hospital. Another issue was whether the ophthalmologist and the anesthesiologist may have been negligent in failing to heed certain warning signs (fever and apprehensions) in the plaintiff which increased the risk of cardiac arrest. A judgment for the defendants was reversed. In Perotti, the plaintiff's decedent, who was confined in the defendant's hospital because he had a mental illness that involved suicidal tendencies, slipped out of his closed ward and while being escorted back to it broke away from his escort and plunged to his death through a closed window. The negligence issues were two: whether the window should have been either barred or made of a stronger type of glass and whether the decedent should have been allowed to escape from the closed ward. The court held for the hospital on the first issue and for the plaintiff on the second. The negative effects of barred windows on the hospital's therapeutic activities provided a strong reason against such windows, and there was no expert evidence that thicker glass than that used by the hospital—bulletproof glass, for example—would have been feasible. The hospital, however, had established a standard requiring confinement of mental patients such as the plaintiff's decedent and had violated its own standard without justification. The court noted: "This was not a case where a determined patient managed to commit suicide 60. 62 Cal. 2d 154, 397 P.2d 161, 41 Cal. Rptr. 577 (1964). 61. 419 F.2d 704 (D.C. Cir. 1969).

Basic Principles of Accident Law · 107 in a mysterious or unexpected fashion. Nor is this a case where a calculated risk was taken for therapeutic reasons with a patient of known suicidal tendencies, nor where a hospital had concluded after examination that a patient was not suicidal and hence did not require precautions." 62 In other words, -pyD was high relative to By, given that the hospital had established a closed ward for such patients and had only to secure it against a patient not shown to be "determined... to commit suicide in a mysterious or unexpected fashion." In both Quintal and Perotti the resolution of the negligence issues seems consistent with the economics of the situations presented. The reference to and deference given customary standards of care in both cases are also appropriate from an economic standpoint. The existence of a preexisting voluntary relationship between doctors and patients suggests that transaction costs were not prohibitive and hence, via the Coase theorem, that customary standards were efficient.

Areas of Strict Liability in Accident Law We now consider whether the choice that the common law has made between strict liability and negligence to govern particular kinds of accidents is consistent with the predictions of the economic model. We said that strict liability was more likely to be the superior regulatory device in cases where achieving optimal accident avoidance requires altering the defendant's activity rather than his care or the plaintiff's activity or care. Is this the pattern of the law? We begin with the rules relating to damage caused by animals. The common law distinguished between domestic animals, such as dogs, pigs, and sheep, and wild animals, such as bears and tigers. The owner's liability for damage caused by wild animals was strict. If B's pet lion clawed A, Β was liable for A's damages even if he had used all reasonable care to prevent the lion from attacking people. For damage caused by domestic animals the owner's liability again was strict—provided that he had notice of the animal's vicious disposition. This principle is the origin of the well-known "one bite" rule for dogs. Before a dog has bitten anyone, its owner will normally have no reason to think it vicious; but once it has bitten someone, he is on notice of its vicious disposition. The law is not really so mechanical: " A single incident does not necessarily place the owner on notice the animal is dangerous or vicious. 62. Id. at 710.

108 · The Economic Structure of Tort Law The test is whether the incident was of such a nature as to lead a reasonable person to believe the dog was sufficiently dangerous as to be likely to cause injury to a person at a later date." 63 Within the class of domestic animals there was an important exception: owners of trespassing cattle or other livestock were strictly liable for property damage (although not for personal injury) caused by their animals.64 We have put in the past tense our statement of the common law relating to liability for damage by animals because important developments have altered it. First, with the growth of zoos and animal parks, there has been a tendency to relax the rule of strict liability for wild animals when they are kept for purposes of exhibition as in a zoo. 65 Second, many western states have rejected the English common law rule of strict liability for property damage caused by wandering livestock.66 The overall pattern, including the secular changes in it, seems consistent with what economic analysis of the choice between strict and negligence liability would predict. In the case of a dog not known to be dangerous, the expected accident cost is too small to warrant the owner either in expending substantial resources to restrain the dog from biting people or in substituting another domestic animal. Moreover, bites by gentle dogs are often due to provocation by the victim rather than to carelessness by the owner. But once a dog is known to be dangerous, the perceived probability of its biting people and hence the expected accident costs associated with the dog rise sharply. Yet as any dog owner knows, it is difficult to restrain a dog at all times. Because care alone may not suffice to avoid an accident, we want the owner of the animal to consider whether the animal is worth keeping—a point about activity level. If vicious dogs were so valuable that the prospect of having to pay repeated damages to victims of dog bites would never deter an owner from keeping such a dog, no allocative purpose would be served by imposing liability for a bite that could not have been prevented by greater care. But because a gentle dog is ordinarily a good substitute for 63. Rolen v. Maryland Casualty Co., 240 So.2d 42, 46 (La. App. 1970). 64. An exception is considered below; see notes 6 7 - 6 8 and accompanying text. On the common law principles governing liability for damages caused by animals see Keeton et al., note 4 above, at 4 9 6 - 5 0 3 ; Glanville L. Williams, Liability for Animals (1939). 65. See, for example, City of Denver v. Kennedy, 29 Colo. App. 15, 476 P.2d 762 (1970), after remand, 31 Colo. App. 561, 506 P.2d 764 (1972). 66. See Wagner v. Bissell, 3 la. 396 (1857); Delaney v. Errickson, 10 Neb. 492, 6 N . W . 600 (1880); Beinhorn v. Griswold, 27 Mont. 79, 69 Pac. 557 (1902); 3 Thomas G. Shearman and Amasa A. Redfield, A Treatise on the Law of Negligence 1719-22 (6th ed. 1913); 1 E. Thomas, Negligence 9 9 8 - 1 0 0 2 (2d ed. 1904).

Basic Principles of Accident Law · 109 a vicious one, imposing strict liability on the owner of a dangerous dog and thereby forcing him to consider whether the benefits of having such a dog exceed the full social costs may yield substantial allocative benefits. Table 4.1 illustrates this point. Row 1 sets out the incomes, expected injuries, and costs of care when Β owns a vicious dog, and rows 2 and 3 the corresponding figures when he substitutes a gentle dog. In row 2 a gentle dog is a good substitute (B's income falls by $15), whereas in row 3 it is a poor substitute (B's income falls by $50, maybe because he lives in a high-crime area where a gentle dog provides inadequate protection). Under a negligence standard Β will choose to keep the vicious dog notwithstanding the availability of the gentle substitute, because his income net of the costs of due care will be higher ($90 compared to $85 or $50). From society's standpoint, however, if a good substitute is available (row 2) the combined incomes of the parties will be higher if a gentle dog is substituted. Strict liability will bring about this result, because the owner's expected income net of damage liability will be $65 with the vicious dog but $85 with the gentle one. In the absence of a good substitute (row 3) he will keep the vicious dog even if he is strictly liable for its bites, because his income will be $65 with and only $50 without the vicious dog in these circumstances. Thus strict liability does not eliminate vicious dogs; it just reduces their number to the point where the only people keeping vicious dogs are those who value them more than the costs such dogs impose on others. Strict liability also saves on information costs in these circumstances. Instead of a court's trying to decide who should be allowed to keep a vicious dog—by estimating for each defendant the reduction in income from substituting a gentle one—strict liability shifts that decision to the owner, the party who is in a better position to make that decision. There are many ways to reduce accident rates efficiently, and no court will have enough information on each potential defendant to determine precisely that defendant's due care or optimum activity level. But as ex-

Table 4.1.

The case of the vicious dog

Activity Keep vicious dog Good substitute available Poor substitute available

/· ($)

/«($)

viy*)D ($)

B(y*) ($)

Net combined incomes ($)

100 100 100

100 85 50

25 0 0

10 0 0

165 185 150

110 · The Economic Structure of Tort Law

plained in the previous chapter, if we are confident for a particular class of accidents that the efficient number of accidents will be attained by most potential injurers' adjusting their activity and by the few remaining injurers' taking due care, the information costs necessary to achieve efficiency will be minimized by placing liability on the people already having this information or able to obtain it at the lowest cost—that is, by imposing strict liability on injurers. The argument for making the owner of a wild animal strictly liable is even stronger than that for vicious dogs. On the one hand, care undertaken by owners and potential victims may not be effective in preventing the animal from endangering people. On the other hand, keeping such an animal is not so valuable an activity that the threat of substantial tort liability will have no effect in inducing people to substitute a less dangerous animal—again an activity-level point. The situation is different when the wild animal is kept not as a pet, curiosity, or watch animal but in a zoo or animal park. In a carefully operated zoo the probability of the animals' harming people is very low, and activity changes are unlikely to be optimal: what would a zoo be without dangerous animals? Hence the trend away from strict liability for such harm. The special rule applicable to straying farm animals and its rejection in the western states further illustrate the economic rationale of strict liability. It is difficult through fencing to prevent cattle, sheep, and other domestic farm animals from straying. Fences cannot practicably be kept in constant good repair, and cattle and sheep are continually finding ways through and around them. These straying animals do considerable damage, particularly in areas of dense cultivation. The alternative of fencing out is also costly in such areas because of the amount of cultivated land that would have to be fenced. Abandoning cultivation would not be a feasible means of reducing this damage either. These circumstances make a rule of strict liability economically attractive, because the rule forces the owner of the animals to consider either a reduction in the size of his herds or flocks or relocating his activity to a less densely cultivated area; and this is the traditional common law rule. But the western states of the United States were not densely cultivated, and livestock raising was not a marginal activity. Fencing out may well have been a more feasible method of controlling damage to crops from cattle and sheep than fencing in because the area under cultivation was smaller than the area being used for ranching. 67 67. The issue w a s debated extensively at the time. See Walter P. Webb, The Great Plains 2 8 2 - 8 3 (1931).

Basic Principles of Accident Law · 111 The analogy to the attractive nuisance doctrine, where fencing out is chosen in preference to fencing in because the area to be fenced out is so much smaller than the area that would have to be fenced in, is apparent. If limiting herd size or substituting other land uses for ranching was not a feasible means of limiting crop damage in the western states, strict liability for such damage would not be the optimal approach—and it was not followed. 68 The greater density of population and cultivation in England than in the United States may also explain the stricter liability for fire damage under English as compared with American common law. 69 For example, in England, in the absence of statutory privilege, a railroad that caused spark damage to crops along its right of way was strictly liable for the damage even if its locomotive had the latest and best spark-arresting equipment. In the United States, a railroad was liable only if negligent, negligence being defined for these purposes as not having the state-ofthe-art spark-arresting equipment. 70 Not only were locomotive sparks a greater hazard in England because of the greater density of population and cultivation, but railroading was probably a more valuable activity in the United States because of the greater distances and more severe climate.71 On both counts, limiting the number of trains was less likely to be a feasible method of damage control here than in England, so strict liability was less likely to be adopted as the liability rule here. Many of the areas of strict liability in the common law are grouped under the rubric of ultrahazardous or abnormally hazardous activities.72 The keeping of wild animals and, in England, the setting of fires are examples of such activities. The roots of this principle in the economics of optimal accident avoidance are apparent in the Restatement's list of the factors to be considered in determining whether an activity is abnormally dangerous: (a)

e x i s t e n c e of a h i g h d e g r e e of risk of s o m e h a r m t o t h e p e r s o n , l a n d o r c h a t t e l s of o t h e r s ;

(b)

likelihood t h a t t h e h a r m t h a t r e s u l t s f r o m it will b e g r e a t ;

(c)

inability t o e l i m i n a t e t h e risk b y t h e e x e r c i s e of r e a s o n a b l e c a r e ;

68. See Buford v. Houtz, 133 U.S. 320 (1890); Beinhorn v. Griswold, note 66 above; Keeton et al., note 4 above, at 540. 69. See id. at 5 4 3 - 4 4 ; Note, "Torts—Fault and Liability—Liability for the Escape of Fire," 16 Va. L. Rev. 174 (1929). 70. Compare Jones v. Festiniog R. Co., L.R. 3 Q.B. 733 (1868), with Burlington & M.R. Co. v. Westover, 4 Neb. 268 (1876). 71. See Patrick O'Brien, The New Economic History of the Railways 49, 78, 86, 89 (1977); Robert W. Fogel, "Notes on the Social Saving Controversy," 39 }. Ecort. Hist. 1, 51 (1979). 72. See Keeton et al., note 4 above, at 545-59.

112 · The Economic Structure of Tort Law (d) extent to which the activity is not a matter of common usage; (e) inappropriateness of the activity to the place where it is carried on; and (f) extent to which its value to the community is outweighed by its dangerous attributes.73 The elements in this definition are, first, a high expected accident cost (factors a and b); second, the impracticability of avoiding accidents through exercising greater care (factor c); third, the feasibility of reducing accidents by curtailing (factors d or f) or relocating (factor e) the activity. The first element when present demonstrates substantial social benefits from reducing the accident rate, the second that this cannot be done economically simply by conducting the activity with greater care, and the third that perhaps it can be done economically by curtailing the activity, because the activity, either in general or in the particular place where it is being conducted, is marginal and can be reduced or eliminated without great social loss. The Restatement's definition of ultrahazardous activities thus coincides with the economic principles that make strict liability the preferred liability rule for some activities. The leading English case on strict liability, Rylands v. Fletcher, 74 from which the modern concept of ultrahazardous activities derives, held that a landowner was strictly liable for any "nonnatural" use of his land— in that case the storage of water in a reservoir. Although not apparent from the opinion, there had been serious accidents involving the bursting of reservoirs, 75 so the expected accident cost may have been substantial. The general principle of Rylands, that of strict liability for nonnatural land uses, was absorbed into the American law as an element in the definition of an ultrahazardous activity—the "not a matter of common usage" provision of the Restatement. But the application of that principle to the storage of water in a reservoir (the narrow holding of Rylands) was rejected, especially in the western states. Because water is abundant in England, the construction of a reservoir was, at least when the case was decided, an unnatural land use in the sense of unusual, hence presumptively not high-valued. The western United States, however, is dry, and the building of reservoirs was not an unnatural use of land in the sense of a marginal use that might be abandoned if the landowner had to bear the costs of accidents caused by it. 76 The more 73. American Law Institute, Restatement (Second) of Torts § 520 (1979). 74. L.R. 3 H.L. 330 (1868). 75. See A. W. B. Simpson, "Legal Liability for Bursting Reservoirs: The Historical Context of Rylands v. Fletcher," 13 J. Legal Stud. 209 (1984). 76. As stated in Turner v. Big Lake Oil Co., 128 Tex. 155, 165-66, 96 S.W.2d 221, 226

Basic Principles of Accident Law · 113 valuable a land use is relative to its alternatives, the less likely it is to be changed by the imposition of liability for accidents that can be avoided only by altering the activity and not by using greater care, and the weaker therefore is the case for strict liability. A traditional example of an activity classified by tort law as ultrahazardous is blasting with explosives. This is, of course, a dangerous activity, and considerable danger remains even after all reasonable care has been used to confine the effects of the blast. Moreover, people who are injured or suffer property damage, especially as a result of debris, from blasting can do little either by taking greater care or by altering their activity to avoid or reduce damage. (As far as altering activity is concerned, because blasting is usually done in connection with construction, which is ubiquitous, potential victims cannot feasibly avoid blasting damage by relocating their activities. This is not true, however, of blasting in mining operations.) Finally, there are substitutes for blasting, so that strict liability may very well alter the level of the activity. In these circumstances a rule of strict liability makes economic sense. At first, tort law distinguished between different types of damage caused by blasting. 77 If the damage was from debris hurled by the explosion, the blaster was a trespasser and was strictly liable. But if the damage was from concussion or vibrations caused by the blasting, the blaster was guilty only of nuisance and therefore liable only if his conduct was unreasonable. This distinction may seem completely artificial, but it is not. The susceptibility of a building to damage from concussion or vibrations can be reduced by changes in design or building materials, by careful choice of location in reference to ground conditions, and by careful maintenance (for example, to prevent cracks from forming, which might magnify the damage from vibrations);78 and especially because (1936): " I n T e x a s w e h a v e conditions v e r y different from those w h i c h obtain in E n g l a n d . A large portion of Texas is a n arid or semi-arid r e g i o n . . . The c o u n t r y is almost w i t h o u t s t r e a m s ; a n d w i t h o u t the storage of w a t e r from rainfall in basins constructed for the p u r p o s e , or to hold w a t e r s p u m p e d from the earth, the great livestock industry of W e s t Texas m u s t p e r i s h . . . " A g a i n , in E n g l a n d there are n o oil wells, n o necessity for using surface storage facilities for i m p o u n d i n g a n d e v a p o r a t i n g salt w a t e r s therefrom. In Texas the situation is d i f f e r e n t . . . P r o d u c i n g oil is o n e of o u r major industries. O n e of the by-products of oil p r o d u c t i o n is salt w a t e r , w h i c h m u s t be disposed of without injury to property or the pollution of streams. T h e construction of basins or p o n d s to hold this salt w a t e r is a necessary part of the oil b u s i n e s s . " 77. See K e e t o n et al., n o t e 4 above, at 5 5 3 - 5 4 ; A n n o t . , 2 0 A . L . R . 2 d 1372 (1951). 78. See U . L a n g e f o r s a n d B. Kihlstrom, The Modern

Technique

of Rock Blasting (3d e d .

1978), especially § 9 . 4 ; L. D o n Leet, Vibrations from Blasting Rock (1960); National Research

114 · The Economic Structure of Tort Law there are other sources of concussion and vibration besides blasting, such as tornados and earthquakes, these changes may well be cost justified. In any event the potential victim is not so helpless in an economic sense as he would be to prevent damage—anyway very infrequent— from mere debris (it would hardly make sense for people to board up their windows to prevent damage from debris thrown their way by blasting), so it is not surprising that liability was less strict. In addition, the causality of damage through vibration is much more complex than that of damage from flying debris, and this, as we shall see in Chapter 8, is another argument against strict liability. This particular objection to strict liability for vibration damage has largely disappeared, however, as a result of research on the effects of blasting, 79 which may help explain the movement toward erasing the distinction between debris on the one hand and concussion and vibration on the other. But blasting has also become much safer and more controllable,80 which is an argument for moving the other way—toward subjecting the entire activity to negligence liability. Therefore we cannot in our present state of knowledge point to the trend toward strict liability as evidence for (or against) the positive economic theory of tort law. Another traditional but now waning example of ultrahazardous activity is flying. Airplane owners were held strictly liable for damage caused to people or property on the ground, although collisions between airplanes and injuries to passengers and crew were governed by the negligence standard. An old case, Guille v. Swan, 81 illustrates the economic basis of strict liability for ground damage. A balloonist descended into the plaintiff's garden, damaging the plaintiff's vegetables, and additional damage was caused when a crowd of onlookers broke into the garden to rescue the balloonist. He was held strictly liable for all the damage. As the court noted, he had no control over the balloon's horizontal motion and hence little ability to control its descent. Thus it was not a case where the accident could have been avoided by the exercise of due care, and there was no suggestion that the balloonist was careless. But the accident could have been prevented by his not ascending in the first place. The reason for the balloon ride is not disclosed in the opinion but the court seems to have regarded such riding, at least over New York City, as a rather frivolous activity. If the activity was not a valuable one, Council, Commission on Surface Mining and Reclamation, Surface Mining of Non-Coal Minerals: App. I, at pp. 2 7 - 2 8 (National Academy of Sciences 1980). 79. See Leet, note 78 above, at 1 0 0 - 3 . 80. See sources cited in note 78 above. 81. 19 Johns. (N.Y.) 381 (1822).

Basic Principles of Accident Law · 115 abandoning it may well have been the optimal method of accident avoidance, in which event strict liability would induce abandonment. It would not have been as efficient for the plaintiff to have switched from gardening to some land use less sensitive to damage from balloons and crowds of onlookers. Thus, a no-liability rule was not an attractive alternative to strict liability. The situation in the early days of the airplane was much the same as in the balloon case. Flying was a marginal activity, avoidance of damage to the subjacent property could not be assured simply by taking care, and the subjacent property owners could not protect themselves by altering their activity. As flying became both an established and a relatively safe activity, the likelihood that strict liability for ground damage would reduce the amount of flying fell. Hence the economic analyst is not surprised that strict liability for ground damage is giving way to negligence liability.82 The situation with regard to airplane collisions has always been different. The parties to a collision are in a symmetrical position insofar as reducing accidents through reducing activity is concerned; moreover, most collisions result from failures to exercise due care. In these circumstances the economic argument for strict liability is weak. Moreover, because each party is both an injurer and a victim, there is no obvious way to apply a strict liability standard. The analysis in this section can be generalized in the following economic hypothesis, to which the cases seem largely to conform. During the early stages of development of a new product or activity the legal system lacks sufficient experience to be able to determine whether the benefits of the product exceed its full costs, including costs to third parties (for example, property owners who suffer ground damage from airplane crashes). One way to generate such information is to hold the producer or user strictly liable for accidents to third parties resulting from the activity. Strict liability forces the innovator to internalize all the costs of his activity. If the activity still flourishes in spite of a strict liability standard, we can be confident that its benefits exceed its full costs or, equivalently, that eliminating or greatly reducing the new activity would not be optimal. At this point the argument in favor of strict liability weakens. Experience has demonstrated that the benefits of the activity exceed its full costs, and society is now being burdened with the greater 82. See Southern California Edison Co. v. Coleman, 150 Cal. App. 2d Supp. 829, 310 P.2d 504 (1957); Elwood v. Bolte, 403 A.2d 869 (N.H. 1979); 1 Lee S. Kreindler, Aviation Accident Law § 6.01[5] (rev. ed. 1980).

116 · The Economic Structure of Tort Law administrative costs associated with an increasing number of claims brought about by the growth of the activity. We would predict, therefore, a shift toward negligence and away from strict liability as a new industry or activity matures. Notice that this hypothesis is the reverse of the "infant industry" or "subsidy" arguments sometimes made for the use of the negligence standard in the nineteenth century. 83 This illustrates the power of the positive economic theory of tort law to generate hypotheses that are at once startling and testable. Two factors limit the scope of the hypothesis, however. First, strict liability may not provide information on the full costs of a new activity because an accident may result in a large number of claims each too small to support a lawsuit. If there is no feasible means of aggregating small claims, we are in effect in a world Of no liability and can offer no hypothesis on the relative advantages of different liability rules. Second, even at the early stages of a new activity, strict liability reduces the incentives of the potential victim to take care or alter his activity level in order to minimize risk. If the victim's care is an important component of due care in the early stages of an activity, as it was for airplane collisions and injuries to pedestrians from automobiles (because the cost of care to the pedestrian often is slight), strict liability would be inefficient, even initially, and we would predict that it would not be adopted as the liability rule. A number of activities have been classified as ultrahazardous, and hence subjected to strict liability, on an ad hoc basis. The considerations used by the courts seem to be those listed by the Restatement in its definition of ultrahazardous activities, a definition that, as we have seen, tracks the economic model closely. In Cities Service Co. v. State, 84 for example, the defendant, in connection with its phosphate rock mining, impounded billions of gallons of phosphatic slimes behind an earthen dam, which broke, causing substantial damage to the surrounding area. The court held the defendant strictly liable for the damage. Potential victims were apparently helpless to avoid or reduce the damage, whereas the defendant, although not careless, might have been induced by strict liability to reconsider its decision to build an earthen dam. Siegler v. Kuhlman85 illustrates a different aspect of the economics of negligence versus strict liability: information costs. The defendant's trailer tank broke free and spilled thousands of gallons of gasoline, which 83. These arguments are discussed in Richard A. Posner, Economic Analysis of Law § 8.2 (3d ed. 1986). 84. 312 So. 2d 799 (Fla. Dist. Ct. App. 1975). 85. 81 Wash. 2d 448, 502 P.2d 1181 (1972).

Basic Principles of Accident Law · 117 exploded, destroying the car driven by the plaintiff's decedent and killing the decedent. In holding that this was a proper case for strict liability, the court pointed out that w h e n large quantities of gasoline being transported by truck explode, the explosion is likely to destroy the evidence necessary to establish whether the accident was caused by negligence. The information costs of applying the negligence test prevented the use of that test in Siegler. Therefore, as between strict liability and no liability, the former was the preferable standard because it was more likely that the defendant could have prevented such an accident at reasonable cost than that the plaintiff's decedent could have done so. National Steel Service Center, Inc. v. Gibbons 86 is a similar case. The defendant, trustee of a bankrupt railroad, was sued for damages caused to the plaintiff's warehouse as a result of the explosion of four tank cars, filled with propane gas, in a derailment. As in Siegler, the explosion destroyed critical evidence, and this was one reason in favor of strict liability. In addition, the large concentration of explosive gas meant that D in our formal economic model was great, implying that activity-level changes might be cost justified if greater care was unavailing. Moreover, the plaintiff was helpless to avoid the accident. In concluding that the rule should be strict liability, the court relied explicitly on the economic theory of tort law. 87 Cases holding that the storage of gunpowder is not an ultrahazardous activity that subjects the owner of the storage facility to strict liability for damage caused by an explosion may seem inconsistent with the blasting cases. If blasting subjects the blaster to strict liability, why should not storage of the material used for blasting also give rise to strict liability? Tuckachinsky v. Lehigh & Co. 88 helps answer the question. The defendant was engaged in coal mining and kept several kegs of black powder near the mine shaft for use in blasting. A bolt of lightning hit the powder and the concussion from the explosion injured the plaintiff, w h o was standing in the doorway of her father's house nearby. The court stated: The evidence in this case shows that the powder magazine had been in use by the defendant company for more than 30 years, and that the plaintiff had resided within about 700 feet of it for some 16 years. Yet there is no testimony to show that any apprehension of danger or any fear of explosion was felt or expressed by any one during that time. No 86. 319 N . W . 2 d 269 (la. 1982). 87. See id. at 272-73, quoting Richard A. Posner, Economic Analysis of Law § 6.11, at pp. 140-41 (2d ed. 1977). 88. 199 Pa. 515, 49 Atl. 308 (1901).

118 · The Economic Structure of Tort Law objection to the location or maintenance of the magazine has been shown. The explosives were stored in small quantities, to meet current needs. Such materials are always dangerous, but, as their use is essential to the work of mining, it is impossible to protect absolutely persons or property in the immediate vicinity. The risk is similar to that arising from the operation of steam boilers and other machinery and apparatus necessary to the prosperity of great communities.89

The difference between Tuckachinsky and the blasting cases is that in Tuckachinsky the plaintiff's optimal activity level was an important variable. In general, people cannot easily relocate to avoid blasting, because blasting may accompany any new construction and new construction is ubiquitous, but they do not have to live 700 feet from a mine shaft, or, as in the Kleebauer case, near a plant that manufacturers high explosives. 90 Tuckachinsky was really a nuisance case. The issue was the optimal pattern of land development, and there was no presumption that the defendant's use of his land to store gunpowder was less appropriate than the plaintiff's use of nearby land as a residence. Of course, the court may have been wrong to think that the defendant could not have done more to prevent the explosion, but it seems to have been looking at the right things. Tuckachinsky may seem inconsistent with the cases that impose strict liability for blasting even when the blasting is done at a permanent site, such as a mine, rather than a temporary construction site. But there is no principle of strict liability for explosions as such; the strict liability is for the deliberate use of explosives in construction, mining, and other activities—and the explosion in Tuckachinsky was accidental. If accidental explosions are rare, ρ may be lower than where the defendant deliberately and routinely causes controlled explosions (which may, however, make D lower). It is on this theory that blasting is classified as an ultrahazardous activity. But the distinction is somewhat frayed in Tuckachinsky itself, where the defendant was engaged in blasting as part of its mining operation, although the particular explosion that injured the plaintiff was accidental. In Chapter 3 we explained that strict liability, to be an efficient rule of liability, requires—unlike negligence liability—a defense of contributory negligence. Nevertheless that defense is less securely a feature of strict liability law than of negligence law, although a related defense, assumption of risk (more fully discussed in the next chapter) is recog89. Id. at 518, 49 Atl. at 309. 90. See Kleebauer v. Western Fuse & Explosives Co., 138 Cal. 497, 71 Pac. 617 (1903).

Basic Principles of Accident Law · 119

nized. In a wild-animal case, for example, the plaintiff is barred from recovery only by "knowingly and unreasonably subjecting himself to the risk that a wild animal... will do harm to his person." 91 Under this rule, one who enters a lion's cage and is mauled by the lion is barred from recovering damages, but the question whether the cost of accident avoidance to the victim was less than the expected accident cost will not be examined beyond this; the Hand formula is not applied. Actually the inconsistency with the economic model in Chapter 3 is superficial. Ultrahazardous activities are by definition ones in which the expected accident cost is high, and often it will be higher than the cost of care to the victim. If your neighbor keeps a pet lion in his backyard, the expected accident cost may seem to justify extraordinary measures of self-protection. But probably the cost of these measures would exceed the cost to the lion's owner of avoiding an accident simply by substituting a less dangerous watch animal. In that event it would be a mistake to apply the Hand formula to the victim's conduct and conclude that he was contributorily negligent and should be barred from collecting damages because of his failure to take such measures. Recall from our earlier discussion of contributory negligence in this chapter that the danger that the defense might induce victims to take precautions that are excessive given the costs of accident avoidance by injurers is avoided, when the liability rule is negligence, by the principle that one is negligent (or contributorily negligent, as the case may be) only if one fails to take a precaution that would be cost justified even if the other party were careful. This principle cannot be applied where the standard is strict liability rather than negligence. If we ask, should the neighbor of the lion's owner have bought himself a suit of armor to protect himself from the lion, and try to answer by determining whether this would be an efficient precaution against hazards created by the nonnegligent possession of a lion, we shall have converted strict liability into negligence; we shall be asking whether the injurer was negligent in order to determine whether the victim was negligent—which is how, in fact, the courts proceed in negligence cases, as we have seen. Thus although in principle (as shown in Chapter 3) a defense of contributory negligence is necessary in order to make sure that strict liability does not require excessive precautions, or activity changes, by potential injurers, the means that 91. Restatement (Second) of Torts § 515(2) (1979); see, for example, Keyser v. Phillips Petroleum Co., 287 So. 2d 364 (Fla. App. 1973). The standard is similar in cases of strict liability for (other) ultrahazardous activities. See Restatement, above, §§ 523, 524(2). The defense of contributory negligence in (strict) products liability cases is discussed in Chapter 10.

120 ' The Economic Structure of Tort Law

the law uses to prevent the defense of contributory negligence from imposing excessive burdens on potential victims is not applicable when the standard is strict liability. What is needed is some other doctrine but one that, like contributory negligence, will shift liability to the victim when his costs of avoidance, whether through activity changes or greater care, are lower than those of the potential injurer. It is not so easy to devise one. Because one of the principal advantages of strict liability is that it economizes on information costs by avoiding an inquiry into the cost-justified measures that the injurer might have taken to avert the accident, a simple comparison of injurers' and victims' accident-avoidance costs for purposes of assessing a defense of contributory negligence would fundamentally alter the character of the strict liability standard. It seems that about the best the law can do is to excuse the injurer when the victim could have avoided the accident at a very low cost, probably therefore lower than the injurer's unmeasured avoidance cost. This seems in fact to be the law's approach. If the victim's cost of avoidance is extremely low (and possibly negative), as where the victim enters the lion's cage, the defense of "knowingly and unreasonably subjecting himself to the risk" comes into play—and achieves the efficient solution. Probably the most important principle of strict liability in tort law (other than in the products area, which is discussed in Chapter 10), at least as measured by the number of cases in which it is applied, is that of respondeat superior, which makes an employer strictly liable for the torts of his employees committed in furtherance of the employment. 92 This principle has sometimes been thought to be an example of the law's sympathy to "deep pocket" arguments. Employees often lack sufficient assets to pay a tort judgment, and respondeat superior allows the victim to reach into the employer's deep pocket. This is an unsatisfactory explanation. The principle of respondeat superior is not recent but dates back to a period in which the common law was not noted for sympathy toward accident victims. Moreover, the employer is strictly liable only for damages resulting from the negligence of his employees. A victim injured by an employee who is exercising due care has no claim against the employer. The economic explanation for respondeat superior focuses first on the 92. For other economic analyses of this doctrine see Lewis A. Kornhauser, "An Economic Analysis of the Choice between Enterprise and Personal Liability for Accidents," 70 Calif. L. Rev. 1345 (1982); Alan O. Sykes, "The Economics of Vicarious Liability/' 93 Yak L.J. 1231 (1984); Note, "An Efficiency Analysis of Vicarious Liability under the Law of Agency," 91 Yale L.J. 186 (1981). For the legal principles see Keeton et al., note 4 above, ch. 13.

Basic Principles of Accident Law · 121 complete helplessness of the accident victim to avoid incorrect employment decisions by exercising care or by altering his activity. This in itself would be an insufficient reason for imposing strict liability if the employer could always or at least most of the time prevent negligence by his employees simply by exercising care in his selection and supervision of them. But employees will sometimes be careless even if they are carefully screened and supervised, if only because their lack of ready assets reduces their financial incentives to take care. And there are a number of activity measures (as distinct from care measures) that an employer can take to reduce accident behavior by his employees, including delegating more work to independent contractors and giving employees simpler tasks requiring less care. However, the most important reason for respondeat superior is the fact already mentioned that employees often cannot pay a tort judgment against them. Our point is not a deep-pocket point; it is that the employer can use the threat of termination as a substitute for employee tort liability in inducing employees to act with due care, and that he will do so only if the employee's carelessness is a cost to him. Making the employer liable for his employee's tort serves to enlist the employer as a substitute enforcer of tort law where the primary enforcement mechanism, a tort action against the immediate tortfeasor, is unworkable. In view of these arguments for strict employer liability for the torts of employees (discussed further in Chapter 7), is there any justification for the common law's refusal to hold parents strictly liable for the torts of their children? One difference is that it is unlikely that imposing liability on parents would significantly affect people's choice of whom to marry or how many children to have; in other words, activity-level adjustments do not seem to be a promising method of reducing torts by childen. 93 Second, children are generally less dangerous than employees; this is, of course, just another reason for not expecting strict liability to induce many activity changes. The major (although now largely abandoned) exception to parental nonliability was consistent with these distinctions: the "family car" rule, 94 under which the owner of the car is liable for the torts of family members committed while they are driving it. Not only does a car make even (or perhaps especially) a child dangerous, but a simple activity change can reduce expected accident costs—not letting the child drive the car. 93. Parents who are negligent in supervising their children are liable for such negligence directly, and principles of vicarious liability do not come into play. 94. See Keeton et al., note 4 above, at 524-27.

222 · The Economic Structure of Tort Law Even a reader who agrees that the areas of strict liability in tort law make good economic sense may still object that we have not shown that the areas of tort law governed by negligence rather than strict liability are not equally or more appropriate for treatment under strict liability. We shall not attempt such a showing here beyond noting that the principal areas of tort law not governed by strict liability involve collisions and professional (chiefly medical) malpractice. Neither is an area suitable for strict liability. In collision cases due care typically requires both parties to exercise care, and there is no presumption that altering the injurer's activity level is a more efficient method of accident avoidance than altering the victim's activity level. Hence there is no allocative advantage to strict liability. Whether there are savings in administrative costs depends on whether the higher information costs of the negligence standard exceed the higher claims costs of the strict liability standard; as a first approximation, it seems best to treat these costs as offsetting. To treat every case in which medical treatment injured the patient as one of strict liability would entail much loss shifting for small allocative gains. Most medical procedures carry a risk of injuring the patient. Under a standard of strict liability every one of these procedures, however carefully administered, would give rise to a claim for damages if injury occurred. It also may be more difficult and therefore more costly to determine in a medical malpractice suit than in an ordinary accident case whether the victim's injury was caused by the injurer's behavior. This would make the information cost of a medical malpractice suit greater than the cost of an ordinary accident case governed by strict liability. Although proving causation would be required under both strict liability and negligence (and there are important economic reasons why this is so, as we shall see in Chapter 8), the fact that the claims cost of a medical malpractice case is great would tend to make considerations of administrative cost favor negligence over strict liability.

.

5

.

Additional Principles of Accident Law

IN THIS CHAPTER w e e x a m i n e s e v e r a l m o r e d o c t r i n e s o f t o r t a c c i d e n t l a w : t h e r e a s o n a b l e - m a n r u l e , 1 c u s t o m , a s s u m p t i o n o f risk, a n d d u t y . S e v e r a l i m p o r t a n t doctrines a r e not d i s c u s s e d systematically, notably p r o o f of n e g l i g e n c e ( i n c l u d i n g t h e d o c t r i n e o f res ipsa loquitur)

and the treatment

of safety s t a t u t e s in n e g l i g e n c e c a s e s , 2 a n d o t h e r s a r e d e f e r r e d t o later c h a p t e r s (especially C h a p t e r s 7, 8, a n d 10).

The Reasonable

Man

If t h e c o s t s t o t h e c o u r t s o f i n f o r m i n g t h e m s e l v e s a b o u t a n i n d i v i d u a l ' s ability to a v o i d a c c i d e n t s w e r e z e r o , t h e y w o u l d s e t a different d u e c a r e level for e a c h individual in e v e r y a c c i d e n t case. A potential injurer w h o w a s v e r y c l u m s y w o u l d h a v e a l o w y* ( w h e r e y* d e n o t e s t h e i n d i v i d u a l i n j u r e r ' s o p t i m a l c a r e l e v e l a n d y* d e n o t e s t h e r e a s o n a b l e - m a n s t a n d a r d set b y t h e c o u r t ) b e c a u s e his i n v e s t m e n t in c a r e w o u l d be

relatively

1. See W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 32 (5th ed. 1984). 2. The doctrine of res ipsa loquitur ("the thing speaks for itself") allows the trier of fact to infer negligence, without direct evidence, when the circumstances of the accident make an inference of negligence more likely than not. In the case of statutes setting standards of health or safety, most courts give either conclusive or presumptive weight to a violation of the statute as demonstrating negligence. This approach could be challenged on the ground that many statutes, including some safety and health statutes, ostensibly passed in the public interest are actually procured by special interest groups desiring to redistribute wealth to themselves and therefore provide a poor guide to optimal regulation, through the tort system, of safety and health. See Richard A. Posner, Tort Law: Cases and Economic Analysis 255-56 (1982), for a brief sketch of the argument. But considerations of separation of powers may naturally incline judges to defer to legislative judgments on matters of health and safety, even if the judges are skeptical of the wisdom of the enactments. We return to this point briefly in Chapter 8.

124 · The Economic Structure of Tort Law unproductive and his marginal cost of care would be relatively high; one who had exceptionally quick reflexes would be highly productive and his marginal cost would be low, so he would have a high y*. The same variation in costs of taking care is possible for potential victims of accidents. Failure to establish individual standards of care for negligence will produce misallocative effects, for two reasons: 3 1. For an individual whose marginal cost of care is relatively low, the standard of care set by the courts will be lower than his y*, in which event he will not take care even though it would reduce expected damages by more than it would cost. He has no incentive to use more than y* care (the reasonable-man standard) because to do so would increase his costs without reducing his expected liability, which, by assumption, is already zero. 2. An individual whose cost of care is slightly higher than average and therefore whose y* is only a little less than y* is likely to have an incentive to bring his level of care up to y*. But this too is inefficient, because the expected reduction in damages from the extra care (y* y*) will be less than the cost of that care. Why would an injurer incur a cost greater than the benefit in reducing expected damages? The answer points up an important distinction between strict liability and negligence. Under strict liability the injurer is liable in damages without regard to his level of care, so he will take additional care only if the reduction in expected damages exceeds the additional cost. Under negligence the injurer is liable for damages if his care falls below y*, but he will avoid all liability if he uses at least y*. Although he has no incentive to use more than y* even if y* > y* (point 1 above), he may bring himself up to the due care standard and use y* even though y* < y*. This will be his income-maximizing choice when the cost of the extra care (y* — y*) is less than p(y*)D, the expected liability payment he avoids by taking y* care (although, by assumption, the extra care costs more than the actual reduction in expected harm that it brings about). 4 3. To simplify the theoretical presentation we focus on the injurer's standard of care, but the analysis applies equally to victim care (χ). 4. Under strict liability, B's net income equals I s - p(0,y,)D - B(y,) where y, is B's level of care. Β would maximize his net income by using y * because that level, by definition, minimizes p(0,y,)D + B(y,). Under the reasonable-man standard of negligence law B's net income equals Is - p(0,y,)D - B(y) when y, y*. Suppose Β is a high cost of care individual, so that his individualized optimal care level, y*, is less than y*. Under the reasonable-man standard Β will use y* or y* depending on whether Is - B(y*) 3? /B - p(0,y*)D - B(y,*) or whether B(y') - B(y,*) 5 p(0,y*)O. Thus, the lower the cost of bringing oneself up to the due care standard—the smaller B(y*) B(y*)—and

Additional Principles of Accident Law · 125 Figure 5.1 is a frequency distribution of y* among a group of potential injurers. Area I denotes injurers who use their own optimum care level, y*. Each of them finds it too costly to bring himself up to y* relative to the benefits of avoiding liability for the victim's damages. Here negligence and strict liability are equivalent. Area II denotes injurers who use y* under a reasonable-man standard even though y* < y*. Area III denotes relatively low cost of care individuals, who use y* but would use y* > y* under strict liability or under an individualized negligence standard. Two further points. First, it is not clear whether more or fewer accidents will occur under a negligence rule with a reasonable-man standard than under strict liability.5 Although some individuals will reduce their care (area III), others will increase it (area II); it is not possible to say which effect will dominate without knowing more about the shape of the frequency distribution and the stringency of the reasonable-man standard. 6 Second, analysis becomes substantially more complicated if we allow for the possibility of victim care and if victims differ in their cost of taking care, as of course they do. Then neither strict liability nor negligence (both with defenses of contributory negligence) will be an individualized standard. Under the former, potential injurers will use Frequency

y' - reasonable

man

standard

Figure 5.1 the greater the expected damages, the more likely is an individual with a higher-thanaverage cost of care to take y* care. 5. Strict liability and an individualized negligence standard would generate identical levels of care and therefore the same number of accidents if changes in activity are ignored. 6. In Figure 5.1 as drawn, area III dominates area II, so more accidents would be expected to occur. If the reasonable-man standard were set higher or if the distribution changed so that more persons were in area II and fewer in area III, the number of accidents could be less than what would occur with strict liability.

126 · The Economic Structure of Tort Law individualized optimal care levels, whereas potential victims having low costs of care will lower their care to the reasonable-man level, and some potential victims having high costs of care will raise their care to the reasonable-man level. The allocative costs of forgoing individual standards of care are undeniable but must be compared with the costs of ascertaining each individual's due care level, an information cost. Information costs are real costs, and it may be efficient to give up the allocative gains of an individualized standard in exchange for a reduction in them. Moreover, the fact that a person may be incapable of a high level of care proves not that he cannot avoid an accident at reasonable cost but only that he cannot avoid it by being more careful. He may be able to avoid it by abandoning or reducing the activity that produces the accident. Although a twelve-year-old cannot attain a high level of care in driving an automobile (in economic terms, it is too costly for him to do so), the social costs of his not driving at all are low. Holding the child to the reasonable-man standard should induce his parents to prevent him from driving whereas holding him to an individualized negligence standard might not. Our analysis of the tradeoffs between individual and aggregate standards of care yields two predictions: 1. When the cost of determining the individual's due care level is low, a departure from that level is more likely to be allowed. This cost is presumably a function of how far the individual's optimal care level deviates from the standard level: the wider the gap, the more easily discoverable it will be by the methods of litigation. 2. The more easily substitutable a reduction in activity is for an increase in care, the less likely is the tort system to allow departures from a single, uniform due care level. Tort law conforms reasonably well to these predictions. With exceptions soon to be explained, the defendant in a negligence case (or the plaintiff, if the issue is contributory negligence) is judged by the standard of the reasonable man, which means the average man (or woman) placed in his situation. 7 An individual whose ability to take care is below average, perhaps because he has poor reflexes, is not excused on that account, 8 and an individual who is above average in his ability to take care—perhaps because of exceptionally good reflexes—generally is not held to a higher standard than an average person would be. 9 Not only 7. 8. 9. Law

See Keeton et al., note 1 above, at 173-93. See id. at 176-77. See Fredericks v. Castora, 241 Pa. Super. 211, 360 A.2d 696 (1976). But see American Institute, Restatement (Second) of Torts § 298, comment d (1965).

Additional Principles of Accident Law · 127 is it costly to measure individual reflexes and the like, and not only are changes in activity substitutable for greater care, but one type of care generally is substitutable for another type. A driver with slow reflexes can easily compensate by driving slower. He incurs a time cost but usually a modest one. There are substitutes even for good judgment. In Vaughan v. Menlove, 10 a farmer ignored warnings about the danger of spontaneous combustion in a hay rack, which caught fire and damaged the property of a neighbor, who sued him. His lawyer argued that the defendant had used his best, although poor, judgment, but the court held that the proper standard was that of the average man. The information cost of determining each injurer's intelligence and ability to make judgments of this sort would be too great to justify departing from the reasonableman standard. A related point is that if such departures were allowed, each injurer would have an incentive to spend resources to convince the court that he was conforming to his individualized but low level of care. Similar problems would arise for victims. All this would add to the cost of litigation. And had the farmer in Vaughan v. Menlove known that he would be liable for the consequences of his misjudgments, he might have listened more attentively to the advice that was given him rather than relying on his own judgment. In types of case where the information costs of departing from the average-man standard are low because the gap between the average individual's due care level and that of the individual defendant is large and palpable, the courts, as predicted, recognize a different standard. A blind man is not held to the same standard of care as a sighted one, provided that he communicates his blindness by carrying a cane; and so with other manifest physical disabilities. 11 And of course a doctor who falls below the standard of care of a reasonable doctor in his locality and specialty cannot defend by showing that he has the skills of a reasonable layman. One effect of these exceptions is to make the remainder of the community—that is, the group subject to the averageman rule—more homogeneous with regard to the capacity to take care, and hence to reduce both the allocative and administrative costs of applying the average-man standard to that group. One exception to the average-man rule that the law has generally refused to recognize is insanity. 12 Considerations both of information 10. 3 Bing. (N.C.) 468, 132 Eng. Rep. 490 (C.P. 1837). 11. See Keeton et al., note 1 above, at 175-76; Argo v. Goodstein, 438 Pa. 468, 265 A.2d 783 (1970). 12. See Keeton et al., note 1 above, at 1072-75; In re Meyer's Guardianship, 218 Wis. 381, 261 N.W. 211 (1935).

128 · The Economic Structure of Tort Law costs and of activity level support this result. Proof of insanity is difficult; even more difficult is establishing the relationship between insanity and care. And people whose insanity is severe enough to affect their ability to avoid physical injury to themselves and others are generally kept under restraint. They are highly dangerous—one might say ultrahazardous—and the same considerations that argue for strict liability for ultrahazardous activities argue for strict liability for the torts of the insane. 13 This illustrates the broader point, which should be borne in mind throughout this section of the chapter, that the choice between an individual and an aggregate standard of care for high cost of care persons (persons who are insane, clumsy, or otherwise incapable of taking care) is analytically similar to the choice between negligence and strict liability. Arguments that support strict liability in some contexts also support the choice of an aggregate rather than an individual standard of care, as in the common law's refusal to evaluate an insane man's level of care by a lower standard than that applied to a sane man. Indeed, we showed earlier in this section that the reasonable-man rule constitutes a pocket of strict liability (area I in Figure 5.1) in negligence law. The same interplay of information costs and activity considerations is found in the law's treatment of children. In general, they are not judged by the adult standard of care but by the standard of care of children of similar age and development to the defendant (or plaintiff, if the plaintiff is a child whose contributory negligence is at issue).14 But they are held to the adult standard of care when engaged in so-called adult activities, such as driving an automobile.15 Because children are obviously and grossly less capable of exercising care in a variety of activities, it would be inappropriate to include them with adults for the purpose of fixing an average level of care by which to judge both groups. The amalgamation would both dilute the adult standard of care, thereby raising the number of accidents caused by adults above the socially optimal level, and in effect impose strict liability on children for their torts. The second result would make sense only if the optimal care of children, as of the dangerously insane, involved constant restraint. As mentioned in the last chapter, it does not. The costs of preventing children from riding 13. The recognition of an insanity defense in criminal law is not inconsistent, for reasons explained in Richard A. Posner, "An Economic Theory of the Criminal Law," 85 Colum. L. Rev. 1193, 1225 (1985). 14. See Dickeson v. Baltimore & Ohio Chicago Terminal R.R., 42 111. 2d 103, 245 N.E.2d 762 (1969). 15. See Daniels v. Evans, 107 N.H. 407, 224 A.2d 63 (1966); Terre Haute First Nat'l Bank v. Stewart, 455 N.E.2d 362 (Ind. App. 1984).

Additional Principles of Accident Law · 129 bicycles, playing in school yards, and the like would exceed the benefits in reduced accidents. Indeed, these benefits are likely to be negligible, since all child-like activities give rise to some level of accidents. Thus strict liability would not have a substantial accident-reducing effect through its effect on activity. Another point that argues for a separate standard of care for children is that the administrative or information costs of distinguishing children from adults is trivial. If children are not to be held to the adult standard of care (or to a standard that averages together adult and child capabilities), what standard should they be held to? A single standard for all children would be inappropriate; it would ignore the enormous differences in optimal care levels between three-year-olds and sixteen-year-olds. But to set a different standard for each age would be inappropriate too, given the different rates at which children mature. The legal standard, which requires comparing the child plaintiff or defendant with children of similar age and development, seems as efficient as any alternatives that suggest themselves. The adult-activities exception to the special standard of care for children is explicable by reference to activity level. Adult activities are defined as such primarily with reference to their dangerousness. Automobiles, motorcycles, tractors, and the like are dangerous instrumentalities in the hands of people who because of youth are incapable of driving them carefully. An alternative to greater care, however, is less activity. That would be a costly alternative for many adults but, as already suggested, is much less so for children. The forgone benefits of not being allowed to drive an automobile at the age of twelve are small; for a forty-year-old they might be high. By holding the child to an adult standard of care, the law encourages children to substitute toward less dangerous activities when the savings from fewer accidents offsets the forgone benefits of changing activities. Put differently, the cost to the child of avoiding liability by bringing himself up to the reasonable-man standard is so great (i.e., the child is in area I in Figure 5.1) that he (or rather his parents) will change the activity rather than risk having to pay substantial damages in the adult activity. The analogy to strict liability for ultrahazardous activities should be evident. And even if children were not more dangerous drivers of automobiles, tractors, and motorcycles than clumsy adults are, the costs to them of avoiding accidents by curtailing their activity would still be much lower. This is an argument for strict liability of children but not of adults. The effect of the average-man rule in imposing strict liability on clumsy adults is mitigated by the fact that there may be care alternatives to sharp reflexes,

130 · The Economic Structure of Tort Law such as driving more slowly. A young child will lack the judgment necessary to utilize these alternatives. There is another major exception to the reasonable-man rule, and it is also consistent with our economic analysis. If a person suffers a sudden seizure—such as a heart attack, stroke, or epileptic fit—which causes an accident, his conduct will be judged by the standard of the average man having whatever medical condition led to the seizure rather than by the standards of the healthy. 16 Costs of information justify this departure from the aggregate standard, even though less activity is an alternative to greater care in such contexts. A person with a heart condition, epilepsy, or some other medical condition that may lead to a seizure could drive less or not at all (most of the cases in which a seizure has caused an accident have been automobile collision cases). But the probability of having a seizure with no warning while driving is small even for people with such conditions. And often the individual is unaware of the condition before the seizure occurs. For these reasons a reasonable-man approach, which would be equivalent to holding the injurer strictly liable, would probably have little allocative effect. It would not alter the injurer's care level (and any activity level change would be small). For he had planned to take due care (y*); it was just that the seizure so increased his cost of care as to drive his level of care far below y*. This individual falls in area I in Figure 5.1. In short, holding the individual to the reasonable-man standard (y*) would increase administrative costs (the costs of transferring wealth by means of the legal system) without creating offsetting allocative benefits. Breunig v. American Family Insurance Co. 17 illustrates the interplay between this principle and the principle that there is no exception for the insane to the average-man rule. While at the wheel of her car, Mrs. Breunig suffered a sudden delusion, which caused her to drive into a truck. The court held that she had been negligent because she had had warning of the deterioration of her mental health, but it also stated that if the delusion had come on without any warning she would not have been deemed negligent. The difference between Breunig and the usual insanity case is that when insanity comes on and causes injury without warning, there is no opportunity to avoid the injury by restraining the insane person. In Breunig, however, the plaintiff was in a position to alter her activity level before the accident occurred. With regard to people who are above average in their ability to take 16. See Lobert v. Pack, 337 Pa. 103, 106-7, 9 A.2d 365, 367 (1939). 17. 45 Wis. 2d 536, 173 N.W.2d 619 (1970).

Additional Principles of Accident Law · 131 care, the law again makes exceptions provided the differences are manifest and hence information costs small. A surgeon is held to the standard of care of surgeons, not the standard of care of the average person in the community who is not a surgeon. The professional standard of care is more extensively discussed under custom in the following section. Sometimes a departure from a uniform national standard of care is warranted by regional differences in the costs of care. The cost of removing ice and snow is lower in northern than in southern climates, because in northern climates it pays to invest more in snow and ice removal equipment and training. So one is not surprised to find that the legal standard of care in ice and snow accident cases has long varied on a regional basis, 18 much as with the locality rule in medical malpractice, which we will examine shortly.

C u s t o m , a n d a N o t e on Disclaimers of Liability Often a particular safety appliance or practice is so widespread in the industry that it can be regarded as customary. Thus, after a certain period in the development of railroading it became customary to equip trains with air brakes, cowcatchers, spark-arresting equipment, and so forth. When a failure to adhere to a customary safety practice causes an accident, this is powerful evidence of negligence and is so treated by the courts. 19 The adoption of a safety practice by most members of the industry shows that its cost is less than its expected benefit in accident avoidance; there is no reason for the industry to adopt the practice otherwise. Nor will a defendant be heard to argue that although other firms in the industry found the device cost justified, it did not, because its costs are higher than those of other firms. Suppose that through some mistake in design an automobile manufacturer produces and installs a brake assembly that is defective according to the industry standard, and by the time it discovers the mistake it has built the defective assembly into a hundred thousand cars and the cost of replacing all these brake assemblies would exceed the expected benefits in accident avoidance. If the manufacturer is allowed to defend against a negligence suit on such a ground, it will not be (adequately) penalized for its design mistake. The case would be harder if the defendant could show that because 18. See Richard A. Posner, " A Theory of Negligence," 1 J. Legal Stud. 29, 66 (1972). 19. See Warburton v. Ν. B. Thayer Co., 75 N.H. 592, 72 Atl. 826 (1909).

232 · The Economic Structure of Tort Law of a scarcity of resources beyond its control it was unable at reasonable cost to install as safe a brake assembly as that used by other manufacturers. We know of no such case but also can think of no legal obstacle (other than possibly the reasonable-man rule) to such a defendant's arguing that its failure to comply with industry custom was not negligent in the circumstances. A separate question would be whether it should have warned its customers that they were getting a less safe brake assembly than normal in the industry; if it should have and failed to do so, they might have a cause of action based on misrepresentation. 20 Notice, though, that if the warning led customers to switch to other manufacturers, who would therefore have to expand their output, those manufacturers might encounter the same resource scarcity and build the same less safe brake assembly. Perhaps in these circumstances the warning should not be required, and the manufacturer not condemned under the reasonable-man standard either. So far we have been discussing the "offensive" use of industry custom—its use by a plaintiff to show that the defendant had been negligent. What of its use by a defendant to argue that because he complied with the industry custom he cannot be held negligent? To answer this question we must reintroduce the fundamental distinction elaborated in Chapter 2 between cases in which transaction costs are low and cases in which they are prohibitively high. If transaction costs are low, an optimal allocation of resources to safety as to other activities will be achieved by negotiation regardless of the liability rule in force. In these circumstances whatever is customary is, at least prima facie, optimal. Hence if it is customary for manufacturers of rotary lawn mowers to install a certain kind of protective guard around the blade, there is a presumption that any manufacturer who installs such a guard is nonnegligent. If a higher standard of safety were optimal, it would pay the manufacturers to provide it, because the incremental price of the lawn mower would more than offset the incremental cost of additional safety. (As a first approximation, consumers will be willing to pay a higher price equal to the expected reduction in accident damages from a safer mower.) This conclusion assumes, however, that consumers have adequate information about the safety characteristics of the products they buy. If they do not, transaction costs may be high despite the fact that there are neither many parties to the transaction nor a problem of bilateral monopoly. 20. By analogy to such cases as Indian Towing Co. v. United States, 350 U.S. 61 (1955), and Erie R. Co. v. Stewart, 40 F.2d 855 (6th Cir. 1930).

Additional Principles of Accident Law · 133 The costs of transactions are prohibitive when the potential victims of accidents are numerous and unidentifiable third parties—travelers at railroad crossings, pedestrians, other drivers. Our automobile case was a mixed example, because brakes serve to protect both the customers of the automobile manufacturer (and are thus analogous to the protective guard in the rotary lawn mower) and third parties—pedestrians and drivers of other cars. Insofar as the benefits of a safety device or practice will enure to people with whom the potential injurer does not have an actual or potential contractual relationship, he will have no incentive, in the absence of law, to provide the benefits. There is no presumption in such a case that the customary level of care is efficient. We are led to predict that compliance with custom will not be a defense in accident cases where transaction costs are high but will be where those costs are low. The legal pattern approximates this pattern, but only very roughly. Custom is rejected as a defense in cases where transaction costs are high but was, traditionally at least, a defense in two of the three most important areas where accidents arise out of contractual relationships: industrial-accident cases (accidents to workers on the job) and professional (especially medical) malpractice cases. It was not a defense in the third area, products cases; but as we shall see in Chapter 10, there are good reasons for regarding that as an area of high transaction costs. However, the courts have never announced a general principle distinguishing the role of custom as a defense in low transaction cost cases from its role in cases of high transaction costs; and in fact the most famous case that rejects custom as a defense was a case of low transaction costs, The T. J. Hooper.21 Two coastal barges sank in a storm while en route from Norfolk to New York City. If the tugs that were towing the barges had been equipped with working radios, the tugs' captains would have learned of the storm and put into the Delaware breakwater, and probably the barges would not have sunk; other tugs in the area that were equipped with radios had pulled into breakwaters without loss. The owners of the barges brought suit against the owners of the tugs, who argued that it was not customary for tugboats to be equipped with radios. Judge Learned Hand rejected the defense of industry custom, stating that "reasonable prudence is in fact common prudence; but strictly it is never its measure; a whole calling may have unduly lagged in the adoption of new and available devices." 22 21. 60 F.2d 737 (2d Cir. 1932). 22. Id. at 740.

134 · The Economic Structure of Tort Law On the facts of The T. ]. Hooper case as they appear in Judge Hand's opinion, this statement is a rejection of the Coase theorem. But the opinion of the lower court in the case suggests that the case was decided correctly on its true facts. It was customary for coastal tugboats to be equipped with radios; 90 percent of such boats were equipped with them, and one of the tugboats involved in the case had a radio—it just was not in working order. In the words of the district judge, "the use of the radio was shown to be so extensive as to amount almost to a universal practice in the navigation of coastwise tugs along the coast." 23 Probably, therefore, the barge owners had negotiated with the defendants on the implicit assumption that the tugs would be equipped with (working) radios. 24 In saying there was no custom, 25 Hand may have been misled by the fact that the radio customarily was supplied by the captain of the tug rather than by the owner. He may have thought that "industry custom" implied that the members of the industry, rather than their employees, customarily supplied radios. But it is unimportant from an economic standpoint whether a particular input is supplied by the employer or the employee. Employees frequently supply their own tools and work clothes, in which event they receive (the Coase theorem implies) a higher wage than they would if the employers supplied them. Presumably, tugboat captains were paid more than otherwise because they customarily brought a radio on board with them. Hand also said that "an adequate receiving set suitable for a coastwise tug can now be got at small cost and is reasonably reliable if kept up; obviously it is a source of great protection to their t o w s . . . Such a set is the ears of the tug to catch the spoken word, just as the master's binoculars are her eyes to see a storm signal ashore." 26 In other words, the benefits of the radio greatly exceeded its costs, and therefore rational contracting parties would have contracted for it, and presumably these 23. 53 F.2d 107, 111 (S.D.N.Y. 1931). Judge Hand did not take issue with the district judge's findings. 24. For accidents that arise in a setting of low transaction costs and where the victim is a contracting party (even indirectly via another contract), the negotiated level of care will be the same as due care. Let V, and Vb denote the gains to each party from the performance of the contract before deducting expected accident costs. The joint net expected gains equal V = Vü + Vb - p(x,y) - A(x) - B(y). Obviously V is maximized by selecting x* and y *, the levels of χ and y that minimize the social costs of accidents. Because transaction costs are low, the parties will negotiate levels of care that maximize the overall or joint gain from the contract. Thus they will select x* and y*. 25. See 60 F.2d at 740. 26. Id. at 739-40.

Additional Principles of Accident Law · 135 parties were rational. This conclusion suggests an alternative basis to industry custom for holding that the loss of the barges was the legal responsibility of the towing company. Consistently with this approach, in an opinion written just three months before The T. }. Hooper Hand had refused to hold a shipowner at fault for having only two ventilators on board: "Shippers are not entitled to the latest design; ships, well built in their time, may still carry cargo unless they become so clearly out of fashion as to be an anachronism. The shipper bargains for no more than usual carriage." 2 7 The assumption seems to have been that it would not have been feasible to add a ventilator, that the ship would have had to be scrapped—implying a high cost of increased safety compared with the cost of having a working radio on board. The weakness of Hand's analysis in The T. J. Hooper remains, however, that if the cost of a safety device beneficial to the customer is much less than the benefit, it is impossible to understand why the device is not customary in the industry—as in fact it appears to have been. An examination of the subsequent citation history of The T. J. Hooper (one of the most heavily cited cases in the tort field, we believe) may cast some light on the issue of custom as a defense in cases where transaction costs are low. We have read every tenth case that cites Judge Hand's opinion, fifteen cases in all. 28 The results are surprising. In more 27. Spang Chalfant & Co. v. Dimon S.S. Corp., 57 F.2d 965, 967 (2d Cir. 1932) (citation omitted). 28. The cases are Salem v. United States Lines Co., 370 U.S. 31, 37 (1962) (issue was whether expert testimony was necessary on safety—no issue of custom as defense); Doe v. District of Columbia, 701 F.2d 948, 964 (D.C. Cir. 1983) (separate opinion) (prisoner tort case); Kane v. Branch Motor Express Co., 290 F.2d 503, 507 (2d Cir. 1961) (another stranger—noncontractual relationship—case); The Doyle, 105 F.2d 113, 117 (3d Cir. 1939) (The T. J. Hooper cited for unrelated proposition); SEC v. Crofters, Inc., 351 F. Supp. 236, 253 (S.D. Ohio 1972) (securities case); Hall v. Ε. I. Du Pont de Nemours & Co., 345 F. Supp. 353, 368 (E.D.N. Y. 1972) (The T. J. Hooper cited for proposition that "entire industries have been held to be below the standard of reasonable care"—no issue of custom as defense); Barrett v. Foster Grant Co., 321 F. Sup. 784, 792 (D.N.H. 1970), rev'd on other grounds, 450 F.2d 1146 (1st Cir. 1971) (defense of custom rejected, employee suing employer's supplier); Silver v. American Export Isbrandtsen Lines, Inc., 310 F. Supp. 681, 684 (E.D. Va. 1970) (controlling weight given custom in maritime personal-injury case); Farrell Lines, Inc. v. S/S Birkenstein, 207 F. Supp. 500, 510 (S.D.N.Y. 1962) (dictum—no issue of custom as defense); Conners-Standard Marine Corp. v. Marine Fuel Transfer Corp., 135 F. Supp. 365, 370 (E.D.N.Y. 1955) (The T. ]. Hooper cited on unrelated point); Gyerman v. United States Lines Co., 7 Cal.3d 488, 501, 498 P.2d 1043, 1051, 102 Cal. Rptr. 795, 803 (1972) (custom relevant but not controlling on contributory negligence); Marietta v. Cliffs Ridge, Inc., 385 Mich. 364, 370,189 N. W.2d 208,219 (1971) (defendant was arguing against, not for, relevance of custom); Nesbitt v. Community Health of South Dade, Inc., 467 So. 2d 711, 714 (Fla. App. 1985) (square rejection of custom as defense, in medical

136 · The Economic Structure of Tort Law than half the cases (eight) the issue of custom as a defense was not presented, and in another the case was not a tort case at all but an SEC enforcement action (Crofters). Of the remaining seven cases, only three squarely reject a defense of custom in a setting of low transaction costs— a recent medical malpractice case (Nesbitt), a case in which a customer was suing a department store (Shafer), and a products liability case (Barrett) in which an employee was suing his employer's supplier, so that there was an indirect contractual relationship. Two cases, Silver and Tonsic, seem actually to have accepted a defense of custom. Perhaps The T. J. Hooper is not quite so mischievous a precedent as one might have feared, when the actual use made of Hand's opinion by later courts is examined. The issue of custom as a defense raises the larger question of the interplay between tort and contract principles. If an accident occurs during performance of a contract as in The T. J. Hooper, why not treat the case as one of breach of contract, where the question would be whether one of the parties had failed to use the agreed-upon level of care? Why bring tort law into the picture at all? The answer may be that in the course of dealing with a myriad of accident cases, many of them in settings of high transaction costs (if most accidents arose out of contracts, perhaps tort law would have developed as a specialized branch of contract law), the courts have evolved a set of rules relating to accident cases that are also useful in deciding accident cases in contract settings— rules relating to damages for personal injury and property damage, victim's conduct, causation, and the like. Where an accident arises from the performance of a contract, the accident features of the case may dominate the contract features. Moreover, an important part of contract law consists of supplying standard provisions that are read into contracts in the absence of specific language to the contrary.29 Such provisions economize on the costs of transactions. Most parties to contracts want the performing party to exercise due care, because that is the level of care that maximizes the joint gain to the parties from the contract.30 So if the courts read the principles of tort law—principles based on economic considerations—into contracts, they give the parties what the parties would have bargained for expressly if they had not relied on the courts to supply terms. malpractice case); Tonsic v. Wagner, 220 Pa. Super. 468, 481, 289 A.2d 138, 144 (1972) (The T. }. Hooper cited by dissent); Shafer v. Η. B. Thomas Co., 53 N.J. Supr. 19, 23, 146 A.2d 483, 485 (App. Div. 1958) (square rejection of custom as defense, in customer case). 29. See Richard A. Posner, Economic Analysis of Law, ch. 4 (3d ed. 1986). 30. See note 24 above.

Additional Principles of Accident Law · 137 Until the tort law of industrial accidents was displaced by workmen's compensation in most industries and modified by employers' liability laws in the remaining few, the employer was obligated only to provide his employee with the safety appliances and other care inputs customary in the industry. 31 The explicit and continuing contractual relationship between employer and employee created a strong presumption that industry custom was optimal. Similarly, the traditional—and still the majority—rule in medical malpractice is that the standard of care is the customary standard in the defendant's locality.32 The law could take the position that although the standard of care in medical malpractice is the custom of the medical-care industry, in determining what that custom is the entire nation should be treated as one locality, and the defendant should thus be held to the national average level of care for his branch of practice. If, however, the standard of medical care varies widely and demonstrably across localities, a nationwide average standard will impose strict liability on those physicians who are practicing in the rural areas and small towns where the customary standard of care is below the national average. The local standard in these areas is below the national average because patients are unwilling to pay for a higher standard of care—not only because incomes are lower in rural areas but, more importantly, because it would be very costly to reach the big-city standard in the absence of large teaching hospitals, concentrations of specialists, and other medical advantages of large cities. Strict liability for mishaps caused by departure from a national standard would ha ν 2 no beneficial allocative effects in this instance but would merely dilute the standard of care applicable to practitioners in large cities, because by definition the national average standard of care is lower than the standard in the largest medical centers. Strict liability is easily avoided simply by making the standard a local rather than a national one. This is an especially attractive solution because the information costs of determining what the local standard is are lower than those of determining what the national average standard is. 33 Perhaps, given improvements in communication, the standard of care 31. See Titus v. Bradford, Β. & K. R. Co., 136 Pa. 618, 20 Atl. 517 (1890). 32. See Gambill v. Stroud, 258 Ark. 766, 531 S.W. 2d 945 (1976); Hemingway v. Ochsner Clinic, 608 F.2d 1040 (5th Cir. 1979); Northern Trust Co. v. Skokie Valley Community Hospital, 81 111. App. 3d 1110, 401 N.E.2d 1246 (1980); Annot., Modern Status of "Locality Rule" in Malpractice Action against Physician Who Is Not a Specialist, 99 A.L.R.3d 1133 (1980). There is an exception for nationally certified medical specialists, see Robbins v. Footer, 553 F.2d 123 (D.C. Cir. 1977); Roberts v. Tardif, 417 A.2d 444 (Me. 1980), which is consistent with the discussion in the text. 33. Compare discussion in Chapter 8 of the "eggshell skull" rule of tort damages.

138 · The Economic Structure of Tort Law in medicine is increasingly a uniform national one. To the extent that this is so it would provide an economic justification for the trend toward rejection of the locality rule. 34 But it would not justify decisions such as Helling v. Carey, 35 which reject conformity even to the national standard as a defense in medical malpractice cases. 36 In Helling the defendant physician had not given the plaintiff the pressure test for glaucoma. As a result of this omission the plaintiff's glaucoma went undetected, and he suffered a serious and irreversible loss of eyesight. The evidence showed on the one hand that the pressure test is cheap and harmless and on the other that it is not customarily (or was not, at the time the case was decided) administered to patients under forty years of age. If the test was as cheap and efficacious as the court thought, it is puzzling why the physicians in the community did not administer it routinely to younger patients. Because due care is by definition the cost-effective level of safety, physicians could have charged a price to the patient equal to or above the cost of the test. It seems unlikely that physicians were conspiring to withhold the test so that their patients would suffer irreversible eye damage and require increased services from them. Helling v. Carey, like The T. J. Hooper, may simply have got the facts wrong. Years before the decision medical texts were strongly recommending that ophthalmologists administer the pressure test routinely to anyone old enough to cooperate. 37 Hence there may not have been a market failure, and the decision may have reached the right result. But the rule it announces is economically unsound—and it is therefore not surprising that it has been rejected by most courts. 38 If, as we have argued, the customary standard of care is presumptively the efficient standard in situations where the costs of a voluntary transaction between the potential injurer and the potential victim are low, it might seem to follow even more clearly that the law would enforce any explicit contractual specification of the standard of care, and thus that it would enforce disclaimers of liability in a contract of sale or other contract. But as we shall see in Chapter 10, disclaimers of liability for negligent personal injury are not enforced. Since, by definition, negligence is the failure to take a precaution that would cost less than its expected benefits, one might think that no rational buyer would ever knowingly consent to a waiver of the seller's liability for negligence; but 34. 35. 36. 37. 38.

Illustrated by Morrison v. MacNamara, 407 A.2d 555 (D.C. App. 1979). 83 Wash. 2d 514, 519 P.2d 981 (1974). See also the Nesbitt case cited in note 28 above. See references cited in Posner, note 1 above, at 288-89. See, for example, Barton v. Owen, 71 Cal. App. 3d 484, 139 Cal. Rptr. 494 (1977).

Additional Principles of Accident Law · 139 we are about to see, in analyzing the defense of assumption of risk, that this point is an oversimplification. In Chapter 10 we will argue that there is nonetheless a reason for refusing to enforce such disclaimers in products liability cases: the probability of a serious accident caused by a defective product is ordinarily so low that it would not pay the consumer to read disclaimers carefully or make product choices on the basis of them. Once transaction costs are seen to include information costs as well as the free-rider and holdout costs that analysts of transaction costs usually focus on, it becomes apparent that negotiations regarding safety may be prohibitively costly even if the parties have a contractual relationship. But this analysis seems not to hold in the medical setting. The risk of untoward consequences is high with respect to many medical procedures, so it is hardly likely that a patient would discard a disclaimer of malpractice liability unread on the ground that the chance of anything going wrong was trivial. The refusal nonetheless to enforce such disclaimers is typically explained on the ground that a person is likely to be too emotionally involved in his illness to be able to make a rational transaction with respect to medical care. But to economists who analyze love, marriage, crime, and accidents from the standpoint of rational choice, such an explanation is not very persuasive, especially since many medical procedures are optional and many medical choices are made by people who are not gravely ill. The law's refusal to enforce disclaimers of liability for medical malpractice (where negligence is not involved) is not explained by our theory of tort law.

Assumption of Risk The examples discussed in the preceding part where tort law rejects (or purports to reject, because in both The T. ]. Hooper and Helling v. Carey the outcome may have been consistent with the underlying economics) the level of safety determined by a free market operating without significant externalities are less significant than one may think. Another doctrine provides a defense to a tort action in a setting of low transaction costs: assumption of risk. A plaintiff who explicitly or implicitly consents to a hazardous activity is barred from complaining if the hazard results in an injury to him. In McLeod Store v. Vinson39 the defendant department store adver39. 213 Ky. 667, 281 S.W. 799 (1926).

140 · The Economic Structure of Tort Law tised that it would conduct a guinea race outside the store and offered a prize for the winner. The plaintiff entered the race and was injured when he fell chasing the guinea. The court held that he could not complain of the hazards of the race; he had voluntarily assumed them. The decision is correct from an economic standpoint. For the plaintiff to have entered the race, the expected value of the prize must have exceeded his expected accident costs, so to dismiss his tort action was in effect to enforce his contract with the defendant. What may seem puzzling about assumption of risk from an economic standpoint is why it is viewed as a defense to a prima facie case of negligence. Where was the negligence in Vinson? And if there was negligence, why would any reasonable person voluntarily assume the risk of being injured by it, when, interpreted in economic terms, negligence implies that both injurer and victim would have been better off ex ante if the injurer had taken greater care? There are several answers to these questions. 1. Even though assumption of risk is technically a defense (the defendant has to plead and prove it), its real significance may simply be in expressing a preference for market compared to judicial evaluations. A judge or jury might think the department store negligent but the fact that the plaintiff was willing to assume the risk of an accident in exchange for the chance of winning the prize shows that it was not. 2. The defense offers a means by which a risk-preferring plaintiff can market his taste for risk. Suppose the expected cost of the accident to the plaintiff in Vinson was greater than the expected value of the guinea race, but he liked risk and so the uncertainty associated both with the risk of loss and the chance of gain was attractive to him. Although this may well be a factor in some cases and is fully compatible with positive economic analysis of tort law, we shall not stress it, because it departs from our economic model, which assumes risk neutrality. 3. The defense may be a means of rectifying some of the misallocative consequences of applying the reasonable-man rule. Suppose that driving a certain type of tractor without a rollover bar would be negligent for the average person because the expected accident costs would exceed the costs of the bar. But A is a driver of above-average skill for whom the costs of the bar exceed the expected accident costs because the probability of an accident to him is very low. Nevertheless an accident occurs and he sues B, the manufacturer of the tractor. There is negligence under the average-man standard and A may not be contributorily negligent because there may have been no reasonable safety measure that he could have adopted to avoid the accident. Nevertheless his action for negli-

Additional Principles of Accident Law · 141 gence is properly barred by the assumption of risk defense, because, given his particular circumstances, the manufacturer's failure to install the bar was not negligent. 40 4. The defense illustrates once again the law's sensitivity to the possibility of avoiding some accidents by less activity rather than more care. In Vinson the plaintiff may have run after the guinea as carefully as he could and his fall may not have been due to any lack of care; nonetheless he could easily have avoided the accident simply by staying out of the race. Efficiency requires that people in his position consider the possibility of avoiding an accident by reducing the level of an activity or abandoning it altogether; the assumption of risk defense forces them to do so. Of course, the same end could be achieved by defining contributory negligence broadly to include a failure to reduce or abandon an activity as well as a failure to take due care, and sometimes this is done; recall our discussion of the Liming case in the previous chapter. We have been discussing the assumption of risk defense in cases, broadly contractual, where the costs of transactions between potential injurer and potential victim are low. Most cases are of this type. They involve either workers injured on the job or participants in or spectators at sporting events. Should the doctrine operate in settings of high transaction costs? If, for example, A is run down by a speeding car driven by B, should Β be permitted to defend against A's action for negligence on the ground that A assumed the risk of being hit by a speeding car because he could have stayed home that day? The economic answer, which is also the legal answer, is no. It would be yes if the optimal rule of liability were no liability, because the expanded notion of assumption of risk would be one route by which such a result could be reached. But once it is decided that no liability is suboptimal, use of assumption of risk would simply undo the decision. So is there no place at all for an assumption of risk defense in settings where the parties are strangers (noncontract, high transaction cost settings)? Not quite. Economic analysis predicts that the defense would be recognized in such cases where the costs to the plaintiff of avoiding the accident by reducing his activity were palpably smaller than the defendant's costs of avoidance. Suppose B, through negligence, starts a fire that spreads to A's house. A runs into the burning house to retrieve 40. A qualification to this analysis is discussed in Chapter 10. Notice also that if A were injured because of negligence on the part of the manufacturer unrelated to the absence of a rollover bar (for example, if the tractor's steering failed), assumption of risk would not apply. Nor should it; this would be an example where the manufacturer took less than the implicitly negotiated level of care.

142 · The Economic Structure of Tort Law his hat, which has little value. The risk of injury to A is great, he is injured, and he sues B. A may have entered the house as carefully as he or anyone could have done, so that contributory negligence would not be a defense if negligence were defined solely with reference to care rather than activity level. But by not entering the house—that is, by changing his activity—A could have avoided the accident at a cost (the loss of the hat) that may be trivial relative to the expected accident cost and also lower than the cost to Β of preventing the fire in the first place. In these circumstances we want A to bear the risk of loss, by analogy to the last clear chance cases, and the law does impose the loss on A, as we saw in the last chapter in connection with the Liming case.

The Duty Limitation; Herein of the Good Samaritan The Limitation in General The standard definition of negligence in law is the failure to comply with the duty of care. The concept of duty is sometimes interjected by defendants in an attempt to avoid the consequences of their negligence. For example, in Moch Co. v. Rensselaer Water Co.41 the defendant had a contract with the city of Rensselaer to supply water at a certain pressure to the homes in the city. The defendant broke its contract through negligence, and as a result the plaintiff's warehouse burned down when the fire department could not draw water from the main because the pressure had failed. The defendant successfully defended against the plaintiff's suit for damages on the ground that it owed no duty to the plaintiff; its contractual duty was to the city. Although the duty concept is not helpful, the decision seems to make economic sense. Bear in mind that contract liability is generally strict liability; excuses for nonperformance of a contract are few. In part this reflects the fact mentioned in a previous chapter that contracts often provide insurance, so that breach of contract may simply be the occurrence of the uncertain event against which the promisor insured the promisee. The contract price will depend on the amount of insurance provided. The water company may have been willing to insure the city against the consequences of a fall in water pressure that the company might not be able to prevent, without necessarily being willing also to insure every owner of property in the community. So unless the parties had intended to provide insurance to nonparties, those nonparties could not sue on the 41. 247 N.Y. 160, 159 N.E. 896 (1928).

Additional Principles of Accident Law · 143 water company's breach of contract. And the fact that the company had broken the contract did not show that it had been negligent. Other objections to liability are the difficulty the company would have had in determining in advance the potential extent of liability (a problem to which we return in Chapter 8) and the fact that the plaintiff and other affected property owners may have been able to prevent or insure against the consequences of fire at a lower cost than the water company. As we shall see in Chapter 10, the no-duty defense was once also important in products liability cases, where it was long the view that the manufacturer of a defective product had no tort duty except to his immediate buyer (the "privity of contract" requirement). Today, however, the principal example of the no-duty principle is the rule that there is no duty to help someone in distress, or, as it is usually said, no duty to be a Good Samaritan. The Economics of Liability for Failing to Be a Good Samaritan The common law refuses to impose liability for failure to assist a stranger in distress no matter how low the costs of assistance would be or how great its benefits. 42 It has been argued that the absence of a liability rule is inefficient and contradicts the positive economic theory of the common law. 43 Suppose that people who failed to make cost-justified rescue attempts were made liable for the consequences of their failure—that is, if the rescuer's inputs are less than y* (the quantity that would minimize the expected losses from the hazard), he is liable for D. The potential rescuer will face a choice between (1) a legally imposed penalty of D if the victim (as we shall call the person in distress, but bear in mind that ordinarily he is not the rescuer's victim) is injured, plus rescue expenditures of C(y0), where O s y 0 < y*; (2) no liability if the victim is injured, but rescue costs of C(y*). Assuming that the liability rule is administered without error, one can easily show, by analogy to our formal model of negligence in Chapter 3, that the second alternative is wealth maximizing. But this overlooks the possibility that potential rescuers will avoid liability not by rescuing but by avoiding activities that give rise to opportunities for rescuing. Because liability is equivalent to a tax on these activities (pro42. See Hurley v. Eddingfield, 156 Ind. 416, 59 Ν.Ε. 1058 (1901); Yania v. Bigan, 397 Pa. 316, 155 A.2d 343 (1959); The Good Samaritan and the Law (James M. Ratcliffe ed. 1966); Jay Silver, "The Duty to Rescue: A Reexamination and Proposal," 26 Wm. & Mary L. Rev. 423 (1985). 43. See Richard A. Epstein, " A Theory of Strict Liability/' 2 J. Legal Stud. 151, 189 (1973).

144 · The Economic Structure of Tort Law vided the inputs necessary to avoid liability, y*, exceed what the rescuer would supply in the absence of a liability rule), liability will induce substitution away from the taxed activity. Assume that activity Η is hazardous, thus creating rescue opportunities, and that the number of victims and rescuers in Η is ν (initially assumed to be fixed at v0) and nh, respectively. There are no victims in the alternative safe activity, S, and therefore no rescue opportunities for the ns persons entering S. 44 Although nh represents the set of rescuers, not all of them will encounter a victim and be called on to rescue. The probability of a person's encountering a victim in Η is inversely related to the ratio of rescuers to victims, as in θ = Q(nh/v) where θ' < 0. (Θ' denotes the derivative of θ with respect to nh/v.) That is, the greater the number of rescuers in Η relative to victims, the less likely it is that any particular rescuer will encounter a victim and be called on to rescue him. A potential rescuer will choose to enter Η or S depending on whether Uh-U°$AUh,

(5.1)

where Uh equals his preliability utility in H, Us his utility in S, and ΔIIh the reduction in his utility in Η because of the liability rule. It follows from Eq. (5.1) that for a given distribution of ΔΙΙΗ among potential rescuers, the greater the substitutability between activities Η and S is, the smaller will be the average difference between Uh and IP and the greater the proportion of potential rescuers who will switch from Η to S in response to a threat of liability. Similarly, for given substitutability between the two activities, the greater is the value of AUh, on average, the greater will be the shift from Η to S. The term ΔΙΙΗ will tend to be greater, the likelier a potential rescuer is to encounter a victim (that is, the greater θ is) and hence the likelier he is to be held liable for failure to rescue, and the greater the difference between the quantity of inputs necessary to avoid liability (y*) and the quantity that would be supplied in the absence of a liability rule (y0).45 Moreover, the disutility of the liability tax to each person remaining in activity Η will tend to be greater, the smaller the number remaining (holding v° constant), because the like44. We assume for the sake of simplicity that an individual is either a nonvictim (potential rescuer) or a victim but not both. 45. Observe that y" will be positive in the absence of liability rules if potential rescuers are altruistic toward victims. Altruism may be triggered, for example, by observing a victim in danger, which in turn makes the rescuer's utility depend, in part, on the victim's utility or income. We discussed the economics of altruistic rescue in "Salvors, Finders, Good Samaritans, and Other Rescuers: An Economic Study of Law and Altruism," 7 J. Legal Stud. 83 (1978).

Additional Principles of Accident Law · 145 lihood will be greater that those remaining will encounter a victim and therefore incur a legal duty to attempt a rescue (Θ' < 0). Merely to note that liability will have substitution effects does not reveal the net effect of liability on the probability of rescue. That depends on both the probability that the victim will be found (as potential rescuers substitute away from hazardous activities) and the probability that once found he will be rescued (defined as pr). Assume that the number of victims found (F) is a function of both the number of potential victims and the number of potential rescuers in H, as in F = F(v,n%

(5.2)

where F is assumed to be homogeneous of degree one (there are no economies of scale) with positive first and negative second derivatives (increases in ν or nh raise F, but at a decreasing rate). 46 The probability that a given victim will be found (which is distinct from Θ, the probability that a given potential rescuer will find or locate a victim) is δ = δ («Vν),

(5.3)

which is a positive function of the ratio of nh to v. In words, the probability that a given victim will be found will rise with the ratio of potential rescuers to potential victims. Thus a decline in the number of potential rescuers in response to the liability rule will reduce the probability that a victim will be found (holding constant the number of victims) but, by reducing the ratio of potential rescuers to victims, will raise Θ, the probability that a given potential rescuer will find a victim.47 Because the unconditional probability of rescue equals hp' (the probability that a given rescuer will find a victim times the probability that the victim, if found, will be rescued), the net effect of liability for failure to rescue is unclear. On the one hand, given that the victim is found, the conditional probability of rescue (pr) will on average rise; on the other hand, the tax imposed by the liability rule will induce some potential rescuers to substitute away from the hazardous activity, reducing the likelihood of finding the victim. We cannot tell a priori which of the two effects will dominate. If the substitution effect dominates, the actual effect of liability will be contrary to its intended effect of increasing the number of rescuers. 46. From the assumption of homogeneity of degree one, Eq. (5.2) can be rewritten as F = vb(nh/v), where δ equals the probability that a victim will be found. Because δF/bv = δ - (nH/v) δ' > 0 and δΡ/δη' = δ' > 0, δ' ( > 0 ) is a positive function of the ratio of nk to v. 47. θ is written as θ = (« ι /ι>)-·δ and δθ/δ(η*/ι>) = - (n*/z>)-2[δ - (nVj>)ö'] < 0 from the assumption that δΡ/δΐ) > 0.

146 · The Economic Structure of Tort Law The relative importance of the offsetting effects on δ and pr has implications for the number of potential victims (previously assumed to be constant) of the hazardous activities. If the effect on \f dominates and the unconditional probability of rescue rises, not only will victims substitute away from their own inputs of safety48 but the number of potential victims in Η will increase as hazardous activities become more attractive. This will generate increases in θ and further substitution away from hazardous activities by potential rescuers. On balance, the equilibrium value of the unconditional probability will increase; otherwise additional victims would not have entered. Alternatively, if the negative effect on δ dominates, victims will substitute toward their own safety, and the supply of victims in the hazardous activity will fall. A further consideration is the cost of administering a liability rule. Legal-error costs in particular might be high because of the difficulty in many settings (for example, on a crowded beach) of identifying potential rescuers. If as a result only a fraction of those who failed to attempt a rescue were held liable, this would reduce the tax effect of a liability rule and so reduce the substitution away from hazardous activities, but it would also reduce the incentive to rescue created by liability. The Legal Position Although the preceding analysis does not prove that the common law's refusal to impose liability for failure to rescue is efficient, it prevents one from concluding that the absence of such a rule necessarily is inefficient. It also suggests that the growing judicial support for an exception in cases where the potential rescuer caused, even if nontortiously, the emergency may be efficient (no stronger statement is possible).49 By identifying a best-placed rescuer, causation reduces the legal-error costs of imposing Good Samaritan duties. Tort writers have noted other exceptions to the no-duty-to-rescue doctrine:50 48. A more complicated liability rule would take account of the victim's inputs of safety and might impose liability for failing to rescue provided the victim took due care. This would eliminate the substitution between one's own safety inputs and the safety inputs of rescuers. 49. The leading case is Montgomery v. National Convoy & Trucking Co., 186 S.C. 167, 195 S.E. 247 (1938). For more recent cases see Zylka v. Leikvoll, 274 Minn. 435,144 N.W.2d 358 (1966); Scatena v. Pittsburgh & New England Trucking, 2 Mass. App. 683, 319 N.E.2d 730 (1974). The last clear chance rule discussed in Chapter 4 could be viewed as an aspect of this principle. 50. See Keeton et al., note 1 above, at 375-85.

Additional Principles of Accident Law · 147 1. A railroad has a duty to assist a passenger who becomes ill on a train. 2. A bar cannot allow an obviously drunk customer to wander off at closing time in circumstances where he is likely to get hurt, even though the bar was not at fault in the customer's getting drunk in the first place. 3. An employer must assist an injured employee at the work site. 4. If a hospital emergency room customarily treats anyone who seeks admittance, it cannot turn away someone obviously in need of immediate medical assistance. All these cases involve an actual or potential contractual relationship between the rescuer and the victim. The issue is the implied terms of the contract. For example, a reasonable implication of the contract of carriage between railroad and passenger is that the former will render assistance to an obviously ill and disabled passenger. The benefits are apparent, whereas the tax effect of a liability rule is absent because the railroad can recoup the cost of assisting the passenger in a higher ticket price—which passengers will gladly pay. Even the emergency-room example fits this mold because the custom of treating all comers invites potentially costly reliance by the ill or injured person who requires immediate assistance. The point to be stressed is that in a contractual setting there need be no concern with the tax effect of liability for nonrescue, because the potential rescuer wants to do business with the victim. The only possible objection to tort liability in these cases is that, in some of them at least, breach of contract might be the more appropriate basis of liability than tort.51 51. Thus, in Randolph's Adm'r v. Snyder, 139 Ky. 159, 129 S.W. 562 (1910), the court held that a suit alleging that the defendant physician had refused to treat the plaintiff's child when sent for, although he had agreed to treat the plaintiff's family by the year in consideration of monthly payments made to him, should have been brought as a contract suit rather than a tort suit. Contrast Hurley v. Eddingfield, 156 Ind. 416, 59 Ν.Ε. 1058 (1901), where the court dismissed a tort action based on the following allegations: "[The defendant-appellee] had been decedent's family physician. Decedent became dangerously ill and sent for appellee. The messenger informed appellee of decedent's violent sickness, tendered him his fee for his services, and stated to him that no other physician was procurable in time to be of any use, and decedent did rely on appellee for medical assistance. Without any reason whatever, appellee refused to render aid to decedent. No other patients were requiring appellee's immediate service, and he could have gone to the relief of decedent if he had been willing to do s o . " There were no allegations that the defendant

148

· The Economic Structure of Tort Law

Richard Epstein has written: Ames insists that his rule1521 would not require the only surgeon in India capable of saving the life of a person with a given affliction to travel across the subcontinent to perform an operation, presumably because the inconvenience and cost would be substantial. But how would he treat the case if some third person were willing to pay him for all his efforts? If the payment is sufficient to induce the surgeon to act, then there is no need for the Good Samaritan doctrine at all. But if it is not, then it is again necessary to compare the costs of the physician with the benefits to his prospective patient. It is hard to know whether Ames would require the forced exchange under these circumstances. But it is at least arguable that under his theory forced exchanges should be required, since the payment might reduce the surgeon's net inconvenience to the point where it was trivial.53 But there is no economic basis for liability in this case. 5 4 Transaction costs are low; the patient (or some third party) need only meet the surgeon's price. Ames's rule is implicitly limited to the high transaction cost case. It is only here that the economic analysis is, as we have explained, inconclusive. had explicitly or implicitly promised to render service to the decedent when asked; the decedent had failed to negotiate the kind of contract involved in the Snyder case. 52. The rule in question would impose liability on one who fails to attempt rescue "when he might do so with little or no inconvenience to himself, and the death or great bodily harm follows as a consequence of his inaction." James Barr Ames, "Law and Morals," 22 Haw. L. Rev. 97, 113 (1908). 53. Epstein, note 43 above, at 199 (footnote omitted). 54. Nor is there any legal basis; see cases discussed in note 51 above.

.

6

.

Intentional Torts and Damages

The Economics of Intentional Wrongdoing The Meanings of

"Intentional"

The focus of our discussion up to this point, reflecting the greater practical importance of the subject, has been on unintentional harms. Not all of these are accidental harms; many nuisances have a probability of 1, as we assumed in Chapter 2. But in none of the cases we have discussed was the defendant trying to harm the plaintiff. This chapter focuses on cases in which this feature is present—the world of the socalled intentional torts, such as assault, battery, and false imprisonment. We shall also discuss concepts closely related to intention in tort law, such as recklessness. And because the award of punitive damages is an important feature of intentional-tort cases (although no longer limited just to them), we shall use the occasion to discuss the general principles of tort damages. The word "intentional" is a chameleon.1 It could be taken to refer simply to knowledge of consequences: in the tort context, to an injury that is actually foreseen, not just foreseeable. But what of the case where the potential injurer's activity level is so great that he knows, with a probability approaching 1, that he will some day inflict an injury? This is the case of a railroad that knows that in the course of a year it is bound to kill or at least injure some of its employees and some travelers at its

1. The terms intent and intentional have vexed philosophers since Aristotle. For some modern discussions see G. Ε. M. Anscombe, Intention (1969); A. Kenny, "Intention and Purpose," 63 J. Philo. 642 (1966); R. Lawrence, Motive and Intention: An Essay in the Appreciation of Action (1972); J. W. Meiland, The Nature of Intention (1970); P. N. O'SuIlivan, Intentions, Motives and Human Action: An Argument for Free Will (1977).

150 · The Economic Structure of Tort Law crossings.2 Can the railroad be said to intend to inflict these injuries merely because it knows they are a highly probable—virtually an inevitable—consequence of its decision, which is voluntary,3 to engage in the railroad business? A second possible meaning of the word "intentional" is "desirous of bringing about a particular result," which in the tort context is an injury. If Β throws a punch at A, he intends to hit him; if he takes A's bike, he intends to take it. The injury—hitting A or taking his property—is not an undesired by-product of other activity. Β might of course just be throwing a punch at A in self-defense or reclaiming a bike that A had stolen from him. But this circumstance would not affect the fact that Β desired and therefore intended to injure A. One might define "intentional" in still a third sense, as (1) desiring to do (2) an unjustified, a wrongful, injury. If this definition were accepted, then in the examples just given Β would not be guilty of inflicting intentional injury. He would not be guilty even if, in throwing a punch, he mistakenly believed that he was acting in self-defense, or, in the bicycle case, if he thought A had taken his bicycle but it was really A's bicycle. We thus have three competing interpretations of the term "intentional tort": 1. Undertaking an activity that the actor knows is almost certain to produce an injury within a short time but not desiring the injury to occur or benefiting from its occurrence; 2. Deliberately inflicting an injury, whether an injury that the injurer believes to be justified or one he believes to be wrongful; 3. Deliberately inflicting an injury that the injurer knows is wrongful. From an economic standpoint the first of our three interpretations is not a helpful guide to legal policy. If an act is consistent with efficient resource allocation, the fact that it is repeated many times does not provide a reason for public intervention; the ratio of costs to benefits is unchanged. This can be shown with the help of the model of accident costs presented in Chapter 3:4 2. Many nuisance cases are of this character, some with lethal consequences also. 3. We ignore the complication created by the fact that railroads usually cannot abandon a service without regulatory permission. 4. Recall that L(x,y) is the expected accident costs per potential victim; p(x,y) is the

Intentional Torts and Damages · 151 L(x,y) = p(x,y)D + A(x) + B(y).

(6.1)

The model assumes a single victim, but now suppose that the number of potential accident victims from a railroad's activities is η and, to make things simple, that each person has an equal probability of being injured, equal damages if he is injured, and equal costs of care. If np is larger than or equal to one (as it would be if, for example, ρ were .001 and η were 2,000), an injury is highly likely (assuming independence, the probability of at least one injury is about 0.86) and therefore it would be classified as an intentional injury under the first interpretation. 5 Yet it makes no economic difference whether π is 1 or 2,000 or 100,000. The levels of due care, x* and y*, are the same whether we desire to minimize L(x,y) or nL(x,y). The injurer's due care level, for example, is found by solving the equation - npyD = nBy and so is unaffected by n, the number of potential victims. 6 No economic purpose is served by classifying the injury in our railroad hypothetical as intentional, and we therefore predict (and find) that the law would refuse to do so. The second interpretation of "intentional"—deliberately inflicting an injury whether or not the injurer believes he is acting wrongfully—does identify a form of conduct economically distinct from the ordinary accident. A person who wants to inflict an injury is more likely to do so than when the injury occurs as a by-product of other activity; Β is more likely to injure A if he is trying to run him down with his car than if A just happens to be a pedestrian sharing the street with B. Hence intent probability of an accident; D is the resulting damages to victim A; χ and y are the inputs of care of A and the injurer, B, respectively; and A(x) and B(y) are the costs of care to A and B. 5. If the probability of not having an accident in one trip is (1 - p), the probability of not having a single accident in η trips (assuming each trip is an independent event) is (1 - p)". Hence the probability of having at least one accident in the entire set of trips is

1 - (1 - p)\

6. The due care level will depend on n, however, if there are economies of scale in care by potential injurers. For example, in the limiting case where a unit of the injurer's care reduces the probability of an injury to every potential victim, due care requires - np^D = B,, which increases with the size of n. This would imply a higher standard of care as η increases. Tort law recognizes this point. A good example is Conway v. O'Brien, 111 F.2d 611, 612 (2d Cir. 1940), rev'd on other grounds, 312 U.S. 492 (1941), where Judge Learned Hand defined the level of a person's care as a function of three factors: "the likelihood that his conduct will injure others, taken with the seriousness of the injury if it happens, and balanced against the interest which he must sacrifice to avoid the risk." The first of these factors is the probability of injuring not a particular person but any person and is equivalent to np(x,y) in our formulation. This point is overlooked in the attack on the economic approach to negligence made in William H. Rodgers, Jr., "Negligence Reconsidered: The Role of Rationality in Tort Theory," 54 So. Calif. L. Rev. 1, 9 (1980).

152 · The Economic Structure of Tort Law in the second sense identifies a class of activities in which the probability of injury (per victim) is higher than in the ordinary accident case. This is true whether or not the injurer believes he is acting wrongfully. The probability of an injury is no less when Β shoots A thinking A is attacking him than when he shoots him because he wants to rob him. To equate "intentional" to "highly probable" may seem to violate both the ordinary and the legal usage of the word "intent," but in fact the equation was suggested by a philosopher.7 The inability to peer inside another person's mind requires that inferences of intent be drawn from behavior. When someone does something that is overwhelmingly likely to produce a specific result, we are properly skeptical when he denies that the result was intended. Nevertheless, it would be a mistake to use the probability theory of intent (a subcategory of our second interpretation) as a basis for always imposing liability on an injurer. Even if the probability of B's injuring A is close to 1 and A's damages are great, the cost to Β of taking care to avoid the injury may exceed either the expected accident cost or the cost to A of avoiding injury. Consider the case of self-defense: Β injures A because Β reasonably believes that if he does not do so he will be injured as or more seriously by A. B's cost of avoiding injuring A includes the injury that Β will suffer if he does not defend himself and so may be great, whereas A might have been able to avoid being injured at a low or even negative cost by not attempting (or appearing to be attempting) to injure Β in the first place. If so, efficient prevention of injuries is encouraged by increasing the costs to A of injuring B, which can be done by a rule that excuses Β from liability for injuring A in selfdefense.8 If the probability theory of intent thus goes too far in some cases, it does not go far enough in others. It would exclude activities that involve a low probability of an injury but a disproportionately lower cost of avoiding the injury. Suppose Β is standing on a bridge over a highway, cars are speeding by, and Β drops a rock on one of the cars, desiring to damage the car and injure its occupants. The probability of B's being successful may be slight, but if the rock damaged A's car, it would do no violence to ordinary language (or, as we shall see, to the economics of the situation) to say that Β had intended to damage it, even though the injury would be excluded from the intentional category under the interpretation of intent that equates "intentional" to "highly probable." 7. See Kenny, note 1 above, at 650. 8. The second part of this chapter deals with self-defense in more detail.

Intentional Torts and Damages • 153 This suggests that our earlier proposition that intent can be inferred from high probability should be supplemented: intent can also be inferred from any combination of probability, severity, and cost of avoidance that shows that the injury was not merely a by-product of lawful activity. Β will not be heard to deny that he wanted to damage the car; there is no other plausible interpretation of his motives. This example illustrates the third interpretation of "intentional"— deliberately inflicting a wrongful injury. Although ρ may be as high as under the second interpretation, this is not necessary. The critical factor is that the costs of avoidance to the injurer are low relative to the social benefits of the activity. Indeed, in examples of pure aggression such as B's stealing A's car, B's beating up A for the sheer joy of it, or B's dropping a rock on A's car, the social cost of avoiding the injury is negative. Real resources expended by Β to commit the injury, as well as the costs to A, would be saved by deterring B. Hence the third meaning of intentional, unlike the first two meanings, establishes a clearcut economic basis for condemning a distinct form of misconduct. A Model of Intentional

Wrongdoing

In the case where the injurer expends resources to increase the probability of injury, the expected loss of A and Β is given by L(x,y) = p(x,y) (D - G) + A(x) + B(y),

(6.2)

where G is the gain to Β from inflicting the injury. Our analysis of unintentional torts implicitly assumed that G = 0 and that inputs of χ and y reduce p. This is not the same as assuming that the unintentional tortfeaser gains nothing from the activity giving rise to the injury. The railroad that injures A at a crossing obtains income from providing transportation services. But its full income is not increased (holding its costs constant) by inflicting the injury. In such a case (provided it is not an alternative care case), due care requires that both parties take care to reduce the probability of an accident (χ* > 0 and y* > 0). With regard to intentional torts, in contrast, it is no longer natural to think of G as zero; and py is now positive rather than negative. These points are connected. If the injurer must expend resources (y) in order to injure rather than avoid injuring someone, he will do so only if he obtains a positive gain from the injury—only if G > 0. Assume, as will usually be true, that the damages to the victim are equal to or greater than the gain to the injurer (D > G). Then D - G is either zero or positive, and in either case the optimal solution is x* =

154 · The Economic Structure of Tort Law y* = 0, for this is where ρ (0,0) = 0 and therefore where the combined net loss to the parties is zero. At any positive values of χ and y, L(x,y) would be positive. An example of a case where D = G would be a simple theft ("conversion" is the tort term) where the object taken was worth the same to the thief as to the victim. An example of D > G is a theft in which the thief injures the victim (battery and conversion) but gets no pleasure from inflicting injury, and again the thing taken is worth the same to both victim and thief. Later we shall consider the more difficult case where D < G. Where the optimal solution is x* = y* = 0, Β should be made liable for the injury to A. In Table 6.1, B, if not liable, will expect a net gain of $20 ($35 - $15) from injuring A, whereas if he is liable he will compare an expected liability of $50 (equal to A's expected damages) with $20 and will decide not to injure A. Liability not only avoids a "transaction" that would not be value maximizing even if costless to make—the expected damages are greater than the expected gain—but also prevents Β from making expenditures (y) to effect the transaction that are totally wasted from a social standpoint. Liability also deflects A from spending $10 to avoid being injured. Because the expected costs of B's behavior include both A's and B's expenditures ($25 combined) plus the net expected damages from that behavior—p(D - G)—of $15, a total of $40 is saved if Β is deterred from attempting to injure A. B's liability should not be affected if A fails to spend the $10 that would be his optimal expenditure on reducing the expected injury if Β were not deterred. As this is an alternative care case in which χ (victim care) is zero, there should be no defense of contributory negligence. We do not want A to spend $10 on self-protection. If he did, Β would not injure A, but there would still be a social loss of $10, which can be avoided by making Β liable. In the class of torts illustrated in Table 6.1 the optimal level of tortious activity is zero. If the law can perfectly deter such torts, it makes no difference (disregarding administrative costs) whether it does so by means of a fine payable to the state or damages payable to the tort victim. In

Table 6.1.

Losses from deliberately wrongful injury

pD ($) No injury attempted Injury attempted

0 50

PG 0 35

($) A(x) ($) 0 10

B(y) ($)

L($)

0 15

0 40

Intentional Torts and Damages · 155 either event, with perfect deterrence there will be no victims, no one will expect to be a victim, and no expenditures on χ will be made. But if enforcement costs are sufficiently large to make perfect deterrence unattainable, a fine paid to the state may be a more efficient method of loss avoidance than damages paid to the victim. If victims are not compensated, potential victims have an incentive to expend resources to avoid being victimized. These preventive expenditures may lower enforcement costs by an even greater amount and may thus represent an appropriate second-best method of tort prevention. Another consideration, however, argues in favor of the damages approach. Sometimes it is difficult to distinguish between the class of wrongs illustrated by Table 6.1 (where D > G) and similar but socially beneficial practices, such as breaking into an empty cabin in the wilderness to prevent starvation (these socially beneficial practices are a subset of the intentional injuries where D < G). Just because the necessary act is, by definition, socially beneficial, it does not follow that B, the actor, should bear no liability for the costs to A. The sensible fine in such cases may not be zero but rather the amount of the costs to the victim—to make sure that the act was committed only when the benefits exceeded the costs. Yet the optimal amount of the activity, even if enforcement costs are disregarded, is positive. In these circumstances the victim should receive damages rather than the state a fine, in order to prevent potential victims from overinvesting from a social standpoint in self-protection. If A, the potential victim, is not entitled to damages, he will expend resources on χ until pxD + Ax = 0. The effect of his expenditures will be either to deter Β from committing the "wrongful" act, because the reduced probability of success (as a result of A's expenditures on x) makes the expected gain to Β from the first unit of y less than the cost of a unit of y, or to reduce B's expenditure on y (assuming A's expenditure reduces the marginal product of y below y*, the loss-minimizing amount). Either outcome will be less efficient than if Β is forced to compensate A for his injury. Since G > D, A's expenditure on χ will lead to a smaller ρ and a greater value of p(D - G) in Eq. (6.2) (for example, increasing it from - 2 to - 1 ) and hence a greater value for L. Similarly, a positive value for χ implies positive costs of A(x) and also greater losses. Where G > D, L is minimized when χ is zero and y is set at y* (where py[D - G] + By = 0), yielding a negative value for L. The intuitive explanation is straightforward. If A is fully compensated for his damages, he will reduce his expenditures on χ (on preventing the injury) to zero, because a positive level of χ would impose costs on

156 · The Economic Structure of Tort Law him with no offsetting benefits. Β in turn will spend the amount necessary to minimize L. Hence the optimal result will be achieved: Β will undertake the activity and A will spend nothing to prevent it. This analysis may clarify our earlier point that there should be no defense of contributory negligence to intentional torts defined in accordance with the third and economically most appropriate interpretation of "intentional." In the case of unintentional torts, compensation is not necessary to induce both parties to behave efficiently. Under a negligence standard, indeed, it is precisely because the victim anticipates that the injurer will take due care and thus that the accident loss will not be shifted to the injurer that the victim is induced to take due care. Similarly, the injurer takes care to avoid liability because he assumes that the victim will not be contributorily negligent. Thus both parties are induced to take due care without the injurer's having to pay compensation to the victim. In the case of intentional torts, as defined here, liability with compensation (as distinct from liability to pay a fine or suffer another sanction that does not compensate the victim) is necessary to induce the parties to behave efficiently if some socially beneficial activities are classified as torts, so that complete deterrence is not optimal, as we are assuming it is for torts of negligence. We now explain our no doubt puzzling suggestion that the class of socially beneficial injuries is not coextensive with the class of injuries where G, the gain to the injurer, is greater than the victim's damages, D. Four cases where G is or seems greater than D must be distinguished. 1. In the first, illustrated by the example of theft from a cabin, the thing taken is more valuable to the thief than to the victim, in the sense— which is the sense of "value" analyzed in Chapter 1—that the thief is willing to pay more for the thing than the victim, its proper owner, would demand to give it up, but transaction costs prevent the transaction from being made voluntarily. There is an intentional injury, but it is value-maximizing. The law does not want to encourage potential victims to spend resources on preventing this kind of taking so it treats the taking as a tort, with the victim entitled to compensation and not subject to a defense of contributory negligence. 2. Next assume that A and Β are neighbors and Β covets A's car. Β would derive greater utility from owning the car than A, but because Β is much poorer than A he cannot buy it from him—so he steals it. In this case G > D in a utilitarian sense, but, as should be evident from Chapter 1, not in the narrower sense in which we compare gains and losses in this book. Our concept of efficiency excludes utility not accom-

Intentional Torts and Damages - 157 panied by willingness to pay. In terms of our model, this is a case in which G is actually less than D. 3. Consider next a variant of the last case: Β would in fact pay A more for the car than A would demand to give it up, but instead Β steals it. Because a market transaction would yield the same increase in value at lower cost—B's cost of stealing the car plus the cost of using the legal system to award damages to A would surely exceed the costs of using the market to transfer the car from A to Β—a market transaction is preferable to a forced exchange. 4. The final and most perplexing case is where Β values the thing taken more than A, in the sense that he would be willing to pay more for it than A would demand to give it up if a market transaction could be arranged; transaction costs are low; but the market transaction is somehow not an adequate substitute for the involuntary taking. An example is a rape in which a major part of the rapist's satisfaction derives from the coercive character of his act.9 Suppose a wealthy and sadistic man would be willing to pay a $100,000 fine each time he committed a rape, and his victims would each feel adequately compensated to receive $50,000 per rape, but if forbidden to rape he would be willing to pay no more than $1,000 to have sex. Here each rape appears to increase social value by $49,000 and there is no market substitute. The case seems similar to our first example, that of the cabin in the woods; there is a gain in value that cannot be realized through the market. The difference is that in the cabin example the coercive taking was a substitute for a market transaction; in the rape example there is, by assumption, no market substitute. If we tie the idea of value to the voluntary processes of the market, there is no increase in value in the rape case because it is not the kind of coercive act that improves the operation of the market, as the theft from the cabin does. In any event, it is understandable why the legal 9. An economic puzzle first discussed in Gary T. Schwartz, "Economics, Wealth Distribution, and Justice," 1979 Wis. L. Rev. 799, 806. For reasons unclear to us, this rather recherche example has become a focus of criticism of the adequacy of the positive economic theory of tort law. See, besides the Schwartz article, Dorsey D. Ellis, Jr., " A n Economic Theory of Intentional Torts: A C o m m e n t , " 3 Intl. Rev. Law & Econ. 45 (1983). Ellis incorrectly attributes to us the position that G, the gain to the injurer, should have no weight in the decision as to whether the conduct engaged in was wrongful. In fact we treat G as a social as well as private benefit, except in the unusual case (illustrated by the rape that is "valued" by the rapist in part because of its coercive nature) where value cannot be assigned to the conduct consistently with the use of the willingness-to-pay criterion that is the basis of the economic concept of value.

158 · The Economic Structure of Tort Law system would not undertake to compare the hypothetical offer price of the rapist with the hypothetical demand price of his victim. The measurement problems are overwhelmingly difficult and hardly worth undertaking to identify the rare rape that may in some sense be thought to increase social wealth. Also, there are market substitutes for coercive sexual acts—substitutes presumptively more efficient than using the legal system to direct the allocation of resources to sex. 10 In discussing the third sense of "intentional," we have assumed that the cost of avoiding injury is strictly negative. This is the clearest case of a wrongful intentional injury, but to regard it as the only one would introduce a discontinuity into the analysis of tort law that is foreign to the way in which economists usually think about problems. Take a case where pD = 100 when y = 1 but 0 when y = 0, meaning that there will be a resource saving of $101 if Β does not expend $1 to inflict the injury; and compare the case where pD = 50 when y = 0 but pD = 0 when y = 1, meaning that there will be a net saving of $49 if Β spends $1 on care. Β might be a driver who has decided to close his eyes while driving down a busy street, in order to rest them. The cost to him of keeping his eyes open is trivial ($1) and the expected accident cost of not doing so is very great, although not so great as it would be if he wanted to run down somebody. It seems arbitrary to place this kind of injury in a completely different category from the deliberately wrongful injury. Table 6.2 expands on this point, summarizes by way of examples our theory of intentional torts, and embeds the theory in our broader economic theory of tort law. For these purposes χ is ignored and y is defined Table 6.2.

Classes of tort cases Expected damages when y = 0

Case 1 2 3 4 5 6 7

Ρ

D($)

pD ($)

y($)

0.002 0.002 0.002 0.900 0.800 0.010 0.900

10,000 10,000 10,000 10,000 10,000 10,000 10,000

20 20 20 9,000 8,000 100 9,000

20,000 15 5 10,000 2 -10 -10

10. For further discussion see Richard A. Posner, "An Economic Theory of the Criminal Law," 85 Colum. L. Rev. 1193, 1198-99 (1985).

Intentional Torts and Damages · 159 as the expenditure by B, the injurer, that is necessary to avert injury to A. Case 1, where the cost of avoiding the accident is disproportionately greater than the expected accident cost, corresponds to what in the early common law was called "inevitable accident"—the kind of accident that could not be prevented at any cost commensurate with the expected accident cost. Case 2 is the standard negligence case, in which the cost of avoidance is less, but not dramatically so, than the expected accident cost. Case 3 corresponds to the tort category known as "gross negligence," defined by Learned Hand as follows: "not only must the interest which he [the injurer] would have had to sacrifice [to avoid the accident] be less than the risk to which he subjects others, but it must so far fail to match that risk that some opprobrium or reproach attaches to him." 1 1 Case 4 we discussed under the interpretation of "intentional" as "highly probable." The injury can be regarded as intentional because it is so probable; the high expected injury cost might even warrant treating the injury as prima facie unlawful; but the cost of avoidance may be, as in case 4 itself, higher than the expected injury cost, so the defendant should have a defense to the prima facie tort. Case 5 is the eye-resting case. A common legal term for this kind of injury is "reckless," but case 6, our earlier example of throwing a rock off a bridge, would often be described as a case of recklessness too. It is treated in law, as in our analysis, as an intentional tort because the costs of avoidance are negative, even though the probability of an injury is low. Case 7 is the deliberately wrongful injury that is also highly probable. Rows 5 through 7 are particularly important to the subject of this chapter. In each case the ratio of y to pD—of the costs of avoidance to the expected injury costs—is extremely low, and sometimes it is negative. It is also low, but not so low, in case 3, gross negligence—obviously a borderline case. The case for prohibition is clear in each of these examples but it is also clear in case 2, simple negligence. On the basis of the analysis thus far, the only respect in which cases 5 through 7 warrant different treatment relates to the defense of contributory negligence. We have shown that there should be no such defense in cases 6 and 7, although of course there should be one for case 2, simple negligence. Case 5 should probably be treated like cases 6 and 7 in this respect; probably the victim could not avoid injury at lower cost than the injurer, whose cost of avoidance is so low. But in this respect there is little or 11. Conway v. O'Brien, 111 F.2d 611, 612 (2d Cir. 1940), rev'd on other grounds, 312 U.S. 492 (1941).

160 · The Economic Structure of Tort Law no difference between case 3 (gross negligence) and case 5. We shall continue our search for a clear-cut economic criterion of intentional tortfeasing by considering, next, remedy. Remedies for Intentional Torts Our formal model implies that awarding only compensatory damages, equal to D in Eq. (6.2) or $10,000 in Table 6.2, will suffice to minimize the social costs of intentional as of unintentional torts. But a more complete economic model would show that punitive damages are proper in a subset of tort cases—namely cases 5 through 7 in Table 6.2. First, suppose we relax our implicit assumption that the probability of identifying the injurer and collecting full damages from him is equal to 1. (To avoid the question of criminal sanctions, we continue to assume that the injurer has sufficient funds to pay the damage award.) In many intentional torts, such as theft (conversion) or robbery (assault and conversion), that probability will often be much less than 1. The average injurer does not bypass the market in these cases because transaction costs are high or prohibitive. On the contrary, those costs typically are low, making market transactions feasible; that is precisely why (on an economic account) theft and robbery and the like are wrongful. Β bypasses the market to avoid having to compensate A. Hence Β has a strong incentive to conceal his identity or engage in other actions that will avoid his being sued. We can therefore expect the probability of identifying and successfully suing Β to be less than 1. If so, optimal damages would be a multiple of the victim's injury; in legal terms, punitive as distinct from merely compensatory damages would be awarded to the victim.12 Even then, potential injurers will simply be made indifferent between using the legal system and using the market to make advantageous transactions, when by hypothesis the market is cheaper because transaction costs are low. If in Table 6.1 the numbers for pD and pG are reversed, so that A's expected damages are only $35 and B's net expected gain is also $35 ($50 - $15), Β will be indifferent between using the market and using the legal system to get what he wants from A. True, there are potential gains to Β as well from using the market if it is cheaper; he can save $15 (minus the costs of using the market). On the other 12. For example, if A's actual damages are $100 but people who commit these torts are identified and successfully sued only half the time, then, assuming risk neutrality and ignoring detection and legal costs, the injurer who is found liable should be made to pay A $200.

Intentional Torts and Damages · 161 hand, if A is a skillful negotiator, he might obtain all of the surplus from a market transaction. To be confident of forcing this transaction into the market where it belongs, we must impose damages greater than $35. How much greater is uncertain. If we were confident that the underlying transaction really should take place only in the market, the size of the damage award would be limited only by the costs of collection. But if damages will sometimes be awarded in cases where high transaction costs make it socially efficient to bypass the market (as in the case of theft from the cabin in the woods), the additional damages should be limited to the costs of using the legal system to make the transaction. A related point is that punitive damages may be justified because it is too costly to measure actual damages accurately without the information provided by a market transaction. Suppose that in a particular class of intentional torts the victim's damages can range from $1 to $1,000, with a mean of $20. If to conserve on measurement costs the courts always or generally awarded $20 in such cases, there would be underdeterrence in many cases. Awarding damages of $1,000 in all cases, although punitive in most, would prevent underdeterrence without burdening the courts with heavy measurement costs. Alcorn v. Mitchell13 illustrates this principle. The defendant spat in the plaintiff's face in a courtroom full of people. The difficulty of measuring a victim's damages in such a case is apparent. By upholding a damage award of $1,000 (in 1872!), described as punitive, the court was implicitly encouraging "transactions" of this type to take place in the market. 14 Of the interpretations of "intentional" described in the preceding section of this chapter, the third—deliberately inflicting a wrongful injury— makes the clearest case for punitive damages. This type of injury usually arises in a situation where market transaction costs are low, the probability of obtaining damages from the injurer is less than 1, and it is costly for the courts to measure damages. As long as this category of wrongs is well defined, so that there is only a small probability of errors of classification, we predict that punitive damages will be awarded; the error costs are likely to be smaller than the gains in deterrence. This condition is more likely to be satisfied in the case of intentional torts, 13. 63 111. 553 (1872). 14. Cases of insult, such as Alcorn v. Mitchell, loom large in the tort law of punitive damages. See Dorsey D. Ellis, Jr., "Fairness and Efficiency in the Law of Punitive Damages," 56 S. Calif. I. Rev. 1, 15-17 (1982). Dan B. Dobbs, Handbook on the law of Remedies: Damages—Equity—Restitution 205 (1973), emphasizes the compensatory role of punitive damages in such cases.

162 · The Economic Structure of Tort Law such as battery, than in the case of unintentional torts governed by a negligence standard, because it is easier to determine whether an act is a battery (an offensive touching of another's person) than whether it is negligent. Should punitive damages be awarded where the defendant's conduct is reckless in the sense of case 5 in Table 6.2? On the one hand, compensatory damages should be adequate—none of the conditions arguing for punitive damages is present in the usual case of recklessness. 15 On the other hand, the costs of awarding punitive damages in a case of recklessness are lower than they would be in a mere negligence case. It is easy to be mistaken about whether conduct is negligent; and some conduct that really is not negligent is deemed negligent by operation of the reasonable-man standard. To award punitive damages in a negligence case would often produce a misallocation of resources. Some clumsy persons, whose costs of taking care were above average, would be led to increase their care to the (average) due care level, even though the added costs of care would exceed the reduction in expected damages that the added care brought about. In the case of recklessness, however, the gap between the expected accident costs and the costs of avoidance is so great that there is little danger of penalizing socially beneficial conduct or inducing excessive care. In these circumstances punitive damages may be defensible as providing a surer deterrent than compensatory damages to conduct that we know we want to deter. We now have two criteria for distinguishing in a functional rather than merely linguistic sense between intentional and unintentional torts: whether there is a defense of contributory negligence and whether punitive damages are awarded. If the law follows economics, it will treat cases 6 and 7 as intentional in this strong sense, and probably case 5 as well. But what about case 4, where an injury is deliberately inflicted but is perhaps justifiable? The fact that the probability of injury is higher in this case than in an accident case argues not for awarding punitive damages but for requiring a greater showing of justification—of costs of avoiding injury that offset the expected injury costs of the activity— than in the ordinary accident case. If the costs of avoidance are shown to exceed the expected injury costs, there is no economic basis for a sanction of any kind, whether compensatory or punitive. But what if the benefits or the avoidance costs are lower than the expected injury 15. This is not true in case 6, which we said the law might also treat as a case of recklessness. The defendant in case 6 would be punished more severely in criminal law than the defendant in case 5—the offense would be second-degree or even first-degree murder rather than (as in case 5) voluntary manslaughter.

Intentional Torts and Damages · 163 costs? Should the judgment against the defendant be merely for compensatory damages or for punitive damages as well? To answer this question we add to Table 6.2 a new row, case 4a, in which ρ is 0.9, D is $10,000, and y is $8,000 (rather than, as in case 4, $10,000). Case 4 may be a situation where Β reasonably believes that killing A is necessary to prevent A from killing him, and case 4a where Β honestly but unreasonably believes this to be so. The cost of avoiding such mistakes may be high, as case 4a assumes, because of the risk to Β if it turns out that A really is menacing him. From the standpoint of economic analysis this is a negligence case, like case 2, and punitive damages would be inappropriate. The question of unreasonable mistake as in case 4a is a clue to distinguishing between the second and third interpretations of "intentional." If the defendant has intentionally inflicted an injury where the real or reasonably apparent costs of avoidance (the alternative formulation is necessary to take care of cases of reasonable mistake) equal or exceed the expected injury costs, he has not committed a wrong, at least in an economic sense. And if he intentionally inflicts an injury through an unreasonable mistake, he is still not a deliberate wrongdoer, the third interpretation of "intentional"; he is merely negligent. Therefore, how the law treats this case (a matter taken up later in this chapter) will be a clue to how closely it tracks the economic analysis.

Defamation and Fraud The question of mistake supplies a clue to understanding one of the most baffling aspects of tort doctrine—the frequent conflation of the intentional and strict liability concepts, which one might have thought diametrically opposed. Defamation, for example, is commonly described as an intentional tort (and clearly is such under the second interpretation of intentional), but it is also frequently described as a strict liability tort because, under common law (that is, disregarding recent constitutional inroads), a mistake by the defendant on any issue other than publication of the defamatory utterance was not a defense to defamation. 16 E. Hulton & Co. v. Jones 17 illustrates this rule. The defendant's newspaper published a story about the amorous peccadilloes of a character called Artemus Jones. The writer of the story thought he had made up the name, and the story was indeed fictitious, but it turned out that 16. See W. Page Keeton et al., Prosser and Keeton on the Law of Torts 802 (5th ed. 1984). 17. 2 K.B. 44 (1909), aff'd, A.C. 20 (H.L.) (1910).

164 · The Economic Structure of Tort Law there was a real Artemus Jones, who sued the defendant for libel and won by showing that his neighbors thought the story was about him. The newspaper did not intend to defame the real Artemus Jones or for that matter the real anybody. In what sense therefore was the newspaper guilty of an intentional tort? The answer lies in a point made earlier, that intent can be inferred from behavior that creates a high probability of injury. 18 By publishing a defamatory story about an Artemus Jones without bothering to inquire whether there was such a person to whom the story might be taken to refer, the newspaper was like a person who walks about in an inhabited area firing a gun with his eyes closed. The probability of his hitting someone is high enough to justify treating the conduct as if it were intentional, and it can be argued that the injury inflicted by the defendant in the Jones case was intentional in the same sense. Jones is a very common last name; Artemus, unusual today, was more common in the nineteenth century. Still, it was not a deliberately wrongful injury. There was a mistake, and whether the defendant's conduct should be deemed intentionally tortious in a sense analytically distinct from the economist's concept of negligence (under which an injury is negligent if it is not cost justified) depends on the cost of avoiding the mistake relative to the expected injury cost. If that cost was very low, we would have a case of recklessness—case 5 in Table 6.2—and punitive damages would be appropriate. Even if the cost was not very low, there is a case for liability, although not for punitive damages. There was nothing at all that Jones could have done to avoid the mistake, whereas the newspaper could have avoided it by ascertaining whether there was an Artemus Jones to whom people might take the story to be referring, by indicating that the story was fictitious, or by not publishing it at all. The newspaper was clearly the lower-cost avoider of the mistake, and because the probability of a serious injury to reputation was high, the newspaper should have been deemed liable for its mistake, and was. 19 In short, mistake may make an intentional tort economically indistin18. Recall, however, that a low probability of injury does not necessarily imply absence of intent. A high probability of injury is a sufficient but not a necessary condition of inferring that the injurer intended the injury. 19. A curious feature of the case, although one not emphasized by the court, is that the plaintiff had once written for the defendant's newspaper. This fact makes it hard to believe that the defendant's conduct was based on a completely innocent mistake. It is not clear from the report of the case whether Jones was awarded punitive damages.

Intentional Torts and Damages · 165 guishable from an unintentional tort, an accident case. It is still possible to describe the injurer's behavior in such a case as intentional under the second interpretation of the word, which focuses on a high probability of injury. But punitive damages would be inappropriate, because the probability of identifying the injurer is near 1, and the costs of using the market, which include the costs of avoiding the mistake, are not obviously negligible (the newspaper could not transact with someone it did not think existed). There is a strong case for compensatory damages, because the optimal level of expenditures on care (or activity-level changes) to reduce the probability of an injury is zero for the victim and positive for the injurer. This is an alternative care case where due care requires χ* = 0 and y* > 0. Hence a rule that holds the injurer strictly liable will create incentives for both parties to take the right amount of care and to achieve the optimal level of activity. This explains why defamation can be regarded as an intentional tort even when the defendant does not know that he is defaming anyone and even though the defamer is strictly liable for mistakes, but it does not explain why defamation based on careless but not reckless mistake should also be classified as an intentional tort. Does it matter, so long as punitive damages are unavailable? We did say that another difference between intentional and unintentional wrongs besides remedy is that a defense of contributory negligence is inappropriate in the former. But this is a moot issue in the case of defamation. The cost to potential victims of avoiding defamation is normally prohibitive (x* = 0), so there is no place for a defense of contributory negligence in defamation cases, whatever the intent of the defendant. When the mistake in a defamation case is reasonable rather than reckless or careless, the question is whether there should be liability at all— that is, whether the right standard is strict liability. The argument for strict liability is a strong one. As already noted, the victims of defamation are almost entirely helpless to do anything to avoid being defamed— except to lapse into obscurity, a costly method of avoiding injury. The injurers, mainly writers and the news media, can avoid defamation not only by exercising greater care (by definition not a feasible option in the case of reasonable mistakes—mistakes that occur even though reasonable care was taken) but also by altering their activity. The newspaper in the Jones case did not have to publish fictitious articles that it represented to be truthful. Defamation, then, is a mixed tort. Liability is sometimes based on intent or recklessness and is sometimes strict. That is no problem as

166 · The Economic Structure of Tort Law long as the law does not confuse the grounds of liability and, for example, award punitive damages in a case where the only basis of defendant's liability is strict liability. The typology set forth in Table 6.2 is also useful in illuminating the development of a tort closely related (at least in economic analysis) to defamation: fraud (usually called "deceit" in tort law) and its modern extension, negligent misrepresentation. 20 Just as punitive damages create less risk of deterring socially desirable behavior when the gap between costs of avoidance and expected damages is very great (as in rows 5 through 7 of Table 6.2), so liability itself creates less such risk in such cases. Because determining the meaning and the truth of representations is difficult, implying substantial error and administrative costs, one is not surprised to find the following progression in the creation of liability for misrepresentation: first, liability for fraudulent statements; second, liability for deliberately misleading omissions; third, liability for negligent misrepresentations and omissions. In the first case ρ is high (because the injurer is trying to inflict an injury) and B(y) is negative. In the second ρ is high but B(y) may be positive, although perhaps low; it costs something to make whatever representations are necessary to prevent silence from being misleading (although it may also cost something to make selective omissions). The third is a conventional negligence case, like case 2 in Table 6.2, where pD and B(y) are close together. One might ask therefore why liability for negligent misrepresentations came later in the evolution of the common law than liability for negligent personal injuries. It may be that D is on average higher in personal-injury cases and therefore less likely to be swamped by error and administrative costs and that the costs of determining due care are lower in a broad range of such cases than the costs of determining the meaning and truth of representations in advertising, stock promotions, sales of businesses, and the like.

T h e Theory (Further) Applied to the Law The Prima Facie Case The prima facie case has a somewhat different structure in the case of intentional torts from that in the case of unintentional torts. Generally, proof that the defendant intentionally injured the plaintiff is enough to 20. See Keeton et al., note 16 above, ch. 18, and the Bishop article cited in the preface. On the economics of misrepresentation see Richard Craswell, "Interpreting Deceptive Advertising," 65 B. U. L. Rev. 657 (1985), and references cited there.

Intentional Torts and Damages • 167 establish the plaintiff's prima facie case (enough to defeat the defendant's motion to dismiss the case), but the defendant can defend by proving that his conduct was reasonable. Thus the plaintiff does not have to prove that the defendant's action was unreasonable, as he must do in a negligence case. 21 For purposes of establishing a prima facie case, intent may be inferred by any of the methods discussed earlier except knowledge of consequences (the first interpretation of "intentional"). That is, it may be inferred from a very high probability of injury, from the fact that the cost to the injurer of avoiding the injury was negative, or from the fact that that cost was trivial relative to the expected costs of injury. 22 But it may not be inferred from the fact that the defendant knew that an injury was likely merely because of the scale of his activity. A railroad-crossing accident is not an intentional tort merely because the railroad knows that some people are bound to be injured at its crossings every year. This pattern is consistent with the economics of intentional harm even though the fact that an injury is intentional in the sense of highly probable does not necessarily warrant a finding of liability. Remember that we are concerned here with the prima facie case. Rarely is an individual justified in deliberately injuring another, so as a matter of economy in pleading and proof it is sensible to place the burden of justification on the defendant. There is no similar presumption of wrongfulness in the case of unintentional injuries, and that is why the prima facie case of an unintentional tort requires proof of unreasonableness as well as injury. An exception to the generalization that it is unusual for someone to be justified in deliberately injuring another is the case of pollution, which is governed by the tort of nuisance. If a factory pollutes a stream by deliberately dumping wastes into it and the pollution amounts to a nuisance, it is an intentional nuisance because the dumping is deliberate. (If the wastes were dumped in the stream by accident, the nuisance would not be intentional although it might be negligent.) But there is 21. In the case of unintentional injuries governed by strict liability rather than negligence, the prima facie case is made out by showing that the defendant caused the plaintiff's injury, except in products liability cases, where the plaintiff normally must show that it was a defect in the defendant's product that caused his injury. 22. The leading treatise on tort law seems to regard as unintentional wrongdoing the case where intent is inferred from the disparity between the costs of avoidance and the expected injury costs, rather than from the high probability of injury. See Keeton et al., note 16 above, at 36, 212-13. But because punitive damages are available and contributory negligence is not a defense (see id. at 212-13), functionally this case is one of intentional tort.

168 ' The Economic Structure of Tort Law no basis for a presumption that intentional nuisances are unjustifiable, so there is no reason to make intentional nuisances prima facie unlawful. Nor has the law done so. Traditionally, as we saw in Chapter 2, nuisances, even if intentional, are unlawful only if shown to be unreasonable. The economics of the problem has triumphed over the conventions of language. Let us examine more closely the prima facie cases of some of the more common intentional torts. So far as personal injury is concerned, the basic intentional tort is battery, usually defined as any deliberate touching of another's person to which the person does not and cannot be presumed to consent. The qualification is important. Β may tap A on the back to gain his attention, clasp his hand to shake it, or brush a speck of dust off A's jacket—deliberate touchings of A's person—but it would be odd to regard these acts as even prima facie batteries. Defining the prima facie tort of battery as any deliberate touching would cast the net of prima facie liability too wide, imposing costs of litigation that would not be justified by the allocative gains from including a host of harmless, although deliberate, touchings within the prima facie scope of tort liability. The opposite extreme would be to define the prima facie tort not as a deliberate touching of a sort not normally consented to but as a touching intended to hurt the victim. Consider the facts in Vosburg v. Putney. 23 The plaintiff and the defendant were schoolboys. The defendant deliberately kicked the plaintiff in the shin. He did not intend to injure the plaintiff seriously, but because (unbeknownst to the defendant) the plaintiff's shin had been weakened by an infection, the kick inflicted a serious injury. The kick was held to be a battery and the plaintiff was allowed to recover damages for the full consequences. This result and the principle it illustrates—that the prima facie tort of battery does not require an intent to cause a "real" injury—make economic sense. Because the kick was deliberate and unauthorized, the probability of some harm, although perhaps only momentary discomfort, was very high, whereas the defendant's cost of avoidance was negative because he expended resources, however trivial, to injure the plaintiff. The plaintiff's costs undoubtedly exceeded the defendant's benefit—D - G > 0 in Eq. (6.2)—precisely because the defendant bore plaintiff no ill will. But a cost-benefit inquiry is in any event not required to decide the liability question. Even if the defendant's gain had exceeded the plaintiff's loss, the economic argument for liability would still have been 23. 80 Wis. 623, 50 N.W. 403 (1891).

Intentional Torts and Damages · 169 compelling because the injury occurred in a setting of low transaction costs. 24 In many battery cases the injury inflicted is slight; Alcorn v. Mitchell, where the defendant spat on the plaintiff, is an example. Even more attenuated is the injury inflicted in the typical assault case. An assault in tort law is a threatening gesture that is intended to and does create in the mind of the victim an apprehension of an immediate battery but does not result in an actual touching of the victim. There is harm, however—the same kind of interference with tranquility and repose, with one's "private space," as in Alcorn and other cases of offensive but not injurious battery—and damages are properly awarded. In all of these cases, 25 although the harm inflicted by the tort is small, the social costs of avoiding the harm are negative, thus strengthening the economic case for liability. What if someone deliberately inflicts emotional distress on another by threats or practical joking or other means that do not rise to the level of an assault because there is no menacing gesture which creates an apprehension of an immediate battery? Although at first the tort law refused to provide any remedy in such cases, later a tort of intentional infliction of emotional distress was recognized. Many courts, however, required the plaintiff to prove that the distress was so severe that it had caused a physical illness. Most states have now abandoned this limitation on recovery. 26 The initial reluctance of tort law and the insistence by some courts on proof of physical as well as emotional injury may reflect the difficulty of establishing a causal link between the defendant's conduct and the plaintiff's injury in these cases. It is one thing to infer emotional harm from the kind of threatening gesture, such as pointing a gun at the plaintiff, that is involved in the usual assault case, and another to infer harm from threats, practical jokes, and the like. If the law were utterly confident that the defendant's conduct was wrongful, it would not have to worry about occasionally making a defendant pay damages far in excess of any actual harm to his victim; these would be a form of punitive damages, performing the same function as in the Alcorn case. But unlike the conduct involved in the typical assault or battery case, the threats and pranks and other outrages involved in cases 24. Awarding damages to the victim that are equal to his actual as distinct from reasonably foreseeable injury, another feature of the Vosburg case, is a general principle of tort damages law—the eggshell-skull principle discussed in Chapter 8. 25. See also note 30 below and accompanying text. 26. For a thorough analysis, see Peter A. Bell, "The Bell Tolls: Toward Full Tort Recovery for Psychic Injury," 36 U. Fla. L. Rev. 33 (1984).

170 ' The Economic Structure of Tort Law of intentional infliction of emotional distress are often ambiguous, and the potential social costs of erroneous liability findings are therefore greater. Heavy damage awards based on erroneous findings of causation could impair lawful activities, such as debt collection, that invite charges of intentional infliction of emotional distress. (See also Chapter 8.) But with advances (halting as they have been) in psychiatric knowledge, courts now feel more confident that correct causal inferences can be drawn, and they are therefore more willing to find tortious conduct even when there is no physical harm. The law governing deliberate takings of or damage to property other than land follows principles broadly similar to those governing personal injury, although called by different names (conversion and trespass to chattels rather than battery and assault), so we will not pause to discuss them. The legal situation with respect to land is somewhat different, however. The basic tort is trespass to land, and the initial peculiarity is the difficulty of classifying the tort between the intentional and strict liability categories. It is often said that trespassing requires a voluntary act— that if Β picks up A and without A's consent hurls him onto C's land, C does not have an action in trespass against A, although he has one against B. 27 But in Louisville R. Co. v. Sweeney, for example, the defendant was held liable in trespass when its streetcar accidentally crashed onto the plaintiff's land. 28 Another peculiarity of the law of trespass to land is that a trespass is frequently said to import or imply damage even if no actual damage takes place. 29 It is easy to see how even a slight battery causes some damage: no one likes to have his person deliberately touched by another outside of the situations in which deliberate touching is sanctioned by social custom and implicit consent. 30 But if Β steps across the boundary onto A's unoccupied, unused land, his action seems not to cause harm to A in even the attenuated sense involved in many of the battery and assault cases. These puzzles are dispelled once the critical economic factors bearing on trespass to land are identified. In the accidental trespass, as that by the vehicle that swerves uncontrollably off the road and into the plain27. See Keeton et al., note 16 above, at 68, n. 6. 28. 157 Ky. 620, 163 S.W. 739 (1914). This is now the minority view; the majority view requires that the entry on the plaintiff's land must be intentional, negligent, or the result of abnormally dangerous conduct. See Keeton et al., note 16 above, at 69. Sweeney itself was overruled in Randall v. Shelton, 293 S.W. 2d 559 (Ky. 1956). 29. See, for example, Dougherty v. Stepp, 18 N.C. 371 (1835), and our discussion in Chapter 2. 30. On the economic basis of the private space concept see Richard A. Posner, The Economics of Justice, ch. 10 (1981).

Intentional Torts and Damages · 171 tiff's house, the case for strict liability rests on the fact that, unlike a collision case (whether involving another vehicle or a pedestrian), the plaintiff can do nothing, either by altering his activity or by increasing his care, to avoid the accident. The fact that it is an accident—that is, a low-probability event—rather than a continuous harm such as a nuisance becomes a reason in favor of strict liability. Although it might be efficient for a potential victim to take care or adjust his activity in the face of a continuous harm, it would not be efficient when the harm was unlikely to occur or at least unlikely to occur frequently. Thus it was not a ridiculous anachronism or evidence of a perverse preference for property values over human life that some common law courts subjected the driver of a vehicle to a rule of negligence for the harm caused to another driver but to strict liability for the harm caused by the vehicle's swerving onto an adjacent landowner's property—even if both harms resulted from the same accident. Notice the parallel to the legal treatment of airplane crashes, discussed in Chapter 4. The explanation for why the "harmless" trespass to land is actionable becomes obvious once opportunity cost is recognized as a form of loss. We must ask what the trespasser wanted by entering the owner's property. Did he want to take a shortcut? To enjoy nature? To pick wildflowers or raspberries? Whatever his purpose, he imposed a cost on the owner—the opportunity cost of the land—that is difficult to measure. Because transaction costs normally are low in this case, the use of the land should be directed by the market. This is done by giving the owner a right of action against the trespasser even if the latter has not harmed him beyond denying him an opportunity to exploit fully the value of his property right.31 Defenses to Intentional

Torts

Some of the most interesting economic questions concerning intentional torts involve the defenses to them. We begin with consent, which plays the same role in the economic theory of intentional torts as assumption of risk plays in the economic theory of unintentional torts. The objection to the conduct that is classified as intentionally tortious is that it is an effort to bypass the market in a setting where transaction costs are low and the conduct itself inefficient—a setting in which the loss function in Eq. (6.2) is minimized if the conduct is deterred. In some circum31. Suits for trespass were also a method for resolving boundary disputes and otherwise determining ownership rights. This is a related point because a property right implies, as we know from Chapter 2, the right to exclusive use of the property.

272 · The Economic Structure of Tort Law stances, however, the conduct will be efficient. The best test of this is a market test. Consent transforms coercion into a mutually desired and therefore value-maximizing transaction: it transforms the brawl into the boxing match. 32 The analysis does not quite hold for the case where consent is inferred from conduct and it later turns out that the victim did not intend to consent. When for example the plaintiff held out his arm in a way that manifested consent to being inoculated although in fact he did not want to be, 33 the defense of consent (perhaps misnamed) served to place liability for a mistake on the party that could avoid it at least cost. That was the plaintiff. He created a misleading impression of consent to an otherwise offensive touching. Another illustration of the economic function of implied consent is the case of the unauthorized surgical procedure.34 If while the patient is under a general anesthetic the surgeon discovers that an operation different from the one that the patient had authorized is called for, "the mere desirability of the operation does not protect the surgeon, who becomes liable for battery . . . It is of course possible that the situation may be one of unforeseen emergency, critical in its nature, which will justify the surgeon in proceeding on the assumption that the patient would consent if he were conscious and understood the situation." 35 Characterizing a situation as "one of unforeseen emergency, critical in its nature" is a way of saying that the costs of an explicit transaction with the patient are prohibitively high. In such a case the law properly allows the market to be bypassed. Consider now the defense of self-defense. The usual justification, when translated into economic terms, is that a threat of legal sanctions cannot influence the conduct of someone who feels that his life is in danger, so that making a person liable in damages for battery when he was acting in self-defense would have no effect on his conduct. But this 32. Consent may not be a defense where there is a statute forbidding unlicensed boxing matches, as in Hudson v. Craft, 33 Cal. 2d 654, 204 P.2d 1 (1949). But in such a case the issue for the court is what rule best promotes the statute, not what rule best promotes the basic common law policy of facilitating voluntary transactions. Denying the defense is probably the best rule from the standpoint of enforcing the statute and thus deterring the activity: it penalizes the winner, and although it also reduces the costs to the loser and thus "encourages" the boxer who expects to lose, probably most people who fight expect to win. Hence there will be a greater deterrent effect from making the winner liable than there would be under the anti-loser rule that consent is a defense. 33. O'Brien v. Cunard S.S. Co., 154 Mass. 272, 28 N.E. 266 (1891). 34. See, for example, Möhr v. Williams, 95 Minn. 261, 104 N.W. 12 (1905). 35. Keeton et al., note 16 above, at 119 (footnotes omitted).

Intentional Torts and Damages · 173 justification is compelling only in the limited class of cases where the battery defended against is (or seems to the prospective victim to be) truly life threatening and only one course of action can possibly avert it. Where these conditions do not hold, the threat of liability for battery can be expected to affect victim behavior, although perhaps only slightly. It will induce greater care in evaluating the danger posed by an assailant (or apparent assailant) and affect the choice among possible methods of responding. A cleaner economic explanation for the defense is that it has a deterrent function. It reduces the probability of injury to the intended victim and raises the costs to the injurer of inflicting the injury. In terms of Eq. (6.2), self-defense is an expenditure by A of A(x) that both lowers ρ and raises B(y) for a given quantity of y. Even if court-awarded damages are inadequate to deter the injurer (for example, because his wealth is less than the victim's damages), he may be deterred by knowing that if he tries to commit the tort, the victim's expenditures on self-defense may frustrate the attempt and perhaps impose (other) costs on him, the injurer. Formally, a tort will be committed only if the tortfeasor's expected net benefit is positive, that is, only if G» = p(x,y) (G - D) -

B(y) > 0,

(6.3)

where GB is his net benefit, D his expected damage liability (where D > G > D—that is, the victim's injury exceeds the injurer's gain which in turn exceeds the letter's expected damage liability), and p(x,y), G, and B(y) are defined as before. Self-defense will lower p, raise B(y) (because any injury to B, the injurer, will be a part of his costs), and thus make it more likely that GB will be negative. Tort law allows only reasonable force to be used to defend oneself, and the economic explanation for this limitation is not self-evident, as one might think. Once conduct is identified as socially undesirable—and the conduct involved in the usual battery certainly fits this description— then, as we have pointed out, the optimal sanction is limited in severity only by the cost of imposing the sanction and not by the injury suffered by the victim (or even by that injury divided by the probability of apprehending and punishing the assailant). And the fact that a battery was committed or attempted shows that the legal penalties were not sufficient to deter it; maybe an unlimited right of self-defense would be sufficient. There is always the possibility of mistake—the battery may be unintended or the assailant laboring under a mistake as to the justifiability of his conduct. But this argues for applying a principle of reasonable force in those cases—not in every battery case. Yet the com-

174 · The Economic Structure of Tort Law mon law rule is that the victim may not use unlimited force to repel even fully deliberate, knowingly wrongful battery.36 A possible explanation is the following. The fact that a battery is being attempted despite the tort and criminal sanctions attached to such conduct implies that the assailant is very difficult to deter. If he is completely undeterrable, the right of self-defense should be confined to what might be called "damage limitation"—to efforts that minimize the joint costs to assailant and victim rather than just the costs of the attack to the victim. The law does not go quite so far; it does not say that if you are being attacked by a raving lunatic, and you can avoid serious but not deadly injury to yourself only by killing him, you may not do so. But to the extent that the concept of excessive force is used to limit the permissible scope of self-defense, it has the effect of subordinating deterrent to damage-limitation objectives, and this may be the proper approach if the fact of an attack is taken to indicate that the assailant is undeterrable. Courvoisier v. Raymond37 illustrates the interesting group of cases in which the person defending himself is mistaken in believing that he is under attack by the person whom he injures. The defendant was standing in his doorway and rioters were milling about in a crowd outside. The plaintiff detached himself from the crowd and approached the defendant hand in pocket. Thinking that the plaintiff was one of the rioters and about to attack him, the defendant shot the plaintiff, who was actually a plainsclothes policeman trying to help him. The plaintiff sued for battery. The court held that plaintiff could not win if the defendant was reasonable in believing that the plaintiff was about to attack him without justification. A seemingly more difficult case of this type is Morris v. Piatt,38 where the plaintiff alleged that he was a mere bystander whom the defendant shot accidentally while aiming at the people who were attacking him. Again the defense of self-defense was upheld. Can the result in either case be justified by reference to either a deterrent or damage-limitation theory of self-defense? From the facts of Courvoisier it appears that the plaintiff could have dispelled the mistake by identifying himself to the defendant as a policeman. The plaintiff was thus the lower-cost accident avoider, the "accident" being the mistake about who he was. It is no answer that it was the defendant's mistake. The mistake was made by both plaintiff and defendant, and 36. See, for example, Germolus v. Sausser, 83 Minn. 141, 95 N.W. 946 (1901). 37. 23 Colo. 113, 47 Pac. 284 (1896). 38. 32 Conn. 75 (1864).

Intentional Torts and Damages · 175 the economic question is, who could have avoided the accident at lower cost? Piatt is a harder case. A bystander is not self-evidently a lowercost accident avoider than his mistaken assailant. But the statement of facts in the Piatt opinion indicates that the plaintiff was a member of a crowd other members of which were attacking the defendant, and the plaintiff could have avoided the injury to himself by leaving the crowd. Even if he could not have avoided injury at any reasonable cost, there would have been no allocative benefit to imposing liability unless the defendant could have avoided the accident at a cost less than the expected cost of the accident; and if the defendant's mistake was a reasonable one, this implies he could not have. Our analysis further implies that if the defendant had been negligent in his mistaken belief that he was under attack, his liability should have been limited to compensatory damages. There is no case for punitive damages because, from an economic standpoint, the tort is one of negligence. 39 The defense of property raises some interesting questions. We can distinguish three types of case: (1) The defendant uses force or the threat of force to protect his property from being taken or destroyed; (2) the defendant damages or destroys another's property to protect his own; (3) the defendant damages or destroys another's property to protect the property of third persons. 1. The governing principle in the first class of cases is that a battery is lawful where necessary to protect a property right, but only if the battery does not involve wounding or killing. A landowner is privileged to carry the trespasser off his property, although that is a battery, or even to shove or drag him roughly off the property, but he may not injure seriously or kill him. 40 Limiting the use of force to protect property would make unimpeachably good economic sense under the damagelimitation theory of defenses to intentional torts discussed ^bove, if human life and bodily integrity were always more valuable than purely property interests, but they are not. If a vandal were wounded in an effort to prevent him from slashing the Mona Lisa, the property interest thus protected would be worth more than the interest in bodily integrity that was being invaded. But it does not follow that because a case can be hypothesized in which mechanical application of the limitations on 39. If we let D" equal the victim's damages and p(x,y) the probability of this injury, then L = p(x,y)D' + A(x) + B(y), where both p, and ρ take negative values. Both parties will have an incentive to take due care if compensatory damages are awarded when χ a x* and y < y*, where x* and y* are the levels of care that minimize L. 40. The leading case is still M'llvoy v. Cockran, 9 Ky. 291 (1820). There is a cognate right to use limited force to protect chattels (goods as distinct from real estate).

176 · The Economic Structure of Tort Law the use of force to defend property would yield an economically incorrect result, these limitations are economically incorrect. To our knowledge, no case remotely like our Mona Lisa hypothetical has ever arisen, and one can only speculate on how such a case would be decided. More important, the infrequency of cases in which the property interest at stake is worth more than the personal interest that would have to be sacrificed suggests that the law's dichotomous treatment of property and bodily integrity at least approximates the underlying economic realities. In the last part of this chapter we show why as a matter of economics an injury to property is indeed very unlikely to justify inflicting a serious bodily injury. An interesting subset of cases in which force is used to protect property involves spring guns. Normally set in unoccupied commercial premises, they are designed to be tripped by and to wound an intruder. A mechanical application of the excessive-force rule would be that because deadly force is being used to protect a mere property interest, the spring gun is impermissible and anyone wounded by it (except perhaps a felonious intruder) is entitled to recover damages. There is more than a trace of this view in the cases and literature on spring guns, including the Restatement of Torts, but many of the cases involve the interpretation of statutes regulating or (more commonly) banning spring guns.41 The common law position was more complex and economically more sensible. In the famous case of Bird v. Holbrook,42 the defendant owned a garden in a London suburb where he grew valuable tulips for sale. He lived about a mile away. There had recently been a theft from the garden even though it was walled. The defendant set a spring gun in the avowed hope of wounding the thief when he returned to steal more tulips, as the defendant expected him to do. The defendant did not put up any notices that he had set a spring gun; he wanted the thief to enter the garden and trip the gun. The plaintiff, a young man with no predatory intent, entered the garden during the daytime to recover a peahen that had strayed into the garden. He accidentally tripped the spring gun and was seriously wounded. He sued the defendant and recovered damages. The situation described in Bird v. Holbrook is illustrated numerically in Table 6.3. If the owner does nothing (row 1), his expected tulip loss is $100; if he sets a spring gun without posting notices (row 2), the loss 41. For discussion and citations see Richard A. Posner, "Killing or Wounding to Protect a Property Interest," 14 ]. Law & Econ. 210 (1971). 42. 4 Bing. 628, 130 Eng. Rep. 119 (1828).

Intentional Torts and Damages · 177 Table 6.3.

Net benefits of setting a spring gun

Owner's behavior Does nothing Sets spring gun Sets spring gun and posts notice Sets spring gun only at night

Expected tulip loss

($)

100 10 12 20

Cost of spring gun

($) —

Cost of Expected notice injury

($)

($)

—

—

10 10 10

1 —

Expected loss

($)

25 2

100 45 25

1

31

falls to $10. Suppose the cost of the spring gun is only $10 and setting it reduces the net expected loss by $65 (the $90 expected reduction in tulip loss minus an expected injury of $25). It may appear that the owner cannot be negligent, because a $10 expenditure on a spring gun yields benefits of $65. But this is wrong; the correct approach (and the one the court took in Bird) is to compare the marginal benefits and marginal costs of added precautions. On this basis the efficient solution is to set a spring gun and post a notice (row 3), because the additional $1 for the notice reduces expected losses by $21—a $23 reduction in expected injury minus a $2 increase in expected tulip losses. 43 Row 4 demonstrates another alternative—setting the gun only at night. Although this is probably preferable to row 2 because of the savings in injury costs to trespassers who do not intend to steal tulips, it may be inferior to setting the gun and posting a notice, because of the increased tulip losses as thieves shift their activity to the daytime. Consistent with this analysis, the court in Bird v. Holbrook did not hold that setting spring guns to defend property interests is always contrary to common law. But it noted two interrelated facts that defeated the defendant's claim of privilege (row 2). First, the defendant was not trying to deter a theft but to punish the thief: the motive was vengeance. The economic theory of tort and criminal law—and therefore presumably of the self-help remedies authorized by that law—is deterrent rather than retributive.44 Second, in his zeal for revenge, the defendant overlooked the possibility that an innocent or even privileged trespasser 43. We assume that the notice adds $2 to the expected tulip losses because it makes it more likely that the potential thief will be able to devise a method of avoiding injury from the spring gun (for example, by disarming it). 44. See Posner, note 30 above, ch. 7.

178 · The Economic Structure of Tort Law might enter the garden (and arguably, as we are about to see, the plaintiff in Bird v. Holbrook was privileged to pursue the peahen into defendant's garden, assuming that he could recapture the bird without doing disproportionate damage to the tulips). Of course a notice would probably be unavailing at night, but the probability that an innocent trespasser would enter the garden at night was lower than the probability that one would do so during the day. Notice seems the key economic issue in this case. Posting a notice accommodates the legitimate interests of the owner of the garden—who because he lived at a distance from the garden and because there were no police in the London suburbs in the early nineteenth century may not have been able to protect the garden from trespassers at reasonable cost other than by a spring gun—with the interest in bodily integrity invoked by the plaintiff's unfortunate entry into the garden. When adequate notice is given, the tradeoff is no longer between life and property but between a small probability of injuring or killing an intruder and a property interest difficult to protect by alternative means. 45 2. We turn now to cases, illustrated by Vincent v. Lake Erie Transportation Co. 46 and Whalley v. Lancashire & Yorkshire Ry. Co. 47 where the defense to a suit for property damage is that the damage was necessary to avert a greater harm to the defendant's property. In Vincent the defendant's ship was moored to the plaintiff's pier when a storm arose. Rather than put out to open water, the defendant's crew lashed the ship more tightly to the pier; the ship was saved but the pier damaged. No provision of the contract between the parties obligated the plaintiff to provide a berth for the defendant's ship during the storm. The plaintiff was held entitled to recover for the damage to its pier, although there is no suggestion that the defendant had acted improperly in securing the ship to the pier. In fact, we may surmise from Ploof v. Putnam 48 and from the excessive-force cases discussed above (including Bird) that the plaintiff in Vincent would have been guilty of a tort if it had prevented the defendant from securing the ship to the pier. Recall from Chapter 2 that in Ploof the defendant's employee had prevented the plaintiff from mooring at the defendant's pier during a storm, causing 45. In the latest spring gun case we have found, McKinsey v. Wade, 136 Ga. App. 109, 220 S.E.2d 30 (1975), the defendant booby-trapped a cigarette vending machine in his liquor store with dynamite, which exploded and killed a trespasser. Again there was no notice; it is the nature of a booby trap to operate by surprise. 46. 100 Minn. 456, 124 N.W. 221 (1910). 47. 13 Q.B.D. 131 (1884). 48. 81 Vt. 471, 71 Atl. 188 (1908).

Intentional Torts and Damages · 179 injury to him and his property, and this conduct was held to be tortious. Because the defendant's pier was not damaged, there was no issue of an offsetting damage claim by the defendant against the plaintiff, as in Vincent. Ploof was also an easier case because the defendant was endangering the plaintiff's life in order to protect a slight property interest (the right to undisturbed possession of his pier). The emergency precluded a transaction in which the defendant would have negotiated with the plaintiff for compensation for using the pier, so it was not the ordinary trespass case where the trespasser's conduct imposes an opportunity cost on the owner of the property and transaction costs are low. Vincent raises two issues not involved in Ploof. The first is whether property may be damaged or destroyed to protect other property as well as to protect personal safety. The court's implicit answer is yes, that the defendant had the right to secure its ship to the plaintiff's pier even though no contract entitled it to do so. Because of the sudden storm, there was no time for a negotiation between defendant and plaintiff.49 Because the expected benefits of saving the ship exceeded the expected costs in damage to the dock, it would have thwarted a wealth-maximizing although involuntary exchange for the plaintiff's employees to have cut the ropes. Ploof and Vincent together illustrate the cases, conventionally grouped under the rubric of "trespass of necessity" or "private necessity," in which high transaction costs make a trespass privileged. The second issue in Vincent but not Ploof is whether the defendant, even though it acted reasonably in securing its ship to the plaintiff's pier, should pay compensation for the resulting damage. In other words, was the standard applicable to the defendant's conduct strict liability or negligence (that is, unreasonableness)? If the latter, the defendant would not have been liable—the expected cost of putting out to open water was greater than the expected cost of damage to the pier. The court may well have been correct in making the defendant strictly liable. The reasons for our conclusion have to do with both the injurer's and the victim's incentives. If victims are not compensated, fewer piers may be built in the future; the analogy to the analysis of Good Samaritan liability in the previous chapter should be apparent. Moreover, strict liability will induce ship owners and masters to consider how they can avoid accidents not only by handling their ships more carefully but also by altering the 49. A negotiation conducted after the storm began would have had a strong element of bilateral monopoly, which as we know from Chapter 2 is a source of high transaction costs. See William M. Landes and Richard A. Posner, "Salvors, Finders, Good Samaritans, and Other Rescuers: An Economic Study of Law and Altruism," 7 J. Legal Stud. 83, 91 (1978).

180 · The Economic Structure of Tort Law frequency or season of sailings—activity adjustments that would probably be overlooked by a court applying a negligence standard. The defendant in the Whalley case was a railroad. As a result of an unprecedented rainfall, a large quantity of water accumulated against one side of the railroad's embankment and threatened to cave it in. To save the embankment the defendant cut trenches through it which conducted the accumulated water onto the plaintiff's land, damaging his crops. The defendant showed that it had acted reasonably to save the embankment. Even so, it was held liable. There is no suggestion that the plaintiff should not have been farming near the embankment in order to avoid a flood that was highly improbable, resulting as it did from an "unprecedented" rainfall. And, granted that the defendant was careful in cutting trenches through its property and was acting to save property more valuable than the plaintiff's crops, it could have prevented harm to the plaintiff either by cutting the trenches before the rainfall occurred or by altering the number, size, or direction of the trenches. In principle all of these options could be canvassed as part of a negligence inquiry, but because the defendant's costs of taking care or altering its activity level were probably lower than those of the plaintiff, a rule of strict liability made better economic sense. 3. Where property is damaged or destroyed to save the property of a third party, negligence rather than strict liability is often the standard used. This class of cases goes by the name "public necessity," to distinguish it from the private-necessity cases like Vincent and Whalley that we have just been discussing. Cases of public necessity involve things like making firebreaks, burning contaminated materials, and destroying valuable property to deny it to an invading army. 50 The economically and legally defining characteristic of a public-necessity case which excuses the defendant from having to pay damages unless he has acted carelessly is that the defendant's conduct confers external benefits. That is what public as distinct from private means in these cases—a happy confluence of the legal and economic meanings. Because the defendant does not reap the full benefits of his act, neither should he have to pay the full costs. Otherwise there will be too little of his activity.51 Of course, 50. See Merrill v. City of Wheaton, 379 111. 504, 41 N.E.2d 508 (1942); Keeton et al., note 16 above, at 146; 18 Eugene McQuillin, The Law of Municipal Corporations § 53:58 (3d ed., Stephen M. Flanagan ed., 1984). 51. A striking example of this danger is reported concerning the great London fire in the seventeenth century: "[T]he Lord Mayor of London, in 1666, when that city was on fire, would not give directions for, or consent to, the pulling down of forty wooden houses, or to the removing the furniture, etc. belonging to the Lawyers of the Temple, then on

Intentional Torts and Damages · 181 the mere fact that the beneficiary of the damage is not the defendant is not enough to make the case one of public rather than private necessity. Benefits are external in a sense relevant to economic policy only when the beneficiaries are sufficiently numerous to make the costs of voluntary transactions with their benefactor prohibitively high or when some other factor prevents voluntary internalization of the benefits. We know of no case in which a defense of private necessity has been recognized where transaction costs would have been low.

Intent Revisited The courts have been willing to attach the label "intentional" to an injury that is intentionally wrongful in an economic sense although not literally an intentional injury to the victim. Consider the well-known doctrine of "transferred intent." If Β shoots at A and by accident hits C, a bystander, instead, Β is liable in battery to C even though he did not intend to injure him. This is the economically correct rule because the potential injury to C is an additional expected cost in the loss function of Eq. (6.2). the Circuit, for fear he should be answerable for a trespass; and in consequence of this conduct half that great city was burnt." Respublica v. Sparkhawk, 1 U.S. (1 Dall.) 357, 363 (Pa. Sup. Ct. 1788), quoted in Charles O. Gregory, Harry Kalven, Jr., and Richard A. Epstein, Cases and Materials on Torts 48 (4th ed. 1984). The story told in the Respublica opinion has been confirmed by a historian. See W. G. Bell, The Great Fire of London in 1666, at 29-30 (1920). The most important example in tort law of externalizing accident costs in order to encourage the provision of public goods is the tort immunity of charitable enterprises. At a time when low incomes depressed the level of charitable giving, to have made charitable enterprises (particularly hospitals) bear their full accident costs might have driven many of them out of business. The consequence would have been less charity than was desired (bearing in mind the free-rider problem that prevents a free market from providing the desired amount of charity). As incomes rose (charity is a superior good) and the government increased public support to hospitals and other traditionally charitable enterprises, the benefits from allowing charities to externalize their accident costs declined, and, not surprisingly, the charitable immunity was phased out—a process nearly complete by the middle 1970s. On the rise and decline of the immunity, with special reference to hospitals, see Bradley C. Canon and Dean Jaros, "The Impact of Changes in Legal Doctrine: The Abrogation of Charitable Immunity," 13 Law & Soc'y Rev. 969 (1979); on the enormous increases in both private philanthropic and governmental contributions to hospitals, especially between 1940 and 1970, see Caryl P. Haskins, "The Role of Private Philanthropy and Public Support of Science in the United States," in 2 Research Papers Sponsored by the Commission on Private Philanthropy and Public Needs 631, 642-43 (Dept. of Treasury 1977) (tabs. 1 and 2). Hospitals, as frequent targets of medical malpractice suits, were the principal beneficiaries of the charitable immunity. See also Eli Ginzberg, Hospitals and Philanthropy, in Philanthropy and Public Policy 73, 91-92 (Frank G. Dickinson ed., Natl. Bur. Econ. Research 1962).

182 · The Economic Structure of Tort Law Only if Β is liable to C for the latter's injury will both Β and C have incentives to refrain from making inefficient expenditures (C to avoid spending on self-protection, which is a more costly alternative than B's refraining from shooting A). Efforts have occasionally been made, notably in the Pinto accident litigation, to attach the label of "intentional" to the case where the injurer consciously adverted to the possibility (which, depending on the scale of the activity, may approach certainty) that someone would be injured. During the development of the Pinto automobile Ford Motor Company made a cost-benefit analysis of the Pinto's fuel tank. The analysis predicted that hundreds of additional deaths and serious injuries would occur unless the fuel tank was altered. This deliberate advertence to the possibility—indeed the virtual certainty—of death and injury was the basis for the prosecution in Indiana of Ford as a criminal for its decision not to alter the fuel tank.52 This attempt to convert the accidental tort of a large company into the equivalent of an intentional tort because the scale of the activity made it fairly certain that accidents would occur failed, as from an economic standpoint it should have. 53 Another question about the mental element in intentional torts concerns the defense of insanity. There is in general no such defense in intentional tort cases just as there is none in accidental tort cases. 54 The usual reason given for this result is that the element of intent required to make out a prima facie case of intentional tort is so attenuated (an intent to injure but not necessarily to injure seriously or wrongfully) that normally even an insane person can form the requisite intent and so can be deterred by the threat of a legal sanction.55 But where the 52. The case is discussed in Richard A. Epstein, "Is Pinto a Criminal?" Regulation, March/April 1980, at 15. Grimshaw v. Ford Motor Co., 119 Cal. 3d 757, 174 Cal. Rptr. 348 (1981), upheld an award of $3.5 million in punitive damages in a suit against Ford for a fatal accident resulting from the fragility of the Pinto's fuel tank. The evidence relied on by the court showed that Ford could have eliminated the safety hazard, which its own engineers considered very great, at a trivial cost—perhaps as little as $4 per car. The facts indicated gross negligence, and the economic model suggests that an award of punitive damages is appropriate in such a case. A similar case, involving the Mustang's fuel tank, is Ford Motor Co. v. Stubblefield, 171 Ga. App. 331, 319 S.E.2d 470 (1984). 53. This is not to say that reckless conduct by corporations should never be punished criminally (see Posner, note 10 above, at 1228-29); it merely indicates that the first meaning of "intentional" analyzed earlier in this chapter is not a sound basis for imposing either tort or criminal liability. 54. See Keeton et al., note 16 above, at 1072-74. 55. On the economics of the insanity defense in criminal law see Posner, note 10 above, at 1223-25.

Intentional Torts and Damages · 183 insanity is of such a nature as to prevent the injurer from forming the requisite intent, he may be excused. 56 There is a possible economic justification for the distinction. As we have seen, requiring the injurer to compensate the victim both deters intentional torts and incidentally deflects potential victims from taking precautions that would reduce the parties' joint wealth. But a person who is not deterrable by reason of insanity will not be led to this result by liability. Assume that the harm to the victim is $100 and the gain to the insane injurer $5. The ideal solution would be that the tort not be committed; this is the result when a "rational" injurer is made liable for $100 (and there is no uncertainty as to whether he will be made to pay). But if the injurer is not fully deterrable, he will act as if his expected liability is, say, only $2 (k x D where k = .02 is the discount factor due to insanity) and commit the tort. Suppose the victim could spend $5 to reduce his damages from $100 to $10. If he were fully compensated, he would have no incentive to do this, because the $90 reduction in injury costs would be exactly offset by a $90 reduction in his damage award. But because the injurer is insane, the $5 expenditure is efficient; it leads to a net reduction in the joint loss of the two parties of $85. If the victim is not compensated, he will have an incentive to make the $5 expenditure. Although this outcome is less efficient than if the tort were not committed in the first instance, it is efficient given that the injurer cannot be deterred. This analysis is incomplete, however, because it neglects the effect of liability in giving an incentive to the custodians and beneficiaries of the insane person's wealth to prevent him from committing torts. This is the traditional rationale for making the insane liable. An ingenious accommodation is suggested by the forewarning cases such as Breunig v. American Family Insurance Co., 57 discussed in Chapter 4; lack of forewarning shows that liability would have no deterrent effect either on the insane person himself or on those who have an interest in conserving his wealth. Although Breunig was an accident case rather than an intentional tort case, from an economic standpoint the question of the liability of the insane has little to do with whether the insane person is charged with an intentional or an unintentional tort. It is therefore not surprising that the cases do not distinguish clearly between the two types of tort in this regard. This is further evidence that the law gives "intentional" a functional rather than a literal signification. It worries 56. See Keeton et al., note 16 above, at 1074. 57. 45 Wis. 2d 536, 173 N.W.2d 619 (1970).

184 · The Economic Structure of Tort Law less about the mental state than about how to optimize losses due to injuries.

Punitive

Damages

Our analysis implies that punitive damages should not be awarded in all cases of intentional tort but only in those that involve deliberate wrongdoing or recklessness. (Gross negligence is a borderline case.) If although classified as intentional the tort was committed through a mistake that may have been careless but was not reckless, punitive damages would be inappropriate. This is the pattern found in the cases, as summarized in the leading treatise on torts: "Where the defendant's wrongdoing has been intentional and deliberate, and has the character of outrage frequently associated with crime, all but a few courts have permitted the jury to award in the tort action 'punitive' or 'exemplary' damages." 58 Besides cases of deliberate wrongdoing, punitive damages are available when the defendant's conduct indicates "such a conscious and deliberate disregard of the interests of others that the conduct may be called willful or wanton" 59 or "reckless," the latter defined as a type of behavior between "proceeding with knowledge that the harm is substantially certain to occur, and the mere unreasonable risk of harm to another involved in ordinary negligence." 60 But punitive damages are not automatically awarded in intentional tort cases, because "it is not so much the particular tort committed as the defendant's motives and conduct in committing it which will be important as the basis of the award." 61 To determine whether the formal legal standard has been consistently applied, we examined a sample of recent tort cases in which awards of punitive damages were affirmed on appeal. The sample consists of all the common law tort decisions affirming such awards reported in the most recent volume of each of the West Publishing Company's case reporters, forty-six cases in all (a list is available from the authors on request). As expected, most involve deliberate torts such as fraud, retaliatory discharge, libel, and battery: thirty-three of the forty-six cases 58. Keeton et al., note 16 above, at 9 (footnote omitted, emphasis added). See also Dobbs, note 14 above, at 2 0 4 - 2 1 , for a succinct discussion of the law of punitive damages. 59. Keeton et al., note 16 above, at 10. 60. Id. at 212. 61. Id. at 11 (footnote omitted). See also C. T. McCormick, Handbook on the Law of Damages 280 (1935). For economic analyses of punitive damages see "Symposium: Punitive Damages Articles," 56 So. Calif. L. Rev. 1 (1982).

Intentional Torts and Damages · 185 are of this type (72 percent). This is 23 percent of the intentional tort cases in the volumes examined. The other thirteen cases where punitive damages were upheld are accident cases. As expected, this is a much smaller percentage of all accident cases in the volumes examined: only 2 percent. Thus punitive damages are more than ten times as likely to be awarded in intentional than in unintentional tort cases. And accident cases in which punitive damages are awarded are probably overrepresented in a sample of appellate cases. The award of punitive damages in such cases is more problematic than in cases of deliberate injury, so they are more likely to generate an appeal on the issue of punitive damages. In all but two of the thirteen accident cases, the evidence of gross negligence or recklessness is plain. In one, the evidence is borderline: the defendant was held to be grossly negligent for having served liquor at its bar to a visibly drunk customer who it was known was going to drive away from the bar.62 In the other, a case of dental malpractice, the award of punitive damages seems wrong, at least on the facts recited in the court's opinion; they show simple negligence and the opinion itself wavers between describing the defendant's conduct as negligent and grossly negligent.63 Nevertheless the cases as a whole are generally congruent with the formal legal standard for awarding punitive damages, which in turn is consistent with the economic model.64

Tort Damages in General In considering the matter of punitive damages we have found a congruence between tort damage law and economic analysis. We would find the same congruence if we expanded our canvass to take in the methods by which courts value lost earnings in tort cases where the victim is killed or disabled.65 Courts make subtle, if sometimes imperfect, 62. Pfeifer v. Copperstone Restaurant & Lounge, Inc. 71 Ore. App. 599, 693 P.2d 644 (1985). 63. Costa v. Storm, 682 S.W.2d 599 (Tex. Civ. App. 1984). 64. The much greater likelihood of an award of punitive damages in intentional than in nonintentional tort cases is confirmed by a recent empirical study of jury verdicts in San Francisco and Cook County, Illinois. See Mark A. Peterson, Punitive Damages: Preliminary Empirical Findings 16-17 (Rand N-2342-ICJ Aug. 1985). We discuss punitive damages further in Chapter 10, which deals with products liability. 65. See Lloyd Cohen, "Toward an Economic Theory of the Measurement of Damages in a Wrongful Death Action," 34 Emory L. J. 295 (1985); Richard A. Posner, Economic Analysis of Law §§ 6.11-6.13 (3d ed. 1986); Keith S. Rosenn, Law and Inflation 220-34 (1982);

186 · The Economic Structure of Tort Law adjustments for inflation and productivity. They also realize that an award of lost earnings in a disability case ought to replace the plaintiff's actual earnings rather than be an average figure66—an example of the unwillingness of the courts to use tort law to redistribute wealth. By awarding damages for "pain and suffering," including disfigurement, tort judges and jurors implicitly recognize, as many critics of tort law do not, that pain and the kind of suffering that ensues when one loses an arm or a leg, is blinded, or suffers some other grievous and permanent (or long-lasting) injury inflict real even though nonpecuniary cost. 67 They also recognize, as we noted in Chapter 2, the disincentive effects of awarding damages periodically rather than in a lump sum. Although in these and other ways the damage rules of the tort system display a characteristic economic intuition, in one very important respect they seem out of phase with economics. This is the matter of valuing a life in a tort case where the victim dies ("wrongful death"). 68 Primitive and early societies, including Anglo-Saxon England, realized that a fatal injury inflicts a real loss for which compensation ought to be paid. But at some point in the evolution of the English common law the rule was changed and a tort was deemed to die with the plaintiff.69 Whether the tort killed him or he died of unrelated causes before he won a tort judgment, he got nothing. All this may conceivably have made some sense on the assumption that the only function of damages in tort law is to make provision for consumption expenditures. If the tort victim dies he has no more consumption needs; his survivors have, Gary A. Anderson and David L. Roberts, "Economic Theory and the Present Value of Future Lost Earnings: An Integration, Unification, and Simplification of Court Adopted Methodologies," 39 U. Miami L. Rev. 723 (1985); Ralph J. Brown and Dennis A. Johnson, "Wrongful Death and Personal Injury: Economics and the Law," 29 S. Dak. L. Rev. 1 (1983); Comment, "Inflation, Productivity and the Total Offset Method of Calculating Damages for Lost Future Earnings," 49 U. Chi. L. Rev. 1003 (1982); Sumner J. LaCroix and H. Laurence Miller, Jr., "Lost Earnings Calculations and Tort Law: Reflections on the Pfeifer Decision," 8 U. Hawaii L. Rev. 31 (1986); Culver v. Slater Boat Co., 723 F. 2d 114 (5th Cir. 1983) (en banc). 66. This is really just an application of the eggshell-skull rule, because having a high income, like having a thin skull, makes one more vulnerable to a serious injury from an accident or attack. 67. See Neil K. Komesar, "Toward a General Theory of Personal Injury Loss," 3 ]. Legal Stud. 456 (1974). 68. See Keeton et al, note 16 above, ch. 23, on the law of damages for wrongful death. 69. For good discussions of the history, see Wex S. Malone, "The Genesis of Wrongful Death," 17 Stan. L. Rev. 1043 (1965); T. A. Smedley, "Wrongful Death—Bases of the Common Law Rules," 13 Vand. L. Rev. 605 (1960); Percy H. Winfield, "Death as Affecting Liability in Tort," 29 Colum. L. Rev. 239 (1929); Gustavus Hay, Jr., "Death as a Civil Cause of Action in Massachusetts," 7 Harv. I. Rev. 170 (1893).

Intentional Torts and Damages · 187 but in a poor society even the breadwinner may earn only trivially more than he consumes. The economic purpose of tort law, however, is not to keep tort victims off the dole; it is to internalize the costs of accidents. By the middle of the nineteenth century the law in England and America had come around to awarding damages in death cases. But it is significant that in both countries statutes were required, the common law courts having been adamant in their refusal to award damages for wrongful death. The statutes, however, were deficient in two respects: they placed artificially low ceilings on the total dollar recovery, and they limited that recovery to the survivors' pecuniary loss—the present value of the decedent's lost future earnings minus the present value of his cut-off personal consumption. The first deficiency was eventually cured by statutory amendments but the second remains a general feature of wrongful-death statutes, occasionally modified for accident victims (such as children) who have no measurable future earnings. 70 The limitation of damages to survivors' pecuniary loss is very peculiar. It implicitly assumes—if, as we generally believe to be the case, tort law seeks to internalize the costs of accidents—that the average person derives no utility from living. He does not work for himself, he works solely for his family. This cannot be right, and it results in a systematic underestimation of damages in wrongful-death cases. Against trying to correct this deficiency in the law of tort damages it can be argued that there is no way to determine what the utility of a person's life is. Perhaps that utility cannot be measured even in principle. If you asked someone how much money he would demand to give up his life, he probably would answer that no finite amount of money would be enough. Yet we know that people do not really set an infinite value on their lives. If they did, they would invest much more in avoiding physical dangers than they do. The answer to this puzzle, which provides a clue to the proper measure of damages in a death case, requires recognition of the nonlinear relationship between the value of life and the risk of death. As is apparent from even the most casual observation, people will often assume very slight risks of death for very small monetary (or monetizable) gains; this is obvious just from watching people cross the street. As the risk increases, however, the compensating equivalent rises at a faster rate until it becomes (for people who are not extreme altruists) infinite when the 70. Another inexplicable feature of the wrongful-death statutes—and one that survives in most states to this day—is that punitive damages cannot be awarded in cases of wrongful death. See Ellis, note 14 above, at 26 n. 122.

188 · The Economic Structure of Tort Law

Figure 6.1

risk becomes a certainty. This relationship, shown graphically in Figure 6.1, may seem to preclude the rational estimation of D, for purposes of applying the Hand formula, in any case where the victim dies. But it does not; all it shows is that D in such cases is not independent of p, the risk of the (fatal) accident. Actually, the analysis shows how the tort system could estimate optimal damages in a death case. To use the Hand formula to determine negligence, one does not actually have to know what D is; all one has to know is the cost of an additional precaution (By) and how much the potential victim would demand to accept the added risk of death if the precaution is not taken. 71 Suppose we know that the average person would demand $1 to assume a .00001 increase in the probability of being killed. Then $1 is the expected reduction in damages (pyD) from an extra precaution that would eliminate the risk. Since we know pyD and py (.00001), we can compute D at the relevant level of risk; it is $100,000. 72 So if there is some precaution (By) that costs less than $1 and if a potential injurer fails to take it and a fatal accident results, the correct damage award is $100,000. This is not to say that the victim's life is worth that amount—that would be a meaningless assertion—but rather that this is the amount of damages that if awarded to the victim will impart the correct incentives to potential injurers. Because there is a good deal of statistical evidence on the compensation that people demand to assume small risks of death—which are the kind involved in most accident cases (if ρ is high, as we have seen, the case is apt not to be regarded in law as a mere accident case)—this may be 71. See Posner, note 65 above, § 6.12. 72. From the nonlinear relationship in Figure 6.1 we can write expected accident damages as p(y)D(p) where D is a positive function of p. The social loss (assuming an alternative care case where x* = 0) is L = p(y)D(p) + B(y), and due care requires pyD + p(8D/8p)py + By = 0, or — p y D(l + E) = By, where £ is the elasticity of damages with respect to p. Ε is positive and increases with ρ (and approaches infinity in Figure 6.1 as ρ approaches 1). The example in the text assumes that the average individual would demand $1 to bear the increase in the risk of death of .00001. Thus, D(1 + E) = $100,000, which would approximately equal D if the probability of death is so low that the elasticity is close to zero. Even if Ε were substantially above zero, damages equal to $100,000 would create the correct incentive for the injurer to take due care.—to take the precaution if it costs less than $1.

Intentional Torts and Damages · 189 a feasible as well as theoretically correct method of estimating tort damages. The tort system shows, as yet, no signs of moving in this direction; it continues to be wedded to the pecuniary-loss measure, which bears no necessary relationship to the economically correct measure. Several mitigating factors should be noted, however: the passage of wrongfuldeath statutes in the nineteenth century may have stifled common law development; only recently has it become clear how tort damages ought to be calculated in death cases; and the statistical estimates of the compensating differentials that people demand to assume a risk of death or serious injury vary so greatly that it is not clear that the method is yet usable. 73 We end this chapter with two observations based on the nonlinear relationship between the risk of death and the value of life. First, this relationship has an important implication for the analysis of intentional torts. In cases where the tortfeasor is trying to kill the victim or is deliberately creating a high risk of death, the optimal damages soar, because ρ is close to one. 74 But this fact is not reflected in any adjustment of compensatory damages in death cases; they are measured the same way in intentional as in unintentional wrongful-death cases. This is contrary to the implications of economic theory but is subject to the same mitigating circumstances just noted with respect to the calculation of damages in cases of accidental death. And maybe the necessary adjustment is made via awards of punitive damages. Second, the fact that optimal damages in many intentional-tort cases are thus very high helps to explain the overlap between criminal and tort law. The economic theory of criminal law suggests that the essential purpose of that law is to control harmful externalities inflicted in circumstances where damage remedies are insufficient, normally because the optimal measure of damages (ignoring problems of collection) exceeds the defendant's wealth. 75 This condition is most likely to be fulfilled in cases of deliberate wrongdoing as we have defined this term—and the principal common law crimes are of this character.76

73. For recent summaries of the estimates, showing enormous disparities, see Glenn Blomquist, "The Value of Human Life: An Empirical Perspective," 19 Econ. Inquiry 157 (1981); Albert L. Nichols, Jr., The Regulation of Airborne Benzene, in Incentives for Environmental Protection 145, 1 7 3 - 7 5 (Thomas C. Schelling ed. 1983). On the general problem of estimation see Richard Zeckhauser, "Procedures for Valuing Lives," 23 Pub. Policy 419 (1975). 74. See also Ellis, note 14 above, at 32 n. 148. 75. See Posner, note 10 above; Steven Shavell, "Criminal Law and the Optimal Use of Nonmonetary Sanctions as a Deterrent," 85 Colum. L. Rev. 1232 (1985). 76. See Posner, note 10 above, at 1199-1200.

.

7

.

Joint and Multiple Torts

other than a self-inflicted one, is caused in an economic sense by at least two people—the injurer and the victim. Recognition of this point is fundamental to the Coase theorem. In discussing contributory negligence, assumption of risk, and changes in activity levels by potential victims of nuisances and accidents, we have seen how the law recognizes that the victim is often a lower-cost avoider of injury to himself than the injurer is. 1 The focus of this chapter, however, is on cases in which the victim's injury is caused by more than one injurer. 2 Such cases are of two basic types, the simultaneous joint tort and the successive joint tort. In the first type the victim suffers a single or indivisible injury as a result of the tortious activity of two or more other persons. Simultaneity is not a strict requirement of this type of joint tort, but in the usual case either the tortious acts of the injurers occur at the same time or a single act is a tort by both injurers—as when an employee injures the plaintiff, and his employer is also liable under the doctrine of respondeat superior. The simultaneous joint tort, like a single-injurer tort, can involve either joint care or alternative care. Recall that in a case of joint care optimal accident avoidance requires both parties (here, both injurers) to take care. Two pertinent examples are: a party wall will collapse unless the neighbors who share the wall each maintain it; two hunters each inflict EVERY INJURY,

1. One of the original rationales of the doctine of contributory negligence was that the negligent victim is in effect a joint tortfeasor with the injurer. See William Schofield, "Davies v. Mann: Theory of Contributory Negligence," 3 Harv. L. Rev. 263, 267 (1890). 2. For good summaries of the relevent legal principles see 1 Fowler V. Harper and Fleming James, Jr., The Law of Torts, ch. 10 (1956); W. Page Keeton et al., Prosserand Keeton on the Law of Torts, ch. 8 (5th ed. 1984).

Joint and Multiple Torts · 191 a fatal wound on a third. In an alternative care case optimal accident avoidance requires that only one potential injurer take care. For example, one person drops a match into a can of gasoline that someone else negligently left lying about, causing an explosion that injures a third; or a product defect causes an injury that would have been prevented if either the manufacturer of the defective component had been more careful or the manufacturer of the final product had inspected its components more carefully. A joint tort is successive rather than simultaneous if one tortfeasor aggravates an injury inflicted by the other, as where a driver negligently hits a pedestrian and a physician negligently treats, thereby aggravating, the pedestrian's injury. The difference between this case and the uninteresting case of two completely separate injuries is that if the first tortfeasor had used due care the entire injury to the plaintiff—initial plus incremental damages—might have been avoided, whereas if only the second injurer had used due care only the incremental damages could have been avoided. The principal legal categories that control liability in the joint and multiple tortfeasor contexts are contribution (or no contribution), indemnity, and proximate cause. Where the victim suffers a single injury as a result of the negligence or other wrongdoing of two or more tortfeasors, traditional common law denies contribution. 3 The plaintiff can sue and collect his full damages from either joint tortfeasor (for simplicity, we assume there are only two, but the principle holds for any number) or both. If he obtains a judgment against both, he can satisfy it in whatever proportions he chooses from either or both; the only limitation is that his total recovery may not exceed the amount of the judgment. And a defendant who pays the greater part (perhaps all) of the judgment has no right to obtain any reimbursement from the other defendant, let alone from joint tortfeasors not sued by the plaintiff. In some cases, however, a tortfeasor from whom the plaintiff has obtained partial or full payment of the judgment can get back what he has paid from the other tortfeasor, whether or not that tortfeasor had been a defendant in the plaintiff's case. This is indemnity. It is not contribution because there is no sharing of the judgment cost between the defendants: a defendant entitled to indemnity is entitled to full indemnity. 4 3. A few states have changed the traditional common law approach by judicial decision. Many states, however, have done so by statute. See Table 7.3. 4. The counterpart principle for the special case where the second tortfeasor is also the

192 · The Economic Structure of Tort Law In the case of successive joint torts the second tortfeasor is liable for the aggravation, not the whole, of the plaintiff's injury, but the first tortfeasor is liable for the entire injury if the possibility of negligent aggravation by another was reasonably foreseeable. This result is uncontroversial, but the rule of no contribution as applied in simultaneous joint tort cases has been denounced by most commentators. Prosser and Keeton find an "obvious lack of sense and justice in a rule which permits the entire burden of a loss, for which two defendants were equally, unintentionally responsible, to be shouldered onto one alone, according to the accident of a successful levy of execution, the existence of liability insurance, the plaintiff's whim or spite, or the plaintiff's collusion with the other wrongdoer, while the latter goes scot free." 5 No contribution has also been criticized from an economic standpoint, on the grounds that deterrence "would seem maximized by the certainty that each joint tortfeasor will bear some of the loss"; "the optimum way to achieve deterrence would be to divide the loss between" the negligent injurers; where two products are equally the cause of harm, "since there is little justification for imposing the loss on one product over the other, the loss is best split between them, thereby giving the consumer some additional information about each product through the price/market mechanism"; even where there is a gross disparity in the deterrability of the two tortfeasors—the case where the common law allows indemnity—there should be apportionment so that the less culpable party will be deterred as well as the more culpable one. 6 victim is that the plaintiff's negligence is either a complete bar to the victim's recovery or it is no bar; there is again no apportionment of damages. This is the common law rule of contributory negligence, which as noted in Chapter 3 is giving way to comparative negligence—the counterpart to contribution. 5. Keeton et al., note 2 above, at 3 3 7 - 3 8 . This passage, from the 1984 edition, is unchanged (except for trivial wording changes) from the 1971 edition, the last before Prosser's death. See William L. Prosser, Handbook of the Law of Torts 307 (4th ed. 1971). The distinguished authors of the fifth edition have not kept fully abreast of the economic analysis of their subject; our analysis of contribution, to which they make no reference, was first published in 1980. See William M. Landes and Richard A. Posner, "Joint and Multiple Tortfeasers: An Economic Analysis," 9 ]. Legal Stud. 517 (1980). They do, however, take brief note of the economic analysis of the negligence standard. See Keeton et al., note 2 above, at 173 n. 46. 6. Comment, " T h e Allocation of Loss among Joint Tortfeasors," 41 So. Calif. L. Rev. 728, 730, 733-34, 7 4 6 - 4 7 (1968). The idea that liability must be apportioned among all who contribute to an injury has been advanced by no less distinguished an economist than Kenneth Arrow: "Another problem is that the notion of liability in law is really too simple a concept. Legal liability tends to be an all-or-none proposition. Consider a product such as plastic bags. They are perfectly all right for storing clothes or food but there is a risk

Joint and Multiple Torts · 193 Both Prosser and Keeton's criticism and the economic criticism rest on a confusion between ex ante and ex post. The position of the joint tortfeasors is indeed unequal ex post if they pay unequal shares of the judgment, but it does not follow that it is unequal ex ante. The expected judgment cost may be the same to each tortfeasor; or if greater to one, it is apt to be greater for reasons (such as greater wealth) that would also operate under the system of contribution, because if one joint tortfeasor is unable to pay his share of the judgment the shares of the others are increased to make up the difference. The distinction between ex ante and ex post relates not only to the issue of fairness as framed by Prosser and Keeton but also to the issue of economic efficiency. If, ex ante, each defendant bears a cost (an expected cost) of liability, each defendant will be deterred, even if ex post all but one pay nothing. Whether each defendant is deterred adequately, however, is a more subtle question.

Simultaneous Joint Torts Theoretical Analysis 1. The joint care case. In the joint care case 7 we can write the loss function, L, as that small children will misuse them, with serious consequences. One would hardly want to say that there is any legal liability ascribable to the plastic-bag makers, for even the safest product can be misused. On the other hand, one might argue that a product that can be misused ought to be somewhat discouraged and perhaps some small degree of responsibility should be imposed, particularly if no adequate warning is issued. The law does not permit any such distinctions. Thus, in an automobile case, one party or the other must be found wrong, even though in fact a crash may clearly be due to the fact that both drivers were behaving erratically, and it would be reasonable to have some splitting of responsibility. At present, with some minor exceptions, the law does not permit this, and 1 suppose it would confuse legal proceedings irreparably to start introducing partial causation. Economists are accustomed to the idea that almost nothing happens without the cooperation of a number of factors, and we have large bodies of doctrine devoted to imputing in some appropriate way the consequences of an action to all of its causes. It is for these reasons that this kind of crude liability doctrine seems to be unsuitable in many cases." Kenneth J. Arrow, "Social Responsibility and Economic Efficiency," 21 Pub. Policy 303, 312-13 (1973). 7. We use the following notation, which is similar to that in Chapter 3: χ = injurer A's level of care (x a 0) y = injurer B's level of care (y a 0) 2 = victim C's level of care (z a 0) = A's cost of care (Ax > 0 and Axx a 0)

194 · The Economic Structure of Tort Law L(x,y,z) = p(x,y,z)D + A(x) + B(y) + C(z)

(7.1)

and define χ*, y*, and z* as the levels of care that minimize L. Under a negligence standard, therefore, injurers will not be liable for the victim's loss provided that χ > χ* and y > y*; and even if the injurers' care levels fall below x* and y*, they will not be liable if the victim is contributorily negligent (z < z*). Thus, as shown in the first row of Table 7.1, the victim bears his full damages when he is contributorily negligent, whatever the care taken by injurers. He also bears the full loss when all parties use due care (row 2); but the full loss is shifted to the negligent injurer (A in row 3 and Β in row 4) when the other injurer and the victim use due care. The last row illustrates the typical joint care problem for a nonseparable injury: both injurers are negligent, the victim is not contributorily negligent, and the allocation of the victim's loss between A and Β is uncertain. Let s" and sb be A's and B's respective fractions of the damages, given that both are negligent and C uses due care. Although s" and sb sum to one, under the common law rule of contribution their precise values are uncertain and depend on each injurer's estimate of how the victim will apportion damages between them. More precisely, s" will equal the probability that the victim sues only A plus the probability that both A and Β are sued times A's expected fraction of the damages given that both are sued; sb is defined similarly. The equilibrium level of victim care in Table 7.1 is always z*. This requires some explaining. First, imagine that the victim, C, used less than due care, say care z0. His expected accident and avoidance costs would equal p(x,y,z0)D + C(z0) and would thus depend on the levels of care chosen by A and B. The greater their care, the lower would ρ be and hence the lower the victim's expected costs. Since A and Β will never use more than x* and y* care respectively, C's expected costs, given that he uses z0 care, will be lowest when the care levels of A and Β are x* and y*. But if C uses due care, his expected accident and avoidance costs will be highest when both A and Β also use due care, because in that case, under a negligence standard, none of the victim's loss is shifted to the injurers. The victim's expected loss will equal p{x*,y*,z*)O + C(z*), compared with costs of only C(z*) when he uses due care and one or both injurers use less than due care and hence are B(y) C(z) p{x,y,z) D L(x,y,z)

= = = = =

B's cost of care (By > 0 and Bn > 0) C's cost of care (C ; > 0 and C„ > 0) probability of an accident (px,ps,p2 < 0, p„, pm, p21 > 0) C's damages from accident expected costs of accident and care

Joint and Multiple Torts · 195 Table 7.1. Care levels

Nonseparable injury—expected accident losses A

Β

,x,y z*,x*,y*

A(x) A(x*)

Z*,xa ,y0

p(x0,y*,z*)D + A(x0) A(x*) sy(x0,y0,z*)D + A ( A C )

B(y) B(y*) B(y*) p(x*,y0,z*)D + B(y0) sbp(x0,y0,z*)D + B(y0)

ZO

0

C p(x,y,z0)D + C(z0) p(x*,y*,z*)D

+ C(z*)

C(z*) C(z*) C(z · )

Note: x0 < x*, y0 < y*, and z„ < z*.

liable for the victim's accident losses, D. To show that the victim has an incentive to use due care, therefore, one must show that the victim's best possible outcome under the first situation (both injurers use due care and the victim uses z0) is worse than his worst possible outcome in the second situation (all parties use due care)—that p(x*,y*,z0)D + C(z0) > p(x*,y*,z*)D + C(z*). But this follows simply from our definition of x*, y*, and z* as the values of care that minimize L. Because a negligence standard, with levels of care appropriately defined, will lead the victim to choose the optimal care level, ζ*, each injurer will select a level of care that minimizes his expected losses from liability plus costs of accident avoidance on the assumption that the victim chooses z*. But what will the injurers' levels of care be? Table 7.1 shows that A's expected loss (due to liability for C's damages) depends, in part, on the level of accident avoidance of the other potential injurer, B. Thus if Β chooses y*, A's expected loss will equal either p(x0,y*,z*)D + A(x0) or A(x*), depending on whether A was negligent (row 3) or nonnegligent (row 2). From the definition of x*, y*, and z* as the values of care that minimize L, it follows that p(x0,y*,z*)D + A(x0) > p(x*,y*,z*)O + A(x*) > A(x*), and hence that A would always choose due care if Β were expected to do likewise. If A expected Β to behave negligently (y0), however, he would choose between x0 and x* depending on whether s"p(x0,y0,z*)D + A(x0) > or < A(x*). This choice appears uncertain. If A's expected share of damages, s", is relatively small, it may seem that his savings in costs of care, A(x*) - A(x0), from behaving negligently might more than offset his expected liability losses of s"pD. And by similar reasoning Β might prefer to behave negligently if A was negligent. Further analysis shows, however, that x*,y* is the equilibrium solution under no contribution. A's expenditure on accident avoidance will be less than the due care level (assuming A anticipates that Β is negligent) if

196 · The Economic Structure of Tort Law s°p(x0,y0,z*)D + A(x0) < A(x*).

(7.2)

Similarly, Β will use less than due care (assuming Β expects A to be negligent) if s»p(x0,y0,z*)D + B(y0) < B(y*).

(7.3)

Inequalities (7.2) and (7.3) imply that (s" + sb)p(x0,y0,z*)D + A(x0) + B(y0) < A(x*) + Β (y*).

(7.4)

Because s" + sb = 1, and x*, y*, and z* are the levels of care that minimize L, (s° + sb)p(x0,y0,z*)D + A(x0) + B(y0) > p(x*,y*,z*)D + A(x*) + B(y*) > A(x*) + B(y*). Therefore, (7.4) is a contradiction and inequalities (7.2) and (7.3) cannot simultaneously hold. Suppose (7.2) is true but (7.3) false. Then Β will select y* and from our prior analysis this will induce A to select x*. Alternatively, if (7.2) is false but (7.3) true, A will select x* and Β will be led to choose y*. To summarize, when the negligence standard is combined with the no-contribution rule in a case involving joint care, all parties have incentives to select levels of care that minimize the total cost of an accident. The intuitive explanation is simple: if the total cost of care is less than the expected accident cost—as it must be if failure to take care is negligence—it will always be in the interest of at least one of the tortfeasors to behave carefully and thereby place the whole liability on the other; and once one behaves carefully the other has an incentive to do likewise. Suppose that A and Β share a party wall and C will be injured if it collapses as a result of the failure of A and Β to maintain it adequately. The expected accident cost to C is $100 and the optimal expenditure on maintenance by A and Β is $40 each. The legal rule is no contribution, and let us assume that if A and Β spend nothing on maintenance the expected liability of each is $50, one-half the expected damages of C. 8 A and Β then know that by spending $40 each can reduce his expected liability to zero, and knowing this each will spend $40. The assumption that optimal care requires equal expenditures on accident avoidance by the potential injurers is inessential. Suppose that the optimal method of avoiding the $100 of expected accident costs is 8.This does not necessarily imply that C will get a judgment against both and seek to satisfy it one-half from A and one-half from B. Maybe C will collect the entire judgment from one or the other but the probability that C will levy judgment against A is the same as the probability he will do so against B. Or maybe C will collect part of the judgment from each and the probability that he will do this times the share collected is the same number (an expected damage figure) for each.

Joint and Multiple Torts · 197 for A to spend $79 and Β $1, and as before that the expected liability of A and Β is $50 each. Clearly Β will spend $1 to avoid an expected liability of $50, and once he has done so A will be faced with the prospect of having to pay $100 unless he spends $79 on care—which he will do. The assumption that the expected liability of A and Β is the same is also inessential. Imagine that for some reason A expects to bear 80 percent of the liability cost and Β 20 percent (maybe A is much more solvent than B). Then in the first of our numerical examples, A initially will have the incentive to take care, and once he does so Β will have an incentive to do likewise; in the second, Β will have the initial incentive to take care, followed by A. The only important thing is that the sum of the parties' expected liabilities equal or exceed the expected accident costs. So far we have analyzed only the variable-proportions case, where each party's care has a positive marginal product. But sometimes the probability of an accident depends only on the minimum care taken by the parties. Suppose that A's care level in the party wall illustration falls below x*. It may be that no matter how much added care Β takes in buttressing his side of the wall, the wall will collapse. Or suppose that A and Β simultaneously throw lighted matches into a room containing explosives where a single match would cause an explosion: the impact of the second match on the probabability of an explosion is zero. 9 To model the fixed-proportions case, we can write ρ = p(min x, min y, z*). Because increasing χ when y is less than χ has no effect on p, and χ has a positive cost, it would never be efficient to have a level of care where y is less than x. An identical argument can be made for y. Hence the efficient care level is where x* = y*. With this definition of due care, the analysis of how a negligence standard and a no-contribution rule create incentives for due care in the fixed-proportions case is then identical to that of variable proportions, modeled above. Our analysis assumes that if one party takes optimal care the other will have an incentive to do the same. But as we know from Chapter 3, this assumption may not always hold even if negligence and due care are correctly defined from an economic standpoint. To recapitulate: if care is stochastic; or if negligence is applied on an average rather than 9. Cases illustrating this point in the context of fires set by different tortfeasors that join and then damage the plaintiff's property are numerous, although not recent. See McClellan v. St. P., Μ. & M. Ry., 58 Minn. 104, 59 N.W. 978 (1894); Corey v. Havener, 182 Mass. 250, 56 N.E. 69 (1902); Oulighan v. Butler, 189 Mass. 287, 75 N.E. 726 (1905); Anderson v. Minneapolis, St. P. & S. S. M. Ry., 146 Minn. 430, 179 N.W. 45, 49 (1920) (dictum); Kingston v. Chicago, N.W. Ry., 191 Wis. 610, 211 N.W. 913 (1927).

198 · The Economic Structure of Tort Law on an individual basis, so that some prospective injurers are not deterred because they are below average in their capacity for care; or if negligence contains other pockets of strict liability (such as the doctrine of respondeat superior); or if judges or juries make errors in applying the negligence standard to particular cases, then one party's taking optimal care may not ensure that the other will do so. Suppose that an expected accident cost of $100 can be avoided if A spends $20 on care and Β $30, and the expected liability cost of each if they fail to take care is $50. Suppose further that if Β spends nothing on care the accident will occur whatever A spends on care. But Β thinks that because of legal error he will not be found negligent and therefore does not take care. Knowing this, A will probably spend nothing on care either because A will not be liable for the accident that results from B's "nonnegligent" behavior. Or suppose that although the average person in B's position could (jointly with A) avoid the accident at a cost of $30, the cost to Β would be $60. Β will not take care. A may still spend $20 on care to avoid being held liable for the accident (assuming he could not defend by showing that the accident would have occurred anyway as the result of someone else's negligence), but the accident will occur anyway, and the $20 will be wasted from a social standpoint. Although these are inefficient results—at least if the costs of avoiding them through more finely tailored or accurately administered tort rules are ignored—they do not provide a basis for preferring contribution to no contribution. (They do provide an argument against strict liability.) The same results would obtain under a rule of contribution where the damages were divided by the number of defendants. Other types of contribution rules may yield better results in the particular examples we have given, but we can think of no rule that would produce results allocatively superior overall to those of no contribution. 2. The alternative care case. In contrast to the joint care case, optimal care in the alternative care case requires that only one of the tortfeasors take care—the one whose costs of care are lower. Assume that A is the lower-cost avoider of the accident, at all levels of care; that A's and B's inputs of care are additive (r = χ + y); and that their costs of care are ax and by respectively, where a < b. Expected accident and avoidance costs are L = p(r,z)D + ax + by + C(z).

(7.5)

Because a < b, due care requires that y* equal, and x* and z* exceed, zero. This result can be achieved by a negligence standard combined

Joint and Multiple Torts · 199 with an indemnity rule that shifts all liability eventually to A, although both A and Β would be treated (initially) as joint tortfeasors. Β has no incentive to spend anything on care, because if A takes due care both A and Β will avoid any liability for C's injury; that is, A's due care implies that there is no negligence on the part of either A or Β and therefore neither is liable. Alternatively, if A is negligent (χ < χ*), Β also avoids liability despite his lack of care, because although initially he is liable (provided that C is not contributorily negligent), indemnity allows him to shift his liability to A. Given B's behavior, the victim, C, will choose 2* because his maximum expected costs, p(x*,z*)D + C(z*), will then be lower than his minimum expected costs at a lower level of care, p(x*,z0)D + C(z0). A must choose between complying with the due care standard, χ*, or behaving negligently and using x0. If the former, A's expected costs are ax*; if the latter, p(x0,z*)D + ax0. From the definition of due care, ax* < p(x0/z*)D + ax0; hence A will choose x*. Suppose that instead of indemnity a rule of contribution were used to apportion damages between A and Β when A was negligent and C not contributorily negligent. If initially Β spends nothing on care, A will choose between due care and negligence depending on whether ax* < or > p(x0,z*)s"D + ax0, where s" equals A's expected share of damages under a contribution rule. Now it is no longer clear that A will prefer x*. If s" is low relative to the added costs—a(x* - x0)—of complying with the due care standard, A may decide to use less than due care. If so, Β (who in effect is strictly liable because of A's negligence) will select a level of y that minimizes p(x0 + y,x*)sbD + by. If y is positive, this may reduce A's care even further and induce additional changes in the behavior of A and B. Although the equilibrium level of care is uncertain (given A's negligence), the combined care of A and Β will be less than r*, the efficient level. The distribution of care between A and Β will also tend to be inefficient because some care is likely to be undertaken by B, although by assumption it is more efficient for all the care to be taken by A. 10 If we explicitly allow for strategic behavior, even more dramatic misallocations can result. Imagine a case where expected damages of $220 can be averted either by A's spending $10, or B's spending $40, on care; 10. Given A's negligence, the added care taken by Β will not be sufficient to bring the aggregate level of care up to r* ( = x*). Suppose that Β takes care up to the point where p^Xo + y, z*)sbD = -b. Since b > a and D is weighted by s ' ( < 1), marginal returns are lower and costs higher in equilibrium compared with the efficient level, and hence r < r*.

200 · The Economic Structure of Tort Law and instead of a rule of indemnity, which would (eventually) make A fully liable, the damages are apportioned between them—say A will bear 80 percent and Β 20 percent. A's expected gain from taking care is positive ($10 < $176) but so is B's ($40 < $44), and A may figure that if he does nothing Β will still have an incentive to take care and the accident will be averted at zero cost to A. If Β figures likewise, neither party will take care. Alternatively, uncertainty about the other party's action may induce each to take care, which would be wasteful. The suboptimal results described above are avoided by the common law approach in alternative care situations—indemnity. The problem does not arise in joint care situations, where indemnity is denied; there it is efficient for both parties to take care, and the result of doing so is not to avert the accident but to cast the entire liability onto the other party unless he also takes care. The reader may be wondering why Β should be liable at all if A could have avoided the accident at lower cost; if Β were not liable, the problems just discussed would not arise. In other words, by analogy to our discussion of contributory negligence in Chapter 4, why isn't B's optimal care zero, in which event he would not be liable for the injury to C and the possibility of misallocation would vanish? The answer is that it is sometimes efficient to impose back-up liability in order to achieve a second-best solution in circumstances where the best solution is not feasible. Suppose that A is unresponsive to the threat of liability because he is judgment-proof. He therefore has no incentive to take any care. If Β knows this and is in effect strictly liable for A's negligence because he is deemed a joint tortfeasor with A, then Β will select the level of y that minimizes p(y,z*)D + by. Although this will be below the optimal level of r* ( = x*), it is a second-best solution preferable to zero care for both A and Β—the outcome that would result if A were judgment-proof and Β were not jointly liable. Moreover, Β may be able to induce A to take greater care. For example, if Β is A's employer, Β may be able to use wage adjustments, monitoring, or the threat of firing to induce A to take due care. Being made liable as a joint tortfeasor will give Β an incentive to adopt such measures. 11 Indemnity will enable Β ocasionally to shift the full burden of liability to A—the best solution. A problem arises in alternative care cases when the injurer's costs of care are identical (a = b) and there is a fixed cost of care per injurer. Imagine that L = p(r,z)D + ax+by

+ n + C(z)

(7.6)

11. An informational advantage of treating Β as a joint tortfeasor is discussed later.

Joint and Multiple Torts · 201 where η equals the number of injurers taking care. The optimal result would be for A or Β—it doesn't matter which—to be liable and for the other not to be, because then only the potentially liable injurer would take care and L would be minimized. Even if the same level of care were reached when both injurers were taking care (that is, χ = \r* and y = ( l - \ ) r * , where r* satisfies -prD = a = b and 0 < λ < 1), η and hence L would be greater compared with the case of one injurer's taking care. The common law, however, would deem A and Β joint tortfeasors and not allow indemnity. 12 This creates a problem similar to the one just discussed under apportionment in alternative care cases: both A and Β may take care, yielding an above-minimum value of L; or neither A nor Β may take care, relying on the other to do so. It is difficult to imagine what rule within the repertoire of a common law court would produce optimal results in this situation.

A Closer Look at the Common Law Doctrines 1. Contribution versus no contribution. Our discussion of the case where the joint tortfeasors do not expect to split the victim's accident costs evenly implies that any allocation rule (not just no contribution) under which the sum of the tortfeasors' expected shares (s° + sb) in joint care cases sum to one—for example, an equal-sharing rule (each negligent injurer's share is 1 In, where η is the number of such injurers) or a rule that proportions the shares to the relative negligence of the injurers— would provide incentives for efficient accident avoidance. No contribution, however, undoubtedly is the cheapest to administer. A rule that permits contribution in any form entails costly suits13 among injurers attempting to shift part of the victim's loss away from the defendant(s) on whom the victim levied judgment—suits with ex post distributive 12. Union Stock Yards Co. v. Chicago, Β & Q. R. Co., 196 U.S. 217 (1905), appears to have been such a case. The brake on a railroad car was defective, causing an injury. Both the terminal company and the railroad were negligent for having failed to discover the defect and were treated as joint tortfeasors. Indemnity was denied. So far as one can tell, the cost of an inspection that would have revealed the defect would have been the same to both companies. 13. Anyone who thinks that contribution is procedurally straightforward will be disabused of this notion by dipping into the legal literature. See John G. Fleming, "Report to the Joint Committee of the California Legislature on Tort Liability on the Problems Associated with American Motorcyle Association v. Superior Court," 30 Hastings L. ]. 1464, 1485-1514 (1979); Stuart M. Gordon and Diane R. Crowley, "Indemnity Issues in Settlement of Multi-Party Actions in Comparative Negligence Jurisdictions," 48 Ins. Counsel J. 457 (1981). Some of the complications are discussed below.

202 · The Economic Structure of Tort Law consequences but no impact on incentives for efficient accident avoidance. It may be objected that the added costs of a rule of contribution would be small if the rule were mechanical (dividing the amount of the plaintiff's damages by the number of defendants) 14 rather than judgmental (division based on relative fault, as under comparative-negligence and some joint-tortfeasor statutes). 15 But the simplicity of the mechanical rule is deceptive unless contribution is limited to the tortfeasors actually named in the tort victim's complaint. If it is not so limited, the number of joint tortfeasors and hence the amount payable by each to those from whom the plaintiff collected his judgment will not be known until additional lawsuits are completed. And the tortfeasors against whom judgment was levied in those suits might bring suit against still other tortfeasors. Although these complications are avoided by a rule that limits a defendant to seeking contribution from other named defendants, such a rule yields few of the benefits in fairness or equality that proponents of contribution seek. If the plaintiff fails to name all joint tortfeasors in the complaint or judgment, some will escape having to pay any damages— an "unfair" result to anyone who regards the no-contribution rule as itself unfair. The only way to make contribution "fair" is to allow defendants to seek contribution from joint tortfeasors whether or not they are sued, but this is a costly process that a system of law dominated by efficiency concerns would not consider worthwhile. An added complication is introduced when we consider the interplay between contribution and the settlement of litigation. For obvious reasons grounded in the Coase theorem, most lawsuits are settled rather than tried to judgment. Suppose a defendant settles a tort suit for $10,000; the plaintiff later sues and obtains a judgment for $1 million from the other joint tortfeasor (as usual we shall assume there are only two); and in an action for contribution the court would adjudge the two defendants equally at fault. Should the second defendant be able to get contribution from the first of $500,000 (minus the $10,000 that the first paid, which would be deducted from the $1 million because the plaintiff is not entitled to recover more than his full damages)? If so, the first defendant has gotten little from the settlement, and such an outcome will discourage defendants from settling. Yet that is the traditional rule in cases where 14. See Richard A. Epstein, "Plaintiff's Conduct in Products Liability Actions: Comparative Negligence, Automatic Division and Multiple Parties/' 45 /. Air L. & Comm. 87, 112-16 (1979). 15. See Keeton et al., note 2 above, at 338; Comment, note 6 above, at 735.

Joint and Multiple Torts · 203 contribution is allowed. 16 Another rule, the settlement-bar rule, forbids nonsettling defendants to obtain any contribution from settling ones. In its pure form this rule would create strong incentives to settle but could be unfair to nonsettling defendants (the settlers might have settled for tiny sums, leaving the nonsettlers to bear the lion's share of liability), and a court that applies it will therefore hold a fairness hearing before approving the settlement in order to make sure that the settlement does not shift too much of the joint liability to the nonsettlers. But of course the need for such a hearing encumbers the settlement process. An alternative approach that has been gaining ground is a comparative-fault approach which operates as follows. When the plaintiff gets a judgment against nonsettling defendants, the court determines the share of liability of those defendants and that is all the plaintiff gets. In the previous example the plaintiff would be allowed to collect only $500,000 from the nonsettling defendant; his total recovery would be $510,000. This rule is relatively simple and protects the interests of all defendants, but it results in lower recoveries for plaintiffs,17 and it requires a judgment of relative fault. Two possible economic benefits of contribution in joint care cases remain to be considered. The first is information. Under a rule of no contribution, each prospective tortfeasor is uncertain what share of the accident cost he will bear, and if because of this uncertainty the expected shares fail to add up to 1, misallocations can result, as we noted earlier. This uncertainty could be dispelled only by (1) a mechanical contribution rule applicable to (2) all of the joint tortfeasors rather than just those named in the judgment, and such a rule, because of its second characteristic, would be expensive to administer. Moreover, each tortfeasor may be uncertain about the number of other tortfeasors; if so, the number of injurers may be underestimated and there will be the same type of misallocation as when the shares do not sum to one under a no-contribution rule. Second, contribution is a form of liability insurance, because it reduces the variance of the ex post outcomes in joint tortfeasor cases. Although many joint tortfeasors are effectively risk neutral, either because they are diversified or because they carry other—and adequate—liability in16. This and alternative rules are discussed in Donovan v. Robbins, 752 F.2d 1170, 1179-81 (7th Cir. 1985). 17. The incentives to settle under various rules of contribution are discussed in Frank H. Easterbrook, William M. Landes, and Richard A. Posner, "Contribution among Antitrust Defendants: A Legal and Economic Analysis," 23 /. Law & Ε con. 331 (1980).

204 · The Economic Structure of Tort Law surance, some may be risk averse, 18 and they would gain from a rule of contribution. This would not be a gain within the framework of our model, however. 2. Intentional harms. The rule of no contribution was first announced in a case involving an intentional tort. 19 Prosser and Keeton regarded the extension of the rule to accidental harms as having been inadvertent. 20 In fact, the economic analysis of no contribution in the intentional case is symmetrical with that in the unintentional. Let A and Β be the intentional joint tortfeasors and C their victim, and let C's due care level be zero. Recall from Chapter 6 that in the intentional tort setting χ and y—A's and B's inputs, respectively—are not inputs of care that lower the probability of an accident. Instead, the greater χ and y are, the greater is the expected harm to the victim (px and py are both now positive). We can thus write the expected costs of the intentional joint tort as L(x,y) = p(x,y)(D - G) + A(x) + B(y)

(7.7)

where G is the total gain to the joint tortfeasors A and Β from committing the intentional harm. Suppose, as is most likely, that the victim's damages are at least equal to the tortfeasors' total gain. If so, L is minimized when χ and y are zero, and an efficient liability rule will be one that gives both A and Β an incentive not to commit the intentional harm. A rule of no contribution gives both parties this incentive. 21 18. See discussion of maritime collisions below. 19. See Merryweather v. Nixan, 101 Eng. Rep. 1377 (K.B. 1799). The nature of the intentional tort in Merryweather is unclear. Statutes providing for contribution often except intentional torts. See Keeton et al., note 2 above, at 339. 20. See id. at 337. 21. Consider A's choices. If Β did not commit the intentional tort, A would have a positive expected loss if A did, because p(x,0)(D — G) + A(x) > 0. Thus, A would prefer to avoid the intentional tort because then his expected loss would be zero. In contrast, suppose it were in B's interest to commit the tort. Then A's expected loss would be p{x,y)(s[D - s{G) + A(x), where s? and sj are his share of expected damages and expected gain. It might appear to be in A's interest to commit the intentional tort if his share of damages is small relative to his share of the gain (A's expected loss would be negative compared with a zero loss from not committing the harm). But it could never be in A's interest to do so if it were also in B's interest. Stated differently, it could never be in both parties' interest to commit the tort because that would require p(x,y)(s;D - s|G) + A{x) < 0 and p(x,y)(s\D - s2G) + B ( y ) < 0 .

(7.8) (7.9)

These inequalities imply that p(x,y)(D-G) + A(x) + B(y) < 0, which is false because D - G ä O . Since it would not be in A's interest to commit the tort if it were in B's interest to do so, Β would not find it in his interest to commit the tort either. Thus, a rule of no

Joint and Multiple Torts · 205 3. Indemnity; and another look at respondeat superior. The usual view is that the circumstances under which indemnity, the complete shifting of liability from one joint tortfeasor to another, will be allowed cannot be reduced to a simple and definite rule.22 We disagree. We believe that the key to understanding when indemnity will be allowed and when it will be denied is the economic difference between joint care and alternative care situations. The courts use three principal tests in deciding whether to allow the defendant from whom the plaintiff has collected a judgment to obtain indemnity from other joint tortfeasors. 23 The first is the active-passive negligence test: the passively negligent defendant is entitled to indemnification from the actively negligent. The distinction has implicit reference to disparities in the probable cost of accident avoidance. In Muth v. Urricelqui24 the buyers of a house that settled because the ground beneath it had been improperly graded sued the general contractor and recovered damages. The general contractor in turn sought and obtained indemnification from the subcontractor who had done the actual grading. The negligence of the general contractor—his failure to supervise the grading adequately—was held to be passive negligence compared to the active negligence of the subcontractor, who had performed the inadequate grading. At first glance Muth may seem simply to illustrate our earlier hypothetical where the costs of accident avoidance were $79 to A but only $1 to B, with the general contractor corresponding to A and the subcontractor, with his greater knowledge of grading and therefore presumably superior abilility to avoid the accident, to B. The parallel is deceptive. In the earlier case it was assumed that optimal avoidance required safety measures by both A and Β (joint care). But the likelihood of an accident in Muth would have been negligible if either tortfeasor contribution would create an incentive to avoid the intentional harm when D - G a 0. That is the efficient result, just as in the unintentional case. 22. See, for example, Keeton et al., note 2 above, at 344. 23. For good discussions of indemnity, see Merrrill Lynch, Pierce, Fenner & Smith, Inc. v. First Nat'l Bank of Little Rock, 774 F.2d 909, 917-19 (8th Cir. 1985); Comment, note 6 above, at 737-47; Keeton et al., note 2 above, at 341-45; Harold A. Meriam, Jr., and John V. Thornton, "Indemnity between Tort-Feasors: An Evolving Doctrine in the New York Court of Appeals," 25 N.Y.U. L. Rev. 845 (1950); Robert A. Leflar, "Contribution and Indemnity between Tortfeasors," 81 U. Pa. L. Rev. 130 (1932); E. Eugene Davis, "Indemnity between Negligent Tortfeasors: A Proposed Rationale," 37 Iowa L. Rev. 517 (1952). See also American Law Institute, Restatement (Second) of Torts § 886B (1979). 24. 251 C.A.2d 901, 60 Cal. Rptr. 166 (1967). A number of similar cases are discussed in Meriam and Thornton, note 23 above.

206 · The Economic Structure of Tort Law had exercised due care (alternative care). The optimal procedure was not for both to take care but for the subcontractor alone to do so, and indemnification creates the proper incentives for this result. If only the general contractor were liable, he would probably make an agreement with the subcontractor by which the latter would indemnify him for any liability arising from negligence in the subcontractor's grading. Because the subcontractor is the lower-cost avoider of the accident, the Coase theorem implies that liability will be shifted to him by contract regardless of the initial assignment of liability. The indemnity doctrine makes an explicit contractual shifting unnecessary and thus economizes on transaction costs. This analysis raises the question, however, why the general contractor is liable at all. There are two reasons. First, it is cheaper for the general contractor to get reimbursement of the plaintiff's damages by way of indemnity from the subcontractor than for the plaintiff to get damages from the subcontractor directly. For example, the general contractor could more easily arrange for the bonding of the subcontractor or some other guarantee of his financial responsibility than could the plaintiff. Second, the general contractor has better information than the plaintiff regarding the identity of the ultimately responsible party. The plaintiff may not know who did the grading or even that the grader was responsible for the settling of the house (rather than the subcontractor who laid the foundations).25 This type of consideration is particularly important in products liability cases. Our reference to contractual indemnity suggests the similarity between the active-passive test and another test that courts frequently use to decide whether indemnity will lie, the implied-contractual-immunity test. In Weyerhaeuser S.S. Co. v. Nacirema Co.26 a stevedore injured while unloading a ship recovered damages from the shipowner, and the latter sought indemnity from the company that employed the stevedore. The company had failed to remove a dangerous structure that it had erected when the ship was in another port, while the shipowner had been careless in allowing the structure to remain on board during the voyage from that port. Although the contract between the shipowner 25. There is an analogy here to the question whether indirect purchasers have standing to maintain a suit for antitrust damages against their remote seller. We have argued that direct-purchaser suits are more efficient, in part because the direct purchaser has better information than the indirect. See William M. Landes and Richard A. Posner, "Should Indirect Purchasers Have Standing to Sue under the Antitrust Laws? An Economic Analysis of the Rule of Illinois Brick," 46 U. Chi. L. Rev. 602 (1979). 26. 355 U.S. 563 (1958).

Joint and Multiple Torts · 207 and the stevedore company was silent on safety questions, the Supreme Court interpreted it to imply a duty on the part of the stevedore company to indemnify the shipowner. The approach is similar to that of the Muth case, even though the test was worded differently. A third common indemnity test is primary versus secondary liability. It is used, for example, to allow an employer sued under the doctrine of respondeat superior for the tort of an employee to obtain indemnity from the employee, who, as the actual tortfeasor, could have avoided the entire accident and at less cost than the employer—again an alternative care case. The reader may sense a tension between respondeat superior and our analysis of the Muth case. The doctrine of respondeat superior makes an employer liable for the torts of his employees committed in the course of their employment (basically, on the work premises during work hours). But (in apparent contradiction to Muth) the employer is not liable for the torts of independent contractors.27 This distinction makes economic sense once the legal difference between an employee and an independent contractor is grasped. An employee is one who works at the employer's direction in exchange for a wage. Because the employer supervises the details of the employee's work, the employer is in a good position to prevent the employee from committing torts; and respondeat superior gives the employer a strong incentive to do this, as we saw in Chapter δ.28 But the independent contractor by definition does not work under the employer's (more properly, principal's) direction but rather promises a certain output in exchange for the contract price. Because the principal does not supervise the inputs into this contractual performance, he is not in a good position to prevent the independent contractor's torts and therefore is not liable for those torts in the absence of negligence, as in hiring an independent contractor who the principal should have known was careless. 29 27. See, for example, Keeton et al., note 2 above, at 509-16; Clarence Morris, "The Torts of an Independent Contractor," 29 III. L. Rev 339 (1934). In deciding whether an agent is an employee or an independent contractor for purposes of deciding whether his principal is liable for his torts, courts use a test of "control," which has been examined from an economic standpoint in a recent article. See Alan O. Sykes, "The Economics of Vicarious Liability," 93 Yale L. j. 1231, 1261-80 (1984). The author's conclusion is that the control test is efficient in many respects but could be made more efficient. We consider his discussion to provide further evidence of the general, although imperfect, conformity of tort law to the concept of economic efficiency explored in this book. See also id. at 1232 n. 5. 28. Clarence Morris made this point more than a half century ago. See Morris, note 27 above, at 341. For an even earlier statement (but not the earliest) see Harold J. Laski, "The Basis of Vicarious Liability," 26 Yale L.J. 105, 114 (1916). 29. See Morris, note 27 above, at 3 4 1 - 4 2 .

208 · The Economic Structure of Tort Law The subcontractor in the Muth case was an independent contractor, not an employee of the general contractor. Why therefore should there be the kind of back-up liability for the subcontractor's torts that the law imposes on the employer for his employees' torts? The answer may be that in Muth the general contractor was contractually liable for the tort; and as a matter of contract between him and his customer, it may well have been optimal to make him bear initial liability, because he presumably had better information than his customers and could sort out the ultimate liability with his subcontractors. That feature is missing where, for example, a passerby is hurt by the negligence of a contractor who is doing work on a home, having been hired to do it by the homeowner. In an important class of cases, however, principals are liable to third parties with whom they have no contractual relationship for the torts of independent contractors. If the work of the independent contractor has a high degree of dangerousness, the principal will be deemed to have a "nondelegable duty of care," meaning that he will be liable, whether or not himself negligent, for the contractor's torts. 30 The explanation in economic terms is twofold. The more dangerous an activity is, the higher the cost-justified level of care. What would be duplicative and inefficient care by a principal and independent contractor in a case of average danger may become cost justified when the danger is so great that backup precautions are much more valuable. And the greater the expected accident cost, the less likely is the independent contractor to be able to pay a legal judgment. While on the subject of respondeat superior, it may be well to note the special situation regarding an employee's intentional torts. The same limitation of respondent superior to torts committed in the course of employment (a limitation plainly related to the employer's cost of monitoring the employee's conduct) is imposed. But in the case of an intentional tort the court is likely to insist in addition that the employee in committing the tort have been trying, however misguidedly, to advance the employer's goals, as where the employee assaults a debtor of the employer in an effort to collect the debt. If the employee is actuated by purely personal motives, the employer's practical ability to prevent the tort will be slight. The employer should be able to ensure that the em30. See Atlantic Coast Development v. Napoleon Steel Contractors, 385 So. 2d 676 (Fla. 1980) (hoisting concrete blocks); Krause v. Alamo Nat'l Bank of San Antonio, 586 S.W.2d 202 (Tex. Civ. App. 1979), aff'd, 616 S.W.2d 908 (Tex. 1981) (building demolition); A. R. Boroughs v. Joiner, 337 So. 2d 340 (Ala. 1976) (crop dusting); Note, "Torts—Nondelegable Duty—Direct and Vicarious Liability for Negligence," 44 N.C. L. Rev. 242 (1965).

Joint and Multiple Torts · 209 ployee not only use due care but also avoid overzealous pursuit of the employer's goals, but it is much harder for the employer to screen and monitor employees for purely personal attitudes. We come back to indemnity, which in products liability cases is allowed under the primary-secondary test. Suppose that the manufacturer of some product is strictly liable for a defect in the product, even if the defect is in a component and it would be prohibitively costly for the manufacturer of the final product to detect it. Without indemnity the manufacturer of the defective component would not have sufficient incentive to take due care because the manufacturer of the final product would bear all or some of the accident costs.31 The existence of three overlapping tests for indemnity with uninformative names—plus other tests that we have not discussed32—is regrettable; but each test as applied seems to make good economic sense, and a survey of indemnity cases (admittedly somewhat out of date) indicates that they conform to the economic interpretation of the rules presented above. 33 However, an important qualification should be noted: in states that have repudiated the common law rule of no contribution, the tendency has been to replace indemnity with contribution, yielding (as the economic analysis implies) suboptimal solutions. For example, in Dole v. Dow Chemical Co. 34 a worker was poisoned by inhaling a chemical, and his next of kin brought a negligence suit against the manufacturer, Dow, claiming that Dow itself should have warned the worker about the dangers of the chemical. Dow sought indemnity from the worker's employer, who had been negligent in failing to comply with Dow's instructions regarding the chemical. It was a clear case for indemnity. The accident would have been avoided if the employer had been careful, and it would surely have cost more for Dow to warn the 31. For a good discussion of indemnity in products cases see Note, "Comparative Causation, Indemnity, and the Allocation of Losses between Joint Tortfeasors in Products Liability C a s e s , " 10 St. Mary's L.J. 587, 5 9 8 - 6 1 0 (1979). 32. See Comment, note 6 above, at 739 ("duty owing between tortfeasors"), 741 ("equitable considerations"). For still other formulations see Note, "Contribution and Indemnity—An Examination of the Upheaval in Minnesota Tort Loss Allocation Concepts," 5 Wm. Mitchell L. Rev. 109, 134 (1979); and the Merrill Lynch case cited in note 23 above. 33. See Comment, note 6 above, at 737-47. The writer of the comment describes the indemnity cases as explicable on a deterrence rationale. As he uses the term, however, "deterrence" means seeking to place liability on the tortfeasor who can avoid the accident at lowest cost. The writer does not distinguish joint care from alternative care cases; but he gives no examples of joint care cases in which indemnity was allowed. 34. 30 N.Y.2d 143, 282 N.E.2d 288 (1972).

210 · The Economic Structure of Tort Law worker than for the employer to do so. Yet the court, relying on a statute allowing contribution among joint tortfeasors, held that Dow was entitled only to contribution and not to indemnity. The distinction between joint and alternative care helps to explain another aspect of indemnity—that a difference in the degree of a joint tortfeasor's negligence is not a ground for indemnity. 35 Restated in economic terms—that a difference in the costs of accident avoidance among joint tortfeasors is not a basis for shifting liability to the lowest-cost avoider—this principle may seem patently inconsistent with an economic theory of indeminty. However, as applied, primarily in collision cases, the principle seems merely to delimit joint care situations, where indemnity would be inefficient. For example, in Warner v. Capital Transit Co. 36 a bus stopped suddenly (injuring the plaintiff) when a taxi swerved in front of it. Both the bus and taxi drivers were negligent, but the former less so; nonetheless the bus company was denied indemnity from the taxi company. This was a joint care case; optimal accident avoidance required that both the bus and the taxi driver take care. If indemnity were allowed in such a case, bus drivers would lack adequate incentives to keep their distance and taxi drivers might take excessive care to avoid getting too close in front of buses. 37 It is as if, in the joint care example we gave earlier where the cost of taking care was $79 to A and $1 to B, the law were to allow Β to obtain indemnity from A in cases where neither took care and an accident occurred. Then Β would have no incentive to take care, and A, to avoid the accident, would have to spend more than $80. In one well-known collision case involving two airplanes, indemnity was allowed, 38 and Prosser and Keeton cannot understand the difference in treatment between this case and Warner.39 But in the former the court had found that one of the planes had the last clear chance to avoid the other; that is, just before the accident occurred, only one of the planes could avoid the other. It was thus an alternative care situation.40 4. The special case of maritime collisions. The earliest examples in English 35. See cases collected in A n n o t . , 8 8 A . L . R . 2 d 1355 (1963). 36. 162 F. Supp. 2 5 3 ( D . D . C . 1958). 37. A fortiori,

indemnity is denied in collision cases w h e r e the parties are equally at

fault. See G e o r g e W . Gale L u m b e r Co. v. Bush, 2 2 7 M a s s . 203, 116 N . E . 4 8 0 (1917); Gobble v. Bradford, 2 2 6 Ala. 517, 147 So. 6 1 9 (1933); P a n u s u k v. Seaton, 2 7 7 F. S u p p . 9 7 9 (D. M o n t . 1968). 38. See United Airlines, Inc. v. Wiener, 3 3 5 F . 2 d 379, 4 0 2 (9th Cir. 1964). 39. See K e e t o n et al., note 2 above, at 343. 40. See our discussion of last clear c h a n c e in C h a p t e r 4. F o r other last clear c h a n c e c a s e s in w h i c h indemnity w a s allowed see Leflar, n o t e 2 3 above, at 1 5 1 - 5 4 .

Joint and Multiple Torts · 211 law both of contribution and of comparative negligence are in the maritime field. Maritime collisions have long been governed by a rule of contribution, 41 and in the more common case where only two vessels contribute to the accident, by a rule of comparative rather than contributory negligence. The traditional rule was one of fixed rather than variable apportionment—each party bore one-half his own damages and paid one-half of the damages of the other party—and was criticized as less fair than apportionment on the basis of relative negligence. 42 But the traditional rule is both less costly to administer and, in joint care cases, allocatively efficient. And those are the relevant cases: the optimal prevention of ship collisions presumably requires both ships to exercise some care rather than (as with the torts of an employee subcontractor, component manufacturer, or stevedore company) one to exerise a great deal of care and the other no care. (In alternative care cases the preferred rule would be indemnity.) We find then in the maritime area a long-standing rejection of the allor-nothing approach to a characteristic of the common law until quite recently. 43 How is this to be explained? Here we must step outside our model and notice that risk aversion is more likely to play a major role in maritime tort law than in terrestrial tort law. Until modern times maritime undertakings were exceptionally risky in terms of both the probability of loss and the magnitude of the loss if it occurred. The demand for insurance must have been great, but the provision of market insurance was undeveloped. 44 In these circumstances we would expect the emergence of legal doctrines that provided insurance as well as deterrence. 45 Because comparative negligence and contribution both pre41. See, for e x a m p l e , T h e N o r t h Star, 106 U . S . 17 (1882). 42. See Grant Gilmore and Charles L. Black, Jr., The law of Admiralty 5 2 8 (2d ed. 1975). The rule persisted in A m e r i c a long after it w a s a b a n d o n e d by other nations, but it w a s eventually o v e r t u r n e d in favor of relative negligence. See United States v. Reliable Transfer C o . , 421 U . S . 3 9 7 (1975). The traditional rule is d e f e n d e d on administrative-cost g r o u n d s in Epstein, note 14 above, at 1 1 0 - 1 1 . 43. See Hillier v. S o u t h e r n T o w i n g C o . , 714 F . 2 d 714, 719 (7th Cir. 1983). 44. The earliest forms of market insurance w e r e in fact maritime. See William J. Winter, A Short Sketch of the History and Principles of Marine Insurance (Ins. Society of N . Y . , 2 d ed. 1935); C. F. Trenerry, The Origin and Early History of Insurance,

Including

the Contract of

Bottomry (1926). 45. Notice, h o w e v e r , that the traditional a p p o r t i o n m e n t rule provides only a m o d e s t a m o u n t of insurance. It c o m e s into play w h e n both parties are negligent, shifting losses from the ship with a larger loss to the ship with a smaller loss until both parties' losses are equal. A n o t h e r maritime doctrine that has a n insurance function, the law of general a v e r a g e , is e x a m i n e d in William M. L a n d e s a n d Richard A . Posner, "Salvors, Finders, G o o d Sa-

212 • The Economic Structure of Tort Law serve allocative efficiency, they are efficient methods of insurance, although like all insurance they are not costless, and evidence presented in Chapter 1 suggests that insurance through the tort system is normally more costly than market insurance. 5. The case of uncertainty. Certain torts that are not joint under conditions of full information are treated as joint by the courts. In Summers v. Tice, 46 for instance, two careless hunters fired in the direction of the plaintiff. He was wounded by a single bullet, but it was impossible to determine which of the two hunters had fired that bullet. Both were held liable as joint tortfeasors. As we shall see in the next chapter, this result would be economically incorrect if it could be determined which hunter had fired the bullet. But in the absence of such knowledge, a rule of joint liability is correct. If neither party will know whether his bullet has struck the victim, we want each party to avoid shooting carelessly (to take due care). The case is thus one of joint care, and a rule of joint liability with no contribution is the correct rule in such a case. A further advantage of the approach in Summers is that it places the burden of obtaining information on the parties that are in the best position to carry the burden. In this respect it resembles the joint liability of product and component manufacturers. The continuing litigation over the tort liability of manufacturers of DES47 presents interesting questions regarding the economically appropriate outer bounds of the Summers v. Tice principle. DES is a drug that was administered to pregnant women (to prevent miscarriage) from the late 1940s to 1971, when it was discovered to cause adenosis and sometimes cervical cancer in the grown daughters of women who had taken it. The daughters are unable to identify the particular manufacturer of the drug the mothers took because the relevant sales took place twenty years or more before suit was brought and neither drug companies nor pharmacies keep records for that length of time. Even the relative sales shares of the manufacturers twenty years ago are unknown. Moreover, it is estimated that between ninety-four and three hundred companies manufactured the drug.48 maritans, and Other Rescuers: An Economic Study of Law and Altruism," 7 J. Legal Stud. 83, 1 0 6 - 8 (1978). 46. 33 Cal. 2d 80, 199 P.2d 1 (1948). 47. See Note, "DES and a Proposed Theory of Enterprise Liability," 46 Fordham L. Rev. 963, 9 6 7 - 7 3 (1978). The landmark decision allowing DES plaintiffs to hold manufacturers liable although they are unable to identify which manufacturer supplied the drug that injured them is Sindell v. Abbott Laboratories, 26 Cal. 3d 588, 607 P.2d 924 (1980). 48. See Note, note 47 above, at 964 n. 3.

Joint and Multiple Torts · 213 Logically, this situation is the same as that of Summers v. Tice if one is confident that each of the manufacturers of DES was negligent in failing to test for second-generation effects at the time they sold the drug for eventual resale to the mothers of the plaintiffs. But because of the passage of time since the alleged negligence, the use by the tort system of average rather than individual care as the basis for determining negligence, and the very large number of joint tortfeasors, a more than usual danger exists that joint liability will have misallocative effects. To illustrate, suppose that there are ten manufacturers of DES, each producing one-tenth of the drug's total output; if the drug is defective, all will be held liable for negligence; total expected accident costs are $1 million ($100,000 per manufacturer); and the costs of avoiding the accident are $80,000 each to five of the manufacturers and $110,000 each to the other five. Each member of the first group knows that if he spends $80,000 on avoidance, the benefit in reducing his expected liability will be only $50,000 (because he will remain liable for the $50,000 in expected accident costs that the other five manufacturers have no incentive to avoid). 49 Therefore the precautions will not be taken, with the result that in this case joint liability leads to the same result as no liability! Whether this danger is greater than the danger that, if the approach of Summers v. Tice is not used in DES cases, none of the manufacturers of DES, however negligent they may have been, will bear any tort liability is clearly a very difficult question, which we shall not attempt to answer. 6. What is a joint tortfeasor? In the DES cases the issue from an economic standpoint is not contribution versus no contribution but individual versus joint liability, because we know from our earlier discussion that in such cases neither approach will produce efficient results. This raises the question of when two (or more) injurers shall be deemed joint so as to trigger the rules applicable to joint tortfeasors. From an economic standpoint, treatment of multiple tortfeasors as joint is appropriate in every inseparable (single) injury case (including the case where the injury would be inseparable under conditions of full information, as in Summers v. Tice). 50 This seems the legal position as well, judging from the following effort to define a joint tort: 49. The $50,000 figures is obtained by assuming that, ex ante, each manufacturer bears one-tenth of the total liability of the group. The total liability is $500,000 if all five manufacturers who can avoid the accident do so. 50. But imagine that in Summers v. Tice each defendant had shot the plaintiff in a different part of the body and neither wound was more serious by virtue of the other. It is clear, as a matter of both economics and law, that this case should be analyzed as two individual injuries.

214 · The Economic Structure of Tort Law Generally, a joint tort has been found by courts in four circumstances: (1) where there has been a concert of action between two or more actors resulting in injury to the plaintiff; (2) where two or more parties have failed to perform a common duty owed to the plaintiff; (3) where there has been vicarious liability imposed upon a master for the acts of his agent which have caused an injury to the plaintiff; and (4) where the separate negligent acts of independent actors have concurred to produce a single, indivisible injury to the plaintiff.51

Points (1) and (2) appear to refer to the joint care cases; (2) may also refer to Summers v. Tice. The third (respondeat superior) is a case of alternative care where, however, we suggested that joint liability with indemnity was probably more efficient than placing liability on the lower-cost avoider alone. The fourth situation could include cases of both joint and alternative care, but we are aware of no cases where joint liability has been imposed in an alternative care setting when it would have been more efficient to impose liability on the lower-cost accident avoider alone. An important and perhaps efficient limitation of the joint tortfeasor concept should be noted: mere knowledge does not convert a bystander into a joint tortfeasor. This result is necessary to preserve the no-duty-to-rescue rule of the common law—a rule that may itself have an economic basis, as we saw in Chapter 5. Treating two or more injurers (or possible injurers) as joint tortfeasors will produce inefficient results in certain cases. The general case is where one of the injurers will not take care to avoid the accident, with the result that the others may not either. This problem is to some extent the inevitable result of the stochastic nature of care, the difficulty of evaluating due care on an individual as distinct from average or group basis, other pockets of strict liability in negligence law, and the possibility of errors in applying the negligence standard in particular cases. An avoidable source of the problem, however, derives from the imposition of strict liability in joint tortfeasor situations. Suppose that we modify our original loss function (Eq. 7.1) and assume that ζ is fixed at z. Optimal care still requires that A and Β each spend on care up to the point where marginal cost equals the marginal reduction in expected damages (pxD and pyD for A and B, respectively). This in turn requires that the reduction in the probability of an accident resulting from each party's care be weighted by the victim's full damages. We showed earlier that a negligence standard (with or without contribution) would do this. Now suppose that both A and Β are strictly liable 51. Note, note 34 above, at 1319.

Joint and Multiple Torts · 215 and each expects to pay half the victim's damages. If so, each will face a potential liability of .5pD and each will take less than optimal care. To put this differently, because neither party can avoid liability by behaving nonnegligently, each party's care will confer benefits on the other by reducing the other's expected liability. In the absence of cooperation neither will have an incentive to take account of these external benefits, and as a result care will be suboptimal. Suppose that ten polluters of a stream impose pollution damages equal to $10 million in the aggregate. Five polluters could reduce damages by $5 million ($1 million each) by undertaking pollution control measures at a loss of $600,000 each. The other five could also reduce damages by $1 million each but at a cost of $1.4 million each. Under a negligence standard the first five will have incentives to avoid polluting and society will be wealthier by $2 million. But under strict liability, with each polluter jointly liable for the total of $10 million in pollution costs, the five polluters who can avoid pollution at a cost of $600,000 each will not do so, because each will still have an expected liability of $500,000 (assuming either a no-contribution rule with each polluter equally likely to be sued for the remaining damage or a contribution rule in which damages are divided by the number of defendants). This inefficient result is avoided either by not treating the polluters as joint tortfeasors or by making the rule of liability negligence rather than strict liability. Here, then, is a subtle cost of strict liability compared with negligence.

Successive Joint Torts The successive joint tort is a combination of an individual and a joint tort. Let p"(x,z) equal the probability of an accident between A and C, and D" the resulting damage to C. This is the individual or separable component because it involves a single victim and single injurer. Suppose, however, that there is also a positive probability of C's being involved in a subsequent accident with Β that is conditional on the initial accident between A and C taking place. By definition the second accident can occur only if the first occurs. This is the joint component because expenditures on care by both A and Β (and C) will reduce the expected losses from the second accident. Let pb{y) equal the conditional probability of the second accident, which to simplify we assume depends only on B's care, and let Db equal the resulting incremental damages to C. 52 52. Suppose that ρ" equals the unconditional probability of the accident between Β and

216 · The Economic Structure of Tort Law For example, suppose that A runs down C and fractures his skull, and B, whose headlights are broken, fails to see C lying helpless in the street and hits him, breaking his leg. Or let Β fail to maintain a guard rail alongside the road and C, after having his skull fractured by A, falls through the defective railing and breaks his leg. In both instances B's lack of care increases the probability of additional harm to C. The expected loss of a successive joint tort is given by L{x,y,z) = p°(x,z){D" + V\y)D»\ + A(x) + B(y) + C(z).

(7.10)

As before, χ*, y*, and z* are the levels of care that minimize Eq. (7.10). 53 Observe in Eq. (7.10) that B's expenditure on care is undertaken prior to the accident between A and C, and that the efficient care levels for A and C depend on both the initial amount of damage, D", and the expected additional damages to C from the second injury. 54 C. Then p" = p"pb + p"°pb„„ where p" and p™ are the respective accident and no-accident probabilities between A and C, and pb and pb„, are the respective conditional probabilities of the second accident. To focus on the conditional-injury aspect, we assume that pj„ = 0 so that p" = p"pj. To simplify the notation we write pb instead of pj in the text. There are two other possibilities: (1) The probability of the accident between Β and C is independent of whether or not the initial accident took place; in this case p" = pb = pba, and we no longer have a second injury conditional on the first. (2) pb > pj„ but pb„ > 0; that is, the likelihood of the second injury is greater if the first occurs, but there is a positive probability that the second will occur even if the first does not. Alternatives (1) and (2) have different implications for the efficient levels of care and appropriate liability rules compared with the assumption that ρ J, = 0. 53. The first-order conditions are p;(D" + pbD") + A, = 0, p"pbD" + By = 0, and p"z(D' + p'D') + C2 = 0, which are satisfied at x*, y*, and z*. Because the second partial derivatives of the accident probabilities are positive and the marginal costs of avoidance are either constant or increasing, the second-order conditions for a minimum are also satisfied. 54. One could specify Eq. (7.10) differently by assuming that B's expenditures on care were not preparatory but were conditional on the accident between A and Β occurring. Then p"B(y) would substitute for B(y) and this in turn would alter slightly the efficient care levels—it would increase the expected gain from A's and C's care because the likelihood of B's undertaking expenditures on care is dependent on the accident between A and C taking place. The examples in the text imply preparatory expenditures by Β (for example, maintenance of safety equipment) but one can easily give examples of conditional expenditures by B, such as on medical treatment that would reduce the probability of additional injury to C. For example, suppose that improper medical treatment would leave C paralyzed for life. Although the choice between making B's expenditure conditional or unconditional has only a marginal effect on the model, it has an important implication for the appropriate legal rule of damages. If B's expenditure is conditional, then an efficient liability rule would hold A and C liable for B's expenditure on care (depending on whether A or C were negligent). The typical case of conditional expenditure is where Β attempts to rescue C

Joint and Multiple Torts · 217 If A is negligent and C is not contributorily negligent, the common law rule of liability holds A liable for the full amount of C's damages— both initial damages of D" and incremental damages of Db. If Β were also negligent, he would be liable up to the amount of Db. If both A and Β were negligent and C not contributorily negligent, C would have the option of suing A for D" + Db, or both A and Β for D" + Db, with each injurer's maximum liability defined as above.55 The application of the no-contribution rule to a successive joint tort means that A and Β have no rights to recover from each other for damages that one has paid but the other was jointly liable for (if A pays Db, A has no right to recover part of Dh from B). If we assume either that all three parties are negligent or that Β and C are negligent while A is not, C will not be able to recover D", because, by assumption, C is contributorily negligent in the accident with A. But his negligence will not bar him from recovering Db from Β (if Β is negligent but A is not) or from either A or Β (if both are negligent).56 Table 7.2 summarizes each party's expected accident loss and cost of avoidance for different levels of care, under the rules set forth above. The proof that the liability rules summarized in Table 7.2 create incentives for injurers and victims to undertake efficient expenditures on accident avoidance is complex,57 but the intuitive explanation is simple. The second injurer cannot avert the first injury, so no allocative purpose would be served by making him liable for that injury. But the first injurer could have avoided the incremental as well as initial damages—the damages inflicted by the second tortfeasor as well as those inflicted by himself—by avoiding the accident. The avoidance of the second injury is thus a benefit of the first tortfeasor's taking care. We want him to

(for example, Β is a doctor) and high transaction costs prevent Β from negotiating with either A or C. If Β is not a professional rescuer, Β will recover his costs only if he is injured in the rescue attempt. If Β is a professional rescuer (a doctor, for example), he is entitled to compensation for his costs. A possible economic justification for the difference in treatment is that the costs incurred by the nonprofessional are usually trivial (the costs of a phone call) and therefore altruism will be a sufficient inducement to rescue. See Landes and Posner, note 45 above, at 110. 55. For example, C might recover D" + KDk from A and (1 - k)Db from Β where 0 < λ < 1. 56. Can C recover Db from A where A and C are negligent but Β is not? We have not discovered any cases on this point, and we proceed on the assumption that C cannot recover Db from A. The analysis, however, would not be affected by modifying this assumption and allowing C to recover from A. 57. And therefore relegated to the appendix to this chapter.

ο Ν Ν

ο Ν υ

υ

+

+ +

U *

ο

Ν

υ

Q

u

*

+ +

υ

δ *

Q

υ

*

+

b

3 *

Ν Ν Q Q 1χ a. U U * ο ο + + Ν Ν * " Ν Q Q ο * ν

la. Q + Q

Η,

Η,

ο a. a. ο Ν Ν

*

a.

μ

>?

*, *** &

a.

κ

;εγ 0)

χ,

Μ) C

ε3

κ

·

3 «)

Q Vi

•b Q

δ " CQ

+ Q

*

CQ

*

»

*

Φ Ο. Χ

>>

«

ν

Ix * Ν * 1

a.

u > •J3 υ ν Ο-

a.

υ

Λ c

C υ •υ Ό

κ

υ

α. χ 0)

ο *

χ < ·< ρ-< Λ

χ

+

tu u

>

ζ

χ 5

Joint and Multiple Torts · 219 weigh this benefit in deciding how much care to take, and he can be induced to do so by being made liable for the misconduct of the second tortfeasor. The common law allocates liability in accordance with the above economic principles. As Prosser and Keeton explain: If an automobile negligently driven by defendant A strikes the plaintiff, causing a skull facture, and leaves the plaintiff helpless on the highway, where shortly afterward a second automobile, negligently driven by defendant B, runs over the plaintiff and breaks a leg, A will be liable for both injuries, for when the plaintiff was left in the highway, it was reasonably to be anticipated that a second car would run the plaintiff down. But defendant Β should be liable only for the broken leg, since Β had no part in causing the fractured skull, and could not foresee or avoid it. On the same basis, an original wrongdoer may be liable for the additional harm inflicted by the negligent treatment of the victim by a physician, but the physician will not be liable for the original injury. 58

The M o v e m e n t to Contribution The common law rule of no contribution among joint tortfeasors has been abrogated by statute in a number of states and by judicial decision in a few. Table 7.3 lists the position of each state and the District of Columbia, indicating when the state abandoned the common law rule, whether it did so by statute or judicial decision, and whether the rule of contribution that replaced the common law rule was broad (defined as allowing contribution from all joint tortfeasors whether or not named in the complaint or judgment) or narrow (defined as allowing contribution only from joint tortfeasors named in the judgment). Because contribution seems to be a less efficient rule than no contribution, the question arises why so many states have abandoned the common law approach. We do not have a complete answer to that question, as we have no idea what politically effective interest group is benefited by contribution among joint tortfeasors. However, whatever the benefits may be, presumably they are weighted against the efficiency losses, which are costs to some group, in the political decision-making process. Therefore, assuming the benefits are distributed randomly across states, we would expect contribution to be found more often in states that do not weight efficiency heavily (as judged by their public policies) 58. Keeton et al., note 2 above, at 352 (footnotes omitted). On contributory negligence in successive torts cf. id. at 4 5 7 - 5 9 .

220 · The Economic Structure of Tort Law Table 7.3.

The movement to contribution (year of adoption) Statute

State

Decision

Broad

Narrow

Alabama3 Alaska Arizona Arkansas California Colorado Connecticut' Delaware District of Columbia Florida Georgia Hawaii Idaho Illinois Indiana" Iowa Kansas' Kentucky Louisiana Maine Maryland Massachusets Michigan Minnesota Mississippi Missouri Montana Nebraska* Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon

1970 1984 1941 1958 1977 1953 1949 br 1975 1966 1941 1971 1979 1956 br 1955 1933 na 1918 na 1941 1963 1957 1926 br 1952 1939 1981 1973 1971 1952 1947 1974 1968 1957

1962

1976 1978 1971

Joint and Multiple Torts · 221 1951 1940

Pennsylvania Rhode Island South Carolina

1960

South Dakota Tennessee

1950 br

Texas Utah

1944 br

1925 1973 1970

Vermont Virginia Washington West Virginia Wisconsin Wyoming

1968

1950 1982 1958 1981 br 1976

a. Has retained no-contribution rule.

than in states that do weight it heavily. Because, in general (and with the exceptions again presumably distributed randomly across states), efficiency is fostered by reliance on markets to allocate resources and depressed by reliance on governmental allocation, the size of the public sector in the state, which can in turn by proxied by the ratio of per capita state and local expenditures or taxes to per capita income, should be positively related to the probability that the state has replaced the common law rule of no contribution with a rule of contribution. Of course the use of these proxies implies that the optimal size of the public sector is proportionately equal in every state, an artificial assumption, and one that may deprive our results of much significance. Table 7.4 presents the results of our attempts to test this hypothesis. Linear probability equations were estimated across states for two dependent variables: the presence or absence of a rule permitting contribution (CONT) and the presence or absence of a broad contribution rule (BROAD). The independent variables include per capita state and local expenditures (EXP), per capita state and local taxes (TAX), the progressivity of the state income tax (PROG) as measured by the difference between the maximum and minimum state personal income tax rate, and per capita income (Y). Table 7.4 indicates some, although only limited, support for our hypothesis. There is a positive and significant relationship between the government-expenditures variable and the probability that a state allows contribution (equation 1) and has a broad contribution rule (equation 2). The state tax variable, which is a less comprehensive measure of the

222 · The Economic Structure of Tort Law Table 7.4. Regression analysis of contribution statutes (f-statistics in parentheses, 51 observations)' Equation no.

Dependent variable

1

CONT* BROADb

2

CONTb

3

BROADb

4

Constant 0.26 (0.08) -2.70 (0.56) 2.00 (0.55) -0.84 (0.16)

EXPC

TAX'

0.49 (1.58) 0.77 (1.74)

PROG 0.01

(0.60) -0.01

(0.71) 0.43 (1.49) 0.51 (1.21)

-0.01

(0.77) -0.01

(0.42)

Yc

-0.34 (0.74) -0.27 (0.40) -0.47 0.87 -0.23 (0.29)

R2

0.02 0.02 0.02 0.02

Source: Independent variables are taken from U.S. Bureau of the Census, Statistical Abstract of the United States: 1985 (105th ed.), Washington, D . C . , 1984, Tables N o . 449, 457, a n d 753. a. Alaska is excluded from analysis because its per capita taxes and expenditures are roughly four times the nationwide average. b. CONT = 1 if contribution allowed, 0 otherwise. BROAD = 1 if broad contribution rule, 0 otherwise. (See Table 7.3.) c. EXP, TAX, a n d Y are in logs. EXP a n d TAX are for 1982, and Y is for 1981.

relative size of a state's public sector, as expected has a positive but less significant effect on both dependent variables (equations 3 and 4). The remaining two variables, the level of state per capita income and state tax progressivity, were statistically insignificant in all equations. We estimated similar equations using the presence or absence of comparative negligence as the dependent variable. (For data on comparative negligence, see Table 3.8.) We found the same relationship—a positive although only marginally significant effect of government expenditures on the probability of having a comparative negligence rule.

Inducing Breach of Contract The analysis of joint tortfeasors developed in this chapter is applicable to any case in which two actors create a single injury, even if the actors are not technically joint tortfeasors. One such case is that of tortiously inducing breach of contract. 59 If A induces Β to break B's contract with 59. See Keeton et al., note 2 above, at 9 7 8 - 1 0 0 5 ; 1 Harper and James, note 2 above, at 489-510.

Joint and Multiple Torts · 223 C, C incurs a single injury (the loss of the expected profits of the contract) as a result of the conduct of both A and B. A and Β are not joint tortfeasors because breach of contract is not a tort and therefore Β is not a tortfeasor. Nevertheless, our analysis of joint tortfeasors is applicable and can explain why A is treated as a tortfeasor. But we confront at the outset a puzzle, or series of related puzzles. If the contract is worth more to A than to C, why isn't it efficient for Β to break the contract and recontract with A?60 Stated otherwise, if C is fully compensated for the costs of the breach to him in his breach of contract action against B, why should he also have a cause of action against A? And why should punitive damages be available in C's tort suit against A, when, in a suit against B, C would be entitled only to contract damages, which are compensatory and never punitive? Won't punitive damages discourage efficient breaches? To answer these questions we begin by noting that the inducement case is one of alternative care. The injury to C will be avoided if A does not induce the breach or if Β does not commit it. In any alternative care case it is possible either to place the full liability on the lower-cost harm avoider at the outset, and not impose joint liability at all, or to impose joint liability together with some device (indemnity, in explicit joint tort cases) for shifting the full liability ultimately to the cheaper harm avoider. Why might joint liability (C can sue either Β for breach of contract or A for the tort of inducing the breach) be the law's solution in the breachof-contract case? First, one of the parties might be insolvent. For example, suppose Β in the above example is C's best salesman, and A lures Β away to work for him, A, instead. C has a cause of action for breach of contract against B, assuming, of course, that Β broke his contract in leaving C's employment; but B, an individual, may be judgmentproof, so A is made liable as well. There is an obvious analogy to the joint liability of an employer and his employee for the latter's torts committed within the scope of employment (respondeat superior). A second possible reason for joint liability is that although Β clearly has broken his contract and is therefore liable to C (and suppose Β is fully solvent), A is the lower-cost avoider of the harm so that we want the full liability to come to rest ultimately on him. Suppose, for example, that Β has guaranteed C the delivery of some good, but A, a labor union, wrongfully organizes a boycott against Β that prevents Β from honoring his delivery obligation to C. Although Β is in breach of contract, A, who 60. It is only a partial answer that, as we shall see, the inducer of the breach does not always want to make a contract with the breaching party.

224 · The Economic Structure of Tort Law deliberately and wrongfully interfered with B's honoring the contract, should bear the ultimate liability. A could have prevented the injury to C by not acting wrongfully, but there was little or nothing Β could do to honor his contract given A's wrongful acts. For generally liability for breach of contract is strict; that is, the victim of the breach is not required to show that the other party acted negligently or willfully in breaking the contract.61 In a joint tort case where the liability of one of the tortfeasors is based on strict liability and that of the other on negligence or willfulness, the case for indemnity, as we have seen, is a strong one, provided that the case is indeed one of alternative care. Hence, if A and Β in the boycott case were treated as joint tortfeasors, Β would have a right of indemnity against A. But A and Β are not joint tortfeasors, so Β does not have a right of indemnity against A. However, the same result is achieved indirectly: the fact that punitive damages are available in tort actions for inducing breach of contract but not in contract damages will induce C to sue A rather than Β because he can obtain a larger judgment. To see to what extent the joint tort analogy actually explains the pattern of cases alleging wrongful inducing of a breach of contract, we collected, from a variety of secondary sources, a representative sample of such cases.62 As shown in Table 7.5, more than 85 percent of the cases in our sample are classified in categories where either there is a substantial danger that the breaker of the contract will be judgment-proof (or the sum involved too small to warrant suit) or the third party who induced the breach could probably have avoided the harm at lower cost than the contract breaker. In the former group of cases we find, for example, lawyers (for individuals) whose clients have been lured away by other lawyers (and the same for real estate and other brokers) plus the salesman case discussed earlier. In the latter group of cases we find the union case already discussed and others closely related to it. The largest group of cases where the inducer is the lower-cost avoider of the harm involve slander and other forms of misrepresentation. For example, in one case in our sample, American Surety Co. v. Schottenbauer,63 the plaintiff, an employee, was discharged by his employer because the defendant, the employer's workmen's compensation insurer, falsely advised the employer that the employee had a diseased finger which was creating a serious risk of accident. This advice was a tortious inducement to the 61. See Richard A. Posner, Economic Analysis of Law §§ 4.8, 6.5 (3d ed. 1986); and discussion of the Moch case in Chapter 5 of this book. 62. A list of the cases is available from the authors upon request. 63. 257 F.2d 6 (8th Cir. 1958).

Joint and Multiple Torts · 225 Table 7.5.

Cases involving tort of inducing breach of contract Solvency problems

Kind of case

Other

6

Suits by lawyers Suits by real estate brokers Breach by employees Other probable solvency cases Slander and misrepresentation Improper influence Strikes and boycotts Conspiracy Competition Other cheaper-cost-avoider cases Miscellaneous

15 22 5 18 9 11

6

15 11 18 48 (35%)

Total

Inducer is lower-cost avoider

70 (51%)

18 (13%)

employer to break his contract with the employee. Clearly, as between the employer and the insurance company, the latter was the lower-cost avoider of the mistaken discharge of the employee. It had the relevant expertise in evaluating the condition of his finger. The employer relied on the insurance company's superior expertise in discharging the employee and could have ascertained the true condition of the finger only after an investigation that would have been more costly to it than to the insurance company.

Appendix. Proof That Liability Rules in Table 7.2 Create Correct Incentives to Take Care 1. It is never in the interest of all three parties, acting independently, to behave negligently and select the x0, y0, and z0 levels of care. If the contrary were true, it would follow from Table 7.2 that A(x*) > p°(x0, z0)ph(y0)saDb + Λ(χ0);

(7.A1)

B(y*) > p (x0,z0)p (y0)s D

(7.A2)

a

b

b

b

+ B(y0);

C(z*) > P"(XQ,Z0)D" + C(z0).

(7.A3)

That is, for each party to behave negligently, his costs of using due care (given that the others are negligent) must exceed his expected costs of

226 · The Economic Structure of Tort Law behaving negligently. If the above inequalities hold, then A(x*) + B(y*) + C(z*) > L(x0,y0,z0). This must be false, however, since A(x*) + B(y*) + C(z*) < L(x*,y*,z*) < L(x0,y0,z0). Thus, all three parties would not be negligent, because if any two were then it would always be in the interest of the third party to choose a nonnegligent level of care. 2. Suppose the victim uses due care. Would a feasible outcome be one where both injurers are negligent? If it were, then A(x*) > pa(x0,z*)[D° + f(y0)s°Db]

+ A(x0)

B(y*) > p"(Xo,z*)p (y0)s D + B(y0), b

b

b

(7.A4) (7.A5)

which implies that A(x*) + B(y*) > L(x0,y0,z*) - C(z*). But this is false, because A{x*) + B(y*) < L{x*,y*,z*) - C(z*) < L(x0/y0,z*) - C(z*). Therefore, if the victim uses due care, at least one of the two injurers will find his expected costs minimized by also using due care. Suppose Β chooses y*. Then for A to choose an inefficient care level implies that A{x*) > pa(xQrz*)[Da + pb(y*)Db] + A(x0).

(7.A6)

This must be false, because, by definition, x* minimizes the right-hand side of (7.A6). Hence, if A expects that Β and C will choose due care, A will also be led to choose due care. A similar argument can be used to show that if Β expects A and C to use due care, B's expected costs will be minimized by choosing due care. For Β not to choose y* requires that B(y*) > p"(x* ,z*)pb(y0)Db + B(y0), which is false because y* minimizes the right-hand side of the latter expression. 3. For the victim to have an incentive to behave negligently, given that the injurers are not negligent, implies from Table 7.2 that p%x*,z0)[D" + pb(y*)Db] + C(z0) < p°(x*,z*)[D° + pb(y*)Db] + C(z*).

(7.A7)

Because L(x*,y*,z*) < L{x*,y*,z0), it follows that L(x*,y*,x*) - A(x*) B(y*) < L(x*,y*,z0) - Λ(χ*) - B(y*). Therefore, (7.A7) is false and hence, given x* and y*, the victim chooses z*. 4. If (x*,y0,z0) (one injurer careful, the other and the victim careless) were a feasible solution, Β would choose to be negligent if B(y*) > pa(x*,z0)pb(y0)Db

+ B(y0).

(7.A8)

We know from the definition of y*, x*, and z* that B{y*) < pa(x*rz*)pb(y*)Db + B(y0) < pa(x*,z0)pb(y0)Db + B(y0) and hence that (7.A8) is false. Because Β now selects y*, this initially yields the x*, y*, and z0 that we have previously shown is not a possible solution. Thus, the χ*, y0, and z0 outcome is dominated by the efficient solution.

Joint and Multiple Torts · 227 5. Finally, consider the only remaining inefficient outcome, x0, y*, and z0. For C to select z0 implies from Table 7.2 that C(z>) > p"(x0,z0)[D' + p\y*)Db] + C(z0).

(7.A9)

We know, however, that C(z*) < p'(x*,z*)[D" + pb(y*)Db] + C(z*) < p"(x0,z0)[D)

Accident (p\nv) No violation (1 - Φ ) No accident (1 - p\nv)r

Figure 8.1

232 · The Economic Structure of Tort Law Table 8.1.

Numerical example of outcomes

Violation No violation

φ

1 - φ

.6 —

— .4

p\v p\nv .5 —

— .5

D($)

Expected damages ($)

100 100

30 20

and of not violating the statute, respectively. But suppose that the injurer spent more resources to avoid violating the standard and as a result φ = .1 and 1 - φ = .9. The expected damages of violating and of not violating the statute would now be $5 and $45. Although expected damages in the violation sequence are greatly reduced (from $30 to $5), this reduction is offset by greater expected damages in the compliance sequence. Total expected damages, therefore, remain constant at $50. Even if the probability of violating the standard is reduced to zero, total expected damages remain at $50. The reason for this result is that the conditional probabilities of an accident (p\v and p\nv) when the standard is violated and when it is not violated are identical, so that adding units of ζ will not lower expected damages. From an economic standpoint, therefore, violating the standard is not negligent or, if one wants to use the word, is not a cause of the accident, because expected accident costs (ignoring the costs of compliance) do not decrease with greater levels of compliance. But if we modified the numbers in Table 8.1 to allow p\v > p\nv, then expected accident costs would fall as the probability of complying with the standard rose, and from an economic standpoint violating the standard could be found to have caused the accident. A more complete analysis, however, requires that we take account of the costs of compliance (the costs of z) before determining negligence and causation. We do this by first rewriting our standard loss function L(x,y) = p(x,y)D + A(x) + B(y) as

L = (φ)(ρ|υ)Ε> + (1 - ){p\nv)D + A(x) + B(y,z).

(8.2) (8.3)

The injurer's and victim's due care levels, x* and y*, are determined as before by balancing the reduction in expected accident damages (marginal benefits) against the marginal cost of care. This requires setting the partial derivatives of Eq. (8.3) with respect to χ and y equal to zero. The resulting marginal benefits of care are now the weighted averages

Causation · 233 of the reduction in conditional probabilities—the weights being φ and 1 - φ respectively.8 Minimizing L with respect to ζ yields -z(p\v-p\nv)D = B2,

(8.4)

where - φ2 is the marginal reduction in the probability of violating the applicable standard of care and Bz is the marginal cost of z. The efficient level of care, z, depends on the value of the parenthetical expression, p\v - p\nv, that is, on the difference between conditional probabilities of the accident. If it is zero, the optimal care (z*) is also zero, and the defendant's failure to take care should not be deemed actionable negligence or, alternatively, the cause (or a cause) of the accident. But if p\v — p\nv is positive, z* will also be positive provided that the cost of the first unit of ζ is less than the expected reduction in damages brought about by that first unit. The injurer would be deemed negligent and to have caused the injury and thus to be liable, if ζ was such that the benefits from additional efforts to comply (the left-hand side of Eq. 8.4) were greater than Bz, the marginal costs of those efforts. 9 Eq. (8.4) refines the Hand formula by making explicit that taking care to lower the probability of negligent behavior must, if it is to have economic significance, also reduce the probability of an accident. The refinement requires that pz be multiplied by the difference between the probability that the accident will occur if care is not taken (p\v) and the probability that it will occur even if care is taken (p\nv). This redefinition enables us to analyze the cases discussed later without invoking causal concepts. Nor need we confine the analysis to negligence cases. For example, as we know from Chapter 3, someone who is subject to strict liability will have an incentive to take due care in the Hand formula sense in order to minimize his liability.

8. W e have -pxD = A, and - p„D = By where px = $d(p\v)!dx + (1 - )d(p\nv)/dx, and py is similarly defined. 9. The model can easily be extended to answer the question whether the victim's violation of an applicable level of care (for example, by trespassing) caused the accident or should be deemed contributory negligence. Now φ would refer to the probability of the victim's violation, ζ to the victim's inputs to reduce this probability, and A. to the victim's marginal cost of z.

234 · The Economic Structure of Tort Law

Classes of Tort Cases Where Causation Is an Issue Cases Where the Accident Would Have Happened

Anyway

In many so-called causation cases the defendant is excused from liability because the accident would have occurred even if the defendant had taken care. For example, in Rouleau v. Blotner10 the defendant failed to give a turn signal and collided with the plaintiff. The evidence showed that the plaintiff's driver would not have seen the signal even if it had been given; therefore the accident would have occurred even if the defendant had not been negligent. A similar case is City of Piqua v. Morris.11 The defendant stored water for its reservoirs in ponds that had overflow wickets. Because of the defendant's negligence, the wickets became stopped up. There was a flood, the wickets could not carry off the excess water, and the plaintiff's farm was damaged. But the flood was so big that the wickets could not have carried away the water even if they had not been stopped up; thus the plaintiffs damage would have been just as great if the wickets had been clear. The plaintiff was held not to be entitled to recover any damages. In a case that is typical of the many statutory cases where liability has been denied on this ground, Weeks v. McNulty, 12 the defendant failed to put fire escapes on his hotel as required by statute, and the plaintiff's decedent died in a fire. But the evidence was that the fire escape would not have averted his death. He tried to get out only by the door; he could have jumped from the window to an abutting building as other guests had done, despite the absence of a fire escape. Using Weeks as an example, let φ be the probability of not complying with a statute requiring fire escapes on the building. Evidence that the fire escapes would not have saved Weeks means, in the context of our model, that the probability of his death was independent of whether or not a violation occurred—that is, that the conditional probabilities in Eq. (8.4) were equal and hence that due care with respect to ζ was zero. There may seem an element of paradox in asserting that McNulty was not really negligent, when the premise of the court's decision was that he had been negligent per se but should be excused for want of a causal relationship between negligence and injury. But this just shows that the courts do in two steps what the economist, with the refined version of the Hand formula, does in one. The court first asks whether there is 10. 84 N.H. 539, 152 Atl. 916 (1931). A similar but more recent case is Peterson v. Nielsen, 9 Utah 2d 302, 343 P.2d 731 (1959). 11. 98 Ohio St. 42, 120 N.E. 300 (1918). 12. 101 Tenn. 495, 48 S.W. 809 (1898).

Causation · 235 negligence in a rather crude sense—a failure to do something that in the general run of cases would be socially cost justified. Then, under the rubric of causation, it asks whether taking additional care would have been socially cost justified in the particular circumstances of the case, given other factors that may have made additional care fruitless. Our analysis of Weeks v. McNulty has been criticized on the ground that we use an ex post analysis to conclude that, ex ante, the hotel's owner used due care. 13 The criticism is incorrect. The point is that the owner's carelessness did not in fact make it more likely ex ante that Weeks would die, given the particular circumstances of the fire. For we know, although after the fact, that even if the hotel owner had been careful, Weeks would have died anyway. This is not to say that the owner necessarily gets off scot-free; if his carelessness resulted in death or injury to others, he would be liable to them, and the threat of this liability will give him the correct incentive to take care. 14 As we shall see in the next chapter, there is even a sense, and an important one, in which McNulty could be thought to have harmed all, including Weeks, whom he exposed to danger: when McNulty was careless, but before there was a fire, he reduced the life expectancy of each of his guests by his failure to install fire escapes. A theory of tort damage could be constructed—may indeed, as we shall see in the next chapter, already exist in embryo—in which such harms are actionable. Then the estate of Weeks could collect damages, together with all of the other guests. But his damages would be the costs of a reduced life expectancy, not the costs of death, and it was the latter that his estate sought. In the City of Piqua case, φ is the probabililty of a big flood. Unlike the situation in Weeks, ζ is not a factor in the analysis because the occurrence of such a flood is independent of the defendant's expenditures on 2. The conditional probabilities are now the probabilities of damage from letting wickets get stopped up given a big flood (p\v) and given normal flooding (p\nv).15 The optimal level of the defendant's care (y) in cleaning wickets is reached when 13. Richard W. Wright, "Actual Causation vs. Probabilistic Linkage: The Bane of Economic Analysis," 14 /. Legal Stud. 435, 4 5 3 - 5 4 (1985). 14. To see this, expand Figure 8 . 1 — a n d hence the loss function in Eq. (8.3)—to include circumstances in which fire escapes would reduce the conditional probability of an accident, circumstances in which it would not, and circumstances in which fire escapes were installed. Holding the defendant liable for failing to install fire escapes in the first but not the second set of circumstances (or of course the third) would be sufficient to give him an incentive to install fire escapes if they are required for due care. 15. We assume only two types of weather, but the analysis can easily be generalized to more.

236 · The Economic Structure of Tort Law - [φθ(ρ|σ)% + (1 - )d(p\nv)/dy]D = B,

(8.5)

Notice that d(p\v)/dy will equal zero because expenditures to clean wickets will not lower the probability of damages from stopped-up wickets when there is a big flood. In other words, the marginal product of care given such a flood is zero. Holding a defendant liable only if normal flooding occurs (the nv state above) is sufficient to give owners of dams the proper incentive to take due care by balancing the costs of cleaning wickets against the expected reduction in damages. To hold a defendant liable for the consequences of an accident that would have occurred even if he had not been negligent (in the economic sense) will yield no allocative gain because, by assumption, liability will not deter a similar accident in the future. Adjudicating liability and transferring money through the legal process from defendant to plaintiff will cost something, and the costs will yield no social benefit if we ignore, as we do throughout this book, possible differences between victims and injurers in attitudes toward risk. There is also the danger that if the defendant is liable for accidents that he cannot prevent, he will use excessive care. True, if the standard is negligence, if negligence has no stochastic element, if the negligence standard is applied on an individual rather than average basis (and has no other pockets of strict liability, either) and if the standard is applied correctly in every case, then liability of however broad a scope will have only one effect—to induce the defendant to take due care and thereby avoid liability. But as noted in earlier chapters, these conditions are too severe; and when they are relaxed, liability for accidents that would not be prevented by the exercise of due care, like the imposition of punitive damages (see Chapter 6), will have misallocative effects. This is clearest when the liability standard is strict liability rather than negligence. Consider an ultrahazardous activity such as blasting where strict liability is the legal standard. We can write the expected incomes of the victim (A) and injurer (B) when blasting occurs and when it does not as follows: Blasting: I0 = V + Ib0 - p(y*)D No blasting: i, = 1" + If

B{y*)

(8.6) (8.7)

I„ and /, are total net income with and without blasting, respectively. /a is A's income, and 1$ and l\ are B's income with and without blasting. We assume that it is too costly for the victim to alter his activity level (for example, by moving) to avoid damages (D) from blasting or to take care to reduce his expected damages (that is, we assume that x* = 0).

Causation · 237 This set of assumptions provides, of course, the economic rationale for imposing strict liability on ultrahazardous activities. It follows from Eqs. (8.6) and (8.7) that blasting is a socially valuable activity if I0 > /,, which requires that (P0 - I\) - p(y*)D - B(y*) > 0.

(8.8)

That is, the injurer's gain in income from blasting must exceed the sum of expected damages and costs of care. With strict liability, the injurer's decision will be the socially correct one. He will have an incentive to use y* (due care) if he goes ahead and blasts; he will decide to blast only if his net gain, which includes a deduction of his expected liability of p(y*)D, is positive; therefore no direct judicial control of the level of blasting will be necessary. Now let Ζ equal the expected damages to A that would result whether or not Β does blasting. Suppose that the foundation of A's house is so weak that it would collapse soon whether or not Β blasts. Obviously the socially correct decision on blasting is the same as before because Ζ is deducted from both Eqs. (8.6) and (8.7) and therefore does not enter (8.8). Because Β will be liable for Ζ only if he engages in blasting and thus can avoid liability by not blasting, he will choose to blast if (ig - JJ) - Ζ - p(y*)D - B(y*) > 0

(8.9)

but not otherwise; and in some circumstances, when (8.9) is negative, he will not blast when it is efficient to do so, and there will be too little blasting. For example, suppose that lb0 - I\ = $50, Ζ = $20, p(y*)D = $25, and B(y*) = $10. There is a net gain of $15 from blasting (a $50 gain in gross income to Β minus $35 in expected damages and cost of care). Yet if Β is liable for Ζ if he blasts, his net income from blasting will be — $5 and he is better off not blasting. Because strict liability is present to some degree even when the formal liability standard is negligence, the above analysis applies pro tanto to negligence cases: imposing liability for negligence where care would not have avoided the accident could produce an inefficient reduction in the injurer's activity level as well as administrative costs not offset by any allocative gain. The Injury Would Not Have Occurred If the Defendant Had Not Been Careless, But the Defendant's Negligence Did Not Increase the Probability of an Accident Ex Ante In an important class of cases, taking care would avoid an accident but liability is denied because, ex ante, the accident was not more probable

238 · The Economic Structure of Tort Law on account of failure to take care. In Berry v. Sugar Notch Borough, 16 for example, a case in which the causal issue pertained to care by the victim rather than by the injurer, the plaintiff was the motorman of a trolley. While running the trolley at an excessive rare of speed, he was injured when a tree crashed down and crushed the roof of the trolley car. The defendant municipality had been negligent in failing to remove the tree; but had it not been for the motorman's negligence in driving at an excessive rate of speed the accident would not have occurred, because if the motorman had been driving more slowly he would not have been under the tree when it fell. Because the probability that the tree would fall on a trolley car was not increased by the motorman's decision to speed, p\nv was no less that p\v,17 and the defense of contributory negligence was correctly rejected. An expenditure on ζ by the motorman would have had no expected benefit in the form of a reduced likelihood that he would be hit by a falling tree. Wright challenges our analysis of Berry as being inconsistent with our analysis of Weeks v. McNulty. 18 He argues that, ex post, we know that the motorman should have driven either slower or faster, because either way he would have avoided the accident and at a cost less than the accident cost. Thus if Berry, unlike McNulty, had been careful (driven slower), the accident would have been averted. But this does not provide a good argument for holding Berry liable (in the sense of unable to recover damages for his injury). The effect would be to induce him and others like him to drive more slowly, but there would be no less damage from falling trees; the incremental expenditure on care would buy no reduction in expected accident costs. The same would be true of making McNulty liable for deaths that would have happened anyway. He would spend more on safety but his additional expenditures would, by definition, save no one who would have died in the absence of added precautions. A case similar to Berry is Ehret v. Village of Scarsdale. 19 The plaintiff's decedent entered a recently completed and not yet occupied house and 16. 191 Pa. 345, 43 Atl. 240 (1899). For other examples see Fowlks v. Southern Ry. Co., 96 Va. 742, 32 S.E. 464 (1899); Central of Georgia Ry. v. Price, 106 Ga. 176, 32 S.E. 77 (1898). 17. The probability of the tree's hitting the car may actually have been reduced by speeding because the trolley was able to cover the same distance in less time. Excessive speed, however, might increase the force with which the tree hit the trolley (although not if the tree fell directly on the trolley rather than at least partly in front of it) and hence increase the amount of damages if an accident occurred. 18. See Wright, note 13 above, at 454-55. 19. 269 N.Y. 198, 199 N.E. 56 (1935). See also Metcalf v. City of Cortland, 56 A.D.2d 959, 392 N.Y.S.2d 719 (1977).

Causation · 239 (apparently) went to sleep there. His entry was unauthorized; he was a trespasser. While asleep, he was asphyxiated because the defendant developer had, through negligence, spliced the gas main in the street into a pipe leading into the house. 20 The defendant argued that it owed no duty of care to a trespasser. The plaintiff replied that the trespass was not a legal cause of the asphyxiation and therefore should not bar recovery. The court agreed with the plaintiff, in part on the ground that the asphyxiation was an unforeseeable result of the trespass. Although the house had not been sold, it was completed and ready for sale, and hence there was no reason for the plaintiff's decedent or a new owner (if the house had been sold) to think that he might be asphyxiated if he spent the night in it. But there is no need to invoke concepts of foreseeability to explain the result; the case could just as easily have been decided on the ground that the decedent's wrongful conduct, the trespass, did not increase the probability of injury. The danger of asphyxiation would have been the same, ex ante, whether the person staying in the house was an owner or a trespasser. In formal terms, p\nv was equal to p\v in Eq. (8.4).21 But if it had been less, foresight might then be relevant in the following sense. The marginal costs of precautions to the plaintiff (Λ2) include the costs of determining whether there is a need to take a precaution to avoid being injured while trespassing. When these costs are factored in, it will sometimes be economically efficient not to deem the plaintiff's trespass contributory negligence. Properly measured, Az may exceed the marginal benefits of ζ in Eq. (8.4). 22

Cases Where It Is Not Clear Whether the Accident Would Have Occurred Anyway In many cases it is not clear whether the accident would have occurred if the defendant had not been negligent. Weeks may have been such a case. Perhaps, as the court thought, if Weeks hadn't noticed that the 20. The other defendant, the Village of Scarsdale, was held not to have been negligent in issuing a permit to the developer to open the street and lay a drain pipe. 21. Herrick v. Wixom, 121 Mich. 384, 80 N.W. 117 (1899), is a similar case where the victim's trespass did not excuse the defendant's negligence. The plaintiff was a trespasser at a circus when a firecracker negligently exploded during a performance and injured one of his eyes. Either the probability of the plaintiff's injury was independent of whether he had a ticket or was a trespasser, or the probability that someone in the audience would be injured was only trivially increased by adding a trespasser to the audience; either way, p\v — p\nv was zero or close to it. 22. Ordinarily the cost of avoiding trespassing is so low that the efficient solution is to hold the victim contributorily negligent. See discussion of last clear chance in Chapter 5.

240 ' The Economic Structure of Tort Law roof of the abutting building was a convenient escape route, he would not have noticed a fire escape either. But of course this is not certain. Another and clearer—or rather, less clear—case is New York Cent. R. Co. v. Grimstad.23 The plaintiff's decedent fell over the side of his barge, which because of the defendant's negligence did not carry a life buoy. Grimstad's wife noticed him overboard and ran into the cabin for a line, but when she emerged he had disappeared. The court held that the defendant was not liable for Grimstad's death, because it appeared that he would have drowned before his wife could have thrown a life buoy to him. In terms of Eq. (8.4), p\v - p\nv (the probability that Grimstad would have drowned if there was no life buoy on board minus the corresponding probability if there was one) was positive, but so small that the optimal ζ was close to zero. Grimstad thus differs only in degree, perhaps slight degree, from Piqua and Berry. For even if p\v — p\nv is not zero, if it is very small the costs of legal proceedings are apt to exceed the allocative benefits from inducing potential injurers to spend additional resources on ζ amount of care. Shavell has argued that the costs of legal proceedings provide no basis for excusing liability for freak (highly improbable) accidents, because those costs will be slight ex ante if accidents, and hence cases, are infrequent.24 But the probability of an accident was not low in Grimstad, and hence if recovery were allowed in such cases there could be many of them. Rather, it was the incremental probability of injury from failing to take care, p\v — p\nv, that was low, and this in turn reduced z* without reducing the administrative costs that the legal system would incur if liability were imposed. Another problem with imposing liability in Grimstad-type situations is similar to that discussed earlier when p\v - p\nv = 0. Liability can lead potential injurers to avoid socially valuable activities when the gain from remaining in the activity is less than their expected liability payments for accidents that would not have been avoided by their taking care.25 In a subsequent case, Kirincich v. Standard Dredging Co.,26 again the plaintiff s decedent fell overboard and drowned. The defendant's dredge, on which the decedent worked, was not equipped with adequate life23. 264 Fed. 334 (2d Cir. 1920). 24. See Steven Shavell, " A n Analysis of Causation and the Scope of Liability in the Law of Torts," 9 J. Legal Stud. 463, 483 (1980). 25. Liability creates a similar problem on the victim's side as well: if the victim is fully compensated, too many persons who cannot swim will be induced to board ships. 26. 112 F.2d 163 (3d Cir. 1940).

Causation · 241 saving equipment. But this time the appellate court reversed a judgment for the defendant. Kirincich had actually reached for the line that his shipmates had tossed to him, and the court thought that if the object had been larger and more buoyant he would have been able to grasp and hold on to it. The difference between the two cases is not that Grimstad would have died even if the defendant in that case had taken care and Kirincich would have been saved, but that there was a much greater probability that care would have been effective in Kirincich's case than in Grimstad's (that is, p\v - p\nv was greater in Kirincich than in Grimstad). When p\v - p\nv is insubstantial, the costs of legal proceedings, the possibility of deterring socially valuable activities, and excessive injurer care are likely to be the dominant considerations; hence liability will be unwarranted. When p\v - p\nv is substantial, expenditures on care are likely to have a larger effect in reducing the probability of an accident. Hence liability is likely to induce precautions of greater social value in the second type of case. A difficult group of cases involve cancers of traumatic origin. For example, in McGrath v. Irving27 the plaintiff was thrown against the side of a car in a collision, inhaled some glass, and later developed cancer of the larynx. A laryngologist testified at trial that in his opinion "the events, the accident, the inhalation of glass were a cause of accelerated development or growth of [plaintiff's] cancer." 28 Other physicians testified that they thought there was no causal relationship between the accident and the cancer. The plaintiff's expert was evidently not certain that plaintiff would not have gotten cancer of the larynx if she had not inhaled the glass. His reference to "accelerated development or growth" implies that he thought she had some latent or incipient cancer which might have grown even without the irritation produced by the inhalation of the glass. The case was thus one of probability well short of certainty. In McGrath, although no effort at quantification was made, the court thought the jury had found that it was more probable than not that the cancer would not have occurred if the defendant had been careful and avoided the accident (that is, p\v - p\nv was substantial). With McGrath compare Doumitt v. Diemer, 29 where the plaintiff sued a firm of x-ray specialists for disfiguring her by negligently treating a 27. 24 A.D.2d 236, 265 N.Y.S.2d 376 (1965). See also National Dairy Products Corp. v. Durham, 115 Ga. App. 420, 154 S.E.2d 752 (1967); Annot., 2 A.L.R.3d 384 (1965); and the excellent discussion in Bert Black and David E. Lilienfeld, "Epidemiologic Proof in Toxic Tort Litigation," 52 Fordham L. Rev. 732 (1984). 28. 24 A.D.2d at 237, 265 N.Y.S.2d at 378. 29. 144 Ore. 36, 23 P.2d 918 (1933).

242 · The Economic Structure of Tort Law tubercular gland in her neck by means of x-rays. The plaintiff's own medical witness testified to the effect that it was impossible to tell whether the disfigurement would not have occurred had it not been for the xray treatment, because the underlying tubercular condition might have been disfiguring. That is, it was impossible to tell whether p\v — p\nv was positive or zero. The court ordered the complaint dismissed. Cases Where the Identity of the Victim Is Unclear In some cases it may be certain that injuries have occurred because of someone's negligence but highly improbable that any given injury is the result of that negligence. For reasons about to become clear, cases acknowledged to be of this character are nonexistent in tort law, but they are an interesting class of cases in principle because they illustrate a shortcoming of the common law system from an economic standpoint. Suppose that a nuclear reactor emits radiation into the surrounding area; either the emission is negligent or the standard of liability is strict, as probably it would be; and it is known with certainty that the effect of the emission will be to increase the number of cancers in the area of exposure from 100 to 111 over the next twenty years. The owner of the reactor should be liable for the cost of 11 cancers, because that is the number that would have been avoided if he had been careful. But we do not know which of the 111 cancers would not have occurred if the utility had not emitted the radioactivity. The probability will be only 10 percent that any one of the 111 victims would not have gotten cancer if the utility had not emitted radioactivity, and this probability is well below the threshold at which liability is, or should be, imposed; if a 10 percent probability were sufficient for liability, then every one of the 111 cancer victims could recover damages from the utility, resulting in grossly excessive damages. The solution to this problem within the framework of the common law tort system is not obvious, and we devote the next chapter to its consideration. Cases Where the Injurer's Identity Is Unclear In Summers v. Tice, discussed in the preceding chapter, two careless hunters fired in the direction of the plaintiff, who was struck by a single bullet; and both hunters were held liable as joint tortfeasors although only one had caused the injury. Here is a clear example of liability imposed on an individual (one of the hunters) whose act was not a cause of the plaintiff's injury. Yet we know from the previous chapter that

Causation · 243 this result is consistent with the economics of the problem, because if both defendants are jointly liable when they use less than due care, it will create incentives for both to take due care. This is some evidence that economics provides a clearer guide to understanding the structure of tort law than an approach that relies on noneconomic concepts such as causation.

The Probability of the Injury Was So Low That Either the Costs of Foresight or Administrative Costs Make Legal Proceedings Unwarranted We move now to a diverse class of cases where p\n - p\nv is positive, and where, unlike the Grimstad or nuclear reactor cases, there is no danger of imposing liability for injuries that might have occurred anyway (that is, p\nv is approximately equal to zero)—yet still the courts refuse to impose liability for the accidents that do occur. We begin with the class of cases in which people endangered but not physically injured by negligent conduct seek to recover damages for mental distress. If A suffered mental distress as a result of witnessing an accident caused by B's negligence, courts traditionally would not award damages, although recently most states have abandoned the traditional rule.30 We can analyze this case by positing two possible states of the world, s and s', with p(s) and p(s') denoting the probabilities that each state will occur. The sum of p(s) + p(s') is l. 3 1 In the type of case under consideration, s is the state where if an accident occurs there is personal or property damage but the eyewitness to the accident suffers no significant nervous injury, and s' is the state where the only injury if an accident occurs is nervous injury to the eyewitness. 32 Let p(y\s) and p(y\s') denote the conditional probabilities of the accident given states s and s' respectively. To simplify, we assume that the plaintiff's care is fixed and 30. An intermediate position adopted by some courts was that recovery for purely mental distress would be allowed if the victim was within the range of ordinary physical peril from the defendant's conduct. For the evolution of the case law see Dulieu v. White & Sons, 1901 2 K.B. 669; Waube v. Warrington, 216 Wis. 603, 258 N.W. 497 (1935); Strazza v. McKittrick, 146 Conn. 714, 156 A.2d 149 (1959); Dziokonski v. Babineau, 375 Mass. 555, 380 N.E.2d 1295 (1978); Bovsun v. Sanperi, 61 N.Y.2d 219, 461 N.E.2d 843 (1984); and for a fine and up-to-date general analysis see Peter A. Bell, "The Bell Tolls: Toward Full Tort Recovery for Psychic Injury," 36 U. Fla. L. Rev. 333 (1984). 31. One can easily generalize the analysis to η possible states of the world where Ip(s') = 1. 32. We could make the analysis more realistic by adding a third state where both mental injury to the witness and physical or property damage occur, but this complication would not alter our conclusions.

244 • The Economic Structure of Tort Law equal to due care, so that the conditional probabilities depend only on the potential injurer's care. We can write L as L = p(s) · p(y|s)D* + p(s') • p(y|s')Ds' + B(y),

(8.10)

where D5 and D5' denote damages in each state given the accident. In Eq. (8.10), L is minimized when - [p(s)py-J> + p(s> yk D»'] = B„,

(8.11)

where pys and py\s· denote partial derivatives. Thus we want the potential injurer to add inputs of care, y, up to the point where their additional cost is just equal to the sum of the expected reduction in damages in the two states of the world. If, however, p(s') is very low, as it surely is in most cases where the plaintiff merely sees or hears about an injury to someone else, the impact of the second term of the left-hand side of Eq. (8.11) on due care is likely to be small. As Shavell has pointed out, this fact in itself is not necessarily decisive.33 First, the mariginal product of care (py|s) or the amount of damages in s' may be large, tending to offset the effect of a low p(s') on care. Second, even if the incremental expenditure required for due care is tiny because there is a very low probability of an accident, the administrative costs of invoking the legal process if such an accident occurs will also be tiny, ex ante, because such cases will be very rare. Nevertheless, if p(s') is very low, courts will have difficulty calculating the increment in care (y). And because the costs of foresight are positive, potential injurers may not take precautions (even precautions that would be cost justified if the costs of foresight were zero) against extremely low-probability injuries, in which event liability will have zero allocative effect just as if p(s') were actually zero. Two additional considerations support denial of liability in the nervous injury cases. First, the average nervous injury to a witness of an accident is small relative to the average physical injury to the (other) accident victim. Hence there is little likelihood that the effect on optimal care of a very low probability of state s' would be offset because Ds' was much larger than Ds. Second, the administrative costs of nervous injury cases are (or were) considerable because of the difficulty and uncertainty of proof of nervous injuries, 34 and the more costly it is to establish liability 33. See Shavell, note 24 above, at 483. 34. This makes the nervous injury case analogous to the Grimstad case. We could redefine s' as the state where one develops nervous injury and p(y\s') as the probability of nervous injury if one witnesses an accident minus the probability of nervous injury if one does not witness an accident (analogous to p\v - p\nv in Grimstad). Because it is difficult to

Causation · 245 accurately, the greater must be the effect of liability on due care to make a legal proceeding socially worthwhile. The second point may offer a clue to the trend away from the traditional common law rule. With society's greater knowledge of mental illness, it is easier today than it once was to make reliable diagnoses of mental conditions and so distinguish spurious from genuine claims of traumatic mental injury. Administrative costs may also explain why common law courts traditionally did allow damages for mental distress when the plaintiff had been physically as well as mentally injured. 35 Because the victim would be in court anyway to sue for his physical injuries, the incremental cost of establishing damages for nervous injury would be less than the cost of a separate lawsuit by a witness to the accident. The attempt to balance allocative gains with administrative costs is visible in the various rules of thumb in modern law that have replaced the traditional approach of limiting damages for nervous injuries to the victim of physical injury. For example, if the right to get such damages is limited to persons physically endangered even though not actually physically injured, the probable damages will be on average higher relative to the costs of verifying their existence than if any witness to the accident could sue for such damages, and much higher than if someone who merely heard about the accident could sue. Many people are distressed to hear of airplane crashes but it would be inefficient to allow all of them to sue the airline. The idea that the low probability of an accident should excuse the injurer's negligence has been resisted on the ground that the particular injury resulting from the negligent conduct is always a low-probability event and the sum of many low-probability events that due care would avert may be a high probability of a serious accident. 36 If A drives at a high speed with his eyes closed, the probability that he will cause some injury is considerable, but the probability that he will injure a particular person in a particular way is slight. In Ferroggiaro v. Bowline37 the defendant, driving carelessly near an intersection, ran off the road and hit a power transmission pole which controlled the traffic lights at the intersection. The collision extinguished the lights and as a result two determine whether the nervous injury would have occurred anyway (p\nv is positive and therefore - p\nv is very small), the case can be analyzed just as Grimstad was. 35. As in the Waube case, note 30 above. 36. This point is stressed by Shavell, note 24 above, at 483-84, citing earlier scholars who had made the point. 37. 153 C.A.2d 759, 315 P.2d 446 (1957).

246 · The Economic Structure of Tort Law cars collided at the intersection. A passenger in one of the cars was injured. Despite the freak character of the accident, the court held that he could recover damages from the defendant. The defendant's careless driving created a danger of an automobile accident; the fact that the precise manner of the accident could not have been foreseen did not provide a reason for excusing his conduct. The most famous freak-accident case is Palsgraf v. Long Island R. Co. 38 The railroad's conductor was careless in trying to help a passenger board the train after the train began moving. The passenger stumbled and dropped an unmarked package on the platform. The package contained fireworks which exploded on impact. The explosion panicked the crowd standing on the platform. The crowd bolted and knocked over a heavy scale at the other end of the platform, which hit Mrs. Palsgraf and injured her. 39 The court held that the railroad was not liable for her injury. This was not a case like Ferroggiaro where careless conduct created a large number of low-probability accident possibilities. Rather, Palsgraf was like the nervous injury cases in involving two distinct states of the world, one far less probable than the other. The probable state of the world (s) was that the conductor would injure the passenger by helping him to board a moving train; the improbable state (s') was that if the passenger stumbled one or more people on the platform would suffer a serious injury. Making the railroad liable for injuries to nonpassengers in this situation would have had no appreciable effect on the railroad's level of care, because the costs of identifying such a state of the world would have exceeded the benefits to the railroad from taking precautions against its occurrence. In another famous proximate-cause case, Watson v. Kentucky & Indiana Bridge & Ry. Co., 40 the defendant through negligence allowed a railroad tank car to be derailed and leak gasoline, which exploded when a bystander deliberately put a match to it. The explosion injured the plaintiff. The court held that this act of arson was so unforeseeable that the defendant was not liable for the plaintiff's injury. If the arson had been foreseeable, the defendant and the arsonist would have been treated as joint tortfeasors, and we know from the preceding chapter that the defendant would in that event have been liable for the plaintiff's full damages. But the possibility of arson was so slight (it was an act of pure 38. 248 N.Y. 339, 162 N.E. 99 (1928). 39. The court's opinion states that the explosion itself knocked the scale over but apparently it was the crowd that knocked it over. See John T. Noonan, Jr., Persons and Masks of the Law 119 (1976). 40. 137 Ky. 619, 126 S.W. 146 (1910).

Causation · 247 malice, with no possibility of pecuniary or any other benefit—except the delights of pyromania) that, as in Palsgraf, the defendant would not have taken account of it in deciding how much care to use; so imposing liability would have had no allocative effect. This result can be further defended by recalling from Chapter 4 the one-bite rule that governs liability for dog bites. Put more precisely, the rule is that the owner is liable only if he has reason to suspect the dog's vicious disposition; and ordinarily there is no reason to suspect it until the dog has bitten someone. Even if the probability of the dog's biting someone is very high, the owner will not be liable unless he has reason to know it is high. Otherwise, as we have said, liability will have no allocative effect. The Palsgraf and Watson cases differ from the first-bite example in that in both those cases the defendant was negligent with respect to the accident probability that was known; therefore it may seem harmless to make the defendant liable for an unforeseeable type of accident as well. It may indeed be harmless, because if the accident is unforeseeable then so is the imposition of liability for its consequences. Hence there is no danger as with some of the other causal cases we have discussed of inducing too much care. But if imposing liability will have no allocative effect, it will confer no economic benefit; it will merely require a costly transfer payment. Palsgraf involved not only the first element in the nervous-injury cases, a low p(s'), but the third as well, high administrative costs. The causal sequence alleged in the Palsgraf case, although possible, was not only improbable but implausible. Maybe the injury to Mrs. Palsgraf occurred as alleged but it seems as or more likely that she took the opportunity afforded by her fortuitous proximity to the explosion of the fireworks to seek damages for an unrelated condition. This suspicion is reinforced by the fact that the principal symptom for which she sought damages was a stammer. 41 The causal sequence in Ferroggiaro was improbable but there was no doubt that it occurred precisely as alleged. Another class of cases where administrative costs are high, although for a slightly different reason, is where a statute designed for one purpose is invoked for another. In Gorris v. Scott, 42 the plaintiff's animals were washed overboard and lost in a storm. The defendant's ship was not equipped with pens, as required by statute in order to prevent diseases from spreading among the animals. Had the ship been so 41. See Noonan, note 39 above, at 127. 42. 9 L.R.-Ex. 125 (1874). See Grady, note 7 above, at 382-83, for a different economic approach to this case from ours.

248

· The Economic Structure of Tort Law

equipped, the plaintiff's animals would not have been washed overboard in the storm; but the court excused the defendant from liability on the ground that the statute did not create a duty to avoid the type of accident that had occurred. At first glance the decision seems clearly wrong from an economic standpoint. Because the pens were required anyway, the incremental cost of care to prevent the animals from being washed overboard was zero, which had to be less than the expected cost of such an accident. But this reasoning assumes that the cost of the pens really was less than the expected cost of disease that the pens would have prevented, and it may not have been, depending on the motivation for the statute. Granted, even if the cost of the pen was not less than the expected disease cost, it may have been less than the sum of the expected costs of disease and of the animals' being washed overboard by a storm. Formally, if we have two classes of injuries, 1 and 2, where 1 is disease and 2 is being washed overboard, the loss function (ignoring victim care) is L = f{y)W

+ ?{y)O> + B{y).

(8.12)

Due care requires the potential injurer to set -(PP1

+ rfD2) = By

(8.13)

or - XpyD' = By, if there are more than two classes. But there is a practical problem with using this insight as a basis for liability in Gorns-type cases: it requires the court to consider a type of accident not before it in the litigation—in Gorris, to weigh the costs of the pens against their benefits in disease prevention when the only accident being litigated involved the washing overboard of plaintiff's animals in a storm. Alternatively, one could argue that limiting liability to state 1 (disease spreads because pens were not in place) creates sufficient incentives to take care and saves the administrative costs of legal claims in state 2. The court in Gorris could simply have taken for granted that the statute reflected a correct cost-benefit analysis, in which event, as we have seen, the incremental cost of due care to avoid the washing overboard was zero and had to be less than the incremental benefit. This approach is frequently taken in statutory due care cases under the name of "negligence per se," the rule that violation of a statutory standard of care is deemed negligence without opportunity for the defendant to show otherwise through the Hand formula or some related technique. Even so, or perhaps especially so, violation of the statute might not be negligent for a particular defendant, given his costs of compliance relevant to the

Causation · 249 benefits of violation. In such a case, adding to the penalties of the statute by making the defendant liable for an unrelated harm could produce overdeterrence. Here the court's causal analysis serves the purpose suggested earlier of refining the legal analysis of negligence. The legal concept of negligence sometimes diverges from the economic—notably in the idea of negligence per se based on breach of a statutory duty. The legal concept of cause comes to the rescue in these cases. The combination of negligence analysis and causation analysis equates to the economic concept of negligence and allows the optimal result to be achieved. Our discussion of Gorris is not intended to imply that courts always should, or always do, ignore the jointness of expenditures on due care. Suppose that there are η possible victims of a particular injurer, B, and an increase in B's input of y would reduce each victim's probability of an accident—that is, B's inputs of care are joint, as they were in Gorris. The loss function for the η victims may be written as L = np(x,y)D + nA(x) + By, and the injurer's efficient care level requires that - npyD = By. That is, the greater the number of potential victims, the greater the level of y that is required for due care. Consider the maintenance of a railroad crossing. The greater the number of travelers at the crossing, the greater will be the expected reduction in damages from a given maintenance expenditure, and hence the greater the railroad's due care level. The legal standard implicitly recognizes this type of jointness. 43 For example, Learned Hand in Conway v. O'Brien defined the level of care as a function of three factors: "the likelihood that [the defendant's] conduct will injure others, taken with the seriousness of the injury if it happens, and balanced against the interest which he must sacrifice to avoid the risk."44 The first factor is the probability of injuring not a particular person but any person and is equivalent to np(x,y). We turn now to two classes of cases in which, although the probability of the state of the world in which the accident occurred is, ex ante, slight, administrative costs are also small, with the result that liability is upheld against proximate-cause arguments by defendants. The first class consists of cases where victims are allowed to recover their full damages even though the severity of the injury was due to a preexisting condition of which the defendant could not have known. In the most famous of these "eggshell-skull" cases, Vosburg v. Putney, 45 the defendant, a schoolboy, kicked a fellow schoolboy on the shin. Unbeknownst to the 43. See id. at 381-85, and our discussion of this point in Chapter 4. 44. I l l F.2d 611, 612 (2d Cir. 1940), rev'd on other grounds, 312 U.S. 492 (1941). 45. 80 Wis. 523, 50 N.W. 403 (1891). Vosburg was discussed for another point in Chapter 6.

250 · The Economic Structure of Tort Law defendant, the plaintiff's shin had been weakened by an infection and as a result the kick caused a serious injury not intended by the defendant. In these and most other eggshell-skull or preexisting-condition cases the plaintiff is allowed to recover his full damages. The allocative gain is clear. If people with above-average susceptibility to injury were never allowed to recover full damages for the negligent injuries inflicted on them, the expected liability costs to potential injurers would be below the actual injury costs. There would be an artificial truncation of damages liability. Also, potential accident victims having eggshell skulls would be induced to take excessive precautions, whereas potential victims having rock-hard skulls might, if entitled to average rather than their actual damages, take insufficient precautions. And although many of the cases, such as Vosburg itself, have involved very low-probability events, it would be difficult to articulate and apply a rule that said that only the unusually susceptible could not recover damages; that everyone who was above average in susceptibility, but not too far above average, could recover. The administrative costs of determining the injury actually suffered by these victims are low, because they are already before the court. Notice, however, that to avoid excessive liability the damages to the eggshell-skull victim must be discounted by the probability that he would have escaped injury unscathed but for the defendant's tort. That probability was high in Vosburg. But suppose instead that the defendant's assault had triggered a latent psychosis in the plaintiff. Then there would be a significant probability that some other, nonliable cause would have triggered the psychosis. The plaintiff's damages should be reduced to reflect this probability. The law has taken this step too. 46 The second class of cases in which administrative costs are low are the rescue cases, illustrated by Wagner v. International Ry. Co. 47 The plaintiff and his cousin were passengers on the defendant's train. Through the defendant's negligence the cousin fell off the train. The plaintiff joined the rescue party and was injured. He was allowed to recover his damages from the defendant, because of the defendant's negligence in the injury to his cousin. This is the standard result in these cases, unless the plaintiff himself behaves negligently in carrying out the rescue effort. There can be no argument that allowing recovery of damages will entail additional lawsuits. The contrary is true, as should be evident from our 46. See, for example, Steinhauser v. Hertz Corp., 421 F.2d 1169, 1173-74 (2d Cir. 1970). 47. 232 N.Y. 176,133 N.E. 437 (1921). A similar case is Guarino v. Mine Safety Appliance Co., 25 N.Y.2d 460, 255 N.E.2d 173 (1969)—as well as the Eckert case discussed in Chapter 4.

Causation · 251 discussion of Eckert v. Long Island R. Co. in Chapter 4. To the extent that cost-justified rescue efforts are undertaken—and a rule allowing the rescuer to recover damages for injuries incurred in the course of the rescue attempt encourages such undertakings—the number of accident victims, and hence the number of lawsuits, will be fewer than if recovery is not allowed. Liability for Economic Loss Our final example of how lawyers' references to cause may conceal economic insight is the misleadingly named economic-loss cases, well illustrated by Rickards v. Sun Oil Co. 48 The defendant's barge negligently hit a drawbridge and wrecked it, thereby cutting off from the mainland the island on which the plaintiff's businesses were located. These businesses (hotels, cafes, a gas station and repair shop, and so on) suffered pecuniary losses as a result, but recovery was denied on proximatecause grounds. This is the usual result in cases where the only injury complained of from a collision or other physical harm is a business injury involving no physical harm to the plaintiff or his property. Thus, for example, if A through negligence damages B's plant, and Β as a result cannot fulfill his contract with C, C cannot sue A for the lost profits of the contract. The economic explanation is that, in Rickards for example, the accident to the drawbridge created external benefits as well as external costs. The business lost by merchants on the island presumably was recouped by merchants on the mainland. Because there is no way that the tortfeasor can obtain restitution of these benefits, if he were liable for this kind of accident he would be overdeterred, just as would happen if punitive damages were awarded in an ordinary negligence case or a defendant were made liable for an injury he had not caused. In a sense, the latter was true in Rickards; if the defendant had been liable for the island merchants' losses, it would have borne a cost greater than the net cost created by the accident. In the Oppen case, 49 discussed briefly in Chapter 4, the plaintiffs were commercial fishermen whose businesses were injured when the de48. 23 N.J. Misc. 89, 41 A.2d 267 (1945). For an up-to-date judicial analysis of the economic-loss cases that contains some economic analysis and references see Barber Lines A/S v. M/V Donau Maru, 764 F.2d 50 (1st Cir. 1985) (Breyer, J.). And for economic analyses in the spirit of our discussion see Grip-Pak, Inc. v. Illinois Tool Works, Inc., 694 F.2d 466, 4 7 3 - 7 4 (7th Cir. 1982); W. Bishop, "Economic Loss in Tort," 2 Oxford }. Legal Stud. 1 (1982); Richard A. Posner, Tort Law: Cases and Economic Analysis 4 6 7 - 6 9 (1982); Mario J. Rizzo, " A Theory of Economic Loss in the Law of Torts," 11 ]. Legal Stud. 281 (1982). 49. Union Oil Co. v. Oppen, 501 F.2d 558 (9th Cir. 1974).

252 · The Economic Structure of Tort Law fendant's negligent oil spill killed the fish on which the plaintiffs depended. If the supply of fish were so elastic that the reduction in catch caused by the oil spill was fully offset by an increase in catch elsewhere, then damages should be denied, because neither fishermen as a group, nor consumers, would have been injured by the spill. The court, however, allowed damages in Oppen and this seems the economically correct result. The spill probably caused a net reduction in the number of fish caught. The benefits to other fishermen were therefore more than offset by the losses to consumers from paying higher prices for fish and from substituting other foods. Because the plaintiffs did not own the fish, they could not sue for the destruction of property; hence their loss was superficially the same kind of pure business loss that was involved in Rickards. But in that case the owner of the bridge could have sued to recover the deadweight losses caused by the accident; there was no one in Oppen better placed than the fishermen plaintiffs to sue for those losses, so they properly were allowed to do so.50 Rabin has suggested that the economic-loss cases cannot be explained on economic grounds, and he has proposed—a rarity in noneconomic torts scholarship—a positive theory of the cases.51 His explicit criticism of the economic explanation set forth above is that many cases allow the recovery of purely pecuniary losses, that is to say, losses offset elsewhere in the economic system, so that there is no net social cost. The examples he gives, however, are inapt.52 He thinks that the collateral-benefits rule, whereby a plaintiff is not required to deduct from his damages the proceeds he may have obtained from an insurance policy covering the event triggered by the defendant's tort, allows the plaintiff to recover double. It does not; the plaintiff paid for the benefit in his insurance premium.53 The converse case, which Rabin also seems to think makes no economic sense, is the law's refusal to allow a life 50. Another federal court of appeals, in another oil spill case, recently upheld the denial (on economic-loss grounds) of damages sought by marina and boat operators, shipping interests that incurred losses from rerouting or delay, seafood companies not actually engaged in fishing, and recreational fishermen, leaving just commercial fishermen to seek damages, as in Oppen. Louisiana ex rel. Guste v. MA7 Testbank, 752 F.2d 1019 (5th Cir. 1985) (en banc). Although some of the exclusions make economic sense, especially the exclusion of marina and boat operators, whose losses may well have been largely offset by gains to marina and boat operators elsewhere, others do not. In particular there seems to be no economic distinction between commercial and recreational fishermen. The opinion does not discuss externalities and its attempt to distinguish the Oppen case is perfunctory. 51. See Robert L. Rabin, "Tort Recovery for Negligently Inflicted Economic Loss: A Reassessment," 37 Stan. L. Rev. 1513 (1985). 52. See id. at 1535 n. 72. 53. See Richard A. Posner, Economic Analysis of Law § 6.13 (3d ed. 1986).

Causation · 253 insurance company to recover damages for the wrongful death of the insured. 5 4 But the insurance company has sustained no loss; it was compensated ex ante for bearing the risk of death in the premium that the insured paid for the policy. Rabin's other example is the third-party beneficiary case. We saw in Chapter 5 that liability was denied in the Moch case because it seemed unlikely that the parties to the water contract had intended to confer benefits on the city's property owners. Where they do intend to confer benefits on a third party, liability is appropriate whether on tort or contract principles. They may have so intended in the J'Aire case, 55 which Rabin discusses at length. The plaintiff operated a restaurant on premises leased from the county. The county m a d e a contract with the defendant for renovations to the restaurant. The defendant, through negligence, unreasonably delayed in making the renovations, and the plaintiff was allowed to sue for the one month's profits lost as a result of the delay. The work was being done primarily for the benefit of the plaintiff rather than his lessor, and the defendant may have implicitly guaranteed the plaintiff against the financial consequences of an avoidable delay. The other cases that Rabin cites also fit the pattern sketched in our discussion of the Rickards and Oppen cases or are explicable by reference to administrative costs. Thus Rabin discusses two cases in which liability was denied that are just like Rickards: "Consider, for example, the plight of a plant operator whose customers are blocked from entry for an extended period w h e n the negligent spillage of a flammable substance requires the closure of the plant's sole access road. Or, consider the misfortune of a group of restaurant, bar, and motel owners w h e n the closing of a negligently manufactured bridge severs their commerce." 5 6 He contrasts these cases with Maez, in which "a drunken driver crashed into a power pole, smashing it and disrupting the electic current. As a result, there was a power surge in the plaintiff's factory that burned out an ammonia transformer and idled two employees for a short time. The plaintiff successfully brought an action for both the property damage to the machine and the employees' lost wages for which he was responsible." 57 Rabin believes that there is some economic inconsistency in al54. See Rabin, note 51 above, at 1524 and n. 37. 55. J'Aire Corp. v. Gregory, 24 Cal. 3d 799, 598 P.2d 60 (1979). 56. Rabin, note 51 above, at 1513 (footnotes omitted). The cases are Dundee Cement Co. v. Chemical Laboratories, Inc., 712 F.2d 1166 (7th Cir. 1983), and Nebraska Innkeepers, Inc. v. Pittsburgh-Des Moines Corp., 345 N.W.2d 124 (la. 1984). 57. Rabin, note 51 above, at 1516, discussing George A. Harmel & Co. v. Maez, 92 Cal. App. 3d 963, 155 Cal. Rptr. 337 (1979).

254 · The Economic Structure of Tort Law lowing recovery in this case and not the others. There is not. The drunken driver case involves a real social cost, measured by the cost of repairing the transformer and the opportunity costs of the employees' time. It is as if Rickards had been the owner of the bridge rather than a merchant on the island cut off from the mainland by the accident. It is true that the employees themselves would probably not have been able to recover their lost wages if they had sued, but there are two good economic reasons for this difference. The first is, once again, the risk of duplicate recovery (and hence overdeterrence); the worker's lost wages may overstate the social cost of the accident. Consider Stevenson v. East Ohio Gas Co., 58 where the plaintiff was denied recovery of damages for losing a week's work when an explosion caused by the defendant's negligence prevented the plaintiff from getting to his place of work. The plaintiff's employer may have made up the plaintiff's lost product by expanding production from another plant or deferring someone's vacation; therefore the net social cost may have been less than the loss in the plaintiffs' wages. But in Mineral Industries Inc. v. George, 59 where as in Maez recovery of damages for lost wages was allowed, the plaintiff was the employer rather than the employee; to the extent that the employer could make up for the loss of the employee by some adjustment elsewhere in his operation, his damages were automatically reduced. Another objection to allowing employees to sue for lost wages is the administrative costs of allowing numerous small claims. That may also be the reason for the suggestion in Petition of Kinsman Transit Co. 60 that if a driver negligently caused an accident in the Brooklyn Battery Tunnel that delayed traffic, he would not be liable for the costs of that delay. Although significant in aggregate, the cost would still not be great enough (in all likelihood) to warrant the expense of a class action, and individual lawsuits would cost far more than any possible allocative gain. So we do not think the cases are an economic hodgepodge. We also do not think that Rabin has been successful in putting forward an alternative positive theory. His "theory" is that the denial of liability in economic-loss cases rests on "a deep abhorrence to the notion of disproportionate penalties for wrongful behavior," 61 as reflected, for example, in the constitutional prohibition against cruel and unusual punishments. The issue in the economic-loss cases, however, is not 58. 59. 60. 61.

47 Ohio L. Abs. 586, 73 N.E.2d 200 (Ohio App. 1946). 44 Misc. 2d 764, 255 N.Y.S.2d 114 (N.Y. Sup. Ct. 1965). 388 F.2d 821, 835 n. 8 (2d Cir. 1968), discussed in Rabin, note 51 above, at 1536. Id. at 1534.

Causation

·

255

p u n i s h m e n t o r d i s p r o p o r t i o n b u t e x a c t c o m p e n s a t i o n : t h e i n j u r e r is j u s t being asked to m a k e g o o d his victim's losses. Certainly as b e t w e e n the i n n o c e n t v i c t i m s a n d t h e n e g l i g e n t o r o t h e r w i s e m i s b e h a v i n g i n j u r e r , it is h a r d t o j u s t i f y p l a c i n g t h e e n t i r e b u r d e n o n t h e v i c t i m w i t h o u t t a k i n g a c c o u n t o f e c o n o m i c c o n s i d e r a t i o n s ( s p e c i f i c a l l y t h e c o n f e r r a l of b e n e f i t s on others, such as Rickards' competitors). A n d notice the moral peculiarity o f e x c u s i n g t h e i n j u r e r f r o m all liability m e r e l y b e c a u s e a f r a c t i o n o f t h a t liability s e e m s u n d u e . T h i s m a y m a k e e c o n o m i c s e n s e , a s w e h a v e a r g u e d , b u t w e d o n o t s e e w h a t n o n e c o n o m i c s e n s e it m a k e s . 6 2 62. On the failure of torts scholarship to produce persuasive positive theories of liability in opposition to the economic see David Partlett, "Economic Analysis in the Law of Torts," in Law and Economics 59, 74 (Ross Cranston and Anne Schick eds. 1982); Richard A. Posner, "The Concept of Corrective Justice in Recent Theories of Tort Law," 10 }. Legal Stud. 187 (1981); George J. Stigler, "The Law and Economics of Public Policy: A Plea to the Scholars," 1 /. Legal Stud. 1, 3 - 4 (1972). For a wide-ranging survey of moral theories of tort law by a philosopher, who concludes that none has much explanatory power, see Jules L. Coleman, "Moral Theories of Torts: Their Scope and Limits," 1 Law & Philo. 371 (1982), 2 id. at 5 (1983); and for an interesting but highly preliminary effort to provide such a theory for the law of negligence see Ernest J. Weinrib, "Toward a Moral Theory of Negligence Law," 2 Law b Philo. 37 (1983). Of course there is much normative writing on tort law from the standpoint of moral theory, a notable example being Richard A. Epstein, "A Theory of Strict Liability," 2 J. Legal Stud. 151 (1973).

•

9

·

Catastrophic Personal Injuries

regulation of safety and health is sometimes pointed to as evidence that tort law is doctrinally as well as practically incapable of optimizing the hazards posed by accidents, pollution, and other harms to safety and health. There is merit to this suggestion; tort law does have its limits, and direct regulation has its place.1 Those limits are not, however, as disabling as sometimes believed. We shall argue this point by reference to the catastrophic accident. Any accident that results in death or a serious permanent injury is a catastrophe to the victim, but we shall limit our definition to accidents in which many victims are killed or seriously injured. Illustrations are the crash of a large passenger plane, a nuclear reactor accident that emits radiation over a wide area, the sale of a drug such as DES that turns out to have adverse long-run effects on human health, and the discharge of asbestos fibers into the air in workplaces, again with delayed effects. Tort law may seem incapable of dealing with catastrophic accidents for three reasons: the existence of multiple victims makes it less likely that the injurer can pay all the accident claims than is the case in a singlevictim accident; the causal link to a particular victim is often unclear; and the injury may manifest itself so many years after the accident ADMINISTRATIVE

1. See Steven Shavell, "Liability for Harm versus Regulation of Safety/' 13 }. Legal Stud. 357 (1984); Steven Shavell, " A Model of the Optimal Use of Liability and Safety Regulat i o n / ' 15 Rand }. Econ. 271 (1984); Donald Wittman, "Prior Regulation versus Post Liability: The Choice between Input and Output Monitoring," 6 ]. Legal Stud. 193 (1977). On the specific subject matter of this chapter see Symposium, "Catastrophic Personal Injuries," 13 }. Legal Stud. 415 (1984); Steven Shavell, "Uncertainty over Causation and the Determination of Civil Liability," 28 J. Law & Econ. 587 (1985).

Catastrophic Personal Injuries · 257 occurred that proof of the defendant's negligence may be impossible to adduce. The last two points are related, because the longer the delay between accident and injury, the harder it is to exclude other causal factors. These points may be overstated, however. The concern with solvency overlooks the fact that the relevant cost in considering whether a potential injurer has an incentive to take the right amount of care is not necessarily the cost of the accident if it occurs but the cost of preventing it; if that cost is less than the potential injurer's wealth, he may have adequate incentives to take the right amount of care rather than let the accident happen and exhaust his assets. But, surprisingly, we shall see that this is less likely to be true if the standard of liability is strict liability than if it is negligence. If strict liability is for this reason rejected as the standard in catastrophic accident cases, the third point in the criticism of tort law—the difficulty of proving negligence many years after the accident—comes to the fore. But it is overstated too. Granted, a strict liability case is simpler to prove than a negligence case, and the difference is especially dramatic if a long time has elapsed between the accident and the full-blown injury and hence (it is usually assumed) between the accident and trial. It is one thing to show that the defendant did something twenty years ago that causes harm today; it is much harder to show that he failed to conform to the standard of care applicable to his activity twenty years ago. But tort law has doctrinal resources for overcoming the problem. If the injury on which the victim of a catastrophic accident can base a tort suit is defined not as the full-fledged illness that blooms many years after the accident but as the impact or exposure that planted the seeds of the illness, the trial can be held soon after the accident. Under this approach the victim's damages will be computed in terms of reduced life expectancy, future pain and suffering, and future medical costs, rather than in terms of realized loss. This approach also solves automatically the problem of the accident known to cause injury to many people, when it is uncertain which people they are; all the potential victims can be compensated for their ex ante loss, measured at the time of suit. There is considerable precedent in tort law for such an approach, although the approach is not yet well established. If as we believe the essential logic of tort law is economics, we can expect the law to move toward adoption of the suggested approach as the courts gain greater experience with mass-disaster cases.

258 · The Economic Structure of Tort Law A Model of Catastrophic Accidents We begin by analyzing the case where there is one injurer but multiple victims. To simplify exposition we assume for the time being that the level of victim care can have no effect on the frequency of accidents. If there are Ν potential victims (for example, people living near a nuclear reactor) and damages are the same for each victim if an accident occurs, the loss function is then L = Np(y)D + B(y).

(9.1)

Due care for the injurer is derived by multiplying the reduction in expected damages per victim by Ν and setting the resulting expression equal to the marginal cost of care. This poses no problem and is a trivial extension of the basic model. Either a negligence or a strict liability standard would provide incentives for the injurer to take due care with respect to the entire set of potential victims. But now suppose that the injurer's assets are insufficient to cover the damages of all victims in the event of an accident. Although this is always possible in a single-victim case, it is more likely in a multiple-victim case. If the injurer's wealth has some upper limit, W, as it must, and D is also fixed, then as Ν increases there will come a point where ND exceeds W. This problem is compounded by another: in catastrophic injury cases the damages per victim are also likely to be large even though both the probability of an accident and the expected damages may be small. And the likelihood that the injurer's assets will be insufficient to cover the victims' damages depends on both the damages per victim and the number of victims. This ignores, however, the fact that W is likely to be positively correlated with N, because usually it takes a larger enterprise to injure more victims. The typical injurer of single victims is an automobile driver, and although the damage he can do is limited, it may exceed his ability to pay (or his insurance coverage). The typical injurer of many victims is a business firm, and the more numerous the victims the larger the firm is apt to be. But the relationship between size of firm and number of potential victims is not one to one; one driver can injure one or a few victims, but one firm that makes a dangerous product or emits dangerous pollutants can injure a substantial multiple of the number of its owners. So let us assume that the problem of limited solvency, that is, ND >W — Β (y), is greater in multiple-victim cases; let D equal the maximum amount per victim that the injurer can be made to pay, so that ND = W - B(y); and consider what difference it makes whether the liability stan-

Catastrophic Personal Injuries · 259 dard is negligence or strict liability. Under strict liability the level of care will tend to be less than due care2 because the injurer will equate - NpyO to (1 - p)By instead of - NpyD to By. Assuming a small probability of an accident (a proper assumption in the case of a catastrophic accident), the solvency limitation dominates the discounting of By by (1 p), and the injurer will take care equal to yu which is less than y*. But under a negligence standard the injurer will either use y1 inputs of care or bring his level of care up to the due care standard, y*. Paradoxically, then, a negligence standard may give the injurer a greater incentive to use due care than strict liability would do. The reason is that by increasing his level of care from y1 to y*, the injurer avoids all liability for damages in the event of an accident, whereas under strict liability expected damages fall as care rises but the injurer still pays something; the benefits to him of additional investments in care are therefore smaller. Figure 9.1 illustrates this point. Under strict liability y1 is the level of care that equates the injurer's marginal benefits to his marginal costs of care given the limitation on his ability to pay full damages. From the injurer's standpoint the added cost of bringing his care up to due care (y*) is area c + d, whereas the corresponding benefit to him—the reduction in expected damages—is d. From a social standpoint, however,

i>P,D

/ B ,

a V

/

V® ' d; ; ι— 1 y, V

\ e

\ Care

Figure 9.1

2. U n d e r strict liability the injurer seeks to minimize Np(y)D + B(y), which can be rewritten as p(y)[W - B(y)] + B(y). Minimizing the latter with respect to y yields Np y D + (1 - p)By = 0. Without a solvency limitation the injurer minimizes Np(y)D + S(y), which yields Np y D + By = 0.

260 · The Economic Structure of Tort Law the benefits are greater than the costs, for (a + b + c + d) exceeds (b + c + d). Thus strict liability yields an efficiency loss equal to a. Under a negligence standard the injurer is faced with two choices. 3 He can select y1 and bear expected damages of d + e, or he can take due care, incurring an added cost of c + d but avoiding all liability for expected damages of d + e. Whether the injurer uses due care (1/*) or yt inputs of care therefore depends on whether area e is larger or smaller than c, which depends in turn on the cost of added care, the difference between D and D, and the effectiveness of added care in reducing expected damages. 4 The analysis is similar if the most efficient method of accident avoidance is for the potential injurer to discontinue his activity rather than reduce the probability of the accident by being more careful. Whether it is or not will depend on whether V § Np(y*)D + B(y*),

(9.2)

where V is the value of this activity minus the value of his next best activity (V excludes expected damages and costs of care). If we assume a ceiling on the injurer's ability to pay, strict liability will generate too much activity (substitute D and y1 above) and negligence either the same amount of activity as strict liability (when the injurer uses y j or more (when the injurer uses y* to avoid paying any damages). This result alters the usual comparison between strict liability and negligence, where a strict liability standard always induces a lower activity level by the injurer than a negligence standard does.

Extension to Causal Uncertainty To analyze the basic problems of catastrophic accidents—causal uncertainty 5 and the passage of time between accident and injury—we 3. At y, expected damages are eliminated in the amount of the area under the -Np¥D curve from 0 to j/J. Expected damages of d + e—the area under the -NpyD curve to the right of y,—still remain. Note that D declines as y increases because ND = W - B(y). 4. We have shown that strict liability leads to y, units of care and negligence to either yj or y* units of care. But if the legal standard of care can be varied, it is always possible to set the standard somewhere in between y, and y* so that the injurer under a negligence standard will choose to comply with the standard instead of using y,. Thus, under a second-best standard, negligence will dominate strict liability because care between and y* is preferable to y,. 5. See, besides Chapter 7 of this book and references cited there, David Kaye, "The Limits of the Preponderance of the Evidence Standard: Justifiably Naked Statistical Evi-

Catastrophic Personal Injuries · 261 consider a concrete example. Suppose that an accident at a nuclear reactor power plant would produce an emission of radiation that might injure many persons living nearby. We shall denote by a the probability that each of Ν exposed persons will incur actual damages (D per person) if the accident occurs. There is also a positive probability b ( < a ) that each person will incur the same harm even if there is no accident. For example, a might be the probability that an exposed person will develop cancer given the increase in the level of radiation as a result of the accident and b the probability that he will develop cancer even if there is no increase. Provisionally we assume that a and b are unaffected by either injurer or victim expenditures on care and that p(y), the probability of a nuclear reactor accident, depends only on y, the insurer's inputs of care. Then we can write the loss function (including expected losses if there is no accident) as

or

L = Np(y)aD + N[ 1 - p(y)]bD + B(y)

(9.3)

L = Np(y)(a -

(9.4)

b)D + NbD +

B(y),

and due care as -Npy(a

-

b)D =

By.

(9.5)

From Eq. (9.4) we see that Np(y)(a - b)D is the increase in expected damages due to the nuclear reactor accident and NbD the expected damages in the absence of any accident. Notice that uncertainty about causation exists at the individual but not at the aggregate level. If, for example, Ν = 1,000, p(y*) = .10, a = .11, and b = .10, a nuclear reactor accident will cause ten additional cancers—l,000(.ll — .10)—a fact known with near certainty from the law of large numbers. Yet for each of the 110 persons in the exposed population who develops cancer, there is uncertainty about whether it was the exposure to radiation or something else that caused his cancer (caused it in the sense that, if he had not been exposed, he would not have gotten cancer). Indeed, because 100 persons would have gotten cancer anyway, it is more likely than not—ten times more likely, in fact—that the reactor accident did not cause any given individual's cancer. 6 dence and Multiple Causation," 1982 A.B.F. Res. J. 487; Glen O. Robinson, "Multiple Causation in Tort Law: Reflections on the DES Cases," 68 Va. L. Rev. 713 (1982). 6. Additional assumptions could of course be added to the example that might enable the law to reduce (but not eliminate) individual uncertainty. For example, people closer to the accident or receiving greater doses of radiation may have higher probabilities (higher

262 · The Economic Structure of Tort Law The dilemma is apparent. If none of the victims can recover damages, the defendant will pay nothing, although in fact he harmed ten people (we just do not know which ten). But if all the cancer victims are allowed to recover damages, the defendant will be forced to pay ten times the actual damages that the accident caused. It may seem that if the first horn of the dilemma results (as it does) in underdeterrence, the second must result in overdeterrence. But as should be clear by now, this is not true in the formal model of negligence. If due care is defined with respect to the additional damages, (a -b)D, caused by the injurer (see Eq. 9.5), there is no overdeterrence as long as the standard is negligence: the injurer takes due care, y*, and avoids any potential liability. The fact that if the accident did occur the damages would be greater than the victim's incremental loss is of only academic interest to the potential injurer, because he will take due care and avoid paying any damages. But if negligence is erroneously defined with respect to total damages, so that due care requires a level of y that minimizes Np(y)aD instead of Np(y)(a -b)D, the injurer must take more care than y* to avoid liability. Or if the legal standard is strict liability and the injurer pays D to each cancer victim if an accident occurs, the injurer will minimize Np(y)aD instead of Np(y)(a - b)D, resulting in too much care. As emphasized throughout this book, the formal model is incomplete because it omits the possibility of finding negligence in a case of unavoidable accident. Once that possibility is admitted, the negligence standard is likely to produce overdeterrence (even if due care is defined with respect to incremental expected damages) if the victim's damages are much greater than the increment in those damages that is actually due to the tortfeasor, if potential defendants can reduce the likelihood of an erroneous finding of liability by taking more than due care, and if persons with higher than average cost of care bring their care up to the reasonable-man standard. Overdeterrence may also induce an excessive reduction in the injurer's activity level—for example, substituting coal for nuclear power even if the latter is cheaper when all accident costs are factored in.

fl's) of developing cancer. Similarly, certain classes of individuals (such as nonsmokers or persons with a lower incidence of cancer in their families) may have lower probabilities of developing cancer in the absence of the accident (lower b's). Thus one might be able to show that for some victims the odds that the accident caused cancer were great (a\b was high) whereas for others they were low (alb was low). Our analysis ignores these complications.

Catastrophic Personal Injuries · 263 Alternative Approaches to Causal Uncertainty The tort law has resources for dealing with the problems identified in the preceding section. Before showing this, however, we shall briefly discuss the alternative damage rules that courts could use to deal with causal uncertainty. We continue to use the example of the nuclear reactor accident. 1. All or nothing. In a garden-variety common law tort case each victim, to win, must show that his injury probably would not have occurred but for the accident, which implies that a - b is close to one (that is, a is close to one and b is close to zero). Given the administrative costs of determining probabilities a and b with any exactness, such a rule tends to be efficient in a single-victim tort case, as we saw in Chapter 8. In contrast, in the multiple-victim case the all-or-nothing character of the traditional approach could create a significant disincentive for the injurer to invest in safety. Because a - b would be small for each plaintiff in the nuclear reactor accident example, no plaintiff would prevail. Knowing this, the potential injurer would not have an incentive to take due care. 2. Compensating all exposed persons. If an accident occurs, each exposed person is harmed in a probabilistic sense because he faces a higher probability (a instead of b) of developing cancer in the future. In principle it should be possible to compensate potential victims appropriately for the added risk, thereby creating the right incentives for potential injurers to take care. In effect, each potential victim would be compensated for his actual damages discounted by the probability that he would not have suffered them anyway. Assume from our previous example that Ν = 1,000, a = .11, and b = .10 and also that damages are $1,000 per cancer victim. Substituting these numbers into our equation for due care (Eq. 9.5) yields l,000P y (.01)($l,000) = Br Awarding each exposed potential victim his expected incremental harm of $10 (equal to actual harm times the incremental probability that it will occur) will create an incentive for the injurer both to take due care and to select the correct activity level, because his liability is equal to the expected damages from his activity. Similarly, under a negligence standard the injurer will choose y* because his expected liability plus costs of care will equal B(y*), which is less than his expected damages and costs of care for y < y*. More generally, an incentive for the injurer to behave efficiently is created by awarding damages to each of the Ν exposed persons equal to D* = (a - b)D.

(9.6)

264 · The Economic Structure of Tort Law 3. Compensating victims ex post, including some who would have been injured anyway. This rule limits recovery to exposed persons who subsequently develop cancer. If a = .11 and Ν = 1,000,110 persons will develop cancer sometime after being exposed to the radiation from the reactor accident, and they would under this rule be awarded damages. But each person's damages would be lower than the assumed harm of $1,000 per cancer, because the harm would be discounted to reflect the fact that 100 of the 110 successful plaintiffs would have gotten cancer anyway. One can show that a damage award per ex post injury equal to D** = [(a - b)/a]D,

(9.7)

where (a - b)/a is the discount factor, would create an incentive for the injurer both to take due care and to choose the correct activity level. Because Na persons (the number who actually developed cancer) would be awarded damages of D** per person, the injurer's expected liability and cost of care (assuming a strict liability standard) would equal NapD** + B(y) = Np(a - b)D + B(y). Minimizing this expression with respect to y yields y*, the due care standard 7 (see Eq. 9.5). 4. Compensating victims ex post whose injury was caused by the accident. We mention this rule for completeness although by assumption it cannot be used because there is no way of knowing which 10 (Na - Nb) of the 110 cancer victims would not have developed cancer but for the nuclear reactor accident. Under this rule each of the 10 cancer victims would be fully compensated (at $1,000) and the injurer would have an incentive both to use due care and to select the correct activity level (assuming strict liability). Notice that this rule (when workable) produces the correct amount of deterrence even though the "harm" of exposure is ignored. Whether to deem exposure a harm for tort purposes is a choice variable for the tort system rather than a given. We now consider the relative advantages of the second and third damages rules (the all-or-nothing rule has already been examined and the fourth rule, by assumption, cannot be used). We remind the reader that the problems of causal uncertainty and of long delay between accident and full-blown injury are related. If the day after the accident at Three Mile Island three residents of Harrisburg, Pennsylvania, had suddenly lost all their hair, it would have been reasonably clear, at least in the absence of any other explanation, that they were victims of the accident. But if twenty years after the accident there are three excess 7. The injurer would also select y* under a negligence standard, assuming that due care is defined as the level of y that minimizes NapD** + B(y).

Catastrophic Personal Injuries · 265 cancer deaths in Harrisburg, it may be impossible to determine which three cancer deaths would not have occurred but for the accident, because lapse of time creates an opportunity for all sorts of other causal factors to come into play. Of course, not everything that follows closely in time upon some event is a result of that event, but it is much easier to exclude other factors when the relevant time period is short than when it is long. An interesting exception to this generalization, however, is the case where negligence occurs in the course of a rescue attempt. For example, Grimstad falls overboard from a boat not equipped with a life preserver and drowns, but it is uncertain whether he would have been saved if a life preserver had been available. Or, through medical negligence a man's lung cancer is not detected as early as it should have been but he probably would have died anyway if it had been detected sooner. 8 Because most people are not dying at the time they are struck down by a negligent injury, these cases are of limited practical importance. Most cases of causal uncertainty involve a delayed-reaction injury, because it is only over a long period of time that the cumulative risk from other, independent causes becomes highly significant. If as in the numerical example in the preceding section we assume that the probability will ordinarily be small that the delayed-reaction victim would not have fallen ill but for the defendant's act (less than 10 percent in our example), the result indicated by conventional tort thinking is no liability and leads as we have said to underdeterrence. A notvery-satisfactory alternative is to wait until the 110 victims develop cancer and give each victim D** or 9 percent—(a - b)/a—of his damages. This may be too little to give the victims an adequate incentive to bring suit, in which event there would again be underdeterrence; in addition, it is an approach so much at variance with traditional tort law thinking that it could not be adopted without a profound revolution in that thinking. Another approach, one within the boundaries of traditional tort thinking, is the second rule examined above—compensating all persons whom the defendant's negligence exposes to risk. 9 Although few tort lawyers 8. A s in H e r s k o v i t s v. G r o u p Health C o o p . , 99 W a s h . 2 d 609, 6 6 4 P . 2 d 4 7 4 (1983). C o m p a r e Kallenberg v. Beth Israel Hospital, 4 5 A . D . 2 d 177, 3 5 7 N . Y . S . 2 d 5 0 8 (1974), aff'd, 37 N . Y . 2 d 719, 3 3 7 N . E . 2 d 128 (1975); Stubbs v. City of Rochester, 2 2 6 N . Y . 516, 124 N . E . 137 (1919); a n d c a s e s discussed in C h a p t e r 7 of this book; in Wolfstone a n d Wolfstone, " R e c o v e r y of D a m a g e s for the Loss of a C h a n c e , " 1978 Pers. Injury Ann.

744;

and in D e P a s s v. United States, 721 F . 2 d 203, 2 0 6 (7th Cir. 1983) (dissenting opinion). 9. Discussed in the c o n t e x t of bankrupt m a s s tortfeasors in Mark J. Roe, " B a n k r u p t c y

266 · The Economic Structure of Tort Law would describe this as a "black letter" rule of tort law, it might be described as a latent or incipient such rule. Consider how statutes of limitations are applied in tort cases. The typical tort statute of limitations defines the occurrence that begins the statute's running as the "accrual" of the victim's cause of action. 10 Because there is no tort without some injury,11 a cause of action for negligence or for injury owing to a defective product does not accrue until there is something that can fairly be described as an injury. 12 But it need not be a full-blown injury; under the law of some states the inhalation of asbestos fibers is deemed a sufficient injury to start the statute of limitations running, even though the fullblown disease (asbestosis or some other asbestos-related disease) may not appear until many years later. 13 Even in states that use a "discovery" rule, thereby suspending the statute of limitations until the disease is more advanced, 14 the injured party may possibly be able to sue as soon as the cause of action was brought into existence by (undiagnosable) injury, even if for purposes of applying the statute of limitations the

and Mass Tort," 84 Colum. L. Rev. 846 (1984). The Roe article illustrates the rich legal literature (much of it with an economic flavor) dealing with mass-disaster torts. For other illustrations see the Robinson article cited in note 5 above; Volkmar J. Hartje, "Oil Pollution Caused by Tanker Accidents: Liability versus Regulation," 24 Nat. Res. ]. 41 (1984); Glen O. Robinson, "Probabilistic Causation and Compensation for Tortious Risk," 14 /. Legal Stud. 779 (1985); Roger H. Transgrud, "Joinder Alternatives in Mass Tort Litigation," 70 Cornell L. Rev. 779 (1985); Note, "A Suggested Remedy for Toxic Injury: Class Actions, Epidemiology, and Economic Efficiency," 26 Wm. & Mary L. Rev. 497 (1985); Note, "The Inapplicability of Traditional Tort Analysis to Environmental Risks: The Example of Toxic Waste Pollution Victim Compensation," 35 Stan. L. Rev. 575 (1983); Note, "Mass Tort Claims and the Corporate Tortfeasor: Bankruptcy Reorganization and Legislative Compensation versus the Common-Law Tort System," 61 Tex. L. Rev. 1297 (1983). Compare the procedure established by the Price-Anderson Act for compensating victims of nuclear reactor accidents. See 42 U.S.C. § 2210(o)(3); Duke Power Co. v. Carolina Environmental Study Group, Inc., 438 U.S. 59, 92 (1978). 10. See, for example, the Minnesota statute discussed in Karjala v. Johns-Manville Products Corp., 523 F.2d 155 (8th Cir. 1975). 11. See, for example, W. Page Keeton et al., Prosser and Keeton on the Law of Torts 4 (5th ed. 1984). 12. Many cases hold that the statute of limitations begins to run "upon the occurrence of the injury, regardless of whether the full extent of the disability is known at that time." Fletcher v. Union Pac. R. R., 621 F.2d 902, 906 (8th Cir. 1980). 13. See, for example, Steinhardt v. Johns-Manville Corp., 54 N.Y.2d 1008, 1010-11, 430 N.E.2d 1297, 1299 (1981). 14. See, for example, Neubauer v. Owens-Corning Fiberglass Corp., 686 F.2d 570 (7th Cir. 1982); and on the varieties of accrual concepts used in asbestosis cases alone see Note, "Manville: Good Faith Reorganization or 'Insulated' Bankruptcy," 12 Hofstra L. Rev. 121 (1983). See also Urie v. Thompson, 337 U.S. 163, 169-71 (1949); Nivens v. Signal Oil & Gas Co., 520 F.2d 1019, 1024, modified on other grounds, 523 F.2d 182 (5th Cir. 1975).

Catastrophic Personal Injuries · 267 cause of action did not accrue until much later: to sue, in short for a probabilistic injury. 15 So if the accident victim knows that he has been exposed to or has ingested a dangerous substance, traditional tort doctrine may allow him— and may even, depending on how a particular state interprets its statute of limitations, require him—to bring suit immediately, without waiting until he actually becomes ill. Even if he could wait, the problem of proving causation might create an incentive for him to sue as soon as he was exposed—provided that damages could be calculated then. They could be, although this is not yet widely understood. When a tort victim's life expectancy has been shortened as a result of an accident for which the defendant is liable under tort principles, then, if one assumes that reduction in life expectancy is compensable, the damages to which the victim will be entitled are the sum (discounted to present value) of the damages that the victim's estate would have received in every future year if he had died in the accident, multiplied by the incremental probability of death associated with the accident. Suppose, for example, that if he had been killed in the twentieth year after the accident, his estate would have been entitled to $100,000. If the accident created a 1 percent probability of his not surviving that long, then he would be entitled to $1,000 in damages, discounted to present value. Expected future medical expenses, future pain and suffering, future disfigurement, and so forth would be treated similarly. Although the computation of damages in such a case would be probabilistic, in the sense that it would give many accident victims a windfall when viewed ex post, this is no different from the situation created by computing lost future earnings and other expected rather than realized losses in ordinary tort cases. These computations are based on probabilities (for example, life expectancies) and therefore produce windfalls for anyone who would not in fact have lived to his expected age—which means half the plaintiffs. This approach to the mass-disaster case solves the problem that is created when, after the consequences of an accident have been realized, a group of people present themselves in court asking for full damages 15. See Moore v. Jackson Park Hospital, 95 111. 2d 223, 232, 447 N.E.2d 408, 411-12 (1983); Comment, "Increased Risk of Disease from Hazardous Waste: A Proposal for Judicial Relief," 60 Wash. L. Rev. 635 (1985); Fournier J. Gayle III and James L. Goyer III, "Recovery for Cancerphobia and Increased Risk of Cancer," 15 Cumberland L. Rev. 723, 736-43 (1985); Comment, "Medical Malpractice: The Right to Recover for the Loss of a Chance of Survival," 12 Pepperdine L. Rev. 973 (1985); Joseph H. King, Jr., "Causation, Valuation, and Chance in Personal Injury Torts Involving Preexisting and Future Consequences," 90 Yale L.J. 1353 (1981). But see Morrissy v. Eli Lilly & Co., 76 111. App. 3d 753, 761, 394 N.E.2d 1369, 1376 (1979).

268 ' The Economic Structure of Tort Law

although it is clear that most of them have suffered no harm from the accident (90 percent in our nuclear example). Ex post only a fraction of the accident victims are losers but ex ante they all are, so the tort system should encounter no insuperable conceptual difficulties in awarding damages to all of them. And, as a matter of fact, although damages for loss of life expectancy as distinct from loss of life itself are rarely sought, there is authority for awarding them.16 The approach that we have sketched is not just an achievable reform within the basic framework of tort law but may actually be an existing part of the doctrinal repertoire of that law. The only reason that it is not a larger part may be that the delayed-action injury with quantifiable effects is a newcomer to the tort scene. We emphasize this qualification: that an injury can have serious but long-delayed consequence has long been known; what is new, because it depends on sophisticated statistical analyses of large groups of people over long periods of time, is even minimally reliable estimates of the actual probability of the long-delayed consequences. We predict—and it is one of the more readily although not immediately testable hypotheses of the positive economic theory of tort law—that in the years to come the approach sketched in this chapter will become an important method of establishing tort liability and measuring damages in delayed-consequences cases. There is still to be considered, however, a major practical problem with the approach: if the expected loss is very small on an individual basis although large in the aggregate, no individual may have an incentive to bear the costs of suit. The class action may solve this problem. With a class action it should not be critical how many people are exposed compared with the number who will actually suffer injury as a result of the exposure: there will be only one suit—a class suit—regardless of the number of plaintiffs.17 True, the more class members there are, the higher will be the costs of notice and payment in the class action. But states could choose to relax, to some extent anyway, both the requirement of individual notice to class members (as distinct from the much cheaper if much less effective method of notice by publication in a newspaper) and the requirement of individual payment; from an economic standpoint, the critical point is that the defendant pays, not that the plaintiff receives. Of course the class action is no panacea even if the current judicial 16. See discussion and references in the DePass dissent, note 8 above, 721 F.2d at 208. 17. We ignore the fact that some members of the class may decide to opt out and bring their own suit.

Catastrophic Personal Injuries · 269 opposition 18 to its use in mass-disaster cases can be overcome. Realistically, the lower the probability of injury (a in Eq. 9.4) is, and hence the larger is the ratio of Ν to Na—that is, the ratio of potential plaintiffs under the ex ante approach to potential plaintiffs under the more conventional ex post approach—the less attractive will the ex ante approach be. Moreover, the ex ante approach is not feasible when people have no reason to suspect danger at the time of the accident or, what is more likely, do not have enough information to assess the magnitude of the danger. That information is necessary not only to motivate victims to sue but also to enable the courts to estimate the accident costs and therefore decide whether the defendants were negligent in failing to prevent the accident. And notice that in determining negligence under either the ex ante or the ex post approach, we assume that only p, the probability of an accident, depends on the injurer's level of care. If both ρ and a, the probability of an injury given an accident, depended on the injurer's care, the information required and hence the difficulty of determining negligence would be even greater. But the problem of information is not as one might expect a powerful argument for direct regulation in lieu of tort law. Regulation will not work unless the government spends resources on obtaining information in advance about the consequences of an accident should one occur. If it does obtain the information, it can disseminate it widely so that people have an incentive to sue and the courts have the information they need to determine negligence. There is no obvious reason why the government should use the information to power a regulatory scheme. If the most efficient information gatherers are not government officials but potential injurers, an argument appears for combining strict liability with a discovery statute of limitations, thus allowing accident victims to sue when they discover the consequences of the accident and placing maximum pressure on potential injurers to gather information in advance about the likely consequences of their activities. At the same time it will simplify the trial of lawsuits necessarily brought long after the conduct that gave rise to the suit by removing the issue of negligence from the case.

Care by Victims of Mass-Disaster Torts We have assumed up to now that potential victims can do nothing to avoid or mitigate the severity of catastrophic accidents. This assumption 18. Illustrated by Yandle v. PPG Industries, Inc., 65 F.R.D. 566 (E.D. Tex. 1974). For a good discussion see Transgrud, note 9 above, at 788-96.

270 · The Economic Structure of Tort Law is artificial and will now be relaxed, and we shall consider how the amount of care a victim will take is affected by the various liability approaches examined above. 19 Assume first that the victim's care (x) is equally productive in both the accident and no-accident states. For example, suppose that following a high-fiber diet or refraining from smoking reduces the probability of developing cancer to the same extent whether or not there is a nuclear reactor accident—that is, ax = bx. Due care for a victim20 is derived by first adding the total cost of victim care, NA(x), to Eq. (9.3) and then minimizing with respect to x, yielding - bxD = Ax. Thus x* requires each potential victim to spend on χ until the reduction in expected damages, which is independent of the likelihood of an accident, equals the marginal cost of care. Under the ex ante approach explored in the preceding section (compensating all potential victims), a potential victim has an incentive to take due care (x*) even if there is no defense of contributory negligence. 21 Because the victim is compensated ex ante for his expected incremental damages, 2 2 he minimizes bD + A(x), which yields x*. Now assume that the victim's care is relatively more productive if an accident occurs—that is, that —ax> - bx. For example, suppose that smoking less reduces the probability of getting cancer by a greater amount if there is a nuclear reactor accident. To simplify, let bx = 0, so that taking care reduces the probability of an injury only if an accident occurs. 19. We need not consider explicitly the effect of different damage rules on the victim's activity level. As shown in Chapter 3, a potential victim will make the correct activity choice (for example, the choice between living near a nuclear reactor plant or elsewhere) only when he faces the full expected damages from an accident. Therefore, under strict liability victims will make incorrect activity choices but injurers will make correct ones, whereas under negligence victims will have incentives to make correct activity choices but injurers will not. 20. Because we assume that all victims are identical, (9.3) becomes L = Np(y)aD + N-

[1 - p(y)]bD + NA(x) + B(y), and victim due care requires paxD + (1 - p)bxD + A, = 0.

Assuming ax = bx, victim due care requires bxD + Ax = 0. 21. We say "even if there is no defense of contributory negligence" because in the usual analysis of strict liability (without a defense of contributory negligence) the victim takes less than due care. He is fully compensated for his injury, so taking care imposes a cost on him without an offsetting benefit to him. This in turn reduces his incentive to take care. See Chapter 3. 22. Under the ex ante approach the victim's expected losses minus expected compen-

sation equal p{y)aD + [1 - p(y)]bD + A(x) - p(y)(a - b)D, or simply bD + A(x). The ex

post approach (compensating victims when the full-blown injury occurs but discounting damages by (a — b)la) also yields x*. After the accident (but before compensation) the victim's expected damages plus costs of care equal (a - b)D + bD + A(x). Subtracting D** discounted by the probability of developing cancer (a) also yields bD + A(x), and minimizing this term with respect to χ yields x*.

Catastrophic Personal Injuries · 271 Because we have already analyzed the case where - ax = - bx > 0, assuming that bx = 0 is equivalent to analyzing the effect of additional care that is relatively more productive if an accident occurs. We assume that a > b even though -ax > 0 and bx = 0. Two cases must be distinguished: in the first, care is undertaken prior to a possible accident (prior care); in the second, care is undertaken only after the accident occurs (after care). Heretofore we have not distinguished between prior and after care because the accident and injury were assumed to be simultaneous—although there is in fact a tort doctrine of avoidable consequences designed to induce the taking of after care. 23 For catastrophic accidents however, the actual injury may not show up until long after the accident. In the case of prior care, due care requires that -paxD = Ax, which is obtained by discounting the expected reduction in damages (-axD) by the probability that an accident will occur (p) and setting it equal to the marginal cost of care. Unlike the situation in our earlier case where ax = bx (the benefits from victim care are independent of whether the accident occurs), a defense of contributory negligence is needed to ensure that the victim's care will not fall short of due care even if he is compensated immediately after the accident. Because the victim's expected compensation equals either pD* or paD**, it will exactly offset his expected incremental damages from an accident of p(a - b)D (see Eq. 9.4). After expected compensation is subtracted, his expected losses will be bD = A(x). But by assumption χ has no effect on b, so the victim's care will be zero. And yet the problem of too little care by the victim may not actually be a serious one in the example under consideration. Although a highfiber diet or refraining from smoking may have a greater impact on a than on b (the essential assumption of our analysis), still both are likely to have significant effects on reducing b,2i and when one discounts that slightly greater effect on a than b by the small probability of a nuclear reactor accident, the incremental expected benefit of additional care is likely to be negligible. Thus it may not be important to have a doctrine of contributory negligence (limited to preaccident carelessness) in catastrophic accident cases. Now consider victim care undertaken after an accident occurs (after care) but before the consequences of the accident are felt. For example, 23. See, for example, Keeton et al., note 11 above, at 4 5 8 - 5 9 . 24. We can show this as follows. Assume - a x > - b x > 0. Due care requires (for prior care) p(ax- b,)D + bxD + Ax = 0. Under strict liability, the victim sets bxD + A, = 0. This yields less than due care, but not much less if both ρ and the difference between a, and bx are small.

272 · The Economic Structure of Tort Law suppose that switching to a high-fiber diet or stopping smoking reduces the probability of developing cancer only if one has been exposed to excessive radiation. From Eq. (9.3) we know that due care requires — axD = Ax,25 and we consider the choice between compensating all potential victims immediately after the accident by D* per victim (see Eq. 9.6)— the ex ante approach—and compensating victims who subsequently develop cancer by D** per victim (see Eq. 9.7)—the ex post approach. Under the former approach, each victim, after receiving compensation for possible future harm, still faces an expected harm of (a - b)D + bD. To minimize this harm he will set — axD = Ax, which yields the due care level, x*. In short, having been compensated initially, the victim has the appropriate incentive to take due care, because he receives its full benefits—lower expected harm in the future. In contrast, if compensation is delayed until the exposed person develops cancer, potential victims will not have adequate incentives to take care even if we assume that their compensation is discounted by the probability that they would have developed cancer anyway. After the accident each victim's expected compensation is aD** (we multiply D** by the probability of developing cancer). Deducting this expected future compensation yields (a — b)D + bD -aD** + A(x). Substituting for aD** yields an expected loss of bD + A(x), and hence the victim takes no care. Recall the analysis in Chapter 2 of the choice between periodic and lump-sum damages for a continuing tort such as a nuisance. The argument for the lump sum is that it does not impair the victim's incentive to invest in avoiding future harm from the continuing tort. The potentially offsetting advantage of periodic payment is that it gives the injurer an incentive to develop ways of avoiding future harm. But that advantage is absent in the present situation, because after the accident has occurred the injurer can do nothing to reduce its consequences as they unfold over the future; those consequences can be avoided only by the victims' investing in care, as by changing diet or quitting smoking. To conclude, if the analysis in this chapter is sound, the tort system is more robust in the face of the difficult problems generated by modern technology than is often assumed. Some progress is discernible in adapting tort doctrines to the economics of these problems, the key doctrinal requirements being (1) that tort injury be deemed to occur (or occur first) at the time of exposure to a delayed-action harm, before symptoms of injury or disease are manifested, and (2) that tort damages be recoverable for loss of life expectancy. 25. We have L = NpaD + N( 1 - p)bD + NpA(x) + B(y), which yields axD + Az = 0.

10 Products Liability and Industrial-Accident Law as a willful interference with the free market in goods and services, modern products liability law seems at first glance a powerful contradiction to the efficiency theory of the common law and hence an apt subject with which to draw a book devoted to testing the theory toward its close. 1 A concluding note in this chapter will contrast the evolution of products liability law with the evolution of the law governing industrial accidents—accidents arising from the employment relation, which is also a contractual relation, like that between the seller and buyer of a product. As throughout the book our model assumes risk neutrality and therefore ignores the insurance effects of strict liability (see Chapter 3). Not only can the main body of products liability rules be explained without introducing risk aversion but the assumption of risk neutrality is plausible. Both potential victims and potential injurers can insure, through accident insurance and liability insurance respectively, against the consequences of product-related injuries. And if insurance is actuarially fair, a risk-averse individual will act as if he is risk neutral with respect to expenditures on safety. In addition, the fact that the defendants in products liability cases are invariably corporations, and usually publicly held ones, provides additional "liability insurance" through the doctrine of limited liability of corporate shareholders; the ability of shareholders to diversify their portfolios creates an incentive for corporations to act as if they are risk neutral with respect to expenditures on safety. Insurance is incomplete and costly, but it is not clear what implications this has HEAVILY CRITICIZED

1. Richard A. Epstein, Modern Products Liability Law: A Legal Revolution (1980), illustrates this type of criticism. The extensive economics literature on products liability is referenced in William M. Landes and Richard A. Posner, " A Positive Economic Analysis of Products Liability," 14 /. Legal Stud. 535, 536 n. 3 (1985).

274 · The Economic Structure of Tort Law for policy. Because strict liability gives victims more protection and injurers less, the disutility of risk may be less or greater under strict liability than under negligence. We assume that all relevant markets are (reasonably) competitive. We also assume that consumers do not have perfect information, because information is a costly good. But we do not assume that consumers have psychological traits that cause them to misperceive risks systematically. Our analysis centers on seven ways in which a product might be thought to cause a personal injury: 1. The product might be designed defectively—for example, an automobile designed with a gas tank that explodes on the slightest impact. 2. There might be a failure in the manufacturing process (process failure), so that, for example, mouse parts get into a soft-drink bottle. 3. A component that the manufacturer bought from someone else might fail as a result of defective design, process failure, or failure of a component of the component. 4. The product might be highly, perhaps unavoidably, dangerous in its ordinary, intended use—for example, a high-speed motorcycle, a cosmetic to which many people are allergic, or blood with undetectable hepatitis virus—at least if the user is not warned of the danger. 5. The product might not be dangerous or defective, yet an occasional accident might result from its use as when the cork of a champagne bottle injures the eye of the person drawing the cork. 6. The product might simply wear out, like a tire that explodes after being driven on for 100,000 miles. 7. The product might have been mishandled by the dealer or the ultimate consumer.

Some Economics of Products Liability How might different liability rules affect the level of safety and the accident rate for product injuries? To simplify exposition, our formal analysis assumes that the manufacturer sells directly to the consumer who is the victim of the injury, and thus the analysis excludes wholesalers and other middlemen as well as cases where the victim is a member of the purchaser's family or a bystander (for example, a pedestrian struck by pieces of an exploding tire). Later we shall consider how admitting these complications might affect the analysis.2 2. The joint liability of manufacturer and middleman to an injured consumer, a special

Products Liability and Industrial-Accident The Efficient

Law · 275

Solution

The dollar benefit (w) to the consumer from the marginal unit of the product consumed, after deducting his expected accident losses plus his costs of taking care to prevent accidents, is given by u =

v(q) -

sp(x,y)D -

A{x),

(10.1)

where q is the number of units purchased (output), v(q) the gross marginal benefit (vq < 0), and s the share of the costs of the product injury that is borne by the consumer; the other terms are defined as before. Inputs of care by consumers include confining the use of the product to the intended use (not shaving with an electric saw) and to a reasonable intensity of use (not riding on tires for more than 50,000 miles). For manufacturers, inputs of care include expenditures on safe design and on carefully inspecting the product to screen out units likely to injure consumers. Besides taking more care, manufacturers can reduce the number of product accidents by selling fewer units of the product, because total accidents equal pq. These two ways of reducing accidents (care and output) correspond to the recurrent distinction in this book between taking more care while engaged in a dangerous activity and reducing the amount of the activity to lessen the probability of an accident. A reduction in output that reduces the total number of accidents is equivalent to a change in activity level. To simplify further, let all manufacturing firms face identical and constant marginal cost, c. The costs (z) per unit of output include both c and the firm's expected liability and costs of care, as in ζ = c + (1 -

s)p(x,y)D +

B(y).

(10.2)

The sum of expected damages and costs of care is L(x,y)=

p(x,y)D + A(x) + B(y).

(10.3)

Net benefits, or welfare, are maximized by choosing values for q, x, and y that maximize W, the difference between total benefits and costs, (10.4) which yields v(q) = c + -pxD=

Ax,

L(x,y),

(10.5) (10.6)

case of joint tortfeasor liability, was discussed briefly in Chapter 7. See also notes 24 and 25 and accompanying text below.

276 · The Economic Structure of Tort Law and

- p y D = Br

(10.7)

Let q*, and as before x* and y*, denote the values of the three variables that satisfy Eqs. (10.5) through (10.7). Thus, q* is the optimal activity level—the level at which the marginal benefit of an additional unit of output equals its full marginal costs, including expected damages and the costs of care—and x* and y* are the due care levels. Because x* and y* also minimize L(x,y), choosing the optimal values of q, x, and y can be viewed as a two-step process: first choose the levels of care that minimize L(x,y)·, then choose the output that equates the marginal gross benefit of the product to its full marginal cost. No Liability (Caveat Emptor) How would the marketplace determine the level of safety and number of accidents in the absence of liability? Suppose no liability led firms to take no care at first. Then consumers would choose a level of care, say xu that minimized the expected damages plus the costs of care for each unit of the good purchased (consumers would behave optimally given that manufacturers take no care), and they would keep buying the product until its net value (see Eq. 10.1) equaled its price (c). Rearranging terms yields u( L(x*, y*). This conclusion assumes that either both x* and y* are positive—a joint care situation—or that y* is positive and x* is zero. Whether consumers use more or less care than x* depends on whether χ and y are substitutable or complementary inputs (if the former, consumers will use more, and if the latter less—but in any event they will use the wrong amount). If the parties are fully informed about accident probabilities and safety levels and the costs of negotiating and monitoring contracts are low, the output and care levels in Eq. (10.8) will not be the equilibrium values. Because consumers will be willing to pay more if expected damages are lower, each manufacturer will have an incentive to take care, provided that the higher price that consumers are willing to offer is enough to offset the cost of taking care. The manufacturer will maximize profits (IT) by choosing y to maximize 7Γ = [o( 0), y1 will result in too little output relative to q*, because v(q) = c + L(0,y]) > c + L(x*,y*). But whether this is the final outcome again depends on the importance of transaction costs. If they are unimportant, firms will have an incentive to negotiate with consumers, who will find it in their self-interest to agree to take care until the reduction in price (which equals the firm's reduction in unit costs of -pxD) from additional care just equals the marginal cost of care (Λχ). Because firms also set - pyD equal to By, the equilibrium care levels will equal the due care levels, x* and y*. Finally, consumers will buy the product until its net price, v(q) - A(x*), equals its full marginal cost, c + p(x*,y*)D + B(y*). This will yield the optimal output given by Eq. (10.5). This is another illustration of the Coase thorem. 3. Of course, transaction costs would be more significant if a middleman were in the picture or if the user of the product were different from the buyer. But in our view these are not the major sources of substantial transaction costs in the product-injury setting; we discuss the major sources below.

278 ' The Economic Structure of Tort Law There is, however, an asymmetry between no liability and strict liability. Under no liability, if the seller wants to take more care and induce consumers to take less care, all he has to do is announce that he is making his product safer and raise his price to cover the cost of the additional safety. Consumers will cut back on their own investment in care. No transaction is necessary. But under strict liability, the seller who wants to induce the consumer to take greater care in exchange for a lower price has to have some way of monitoring the consumer's care or otherwise enforcing the deal. One possibility is for the consumer to waive his legal right to compensation if he takes less than the agreed upon level of care. Negligence Under a negligence approach the manufacturer would be liable only if he failed to take due care. Like a rule of no liability or strict liability, negligence will lead eventually to an efficient solution, provided transaction costs are zero. But it will also lead to the efficient outcome even when those costs prevent the parties from negotiating about levels of safety. To see this, suppose that neither party can observe (or discover at a reasonable cost) what the other party is spending on safety, so that it is too costly to write and monitor a contract specifying each party's level of care, but that each party knows the expected injury costs of the product and how different levels of care can affect those costs. We assume a joint care situation. Table 10.1 illustrates the consumer's expected costs (care plus injury) at different safety levels; x0 and y0 represent the levels of consumer (C) and manufacturer (M) care that are less than due care. Notice that if Μ uses y0 units of care and C uses x*, C's expected damages and costs of care will be A{x*), because Μ will be deemed negligent and therefore will be held liable for C's injury. For all other combinations of care in Table 10.1, C will be unable to recover damages. Table 10.1.

Consumer's expected costs

Manufacturer's care

I/o y*

Consumer's care x0

x*

p(x0,y0)D + A(x0) p(x0,y*)D + A(x0)

A(x*) p(x*,y*)D + A(x*)

Products Liability and Industrial-Accident Law · 279 Now consider C's choice of care, and for the moment assume that the price of the product is independent of the amount of care taken in its design and manufacture. If C expects Μ to use y0, then C will minimize the full price of the product, including his costs of care and expected injury, by choosing x*. This follows from Table 10.1 because p(x0,y0)D + A(x0) = L(x0/y0) - B(y0) > L(x*,y*) - B ( y 0 ) > A(x*). Alternatively, if C expects Μ to take y* care, C will also choose χ*, for then p(xg,y*)D + A(x0) = L(x0,y*) -B(y*)>L(x*,y*) - B{y*) = p{x*,y*)D + A{x*). Although Μ cannot observe C's level of care, Μ will choose his own level of care knowing that it is in C's interest always to take due care. This means that M's expected costs will be c + p(x*,y0)D + B(y0) if he uses y0, and c + B(y*) if he uses y*. Because p(x*,y0)D + B(y0) = L(x*,y0) - Λ(χ*) > B(y*), Μ will minimize his expected costs by choosing y*. Our analysis, however, is incomplete. C and Μ are not strangers. They have entered into a contract for the sale of a product, a contract specifying both price and quantity, and that price may depend on which party is liable for C's injury. One might argue, for example, that C would expect to pay a higher price under a negligence standard if he uses x* instead of x 0 while Μ uses y 0 . Because Μ will be liable for failing to use due care, M's costs will be higher and hence his price will be higher too. This suggests that C will be able to infer how much care Μ takes from the price that Μ charges (higher prices imply less care), even though C cannot observe M's care directly. If so, this will create an opportunity for Μ to behave strategically, as by taking due care but charging a price greater than c + B(y*) to make C think that Μ has taken less than due care. Μ stands to earn greater profits now because price exceeds his unit costs, but if C believes that Μ is behaving negligently, he will pay this higher price because he expects to be compensated for any injury caused by a defect in the product. Will C be misled? Not if he realizes that Μ has an incentive to mislead him by charging a higher price. For then he will reason that M's price conveys no information about M's level of care. In terms of Table 10.1 this means that C will choose his level of care on the assumption that the price he pays for the product provides no information on and therefore is independent of M's level of care. Our analysis of Table 10.1 showed that the result will be that both C and Μ take due care. Thus, under a negligence rule both parties will take due care, C will not recover damages from Μ if an injury occurs, and both parties will be at the optimal level of care. 4 4. This analysis assumes, of course, that C could in principle infer safety from price. If this assumption is ruled out as inconsistent with our assumption that transaction costs

280 · The Economic Structure of Tort Law The basic reason why a negligence rule induces optimal care by both parties and strict liability does not is that we have defined due care to require care by the victim as well as by the injurer. 5 Although it may seem obvious, we add that if prohibitive transaction costs imply that the parties (or just the consumer) lack information on the dangerousness of a product and on how different safety measures affect that dangerousness, there is no reason to believe that negligence will produce an efficient outcome.

Why Not a Contractual Solution? Before going further, we take up the fundamental economic puzzle of products liability law: the injurer and the victim have a contractual relationship, so why shouldn't they be left to work out the optimal combination of safety precautions contractually? Why isn't no liability optimal? The answer is that contracts are costly to make and that the costs may well exceed the benefits, relative to regulation by tort law, when the contingencies that would be regulated by contract—death or personal injury from using a product—are extremely remote. 6 It hardly pays, when buying a case of beer, to enter into a contract specifying rights and duties in the event that one of the bottles of beer explodes in your face. It may seem, however, that for an expensive product such as an automobile, where the parties have a written contract anyway, it would cost little to add a clause dealing with even a remote contingency. But the greatest cost would not be the direct cost of drafting; it would be are prohibitive owing to the high cost of information, the conclusion that negligence results in optimal care follows even more directly—provided only that the consumer knows that he cannot get damages if he is careless. 5. If we assume that strict liability contains a defense of contributory negligence, then it too will lead to optimal levels of care by both parties. 6. Unfortunately, we have not been able to find any estimates of rates of accidents caused by defective or dangerous products. The Consumer Product Commission does publish every year detailed statistics of what it calls product accidents, but the statistics do not discriminate between an accident that occurs merely while using a product and an accident attributable to the product in some interesting sense. Furthermore, the only statistics are the number of accidents; the commission makes no effort to compute an accident rate (the number of accidents divided by some measure of use of the product involved in the accidents). If one took the statistics literally, one would conclude that beds are much more dangerous than chain saws. So we must appeal to the reader's intuition that product accidents caused in some sense by the product and not just the user are extremely rare, with some important exceptions that we take up later.

Products Liability and Industrial-Accident Law · 281 the cost of information. The inclusion of such a clause would not serve its intended purpose unless the consumer knew something about the costs of alternative safety measures that the producer might take and about the safety of competing products and brands. But the cost of generating that information, and particularly the cost to the consumer of absorbing it, may well be disproportionate to the benefit of a negotiated (as distinct from imposed-by-law) level of safety. What makes the problem of information serious is that the consumer may not have much intuitive feel for an extremely small risk. This may be why we observe consumer product warranties for defects in workmanship and material that do not cause personal injury, which are common defects, but not for defects that do cause such injury, which are uncommon because so much more costly to the consumer 7 —and why it is irrelevant that if consumers are well informed about the risks of product injuries, formal contracts are not necessary to obtain optimal safety under a regime of no liability. Good information about remote risks of accident cannot be assumed. It is true that a market can operate efficiently even if some consumers are poorly informed; they can take a free ride on the well informed, provided that manufacturers cannot discriminate in price or quality between different consumers on the basis of how well informed the consumers are. But it may well be that only a trivial fraction of consumers have any feel for the very low risks associated with most products—too small a fraction to influence manufacturers' decisions on safety. All this leaves unexplained, however, why in those instances (perhaps few) where seller and buyer may want to cut their own deal the law does not allow them to do so; why, in other words, it does not enforce the seller's disclaimer of liability.8 The answer is suggested by our earlier discussion: given the high costs (relative to benefits) of information about an extremely low-probability event, 9 it may not pay the consumer to study a disclaimer of liability carefully even if it is clear and conspicuous. A manufacturer will reap little consumer ill will from fooling consumers with a disclaimer that they fail to read, because product accidents are so rare anyway; and for the same reason competing manufacturers will 7. Such clauses may not be included simply because they are unenforceable. 8. Section 2 - 7 1 9 ( 3 ) of the Uniform Commercial Code, which in this respect is declaratory of the general approach of the modern law, establishes a presumption against the enforceability of waivers of liability for personal injury. See, for example, Turner v. International Harvester Co., 133 N.J. Super. 277, 336 A.2d 62 (1975). 9. Such as catching anthrax from a shaving brush, as in S. H. Kress & Co. v. Lindsey, 262 Fed. 331 (5th Cir. 1919).

282 · The Economic Structure of Tort Law not find it profitable to try to compete by offering to disclaim disclaimers. High information costs relative to the benefits of the information may defeat voluntary contracting. Consistently with this analysis, the law is much more likely to enforce disclaimers of liability to a business purchaser, as where the buyer of a machine agrees to indemnify the seller if the latter is held liable for personal injury to a user of the machine. 10 Our analysis can be refined by examining the enforcement of disclaimers of liability under a negligence standard. It is a little hard to see why a consumer would voluntarily—that is, with full knowledge of what he was doing—agree to absolve the manufacturer from liability for negligence. That would imply that the consumer preferred less to more income—preferred a $1 savings in price from accepting the disclaimer to, say, a $2 savings in expected damages from a safety improvement that would cost the manufacturer (and hence, in a competitive market, the consumer) less than $2; for otherwise the manufacturer would not be negligent in failing to make the improvement. Thus if we observe a clause disclaiming negligence, the presumption is that the consumer has been deceived; more precisely, that high information costs have prevented a value-maximizing exchange—one where the consumer would voluntarily pay a higher price in exchange for the manufacturer's taking cost-justified care. In principle, the presumption could be overcome by showing that the manufacturer's customers were limited to a group of risk preferrers—who would prefer a dollar reduction in price to avoiding an uncertain expected loss of more than a dollar—or to persons who, being especially efficient at taking care, would prefer to substitute more of their own care for what would otherwise be efficient care by the manufacturer (y* > 0 for customers of ordinary skill but y* = 0 for a highly skilled group of customers). We know of nothing in the law that would prevent a manufacturer from trying to overcome the presumption in this way, but we also know of no case where a manufacturer has tried to do so; we shall offer in a moment a conjecture as to why. What if the disclaimer seeks only to avoid strict liability, not liability for negligence? It may seem that consumers would sometimes be willing to waive the manufacturer's liability in exchange for a reduction in the price of the product, provided that the manufacturer continued to take due care. But there are several difficulties: 1. When negligence is replaced by strict liability, this does not mean that there are no negligent accidents but only that the legal system will 10. See, for example, Berry v. V. Ponte & Sons, 166 N.J. Super. 513, 400 A.2d 114 (1979).

Products Liability and Industrial-Accident Law · 283 no longer bother to determine, if an accident occurs and suit is brought, whether the defendant was negligent. If disclaimers of liability were enforced in cases where the manufacturer had not been negligent, it would be necessary in passing on the validity of the disclaimer to determine in every case whether the manufacturer had been negligent; so some of the administrative cost savings from replacing negligence by strict liability (a more complex by a simpler standard) would be lost. 2. Previous chapters presented evidence that the common law adopts strict liability as the liability rule only in classes of cases where it is a more efficient method of accident prevention than negligence. In products cases this may be because a reduction in the manufacturer's output—which a court applying a negligence standard would probably not consider a method of preventing accidents that is required by due care— is necessary to limit the number of accidents optimally. Then a disclaimer of strict liability will have the same paradoxical character as disclaiming liability for negligence, except in a minority of cases where the accident is unavoidable by any cost-justified measure. The consumer will be paying less but the price saving will be more than offset by higher expected accident costs to him. A well-informed consumer would not agree to such a disclaimer, provided, as before, that he was not a risk preferrer and not much more capable than the average person of using the product safely. 3. Strict liability is a misleading term when applied to products liability. As we shall see, much of what is called strict products liability really is negligence liability; this is true, for example, in most cases where liability is based on a design defect or on a manufacturer's failure to warn of a known hazard. In these cases the analysis of disclaimers is identical to that in negligence cases. But again these considerations should (and in law do) merely create a presumption against enforcing disclaimers of liability; so again we must ask why efforts to rebut the presumption are so rare. An answer is suggested by noting that the danger to the consumer is in the product itself and that if the danger is obvious to the consumer he will be deemed to have assumed the risk of an accident if one occurs (subject to exceptions and qualifications discussed later in this chapter) and the manufacturer will not be liable. Therefore, if the manufacturer wants to disclaim liability, he need only take steps to make sure that the danger in his product is obvious to the consumer. The danger may be obvious from casual inspection but if not it usually can be made obvious by a warning on the label or by descriptive literature included in the package in which the product is sold. In effect, the law enforces an implicit disclaimer

284 · The Economic Structure of Tort Law when the costs of information to the consumer are low, as is implied by calling the danger "obvious." This solution is consistent with the economics of the problem.

Specific Doctrines of Products Liability Law We have seen that the law's decision not to leave the regulation of product safety entirely to the market appears to be an economically rational one, but we have yet to consider the rationality of the specific doctrines of products liability law.11 Our focus will be on the modern doctrines, but historical developments will be examined briefly first. Historical Trends The two most important historical developments in the law of products liability are the abandonment in most states of the doctrine of privity of contract, which barred consumers from suing the manufacturer unless they had bought the product directly from him, and the gradual supplanting of negligence liability by strict liability. Both developments may be related, although perhaps only loosely, to the growing complexity of products. In the nineteenth century most consumer products were "simple" in the sense that the consumer could ascertain their qualities at low cost. These products were what are called in economics "search" or "inspection" goods (a cantaloupe, whose ripeness can be determined by squeezing, is an example of such a good, at least with respect to tactile qualities such as ripeness). "Experience" goods require use rather than merely touching or inspection to reveal their qualites; any sealed packaged product is an experience good.12 "Credence" goods—well illustrated by many modern medicines, by products such as automobiles where durability is important to the consumer, and by a wide range of services including legal education, automobile repairs, computer dating 11. These doctrines are summarized in (among other places) Epstein's book (note 1 above); James E. Beasley, Products Liability and the Unreasonably Dangerous Requirement (1981); and W. Page Keeton et al., Prasser and Keeton on the Law of Torts, ch. 17 (5th ed. 1984). We have already discussed one specific doctrine—the presumption against the enforceability of disclaimers of liability for personal injuries—and alluded to another (assumption of risk). 12. There were few "shelf goods" (prepackaged goods) in 1870. See James E. Nichols, " T h e Grocery Trade/'in 2 One Hundred Years of American Commerce 595, 598 (Chauncey M. Depew ed. 1895).

Products Liability and Industrial-Accident Law · 285 services, and psychoanalysis—may not reveal their true attributes even after substantial use. 13 As one goes up this ladder, the costs of information to the consumer rise. Because most consumer goods in the nineteenth century were search goods with respect to the danger of personal injury, it would probably have been as easy or almost as easy for the consumer (and even easier for the middleman who sold the product to him) to prevent an accident from a defective product as for the manufacturer to do so. If so, there was little reason to make the manufacturer liable to the consumer, especially because a middleman—against whom the consumer could recover damages on a theory of negligence—had to be in the picture for the manufacturer to have a defense of lack of privity. The consumer probably is more helpless today than he once was. The growth in the technical complexity of products (the horse and buggy giving way to the automobile, patent medicines giving way to modern medicines, fresh foods giving way to canned and frozen foods, and so on) has been accompanied by a relative decline in the technical knowledge of consumers as consumers. At a time when most people lived on farms that were largely self-sufficient, most consumers were knowledgeable regarding the (few) products that they bought. This is no longer true. Partially offsetting trends, however, are higher literacy, more education, and the emergence of consumer-information intermediaries, such as department stores and Consumer Reports—but that emergence is itself a response to the growing complexity of products. Two other factors besides greater simplicity may explain the lack of interest of the nineteenth-century legal system in manufacturer liability. First, much production was carried on either on the farm or by individual craftsmen, often working at home. 14 Often the producers were smaller than the middlemen, were unidentifiable (in the case of most farm products), and were judgment-proof. It made more sense to make the consumer look to the retail dealer for responsibility for defective products than to the manufacturer. 15 Second, advances in scientific knowledge have made it easier to ascertain the point in the chain of distribution at 13. For economic analysis of these distinctions see Michael R. Darby and Edi Kami, "Free Competition and the Optimal Amount of F r a u d , " 16 /. Law & Econ. 67 (1973); Phillip Nelson, "Information and Consumer Behavior," 78 J. Pol. Econ. 311 (1970). 14. See 1 Victor S. Clark, History of Manufactures in the United States 438, ch. 17 (1929); 2 id. at 504, ch. 43; 3 id. at 263, ch. 18; Rolk Milton Tryon, Household Manufactures in the United States 1640-1860: A Study in Industrial History, ch. 8 (1917); Harvey A. Wooster, " A Forgotten Factor in American Industrial History," 16 Am. Econ. Rev. 14 (1926). 15. See Carl Crow, The Great American Customer 147 (1943).

286 · The Economic Structure of Tort Law which a product that caused an accident became defective. If a person gets violently ill after eating canned salmon, it is quite obvious that the fault lies with the canner; tainted food and poisoned drugs were the earliest exceptions to the privity limitation.16 But if, in the nineteenth century, a wheel fell off one's coach several years after one bought the coach from a dealer, pinpointing the cause might well have been impossible. The movement from negligence to strict liability can be dated, although only very roughly, from the Henningsen decision in I960, 17 as the earlier movement, from no liability to liability for negligence, can be dated roughly from Judge Cardozo's famous opinion in MacPherson v. Buick Motor Co. in 1916.18 We shall see that the concept of strict liability that is applied in products cases has a significant component of negligence thinking; nevertheless the term is not a complete misnomer in such cases. The relevant differences between strict liability and negligence are two (see also Chapters 3 and 4): 1. Recall that strict liability gives potential injurers, but not potential victims, an incentive to reduce the number of accidents by reducing the level of their activity, whereas negligence creates the opposite incentive—for the potential victim but not the potential injurer to reduce the level of his activity. In the context of products liability, however, there may seem to be no difference between the manufacturer's reducing his activity and the consumer's reducing his own activity. The manufacturer reduces his activity by producing less, the consumer by buying less (although also perhaps by using the product less, or less intensely); and since what is produced is bought, how can one speak of preferring one or the other party to reduce his activity level? The answer has to do with costs of information. If information about 16. See, for example, Thomas v. Winchester, 6 N.Y. 397 (1852). 17. Henningsen v. Bloomfield Motors, Inc., 32 N.J. 358, 161 A.2d 69 (1960). Technically, Henningsen was not a tort case at all but a breach of warranty case. However, as interpreted in Henningsen, the manufacturer's implied warranty of fitness did not require privity of contract with the consumer and could not be disclaimed, so it was essentially the same thing as strict tort liability; before Henningsen a products liability suit based on a theory of implied warranty was likely to founder on lack of privity, on a disclaimer of liability, or on both. Later cases jettisoned the implied-warranty theory and replaced it with outright strict tort liability. The differences are not great; the major difference is that in an impliedwarranty case the statute of limitations runs from the date of sale rather than the date of injury. See Phipps v. General Motors Corp., 278 Md. 337, 3 4 9 - 5 0 , 363 A.2d 955, 9 6 1 - 6 2 (1976). 18. 217 N.Y. 382, 111 N.E. 1050 (1916).

Products Liability and Industrial-Accident Law · 287 the likelihood of product injuries is costly to obtain, the law, if it wants to achieve efficiency, will place liability (other things being equal) on the party who has the information or can obtain it at lower cost. Suppose that party is the manufacturer. Under strict liability his costs and hence price will increase to offset his expected liability, and consumers will be led to buy less. In this way the manufacturer signals to consumers the dangerousness of the product, and the optimal output results (see Eq. 10.5). 19 Now consider what happens under a negligence regime when the consumer is poorly informed. Possibly he will not reduce his purchases at all, because he assumes that the product is safe. Alternatively his lack of information may cause him to exaggerate the product's dangerousness and buy too little. In either case output will diverge from the optimal level. True, if the consumer is just as likely to overstate as to understate the product's dangerousness, on average he will buy the optimal amount. Yet in most cases he will buy too little or too much. A rule of strict liability, in contrast, leads to the optimal amount in every case. So if the technology of accident prevention shifts in favor of preventing accidents by reducing the activity of relatively well-informed injurers, and victims are relatively ill informed, we would predict a shift toward strict liability. If the asymmetry of information were reversed, negligence would be the preferred rule. Well-informed consumers would reduce their purchases optimally to take account of their expected injuries, because they would know that they would not be compensated if an accident occurred. 2. A claim of strict liability is cheaper to process than a claim of negligence, because the issue of the injurer's negligence is eliminated. But there may be more strict liability claims than there would be negligence claims, because every accident—not just the accidents that could be avoided by the injurer's taking the right amount of care—will give rise to prima facie liability. This is true even if there is a defense of contributory negligence to strict liability; the same defense is available, of course, in negligence cases. This potential disadvantage of strict liability will be small, however, in any class of cases where the technology of accident prevention enables most product accidents to be prevented by the manufacturer's either taking more care or reducing his activity level. Another development in the history of products liability is mysterious: the late extension of liability to injuries to bystanders as distinct from consumers—that is, persons not in the chain of title from the producer. 19. This example assumes that the principal way to reduce the number of accidents is to reduce output.

288 · The Economic Structure of Tort Law The problem is not the extension as such but how long it has taken to be made. The economic argument for strict products liability to bystanders is stronger than the argument for strict liability to people in the chain of title.20 With the bystander there is no possibility of a contractual solution, and he is probably helpless to avoid an accident caused by a defective product. An Empirical Test We attempt to test our economic explanation for the gradual abandonment of the privity doctrine by estimating a regression equation that relates the year that privity was abandoned in a state (the YEAR variable) to urbanization, per capita income, the stock of automobiles, and several other economic and demographic variables that may be proxies for product complexity. Table 10.2 lists the year each state rejected the privity doctrine. Ten states rejected privity by statute (indicated by an s next to the year). The others rejected it first by judicial decision. We estimated the following regression equation (f-statistics in parentheses): YEAR = 1991 - .74 URB + 18.6S + .004 INC - .39 AUTO (77) (3.5) (4.4) (0.3) (1.0) - .52 AGR + .01 MFG + .02 DN - .84 ILT + 47 Β (1.6) (.03) (1.0) (1.5) (2.2) R2 = .55 The variables are as follows: URB = percent urban population: S = dummy variable that takes the value 1 if privity is rejected by statute and 0 otherwise: INC = per capita income; AUTO = population per registered automobile; AGR = percent agricultural workers; MFG = percent manufacturing workers; DN = population density; ILT = percent population illiterate; and Β = percent black population. All variables (except for YEAR and S) are for the year 1920 or as close to 1920 as the data permit. Thus, the regression predicts the date that privity was abandoned from values of the independent variables in 1920. As predicted, the more urban a state's population (as of 1920) and hence the less likely its consumers were to have first-hand familiarity with the products they bought and the smaller the fraction of products they bought that were produced by farmers and individual craftsmen, 20. As pointed out in Elmore v. American Motors Corp., 70 Cal. 2d 578, 586, 451 P.2d 84, 89 (1969). See also Epstein, note 1 above, at 59.

Products Liability and Industrial-Accident Law · 289 Table 10.2. Rejection of privity requirement beyond previously recognized exceptions State Alabama Arizona Arkansas California Colorado Connecticut Delaware Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Montana

Date

State

Date

1939 1967(s) 1949 1934 1963(s) 1944 1950 1956 1933 1967(s) 1934 1938 1960 1953 1929 1929 1963(s) 1951 1946 1939 1931 1966 1927 1947

Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming

1954 1962 1936 1925 1940 1916 1933 1965(s) 1927 1934 1961(s) 1944 1921 1961 1966(s) 1947 1934 1953 1965 1964(s) 1938 1974(s) 1928 1961 (s)

the earlier it abandoned privity. States that rejected privity by statute did so on average 18.6 years later than states that rejected privity by judicial decision, implying that legislatures intervene in common law fields such as torts only after the judiciary has failed to alter existing doctrine. Of the remaining variables, only the proportion black is significant. Because the value of this variable (in 1920) is much higher in southern states, its positive regression coefficient shows that southern states abandoned privity later than other states. We are reluctant to attribute this to differences in product complexity associated with a lower level of economic development in the South, because the income and manufacturing variables are both statistically insignificant. Moreover,

290 · The Economic Structure of Tort Law

the auto and agricultural variables, which may also be associated with product complexity, have the wrong signs, although neither is significant. The population-density and proportion-illiterate variables are also not significant. But the negative regression coefficient on the proportion illiterate may suggest that where consumers have less information on a product's attributes and are less able to take care (because they cannot read), states are more likely to abandon the privity requirement—as our analysis would predict. Table 10.3 provides further evidence of the powerful effect of urbanization on the abandonment of privity. This table computes values for urbanization at ten-year intervals from 1930 to 1970 for states that eventually abandoned privity. It shows, for example, that the percentage of people living in cities averaged 62.5 percent in 1930 in states that had abandoned privity by 1930 compared with 42.6 percent in states that had not (but were to do so eventually by judicial decision) and 33.3 percent in states that were later to reject privity by statute. Between 1930 and 1940 thirteen more states abandoned privity by judicial decision. Once again, these were more urbanized states than those that retained privity (48.6 compared with 42.1 and 35.6 percent). States that rejected

Table 10.3. Privity and urbanization 1930 Nonstatutory No. states that rejected privity Percent urban No. states that retained privity" Percent urban

8

1940 13

1950

1960

1970

1974

7

6

4

—

64.9

51.2

—

4

—

—

51.5

—

—

—

9

1

—

60.4

38.9

1

—

38.9

—

62.5

48.6

59.3

30

17

10

42.6

42.1

48.1

Statutory No. states that rejected privity Percent urban No. states that retained privity" Percent urban

10

10

10

10

33.3

35.6

45.8

53.4

a. As of date shown; later rejected it.

Products Liability and Industrial-Accident Law · 291 privity between 1940 and 1950 were also more urbanized than states that retained it (59.3 compared with 48.1 and 45.8 percent), and the relationship between urbanization and the rejection of privity continues to hold for the remaining decades. Table 10.3 also indicates that privity is rarely rejected unless a majority of the population lives in cities; only once (between 1930 and 1940) was privity rejected when less than 50 percent of the population was urbanized, and urbanization was always less than 50 percent in states that retained privity during the period 1930 to 1950. Modern

Doctrines

We now consider how the modern regime of products liability law deals with the seven types of accident mentioned at the outset of this chapter and whether it deals with them in what is at least prima facie an efficient fashion. 1. Defective design. Little need be said about defective design because the courts follow an explicit Hand formula approach. 21 This may surprise the reader, because defective-design cases are governed by strict liability. But liability is strict in products cases only if a defect is shown, or, as we shall see, if the product, although not defective in the sense of improvable at a cost lower than the reduction in accident costs that the 21. See Epstein, note 1 above, at 83; Phillips v. Kimwood Machine Co., 269 Ore. 485, 495-96, 525 P.2d 1033, 1038 (1974); Knitz v. Minster Machine Co., 69 Ohio St.2d 460, 46566, 432 N.E.2d 814, 818 (1982); Brown v. Link Belt Division, 666 F.2d 110, 115 (5th Cir. 1982); Birchfield v. International Harvester Co., 726 F.2d 1131, 1136 (6th Cir. 1984). We have been criticized for ignoring other tests used by courts (sometimes by the same courts that use the Hand formula approach). See Malcolm E. Wheeler, "Comment on Landes and Posner," 14 ]. Legal Stud. 575, 575-77 (1985). The principal other test, however, the consumer expectation test—whether the product "was dangerous to an extent beyond what would be contemplated by the ordinary and prudent buyer," Kaufman v. Meditec, Inc., 353 N.W.2d 297, 300 (N. Dak. 1984)—is consistent with a market model of products liability. The author is on firmer ground in criticizing courts that determine the level of a product's risk according to scientific knowledge at the time of trial rather than when the product was made, a problem with durable products and products that cause long-delayed reactions of the sort discussed in Chapter 9. But he neglects to consider, as a possible justification for such an approach, the difficulty of determining the level of knowledge in the distant past. The most serious flaw in the article is its reliance on extreme cases to illustrate supposed central tendencies of modern products liability law. A notable example is the repeated citation of Beshada v. Johns-Manville Products Corp., 90 N.J. 191, 447 A.2d 539 (1982), without informing the reader that New Jersey is not a typical state as far as tort law is concerned or that Beshada was narrowly limited in a subsequent opinion of the New Jersey Supreme Court, Feldman v. Lederle Laboratories, 97 N.J. 429, 479 A.2d 374 (1984); see Robert L. Rabin, "Indeterminate Risk and Tort Reform: Comment on Calabresi and Klevorick," 14 }. Legal Stud. 633, 634-36 (1985).

292 · The Economic Structure of Tort Law improvement would bring about, is "unreasonably dangerous." And "defect" is determined in cost-benefit terms: a product is defective if it could have been made safer at a lower cost than the benefit in reducing expected accident costs. 22 Because, as we know from the preceding discussion (and earlier, from Chapters 4 and 7), negligence tends to be the preferred liability rule in joint care situations, it may seem surprising that negligence would in effect be the applicable rule for design defects; the consumer would appear to be quite helpless to prevent accidents arising from such defects. But this is not true. Design defects typically involve durable goods, where the likelihood of an accident depends both on the product's design and the method and intensity of use by the consumer. For example, a person who drives too fast and is injured when his car runs off the road could argue that his injury would have been avoided if the car had been designed differently (say, more like a tank). No doubt he would be right; it is always possible to change a product's design to make it safer. Yet it is more efficient for the driver to drive more slowly than for the manufacturer to turn the car into a tank. If the manufacturer were liable for any and every injury that could have been avoided by a different design, consumers would have less incentive either to take care or to alter their activity level, because they would know that they would always be compensated for an injury.23 That is, if we are right that many product accidents that could be avoided by a different design arise in joint care situations, negligence is the more efficient rule than strict liability for design defects. It may be more accurate to describe the liability standard for design defects as negligence without a defense of contributory negligence, because once a design defect is found to have caused the accident, the manufacturer is liable without regard to the level of victim (consumer) care. But as we saw in Chapter 4, if the liability rule is negligence, a defense of contributory negligence is usually unnecessary to create an incentive for the victim to take care, because negligence is failure to take the care that would be optimal if the victim were also using due care— not the care (for example, making a car as strong as a tank) that would be optimal if the victim were negligent. Contributory negligence is rel22. For illustrative cases see Rourke v. Garza, 530 S.W.2d 794, 799 (Tex. 1975); Rivera v. Rockford Machine & Tool Co., 1 111. App. 3d 641, 647-48, 274 N.E.2d 828, 832-33 (1971); McCormack v. Hankscraft Co., 278 Minn. 322, 335, 154 N.W.2d. 488, 498 (1967). 23. Of course, compensation is unlikely to be 100 percent in cases of death or serious injury. But the greater it is, the smaller is the incentive of potential victims to take costly measures for preventing accidents.

Products Liability and Industrial-Accident Law · 293 evant mainly when both parties are negligent, in which case it serves to economize on administrative costs. 2. Process failure. The concept of strict liability has more bite here than in the defective-design setting. Suppose that the cost-justified level of hygiene in a soft-drink plant allows mouse parts to get into an average of one in every million bottles. By assumption there is no negligence; yet the bottle that contains the mouse parts is defective as a matter of law, and if the consumer is made ill, the manufacturer will be liable. This result makes economic sense, however, because of the asymmetrical position of the manufacturer and the consumer with regard to avoiding this type of accident. The consumer can do nothing at reasonable cost, in the short or long run, to prevent the accident. It is out of the question that he should minutely inspect each soda bottle on the one-in-a-million chance that it contains mouse parts. (If he does notice the mouse parts and drinks the soda anyway, he will be barred by the doctrine of assumption of risk from obtaining damages.) But although by assumption there is nothing cost-justified the manufacturer can do at present to reduce the probability of the accident (otherwise he would be negligent), advances in technology may some day enable him to reduce it at a lower cost than the benefit in reducing expected accident costs. Under a negligence standard the manufacturer will simply wait until the technology is developed; strict liability will give him an incentive to foster its development. True, it will simultaneously reduce the incentive of the consumer (or of businesses that might sell to the consumer) to develop better inspection methods, but it is a reasonable guess that little would be gained from preserving such an incentive. In addition, strict liability will result in a small price increase (to cover the cost of the occasional tort judgment), which will in turn result in (1) a small reduction in output and (2) consumers' substituting other, safer products. It is true that to a perfectly informed consumer the price increase resulting from strict liability would not be a net cost; it would just offset the benefit he would be receiving from shifting to the manufacturer the cost of injuries caused by the product. But now recurs the point that it does not pay consumers to obtain information about small product risks. Suppose that in making his purchasing decisions the consumer will treat minute risks as zero risks (that is, they will not affect his decisions), not because of any psychological peculiarity but simply because it does not pay to compute and try to avoid extremely remote contingencies (a point made in a different context in Chapter 8). Then in a world of no liability, output will exceed the optimal output (see Eq. 10.5) and the manufacturer will fail to take due care. Shifting to strict liability will result in a

294 · The Economic Structure of Tort Law price increase to cover the manufacturer's expected products liability costs, and the higher price will be perceived by the consumer as an increase in the full cost of the product (because he does not realize that the price increase is compensating him for an expected accident cost), and there will be some substitution away from the product. Thus, as noted earlier, the information about risk is impounded in the higher price and "communicated" to the consumer in a form that he can understand without investing in costly information. Finally, because the manufacturer aggregates the expected accident costs of all his present and future customers, an imperceptible cost to the individual consumer becomes a noticeable cost to the manufacturer, inducing him to look for ways of making long-run improvements in safety. The process just described illustrates our earlier point that strict liability is a sensible rule when society wants to influence the level of the potential injurer's activity. Assuming that the costs of information are much higher to the buyer than to the seller, a negligence rule would not give the buyer the correct incentive to substitute away from the product. A rule of strict liability, by inducing the seller to sell less, automatically reduces the buyer's purchases. The only way to induce a change in the mutual activity of seller and buyer, given prohibitive information costs to the buyer, is to give the seller an incentive to change the activity level; and strict liability does that. 3. Component defect or failure. The analysis of component failure is similar to that of process failure. The consumer usually can do nothing to detect a defect in a component of the product he buys. The manufacturer of the final product, although by definition not negligent in failing to detect the defect in the component (if he were, there would be no need to invoke a distinctive body of strict products liability law), is less helpless than the consumer. 24 The manufacturer can more easily identify the component that failed than the consumer can, and he can take appropriate action, legal or business, against the supplier of the component. He can also raise his price to cover potential liability costs and so induce substitution by the consumer away from the products that have this hazard. And he has an incentive to explore long-run improvements in inspecting for component defects. 25 24. Some such cases involve an intermediate buyer between the final manufacturer and the ultimate consumer. An example is Goldberg v. Kollsman Instrument Corp., 12 N.Y.2d 432, 191 N.E.2d 81 (1963), where the court held an airplane manufacturer liable for an accident caused by a defective altimeter bought from another manufacturer, but not the airline that had bought (and that operated) the plane that crashed. 25. We analyzed another aspect of liability for component defects—indemnity of the final manufacturer by the component manufacturer—in Chapter 7.

Products Liability and Industrial-Accident Law · 295 4. Unavoidably dangerous products. The next type of accident arises from the inherent (unavoidable) dangerousness of the product. Cigarettes are hazardous to health, but not because they are defectively designed or produced; in the present state of technical knowledge the danger is largely, maybe entirely, inseparable from the qualities that make the product desirable to consumers. Even so, the danger may be great enough to cast prima facie liability on the manufacturer; what defenses should be available will be discussed later. Properly designed, produced, and tested vaccines and other medicines may cause serious allergic reactions in a small fraction of the population whose members often cannot be identified. Asbestos fibers, if inhaled, may cause serious lung diseases that are unavoidable by making asbestos differently. The difference between this type of accident and those previously discussed is that often (not always, however) the hazard, because it is inherent in the product— present in every batch—is apparent to the consumer and thus can be avoided simply by not buying the product. Notice that the very fact that both the probability of and costs of harm are high may reduce the costs of information to the consumer, assuming that the probability and magnitude of harm are widely known. When the consumer's information costs are low, the optimal rule may (as suggested earlier) be no liability. It economizes on administrative costs, and also sorts consumers into groups on the basis of differences in their expected damages (or costs of care) and thereby improves the allocation of resources. The second point is illustrated in Table 10.4. Consider two equal-sized groups of consumers. Group A has low expected damages, whereas Β has high expected damages (possibly because of high earnings or high costs of taking care). A manufacturer unable to discover at reasonable cost which group a consumer belonged to would, under strict liability, charge each consumer a price of $9—$1 to cover manufacturing costs, including a competitive rate of return on capital, plus $8 to cover expected damages—V2($ll) + '/2($5). Consumers in both groups would continue to buy the product, because each consumer was receiving a positive net benefit. This result would be inefficient. The social cost of selling to persons in Β is $12, not $9.

Table 10.4. Comparison of two consumer groups Consumer group A Β

Value ($)

Expected damages ($)

Costs ($)

10 10

5 11

1 1

296 · The Economic Structure of Tort Law Under no liability, price would drop to $1. Group A consumers would continue to purchase the product (indeed they might buy more of it, because they would be receiving a greater net benefit than before), whereas group Β ones would not. This is the result we want, because Β consumers receive a net benefit (after deducting expected damages) of minus $1. The argument for no liability is weakened if the hazard cannot be avoided as easily as we have just been assuming, because it is not well known to consumers. For example, the hazards of DES to the fetus were not understood when DES was being dispensed as an antiabortifacient. Although those hazards may not have been known, or even knowable at reasonable cost, to the manufacturers of DES either, that would just mean that the manufacturers were not negligent. There would still be an argument for strict liability, familiar from our discussion of process and component failures: the manufacturer may be better able than the consumer to take direct or indirect measures to reduce the hazard. Merely by charging a higher price to compensate for expected liability costs the manufacturer will induce substitution toward products expected to be safer, and that is an important method of accident control. This assumes, however, that the manufacturer has or should have some expectation of danger, although not necessarily of the specific danger. If the costs of information are as high to the manufacturer as to the consumer, strict liability will not lead to greater safety in the short or long run; it will just impose administrative costs, including the costs of determining causation, because some victims might have developed similar conditions even if their mothers had not taken DES. There is judicial authority for excusing the manufacturer from liability in such a case. 26 For the case where, although the product may be dangerous enough to cast prima facie liability on the manufacturer the consumer's knowledge of the risk makes it optimal under our previous analysis to excuse the manufacturer from liability, products liability law needs a concept of victim fault—and has one. For despite the law's reluctance to enforce waivers of liability, the manufacturer is allowed to escape liability for accidents caused by a highly dangerous product by clearly warning consumers of the danger—provided the warning enables the consumer to protect himself from those hazards at reasonable cost. 27 For example, if the manufacturer of a vaccine warns users that they may have a 26. See, for example, Basko v. Sterling Drug, Inc., 416 F.2d 417, 426 (2d Cir. 1969); O'Hare v. Merck & Co., 381 F.2d 286, 291 (8th Cir. 1967). 27. See, for example, Hollenbeck v. Ramset Fasteners, Inc., 267 N.C. 401, 148 S.E.2d 287 (1966); Boyd v. Thompson-Hayward Chem. Co., 450 S.W.2d 937 (Tex. Civ. App. 1970); Formella v. Ciba-Geigy Corp., 100 Mich. App. 649, 300 N.W.2d 356 (1980).

Products Liability and Industrial-Accident Law · 297 reaction to it if they are allergic to eggs or strawberries, a user who ignores the warning will not be able to recover damages. 28 Or if a manufacturer of motorcycles warns users that there is a danger of "wobble" at high speeds, he is off the liability hook; the warning enables consumers to protect themselves from the danger by not driving at those speeds. The attractive feature of the duty-to-warn idea from an economic standpoint is that it maximizes consumer choice. Instead of changing the product for everyone in a way that may displease many, the manufacturer in effect channels particular consumers toward substitute products that are more suitable for their particular needs, as in Table 10.4. It is not inconsistent to allow warnings to defeat liability but refuse to enforce disclaimers of liability. The warnings come into play only when the product is highly dangerous, either to all users or to a defined class of users. The benefits of information to consumers are very great in such a case, and consumers can therefore be expected to make some investment in absorbing the information conveyed by the manufacturer. The purchaser of a high-speed motorcycle will read any warnings in the owner's manual carefully; a person who knows that he is allergic to peanut oil will scrutinize carefully the list of ingredients on the labels of canned goods. Most product hazards, however, are very small; as suggested earlier, it does not pay the consumer to inspect a box of cigars carefully to see whether it contains a disclaimer of liability should he be injured by a cigar exploding in his face. 5. Unavoidably slightly dangerous products. Our fifth type of product accident differs from the fourth only in that the inherent danger is small. A typewriter is heavy, and if it falls on someone's toes it can cause a serious injury, but the probability of its doing so is very small. To impose strict liability when it is unlikely to affect the allocation of resources creates administrative costs with no expectation of offsetting benefits, and that would be the result of making manufacturers strictly liable for every trivial hazard. They would not seek ways of making typewriters lighter in order to avert liability for crushed toes. And the trivially higher price that the manufacturers would charge to cover the expected liability costs would not induce consumers to substitute toward safer products; there are no safer substitutes. The situation is different if one can identify a class of highly dangerous products, the only class of unavoidably dangerous products that subject their manufacturers to strict liability for accidents resulting from the 28. See, for example, Arthur W. Murphy, Kenneth V. Santagata, and Frank P. Grad, The Law of Products Liability: Problems and Policies, pt. 1 (1982).

298 ' The Economic Structure of Tort Law inherent danger. 29 Here the potential gains from strict liability are substantial, provided consumers are not well informed about the likely dangers of the product. Strict liability may lead the manufacturer not only to search for ways of making his product safer but also to raise the price significantly in order to cover expected liability costs that are significant. This in turn will cause substitution toward safer products—and there are safer products. (Such substitution would take place under negligence only if consumers were as well informed as manufacturers.) The analysis is similar to that of strict liability for ultrahazardous activities (see Chapter 4), a category that if our analysis is correct includes the unavoidably highly dangerous product as a special case. 6. Worn-out product. Our sixth type of accident, one caused by a product that has become dangerous simply by long use, provides a bridge to the last type—that caused by misuse or other fault not attributable to the manufacturer. A tire that has been driven on for 200,000 miles is extremely dangerous; the danger is avoidable at lower cost by the consumer's discarding the tire than by the manufacturer's trying to make a tire that never wears out; and the danger of worn-out tires is so well known that the manufacturer is not required to warn consumers in order to galvanize the least-cost method of accident avoidance. 30 The manufacturer is thus induced to take care to reduce the probability of accidents caused by normal use; the consumer is induced to take care by timely replacement of worn-out tires; and liability properly is excused if the consumer fails to exercise due care. 7. Victim or third-party fault. In our final accident type the product is not dangerous, and so it is straightforward to describe the accident as resulting entirely from the misuse of the product by someone other than the manufacturer—sometimes the ultimate consumer (and victim), sometimes a dealer or other intermediary between manufacturer and consumer. If a person eats a quart of ice cream and gets a stomach ache, he has only himself to blame. The more interesting case is where the manufacturer is prima facie liable but asserts victim or third-party conduct as a defense. We shall distinguish three types of such conduct: (a) The concept of assumption of risk is operative when a product is defective or unreasonably dangerous but the victim knows of the hazard, 29. See, for example, Henkel v. R and S Bottling Co., 323 N . W . 2 d 185 (la. 1982); Metal Window Products Co. v. Magnusen, 485 S.W.2d 355 (Tex. Civ. App. 1972); American Law Institute, Restatement of Torts (Second) § 402A, comments j, k (1965). We are speaking of prima facie strict liability; the consumer's knowledge of the risk may be a defense. 30. See, for example, Mullen v. General Motors Corp., 32 111. App. 3d 122, 129, 336 Ν.Ε.2d 338, 344 (1975.)

Products Liability and Industrial-Accident Law · 299 encounters it voluntarily, and is injured; he is not allowed to recover damages. This is the most important ground on which cigarette smokers who get lung cancer or other cigarette-related diseases have lost their products liability suits against the cigarette manufacturers. These plaintiffs' lack of success is a significant bit of evidence for the economic rationality of products liability law. (b) "Mishandling" refers to the situation where either an intermediate seller or the ultimate consumer has caused or contributed to the accident either by handling it carelessly or by putting it to a use for which it was not intended. 31 In either case the manufacturer will not be liable. (c) Contributory negligence in a product case might be unrelated to mishandling. Suppose that although the automobile was defectively designed the accident would not have occurred if the driver had not been going 100 miles an hour. And the speed did not aggravate the defect. Maybe the brakes were defective and would have failed at 20 miles an hour—but at that speed the driver would have been able to maneuver around the oncoming car in time. As should be apparent from Chapter 8, the standard analysis again indicates that there should not be liability, and generally there is not. Notice, by comparison with Chapter 4, that there is a broader range of defenses of victim conduct in products liability cases than in other strict liability cases. 32 The economic explanation is that most products liability cases do not involve ultrahazardous activity (some do, as we saw), that is, a high Ρ in the Hand formula. The fact that an activity is ultrahazardous implies that accident avoidance brought about by reducing the amount of the activity (substitution toward other activities) is likely to be more productive than victim self-protection. The presumption is weaker in the products liability area, and the law seems to have responded accordingly. Products liability law not only incorporates a fairly broad range of defenses based on victim misconduct but contains at least one interesting refinement that seems economically justified, although it is highly controversial and has no counterpart that we know of in general tort law. This relates to the obvious-danger defense, a variant or component of assumption of risk. If a manufacturer sells a machine with exposed moving parts that are dangerous to anyone who touches them and a 31. For good examples of these different situations see McDevitt v. Standard Oil Co., 391 F.2d 364 (5th Cir. 1968); Ford Motor Co. v. Matthews, 291 So.2d 169 (Miss. 1974). 32. See West v. Caterpillar Tractor Co., 336 So.2d 80 (Fla. 1976); Cyr v. B. Offen & Co., 501 F.2d 1145 (1st Cir. 1974); Dix W. Noel, "Defective Products: Abnormal Use, Contributory Negligence, and Assumption of Risk," 25 Vand. L. Rev. 93, 96 (1972).

300

· The Economic

Structure

of Tort

Law

worker is hurt doing just that, the manufacturer will have a prima facie defense that the worker assumed the risk when he put his hand in the machine. But in a growing number of states the worker will be allowed to rebut the defense by showing either that he acted reasonably in reference to the risk33 or that the manufacturer could have shielded the moving parts at trivial cost.34 The second exception is not just a token of sympathy for accident victims. The economic purpose of a defense of contributory negligence or assumption of risk is to identify the participant in the accident who could have avoided it at lowest cost and to make sure that he bears the cost of the accident. Even though the worker may have been negligent in the sense that the cost of avoiding the accident was less than the expected accident cost, the manufacturer may have been more negligent because the cost of shielding the moving parts was less than the cost of care to the worker. If so, economic analysis indicates that the manufacturer should be liable rather than the worker. For this is simply an alternative care case, where either party could prevent the accident at less cost than both together, and it is cheaper for the manufacturer than for the consumer to do so. The problem with "foreseeable misuse," as this defense to a defense is sometimes called, is that if pushed too far it would eliminate the defense of contributory negligence entirely rather than merely confine it to cases where the victim really was the lower-cost avoider of the accident. It is well known, hence foreseeable, that there are some extremely careless (high-cost-of-care) people. But if this makes the manufacturer fully liable to all consumers because the cost of care is prohibitive for some, then ordinary consumers, as distinct from extremely careless ones, will have no incentive (more realistically—because tort damages for severe personal injuries usually undercompensate—less incentive) to avoid the Occident, even though by assumption they can do so at lower cost than the manufacturer. For extremely careless persons the efficient solution is not so clearcut. Maybe they should stop using the product or invest in becoming more skilled at taking care. And even if the most efficient solution would be for the manufacturer to take greater care, the proper liability rule may still be no liability, provided that the extremely careless are a small fraction of all consumers and the manufacturer cannot identify and take special precautions in advance with reference to them. On these as33. See 2 Louis R. Frumer and Melvin I. Friedman, Products Liability § 16A[5][f] (1984). 34. See, for example, Forrest City Machine Works, Inc. v. Aderhold, 273 Ark. 33, 616 S.W.2d 720 (1981).

Products Liability and Industrial-Accident Law · 301 sumptions there are two possibilities. The first is that the manufacturer will do nothing to increase the safety of the product, preferring instead to pay an occasional liability judgment. There will be no allocative benefit of liability and some administrative cost. This might seem the likelier possibility if as we have assumed the fraction of extraordinarily careless consumers is very small. But bear in mind that if out of solicitude for these consumers the law abolishes all defenses based on victim fault, the incentive of ordinary consumers to take care in using products will be diminished, and the manufacturer may decide to compensate by making the product safer—"idiot-proof"—even though only a small fraction of his consumers are idiots in this sense. (This is the second possibility.) If this happens, however, then by assumption a less efficient method of accident prevention (manufacturer care) is being substituted for a more efficient one (consumer care). This assumes that the fraction of extraordinarily careless consumers is not so large that it becomes efficient for the manufacturer to take care for their sake, even though the result is to reduce the incentives of ordinary consumers to take care. But if the fraction were large enough, we would no longer be speaking of extraordinarily careless consumers; these would be the ordinarily careful and the others would be extraordinarily careful. When a product accident is due to misuse by the dealer or someone else intermediate between the manufacturer and the victim, we are in the realm of intervening causes. An example is provided by the recent (and largely unsuccessful) efforts of victims of gunshot wounds inflicted by criminals to obtain damages from the gun's manufacturer on a theory of strict liability.35 Of course the gun is not defective in any ordinary sense—rather the opposite—but the argument is that it is unreasonably dangerous. The argument for liability can be strengthened by pointing out that the criminal who pulls the trigger may be insolvent and therefore not amenable to deterrence through tort law and that the manufacturer, although he cannot control the criminal directly, will, by raising the price of his guns (to compensate for expected liability costs to the victims of any criminals who possess his guns), and so reducing his output, reduce the amount of criminal use of guns. The economic objection to this method of crime control is that it raises the price to all users of guns, the lawful and unlawful alike (indeed, it increases the costs of the police!), whereas more vigorous enforcement of the criminal law would 35. See Adkinson v. Rossi Arms Co., 659 P.2d 1236 (Alaska 1983); Note, "Handguns and Products Liability," 97 Haw. L. Rev. 1912 (1984); Note, "Manufacturers' Strict Liability for Handgun Injuries: An Economic Analysis," 73 Ga. L. Rev. 1437 (1985).

302 · The Economic Structure of Tort Law concentrate those costs on the criminals. If, however, it were possible to identify a class of guns used mainly for illegal purposes ("Saturday night specials"?), a stronger case could be made for strict liability as a method of trying to price them off the market. 36 8. Damages. Damage issues in products liability cases are in the main no different from those in tort cases generally. The only issue that requires separate consideration is the occasional award of punitive damages in products liability cases. At first glance it seems highly inefficient to award such damages in any strict liability case, because if liability is strict there will be cases where the manufacturer will, if penalized, spend more money on care than the benefits in added safety—he will, in other words, overinvest in safety. For example, if the expected accident cost is $5, the cost of prevention is $6, and the expected punitive damages are $2, a wasteful precaution (because $6 > $5) will be taken (because $7 > $6). See Chapter 6. But as a matter of fact punitive damages are not awarded routinely in product liability cases, but only where there are circumstances of aggravation. These include reckless indifference to safety, gross negligence, and concealment by the manufacturer of the dangerousness of his product from consumers. 37 The last is a form of fraud, which is an intentional tort, and the economic case for punitive damages is the same as in other intentional-tort cases. Punitive damages for reckless conduct may also be defensible on economic grounds, as we saw in Chapter 6. Because so much concern has been expressed about runaway juries imposing colossal punitive damage awards in products liability cases, we attempted to estimate the quantitative importance of such awards by examining all reported products liability cases in the federal courts of appeals from the beginning of 1982 through November 1984. Before we describe the sample cases and the outcomes we note that the sample itself is likely to be unrepresentative of the universe of products liability cases—but unrepresentative in a direction that tends to bias our estimation against the hypothesis that we test, and reject: that punitive damages are being overused in modern products liability law. The economic theory of litigation predicts that the cases most likely to be litigated and appealed will be those where the law is most uncertain and where 36. As held—after the above was written—in Kelley v. R.G. Industries, Inc., 497 A.2d 1143 (Md. App. 1985). 37. See David G. Owen, "Punitive Damages in Products Liability Litigation," 74 Mich. L. Rev. 1257, 1325-71 (1976); David G. Owen, "Problems in Assessing Punitive Damages against Manufacturers of Defective Products," 49 U. Chi. L. Rev. 1, 28 (1982).

Products Liability and Industrial-Accident Law · 303 Table 1 0 . 5 . P r o d u c t s liability cases: Federal c o u r t s of appeals, 1982-November 1984 T y p e of case

Design defect

Process failure

Component defect failure

Unavoidably dangerous

Total

141

11

9

11

172

55

3

4

5

67

Reversed

19

2

1

2

24

Defendant wins

86

8

5

6

105

Reversed

37

4

2

3

46

N e t plaintiff

73

5

5

6

89

68

6

4

5

83

Total c a s e s Plaintiff w i n s in trial c o u r t

in trial c o u r t

wins" Net defendant winsb a. trial b. trial

Plaintiffs' wins in trial court less reversals plus reversals when defendant wins in court. Defendants' wins in trial court less reversals plus reversals when plaintiff wins in court.

damages are largest. 38 This suggests that our sample will overstate the relative importance of punitive damages in products liability cases; an award of punitive damages will increase the probability of an appeal. Of course, if parties are highly risk averse, this would tend to raise the settlement rate in cases where punitive damages are likely to be awarded if the case is tried. But we have offered reasons for doubting the importance of risk aversion in the products liability area. Table 10.5 classifies our sample cases by type of product injury and includes a breakdown of the outcomes in both the trial and appellate courts. Not surprisingly, design-defect claims, where the standards of liability are less certain than for other areas of products liability, comprise the greatest proportion (more than 80 percent) of our cases. We also find, consistently with a recent economic study of litigation,39 that plain38. See William M. Landes, "An Economic Analysis of the Courts," 14 J. Law & Ecoti. 61 (1971); Richard A. Posner, "An Economic Approach to Legal Procedure and Judicial Administration," 2 J. Legal Stud. 399 (1973). 39. See George L. Priest and Benjamin Klein, "The Selection of Disputes for Litigation," 13 /. Legal Stud. 1 (1984).

304 · The Economic Structure of Tort Law tiffs win about 50 percent of litigated cases. Interestingly, this is not true in the trial court. Our 50 percent figure is due to the greater frequency of reversals when the defendant has won in the lower court. Table 10.6 provides data on punitive damages. Notice the relative insignificance of punitive damages (made even more striking by the likelihood that a sample limited to appellate cases will overstate the importance of punitive damages in products liability cases). Of the 172 cases in the sample, punitive damages were upheld on appeal in only four instances, which is fewer than 3 percent of the cases and fewer than 5 percent of the cases that plaintiffs won. Observe also that punitive damage awards are more likely to be reversed than are other outcomes. For example, six of ten punitive damage awards were reversed and one more was reduced by two-thirds. In comparison, fewer than one-third of the decisions in favor of plaintiffs but not awarding punitive damages were reversed. An extension of our study through mid-1985 yielded a total of fortyeight more cases (again mostly design-defect cases), in only one of which punitive damages were upheld on appeal, as shown in Table 10.7. This is only 2 percent of the total number of cases and 4 percent of the cases that the plaintiffs won. (Notice that again the plaintiffs won half the cases.)

Table 10.6.

Punitive damage awards

Trial court Actual ($)

Punitive ($)

Ratio of actual to punitive

400,000 165,000 1,060,000 950,000 100,000 391,500 560,000 175,000 350,000 127,000

100,000 35,000 1,000,000 500,000 200,000 625,000 440,000 210,000 300,000 500,000

4.0 4.7 1.1 1.9 0.5 0.6 1.3 0.8 1.2 0.3

Appellate court: Punitive damages reversed? Yes Yes Reduced® No Yes Yes Yesb Yes No No

a. Punitive damages reduced to $300,000 and actual damages to $290,000 or new trial on damages only. b. Remanded with instructions that punitive damages would be proper if "reckless disregard for public safety" found.

Products Liability and Industrial-Accident Law · 305 Table 10.7.

Punitive damage awards—follow-up sample

Trial court Actual ($)

Punitive ($)

390,000 1,500,000 1,000,000

3,900,000 750,000 3,000,000

Ratio of actual to punitive

Appellate court: Punitive damages reversed?

0.1

Yes

2.0 0.3

No Yes

But might a sample limited to federal courts understate the frequency of punitive damages? A plaintiff who wants to be in state court can usually defeat the attempt of an out-of-state defendant to remove the case to federal court under the diversity jurisdiction by joining as an additional defendant the dealer who sold the product to the plaintiff. The dealer will usually be a local firm, so citizens of the same state will be on both sides of the suit and the federal court will have no jurisdiction. Hence if state courts are more sympathetic to punitive damage claims, we might expect such cases to show up more in state than federal courts. We therefore examined all the product liability cases reported in the ten most recent volumes of each of the West Publishing Company regional reporters, excluding New York and California cases for reasons to be explained shortly. This exercise yielded a sample of 119 cases, all decided either late in 1984 or in 1985. The results are shown in Tables 10.8 and 10.9, which follow the format of Tables 10.5 and 10.6. As in the federal court samples, design-defect cases greatly predominate and plaintiffs win about 50 percent of our cases. Yet the exact percentage is actually lower than in the combined federal sample—49 percent versus 51 percent. And notice that in less than 2 percent of all the cases and in only about 3 percent of the cases in which plaintiffs prevailed were punitive damages upheld. The percentages are lower than in the federal court samples. We examined the New York and California cases separately to see whether punitive damages are more generously awarded in some of the more "liberal" states. In the twenty New York and California cases in our sample, punitive damages were awarded in none. Again plaintiffs won 50 percent of the cases, and again design-defect cases predominated (90 percent). The insignificance of punitive damages in our samples is evidence that they are not being routinely awarded. Out of a total of 359 cases in all of our samples, punitive damages were allowed in only seven—2 per-

306 · The Economic Structure of Tort Law Table 10.8. 1985

Products liability cases: State appellate courts, December 1984Design defect

Process failure

Component defect/failure

Unavoidably dangerous

Total

103

9

0

7

119

Plaintiff wins in trial court Reversed

34

5

0

1

40

10

1

0

0

11

Defendant wins

69

4

0

6

79

Reversed

25

2

0

2

29

Net plaintiff wins

49

6

0

3

58

Net defendant wins

54

3

0

7

61

Total cases

in trial court

Table 10.9.

Punitive damages awards

Trial court: Actual ($)

Punitive ($)

Ratio of actual to punitive

349,500 1,107,000

304,500 750,000

1.15 1.50

No No

100,000

450,000

3.22

Yes

Appellate court: Punitive damages reversed?

cent. (The average award of punitive damages in those seven cases was less than $500,000, slightly more than the average actual damages awarded in those cases.) The results of our study are consistent with the results of a recent study by the Rand Corporation's Institute for Civil Justice of all civil jury trials in San Francisco and in Cook County, Illinois (the county in which Chicago is located) between 1960 and 1984.40 In this entire period there were only eight awards of punitive damages in products liability cases. In percentage terms, only eight-tenths of 1 percent of all jury trials in 40. Mark A. Peterson, Punitive Damages: Preliminary Empirical Findings 1 6 - 1 7 ( N - 2 3 4 2 ICJ Aug. 1985). See also Stephen Daniels, "Punitive Damages: The Real Story," ABAJ, Aug. 1, 1986, at 60.

Products Liability and Industrial-Accident Law · 307 San Francisco in products liability cases resulted in such an award; the corresponding figure for Cook County was only one-tenth of 1 percent. These figures, which as expected are lower than those for our samples of appellate cases, actually overstate the success of plaintiffs in obtaining punitive damage awards, because many jury awards are reduced by the trial judge or on appeal. Unfortunately, the Rand study does not indicate the dates of the punitive damage awards, so we do not know how many of the eight awards fell in a period comparable with our study. The Rand study does show an increase in punitive damage awards in personal injury cases in the period 1980 to 1984—modest in San Francisco, steep in Chicago—but does not break down the increase by type of case.41

New Developments Two recent developments are putting strains on products liability law, viewed economically. The first is the increasing durability of goods. It is not unusual for machinery to have a useful life of many years. A defect may appear and cause an accident many years after the machine had been produced. Proof of defect and defenses against such proof are of course more difficult to adduce with passage of time, increasing the risk of legal error. Nevertheless under accepted notions of how statutes of limitations work, the time for suit would not begin to run until the accident occurred. Another problem is the long-delayed consequence of a product defect or danger, as in the asbestosis, dioxin, and DES cases. There were delayed reactions in earlier times, too, but scientific knowledge was not sufficiently advanced then to trace out the causal sequences. Nevertheless, Chapter 9 suggested that tort law may be able to cope with this problem. There is a movement in the states to adopt an outside time limit for bringing products suits, measured from the date the product was sold rather than the date the injury occurred or was discovered.42 Ten years is a common period. The effect of such an arbitrary cutoff date is, by limiting the manufacturer's exposure, to reduce the effect of strict liability in creating incentives to make safer products or steer consumers to substitute products that are safer. But offsetting this is the reduction in the risk of legal error, a reduction brought about by cutting off liability for remote acts. We are not prepared to evaluate this tradeoff beyond 41. Peterson, note 40 above, at 24-25. 42. See 3 Frumer and Friedman, note 33 above, at § 16C[2][i],

308 · The Economic Structure of Tort Law noting that it cannot be condemned out of hand on economic grounds. Another modern development is the growth of both domestic and international trade; as a result products are more likely than they were in the nineteenth century to be produced in one state and consumed in another or produced abroad and consumed domestically. Because the consuming state sets the products liability standard (that is where the accident occurs, and the law of the place of the accident is, under traditional conflict-of-laws principles and on the whole still today, the law that governs liability issues arising out of an accident), and because it may be very costly for a manufacturer to design or manufacture his product differently for sale in different markets, the consuming state may, through the products liability standards that it sets, affect consumers in other states by inducing the manufacturer to alter his product for all his customers. But the consumers in other states are not represented in the political process of the state that is setting the standards, and those consumers may have different preferences regarding safety. This problem provides an economic basis for recent proposals to enact national limitations on state products liability laws. 43 Of course, it does not follow that such limitations would be efficient, and it does not tell us what form they should take.

A Glance at industrial-Accident Law We conclude this chapter with a brief glance at a problem that is related from aft economic standpoint to that of product accidents: that of accidents arising out of the employment relationship. When a worker is injured on the job he may wish to sue his employer, much as an injured consumer may wish to sue the seller of the product that injured him. In both cases the tort arises out of a contractual relationship and it may Seem that the law would treat them symmetrically, but it does not and never has. Industrial-accident law, as the law governing the employer's liability to an employee injured on the job is usually known, was a well-developed body of tort law in the nineteenth century. 44 The employer was 43. For a good discussion see David A. Rice, "Product Quality Laws and the Economics of Federalism," 65 B.U. L. Rev. 1 (1985). 44. Described in Richard A. Posner, " A Theory of Negligence," 1 J. Legal Stud. 29, 4 4 46, 6 7 - 7 1 (1972); Crystal Eastman, Work-Accidents and the Law, chs. 1 2 - 1 3 (1910); Arthur Gray Powell, "Some Phases of the Law of Master and Servant: An Attempt at Rationalization," 10 Colum. L. Rev. 1 (1910).

Products Liability and Industrial-Accident Law · 309 liable only if negligent and could defend by reference to the employee's assumption of risk or contributory negligence. The rule of respondeat superior was not applied as such (the significance of the qualification will become clear). Therefore, if an employee was injured by the negligence of another employee, the employer was not liable (this was known as the fellow-servant rule) provided that the employer had not been negligent in hiring, supervising, or failing to fire the employee who caused the injury. Although some courts limited the application of the fellow-servant rule to workers working in some proximity to each other, so that it would be realistic to expect them to monitor each other's safety and report careless conduct to supervisory personnel, most courts rejected the proximity limitation.45 The doctrinal structure seems broadly consistent with efficiency. The worker who assumes the risk of unsafe working conditions (presumably in exchange for a higher wage—the phenomenon of compensating differentials pointed out by Adam Smith and attested to by a large modern literature),46 is careless, or fails to report a careless fellow worker who later injures him is barred from recovering damages—in the first case as a quid pro quo, in the second and third cases to deter inefficient conduct. 47 Although the fellow-servant rule has no counterpart in accident law involving either strangers or consumers, this is because strangers and consumers are in no position to observe and report on the safety behavior of employees of the injurer before the injury occurs. Thus it is possible to reconcile the fellow-servant rule with the rule of respondeat superior in ordinary accident cases and the rule of strict liability in prod45. See id. at 3 1 - 3 2 . 46. See Richard Thaler and Sherwin Rosen, "The Value of Saving a Life: Evidence From the Labor Market/' in Household Production and Consumption 265 (Nester Ε. Terleckyj ed. 1975); Craig A. Olson, " A n Analysis of Wage Differentials Received by Workers on Dangerous Jobs," 16 }. Human Resources 167 (1981); W. Kip Viscusi, Employment Hazards: An Investigation of Market Performance, pt. Ill (1979); W. Kip Viscusi, Risk by Choice: Regulating Health Safety in the Workplace, ch. 3 (1983). The phenomenon of wage premia for dangerous work was well known to contemporary scholars of industrial-accident law. See, for example, Powell, note 44 above, at 30. The phenomenon is overlooked by those who view nineteenth-century industrial-accident law as a device for lowering the costs of industry. See, for example, Lawrence M. Friedman and Jack Ladinsky, "Social Change and the Law of Industrial Accidents," 67 Colum. L. Rev. 50, 58 (1967). 47. Francis M. Burdick, "Is Law the Expression of Class Selfishness?," 25 Harv. L. Rev. 349, 363 (1912), gives a remarkable example of the logic of the fellow-servant rule: a p r e Civil-War case from Georgia where a slaveowner was held not barred from recovering damages for the death of his slave at the hands of fellow servants who were free men, on the ground that the slave would not have dared to take any steps to make his free fellow servants work more safely!

310 • The Economic Structure of Tort Law ucts cases. Moreover, although the limitations on employers' liability to their employees has long been thought an example of the inhumanity and class bias of nineteenth-century tort law, their liability was greater than that of a manufacturer to the consumers of his defective products, because a suit by an employee did not encounter the problem of lack of privity of contract. Whether in practice the tort law of industrial accidents was efficient is a debated and inconclusive question. 48 The principal doubts concern the information of employees and the competitiveness of labor markets. We shall make no attempt to resolve the debate but merely note the formal adequacy of industrial-accident law in a model of efficient rules of tort law—although with the following qualification: the fellow-servant rule dated from the 1840s in this country and became progressively less realistic with the growth of the size of firms. The rejection of a proximity limitation is a puzzle. Beginning around the turn of the century Congress and the state legislatures made dramatic inroads into the tort law of industrial accidents. The workmen's compensation movement substituted for common law tort liability a form of strict liability, with no defenses of assumption of risk or contributory negligence and of course no fellow-servant rule but also with a fixed schedule of damages less comprehensive than common law tort damages. And workmen's compensation was administered by state regulatory agencies rather than by courts. Almost all workers are now covered by workmen's compensation—except workers in the railroad and maritime industries. They retain their common law tort rights and in amplified form: they can recover damages on proof of minimal negligence, there is no fellow-servant rule or defense of assumption of risk, disclaimers of liability are not enforced, and contributory negligence is replaced by comparative negligence. These developments do not seem efficient. Subject to a qualification noted in the next paragraph, workmen's compensation dilutes the worker's incentive to take care, 49 although it does not eliminate it; because 48. Compare Posner, note 4 above; James L. Croyle, "Industrial Accident Liability Policy of the Early Twentieth Century," 7 J. Legal Stud. 279 (1978), and Richard A. Epstein, "The Historical Origins and Economic Structure of Workers' Compensation L a w / ' 16 Ga. L. Rev. 775, 7 7 7 - 8 6 (1982), with Gary T. Schwartz, "Tort Law and the Economy in NineteenthCentury America: A Reinterpretation," 90 Yale L.J. 1681 (1981); James R. Chelius, "Liability for Industrial Accidents: A Comparison of Neligence and Strict Liability Systems," 5 /. Legal Stud. 293 (1976), and C. G. Veljanovski, "The Employment and Safety Effects of Employers' Liability," 29 Scottish J. Pol. Econ. 256 (1982). 49. For some empirical evidence of this see Richard J. Butler, " W a g e and Injury Rate

Products Liability and Industrial-Accident Law · 311 workmen's compensation benefits do not produce full compensation, the worker retains a monetary as well as nonmonetary incentive to take some care. The Federal Employers Liability Act and the Jones Act (governing the tort rights of railroad and maritime workers, respectively) create something close to strict liability and remove important defenses, notably assumption of risk. These developments, too, are puzzling in light of the Coase theorem. Although workers may face some information costs about workplace safety, those costs are probably much lower than in the products context, especially given the role of unions as information intermediaries. If so, any necessary adjustments in conventional tort principles could be left to be worked out by contract between employers and employees. The Coase theorem also implies that whatever the rule of liability is, workers and employers would recontract to reach the efficient safety levels, provided that transaction costs are not prohibitive. Thus, an inefficient industrial-accident law may have little impact on safety and accidents and merely increase transaction costs. However, with disclaimers of liability unenforceable, contracting around workmen's compensation laws and employers liability acts becomes difficult or even impossible. Statutes present many puzzles to the positive economic theorist of law, as we mentioned in discussing wrongful death statutes in Chapter 6. For purposes of our study, a study of judge-made rather than statutory law, it seems enough to point out that the specific doctrines of tort industrial-accident law, although now superseded by legislation, seem generally to have been efficient, at least formally, and to have been economically consistent with, although different from, the corresponding doctrines applicable to accidents between strangers, on the one hand, and between sellers and buyers of products, on the other. Responses to Shifting Levels of Workers' Compensation in Safety and the Work Force," in Incentives and Disincentives in Workers' Compensation 61 (John D. Worrall ed. 1983).

. 11 . A Brief Summary, with Suggestions for Further Research in this book has been to expound and test the hypothesis that the rules of the Anglo-American common law of torts are best explained as if designed to promote efficiency in the sense of minimizing the sum of expected damages and costs of care; or, stated differently, that the structure of the common law of torts is economic in character. This hypothesis, which we also refer to as the positive economic theory of tort law, is part of the broader hypothesis that common law rules in general, not just those in the tort field, have an implicit economic structure. The logic of the common law is an economic logic. The particular form of the hypothesis that we have defended is a modest one. We do not contend that all rules of tort law are efficient— only that most are. And we do not contend that the rules as they are actually implemented on the firing line—by judges, lawyers, jurors, claims adjusters, safety engineers, business executives, and insurance regulators—achieve an efficient allocation of resources to safety, at least when efficiency is conceived apart from institutional frictions. There is a gap betwen the law in the books and in action; our focus in this book, however, has been on the former. We argue that the law creates incentives for parties to behave efficiently rather than that they actually behave so. The body of tort rules—vast, reticulated, controversial—provides a sufficient object of study for one book. It would be absurd to think every rule of tort law efficient, when one considers the weak incentives of judges to achieve efficient results or indeed results consistent with any goal. The institution of judicial independence guarantees a certain randomness in the judicial product. What is nevertheless remarkable is the degree to which this large body of law (and again we stress that we are talking about the law in the books and not necessarily the law in action) can be reduced to a handful O U R OBJECTIVE

Suggestions for Further Research · 313 of equations. But this result is startling because of its unfamiliarity rather than because of any inherent implausibility in assigning an economic function to common law rules. As explained in Chapter 1, the reader need not be convinced that efficiency ought to be the sole or even primary guide to social choice or that judges are perfect agents of the legislatures that set the goals of government. All that is necessary is that the reader accept two possibilities. The first is that because the courts, especially in such fields as tort law, are poorly equipped to develop and implement doctrines that will effectively redistribute wealth to particular interest groups (rather than randomly), the forces that control the political system will find it in their self-interest to give the courts the function of maximizing the size of the economic pie and to other institutions the function of rearranging the slices of the pie in accordance with the balance of political power. The second is that judges are sufficiently good although not perfect agents of those dominant forces to be relied on to carry out their assignment with some although not perfect consistency, imagination, and diligence. The evidence for the positive economic theory of tort law is considerable. Chapters 2 through 10 of this book ranged widely over the doctrines of tort law. We considered not only the negligence standard itself, where the economic calculus is most visibly at work, but the division between negligence and strict liability; the defenses to negligence and strict liability (including such novelties as foreseeable misuse); the concepts of duty and custom; the rules governing causation and joint liability; the principle of vicarious liability, as in the doctrine of respondeat superior, and the limits and exceptions to that doctrine; the torts of deceit, defamation, and interference with contract, as well as the torts that protect against personal injury and property damage; and much else besides. We not only considered individual cases and doctrines but essayed comparisons over time and across jurisdictions in an effort to correlate doctrinal variation with economic differences. Although our analysis is mostly qualitative, we have offered some statistics, notably on the award of punitive damages, in support of the economic hypothesis. We have considered not only the gross economic distinctions that inform tort law but the subtle ones, such as the difference between joint and alternative care; and this shows that we are not just engaged in translating common sense into jargon and calling the result "economics." We are mindful of the dangers of rationalization. Knowing what the rules of tort law are, we are in a position to construct an economic rationale for each one, working backward. It is partly to reduce this

314 · The Economic Structure of Tort Law danger that we have kept our economic model extremely simple, as by assuming risk neutrality, which greatly reduces our degrees of freedom. The explanatory power of the simple model seems great. Think back to the large number of topics discussed in the previous chapters of this book, in addition to those listed above: the different remedies in nuisance and trespass cases and the special doctrines of nuisance law (for example, permanent versus temporary nuisance); the different rules applicable to domestic and wild animals, to straying livestock, to sparks, and to blasting; the reasonable-man rule, last clear chance, attractive nuisance, and other special doctrines of negligence law; the gradations of tort culpability, ranging from strict liability to simple and then gross negligence and to recklessness and intentional wrongdoing; the economic-loss cases; the rules governing tort damages, such as the eggshell-skull rule; the evolution of products liability law; and many others. The large fraction of tort rules, weighting number by importance, that conform to our economic model are evidence that the model captures an important part of legal reality. For those readers for whom economic efficiency is not merely a descriptive theory but a call for action, we have suggested several areas in which tort law might be changed to make it conform more closely to the theory; perhaps these changes would just nudge it a little faster along its natural line of development. In particular we have suggested a method of computing the pecuniary value of the loss of utility caused when a person is killed as a result of a tort and a method of making tort law a more effective deterrent in situations where tortious conduct gives rise to a low probability of illness or death but the total harm eventually suffered is considerable. Yet this book stands only at the threshold of economic research on the tort system. Not only are our occasional suggestions for reform merely sketched, with perhaps unconvincingly broad strokes, but even in the area of our primary interest—testing the positive economic theory of tort law—the book is only a beginning. For example, we have noted several important areas where the common law of torts seems not to be efficient: the rejection of custom as a defense, along with the refusal to enforce disclaimers of liability for negligence, even in areas of low transaction costs; the refusal in death cases to compensate for the loss of utility to the victim himself; and the trends in the law to comparative negligence and to contribution among joint tortfeasors—doctrines that impose additional administrative costs on the legal system with no gain in allocative efficiency. One would like to know what forces have generated these anomalies. Our statistical analysis of the comparative neg-

Suggestions for Further Research · 315 ligence and contribution movements, our analysis of a small sample of citations to The T. ]. Hooper, and our speculations on the difficulties of measuring loss of utility in wrongful death cases only scratch the surface of these important questions. Our focus on rules rather than outcomes, moreover, puts us at one remove from the basic data of analysis in a common law system: the actual decisions of the appellate courts that make and apply common law doctrine and the trial courts that apply it. Modest examples of a different approach are, again, our study of the citation history of The T. J. Hooper in Chapter 5; in Chapter 4 our study of the negligence cases in the Gregory, Kalven, and Epstein casebook; and in Chapters 6 and 10 our tabulations of punitive damage awards on appeal. Much more could be done in the way of systematic analysis of tort cases—for example, by constructing a sample of tort cases weighted by the number of citations to each case, such number being a proxy for the importance of the case in shaping the law. A particularly promising avenue for further study is illustrated by the attempt in Chapter 10 to relate state products liability rules to urbanization and other variables that could be expected—if the positive economic theory of tort law is correct—to produce differences across states in these rules. Another possibility is to see whether differences across states in the rules of defamation law can be explained by differences in levels of education, on the theory that the more educated a state's population is, the less need it has for a strong defamation law—the more the marketplace in ideas can protect people against injury to reputation caused by false statements. More broadly, differences in the rules of tort law both across states and between United States and other common law jurisdictions can be related to the economic differences that would be expected to produce those differences in rules if tort law is indeed shaped by an implicit concern with promoting economic efficiency. We have made occasional stabs at such comparisons—for example, in discussing the different rules relating to liability for property damage caused by straying livestock in the western and eastern states and the different rules on liability for railroad spark damage in the United States and England; but for the most part we have treated the common law as if it were the law of a single jurisdiction. Although much more should be done to put the positive economic theory of tort law on a firm footing, it is equally important to note that more can be done. The theory can be embedded in a broader economic theory of judicial and legislative action in which such apparent anomalies as the movements to comparative negligence and to contribution may

326 · The Economic Structure of Tort Law disappear. Perhaps, as suggested in Chapter 9, in states that are relatively collectivist judicial rules take on a relatively collectivist hue; perhaps this also depends on how independent from political control the judges are in such states, a matter not explored in our analysis. We have made several suggestions for further empirical work and here is another: if we are right that the only difference between contributory and comparative negligence is that the latter costs more to administer, it should be possible to compare accident rates and insurance costs across states, correcting for other factors, to see whether, as we would predict, the adoption of comparative negligence does not alter the accident rate but does raise liability insurance costs. Such a study would begin to answer the question (on which some evidence was cited in Chapter 1) whether the economic theory of tort law is a fruitful theory of the law in action as well as of the law on the books.

Case Index

Adams v. Bullock, 97-98, 102 Adkinson v. Rossi Arms Co., 301n Alcorn v. Mitchell, 161n, 169 American Surety Co. v. Schottenbauer, 224-25 Anderson v. Minneapolis, 197n A. R. Boroughs v. Joiner, 208n Atlantic Coast Development v. Napoleon Steel Contractors, 208n

Baker v. Pendergast, 90n Barber Lines A/S v. M/V Donau Maru, 251 η Barrett v. Foster Grant Co., 135n, 136 Barton v. Owen, 138n Basko v. Sterling Drug, Inc., 296n Basso v. Miller, 93n Beinhorn v. Griswold, 108n, l l l n Berry v. Sugar Notch Borough, 229n, 238, 240 Berry v. V. Ponte & Sons, 282n Beshada v. Johns-Manville Products Corp., 291n Birchfield v. International Harvester Co., 291n Bird v. Holbrook, 176-78 Blyth v. Birmingham Waterworks Co., 99-100, 102, 103 Bolton v. Stone, 99, 103 Boomer v. Atlantic Cement Co., 44-47 Bovsun v. Sanperi, 243n Boyd v. Thompson-Hayward Chem. Co., 296n

Breunig v. American Family Insurance Co., 130, 183 Brown v. Link Belt Division, 291n Buford v. Houtz, l l l n Burlington & M. R. Co. v. Westover, l l l n Carpenter v. Double R Cattle Co., 50n Central of Georgia Ry. v. Price, 238n Chicago, B. & Q. R. Co. v. Krayenbuhl, 87n Cities Service Co. v. State, 116 City of Denver v. Kennedy, 108n City of Piqua v. Morris, 234-36, 240 Conners-Standard Marine Corp. v. Marine Fuel Transfer Corp., 135n Conway v. O'Brien, 151n, 159n, 249 Cook v. Minneapolis, St. P. & S. S. M. Ry. Co., 229n Cooley v. Public Service Co., 104 Corey v. Havener, 197n Costa v. Storm, 185n Courvoisier v. Raymond, 174-75 Cox v. Hennis Freight Lines, Inc., 90n Culver v. Slater Boat Co., 186n Cyr v. B. Offen & Co., 299n Daniels v. Evans, 128n Davis v. United States, 92n Delaney v. Errickson, 108n DePass v. United States, 265n, 268n Dickeson v. Baltimore & Ohio Chicago Terminal R. R., 128n Doe v. District of Columbia, 135n Dole v. Dow Chemical Co., 209-10

318

· Case

Index

Donovan v. Robbins, 203n Dougherty v. Stepp, 170n Doumitt v. Diemer, 241-42 Doyle, The, 135n Duke Power Co. v. Carolina Environmental Study Group, Inc., 15n, 266n Dulieu v. White & Sons, 243n Dundee Cement Co. v. Chemical Laboratories, Inc., 253n Dziokonski v. Babineau, 243n Eckert v. Long Island R. Co., 100-2, 103, 250n, 251 Ehret v. Village of Scarsdale, 238-39 E. Hulton & Co. v. Jones, 163-65 Elmore v. American Motors Corp., 288n Elwood v. Bolte, 115n Erie R. Co. v. Stewart, 132n Escola v. Coca-Cola Bottling Co., 64n

Guille v. Swan, 114n Guinard v. Knapp-Stout & Co., 64n Gyerman v. United States Lines Co., 135n Hall v. Ε. I. Du Pont de Nemours & Co., 135n Harding v. Philadelphia Rapid Transit Co., 102n Harris v. Johnson, 90n Hauser v. Chicago, R. I. & P. Ry., 105 Helling v. Carey, 138 Hemingway v. Ochsner Clinic, 137n Hendrichs v. Peabody Coal Co., 96-97, 102-3 Henkel v. R and S Bottling Co., 298n Henningsen v. Bloomfield Motors, Inc., 286n Herrick v. Wixom, 239n Herskovits v. Group Health Coop., 265n Hession v. Liberty Asphalt Products, Inc., 89n, 90n Hillier v. Southern Towing Co., 211n Hohenberg v. 77 West 55th Street Associates, 46n Hollenbeck v. Ramset Fasteners, Inc., 296n Hudson v. Craft, 172n Hurley v. Eddingfield, 14n, 147n Hutcherson v. Alexander, 51n

Farrell Lines, Inc. v. S/S Birkenstein, 135n Feldman v. Lederle Laboratories, 291n Ferroggiaro v. Bowline, 245-46, 247 First Nat'l Bank in Albuquerque v. NorAm Agricultural Products, Inc., 64n Fletcher v. Union Pac. R. R., 266n Ford Motor Co. v. Matthews, 299n Ford Motor Co. v. Stubblefield, 182n Formella v. Ciba-Geigy Corp., 296n Forrest City Machine Works, Inc. v. Aderhold, 300n Fowlks v. Southern Ry. Co., 238n Frazee v. St. Louis-San Francisco Ry., 93n Fredericks v. Castora, 126n

J'Aire Corp. v. Gregory, 253 Jones v. Festiniog R. Co., l l l n

Gambill v. Stroud, 137n George A. Harmel & Co. v. Maez, 253-54 George W. Gale Lumber Co. v. Bush, 210n Germolus v. Sausser, 174n Gobble v. Bradford, 210n Goldberg v. Kollsman Instrument Corp., 294n Gorris v. Scott, 247-49 Grimshaw v. Ford Motor Co., 182n Grip-Pak, Inc. v. Illinois Tool Works, Inc., 251n Guarino v. Mine Safety Appliance Co., 250n

Kallenberg v. Beth Israel Hospital, 265n Kane v. Branch Motor Express Co., 135n Karjala v. Johns-Manville Products Corp., 266n Kaufman v. Meditec, 291n Keffe v. Milwaukee & St. Paul Ry., 96n Kelley v. R. G. Industries, Inc., 302n Keyser v. Phillips Petroleum Co., 119n Kingston v. Chicago, N.W. Ry., 197n Kirincich v. Standard Dredging Co., 24041 Kleebauer v. Western Fuse & Explosives Co., 118n Knitz v. Minster Machine Co., 291n

Indian Towing Co. v. United States, 132n In re Meyer's Guardianship, 127n

Case Index Krause v. Alamo Nat'l Bank of San Antonio, 208n Kumkumian v. City of New York, 93-95

LeRoy Fibre Co. v. Chicago, Milwaukee & St. Paul Ry., 89n, 90 Liming v. Illinois Cent. R. Co., 102, 141, 142 Lobert v. Pack, 130n Louisiana ex rel. Guste v. M/V Testbank, 252n Louisville R. Co. v. Sweeney, 170 Lucy Webb Hayes National Training School v. Perotti, 106-7

Mackintosh v. Mackintosh, 86n MacPherson v. Buick Motor Co., 286 Marietta v. Cliffs Ridge, Inc., 135n McClellan v. St. P., Μ. & M. Ry., 197n McCormack v. Hankscraft Co., 292n McDevitt v. Standard Oil Co., 299n McDowall v. Great Western Ry., 105-6 McGrath v. Irving, 241n McKinsey v. Wade, 178n McLeod Store v. Vinson, 139-41 Merrill v. City of Wheaton, 180n Merrill Lynch, Pierce, Fenner & Smith, Inc. v. First Nat'l Bank of Little Rock, 205n, 209n Merryweather v. Nixan, 204n Metal Window Products Co. v. Magnusen, 298n Metcalf v. City of Cortland, 238n M'llvoy v. Cockran, 175n Mineral Industries Inc. v. George, 254 Moch Co. v. Rensselaer Water Co., 14243, 224n, 253 Möhr v. Williams, 172n Monroe Carp Pond Co. v. River Raisin Paper Co., 44n Montgomery v. National Convoy & Trucking Co., 146n Moore v. Jackson Park Hospital, 267n Morris v. Piatt, 174-75 Morrison v. MacNamara, 138n Morrissy v. Eli Lilly & Co., 267n Mullen v. General Motors Corp., 298n Muth v. Urricelqui, 205-8

· 319

National Dairy Products Corp. v. Durham, 241n National Steel Service Center, Inc. v. Gibbons, 117 Nebraska Innkeepers, Inc. v. PittsburghDes Moines Corp., 253n Nesbitt v. Community Health of South Dade, Inc., 135n, 136, 138n Neubauer v. Owens-Corning Fiberglass Corp., 266n New York Cent. R. Co. v. Grimstad, 24041, 243, 244n Nivens v. Signal Oil & Gas Co., 266n Northern Trust Co. v. Skokie Valley Community Hospital, 137n North Star, The, 211n Nussbaum v. Lacopo, 98-99, 102n O'Brien v. Cunard S.S. Co., 172n O'Hare v. Merck & Co., 296n Osborne v. Montgomery, 104 Ouligan v. Butler, 197n Palsgraf v. Long Island R. Co., 246-47 Panusuk v. Seaton, 210n Paris v. Stepney Borough Council, 100 Pendergrast v. Aiken, 50n Petition of Kinsman Transit Co., 254 Pfeifer v. Copperstone Restaurant & Lounge, Inc., 185n Phillips v. Croy, 90n Phillips v. General Motors Corp., 286n Phillips v. Kimwood Machine Co., 64n, 291 Ploof v. Putnam, 43, 178-79 Puchlopek v. Portsmouth Power Co., 93n Quintal v. Laurel Grove Hospital, 106-7 Randall v. Chelton, 170n Randolph's Adm'r v. Snyder, 147n Respublica v. Sparkhawk, 180n Rickards v. Sun Oil Co., 251, 253-55 Rivera v. Rockford Machine & Tool Co., 292n Robbins v. Footer, 137n Robert Addie & Sons (Collieries), Ltd. v. Dumbreck, 93n Roberts v. Tardif, 137n Rolen v. Maryland Casualty Co., 108n

320

• Case

Index

Rose v. Socony-Vacuum Corp., 49n Rouleau v. Blotner, 234 Rourke v. Garza, 292n Rylands v. Fletcher, 112 Salem v. United States Lines Co., 135n Satren v. Hader Co-Operative Cheese Factory, 44n Scatena v. Pittsburgh & New England Trucking, 146n SEC v. Crofters, 135n, 136 Shafer v. Η. B. Thomas Co., 136n S. H. Kress & Co. v. Lindsey, 281n Siegler v. Kuhlman, 116-17 Silver v. American Export Isbrandtsen Lines, Inc., 135n, 136 Sindell v. Abbott Laboratories, 212n Southern California Edison Co. v. Coleman, 115n Spang Chalfant & Co. v. Dimon S.S. Corp., 135n Spur Industries, Inc. v. Del E. Webb Development Co., 51 Steinhardt v. Johns-Manville Corp., 266n Steinhauser v. Hertz Corp., 250n Stevenson v. East Ohio Gas Co., 254 Strazza v. McKittrick, 243n Stubbs v. City of Rochester, 265n Sturges v. Bridgman, 50n Summers v. Tice, 212, 213, 242 Terre Haute First Nat'l Bank v. Stewart, 128n Thomas v. Winchester, 286n Titus v. Bradford, Β. & K. R. Co., 137n T. J. Hooper, The, 133-36, 138, 315 Tonsic v. Wagner, 136n Tuckachinsky v. Lehigh Co., 117-18

Turner v. Big Lake Oil Co., 112n Turner v. International Harvester Co., 281n Union Oil Co. v. Oppen, 105, 251-52, 253 Union Stock Yards Co. v. Chicago, B. & Q. R. Co., 201n United Airlines, Inc. v. Wiener, 210n United States v. Carroll Towing, 85n, 103, 104 United States v. Reliable Transfer Co., 211n United Zink & Chem. Co. v. Britt, 96 Urie v. Thompson, 266n Vaughan v. Menlove, 127 Vincent v. Lake Erie Transportation Co., 43, 178-80 Vosburg v. Putney, 168, 169n, 249-50 Wagner v. Bissell, 108n Wagner v. International Ry. Co., 250-51 Warburton v. Ν. B. Thayer Co., 131n Warner v. Capital Transit Co., 210 Watson v. Kentucky & Indiana Bridge & Ry. Co., 246-47 Waube v. Warrington, 243n, 245n Weeks v. McNulty, 234-35, 238, 239-40 West v. Caterpillar Tractor Co., 299n Weyerhaeuser S.S. Co. v. Nacirema Co., 206-7 Whalley v. Lancashire & Yorkshire Ry. Co., 178, 180 Williams v. Montgomver, 70n Yandle v. PPG Industries, Inc., 269n Zylka v. Leikvoll, 146n

Author Index

American Law Institute (Restatement of Torts), 20, 49n, 70, 88n, 111-12, 116, 119n, 176 Ames, James Barr, 4, 148 Anderson, Gary Α., 186n Anscombe, G. Ε. Μ., 149n Aristotle, 14 Arnould, Richard J., 10η Arrow, Kenneth, 192n Baker, J. H., 2n Baumol, William J., 14n Beasley, James Ε., 284n Becker, Gary S., 15n, 23n, 57n Bell, Peter Α., l l n , 169n, 243n Bell, W. G., 181n Bentham, Jeremy, 4, 6, 7 Bishop, William, 24n, 166n, 251n Black, Bert, 241n Black, Charles L„ Jr., 211n Blackstone, William, 2 Blitch, Carolyn, l l n Blomquist, Glenn, 189n Blum, Walter J., 7n, 9 Borgo, John, 228n, 229n Bork, Robert Η., 18n Bowers, William J., l l n Brenner, Joel Franklin, 52n Brown, John Prather, 8, 54n Brown, Ralph J., 186n Bruce, Christopher J., l l n , 24n Burdick, Francis Μ., 309n Butler, Richard J., 10η, 310n Bynon, Theodora, 21 η

Calabresi, Guido, 6 - 7 , 29n, 30n, 54n, 230n Caldwell, Bruce J., 22n Canon, Bradley C., 181n Cardozo, Benjamin Nathan, 98, 286 Chelius, James R., 25n, 310n Cheung, Steven N. S., 57n Clark, Victor S., 285n Coase, Ronald H., 6 - 8 , 31n, 33-34, 54n, 230 Coe, R. D., 15n Cohen, Lloyd, 185n Coleman, Jules L., 16n, 255n Cook, Philip J., 55n Crandall, Robert W „ 10η, 53n Cranston, Ross, 24n, 255n Craswell, Richard, 166n Crow, Carl, 285n Crowley, Diane R., 201n Croyle, James L., 24n, 310n Curran, Christopher, 80n Daniels, Stephen, 306n Danzon, Patricia Μ., 13n Darby, Michael R., 285n Dardis, Rachel, 10η Darling-Hammond, Linda, 25n Davis, E. Eugene, 205n Demogue, Rene, 5n Demsetz, Harold, 7, 34n Depew, Chauncy Μ., 284n Diamond, Peter Α., 7, 8n, 61n, 72n, 73n Dickinson, Frank G., 181n Dickinson, John J., 5n

322 · Author Index Dobbs, Dan Β., 47η, 161η, 184η Douglas, William Ο., 5η Dukeminier, Jesse, 50η Dworkin, Ronald, 9, 16η Eads, George, l l n Easterbrook, Frank Η., 203n Eastman, Crystal, 308n Ehrlich, Isaac, 57n Ellis, Dorsey D., Jr., 157n, 161n, 187n, 189n Epstein, Richard Α., 49n, 50n, 103, 104n, 143, 147-48, 181η, 182n, 202n, 211n, 228η, 255η, 273η, 288n, 291n, 310n, 315 Flanagan, Stephen Μ., 180n Fleming, John G., 201n Fletcher, George P., 8 Fogel, Robert W., H i n Frech, Η. E„ III, 40n Friedman, David, 55n Friedman, Lawrence Μ., 309n Friedman, Melvin I., 300n, 307n Friedman, Milton, 22n Frumer, Louis R., 300n, 307n Furmstron, Michael, 25n Gayle, Fourmier }., III, 267n Gilmore, Grant, 211n Ginzberg, Eli, 181n Givelber, Daniel ]., l l n Gordon, Stuart Μ., 201n Goyer, James L., III, 267n Grabowski, Henry, l l n Grad, Frank P., 297n Grady, Mark F., 230n, 247n Graham, Daniel Α., 55n Graham, John D., 10η Grayston, Richard W., 10η, l l n Gregory, Charles O., 3n, 5, 7n, 103, 181n, 315 Haddock, David, 80n Hamilton, May Τ., 56n Hand, Learned, 85, 133-35, 151n, 249 Harper, Fowler V., 43η, 190η, 222η Harris, Donald, 25η Hart, Η. L. Α., 228η Hartje, Volkmar, 266η Haskins, Caryl P., 181n

Hay, Gustavus, Jr., 186n Hirshleifer, Jack, 12n, 14n, 23n Holmes, Oliver Wendell, 4, 5, 57n, 89-90, 96 Honore, A. M„ 228n Horwitz, Morton J., 71n Hovenkamp, Herbert, 71 η James, Fleming, Jr., 5, 13n, 43n, 190n, 222n Jaros, Dean, 181n Johnson, Dennis Α., 186n Kakalik, James S., 58n Kalven, Harry, Jr., 7n, 9, 103, 181n, 315 Kami, Edi, 285n Kaye, David, 260n Keeler, Theodore Ε., 53n Keeton, Robert E., 6n Keeton, William R„ 13n Keeton, W. Page, 42η, 44η, 48n, 49n, 50n, 64η, 70η, 88η, 90η, 91n, 96n, 103n, 108η, l l l n , 113η, 120n, 121n, 123n, 126η, 127η, 146η, 163n, 166n, 167n, 170η, 172η, 182η, 183n, 184n, 186n, 190n, 192-93, 202η, 204n, 205n, 207n, 210, 219, 222η, 228η, 266n, 271n, 284n Kenny, Α., 149n, 152n Kihlstrom, Β., 113n, 114n King, Joseph H., Jr., 267n Klein, Benjamin, 303n Kniesner, Thomas J., 25n Komesar, Neil K„ 186n Kornhauser, Lewis Α., 120n Kreindler, Lee S., 115n Krier, James Ε., 50n Kwerel, Evan, 13n Labov, William, 21n LaCroix, Sumner J., 186n Ladinsky, Jack, 309n Landes, Elisabeth Μ., 10η, 11, 61n Langefors, U., 113n, 114n Laski, Harold J., 5n, 207n Latin, Howard, 11-12 Lave, Lester Β., 53n Lawrence, R., 149n Leet, L. Don, 113n, 114n Lefler, Robert Α., 205n, 210n Letwin, William, 22n

Author Lilienfeld, David E.( 241n Lindgren, Bjorn, 10η Lone, James Η., 56n Maclean, G. Α., 57n Maitland, Frederick William, 2n Malone, Wex S., 186n Manson, William D., 24n, 42n Marshall, Alfred, 17 McCormick, C. Τ., 184n McEwin, R. Ian, 10η McQuillin, Eugene, 180n Meiland, J. W., 149n Melamed, A. Douglas, 29n, 30n Meriam, Harold Α., Jr., 205n Merrill, Thomas W., 42n, 50n Michelman, Frank I., 30n Miller, H. Laurence, Jr., 186n Mirrlees, James Α., 61n Morris, Clarence, 5n, 207n Munch, Patricia, 58n Murphy, Arthur W., 297n Nelson, Phillip, 285n Nichols, Albert L„ Jr., 189n Nichols, James Ε., 284n Noel, Dix W„ 299n Noonan, John Τ., 246n, 247n O'Brien, Patrick, l l l n O'Connell, Jeffrey, 6n Olson, Craig Α., 10n, 309n O'Sullivan, P. Ν., 149n Owen, David G., 302n Partlett, David, 24n, 255n Peltzman, Sam, 10η, 15n Peterson, Mark Α., 185n, 306n, 307n Pigou, A. C., 5n, 6 - 7 Polinsky, A. Mitchell, 18η, 31n, 40n, 44n, 51n Pollock, Frederick, 2n Pound, Roscoe, 2n Powell, Arthur Gray, 308n Prichard, M. J., 2n Priest, George L., 303n Prosser, William L„ 42n, 50n, 64n, 19293, 210, 219 Pyle, David J., 10

Index

·

323

Rabin, Robert L„ 3n, 98n, 252-55, 291n Ratcliffe, James Μ., 143n Raynes, Harold Ε., 57n Redfield, Amasa Α., 108n Reuter, Peter, l l n Rice, David Α., 308n Rizzo, Mario J., 24n, 251n Roberts, David L., 186n Robinson, Glen Ο., 261n, 266n Rodgers, William H„ Jr., 151n Roe, Mark J., 265n Rosen, Sherwin, 10η, 309n Rosenberg, Alexander, 22n Rosenfield, Andrew Μ., 57n Rosenn, Keith S., 185n Ross, David, 14n Rubin, Paul Η., 24n Santagata, Kenneth V., 297n Schelling, Thomas C., 189n Schick, Anne, 24, 255n Schofield, William, 4, 190n Schwartz, Gary Τ., 9n, 25n, 58n, 71n, 103n, 157n, 310n Seavey, Warren Α., 66n Shavell, Steven, 8, 13η, 18η, 24n, 61n, 76n, 189η, 228n, 230n, 240n, 244, 245n, 256n Shearman, Thomas G., 108n Silver, Jay, 143n Simpson, A. W. Β., 112n Smedley, Τ. Α., 186n Smith, Adam, 4, 17, 309 Steiner, Joseph Μ., 16n Stigler, George J., 15n, 255n Stuart, Charles, 10η Swan, Peter L., 10η, l l n , 25n Sykes, Alan Ο., 120n, 207n Terleckyj, Nestor Ε., 10η, 309n Terry, Henry T., 4, 71 Thaler, Richard, 10η, 309n Thornton, John V., 205n Transgrud, Roger Η., 266n Trenerry, C. F., 211n Trivers, Robert L., 14n Tryon, Rolk Milton, 285n Veljanovski, Cento G., 20n, 25n, 310n Vickrey, William, 7n Viscusi, W. Kip, 10η, 309n

324 · Author Index Webb, Walter P., 110η Weinrib, Ernest J., 255n Wheeler, Malcolm Ε., 291n White, G. Edward, 9n Wilber, C. Κ., 15n Wiley, Jerry, l l n Williams, Glanville L., 108n Williams, Stephen, 7In Winfield, Percy Η., 186n

Winter, William ]., 211n Wittman, Donald, 13n, 93n, 256n Wolfstone, Leon L., 265n Wolfstone, Thomas J., 265n Wooster, Harvey Α., 285n Worrall, John D., 10η, 311n Wright, Richard W „ 228n, 235n, 238n Zeckhauser, Richard, 189n

Subject Index

Accrual of cause of action, 266 Active-passive negligence test, 205 Activity levels, 40-41, 61, 66-71, 102, 110, 121, 128, 141, 180, 260, 262, 294 Airplane accidents, 114-15, 210 Alternative care, 60-61, 90-92, 165, 19091, 198-201, 206, 209n, 210, 214, 223, 300 Altruism, 144n, 187, 217n Animals, liability for damage caused by, 20, 107-11, 119, 247 Antitrust law, 206n Asbestos cases, 256, 266, 307 Assault, 2, 15, 160, 169 Assumption of risk, 118-19, 120, 139-42, 284n, 296-300, 310 Attractive nuisance, 42, 87n, 95-96, 103, 111 Avoidable consequences, 271 Bankruptcy, 265n Bargaining, 34, 45. See also Bilateral monopoly Battery, 2, 15, 48, 154, 162, 168-69, 17378 Bilateral monopoly, 34, 45-47, 179n Blasting cases, 113-14, 117-18, 236-37 Blindness, and standard of care, 127 Boxing matches, 172n Breach of contract. See Contract Cancer, 241, 262n, 270, 299. See also Nuclear accidents Catastrophic accidents, 259ff.

Causation, 114, 122, 146, 170, 228ff., 286, 296, 307; in catastrophic injury cases, 256-57, 260-62; damages, 235-36, 25253; in economic loss cases, 251-55; in eggshell-skull cases, 249-50; in lowprobability accidents, 245-47; in rescue cases, 250-51; and statutory due care, 247-49; in strict liability cases, 236-37; in uncertain-injurer cases, 242-43; in uncertain-victim cases, 242 Caveat emptor, 276-78 Charitable immunity, 181n Children, liability of, 128-30 Class actions, 254, 268-69 Coase theorem, 6 - 7 , 29, 31-38, 62, 107, 134, 190, 202, 276-77, 311 Collateral-benefits rule, 252 Collision cases, 77-79 Common law, 2, 7 - 8 , 23-24, 31, 52, 146, 191, 201, 265, 313 Comparative fault, 203 Comparative negligence, 27, 80-82, 202, 211-12, 314-15 Compensating differentials, 189, 309 Compensatory damages. See Damages Consent, 171-72 Consumer Product Commission, 280n Contract, 136, 142-43, 147, 206-7, 253, 278ff.; inducing breach of, 222-25 Contribution, 27, 191, 192n, 194-96, 19899, 203-4, 209, 219-22, 315; in maritime cases, 210-12; versus no contribution, 201-4

326 · Subject Index Contributory negligence, 27, 39-41, 50n, 69, 74-77, 88-91, 93, 118-20, 128, 141, 156, 159, 162, 165, 190n, 194, 199, 200, 233n, 239, 270n, 292-93, 309, 310; in catastrophic-injury cases, 271; in products liability cases, 299-300 Conversion, 48, 154, 160, 170. See also Theft Corrective justice, 14 Credence goods, 284 Criminal law, 154-55, 162n, 182n, 189, 301-2 Custom, 27, 107, 131-39, 314 Damages, 44n, 52, 55n, 65n, 74, 77-79, 137n, 154-56, 160, 165, 175, 185-89, 250, 302-7, 310; in catastrophic-injury cases, 257, 262-68, 272; in death cases, 27; periodic versus lump-sum, 46-47; in products liability cases, 295-96, 302-7. See also Deterrence; Punitive damages; Wrongful death Deceit, 166. See also Fraud; Misrepresentation Defamation, 163-66, 315 Defense of property, 175-81 Defenses, 65; to intentional torts, 171-81. See also names of particular defenses Degrees of negligence, 210 DES, 212-13, 256, 296, 307 Design defect, 283, 291-93, 303 Deterrence, 4 - 5 , 9-12, 13, 58, 156, 16162, 172n, 173n, 192, 209n, 254, 262, 264-65 Disclaimers, 27, 138-39, 281-84, 297, 311, 314 Discovery rule, 266-67 Distribution of income and wealth, 15-16, 18-19, 34-36, 40, 123n, 313 Due care. See Negligence Durable goods, 292, 307 Duty of care. See Negligence; Nondelegable duty of care Economic loss, 105, 251-55 Economics and economic models, 12-13, 23, 57 Efficiency, 7, 9, 13, 14, 16-17, 35-38, 4 0 41, 52, 157-58

Eggshell-skull rule, 137n, 169n, 186n, 24950 Eminent domain, 29-30 Emotional injury, 169-70, 243-45. See also Damages Employment. See Industrial-accident law; Respondeat superior Equity, 7n, 45, 103. See also Injunctions Error costs, 146, 166, 198, 307 Ex ante versus ex post, 193, 201-2, 235, 238, 263ff„ 268 Excessive-force rule, 175-78 Expected-utility analysis, 55-58 Experience goods, 284 Ex post. See Ex ante versus ex post Externality, 5n, 6, 13, 32, 36, 48-49, 69n, 105, 180-81, 251, 252, 308. See also Social cost Extortion, 34n Fairness, 9, 14, 19 Federal Employers Liability Act, 3, 15, 311 Fellow-servant rule, 309-10 Foreseeability, 149-50, 239, 244, 246-47, 300 Fraud, 302. See also Deceit; Misrepresentation General average, 211n Good Samaritan, 142ff., 179, 230. See also Rescue Gross negligence, 92, 159, 185, 302 Hand formula, 85-87, 88-89, 91, 96-103, 119, 188, 229-30, 233, 234, 248, 291, 299 Hypersensitive use, 50n Implied-contractual-immunity test, 206-7 Income, 54-55, 124n Incremental costs, 100 Indemnity, 191, 199-201, 205-10, 223; tests for, 205-9 Independent contractor, 205-8 Inducing breach of contract, 222-25 Industrial-accident law, 3n, 308-11 Industrial accidents, 3, 25n, 133, 137 Information costs, 65, 67, 71, 109-10, 11617, 126-31, 139, 203, 212, 269, 274, 28082, 284-85, 286-87, 290, 293-94, 295, 310-11

Subject Index · 327 Injunctions, 4 2 - 4 8 Insanity as tort defense, 127-28, 130, 174, 1 8 2 - 8 4 , 250 Insult, 161n Insurance, 4, 6, 1 0 - 1 1 , 13, 17, 22, 5 5 - 5 8 , 66, 82, 1 4 2 - 4 3 , 2 0 3 - 4 , 211, 253, 273; social, 103. See also Risk Intent, 149n, 164. See also Intentional torts Intentional torts, 149ff., 204, 2 0 8 - 9 , 302 Joint care, 60, 74, 1 9 0 - 9 1 , 200, 205, 209n, 210, 212, 214, 277 Joint tortfeasor, 27, 190n, 191, 193, 201n, 2 1 3 - 1 4 , 275n Jones Act, 3, 311 Judges, 19, 2 2 - 2 3 , 24, 57, 87, 98, 123n, 313 Juries, 302 Justice. See Corrective Justice; Distribution of income and wealth; Efficiency; Fairness

Necessity. See Private necessity; Public necessity Negligence, 3 9 - 4 1 , 61, 6 3 - 6 4 , 7 4 - 7 7 , 8 5 88, 104, 1 2 4 - 3 1 , 1 3 8 - 4 1 , 159, 162, 171, 191, 1 9 4 - 9 9 , 2 1 4 - 1 5 , 257, 274, 2 8 2 - 8 4 , 2 8 6 - 8 7 , 292; in catastrophic-injury cases, 262, 269; degrees of, 91; in Kalven casebook, 103-7; per se, 234, 2 4 8 49; and process failure, 293-94; in products cases, 2 7 8 - 8 0 , 300; versus strict liability, 2 - 4 , 17, 2 5 9 - 6 0 . See also Contributory negligence; Strict liability Negligent misrepresentation. See Misrepresentation No-fault automobile accident compensation laws, 11 Nondelegable duty of care, 208 Nuclear accidents, 242, 256, 261 Nuisance, 6, 25, 3 9 - 4 1 , 4 2 - 4 4 , 47, 59, 70, 113, 118, 150n, 167-68, 171, 272; coming to the, 5 0 - 5 1 ; permanent, 47; public versus private, 4 8 - 4 9 ; temporary, 47

Kaldor-Hicks efficiency, 1 6 - 1 9 Land occupier, tort duties of, 93n. See also Attractive nuisance Last clear chance, 90, 9 2 - 9 5 , 142, 146n, 210, 239n Legal-realist movement, 5 - 6 Liability, parental, 121; rules of, 18, 25, 3 0 - 3 1 , 3 6 - 3 8 , 4 2 - 4 8 , 6 2 - 7 7 , 143-44; vicarious, 121 Life expectancy, 267, 272 Limited liability, 273 Locality rule, 1 3 7 - 3 8 Locomotive-sparks cases, 111 Malpractice. See Medical malpractice Marginal utility of income, 56, 58 Maritime collisions, 2 1 0 - 1 2 Mass-disaster cases. See Catastrophic accidents Master and servant. See Respondeat superior Medical malpractice, 11, 58n, 106, 122, 131, 133, 137, 139, 181n; locality rule, 131 Mishandling defense, 299 Misrepresentation, 132; negligent, 166. See also Deceit; Fraud Monopoly, 32n

One-bite rule, 1 0 7 - 8 Opportunity cost, 171, 179, 254 Optimal care, 60. See also Negligence; Strict liability Pain and suffering, 186, 267 Pareto superiority, 13, 1 6 - 1 7 ; potential, 16 Perfect competition, 12 Pollution, 52, 105, 1 6 7 - 6 8 , 215, 252 Preexisting-condition cases, 250. See also Eggshell-skull rule Price-Anderson Act, 266n Prima facie case, 1 6 6 - 7 1 Primary vs. secondary liability, 207, 209 Private necessity, 43, 179-81 Privilege, 177-78 Privity of contract, 143, 2 8 4 - 8 5 , 2 8 8 - 9 1 , 310 Products liability, l l n , 27, 58n, 133, 1 3 9 41, 167n, 206, 209, 2 1 2 - 1 3 , 273ff.; assumption of risk, 298-300; to bystanders, 287-88; component defect, 294; and contract, 280ff.; contributory negligence, 299-300; damages, 2 9 5 - 9 6 , 3 0 2 - 7 ; design defect, 2 9 1 - 9 3 ; disclaimers, 2 8 1 - 8 4 ; durable goods, 307; effects of trade growth, 308; extremely careless con-

328

· Subject Index

Products liability (cont.) sumers, 300-1; foreseeable misuse defense, 300; historical trends, 284-88; information costs, 280-82, 284-85; inherently dangerous products 295-97; long-delayed consequences, 307; mishandling, 299; misuse by an intermediate, 301-2; process failure, 293-94; slightly dangerous products, 297; thirdparty fault, 298-302; ultrahazardous activity, 299; warnings, 296-97; worn-out products, 298 Property. See Defense of Property Property rights, 25, 29-38, 42-48 Proximate cause, 191, 246-47 Public good, 16, 180 Public necessity, 180-81 Punitive damages, 30, 48, 160-65, 167n, 184-85, 187n, 189, 223-24, 251, 302, 315; in products liability cases, 304-7

Rape, 157-58 Reasonable force, 173-74 Reasonable-man rule, 73, 123-32, 140, 162, 197-98 Reasonable use, 41, 45-46, 49, 50n. See also Negligence Recklessness, 159, 162, 164-65, 182n, 18485 Redistribution of income and wealth. See Distribution of income and wealth Regulation of safety and health, 53n, 256 Rescue, 71, 100-2, 143-48, 214, 217n, 250-51, 265. See also General average; Good Samaritan Res ipsa loquitur, 123n Respondeat superior, 5n, 20, 73, 120-21, 190, 198, 207-8, 214, 223, 309 Restatement of Torts, 20, 49n, 70, 88n, 11112, 116, 119n, 176 Retribution, 14, 177-78 Risk: general attitudes toward, 55-58, 140, 282, 314; risk aversion, 56-57, 82, 204, 211, 273-74, 303; risk neutrality, 273 Search goods, 284 Self-defense, 150, 152, 172-75 Sequential torts, 76n, 82, 92-93 Settlement of litigation, 202-3

Simultaneous joint torts, 190-92; joint care, 193-98 Social cost, 6 - 8 , 45n, 58-62, 68, 73-74, 252, 254. See also Externality Social welfare, 58-59 Spring guns, 176-78 Statutes, 248-49, 311; economic theory of, 219-22 Statutes of limitations, 266, 269, 286, 307 Stochastic care, 72-73, 76, 94-95, 197-98, 214, 236 Strategic behavior, 199-200. See also Extortion Strict liability, 30, 38, 39-41, 51, 61, 63, 107-22, 128-29, 137, 163-65, 170-71, 179-80, 198, 214-15, 273, 282-84, 28687, 309-10; for catastrophic injuries, 257, 259-60, 269; of children, 129-30; contributory negligence, defenses, 7 9 80; disclaimers, 282-83; highly dangerous products, 297-98; and negligence compared, 2 - 4 , 17, 64-73, 124-26, 25960; and process failure, 293-94; and products liability, 274, 277-78 Successive joint torts, 190-92, 215-19 Theft, 156, 160. See also Conversion Third-party beneficiary doctrine, 253 Tort law, 1 - 3 , 23-24; accident cases, 2-3; constitutional torts, 3; defenses, 3; economic analysis of, 6-8; history of, 1 - 3 , 65-66, 71, 116, 186-87, 284-86, 308-11; immunities to, 3; no-fault compensation versus strict liability, 5-6; positive economic theory of, 1, 12, 16, 24, 31, 4 3 44, 45, 82, 104, 116, 312ff.; positive noneconomic theory of, 252-55; as social insurance, 5 Torts. See Intentional torts; names of particular torts Transferred intent, 181-82 Trespass, 42-44, 59, 96, 102n, 113, 17071, 179, 181n, 239; by cattle, 108 Ultrahazardous activities, 51, 111-20, 129, 236-37, 298, 299 Uniform Commercial Code, 281n Utilitarianism, 4 - 5 , 18, 156 Utility, 54-58, 187-88

Subject Index Value, 157-58. See also Efficiency Vicarious liability. See Respondeat superior Warranty, 286n Wealth, 54-55; maximization of, 16ff. See also Efficiency

·

329

Workers' (formerly workmen's) compensation, 3, 25n, 224, 310-11 Writ of trespass, 2 Wrongful death, 27-28, 186-89, 31415