Forthcoming: 72 Fordham L. Rev. 1857 (2004)

RAWLSIAN FAIRNESS AND REGIME CHOICE IN THE LAW OF ACCIDENTS
Gregory C. Keating* The political philosophy of John Rawls is pregnant with implications for the tort theory. Our law of intentional and accidental physical injury is rich with the rhetoric of reasonableness and fairness, and these ideals lie at the heart of Rawls’s political philosophy. The figure of the reasonable person is central both to the law of negligence—where it serves as the master criterion of justified risk imposition—and to the law of intentional torts—where it helps to define the contours of permissible self-defense, the sensibility by which the offensiveness of contact in battery is measured, and the content of the consent given in connection with matters as diverse as The concept of contact sports and medical operations.1 reasonableness figures prominently in strict liability as well. The intentional infliction of unreasonable harm triggers liability for damages in the law of nuisance, and strict liability in general can be fruitfully understood as a form of liability applicable when the conduct which leads to accidental injury is reasonable, but the failure to make reparation for the harm done is unreasonable.2 Principles of fairness figure more prominently in the judicial rhetoric of strict products liability than economic ideas of efficient precaution and efficient insurance do.3
* William T. Dalessi Professor of Law, USC Law School. For instruction and advice, I am grateful to Ken Abraham, Scott Altman, Charles Fried, Richard Fallon, Louis Kaplow, Scott Michelman, Lewis Sargentich, Arthur Ripstein, and Ben Zipursky; to the participants at the conference; and to the participants at a faculty workshop at Harvard Law School. Special thanks are owed to Jim Fleming for organizing the conference and to Ben Zipursky for organizing the torts panel. 1. On reasonableness in negligence law, see Gregory C. Keating, Reasonableness and Rationality in Negligence Theory, 48 Stan. L. Rev. 311 (1996). On the various roles of reasonableness in intentional tort law, see the materials in Robert E. Keeton et al., Tort and Accident Law 30-93 (4th ed. 2004) [hereinafter Tort and Accident Law]. 2. On nuisance, see Tort and Accident Law, supra note 1, at 123-36, 897-917. On strict liability, see id. at 283-84, 294-307; Robert E. Keeton, Conditional Fault in the Law of Torts, 72 Harv. L. Rev. 401 (1959). 3. See James A. Henderson, Jr., Judicial Reliance on Public Policy: An Empirical Analysis of Products Liability Decisions, 59 Geo. Wash. L. Rev. 1570, 1589, 1597 (1991) (noting that “[m]easured by what judges say in their published opinions . . .

1857

1858

FORDHAM LAW REVIEW

[Vol. 72

For the last thirty or so years, however, normatively inclined academic discourse on the law of accidents has been carried on largely in the idioms of efficiency and corrective justice. Powerful and illuminating as much of this work has been, it has tended to obscure the prominence of conceptions of reasonableness and fairness in our law of accidents. John Rawls’s great work provides us with the tools to restore those ideals to an equally prominent place in normative discourse about the law of accidents. My aim in this paper is to put Rawls’s philosophy to work by examining the choice between negligence and strict liability from a fairness perspective. I hope to show both the power and fertility of Rawls’s ideas, and the robustness of fairness concerns in our law of accidents. My particular topic is one that has been touched on before from a Rawlsian perspective. Early in the 1970s, as Rawls’s theory was first bursting upon the legal academy, George Fletcher wrote a celebrated article, Fairness and Utility in Tort Theory.4 Fletcher’s remarkable article connected Rawls’s work with reciprocity (and nonreciprocity) of risk imposition. Fletcher argued that nonreciprocity of risk both characterized realms of strict liability within tort accident law and justified those realms, whereas reciprocity of risk characterized and justified realms of tort accident law that were governed by negligence liability.5 In this Article, I argue against Fletcher’s identification of fairness in the choice between negligence and strict liability with the presence or absence of reciprocity of risk, and in favor of focusing on the fair distribution of the costs of accidental injury among those who benefit from the imposition of the underlying risks. I argue, further, that a distinctively Rawlsian conception of fairness lends support to a powerful general case for preferring strict enterprise liability to negligence liability. The allure of reciprocity of risk as a master criterion of fairness is evident. When risks are reciprocal, they are equal in probability and magnitude and are imposed for equally good reason. The right to impose a risk enhances the freedom of potential injurers and the exercise of that right endangers the security of potential victims. fairness norms, not efficiency norms” predominate, and their predominance increases when they conflict with efficiency rationales). 4. George P. Fletcher, Fairness and Utility in Tort Theory, 85 Harv. L. Rev. 537 (1972). Charles Fried’s An Anatomy of Values 183-206 (1970), also connected Rawls’s political philosophy with reciprocity of risk imposition, but focused less on the choice between negligence and strict liability. The emphasis on reciprocity of risk in tort theory has a long history, running back through the work of Francis H. Bohlen in the early twentieth century, The Rule in Rylands v. Fletcher, 59 U. Pa. L. Rev. 298, 373, 423 (1911) (pts. 1-3), to the opinion of Lord Cairns in Rylands v. Fletcher, 3 L.R.E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.), and the opinion in Losee v. Buchanan, 51 N.Y. 476 (1873). 5. See generally Fletcher, supra note 4.

2004]

THE LAW OF ACCIDENTS

1859

Reciprocity of risk defines a community of equal freedom and mutual benefit. Reciprocity of risk defines a community of equal freedom, because reciprocity exists when risks are equal in probability and gravity. When risks are equal in these respects, people relinquish equal amounts of security and gain equal amounts of liberty. Reciprocity of reasonable risk defines a situation of mutual benefit because risks are reasonable when the benefits of imposing them are greater than the burdens of bearing them. Reciprocity of risk thus defines a fair situation with respect to the distribution of risk. When risks are not reciprocal, risk is not fairly distributed. By prescribing payment for harm done, strict liability redresses, ex post, the ex ante unfairness of nonreciprocal risk. When risks are reciprocal, strict liability is superfluous. When risks are reciprocal, “strict liability does no more than substitute one form of risk for another—the risk of liability for the risk of personal loss.”6 This, in a nutshell, is Fletcher’s argument. Fletcher’s argument is elegant and powerful, but its preoccupation with risk of physical injury—rather than with physical injury itself—is troubling. With rare exceptions, risk of physical injury is cause for concern only because it occasionally erupts into actual injury. Physical injury is what devastates and destroys lives. Physical injury is what gives the law of accidents its moral urgency. Reciprocity of risk defines a circumstance where the burdens and benefits of risk are proportional and to everyone’s benefit. It is more important, however, to make the burdens and benefits of harm—of accidental physical injury—proportional. It is more important to distribute the costs of accidents fairly. Harm itself is distributed fairly only when harm—not risk—is reciprocal. Reciprocity of harm, however, is only found in the law of nuisance, and even then only in the case of the mutual, low level interferences with each other’s use and enjoyment of property that are the subject of the “live and let live” rule of nuisance law.7 Accidental physical injury, however, is rarely reciprocal, and fortunately so. Reasonable risk impositions only occasionally result in accidental physical injury. Harm, therefore, befalls only a few of those exposed to reciprocal risk. In accident law, the alternative to the fair distribution of risk is not the fair distribution of harm, but the fair distribution of the costs of accidents across those who benefit from the imposition of the relevant risks. In accident law, the alternative to the reciprocity of risk criterion is the enterprise liability version of strict
6. Id. at 547. 7. The “live and let live” rule is usually traced to Baron Bramwell’s opinion in Bamford v. Turnley, 122 Eng. Rep. 27 (Ex. Ch. 1862). The opinion gives the following nineteenth century list of examples subject to the “live and let live” rule of no liability for low-level nuisances: “burning weeds, emptying cess-pools, making noises during repairs.” Id. at 32; see also infra note 55 and accompanying text.

1860

FORDHAM LAW REVIEW

[Vol. 72

liability. Enterprise liability in tort pins the costs of accidents— negligent and nonnegligent alike—on the enterprises or activities responsible for them.8 By doing so, enterprise liability distributes the costs of concentrated accidental injuries among participants in the enterprise who benefit from its risk impositions. It makes a difference, I believe, whether ideas of fairness (Rawlsian or otherwise) are best expressed by the idea of reciprocity of risk or by enterprise liability. Identifying fairness with reciprocity of risk leads to the view that fairness ideas in tort find their fullest expression in a common law regime which resembles the common law of accidents at the turn of the twentieth century.9 Identifying fairness with reciprocity of risk leads to an implicit defense of the common law as it was a century ago—to a kind of nostalgia. Identifying fairness with the fair distribution of the costs of accidents leads to a very different view of the proper shape of the law of accidents. Identifying fairness with the fair distribution of harm leads—presumptively—to favoring the expansion of enterprise liability, both within and beyond the tort law of accidents. It leads to seeing a wide variety of administrative schemes, including workers’ compensation, no-fault automobile insurance, industry-wide liability for black lung disease, and even the society-wide liability of the New Zealand Accident Compensation scheme, as continuous with the tort law of accidents.10 Identifying fairness with enterprise liability leads to an agenda for progress and reform. Part I of this Article sketches the core elements of a fairness conception of accidental risk imposition, drawing in some detail on Rawlsian ideas of reasonableness, interpersonal comparison, and fair social cooperation. Part II reconstructs the reciprocity of risk criterion as a way of bringing that conception to bear on the choice between negligence and strict liability in tort law, and then argues against the reciprocity of risk criterion as the principal ground for choosing between these doctrines.11 The latter half of Part II argues
8. See generally Ira S. Bushey & Sons, Inc. v. United States, 398 F.2d 167 (2d Cir. 1968). 9. See Fletcher, supra note 4, at 543-51. 10. Fletcher sees enterprise liability as the expression of loss-spreading ideas entirely independent of fairness, and indeed opposed to fairness. See id. Fletcher describes “loss-shifting in products-liability cases” as “a mechanism of insurance.” Id. at 544 n.24. Fletcher describes insurance arguments for the imposition of strict liability as arguments of “distributive rather than corrective justice.” Id. at 547 n.40. Imposing liability on insurance or loss-spreading grounds “violates the premise of corrective justice, namely that liability should turn on what the defendant has done, rather than on who he is.” Id. Many legal scholars likewise see administrative accident schemes, and even enterprise liability within tort, as animated by ideas which are foreign to the core of tort law. See, e.g., Jules L. Coleman, Risks and Wrongs 395406 (1992); Ernest J. Weinrib, The Idea of Private Law 171-204 (1995). 11. My reconstruction is meant to put the best face on the criterion as a specification of fairness in tort. Fletcher might not accept this reconstruction.

2004]

THE LAW OF ACCIDENTS

1861

that fairness theories should be concerned more with harm than with risk, and more with the distribution of the costs of accidental harm than with the distribution of risk. Irrespective of the initial distribution of risk, it is fairer to distribute the costs of accidents across those who benefit from the imposition of the relevant risks than it is to leave those costs concentrated on random victims. Part III examines how this idea—the idea that harms ought not be concentrated on victims and ought to be dispersed across those who benefit from the risk impositions responsible for those harms—finds expression in enterprise liability. A Rawlsian conception of fairness can and does, I argue, lead us to favor the enterprise liability principle of fairness over the reciprocity of risk criterion, but it does not determine the content of the enterprise liability conception of fairness itself. Part III lays out the understanding of fairness that enterprise liability embodies, and argues that the enterprise liability understanding is instituted in a variety of ways, inside and outside the law of torts. The section places particular emphasis on the expression of enterprise liability ideas by a variety of nonfault administrative plans, ranging in breadth from plans covering particular kinds of injuries—vaccination related health injuries, for example—through plans covering particular kinds of activities—e.g., workplace accidents or automobile related ones—to plans whose reach is society-wide— such as the New Zealand Accident Compensation Plan. No-fault administrative plans illuminate several essential elements of enterprise liability. For one thing, administrative plans are able to implement enterprise liability ideas of fairness in circumstances where tort accident law cannot. No-fault automobile insurance, for example, is able to implement enterprise liability in a setting where the common law is unable to do so, because the common law cannot devise nonfault criteria for sorting injurers and victims in the case of automobile related accidents.12 No-fault administrative plans also illuminate two essential aspects of the idea of fairness at work in enterprise liability. First, the implementation of enterprise liability by these plans vividly illustrates the attenuation of causation implicit in the idea that the costs of the accidents characteristic of an activity should be shared by all those who benefit from that activity. The enterprise liability principle of fairness retains the traditional tort requirement of harm as a condition of liability, but relaxes or attenuates the traditional tort focus on causation, because it holds that accident costs should be dispersed across all those who benefit from the underlying risks. Indeed, the relaxation of causation by these plans is one of the reasons that they can realize enterprise liability
12. As Baron Bramwell put it long ago, “Where two carriages come in collision, if there is no negligence in either it is as much the act of the one driver as of the other that they meet.” Fletcher v. Rylands, 159 Eng. Rep. 737, 744 (Ex. 1865) (Bramwell, J., dissenting), aff’d 3 L.R.-E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.).

1862

FORDHAM LAW REVIEW

[Vol. 72

ideals in circumstances where the common law cannot. Second, no-fault administrative accident plans illustrate the elasticity of the idea of “enterprise.” Enterprise liability blossoms in ever-widening circles, expanding from the common law liability of particular firms through the liability of entire industries, to societywide liability. The fact that it does teaches us much about the form of liability. The benefits of risky activities radiate outward in concentric circles. Nuclear power most benefits those who produce and consume it, but it also benefits those of us who merely happen to live in a society made richer by the presence of nuclear power plants. Generally speaking, the benefits of most risky activities radiate outward until they diminish to the point of being unidentifiable. Specifying the relevant community of benefit is a standing task for enterprise liability schemes; a task whose performance requires the exercise of essentially contestable, normative, political judgment. Part IV examines how even a system of tort accident law animated by a firm commitment to ideas of fairness which find their natural expression in enterprise liability would nonetheless cede substantial chunks of the tort law of accidents to negligence, and even to no liability at all. The barriers to complete common law enterprise liability examined in this section are partly the flip side of the advantages of the administrative schemes studied in Part III, partly the product of other practical problems, and partly the consequence of competing normative considerations. I. A FAIRNESS FRAMEWORK A. The Contours of Fairness The fairness conception that I shall sketch has a number of elements. One of these is a conception of persons. It supposes that we are each equal, independent persons; self-governing agents with purposes to pursue and lives to lead.13 We each have the capacity to lead our lives in accordance with some conception of their point, and a deep interest in living under institutions that enable us to do so. To make our lives answer to our aspirations, we need, among other things, a substantial measure of security—of freedom from accidental injury and death at the hands of others. “Security,” John Stuart Mill remarked: no human being can possibly do without; on it we depend for all our immunity from evil and for the whole value of all and every good, beyond the passing moment; since nothing but the

13. This is a commitment that the fairness conception shares with more libertarian ones. See, e.g., Charles Fried, Right and Wrong 28-29 (1978) (“Respect for Persons”).

2004]

THE LAW OF ACCIDENTS

1863

gratification of the instant could be of any worth to us, if we could be deprived of everything the next instant.14 Our need for security, however, is only half the story. We also need a substantial measure of liberty—of freedom to put others at risk of physical harm in pursuit of our own ends. When we act we put others at peril, even if only very slightly, and even when we act with appropriate caution. If we are not permitted to imperil others—if we cannot endanger their security—we cannot act and so cannot pursue our ends and lead our lives. Maximal security extinguishes liberty and maximal liberty devastates security. Yet substantial measures of each are essential preconditions of effective agency. This is the dilemma at the heart of accident law. When the law of accidents licenses the imposition of a risk, it enhances the freedom of some and imperils the security of others. Those who impose the risk are set free to pursue ends and activities that they value, and their pursuit exposes others to risks of physical harm. When the law of accidents forbids the imposition of some risk, it does the reverse—it curbs the freedom of prospective injurers and enhances the security of potential victims. Risk impositions thus pit the liberty of injurers against the security of victims, and the law of accidents sets the terms on which these competing freedoms are reconciled. The task of the tort law of accidents is to reconcile liberty and security on terms that are both favorable and fair. Favorable terms enable people to pursue their aims and aspirations over the course of complete lives; fair terms reconcile the competing claims of liberty and security in ways that are acceptable even to those they disadvantage.15 The question of how best to reconcile the pursuit of activities we value with the physical and psychological integrity that those activities can jeopardize is, of course, an issue that each of us must face individually. What ends are worth the risks they entail? Are the risks
14. John Stuart Mill, Utilitarianism (1863), reprinted in Utilitarianism and On Liberty 181, 226 (Mary Warnock ed., 2d ed. 2003). 15. This conception of the problem of accidental harm has its roots in the social contract tradition in political theory, especially as carried on by contemporary philosophers such as John Rawls, Tim Scanlon, and Thomas Nagel. “Liberty” and “security” in the sense used here, however, do not identify “primary goods” lexically superior to income and wealth in the manner of the liberties covered by Rawls’s first principle of justice. “Liberty” and “security” are general cover terms designed to characterize, at a fairly high level of generality, the stakes in accidental risk imposition. The burdens and benefits of risk include increases and losses in wealth and income, so there is no question of these freedoms being lexically prior to the primary goods of wealth and income. Thus, in judging the reasonableness of various risk impositions or liability rules, we should assess the significance of gains and losses in wealth and income in terms of their impacts on liberty and security. Why characterize the interests at stake in risk impositions as interests in freedom at all? Because risks and risk reduction affect the space that we have to form, evaluate, and act upon our aims and aspirations.

1864

FORDHAM LAW REVIEW

[Vol. 72

of death and disfigurement that are the price of scaling Mount Everest worth the sense of accomplishment that comes from standing on its summit? Are increased risks of cancer worth bearing as the price of performing pathbreaking medical research? Are increased risks of cancer worth bearing as the price of earning an ordinary paycheck? This kind of individual choice is not, however, the chief concern of the law of accidents. The problem of accidental harm is a problem of social choice—of how best to reconcile the competing claims of liberty and security for a plurality of persons. More fully, the problem of accidental harm requires reconciling the competing claims of liberty and security for a plurality of persons, each of whom is independent, all of whom are equal, and among whom diverse and incommensurable conceptions of the good flourish. Because people are equal and independent, the terms of accidental risk imposition must be ones that equal people might freely accept as legitimate for the governance of their lives in common. Diverse and incommensurable ends and aspirations flourish among free people because the range of valuable activities and valuable ways of life is diverse. Because people have distinct lives to lead, and because their aims and ends diverge, the principles of social choice differ markedly from the principles of individual choice. Individually, it may be rational to expose ourselves to risks that it would be unreasonable— unfair—to impose on others. 1. Rationality and Reasonableness The distinction between reasonableness and rationality is one drawn by ordinary discourse.16 Rationality requires the intelligent pursuit of one’s ends, whatever those ends are. Reasonableness requires taking the impact of one’s conduct on other people into account as a circumstance capable of influencing one’s decisions, and being prepared to govern one’s conduct on a basis acceptable to others whom one’s conduct affects. Rationality and reasonableness may well diverge. It may be perfectly rational for the owner of a dock to demand exorbitant compensation for permitting a ship to tie up at the dock during a gale.17 The gale gives the dock owner a very strong hand to play; playing that hand to the hilt is eminently rational. It may be equally rational for the ship owner—faced, otherwise, with the loss of his ship—to pay exorbitant compensation. The gale gives the
16. See W.M. Sibley, The Rational Versus the Reasonable, 62 Phil. Rev. 554 (1953). Sibley’s description of rationality, which the text follows, is a basic and familiar one, but probably not the only way of specifying the concept, even at a very general level. 17. The circumstance (though not, so far as I know, the bargaining) arose in Vincent v. Lake Erie Transportation Co., 124 N.W. 221 (Minn. 1910). In this circumstance, the doctrine of private necessity applies. The doctrine extinguishes the property right to exclude, and entitles the dock owner only to compensation for harm done.

2004]

THE LAW OF ACCIDENTS

1865

ship owner a correspondingly poor hand to play. But it is also unreasonable for the dock owner to insist on such exorbitant compensation. Reasonable terms of cooperation are terms that the parties to them would regard as fair, were they to find themselves in each other’s shoes. Judgments of reasonableness abstract from inequalities of bargaining power. Reasonable terms of cooperation— fair terms of cooperation—are terms that parties would be prepared to accept absent the coercion of circumstance or, indeed, absent any coercion at all. Reasonable terms of cooperation attract unforced agreement. A hard bargain stuck by a dock owner at the expense of a ship owner whose ship faces all but certain destruction at the hands of a gale unless the dock owner grants permission to dock, is not an unforced agreement. It is the very force of unfavorable circumstance that the dock owner seeks to exploit and which gives the ship owner reason to acquiesce. Driving a hard bargain is rational—but unreasonable—behavior on the dock owner’s part. Drawing on the distinction between rationality and reasonableness, it makes sense to say that risk impositions may be at once rational and unreasonable. We may say, for example, that the rationality of exposing oneself to a risk depends on the end furthered by the exposure, the importance that one attaches to furthering that end, and the efficacy with which the exposure will further the end. The canons of rationality thus give wide rein to individual subjectivity, and are naturally expressed in the language of efficiency. Individuals are free to value the burdens and benefits of risks by any metric they choose, and it is surely natural for them to value burdens and benefits by their own subjective criteria of well-being. It is also rational for individuals to run risks whenever, by their own lights, the expected benefits of so doing exceed the expected costs, and to decline to run risks whenever the expected costs exceed the benefits. The rationality of a risk imposition is not, however, enough to guarantee its reasonableness. It is not necessarily reasonable for people to expose others to risks because—by the potential injurer’s own evaluation of the end furthered by the risk imposition—the benefits of imposing the risk exceed the burdens of having to bear exposure to it. Rational risk impositions are not necessarily reasonable ones because other people have different values and distinct lives. The distinct lives of different people cannot be collapsed into a single life that reaps both the burdens and the benefits of rational risk impositions. Some people will die at the hands of risk impositions whose benefits will accrue to others. In a world of distinct persons who affirm diverse and incommensurable conceptions of the good, there is, moreover, no reason to assume that potential victims of any given risk imposition value the ends pursued through that risk imposition in the same way as the potential injurer imposing the risks does. The fact that you may be prepared to run

1866

FORDHAM LAW REVIEW

[Vol. 72

enormous risks for the advancement of medical knowledge does not mean that I am prepared to do so. The fact that I might be willing to suffer the risks of hurtling down the road at 100 miles per hour does not mean that you are willing to suffer the risks of my doing so. The diverse aims of a plurality of persons cannot be converted into a single scale, so that we may make collectively the same kinds of judgments that we each make individually. Because lives are distinct and values diverse, “sacrificing another for one’s own purposes must be viewed as different from sacrificing oneself.”18 2. Rawlsian Reasonableness The general idea of reasonableness on which we have drawn suggests that risks must be imposed on terms which are acceptable to a plurality of persons with distinct lives and diverse ends and preferences. Political philosophy in a Rawlsian vein19 elaborates on the general idea of reasonableness in a number of ways. First, it supposes that reasonableness and rationality are complementary notions. Reasonable people have diverse ends—diverse conceptions of their own rational advantage; diverse preferences, ends, and aspirations. Second, political philosophy in a Rawlsian vein supposes that reasonable people share the common aim of reaching unforced agreement on fair terms of cooperation. The interactions of reasonable people are thus different from the interactions of purely rational actors. Rational egoists interacting with other rational egoists seek to determine the course of action which will best advance their own interests, given the existence of other rational agents seeking to advance their own interests as best they can. They seek to determine the course of action most likely to maximize their own advantage. Rational agents tend to cooperate on terms that are mutually advantageous—on terms that make everyone better off with respect to their preexisting situation, in the way that Pareto-superior transactions do. Reasonable people, by contrast, do not interact with other reasonable people by seeking their own greatest advantage. Reasonable people seek to cooperate on fair terms with other equal, independent, and reasonable people. Fair terms do not necessarily advantage everyone they affect in the sense of making everyone they affect better off than they were under the preexisting distribution of advantages. Measured against the baseline of preexisting entitlements, the move from an unfair situation to a more fair one does not improve the situation of those who profited from the preexisting unfairness. The enfranchisement of some previously disenfranchised group on the ground that fairness
18. Fried, supra note 4, at 191. 19. I am drawing here principally on the discussions of reasonableness in John Rawls, Political Liberalism 48-66 (1996).

2004]

THE LAW OF ACCIDENTS

1867

requires it, for example, does not improve the lot of those who benefited from the subordination and disenfranchisement of the group. Reasonableness is thus linked to an idea of mutual benefit, but not to the idea of mutual advantage against a preexisting baseline of entitlements. Reasonableness is linked to reciprocity—to terms that define appropriate terms of cooperation among equals. Among equals, fair terms of cooperation are determined not by comparing advantage and disadvantage against the baseline of preexisting entitlements, but by comparing burdens and benefits to those affected under alternative possible principles of cooperation. The reasonableness of preferring a regime of strict liability to one of negligence, for example, depends on comparing burdens to victims under negligence to burdens to injurers under strict liability—not on comparing the advantages to some and the disadvantages to others of moving from a regime of negligence to one of strict liability. 3. Interpersonal Comparison Because the reasonableness of possible terms of cooperation turns on their distribution of the burdens and benefits of cooperation, some criteria for comparing burdens and benefits are necessary. This is a formidable challenge in its own right. When people affirm diverse and incommensurable ends, the criteria of interpersonal comparison, by which burdens and benefits are measured, must be “mutually acceptable to people whose preferences diverge.”20 The overlap in people’s needs (in contrast to the divergence in their aspirations, preferences, and wants) makes this possible. People whose preferences diverge may still need many of the same things: liberty, security, health, income, and wealth, for example. The ambition behind Rawls’s “primary goods” is to identify the institutional conditions (equal liberties) and “all purpose goods” that people need to pursue their diverse ends and aspirations, and to make these goods and institutional conditions the basis of interpersonal comparison of well-being.21 Freedom of action and security are “institutional conditions” akin to the basic liberties so far as accidental risk imposition is concerned. Freedom of action and security are conditions on whose importance people with diverging preferences

20. T. M. Scanlon, Preference and Urgency, 72 J. Phil. 655, 668 (1975). 21. In the parlance of the interpersonal comparison literature, these are generally described as “objective” criteria of interpersonal comparison, in contradistinction to “subjective” ones. “Subjective” criteria of interpersonal comparison evaluate “the level of well-being enjoyed by a person in given material circumstances or the importance for that person of a given benefit or sacrifice . . . solely from the point of view of that person’s tastes and interests.” Id. at 656. Objective criteria appraise burdens and benefits in terms that are “the best available standard of justification . . . mutually acceptable to people whose [aims, ends, and] preferences diverge.” Id. at 668.

1868

FORDHAM LAW REVIEW

[Vol. 72

can agree. Their importance does not depend on affirming any particular preferences or on holding any particular set of ends and aspirations. Their importance depends on having ends and aspirations, and on having a fundamental interest in being able to realize those ends and aspirations over the course of a normal life. Freedom and security are essential conditions for the pursuit of most of the ends human beings hold, especially when we think of pursuing ends over the course of a complete life. Risks of physical harm materialize into physical harm, and physical harm can end the pursuit of one’s aims and objectives entirely. Even when physical harm does not lead to death, it can profoundly disrupt the pursuit of ends and aspirations, rendering the realization of some ends impossible and severely impeding the pursuit of other ends. Conversely, freedom of action is prima facie enabling of the pursuit of one’s ends, whatever they are. Being forbidden to act at all—because the risk of physical injury to others is too great—would cripple the pursuit of almost any end. Being forbidden to act in certain ways—because those ways endanger other people too much—tends to impede the pursuit of at least some ends. The reasonableness of risk impositions thus turns on the way that the impositions reconcile the competing claims of liberty and security. Risk impositions are reasonable when the freedom to impose a risk is more valuable than the foregone security that is the necessary price of that risk imposition. They are unreasonable when the security lost is more valuable than the freedom of action potentially gained. More concrete categories are necessary when we make the judgments about the reasonableness of particular risks and precautions, as we do in negligence law, for example.22 The question before us, however, is different. We are concerned with the reasonableness of choosing negligence over strict liability, and vice-versa. In this context, the relatively abstract account of the interests at stake given by the terms “liberty” and “security” seems sufficient. The choice between negligence and strict liability is a highly general one, and it is natural to think about it in highly general terms. 4. Reasonable Regime Choice and the Original Position Within legal scholarship, Rawls’s work is associated above all with the device of the “original position.”23 The framework that I have sketched omits that device entirely. My reasons for this are threefold. First, the enterprise on which I am embarked is an exercise in non22. For discussion of how the common law of negligence proceeds in this regard, see Fried, supra note 4, at 191-93. See generally Keating, supra note 1. 23. The device of the original position has been seized on by scholars who borrow nothing else from Rawls. See, e.g., Ronald J. Gilson, The Devolution of the Legal Profession: A Demand Side Perspective, 49 Md. L. Rev. 869, 875-77 (1990) (invoking the idea of decision behind a “veil of ignorance”).

2004]

THE LAW OF ACCIDENTS

1869

ideal theory—an essentially Dworkinian endeavor to determine whether expanding the domain of strict enterprise liability and proportionately shrinking the domain of negligence liability makes our existing law of accidents the fairest law that it might be. We are looking for an understanding of fairness that speaks to the distribution of the costs of accidents and tells us how we might best understand and reshape the accident law that we have. Our inquiry is therefore both into and constrained by the two fundamental principles of responsibility for harm done that our law institutes. We are not engaged in an exercise in ideal theory—we are not beginning with a blank slate, setting out to determine the ideal law of accidents. Rawls’s theory of justice, by contrast, is an exercise in ideal theory— an attempt to work out the best conception of justice for persons who are free and equal, and who accept various conditions (e.g., the circumstances of justice) as constraints on the conception that they might choose. Direct application of Rawls’s theory is therefore inappropriate. Second, I believe that if we were to extend the enterprise of A Theory of Justice and inquire—at the legislative stage—into the accident law regime that a just society would adopt, we would discover that Rawls’s principles of justice leave a great deal of latitude in the choice of such regimes, and that the choice of the best regime would depend heavily on the particular historical traditions and present conditions of the society. A theory that leaves open the choice between capitalism and market socialism surely leaves ample room for accident law schemes ranging from the common law of torts through New Zealand style social insurance and direct regulation of risk. Third, the device of the original position bears an ambiguous relation to the central ideas of Rawls’s theory. On the one hand, it illuminates the relation between A Theory of Justice and the social contract tradition. It carries on the social contract tradition because it models the ideal of an ideal, unforced agreement. On the other hand, the device of the original position can also obscure the distance between the idea of reasonable agreement—which is at the heart of Rawls’s view—and the idea of rational choice found in economics and decision theory.24 The device must therefore be used with care and precision, as its loose appropriation in support of instrumental and economic views of tort law shows.25 It seems wise to avoid this danger
24. See Burton Dreben, On Rawls and Political Liberalism, in The Cambridge Companion to Rawls 316 (Samuel Freeman ed., 2003); T.M. Scanlon, Contractualism and Utilitarianism, in Utilitarianism and Beyond 103 (Amartya Sen & Bernard Williams eds., 1982). 25. Charles Fried & David Rosenberg, Making Tort Law: What Should Be Done and Who Should Do It 13-36 (2003), and Louis Kaplow & Steven Shavell, Fairness Versus Welfare 437-43 (2002), both deploy the idea of ex ante choice to support

1870

FORDHAM LAW REVIEW

[Vol. 72

at the outset by steering clear of the original position and emphasizing the distinctive ideas of reasonableness, fair social cooperation, and the independence and equality of persons that characterize not just Rawls’s theory but the school of moral and political philosophy of which it is part. It seems best to bring these ideas to bear by allowing them to saturate our analysis of the fair distribution of accident costs and letting the rich conceptions of fairness and reasonableness found in our law of accidents react upon and reshape the philosophical ideas that we are bringing to bear.26 We shall, therefore, bring Rawlsian ideas to bear directly, using them to enrich our understanding of the ideals of fairness extant in our law of accidents, and asking more informally when reasonable people—concerned with distributing the costs of risky but mutually beneficial activities fairly—would agree to a regime of negligence liability and when they would agree to a regime of strict enterprise liability. Inquiring more informally involves comparing the distribution of burdens and benefits between injurers and victims as classes, and considering the reasons that they might advance in favor of these competing distributions.27 B. Fairness and Choice of a Tort Regime The tort law of accidents is now and long has been divided between realms of negligence and realms of strict liability. The choice between these rival principles of responsibility for harm done is, arguably, the most fundamental choice in the law of torts. Particular conceptions of liability—such as the enterprise liability conception of strict liability— articulate these regimes in specific ways. But a general characterization of the difference between these two principles of responsibility can still be given. The essential distinction (in my view28) is that negligence liability criticizes conduct, whereas strict highly economic interpretations of tort accident law, and invoke Rawls in support of choosing ex ante. 26. At least one interpreter of Rawls has noted that the common law itself is a rich source of ideas of reasonableness. See Dreben, supra note 24, at 316. 27. This procedure is similar to the procedure followed by the more general contractualism of Tim Scanlon. See Thomas M. Scanlon, What We Owe to Each Other 189-247 (1998). 28. Although the division of the law of accidents between realms of negligence and realms of strict liability is a longstanding one, how best to understand the difference between these competing principles of responsibility remains a contested matter among tort scholars. There are, for example, scholars who see strict liability as essentially a surrogate kind of negligence liability, and scholars who see it as a distinct and competitive form of liability. Jules Coleman and Richard Posner are among those who see strict liability as an instrument for the realization of negligence aims. See Coleman, supra note 10, at 226-28 (1992). Judge Posner’s opinion in Konradi v. United States, 919 F.2d 1207 (7th Cir. 1990), and Indiana Harbor Belt R.R. Co. v. American Cyanamid Co., 916 F.2d 1174 (7th Cir. 1990), illustrate this position as well. Posner first articulated this view of strict liability as a servant of negligence objectives

2004]

THE LAW OF ACCIDENTS

1871

liability criticizes merely the failure to make reparation for harm done. The imposition of negligence liability on a defendant condemns the defendant’s conduct as wrongful. Negligent conduct is unreasonable conduct; insufficiently careful conduct. Strict liability does not condemn the conduct of the party who inflicted injury. Rather, it condemns—calls unreasonable—only the failure to repair the injury inflicted. Put differently, the fundamental difference between strict and negligence liability is that under strict liability, the payment of damages to those injured by the characteristic risks of an activity is a condition for the legitimate conduct of an activity,29 whereas under negligence liability, the payment of damages is a matter of redress for the wrongful infringement of the property and physical integrity of others.30 The choice between negligence and strict liability is thus a choice both between leaving nonnegligent accident costs—costs arising out of risks reasonably imposed—on the victims who suffer them and shifting such losses back to the injurers who inflict them, and between reasons for imposing liability. Negligence is liability for unreasonable risk imposition; strict liability is liability for reasonable risk imposition. The reasonableness of preferring negligence to strict liability (and vice-versa) depends on comparing burdens and benefits to injurers and victims under these competing principles. On its face, negligence places greater burdens on victims because negligence requires victims to shoulder the costs of nonnegligent accidents, whereas strict liability in Richard A. Posner, A Theory of Negligence, 1 J. Legal Stud. 29, 42-44 (1972) (discussing respondeat superior). Guido Calabresi and Robert Keeton are among those who see it a distinct and competitive kind of liability. Guido Calabresi, The Costs of Accidents: A Legal and Economic Analysis 68-134 (1970); Keeton, supra note 2. The best characterization of the distinction between the two forms of liability may well depend in part on the best justification of the two forms of liability. The characterization offered here should therefore be viewed as a contestable one. 29. Legal doctrine and rhetoric often come very close to putting the matter this way. For example, Koos v. Roth, 652 P.2d 1255, 1262 (Or. 1982), a leading case on abnormally dangerous activity liability, explains that, under strict liability, “the question is not whether the activity threatens such harm that it should not be continued. The question is who shall pay for harm that has been done.” Loe v. Lenhardt, 362 P.2d 312, 317 (Or. 1961), explains the basis of abnormally dangerous activity liability in language that comes even closer to the language of conditional fault embraced in this Article: “The element of fault, if it can be called that, lies in the deliberate choice by the defendant to inflict a high degree of risk upon his neighbor, even though utmost care is observed in so doing.” The Comment on Clause (c) to Section 520, Abnormally Dangerous Activities, Restatement (Second) of Torts (1977), observes: The utility of [the injurer’s] conduct may be such that he is socially justified in proceeding with his activity, but the unavoidable risk of harm that is inherent in it requires that it be carried on at his peril, rather than at the expense of the innocent person who suffers harm as a result of it. Id. 30. See Gregory C. Keating, The Idea of Fairness in the Law of Enterprise Liability, 95 Mich. L. Rev. 1266, 1308-12 (1997).

1872

FORDHAM LAW REVIEW

[Vol. 72

places prima facie greater burdens on injurers, because strict liability requires injurers to shoulder the costs of nonnegligent accidents. Initially, then, we are asking when it is reasonable to place greater burdens on victims and when it is reasonable to place greater burdens on injurers. 1. Benchmarks of Fairness Judgments regarding the fair distribution of the costs of nonnegligent accidents do not arise in a vacuum. They arise against a background of mutually beneficial cooperative conduct among equal persons. Cooperative practices among equal persons give rise to natural focal points, or benchmarks, of fair division. To see this, consider the case for the difference principle in its simplest form.31 In its simplest form, the case for the difference principle depends on (1) the general idea of society as a cooperative venture among equal persons; and (2) particular features of the basic structure of society, especially the pervasive effects that the basic structure has on the life prospects of those who live under its institutions. Equal participants in a system of social cooperation have prima facie claims to equal shares of the system’s benefits. Reasonable people, participating as equals in a practice which creates both burdens and benefits, would take equal division as the presumptively fair benchmark—the focal point—from which their deliberations about the apportionment of burdens and benefits of social cooperation should start. The “priority of those worst off” which characterizes the difference principle—the fact that the distribution of income and wealth has to be justified especially to them—arises against this benchmark. The claims of those worst off under institutions which permit economic inequalities take on a certain priority, both because those worst off are receiving less than equal shares of the benefits to which they have a prima facie equal claim, and because principles governing permissible inequalities of wealth and income with respect to the basic structure of society have a pervasive and profound effect on the life prospects of those subject to them. Risks of physical injury likewise arise in the course of social cooperation among equal persons. Equality of division is likewise the presumptively fair benchmark when the burdens and benefits of those practices affect those touched by them equally. But practices of risk imposition rarely affect every member of society equally.32 Risk of
31. See John Rawls, Justice As Fairness: A Restatement 122-26 (Erin Kelly ed., 2001). See generally Joshua Cohen, Democratic Equality, 99 Ethics 727 (1989). 32. Practices of risk imposition can have pervasive and profound effects on the life prospects of those they endanger when they involve the imposition of significant risks of death or devastating injury. When this is the case, stringent precautions for the protection of those so endangered, analogous to the priority that the difference principle gives to the claims of those worst off, are appropriate. See Gregory C.

2004]

THE LAW OF ACCIDENTS

1873

physical injury is pervasive in an advanced technological society, but the activities which give rise to it are diverse and often quite particular. The injuries that preoccupy accident law arise from a variety of sources—from driving and flying; from milling, mining, and manufacturing; from producing and consuming pharmaceuticals; and from playing sports and playing with toys. These activities rarely affect every member of society in equal ways. They tend to benefit some more directly than they benefit others, and to burden some more directly than they burden others. The fact that most activities responsible for risks of physical injury burden and benefit people differentially sets the subject matter of accident law apart from the basic structure of society that is the concern of Rawls’s theory of justice. The basic structure of a society has a pervasive and profound effect on the life prospects of every member of society. Practices of risk imposition generally do not affect every member of society equally and pervasively. Generally speaking, practices of risk imposition—flying and driving, milling cotton and refining gasoline—affect people in very different ways and to very different degrees. This differential distribution of burdens and benefits affects the natural focal point from which deliberation about fair division begins. When the benefits of a practice are differentially distributed, the presumptively fair way for equals to distribute the burdens of that practice is in proportion to those differentially distributed benefits. Those who benefit from the imposition of risks should bear the costs of the accidents which result from those risks. 2. Risk Impositions Within and Between Communities Because benefit and burden presumptively should be proportional, it is useful to distinguish two fundamental kinds of cases where practices of risk imposition distribute burden and benefit differently. In the first kind of case, risk impositions occur within “communities of risk.” In the second, risk impositions take place between communities. A “community of risk” is present in its strongest form when potential injurers are also potential victims, and equally so. (In tort law, the risks of the road are often taken to be a rough approximation of a perfect community of risk.) Each member of the community has her security compromised by having to bear risks imposed by others, but each also has her liberty enhanced by being able to impose risks on others. Within a community of risk, risks may be fairly imposed and mutually beneficial in a particularly strong way. When each
Keating, Pressing Precaution Beyond the Point of Cost-Justification, 56 Vand. L. Rev. 653, 697-717 (2003). But it is the existence of pervasive effects on a discernible class of potential victims that is critical in these cases, not the existence of equal and pervasive effects on every member of society.

1874

FORDHAM LAW REVIEW

[Vol. 72

member of a community is equally a potential injurer and a potential victim, risks will be fairly imposed: Each member of the community will be exposed to equal risks by each other member of the community, and will impose equivalent risks on each other member of the community. (In an idealized community of the highway, for instance, each driver will be equally at risk and equally putting others at risk.) If the risks imposed are reasonable ones—if the freedom to impose the risks is worth more than the foregone security that is its price—then each member of the community will also benefit from the right to impose those risks. (Each driver will gain more from the mundane freedom to take his or her car on the road, for example, than he or she loses from having to bear the risks created by the presence of other cars on the road.) Within a perfect community of risk, the burdens and benefits of accidental risk imposition are fairly distributed because they are equally distributed. In a community of equals, equal division is prima facie fair—prima facie reasonable. Risks are imposed by members of one community on members of another community when potential injurers and potential victims engage in distinct activities, which do not impose equivalent risks on one another. When potential injurers play cricket and potential victims walk through their yards,33 members of one community (cricket players) are imposing greater risks on members of the other community (homeowners or pedestrians), and bearing less in the way of exposure to risk. When one party mills and the other party mines,34 because water is a resource for milling and a detriment to mining, the milling party is imposing greater risk on the mining party and bearing less in the way of exposure to risk. Even if the risks that the cricket players and the millers impose are reasonable, the burdens and benefits of those risks are not fairly distributed. Practices of risk imposition which are intermediate between these two poles are both easy to conjure up and common. For example, given the importance that driving has to our daily lives (this from someone who lives in Los Angeles), we may all stand to benefit from the practice of transporting large quantities of gasoline over the roads in tanker trucks, even though this method of transport creates risks of massive explosion, and even though most of us never expect to make use of the legal right to transport gasoline in this manner.35 Residents of Manhattan, for example, generally gain nothing of value from the right to haul gasoline by tanker truck. Indeed, they may drive so infrequently that they gain far less than Angelenos do from this

33. Bolton v. Stone, 1951 A.C. 850 (appeal taken from C.A.). 34. Fletcher v. Rylands, 159 Eng. Rep. 737 (Ex. 1866), aff’d 3 L.R.-E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.). 35. The transport of gasoline in this manner precipitated the death of the victim in Siegler v. Kuhlman, 502 P.2d 1181, 1182 (Wash. 1972).

2004]

THE LAW OF ACCIDENTS

1875

method of transporting gasoline.36 But even Manhattanites may benefit indirectly from the enterprise of transporting gasoline by tanker trucks, even if they do not impose risks on gasoline tankers equivalent to the risks that gasoline tankers impose on them, and even if they benefit less than Angelenos. Their life prospects may be improved by virtue of the prosperity created and sustained by the practice of transporting gasoline by tanker trucks, and even Manhattanites may benefit from the use of private automobiles that the practice enables. In all of these cases, the fairest—most reasonable—distribution of the costs of accidents is open to argument. But the antecedent distribution of burdens and benefits by a practice of risk imposition bears on deliberations of fairness in an important way. The antecedent distribution of burdens and benefits by a practice of risk imposition sets the benchmark or focal point from which deliberations about fair distribution begin. It is presumptively reasonable— presumptively fair—for the burdens of a risky activity to be borne by those who benefit from it. Prima facie, burden and benefit should be proportional. Prima facie, losses should be shifted if their shifting would improve the distribution of burden and benefit. And this is (prima facie) true even if the risks which resulted in those losses were reasonably imposed. Let us retrace our steps. Reasonableness requires (1) taking the impact of one’s conduct on other people into account as a circumstance capable of influencing one’s decisions; and (2) being prepared to govern one’s conduct on a basis acceptable to others affected by one’s conduct. Reasonable people seek to cooperate on fair terms with other equal, independent, and reasonable people. Fair terms enable each person to pursue his or her own aims and ends on terms that all those affected can accept. When risks of physical injury are at issue, the terms on which risks are permissibly imposed and accident costs distributed reconcile competing claims of liberty and security. Liberty and security are institutional conditions which enable people with diverse aims and ends to realize those ends, whatever they happen to be. A substantial measure of each is necessary for people to realize their particular plans and aspirations. The predicament of accident law is that, when risks of physical injury are at stake, liberty and security conflict and the task of accident law is

36. It is tempting to think that they are also exposed to proportionately less risk from this practice of transporting gasoline, so that their lesser benefit is matched by lesser burden. But it is not clear to me that they are at much less risk from the practice. Tractor trailers towing gasoline may create risks of especially great harm in the confined quarters and crowded spaces of Manhattan, even if there are fewer trucks traveling through this area. The risks posed by tractor trailers hauling gasoline may not diminish commensurately with the frequency of tractor trailer trips.

1876

FORDHAM LAW REVIEW

[Vol. 72

to reconcile liberty and security on terms that are fair and therefore acceptable to those they affect. Presumptively, fair terms of cooperation are equal terms. In a cooperative venture among equal persons, it is prima facie fair to distribute burdens and benefits equally. Equal persons should be entitled to impose equal risks on one another, and to share equally in the benefits and burdens of risk impositions. Deliberation about the fair terms of cooperation begins from this starting point. Deliberation about the fair distribution of accident costs, however, moves away from the benchmark of equal division, because practices of risk imposition generally benefit and burden those they affect differentially. When burden and benefit are distributed differentially by a practice of risk imposition, proportionality of benefit and burden—not equality of benefit and burden—is the natural focal point or benchmark from which deliberation over fairness begins. Prima facie, burdens should be shared in proportion to benefits. When accident law chooses between regimes of negligence and strict liability, it is primarily choosing between alternative distributions of the costs of accidents which arise from reasonable risk impositions and therefore should not be prevented. Strict liability and negligence differ primarily in their allocation of the costs of nonnegligent accidents—or so I shall assume for purposes of this Article. Strict liability shifts those costs back onto injurers; negligence leaves those costs on the victims of nonnegligent accidents. Prima facie, the fairness of these regimes depends on whether they bring burden and benefit into alignment with one another. The prima facie clause in the claim made in the last paragraph is important, for two reasons. First, the choice between negligence and strict liability does have precautionary effects, and those precautionary effects themselves raise questions of fairness. Indeed, in my view, the most pressing questions of fairness arise when risks threaten irreparable injury or death, so that fairness cannot be achieved after the fact by making reparation for harm done. When risk impositions threaten severe irreparable injury, it is therefore essential that strict liability induce appropriately great precaution. Irreparable injury cannot be made right after the fact. Justice must be done at the time risk is imposed, by taking sufficiently stringent precautions—or not at all.37 The “all things considered” fairness of a liability regime thus depends both on its precautionary effects and on the way that it distributes the costs of those accidents which should not be prevented. The assumption that I am making—that the precautionary effects of strict liability are, in general, at least as fair and as beneficial as those of negligence—is, I believe, a reasonable assumption. The rhetoric
37. For a discussion of this, see generally Keating, supra note 32.

2004]

THE LAW OF ACCIDENTS

1877

and rationale of both statutory and common law strict liability generally makes the same assumption—that the risks being subjected to strict liability are ones which it is reasonable to impose.38 There are, moreover, good reasons to think that common law rhetoric is right on this score. Economic analysis teaches us that strict liability generally induces greater precaution than negligence liability, because it induces potential injurers to adjust the frequency or intensity with which they conduct their activities, as well as the care with which they do so.39 The fact that strict liability places ultimate responsibility in the hands of potential injurers for choosing between letting accidents happen and preventing them from happening, whereas negligence liability places that responsibility in the hands of juries and judges,40 is further reason to think that strict liability will usually induce more effective precaution than negligence liability does. Potential injurers are likely to have greater expertise with respect to the risks of their activities, and so are likely to be better at devising and executing appropriate precautions. We therefore have good reason to suppose that strict liability will induce more stringent precaution against risks of devastating injury than negligence liability will.41 To be sure, the assumption that strict liability will generally induce appropriate precaution is an assumption, and it only holds true generally. In some circumstances, strict liability probably induces excessive precaution. The imposition of common law strict liability on recreational activities would most likely be undesirable for this In other circumstances, negligence liability may be reason.42 appropriate because we have reason to induce more intense victim precaution. In yet other circumstances, strict liability may fail to induce sufficient precautions, and may induce less precaution than negligence liability. This is a recurring worry with respect to no-fault automobile insurance, a form of strict enterprise liability implemented Even so—even by the mechanism of victim loss insurance.43 acknowledging the existence of these exceptions—there is good reason to assume that common law rhetoric is right to suppose that strict liability is usually at least as effective as negligence in inducing appropriate precaution. We shall therefore proceed on that assumption.
38. George Fletcher notes this as well. See Fletcher, supra note 4, at 543-44 (making this point and citing common law evidence in support of it). 39. See Steven Shavell, Economic Analysis of Accident Law 21-31 (1987) [hereinafter Shavell, Economic Analysis]. See generally Steven Shavell, Strict Liability Versus Negligence, 9 J. Legal Stud. 1 (1980) [hereinafter Shavell, Strict Liability]. 40. See Guido Calabresi & Jon T. Hirschoff, Toward a Test for Strict Liability in Torts, 81 Yale L.J. 1055 (1972). 41. See Keating, supra note 32 (discussing the stringent precaution against significant risks of devastating injury that fairness requires). 42. See infra text accompanying note 137. 43. See infra note 96 and accompanying text.

1878

FORDHAM LAW REVIEW

[Vol. 72

There is, however, a second reason to recognize that the “all things considered” desirability of an accident law regime is not settled solely by the fairness of the way that it distributes the costs of accidents: Considerations of fairness in the distribution of accident costs do not have automatic priority over other normative considerations which bear on the law of accidents. The basic reason that Rawls gives for the priority of the claims of justice over those of efficiency—that the choice of the principles and institutions for the basic structure of society has a pervasive effect on the life prospects of those persons who live under them—does not apply with comparable force to less encompassing institutions such as the institutions of accident law. Within the law of accidents, an analogous priority of fairness arises only when risks of severe, irreparable injury—risks of death or devastation—are at issue. Severe, irreparable injury has a pervasive and irreversible effect on the lives of those upon whom it lights, either ending their lives entirely or scarring them permanently. The power and pervasiveness of these effects make the claims of those put in peril of such harm especially urgent.44 When we are thinking about the general choice between regimes of strict liability and negligence, however, there is no class of persons whose claims have special urgency in the way that the claims of those least advantaged with respect to the basic structure of society or the claims of those whose lives stand to be devastated by risks of death or devastating injury have special urgency. The assumption behind strict enterprise liability is that fairness can generally be done after the fact by the payment of money damages to those physically injured at the hands of an enterprise’s “characteristic risks.” This assumes that the injuries inflicted are generally not severe and irreparable. When injuries are severe and irreparable, fair distribution cannot be fully realized after the fact. Death is the canonical case of a harm that cannot be dispersed and distributed across those who benefited from the risk responsible for it. The case for the priority of fairness over competing values— efficiency, for instance—therefore has a different source and strength. Whatever priority fairness has stems not from the special urgency of the claims of some of those affected by the risks at issue, but from what Martin Stone has felicitously called “the unity of doing and suffering.”45 When people act for their own advantage and benefit
44. See generally Keating, supra note 32. 45. See Martin Stone, On the Idea of Private Law, 9 Can. J.L. & Jurisprudence 235, 247 (1996). The “unity of doing and suffering” is often taken to argue against all forms of strict liability and in favor of negligence liability, with a very strict criterion of actual causation. See Weinrib, supra note 10, at 145-203; Ernest J. Weinrib, The Insurance Justification and Private Law, 14 J. Legal Stud. 681, 683 (1985) (arguing that insurance or cost-shifting rationales have no place in the justification of liability rules

2004]

THE LAW OF ACCIDENTS

1879

from the impositions of risks which harm others, they single themselves out as specially responsible for the harm involved—both because they have acted and because they have benefited. The fact that someone who has neither acted to impose a risk on others, nor benefited from the imposition of that risk, might be the best party to disperse the costs of the harms arising out of that risk is a far more dubious and problematic basis of responsibility.46 People have reason to reject outright principles of responsibility for harm done which predicate responsibility simply on the ability of an agent to realize a socially desirable end. Principles of responsibility which turn on voluntary agency and receipt of benefit are, by contrast, prima facie candidates for reasonable acceptance. Fairness thus has a strong claim to be—as Judge Friendly says in Ira S. Bushey & Sons, Inc. v. United States—an “overarching principle” which trumps the claims of efficiency when they transgress the bounds of fair responsibility. Within those boundaries, however, there is no reason to suppose that fairness has some absolute priority over other normative claims, be they claims of efficiency or of right. The final part of this Article considers how the competing normative claims of property rights, free choice, and even of intangible social goods may alter or displace the claims of fairness.47 Lastly, it is also worth observing that within the outer boundaries fixed by considerations of agency and benefit, the claims of fairness and efficiency cooperate as well as compete. Bushey itself is a case in point. Two basic aspects of efficiency—accident avoidance and loss distribution—point in opposite directions so far as the imposition of liability is concerned. The end of efficient precaution is best furthered by leaving the loss on the plaintiff, whereas the end of efficient loss distribution is best furthered by pinning the loss on the defendant.48 Fairness favors pinning the loss on the defendant, thereby converging on the same

because they do not grow out of the parties’ relationship as doer and sufferer of the same harms). I believe this view mistaken, but I cannot discuss the matter here. In my view, what counts so far as the “unity of doing and suffering” is concerned is the moral relationship between injurer and victim—the relationships of agency and benefit described in the text. 46. The great case law statement of this point is Judge Friendly’s opinion in Ira S. Bushey & Sons, Inc. v. United States, 398 F.2d 167 (2d Cir. 1968): It is true, of course, that in many cases the plaintiff will not be in a position to insure, and so expansion of liability will, at the very least, serve respondeat superior’s loss spreading function. But the fact that the defendant is better able to afford damages is not alone sufficient to justify legal responsibility, and this overarching principle must be taken into account in deciding whether to expand the reach of respondeat superior. Id. at 171 (citations omitted); see Keeton, supra note 2 (arguing that loss-spreading concerns almost never account for the imposition of tort liability and distinguishing such concerns from the fair apportionment of burdens and benefits). 47. See infra Part IV.C. 48. Bushey, 398 F.2d at 170-71.

1880

FORDHAM LAW REVIEW

[Vol. 72

conclusion as the policy of efficient loss distribution.49 More generally, when risks are reparable by the payment of money damages—when the harm done can be undone by the payment of adequate compensation—efficient precaution complements fair reparation. When harm is fully compensable—fully rectifiable by the payment of money damages—fairness can be achieved after injury is inflicted by redistributing accident costs from victims to injurers. Efficient precaution—taking cost-justified precaution and only costjustified precaution—is both consistent with fairness and conducive to its realization. Efficient precaution is consistent with fairness because fairness can be done after the fact of injury by the payment of money damages. Efficient precaution is conducive to the realization of fairness because efficient precaution maximizes the resources available for reparation. With this background in mind, let us reconstruct the reciprocity of risk criterion as a Rawlsian account of when negligence is more reasonable than strict liability, and vice-versa. II. THE RECIPROCITY OF RISK CRITERION The basic claim of the reciprocity of risk criterion is that negligence is appropriate when risks are reciprocal once reasonable care is exercised, whereas strict liability is appropriate when risks are not reciprocal once reasonable care is exercised. Because strict liability and negligence express different judgments about the conduct to which they apply, this claim implicitly asserts that the payment of money damages to the victim of a risk is rightly conceived as redress for wrongful infringement of the victim’s security by unreasonable conduct when the initial distribution of risk is reciprocal, and rightly conceived as a condition for the legitimate conduct of an activity (and paid whenever the activity issues in a “characteristic harm”) when the initial distribution of risk is nonreciprocal. A. Fairness as Reciprocity of Risk The central idea of the reciprocity of risk criterion is that negligence liability fairly apportions the burdens and benefits of risky activities within a community of reasonable risk imposition, whereas strict liability does so when risks are imposed by one community on

49. Bushey involved the flooding of a dry dock by a drunken sailor returning to his ship from shore leave. Id. at 169-70. The end of efficient precaution was best served by leaving the loss on the dry dock, because that would encourage the dry dock to install automatic locks on the valves controlling the flooding and draining of the dock. See id. at 170-71. The end of efficient loss distribution was best served by pinning the loss on the Coast Guard—the larger enterprise. See id. at 171. Fairness favored placing the loss on the Coast Guard because it benefitted from the practice— shore leave—responsible for the accident. See id. at 171-72.

2004]

THE LAW OF ACCIDENTS

1881

another.50 Recall that a “community of risk,” in its strongest form, is one whose members impose identical risks of harm on one another, thereby imposing and being exposed to equivalent risks. A community of reasonable risk imposition is one whose members impose only risks that confer more in the way of benefits on those who impose them than they inflict in the way of burdens on those who are exposed to them. A speed limit of sixty is more reasonable than a speed limit of thirty when—and only when—the freedom to drive thirty miles an hour faster is worth more than the increased risk of injury that accompanies that increase in speed. When risks are reciprocal, then, each person relinquishes an equal amount of security and gains an equal amount of freedom. When reasonable risks are reciprocal, each member of the community that imposes and is exposed to them: (1) relinquishes an equal amount of freedom; (2) gains an equal amount of security; and (3) gains more in the way of freedom than they lose in the way of security. Once due care has been exercised, reciprocity of risk thus defines a community of equal freedom and mutual benefit. Reciprocity of risk defines a community of equal freedom because reciprocity exists when risks are equal in probability and gravity. When risks are equal in these respects, persons relinquish equal amounts of security and gain equal amounts of liberty. Reciprocity of risk defines a community of mutual benefit because each person gains more in freedom than she loses in security, and because—when risks are reasonable—each person’s freedom of action is equally benefited and each person’s security is equally burdened. Subjecting reasonable reciprocal risks to strict liability increases neither freedom nor fairness because strict liability bears only on the distribution of nonnegligent accident costs.51 It does not improve the fairness of the distribution of the burdens and benefits of risk imposition because its adoption simply “substitute[s] one form of risk for another—the risk of liability for the risk of personal loss.”52 Matters are different when risks remain nonreciprocal even though injurers exercise due care. Strict liability properly applies to risks that are reasonable but nonreciprocal. The imposition of nonreciprocal risks is appropriate when those risks are to the long run advantage of the prospective victims that they imperil, but not mutually beneficial in the strong sense that reciprocal risks are. The greater-than-normal risks imposed by transporting thousands of gallons of gasoline in tanker trucks53 illustrate this circumstance. Given the importance of driving in our daily lives, each of us may benefit from the transport of
50. See generally Fletcher, supra note 4. 51. Or so Fletcher’s paper implies. See id. This is a legitimate simplifying assumption, which this Article also accepts. 52. Id. at 547. 53. See Siegler v. Kuhlman, 502 P.2d 1181, 1185 (Wash. 1972).

1882

FORDHAM LAW REVIEW

[Vol. 72

thousands of gallons of gasoline over the roads in gasoline tanker trucks, even though this method of transporting gasoline creates risks of massive explosion, and even though most of us never expect to make use of the legal right to transport vast quantities of gasoline in this manner. But we are not benefited to the extent that we would be if we all routinely drove such vehicles ourselves. The right to impose the greater-than-normal risks of driving gasoline tankers is not one that most of us find valuable. It follows, therefore, that the prospective victims of nonreciprocal risk impositions are not fully compensated for bearing these risks by the right to impose equal risks in turn. The imposition of such nonreciprocal risks is not part of a normal life, and the value of the right to impose such risks does not offset the disvalue of being exposed to them. Subjecting nonreciprocal risks to strict liability offsets this unfairness, insofar as ex post compensation can redress the harms that victims suffer. By ensuring that those injured by nonreciprocal risk impositions are—so far as possible—fully compensated for their injuries, strict liability effects a more robust mutuality of benefit. Risk is unfairly distributed ex ante, but the costs of accidents issuing from those risks are fairly distributed ex post. The damages paid under strict liability are, then, not redress for wrongful infringement of another’s security, but a way of making the distribution of the burdens and benefits of nonreciprocal risks more fair than it would otherwise be. The payment of damages makes reasonable nonreciprocal risk impositions work to the greatest advantage of those who have most reason to object to them—namely, those who benefit only indirectly from them and whose lives, limbs, and property are injured by them. Without damages, victims would still be better off bearing reasonable nonreciprocal risks than forbidding them, because reasonable nonreciprocal risks are to the long run advantage of those they imperil. But they are better off yet if they are compensated in the event that the imposition of those risks injures them. Absent reparation by injurers, victims must, at best, draw on their own resources to repair the harm that they have suffered; reparation spares them this expense. The payment of compensation to the victims of accidents issuing from reasonable but not reciprocal risk impositions is thus a condition for the legitimate conduct of activities whose reasonable risks are nonreciprocal, not redress for harm wrongly inflicted. Reasonable nonreciprocal risks are not mutually beneficial—are not fair—in the strong sense that reasonable reciprocal risks are, because the benefits of reasonable nonreciprocal risks are captured by many and their burdens borne by few. Negligence liability does nothing to rectify this unfairness, because it leaves nonnegligent accident costs concentrated on those unfortunate enough to have suffered them. Strict liability does rectify this unfairness by shifting the costs of nonreciprocal risks

2004]

THE LAW OF ACCIDENTS

1883

of impositions back onto those who impose them and so benefit the most from imposing them. This account of the proper division of labor between negligence and strict liability is appealing in part because it is both responsive to the fact that fault liability involves criticism of conduct, whereas nonfault liability criticizes only the failure to make reparation, and offers a justification for that distinction. When risks are reasonable and reciprocal, negligent conduct is objectionable for two reasons. First, because negligent conduct involves the imposition of a nonreciprocal risk, it disrupts the fair equilibrium of risk. Second, because negligent conduct involves the imposition of an unreasonable risk, negligent conduct is objectionable in a way that conduct whose benefits and burdens are unfairly distributed need not be. When a speed limit of sixty is taken to fix the upper boundary of reasonably safe driving so far as speed is concerned, the judgment being made is that hurtling down the road in a machine capable of inflicting great injury at a higher speed is a game not worth its candle. In this setting, the risks of driving seventy-five are not just nonreciprocal. They unjustifiably endanger, and would so even if they became reciprocal. The conduct which gives rise to negligence liability is objectionable—unreasonable—because it does not show sufficient regard for the security of others. Negligence liability therefore involves the criticism of conduct; it faults conduct, and properly so. Negligence liability properly criticizes conduct because insufficiently careful conduct strikes an unreasonable balance between the injurer’s own interest in being free to impose risk on others and the victim’s security. When risks are nonreciprocal in the way that the risks of transporting gasoline by tanker trucks are nonreciprocal, the underlying conduct is not unjustifiable or unreasonable. Quite the contrary, a society as dependent on private automobile transportation as ours needs gasoline, and thus needs to have it transported to gas stations. The practice of transporting it is justifiable. Strict liability is imposed not because the conduct is objectionable, but because negligence liability distributes the burdens and benefits of the conduct unfairly. Under negligence liability, the lion’s share of the benefits are reaped by those who own and operate gasoline tankers, lesser benefits are reaped (in varying degrees) by the rest of us, and the burdens of the activity are borne by an unlucky few whose fate it is to be immolated by the infrequent explosions that are the inevitable cost of the practice. The only “fault” lies in the failure to make reparation to the unlucky few who bear the burdens of the activity. Strict liability restores mutuality of benefit, so far as possible. It makes those who suffer physical injury at the hands of the practice whole, to the extent that can be done. The payment of damages through strict liability does not express a condemnation of the conduct involved in the

1884

FORDHAM LAW REVIEW

[Vol. 72

infliction of injury, it merely ensures that the activity is conducted on fair terms. B. Risk and Injury Is the reciprocity of risk criterion’s claim to fairness convincing? For a liability rule to be prima facie fair, the rule must distribute the burdens of the risks that it regulates proportionately with the benefits of those risks. If, for example, every member of a community benefits equally from the right to impose a risk, every member of that community ought to bear an equal share of the burdens of that risk— of the accidental injury and death it inflicts. Disproportionate distributions of burdens and benefits are presumptively unfair. Are fully reciprocal risks fair then? The answer is that perfect reciprocity of risk defines a fair situation with respect to the distribution of risk. When risks are fully reciprocal each member of the community of risk is equally benefited by the right to impose risks similar to those they must bear, and equally threatened by the risk impositions that they must bear. The reciprocity of risk criterion does not, however, define a fair distribution of harm, and the distribution of harm is more important than the distribution of risk. It is the ripening of risk into harm—not the chance of such ripening—that is the real burden of risk. Risk rarely impairs the ability to pursue a conception of the good over the course of a complete life. It is harm—physical injury and death—that wreaks havoc with people’s lives.54 Risk can be fairly distributed, even when the costs of the accidental harm which results from that risk is unfairly concentrated, and the distribution of harm matters more than the distribution of risk. Fairness requires that those who benefit equally from the imposition of a risk share equally in the burden of that risk—the loss of life, limb, and property that is its cost. The presence of reciprocal risk thus does not ensure that harm is fairly distributed. For harm to be fairly distributed, reciprocity of harm must be present. Baron Bramwell’s famous “live and let live” rule in nuisance law illustrates reciprocity of harm:
The instances put during the argument, of burning weeds, emptying cess-pools, making noises during repairs, and other instances which would be nuisances if done wantonly or maliciously, nevertheless may be lawfully done. It cannot be said that such acts are not nuisances, because, by the hypothesis, they are; and it cannot be
54. To be sure, there are special cases where exposure to risk is itself a kind of harm. Exposure to carcinogenic toxins and radiation can result in risks of harm that persist long after the exposure ends, and this may count as a harm in itself. See, e.g., In re TMI, 67 F.3d 1103 (3d Cir. 1995) (holding that exposure to radiation beyond a certain threshold fixed by regulation constitutes a harm regardless of subsequent personal injury). But these are exceptional, and distinctively modern, cases. Fletcher clearly has more typical (and traditional) cases in mind.

2004]

THE LAW OF ACCIDENTS

1885

doubted that, if a person maliciously and without cause made close to a dwelling-house the same offensive smells as may be made in emptying a cesspool, an action would lie. Nor can these cases be got rid of as extreme cases, because such cases properly test a principle. Nor can it be said that the jury settle such questions by finding there is no nuisance, though there is. . . . There must be, then, some principle on which such cases must be excepted. It seems to me that that principle may be deduced from the character of these cases, and is this, viz., that those acts necessary for the common and ordinary use and occupation of land and houses may be done, if conveniently done, without subjecting those who do them to an action. . . . There is an obvious necessity for such a principle as I have mentioned. It is as much for the advantage of one owner as of another; for the very nuisance the one complains of, as the result of the ordinary use of his neighbour’s land, he himself will create in the ordinary use of his own, and the reciprocal nuisances are of a comparatively trifling character. The convenience of such a rule may be indicated by calling it a rule of give and take, live and let live . . . .55

When, in “the common and ordinary use and occupation” of the land and houses, neighbors expose each other to modest interferences with each others’ use and enjoyment of property, harm is fairly distributed. When, however, in the ordinary use of the roads, drivers expose each other to similar risks of injury, harm is not likely to be fairly distributed. Fortunately, serious automobile accidents are not so frequent that drivers can routinely expect to be the victims of a nonnegligent accident issuing in serious physical injury one month and to precipitate a nonnegligent accident issuing in serious physical injury the next month. The risks of the road may be fairly distributed— because they are reciprocal—but the nonnegligent harms that issue from those risks are not fairly distributed under a regime of negligence liability. Negligence liability lets those losses lie where they fall, and they fall unevenly. The relative infrequency of serious automobile accidents appears to be a typical feature of the activities governed by the tort law of accidents. (The subject matter of accident law proper differs fundamentally from the subject matter of nuisance law in this respect. Nuisance law is centrally concerned with continuous invasions, with ongoing interferences with the use and enjoyment of property.56) Tort accident law addresses sudden explosions of standing risks into substantial physical injuries. The collapse of Rylands’s reservoir in

55. Bamford v. Turnley, 122 Eng. Rep. 27, 32-33 (Ex. Ch. 1862). 56. See, e.g., Wheat v. Freeman Coal Mining Corp., 319 N.E.2d 290 (Ill. App. Ct. 1974) (concluding that constant invasion by coal dust and smoke from defendant’s mining operation constituted a nuisance); O’Cain v. O’Cain, 473 S.E.2d 460 (S.C. Ct. App. 1996) (concluding that constant odors and flies from defendant’s hog farming operation constituted a nuisance).

1886

FORDHAM LAW REVIEW

[Vol. 72

Rylands v. Fletcher,57 is a typical example; the severing of the plaintiff’s leg and foot in Davis v. Consolidated Rail Corp.58 is another; the immolation of the plaintiff by the overturning of the gasoline tanker in Siegler v. Kuhlman59 is a third. This relative infrequency of injury—especially of nonnegligent injury—undermines the argument that reciprocity of risk defines a fair situation. When harm is infrequent, the ordinary conduct of an activity will not ensure that those who suffer nonnegligent injuries at one point inflict them shortly thereafter, or vice-versa. Reciprocity of risk will not guarantee reciprocity of harm. To be sure, the relative infrequency of injury responsible for making harm nonreciprocal even when risk is reciprocal is a good thing. A world in which people suffered serious physical injuries in automobile accidents one month, and inflicted them through automobile accidents the next month, would be a world in which life would be “nasty, brutish, and short.”60 What, then, is the fairer alternative to reciprocity of risk if reciprocity of harm is neither possible nor desirable? The law of accidents offers an answer and it is that answer—that the costs of harm, if not harm itself, may be fairly distributed by the enterprise form of strict liability—to which we must now turn. III. DISTRIBUTING THE COSTS OF ACCIDENTAL HARM FAIRLY A. The Fairness of Enterprise Liability The theory of enterprise liability asserts that: (1) accident costs should be internalized by the activity responsible for them; and (2) accident costs should be dispersed and distributed among the participants in that activity. Within the law of torts, enterprise liability flowers most fully as a form of strict liability. It has ebbed and flowed throughout the course of the twentieth century but, even when it ebbs, its influence can be found in vicarious liability cases, in abnormally dangerous activity liability cases, and in product liability cases.61 The core idea of fairness embodied by enterprise liability is an idea of fairness in the distribution of harm, namely, the idea that the burdens
57. 3 L.R.-E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.). 58. 788 F.2d 1260 (7th Cir. 1986). Plaintiff, a railroad worker, had crawled underneath a parked train to inspect it and had one leg severed just below the knee, and most of the foot on the other leg sliced off when the train that he was inspecting moved without warning. Id. at 1262. 59. 502 P.2d 1181 (Wash. 1972). 60. Thomas Hobbes, Leviathan 186 (C. B. Macpherson ed., Penguin Books 1968) (1651). 61. See Keating, supra note 30, at 1287-95. Enterprise liability conceptions do, however, sometimes surface within negligence law. See Gregory C. Keating, The Theory of Enterprise Liability and Common Law Strict Liability, 54 Vand. L. Rev. 1285, 1329-33 (2001).

2004]

THE LAW OF ACCIDENTS

1887

of accidental injury should be distributed across those who benefit from the risks which result in those injuries. Enterprise liability thus makes a claim that Fletcher rejects—that the imposition of strict liability on reciprocal risks will distribute the burdens and benefits of risky activities more fairly than negligence liability does. Fletcher remarks that “[w]here the risks are reciprocal among the relevant parties, . . . a rule of strict liability does no more than substitute one form of risk for another—the risk of liability for the risk of personal loss.”62 The plain implication of this remark is that the fair distribution of losses arising out of reciprocal risks is beyond the reach of tort law; the imposition of strict liability on reciprocal risks merely shifts concentrated harm. Enterprise liability insists otherwise—that the imposition of strict liability distributes an otherwise concentrated loss fairly, by dispersing that loss across those who benefit from the activity out of which it arose. We are at an impasse. The roots of that impasse lie in unstated empirical assumptions: Fletcher’s theory and the theory of enterprise liability assume different social worlds.63 Excavating the competing empirical assumptions is essential both to grasping the theories and to escaping the impasse in which we find ourselves. 1. Two Social Worlds The assumption which divides Fletcher and enterprise liability is gestured at by one of Oliver Wendell Holmes’s more famous turns of phrase. Writing in 1897, Holmes observed that “[o]ur law of torts comes from the old days of isolated, ungeneralized wrongs, assaults, slanders, and the like,” whereas “the torts with which our courts are kept busy to-day are mainly the incidents of certain well known businesses . . . railroads, factories, and the like.”64 Implicit in Holmes’s remark is a distinction not just between two kinds of accidents, but between two kinds of social worlds. Stylizing and simplifying, we can call these two worlds the “world of acts” and the “world of activities,” respectively. The “world of acts” is Holmes’s world of “isolated, ungeneralized wrongs.” The “world of activities” is the world in which accidents are the “incidents” of organized enterprises. In the “world of acts,” risks are discrete. The typical actor is an individual or a small firm which creates risk so infrequently that harm is not likely to materialize from any single actor’s conduct. The typical accident materializes out of the activity of isolated, unrelated actors, acting independently (i.e., natural persons or small firms
62. Fletcher, supra note 4, at 547. 63. The importance of such assumptions is stressed in Henry J. Steiner, Moral Argument and Social Vision in the Courts (1987). 64. Oliver Wendell Holmes, The Path of the Law, in Collected Legal Papers 167, 183 (1920). The paper itself was originally delivered in 1897.

1888

FORDHAM LAW REVIEW

[Vol. 72

separately engaging in activities on an occasional basis). Taken as a whole, the activities of these individual actors are diffuse and disorganized, and quite possibly actuarially small. The dogfight that precipitated Brown v. Kendall65 is a representative tort in this world: It arose out of a chance encounter between unrelated parties, neither of whose activities were large enough to make such misfortunes commonplace and expected. In the “world of acts,” then, risks are isolated, “one-shot” events. Harm, when it materializes, is an accidental misfortune. Because actors are small, and risks independent and uncorrelated, liability rules shift, but do not spread, losses. In this world, the imposition of strict liability on reciprocal risks merely “substitute[s] one form of risk for another—the risk of liability for the risk of personal loss,” as Fletcher says.66 A fair distribution of the costs of accidents—of harm—is hard to come by because the distribution of the costs of accidents across the activities that generate them depends upon the underlying activity satisfying basic criteria of insurability. Foremost among these criteria is the law of large numbers.67 But, in the purest form of the “world of acts,” both actors and activities are small. At the opposite pole from the “world of acts” is the “world of activities.” In the “world of activities,” risks are generalized and systemic. Systemic risks arise out of a continuously repeated activity (the manufacturing of Coke bottles, the transportation of gasoline, the supplying of water by a utility) that is actuarially large. “Accidental” harm is statistically certain to result from such risks: If you make enough Coke bottles some are sure to rupture;68 if you transport enough gasoline, some tankers are sure to explode;69 if you leave water mains uninspected in the ground long enough, some are sure to break;70 if you turn loose enough sailors on shore leave, some of them are bound to return to their ships drunk and wreak havoc.71 In the “world of activities,” both actors and activities are large. The cost of accidents can therefore be dispersed and distributed.
65. 60 Mass. (6 Cush.) 292 (1850). 66. Fletcher, supra note 4, at 547. 67. See Tort and Accident Law, supra note 1, at 728-29. 68. See Escola v. Coca-Cola Bottling Co., 150 P.2d 436 (Cal. 1944). 69. See Siegler v. Kuhlman, 502 P.2d 1181 (Wash. 1972). 70. See Lubin v. Iowa City, 131 N.W.2d 765 (Iowa 1964). The waterworks chose not to replace mains until they broke because it was inefficient to inspect the mains for signs of incipient breakage and replace them before they broke. 71. The suit in the Bushey case, 398 F.2d 167 (2d Cir. 1968), arose out of an incident in which a drunken sailor, returning from shore leave late at night to his Coast Guard ship, which was being overhauled in a floating dry dock, opened the valves and flooded the dry dock causing the dry dock to sink and the ship to partially sink. The court, in an opinion by Judge Friendly, affirmed that the conduct was within the scope of employment, because the risk of drunkenness was a risk increased by the Coast Guard’s “long-run activity in spite of all reasonable precautions” on its part, and hence was fairly charged to the Coast Guard. Id. at 171.

2004]

THE LAW OF ACCIDENTS

1889

In the “world of activities,” the typical injury arises not out of the diffuse and disorganized acts of unrelated individuals or small firms, but out of the organized activities of firms that are either large themselves, or are small parts of relatively well-organized enterprises. The defendant in Lubin v. Iowa City (a case where a waterworks left water mains uninspected until they broke) is large in the first sense: A single entity is responsible for the piping of water through underground pipes throughout a city, for laying and maintaining those pipes, for charging consumers for the water so transported, and so on. The transportation of large quantities of gasoline in tanker trucks on highways is large in the second sense: The firms that do the transporting may (or may not be) small and specialized, but they are enmeshed in contractual relationships with those who manufacture and refine the gasoline, those who operate gasoline stations, those who manufacture tractor trailers, and so on.72 In the “world of activities,” accidental harms can be spread across the enterprises that engender those harms. When the law of large numbers is met, risks are not only certain to issue in harms, they are also very likely to issue in harms with predictable regularity. When activities are actuarially large, the accidents that they engender will likewise be predictable and regular, and the costs of those accidents can be factored into the costs of conducting the enterprise. The costs of manufacturing and distributing Coke can include the costs of injuries from exploding Coke bottles; the costs of supplying water to households and businesses can include the costs of the damage caused by broken water mains. The move from the “world of acts” to the “world of activities” thus changes the question of fairness presented by the imposition of strict liability on fairly distributed risks. In the “world of acts,” strict liability, as Fletcher says, merely substitutes “the risk of liability for the risk of personal loss”73—it yields a different, but no fairer, distribution of the financial burdens and benefits of accidental harm. In the “world of acts,” negligence is preferable to strict liability because negligence reconciles liberty and security equally fairly, and less expensively, than strict liability does. In the “world of activities,” however, strict enterprise liability is fairer than negligence. Under enterprise liability, those who benefit from the imposition of particular systemic risks—from the risks of selling Coke in pressurized bottles, or the risks of leaving water mains undisturbed until they break—also bear the financial burdens of the accidents that issue from these risks. In the “world of activities,” unlike the “world of acts,” the extra burdens that strict liability places on the liberty of injurers are less than the extra burdens than negligence places on the security of
72. The perception that the separate actors form a connected enterprise surfaces very explicitly in Siegler, 502 P.2d at 1181. 73. Fletcher, supra note 4, at 547.

1890

FORDHAM LAW REVIEW

[Vol. 72

victims. Negligence leaves concentrated harms on injurers; enterprise liability disperses concentrated loss and distributes it across all the beneficiaries of an enterprise. In the “world of acts,” it is reasonable to impose negligence liability on reciprocal risks. Reasonable injurers may object that the move to strict liability imposes as great a burden on their freedom of action as negligence imposes on the security of victims. Under negligence, the concentrated costs of nonnegligent accidents strike victims like lightning; under strict liability those costs strike injurers like lightning. Because strict liability yields a distribution of accident costs which is no fairer than the distribution under negligence liability, it is reasonable to maximize the size of the pie by preferring the cheaper liability regime. In the “world of activities,” by contrast, the burdens are asymmetrical. By pinning an activity’s accident costs on that activity, enterprise liability distributes the costs of nonnegligent accidents across those who benefit from the underlying risks. Negligence liability leaves the costs of those accidents concentrated on unlucky victims. Those victims have good reason to object: The burden that strict enterprise liability places on injurers is both less than the burden that negligence liability places on victims and more fairly distributed. The burden of strict enterprise liability on injurers is less than the burden of negligence liability on victims because negligence liability leaves concentrated costs on victims, whereas strict enterprise liability disperses those concentrated costs across the enterprise responsible for them. The burdens of enterprise liability are more fairly distributed because enterprise liability pins the costs of nonnegligent accidents on those who benefit from the activity responsible for them. Enterprise liability places the burden on those who benefit. 2. The Facets of Fairness and the Relaxation of Causation The fairness case for enterprise liability, however, is not fully captured by the statement that it distributes the costs of accidents across those who benefit from the underlying risks. Indeed, four distinct facets of the fairness case for enterprise liability can be distinguished. These four elements combine to relax the fairly stiff requirement of causation characteristic of negligence liability in tort. The moral logic of enterprise liability inside tort law is important for two reasons. First, the exploration of that logic allows us to state the case for the fairness of enterprise liability more fully. Second, the exploration of that logic leads to the recognition that enterprise liability may also flourish outside of tort law—in nonfault, administrative accident plans—because it shows that enterprise liability in tort deemphasizes one of the traditional elements of tort liability.

2004]

THE LAW OF ACCIDENTS

1891

The first of the four facets of enterprise liability fairness is one that we have already implicitly stressed—fairness to victims. It is unfair to concentrate the costs of characteristic risk on those who simply happen to suffer injury at the hands of such risk, when those costs might be absorbed by those who impose the characteristic risk. Fairness prescribes proportionality of burden and benefit. Victims who are strangers to the enterprise derive no benefit from it, and it is therefore unfair to ask them to bear a substantial loss when that loss might be dispersed across those who participate in the enterprise and therefore do benefit from it. Victims who are themselves participants in an enterprise share in its benefits, but not in proportion to the detriment they suffer when they are physically harmed by the enterprise. Here, too, enterprise liability is fairer than negligence. It disperses the costs of enterprise-related accidents and distributes them within the enterprise, so that each member of the enterprise bears a share. Second, enterprise liability is fair to injurers because it simply asks them to accept the costs of their choices. Those who create characteristic risks do so for their own advantage, fully expecting to reap the benefits that accrue from imposing those risks. If those who impose characteristic risks choose wisely—if they put others at risk only when they stand to gain more than those they put in peril stand to lose—even under enterprise liability they will normally benefit from the characteristic risks that they impose. If they do not, they have only their poor judgment to blame, and society as a whole has reason to penalize their choices. The Coast Guard lets its sailors loose on shore leave for its own benefit (as well as for theirs) and it reaps the rewards of their shore leave. If the costs of shore leave are greater than the benefits, the Coast Guard has reason to reconsider the practice, and society has reason to discourage it. The conception of responsibility invoked in the last paragraph is a familiar and widely accepted one. We take it for granted, for example, that: the person to whom the income of property or a business will accrue if it does well has normally also to bear the risk of loss if it does badly. In the law of sales, when the right to income or fruits normally passes to the buyer, the risk of deterioration or destruction normally passes to him as well.74

The same point might be made about the purchase of stocks, or even lottery tickets. It is fair to ask agents who choose to act in pursuit of their own interests and stand to profit if things go well to bear the risk of loss when things go badly. Enterprise liability is fair to injurers. Third, enterprise liability is fair because it exacts a just price from injurers for the freedom that tort law confers upon them. Tort law
74. Tony Honoré, Responsibility and Fault 79 (1999).

1892

FORDHAM LAW REVIEW

[Vol. 72

permits potential injurers to put others at risk of physical injury without their knowledge or consent, simply because potential injurers believe that they stand to benefit from those risk impositions. Indeed, tort law requires potential victims to entrust their lives and limbs to persons and entities that stand to profit by imperiling them. This power is of great value to potential injurers: They stand to reap rewards by imposing risks in part because they can choose to impose those risks in circumstances that maximize the benefit they gain from doing so. The price that enterprise liability exacts for this freedom and power is financial responsibility for physical harm, occasioned by the exercise of that power. To induce potential injurers to exercise their power over others responsibly—and to safeguard the security of those others—enterprise liability taxes the exercise of the power to put others at risk when it goes awry and issues in physical harm. Negligence liability taxes the exercise of the power to imperil others only when the injurer has exercised that power without sufficient care.75 Accidental harms attributable to activities that are conducted carefully but at an excessively high level of intensity, or without undertaking justified research that would yield safer ways of proceeding, tend to escape the reach of negligence liability. Strict accountability induces potential injurers—particularly large enterprises—to conduct their activities more carefully. By taxing every exercise of the power to imperil others that issues in an accident characteristic of the enterprise in question, enterprise liability induces injurers to comb through their activities in search of risk reducing precautions. Worthwhile precautions whose omission escapes the eye of negligence law may be induced by the imposition of enterprise liability. Put differently, the freedom to imperil others when and as one sees fit is enormously valuable. Strict accountability for the harm that one does is a fair price to pay for that freedom, especially when paying that price helps to ensure that the power to imperil others is exercised with due regard for their safety. How do we know, though, that an appropriate level of safety—not an excessive one—is induced? Economists appeal to the idea of an optimal level of safety to answer such questions, but that idea is unavailable within a fairness framework.76 The answer it gives has both normative and empirical
75. The ideas in this paragraph draw on the writings of Steve Shavell, Guido Calabresi, and Jon Hirschoff. See Calabresi & Hirschoff, supra note 40; Shavell, Economic Analysis, supra note 39; Shavell, Strict Liability, supra note 39. 76. Much of the economic literature on the effects of enterprise liability is enormously relevant, however. It matters a great deal, for example, whether or not the expansion of liability under the influence of enterprise liability ideas triggered the insurance crisis of the late 1980s. The debate between scholars such as Richard Epstein and George Priest on the one hand, see Richard A. Epstein, Products Liability as an Insurance Market, 14 J. Legal Stud. 645 (1985); George L. Priest, The Current Insurance Crisis and Modern Tort Law, 96 Yale L.J. 1521 (1987) [hereinafter

2004]

THE LAW OF ACCIDENTS

1893

dimensions. Normatively, we have good reason to regard risks of severe, irreparable injury as especially worth avoiding. Accidental physical injury that brings life to a premature close or irreversibly alters its normal course justifies especially stringent precaution. There is good reason to think that strict accountability induces potential injurers to conduct their activities more safely than fault liability does, and little reason to think that it induces too much safety. There is good reason to believe that strict accountability induces potential injurers to conduct their activities more safely than negligence liability does precisely because strict accountability induces potential injurers to regulate the intensity as well as the carefulness with which they conduct their activities, and prompts potential injurers to comb through their practices in pursuit of superior precautions. There is little reason to think that strict accountability induces too much precaution because tort damages do not even attempt to exact a price for all of the harm that accidents wreak— emotional, relational, and economic, as well as physical.77 Indeed, the most grievous harm that accidents inflict generally goes uncompensated, because it is beyond compensation. Tort law does not generally award wrongful death damages for the value of the life that the victim lost.78 The price that strict tort liability exacts therefore seems unlikely, absent special circumstances, to induce excessive precaution. This seems all the more true when we take into account the fact that only a small fraction of valid accident claims are ever pursued.79

Priest, Insurance Crisis]; George L. Priest, Modern Tort Law and Its Reform, 22 Val. U. L. Rev. 1 (1987) [hereinafter Priest, Modern Tort Law], and scholars such as Jon Hanson, Kyle Logue, and Steven Croley on the other hand, is enormously relevant here, see Steven P. Croley & Jon D. Hanson, What Liability Crisis? An Alternative Explanation for Recent Events in Products Liability, 8 Yale J. on Reg. 1 (1991); Jon D. Hanson & Kyle D. Logue, The First-Party Insurance Externality: An Economic Justification for Enterprise Liability, 76 Cornell L. Rev. 129 (1990). If enterprise liability precipitates uninsurability in some range of cases, and if the problem cannot be corrected by various revisions to enterprise liability schemes (e.g., reformulating damages to make them lower and more manageable), then the argument that enterprise liability distributes the benefits and burdens of accidental injury more fairly than negligence simply fails. See infra Part IV.B. 77. See Tort and Accident Law, supra note 1, at 657-76 (discussing recovery for pure emotional and pure economic harm); id. at 700-21 (discussing damages for relational harm); W. Page Keeton et al., Prosser & Keeton on the Law of Torts § 54, at 359-67 (5th ed. 1984) [hereinafter Prosser & Keeton on the Law of Torts] (discussing recovery for “mental disturbance”); id. § 125, at 931-36 (discussing recovery for relational harm). 78. See Tort and Accident Law, supra note 1, at 700-13 (discussing compensation for harm stemming from wrongful deaths); Prosser & Keeton on the Law of Torts, supra note 77, § 127, at 949-54 (discussing damages for wrongful death claims). 79. See Deborah R. Hensler et al., Compensation for Accidental Injuries: Research Design and Methods 110 (1991) (“[O]verall, about one injury in ten leads to an attempt to collect liability compensation.”).

1894

FORDHAM LAW REVIEW

[Vol. 72

The fourth facet of fairness returns us to the general idea of burdenbenefit proportionality: Enterprise liability distributes accident costs among actual and potential injurers more fairly than negligence does. Negligence liability does not require that the costs of accidents—even negligent accidents—be spread among those who create similar risks of harm, whereas enterprise liability does. Enterprise liability asserts: (1) that accident costs should be internalized by the enterprise whose costs they are; and (2) that those costs should be dispersed and distributed among those who constitute the enterprise, and who therefore benefit from its risk impositions. Negligence liability, by contrast, holds that injurers have a duty to make reparation when they injure others through their own carelessness. Negligence liability justifies shifting concentrated losses, whereas enterprise liability justifies dispersing and distributing concentrated losses. To be sure, nothing in negligence liability forbids injurers from insuring against potential liability, but nothing in negligence liability requires it, either. Insurance is not integral to negligence liability, even though insuring against negligence liability is standard modern practice. It is, moreover, important in this regard that insuring against negligence liability makes negligence liability fairer precisely because it moves negligence towards enterprise liability. Negligence liability is often harsh, and problematically so.80 In part, negligence law is harsh because it justifies shifting potentially devastating losses from injurers to victims on the basis of relatively modest acts of wrongdoing. A moment’s carelessness behind the wheel of a car can inflict millions of dollars of harm, and that is enough to bankrupt most drivers. The price that negligence liability exacts can thus seem quite disproportionate to the wrongfulness of the conduct whose blameworthiness justifies the exaction. The ordinary negligence of natural persons is a relatively innocent sort of wrongdoing: The failure to foresee a risk clearly enough, calculate its probability accurately enough, concentrate well enough, or execute a course of action precisely enough, are all instances of ordinary negligence. We are all prone to such mistakes, human frailty being what it is. Yet negligence law is unforgiving. Failures to act as a reasonable person would act in similar circumstances are enough to support liability, even if those failures are the product of ordinary imperfection. And the extent of the ensuing liability can be devastating. So long as we restrict our gaze to the apportionment of costs between a particular injurer and the victim of her negligence,
80. “Average reasonable person” doctrine shows this side of negligence liability most clearly. See Tort and Accident Law, supra note 1, at 345-68; Prosser & Keeton on the Law of Torts, supra note 77, § 32, at 177 (noting that even people with severe mental disabilities are held to the standard of “a normal, prudent person”). Comparative negligence tends to mitigate some of this harshness, because it takes the particularities of the parties into account in apportioning fault.

2004]

THE LAW OF ACCIDENTS

1895

negligence law is exacting and intolerant, but justifiably and fairly so. The activities that negligence liability regulates are unforgiving. Small mistakes can explode into serious injuries. Momentary lapses of attention behind the wheel of a car—or at the helm of a ship or the controls of a plane—can and do destroy human lives. The seriousness of the harm risked by ordinary negligence is good reason to hold actors to strict standards of conduct. And the failure to conform to a norm of reasonable care is a kind of wrongdoing, even if not a particularly egregious one. Wrongdoing fairly exposes wrongdoers to responsibility to repair the harm that they have done. Forgiving wrongful lapses in concentration and failures of foresight would allocate the losses these frailties cause even more unfairly. Why should injured victims absorb the costs of the carelessness that harmed them? Shifting the costs of a negligent injury to the wrongdoer whose inadvertence caused it may be harsh, but it is fairer than letting the loss lie where it fell. Finally, forgiving lapses in concentration and failures of foresight might well encourage carelessness. Forbearance tends to foster the objects of its indulgence. Holding actors accountable for the harmful consequences of their understandable errors is, then, fairer than excusing them. But this does not settle all questions of fairness, nor undermine the argument that enterprise liability is fairer still. The small lapses that very occasionally precipitate large injuries are common indeed. Most of us occasionally let our minds wander while behind the wheel of our cars, give some small risk insufficient consideration, or fail to execute some all too familiar precaution with the precision that it requires. Most of us also usually escape without injuring anyone else. Yet the luck of the draw is all that distinguishes those of us who get away without injuring anyone from those of us who inflict grievous injury. Fate singles out an unlucky few for liability—often massive liability—and fortune spares the rest. Those unlucky few who inflict injury cannot, on balance, claim that they are unjustly held accountable for the harm that their wrongdoing has caused, but they might justly complain that a system under which they alone bear the costs of the injuries they inflict is less fair than one which pools those losses among all those who create similar negligent risks.81 Negligence mitigated by the institution of liability insurance is fairer than negligence detached from that institution. Liability insurance distributes the costs of negligence among all those who are, over the long run, similarly negligent, and that is fairer than leaving the costs of negligence on those whose misfortune it is to have their negligence issue in injury. Luck and luck alone separates the negligent who cause injury from the negligent who do not. It is fairer
81. Jeremy Waldron makes this point forcefully in Moments of Carelessness and Massive Loss, in Philosophical Foundations of Tort Law 387 (David G. Owen ed., 1995).

1896

FORDHAM LAW REVIEW

[Vol. 72

to neutralize the arbitrary effects of luck than to let luck wreak havoc with people’s lives. Just as negligence liability with the institution of liability insurance is fairer among actual and potential victims than is negligence liability without that institution, so too enterprise liability is fairer than negligence liability with insurance. Once negligence liability operates against the background of liability insurance, all that divides it from enterprise liability is its treatment of those accident costs that flow from reasonable risk impositions. Both negligence liability and enterprise liability pool the accident costs that issue from negligent risk impositions among those who are similarly negligent. Negligence liability, however, leaves the nonnegligent accident costs of an activity on the activity’s victims, whereas enterprise liability distributes those costs across the enterprise—across all those who impose the characteristic risks that lead to such accidents. Under negligence liability, victims may disperse the costs of an activity’s nonnegligent accidents by purchasing loss insurance, but victim loss insurance will not generally distribute those costs across those who impose similar risks. Loss insurance will disperse accident costs across an actuarially similar pool of insureds, and the premium will be paid out of the insured’s pocket. Absent special circumstances,82 it will be a matter of mere coincidence if the pool of actuarially similar insureds somehow benefits from the activity responsible for the harm. Demanding that victims insure themselves against accidental losses inflicted upon them by the activities of others and so shoulder the burden of realizing the socially desirable end of loss-dispersion adds “institutional insult to personal injury.”83 When reasonable risk results in accidental harm, chance and chance alone separates those who injure and are injured from those who do not and are not. To leave nonnegligent losses on those whose misfortune it is to suffer them, when we might readily spread these losses among all those who create similar risks of injury, is unfair. When the concentrated costs of nonnegligent accidents might easily be dispersed and distributed across those who benefit from the creation of the relevant risks, the victims of such accidents might reasonably object to a principle of responsibility that leaves the costs of those nonnegligent accidents concentrated on victims. Those who benefit from the imposition of the relevant risks but escape injury at the hand of those risks, by contrast, cannot reasonably object to having nonnegligent accident costs dispersed and distributed across all those who benefit from the imposition of the relevant risks. It may be rational to seek to appropriate the benefits of recurring risk
82. The principal special circumstance is the use of compulsory loss insurance to effect enterprise liability under no-fault automobile insurance. See infra Part III.B.1. 83. The felicitous expression is Jules Coleman’s. Jules L. Coleman, Adding Institutional Insult to Personal Injury, 8 Yale J. on Reg. 223, 230 (1991).

2004]

THE LAW OF ACCIDENTS

1897

imposition for oneself and to thrust the burdens of those risk impositions onto others, but it is not reasonable to do so. Dispersing the nonnegligent accident costs characteristic of an activity across pools of victims who are bound together only by their actuarial similarity is likewise less reasonable than dispersing them across the injurers who create similar risks and benefit from doing so. People who do not benefit from an activity may reasonably object to bearing its costs when those who do benefit might be made to bear its costs with equal ease. In short: Fairness favors dispersing the costs of blameless accidents among all those who create similar risks of such accidents, just as much as it favors dispersing the costs of accidents precipitated by wrongdoing among lucky and unlucky wrongdoers. Pooling the risks of negligent accidents, but not the risks of nonnegligent accidents, is presumptively less fair than pooling both sets of risks. This last argument of fairness highlights both the fact that enterprise liability relaxes the requirement of causation, and also the fact that the logic of fairness at work in enterprise liability criticizes— as arbitrary and unfair—the traditional tort insistence on a fairly rigid sort of causation. When cause and cause alone distinguishes those who injure from those who do not, luck and luck alone distinguishes those who bear liability from those who escape it. Insisting on actual causation of harm as a necessary condition of liability when luck and luck alone determines who causes harm is arbitrary and unjustifiable. There is no good reason why a person unfortunate enough to have her carelessness issue in massive injury should bear massive loss, while many others who have been identically culpable are spared all responsibility.84 Within the law of torts, the basic thrust of enterprise liability is to press for the expansion of liability within traditional domains of strict liability, and to expand the domain of strict liability relative to negligence. Just how far it does and should press in this direction are deeply contested matters in torts scholarship. But the matter cannot be said to be understudied.85 The same cannot be said, however, for
84. With small numbers, this is the lesson of Summers v. Tice, 199 P.2d 1 (Cal. 1948). 85. From the mid-1980s to the mid-1990s, there was a vigorous debate among economically inclined tort scholars over the merits of enterprise liability. For criticism of enterprise liability, see Epstein, supra note 76, at 648-53 (arguing that modern products liability law frustrates the tripartite insurance ideals of diversifying risk, ameliorating adverse selection, and limiting moral hazard); Priest, Insurance Crisis, supra note 76, at 1553 (arguing that first-party insurance is preferable to thirdparty insurance through tort liability because the former can incorporate copayments, whereas the latter cannot); Priest, Modern Tort Law, supra note 76, at 17 (arguing that product manufacturers are in a poor position to acquire adequate information about the riskiness of insureds and cannot charge higher product prices to higher risk purchasers and users); Alan Schwartz, The Case Against Strict Liability, 60 Fordham L. Rev. 819, 820, 832-40 (1992) (arguing that product defects should be subject to a

1898

FORDHAM LAW REVIEW

[Vol. 72

the way in which enterprise liability and its aspiration to distribute the costs of accidents fairly make themselves felt beyond the law of torts. It is to that subject, therefore, that we shall now turn. B. Enterprise Liability Beyond Tort If considerations of fairness favor enterprise liability within tort, they also favor enterprise liability beyond tort. Administrative alternatives to the law of torts—workers’ compensation schemes, nofault automobile insurance, statutory schemes for the compensation of certain kinds of injuries (e.g., ones inflicted by vaccination)—are often thought to express loss-spreading or insurance ideas which have little or nothing in common with the law of torts.86 The claim that these schemes embody the idea that it is better to spread a loss across many people than to leave it concentrated on one person seems correct. But it is unpersuasive to claim that these schemes embody loss-spreading aims to the exclusion of fairness concerns. The idea of fairness that enterprise liability expresses is evident in the law of torts, but it also is evident beyond that law, in administrative schemes that displace the
“market” regime of “free contract” with compulsory disclosure, because strict liability forces consumers to purchase excessive amounts of insurance and inefficiently depresses demand by forcing manufacturers to insure for nonpecuniary harm). The vigorous criticism voiced by Epstein, Priest, Schwartz, and others prompted an equally vigorous defense of enterprise liability by a younger generation of scholars. These scholars argued that third-party insurance is generally more efficient than first-party insurance, especially in the case of product-related accidents, and especially at sorting insureds into suitably narrow risk pools. See Croley & Hanson, supra note 76, at 109-10 (arguing that enterprise liability is stimulating the rise of mutual insurance companies, which are constructing more homogeneous and thus more efficient risk pools); Hanson & Logue, supra note 76, at 137 (arguing that first-party insurers fail to adjust premiums according to consumption choices and that a negligence regime therefore induces manufacturers to make suboptimal investments in product safety, whereas enterprise liability optimizes manufacture care and activity levels). Croley and Hanson have also challenged the argument that the award of nonpecuniary damages is inefficient. See Steven P. Croley & Jon D. Hanson, The Nonpecuniary Costs of Accidents: Pain-and-Suffering Damages in Tort Law, 108 Harv. L. Rev. 1787, 1791-93 (1995) (arguing that proposals to reform the tort system by reducing compensation are not efficient from a deterrence perspective and that tort law may show the existence of otherwise unmet consumer demand for insurance against pain and suffering). 86. Corrective justice theorists in particular often see these schemes as expressing only an insurance ideal that losses should be distributed, and widely so. They see this ideal as entirely different from the conception of responsibility that they find in tort law. An essential element of this claims that these schemes do not embody corrective justice conceptions of responsibility because they drastically de-emphasize causation of harm as a condition of liability. See, e.g., Coleman, supra note 10, at 395-406; Weinrib, supra note 10, at 38-42. If the argument of this section is correct, corrective justice theorists are wrong to believe that insurance justifications alone are capable of justifying these schemes. But they are right to believe that the attenuation of causation is characteristic of these schemes and, indeed, of enterprise liability itself. The logic of fairness supports this attenuation. In criticizing the attenuation of causation, collective justice theorists thus highlight a fundamental difference between the idea of fairness and the idea of corrective justice.

2004]

THE LAW OF ACCIDENTS

1899

law of torts proper. Indeed, the idea of enterprise liability found its first full expression not in the law of torts, but in one of these schemes—namely, workers’ compensation law.87 Administrative versions of enterprise liability warrant our attention for three reasons. First, their continuity with the law of torts is worth establishing, in light of contemporary claims of radical discontinuity. Second, administrative schemes are important because they are capable of instituting the idea of fairness that animates enterprise liability in circumstances where tort law cannot. Third, administrative schemes shed light on the idea of fairness that animates enterprise liability. Nonfault administrative schemes both accentuate the attenuation of causation—the de-emphasis of the causal connection between injurer and victim—that characterizes enterprise liability as a whole, and highlight the elasticity of the idea of “enterprise” itself. 1. Advantages of Administrative Alternatives The most familiar administrative alternatives to tort are workers’ compensation schemes and no-fault automobile insurance. The latter is a particularly illuminating case in point. First, the very idea that nofault automobile insurance institutes a form of enterprise liability may come as a surprise. Enterprise liability in tort is a form of strict injurer liability, associated with liability insurance. No-fault automobile insurance, by contrast, is a form of loss insurance which displaces negligence liability in tort. And loss insurance itself is usually thought of as an alternative to tort liability. Even within the tort law of accidents, the availability of loss insurance has long been conceived as a reason to bound tort liability.88 Loss insurance disperses the costs of a loss that would otherwise be concentrated on its victim. That, indeed, is its very point. In tort cases, loss insurance is less likely than liability insurance to disperse injury costs across those who benefit from the creation of the relevant risk. When victims and injurers are strangers to one another, strict liability coupled with liability

87. See, e.g., Jeremiah Smith, Sequel to Workmen’s Compensation Acts, 27 Harv. L. Rev. 235, 344 (1914); Young B. Smith, Frolic and Detour, 23 Colum. L. Rev. 444, 456 (1923) (addressing the idea that accident costs should be distributed among those who benefit from the enterprise that creates them as a distinctive conception of strict liability, and tracing that idea to the Workmen’s Compensation Acts adopted around the turn of the twentieth century). I discuss the contribution of workers’ compensation schemes to enterprise liability in tort in Gregory C. Keating, The Theory of Enterprise Liability and Common Law Strict Liability, supra note 61. 88. See, e.g., Ryan v. N.Y. Cent. R.R. Co., 35 N.Y. 209, 216 (1866) (holding that negligence liability for starting a fire should not extend beyond the house immediately set afire by the defendant’s negligence, in part because “each man” is “enabled to obtain a reasonable security” by insuring against loss). In the same vein, modern critics of enterprise liability in tort have often favored (victim) loss insurance as an alternative to (injurer) enterprise liability. See Epstein, supra note 76; Priest, Insurance Crisis, supra note 76; Priest, Modern Tort Law, supra note 76.

1900

FORDHAM LAW REVIEW

[Vol. 72

insurance will tend to disperse the costs of characteristic risks across those who benefit from their creation because efficient risk-pooling requires pooling injurers who impose similar risks of injury. This tends to disperse the costs of any given type of nonnegligent accident across those who create similar risks of such accidents. Loss insurance does not have an equally strong tendency to disperse losses across those who benefit from the risks that cause those losses because efficient loss insurance only requires dispersing accident costs across some pool of actuarially similar victims. It pools victims who suffer similar injuries, not injurers who impose similar risks. When injurers and victims are members of the same closed community of risk, however, loss insurance can distribute the costs of that community’s characteristic risks as fairly as liability insurance does. (Among other things, no-fault automobile insurance shows that reciprocity theorists are right to point to the practice of driving as a canonical instance of a “community of risk.”) Under compulsory loss insurance, each member of such a community of risk bears his or her fair share of its characteristic accident costs in the form of a loss insurance premium. Under liability insurance, they bear it in the form of a liability insurance premium. When potential victims are also and equally potential injurers, loss insurance internalizes accident costs as much as strict injurer liability does. No-fault automobile insurance exploits this fact, using mandatory loss insurance to create a kind of enterprise liability. Within a community of risk, then, it may be possible to use either compulsory loss insurance or strict liability to institute enterprise liability and thereby distribute the costs of characteristic risk fairly— across those who benefit from its creation. When compulsory loss insurance or strict liability can both distribute accident costs fairly, the choice between them turns on considerations of administrability, cost, and risk reduction. In the automobile accident context, for instance, no-fault insurance appears cheaper and easier to administer. Cheaper, because it does not require transferring the costs of nonnegligent accidents from victims to injurers. Easier to administer, because in the absence of fault it is hard to attribute automobile related accidents to one party as the “injurer.” This attribution problem is, in fact, so acute that strict liability in its usual form— holding injurers liable for all the physical harms that issue from the characteristic risks of their activity—is not a live alternative to negligence.89 By contrast, it is easy to identify an injury suffered in the course of an automobile accident, and thus easy to implement no-fault automobile insurance.
89. Recall Baron Bramwell: “Where two carriages come in collision, if there is no negligence in either it is as much the act of the one driver as of the other that they meet.” Fletcher v. Rylands, 159 Eng. Rep. 737, 744 (Ex. 1865) (Bramwell, J., dissenting), aff’d 3 L.R.-E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.).

2004]

THE LAW OF ACCIDENTS

1901

No-fault automobile insurance illustrates two general advantages that nontort administrative schemes have over enterprise liability in tort. First, such schemes are often able to solve attribution problems that common law incarnations of enterprise liability cannot. Causal problems—the inability to distinguish injurer from victim in the absence of some fault criterion—prevent the law of torts from imposing strict liability in tort on highway accidents.90 No-fault insurance circumvents this problem. By requiring victims to insure against nonnegligent losses (as well as negligent ones), no-fault insurance is capable of attributing the nonnegligent accident costs of driving to the activity. Compulsory loss insurance attributes the costs of automobile accidents to the activity of driving without requiring us to sort injurers from victims in cases of nonnegligent injury. Tort law is unable to tap the mechanism of compulsory first-party insurance against loss in a similar way. Other administrative schemes solve attribution problems which would bedevil, if not defeat, the common law of torts, by specifying in detail which injuries are to be attributed to a particular activity. The National Childhood Vaccination Act, for example, incorporates a “Vaccine Injury Table,” listing illnesses associated with various vaccines and time periods following the administration of a vaccination within which the first symptom or manifestation of an illness may occur. Proof that an illness occurred within a specific time period creates a rebuttable presumption that the vaccination was its cause. Aggregate statistical connections between exposure and illness establish causation. The second advantage of administrative schemes is that they often can effect enterprise liability in circumstances where the common law cannot, because administrative schemes can exert more control over the mechanisms and institutions of insurance. Enterprise liability in tort must, for the most part, hope that the imposition of strict liability will stimulate the provision of appropriate self or third-party insurance against liability. Administrative schemes, by contrast, can compel the purchase of insurance.91 Compelling insurance against some class of accidents both stimulates the demand for insurance and facilitates its provision. Other things equal, the larger the pool of insureds, the easier it is to spread risk among them.92 Administrative
90. See infra Part IV.A. 91. Compelled insurance is a universal feature of workers’ compensation schemes, for example. See 3 Arthur Larson & Lex K. Larson, Larson’s Workers’ Compensation Law § 150.01 (2003) (“All states require that compensation liability be secured.” (citation omitted)). 92. See Robert I. Mehr et al., Principles of Insurance 35 (8th ed. 1985) (listing “a large group of homogeneous exposure units” as the first of seven criteria that “need to be considered before attempting to operate a successful insurance plan”). Note that a pool of insureds that is larger but less homogenous is not necessarily easier to insure. It depends on whether size dominates homogeneity in the context at hand.

1902

FORDHAM LAW REVIEW

[Vol. 72

schemes can also compel the use of particular insurance mechanisms, as no-fault automobile liability schemes compel the use of first-party insurance against loss. Indeed, administrative schemes can foster the provision of insurance even more directly. Legislatures and administrative agencies can construct appropriate insurance mechanisms, and require the provision of insurance to parties who are either unable to self-insure or unable to purchase private insurance in the marketplace. State-sponsored insurance funds are a familiar part of workers’ compensation law, for example, as is the practice of providing for assignment of rejected risks.93 Nontort administrative schemes raise a host of questions. Some of these raise important issues about the logic of fair distribution of accident costs. What, for example, are we to make of the reduced damages typically found in such schemes?94 Is it fair to trade size of recovery for certainty of recovery? Is it more fair to key damages solely to the kind or severity of injury, as workers’ compensation schemes typically do? Or is it more fair to peg them at the level necessary to restore the particular victim to the level of well-being that she enjoyed prior to her injury, as tort liability generally does? Other questions raise important issues about the role of enterprise liability in reducing risk to a reasonable level. What effect does reducing damages, but making their payment more certain, have on the level of risk-reduction?95 What effect does instituting enterprise liability by victim insurance—instead of by injurer strict liability— have on the level of risk associated with an activity? These are important questions. Making the case for an administrative alternative to negligence liability in tort—for no-fault automobile insurance instead of negligence liability, say—requires addressing these questions. It is not enough to make the case that the administrative scheme distributes accident costs more fairly. The greater fairness of an administrative scheme—when it exists—has only prima facie force. If no-fault automobile insurance, for example, were to distribute accident costs more fairly than negligence but precipitate more fatal automobile accidents,96 it seems likely that its failure as a
93. See Larson & Larson, supra note 91, § 150.01 (“Six states require insurance in an exclusive state fund. Fourteen states have competitive state funds.” (citations omitted)); see also id. § 150.05 (discussing Assigned Risk Practice). In a similar vein, the National Childhood Vaccination Act, 42 U.S.C. §§ 300aa-10 to 300aa-34 (2000), creates a trust fund to pay compensation to those eligible to recover under the Act. 94. See Tort and Accident Law, supra note 1, at 1159-64, 1203-06. 95. For a recent attempt to investigate an aspect of this problem, see Yu-Ping Liao & Michelle J. White, No-fault for Motor Vehicles: An Economic Analysis, 4 Am. L. & Econ. Rev. 258 (2002). 96. See Gary T. Schwartz, Reality in the Economic Analysis of Tort Law: Does Tort Law Really Deter?, 42 UCLA L. Rev. 377, 393-97 (1994) (surveying the literature addressing whether no-fault automobile insurance increases the incidence and/or the severity of automobile accidents). Schwartz discusses the effect of no-fault’s guarantee of compensation to drivers injured through their own negligence on the

2004]

THE LAW OF ACCIDENTS

1903

system of reasonable risk reduction would trump its advantages as a system for the fair distribution of accident costs. But these questions, important as they are, are beyond the scope of our present concerns with the fundamental connections between the idea of fairness and forms of liability. One feature of nontort administrative schemes which we have yet to discuss does, however, go to the fundamental question that concerns us. Administrative schemes sometimes reach beyond enterprise liability to industry-wide liability and even society-wide liability. In doing so, do they take the logic of fairness a step further, or do they expand the notion of “enterprise” so far that it breaks, leaving us with a different kind of liability? Do industry- and societywide nonfault administrative schemes express in whole or in part the ideal of fairness which animates enterprise liability, or do they simply express the ideal of insurance—the idea that losses should be dispersed? For anyone concerned with the inner logic and the limits of enterprise liability, this is an important question. To answer it, even tentatively, however, we need to specify the distinctive features of these forms of liability. 2. Industry- and Society-Wide Liability Industry-wide liability charges accident costs arising from the type of activity conducted by a particular industry to the industry as a whole. An industry-wide fund is established, financed by a flat assessment levied on all of the firms that are members of the industry. Someone injured by the pertinent type of activity recovers from the industry-wide fund—not from the particular firm that injured her—on a nonfault basis. The National Childhood Vaccination Act imposes industry-wide liability, as does the federal scheme for compensating the victims of black lung disease.97 Tort law itself reaches beyond enterprise liability toward industry-wide liability in the special case of market share liability. Under Sindell v. Abbott Laboratories,98 for example, victims whose injuries are caused by generic products may sue every producer of the generic product that injured them, and may recover from each firm in proportion to that firm’s share of the product market. Under this form of market share liability, all the firms that manufacture a particular product share collective responsibility for the product’s accident costs, and each firm pays its proportionate share of those costs.

frequency and severity of accidents. Id. at 396. 97. Title IV of the Federal Coal Mine Health and Safety Act of 1969, Pub. L. No. 91-173, 83 Stat. 742, 792-98, amended by The Black Lung Benefits Act of 1972, 30 U.S.C. §§ 901-945 (2000). 98. 607 P.2d 924 (Cal. 1980).

1904

FORDHAM LAW REVIEW

[Vol. 72

Society-wide liability compensates victims out of general tax revenues: The whole society is the source of reparation. The New Zealand Accident Compensation Scheme is the most famous example, but administrative schemes such as the Price-Anderson Act99—which governs the liability of licensed private operators of nuclear power plants for nuclear accidents—also embodies the idea of societal responsibility. Under society-wide liability, reparation is made not by the firm and its insurer (as under enterprise liability), or by the industry as a whole (as under industry liability), but by society as a whole. For our purposes, the fundamental question here is whether industry- and society-wide liability are simply expressions of the idea of social insurance, or if they also embody the enterprise liability idea of fairness. Social insurance embodies the idea of loss-spreading writ large—the idea that it is better to disperse the costs of significant injuries as widely as possible, rather than leave them concentrated on victims. Nothing in the idea of social insurance requires dispersing costs across those who benefit from their creation. Enterprise liability links loss-spreading to fairness: Accident costs should be internalized by—and distributed across—the enterprises that generate them so that burdens and benefits are fairly proportioned. In important part, industry- and society-wide liability embody the same idea of fairness as enterprise liability. Market share liability holds individual firms responsible for that portion of a product’s accident costs that corresponds to the firm’s share of the product market. By so doing, it apportions financial responsibility in accordance not just with the harm caused by the firm, but in accordance with the benefit that the firm derived from the sale of the product, on one plausible measure of benefit. Market share liability thus institutes the principle of burden-benefit proportionality within an entire product market—or industry. When industry-wide liability is instituted by administrative scheme, it has essentially the same effect, though the extent to which burden-benefit proportionality is realized depends in part on the way the assessment is levied within the industry. Payments in proportion to risk imposed realize burdenbenefit proportionality more fully than a flat assessment on each firm. Enterprise liability radiates out even further in the Price-Anderson Act, which limits the liability of nuclear power plants licensed by the federal government for nuclear accidents resulting from the operation of those plants, and further still in the New Zealand Accident Scheme. Considerations of fairness figure prominently in the justification of both schemes. The Price-Anderson Act illustrates both the establishment of industry-wide liability by administrative act, and the
99. Pub. L. No. 85-256, 71 Stat. 576 (1957) (codified as amended in scattered sections of 42 U.S.C.).

2004]

THE LAW OF ACCIDENTS

1905

movement beyond industry liability to society-wide liability. At the time that its constitutionality was litigated in Duke Power Co. v. Carolina Environmental Study Group,100 the Price-Anderson Act included an industry-wide fund, albeit one that may not have been large enough to cover the full cost of a major nuclear accident. The district court opinion in the case explained the link between such a fund—between industry-wide liability—and the principle of burdenbenefit proportionality:
[A] liability pool . . . requiring either contributions in advance, or liability for assessment on a unit basis or otherwise, of all power companies building or operating nuclear generators . . . would effectively place the responsibility upon the group most directly profiting from any improvement in the costs or usefulness of electric power—the power company stockholders and the customers themselves.101

Can the idea of benefit-burden proportionality be linked to societywide liability? The court thought so:
Another rational alternative [to industry-wide liability] would be to make such accidents a national loss and to pay those damaged out of the federal treasury. This would spread the loss among those who benefited indirectly by having the nation’s power supply increased as well as among those who presumably benefited directly.102

The force of this point—the fairness of society-wide liability—can be seen by considering the rationale of the Supreme Court decision upholding the constitutionality of the Price-Anderson Act, even though its ceiling on total recovery would leave some victims of a major nuclear accident unable to recover from the fund created by the Act. Liability limitation, the Court said, “is an acceptable method for Congress to utilize in encouraging the private development of electric energy by atomic power” when Congress reasonably concludes that the development of such power plants is in the public interest.103 If public benefit is the rationale for encouraging the private provision of nuclear power, it is fairer for society as a whole to share the costs of a nuclear accident, than it is to leave the loss on those unlucky enough to be harmed by such an accident. Benefit-burden proportionality— fairness—favors placing an important share of the liability on the public. The widest scheme of society-wide liability is the New Zealand Accident scheme, which covers all cases of “personal injury by accident.” This scheme expresses a “principle of community
100. 438 U.S. 59 (1978). 101. Carolina Envtl. Study Group, Inc. v. U.S. Atomic Energy Comm’n, 431 F. Supp. 203, 225 (W.D. N.C. 1977). 102. Id. 103. Duke Power Co., 438 U.S. at 86.

1906

FORDHAM LAW REVIEW

[Vol. 72

responsibility.” That principle involves a capacious interpretation of the idea of fairness as proportional sharing of burden and benefit:
[S]ince we all persist in following community activities, which year by year exact a predictable and inevitable price in bodily injury, so should we all share in sustaining those who become the random but statistically necessary victims. The inherent cost of these community purposes should be borne on a basis of equity by the community.104

The first lesson here is that the ideal of fairness may animate industry- and society-wide liability as much or more than the ideal of loss-spreading. The contemporary inclination to see all invocation of insurance considerations in tort law simply as an expression of the idea that losses should be dispersed, not concentrated,105 can blind us to the role being played by considerations of fairness. The idea of fairness that informs enterprise liability makes its presence felt even in the most expansive administrative schemes of accident law; even in schemes which appear, at first blush, to be pure expressions of insurance ideas. A second lesson has more to tell us about the idea of fairness itself. Identifying the relevant community of benefit and burden—the relevant enterprise—is a standing challenge for any form of enterprise liability. A common, and well-taken, criticism of burden-benefit proportionality has it roots in the difficulty of identifying the relevant enterprise. That criticism complains that the idea of burden-benefit proportionality is analytically obscure or incomplete: How can we tell who benefits and in what proportion?106 Appeal to the economic notions of marginal and infra-marginal benefit is out of place here; we are discussing fairness, not efficiency. The variety of forms of enterprise liability extant in our law provide an answer to this question. That answer reveals an analytic idea, but it also displays the role of controversial, contestable, and essentially political judgment. Why do these schemes claim that the characteristic accident costs of an activity may be fairly distributed by industry- and society-wide liability as well as by enterprise liability in tort? The analytic basis of their claim is that the benefits of risky activities radiate outward in concentric circles. Nuclear power most benefits those who produce and consume it, but it also benefits those of us who merely happen to live in a society made wealthier by its presence. This is not a peculiarity of nuclear power. The immediate
104. Report of the Royal Commission of Inquiry, Compensation for Personal Injury in New Zealand para. 56, at 40 (1967). 105. Dispositions in both economically oriented and corrective justice approaches to tort combine to support this tendency. Economically inclined scholarship supposes that accident law should pursue the twin ends of efficient accident avoidance and efficient insurance against accidental injuries whose prevention is inefficient. Corrective justice theories have been disposed to accept economic characterizations of the case for enterprise liability. See supra note 10. 106. I am grateful to Ken Abraham for urging on me the importance of this criticism.

2004]

THE LAW OF ACCIDENTS

1907

benefits of transporting gasoline by tanker trucks accrue to those in the industry; less immediate but still substantial benefits accrue to those who use gasoline regularly; and still smaller benefits accrue to those who rarely or never drive themselves, but who benefit from the productive activity that driving enables. The benefits of activities radiate outward until they diminish to the point where they are no longer identifiable. The determination of who benefits—of what the relevant community of benefit is or ought to be for purposes of apportioning the costs of accidents—is an essentially contestable matter. Deciding whether we should, “all things considered,” move to a regime of accident law with more industry-wide liability, more administrative plans, or even abolish tort law entirely and replace it with a New Zealand style scheme of society-wide liability is a vast undertaking, well beyond the scope of this Article. For our purposes, the main lesson lies in the fact that enterprise liability can be given such widely varying construction, so that fixing the proper scope of enterprise liability requires the exercise of normative and political judgment. Judgments about communities of benefit are eminently political judgments about how we should order our lives in common— and properly so.107 Because risky activities radiate their benefits out across a variety of actors, and because the boundaries of communities of risk may be fixed in narrower and broader ways, the idea of fairness can give rise to industry- and society-wide liability as well as to enterprise liability in tort. Whether the idea of fairness that animates enterprise liability in tort is best embodied—either in general or in a particular instance—by enterprise liability in tort, or by an administrative incarnation of enterprise liability, or by an administrative scheme of industry- or society-wide liability, is a matter that can only be settled by detailed examination of the possibilities at hand. Deciding whether enterprise liability should be instituted at the level of individual firms (as common law enterprise liability usually does), or at the level of a particular risky practice such as driving, or at an industry- or society-wide level, requires both an appraisal of the institutional mechanisms available and the reasons favoring broader and narrower specifications of the relevant community. The exercise of contestable, normative, and political judgment is cause for concern only insofar as it raises the worry that the idea of benefit-burden proportionality is empty or especially obscure.
107. See Honoré, supra note 74, at 91. One can argue that the distribution of risks should, for example as regards motoring, take place at the level not of the individual but of the vehicleowning population or the whole community. The level at which risks should be distributed in a particular area of community life seems pre-eminently a matter of political judgment. Id.

1908

FORDHAM LAW REVIEW

[Vol. 72

The worry is unfounded. Questions about how to interpret burden and benefit arise in particular contexts. Those contexts provide structure and definition, so that the determination of the relevant community of benefit and burden becomes a classic interpretive Legislators question of the sort lawyers routinely confront.108 constructing enterprise liability schemes begin with a pre-theoretic sense of the domain of burden and benefit—be it customers, employees, suppliers, and shareholders of a firm; members of an industry; or society at large—and specify the relevant community more precisely by their choice of liability rules and financing mechanisms. Courts applying and articulating enterprise liability ideas confront much more well-defined interpretive issues. The issues that courts must settle arise within statutory schemes or common law constructs specifying communities of burden and benefit fairly precisely. Drawing on an enterprise liability conception of fairness to fix the contours of the scope of employment rule in vicarious liability law—as Judge Friendly does in the Bushey opinion109—is a classic case in point. In the vicarious liability context, enterprise liability ideas point us towards devising a test for scope of employment which locates the boundaries of a firm’s activity. (Negligence ideas, by contrast, point us towards devising a test which identifies those firm-related accidents which the firm should have prevented.) Identifying the boundary of an activity (e.g., of driving) or of an industry’s impact (e.g., of mining’s health effects) is not a possibility before the court. Because this exercise is an interpretive one through and through, we do not need an independently specified account of burden and benefit. IV. LIMITS AND EXCEPTIONS: THE RESILIENCE OF NEGLIGENCE LIABILITY The coin that we have been studying has a flip side. Even if the greater fairness of enterprise liability were widely accepted, and even if considerations of fairness were widely agreed to be decisive, enterprise liability would not expand to consume the whole of tort accident law. Within tort law, negligence liability would prove remarkably resilient, for at least three reasons. First, it may often be impossible for the common law to attribute accidents to activities without the benefit of a fault criterion. Second, risks may sometimes be uninsurable, in which case enterprise liability will not be able to realize its distinctive aspirations and the imposition of strict liability will be no fairer than the imposition of negligence liability. Third,
108. I mean “interpretive” in the sense made famous by Ronald Dworkin. See, e.g., Ronald Dworkin, Law’s Empire (1986). 109. Ira S. Bushey & Sons, Inc. v. United States, 398 F.2d 167, 171-72 (2d Cir. 1968).

2004]

THE LAW OF ACCIDENTS

1909

competing normative considerations may overcome the presumption that fairness generates in favor of enterprise liability. Let us consider these in turn. A. Attributing Accidents to Activities The fact that enterprise liability relaxes traditional tort requirements of causation should not blind us to the fact that it requires attributing accidents to activities. Enterprise liability is liability for “characteristic risk.” The common law is not always able to isolate the “characteristic” risks of certain activities. 1. “Characteristic Risk” and the Confluence of Activities “Characteristic risks” are reasonable risks of a particular kind of injury which exceed the background level of risk, and flow from the long-run activity of an enterprise. These risks are reasonable because we are better off bearing them than preventing them—the “cost” of prevention exceeds the benefit. They “flow from [an enterprise’s] long-run activity in spite of all reasonable precautions on [its] part.”110 They are characteristic of an enterprise because the enterprise’s presence in the world increases the incidence of the risk above its normal, background level. The ill-fated drunken sailor whose trespass precipitated the flooding of the dry dock in Bushey is one justly famous case in point. Drunkenness, Judge Friendly remarked, is “the condition for which seamen are famed,”111 and increased drunkenness in the vicinity of berthed Coast Guard vessels is something that reasonable people might take to be characteristic of the Coast Guard’s enterprise. The long run effect of turning sailors loose on shore leave may well be to increase the incidence of drunkenness in the vicinity of the vessels from which they are dispatched. What might be a parallel example of a background risk? The risk of increased traffic accidents created by turning sailors loose on shore leave on the docks of Brooklyn might be an example. Common sense suggests that sailors on shore leave in Brooklyn are not unusually prone to precipitate automobile accidents in the vicinity of the vessels from which they disembark. It seems unlikely that sailors discharged on the docks of Brooklyn will be driving back and forth between shore and ship with unusual intensity. Those who do rent cars seem likely to use them to travel a considerable distance from the waterfront, dispersing quickly so that the risks they create merge rapidly into the ordinary risks of the road. This, of course, is very much a contingent fact. In other settings—say, a military base located in Southern California, where driving is ubiquitous—military
110. Id. at 171. 111. Id. at 168.

1910

FORDHAM LAW REVIEW

[Vol. 72

personnel may well be prone to venture forth from their bases on leave in their own cars. In that setting, if leave continues to increase drunkenness, it should also lead to a characteristically greater risk of automobile accidents.112 The idea of an occupation’s “characteristic risks” or of a firm’s “characteristic risks” is a comprehensible one, relatively welldeveloped in the case law. Even so, hard cases are easy to conjure up. Suppose that members of particular occupations—salespeople of various kinds, perhaps—both drive and telephone a great deal in the course of their work. Suppose further that some such salesperson, commuting to work in the morning or back home in the evening, culpably causes an accident because she is talking on her phone. Should we consider her accident to be a characteristic risk of her employer’s enterprise? Accidents that occur in the course of commuting are generally not counted as characteristic risks of particular employments.113 People are likely to commute to some job. If they weren’t commuting to this particular job, they would be commuting to another one, so the connection between a particular job and a particular commute is generally coincidental not characteristic. Should this general rule apply to our talkative salesperson? Should it apply if the phone is fixed in the car and paid for by the business, but not if it is fully mobile and paid for by the salesperson? Should it matter if the call was a personal or a business one? Should it matter if the salesperson was leaving work late or going to work early because the press of business required her to put in a long day? Should the background level of risk—the prevalence of talking on the phone while driving—in this community be critical? Cases plainly exist where it is hard to say if the negligence involved is a characteristic risk of the firm’s activity, something peculiar to the life of the driver in question (she was distracted by a furious argument with her father-in-law who just happened to call while she was driving), or a feature of the entire community of drivers in the area where she lives and works (no social sanction is attached to talking on the phone while driving). Within a well-established enterprise liability doctrine such as the vicarious liability of employers for the torts of their employees, hard cases of this sort—close cases—are by and large manageable. But there are important classes of cases where common law efforts to construct adequate criteria of “characteristic risk” face

112. See Taber v. Maine, 45 F.3d 598, 600 (2d Cir. 1995) (applying the law of Guam and holding the United States Government vicariously liable for an automobile accident involving a navy serviceman who drove off base while on liberty after a “grueling 24 hour duty shift”). 113. See Konradi v. United States, 919 F.2d 1207, 1209 (7th Cir. 1990) (“The general rule is that an employee is not within the scope of his employment when commuting to or from his job.”).

2004]

THE LAW OF ACCIDENTS

1911

great obstacles, because it is difficult to disentangle the long-run increase in accidents effected by a particular enterprise. All accidents arise at the intersection of two or more activities. In some circumstances, it is impossible for the common law of torts to attribute responsibility for an accident to one of the parties to it without employing some criterion of fault.114 Highway accidents are the canonical case.115 In the absence of norms—usually statutes— specifying duties of precaution, rights of way, and so on, it is often impossible to attribute responsibility for accidental injury. In the absence of crosswalks, we may not be able to say if a pedestrian or a driver was responsible for an accident between the two. In the absence of rules ordering priorities among vehicles at four-way intersections, we may not be able to say whose activity is responsible for an accident between two cars at such an intersection. Highway accidents are (to my mind) somewhat atypical in that acute problems of attribution arise even in cases where accidents arise at the intersection of only two activities—when only two vehicles are involved, or only one vehicle and a pedestrian. Problems of causal indeterminacy tend to be more acute when accidents arise at the intersection of multiple activities. When a bus bearing schoolchildren is struck by a train at a railroad crossing, for example, the accident arises at the intersection of numerous activities. The railroad contributes to the accident by its design and operation of both the crossing and the train. Surface grade crossings can be replaced by
114. Stephen R. Perry argues in The Impossibility of General Strict Liability, 1 Can. J.L. & Jurisprudence 147 (1988), that “general strict liability” is impossible because we cannot attribute accidents to activities without employing fault criteria. Arthur Ripstein and Jules Coleman essentially accept Perry’s arguments in Mischief and Misfortune, 41 McGill L.J. 91, 107 (1995). See also Arthur Ripstein, Equality, Responsibility, and the Law 32-53 (1999). If Perry means what Bramwell means, see supra, note 12—that universal strict liability is impossible—then I agree with him. If he means, as he sometimes seems to, that there are no effective strict liability attribution rules—because all effective strict liability rules are fault rules in disguise— then I disagree with him. Some strict liability attribution rules—the “scope of the employment” test in vicarious liability law, the “scope of the risk” test for abnormally dangerous activity liability, the manufacturing defect test in product liability law, and the “arising out of and in the course of employment” test in workers’ compensation law, come to mind—connect accidents and activities as effectively as fault criteria in many circumstances. Whether or not satisfactory strict attribution rules can be devised in a particular context depends on the features of the context. 115. See, e.g., Fletcher v. Rylands, 159 Eng. Rep. 737, 744 (Ex. 1865) (Bramwell, J., dissenting), aff’d 3 L.R.-E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.). The thrust of Bramwell’s opinion is strongly supportive of a regime of strict liability for accidents among strangers. The Exchequer Chamber entered judgment in favor of the defendants on the ground that they were not negligent and Bramwell dissented “on the plain ground that the defendants have caused water to flow into the plaintiff’s mines, which but for their, the defendants’, act would not have gone there.” Id. His position, then, is that it is impossible to attribute a highway collision to one party rather than another absent fault on one driver’s part, and this makes it impossible to impose strict liability on highway accidents.

1912

FORDHAM LAW REVIEW

[Vol. 72

underpasses and overpasses; warning bells and horns can be more and less piercing; gates can bar access to the tracks more and less effectively; the schedule and maintenance of the train can affect the incidence of accidents; as can the selection and training of the engineers who operate the trains. The traffic department of the city can affect the incidence of accidents by its designing of roads, routing of traffic, and timing of lights. The school department can affect the incidence of accidents by its choice of routes and its selection and training of drivers. The manufacturer of the bus can, by its choice of design, affect the ease with which accidents can be avoided and the severity of the injuries that they cause. Buses can be more or less maneuverable, more or less soundproof, and more or less crashworthy. It can be extremely difficult and even impossible to devise nonfault attribution rules—such as the “scope of the employment” test for vicarious liability or the “scope of the risk” test for abnormally dangerous activities—to apportion responsibility for accidents that arise at the intersection of multiple activities.116 Negligence norms must often be deployed by default. Fairness may require that each of these enterprises—railroading, designing automobiles, transporting schoolchildren, designing traffic systems—bear its characteristic accident costs, but complexity tends to defeat the identification of characteristic risks. Fault liability therefore tends to expand beyond the boundaries within which fairness wants to confine it. 2. Background Risk Risks which are very remote—very, very low in probability— present another problem for enterprise liability. Enterprise liability cannot be instituted by the common law when the nonnegligent risks of an activity are so low that they simply merge into the general background risks of living. Risk of physical harm—diminished security—is the byproduct of productive activity. Some risk of serious physical injury and death is the price of activity, of freedom to act. We cannot farm or build or barbeque or drive or fly without taking and imposing risks of devastating injury. Reasonable precaution cannot eliminate all risk. Risks whose existence is the unavoidable price of activity are the background risks of social life; the background
116. That is not to say that it is always impossible to devise strict liability rules for accidents that arise at the intersection of multiple activities. Product liability law proves otherwise. Defect tests which go beyond negligence are familiar in modern product liability law. The consumer expectation test is a nonfault test, for example, and the risk-utility test goes beyond fault liability when it judges the adequacy of a product by the knowledge available at the time of trial instead of the knowledge available at the time the product was marketed, or when it relaxes the strict kind of feasible alternative design requirement found in the Restatement Third’s formulation of the defect test.

2004]

THE LAW OF ACCIDENTS

1913

against which “characteristic risks” arise. “Background” risks are typical of social life in general; they are the price of a number of different activities whose separate contributions cannot be disentangled. With certain inescapable variations, ordinary activities, carefully conducted, produce mutually imposed and mutually beneficial risks.117 Very, very low probability risks of massive damage from fire, for example, are created by a wide range of activities: smoking cigars, pipes, and cigarettes; barbecuing in the backyard; using electrical appliances or gas stoves; driving gasoline powered vehicles; and so on. Equal background risk matures into unequal harm. In the long run, even very, very low probability harms inevitably issue in serious injury and death. Fate condemns a small number of us to be victims of the ordinary nonnegligent risks of social life and fortune spares the rest. This inequality of harm may well be unfair. Perhaps the inevitable accidents that issue from the background risks of social life ought to be shared across society as costs of living.118 But the unfairness is beyond the rectification of tort accident law, which can only attribute accidents to particular activities. It is arbitrary and unfair to attribute the inevitable accidents that are the fruit of background risk to any particular activity. The connection between any such accident and any particular activity is merely fortuitous, a matter of coincidence, not causation. It was merely bad luck, for example, that the child plaintiff in Van Skike v. Zussman119 was inspired to play with fire and so suffer serious injury by winning a toy lighter—a miniature plastic plaything incapable of being lit—as a prize in a gumball machine. He might have been inspired to play with fire by any of a number of other things—by a gas grill or a stove burner or a campfire, by seeing a fire on television, or by some burst of childhood curiosity sparked by something other than exposure to fire. The connection between this child’s injury and this particular product—this toy lighter—is coincidental. It is therefore unfair to pin responsibility for this accident on the parties responsible for selling the toy lighter to the child. Because accidents that issue from background risks cannot fairly be attributed to any particular activity,120 tort accident law must let the
117. See Fried, supra note 4, at 192-93. 118. Note that a nontort version of enterprise liability, like the New Zealand Accident Compensation plan, could attribute background risks to an activity, namely, to social life. 119. In Van Skike v. Zussman, 318 N.E.2d 244, 246 (Ill. App. Ct. 1974), a small child “won” a toy cigarette lighter as a prize from a gumball machine, purchased lighter fluid, and set himself on fire when he attempted to fill the toy lighter with the lighter fluid. 120. The fire in Van Skike also shows how “industry-wide” liability as well as “society-wide” liability might be able to institute enterprise liability in circumstances

1914

FORDHAM LAW REVIEW

[Vol. 72

losses that flow from background risks lie where they fall. For tort law, the unfairness of the unequal harm that issues from background risk is an inevitable one. When the nonnegligent risks of an activity are so low that they disappear into the standing level of background risk, negligence liability is at least as fair as strict liability, and more practicable. Strict enterprise liability is at least as unfair as negligence because it can only connect injuries to activities in an arbitrary way.121 And there is no reason that any particular activity should bear any particular background risk. Strict enterprise liability is less practicable than negligence because there is no distinctive risk for strict liability to attach itself to, so the attribution of accidents to activities must, inevitably, be erratic. B. Uninsurable Risks Enterprise liability in tort supposes that we live in the “world of activities,” and that the costs of accidents can therefore be distributed across those who benefit from the imposition of the risks that issue in those accidents. It supposes that the risks subject to strict liability are insurable, so that the imposition of strict liability does not merely “substitute one form of risk for another—the risk of liability for the risk of personal loss.”122 In the “world of acts”—Holmes’s world of “isolated, ungeneralized wrongs”—inability to insure against accidental injury may have been the norm. The sine qua non of insurability is a sufficiently large activity. The supposition that we have moved from the “world of acts” into the “world of activities” is a stylized but surely substantially accurate one. Even so, the supposition that all risks are insurable is too strong. Even within a “world of activities,” some actors and activities may still operate in the where the common law cannot. The fact that the connection between this child’s injury and the particular product—the toy lighter—whose use occasioned it is coincidental does not prove that this accident is unconnected to any identifiable activity. It might make perfect sense to classify this accident as a risk of the broader activity of “taming fire and putting it to productive use.” The risk that children will take an interest in fire might be classified as a standing risk of having lighters, matches, ovens, stoves, water heaters, barbecues, candles, chimneys, cigarettes, and countless other ordinary products and activities that make constructive use of fire. So conceived, the plaintiff’s injury does issue from the “characteristic risk” of an activity, namely, the activity of “putting fire to use.” It might be, that is, a risk: (1) associated with the presence of a particular activity; and (2) that persists after all reasonable precautions to reduce it have been taken. If we had at our disposal an “industry-wide liability” scheme covering the various particular activities that constitute the broader activity of “putting fire to use,” it might make perfect sense not to count this accident as a background risk of living but as a cost of using fire. 121. Arguably, strict liability is more unfair than negligence in this circumstance. If there is no justice done by shifting a loss, it may be more unfair to shift it than to let it lie. The very act of doing so may, in this circumstance, be an injustice. So, too, it may be unfair to consume resources that would otherwise be available for another use when there is nothing to be gained from their consumption. 122. Fletcher, supra note 4, at 547; see Brown v. Collins, 53 N.H. 442, 446 (1873).

2004]

THE LAW OF ACCIDENTS

1915

“world of acts.” More than that, the liability system experienced what was widely termed an “insurance crisis” in the mid- to late- 1980s; a crisis which called into question the possibility and desirability of enterprise liability in tort. Prompted by this “insurance crisis,” an important strand of legal scholarship argued forcefully that the long expansion of enterprise liability over the course of the twentieth century had precipitated an insurance crisis.123 The extension of enterprise liability to a wide range of activities for which liability insurance was ill-suited had caused a widespread withdrawal of insurance coverage; that withdrawal had led, in turn, to the withdrawal of important services. The nonnegligent risks of small children being abused by caregivers at day care centers was an ostensible case in point.124 Day care centers, the argument ran, were generally too small to self-insure, and thirdparty insurers were not able to identify the relative riskiness of prospective insureds cheaply and accurately enough to supply such insurance. Deprived of insurance, day care centers were forced to close, leaving working parents in the lurch and the prospective victims of child abuse—the children themselves—arguably worse off than they would have been had enterprise liability not been imposed. The argument that the extension of enterprise liability precipitated a crisis of insurability (rather than, say, a temporary disruption of such markets leading to an efficient readjustment) was forcefully challenged by another, equally important strand of scholarship, in the early- and mid-1990s.125 And the matter remains eminently debatable. For our purposes, however, it will do to suppose that some risks are uninsurable, even in the modern “world of activities,” and to leave to another day the task of determining just how broad the range of uninsurable risks is. Our concern is normative: What is the fair liability rule for “uninsurable risks”? When risks are uninsurable, Fletcher’s reciprocity of risk criterion regains its attractiveness. Fairness requires proportionality of benefit and burden. Within a community of reciprocal risk, everyone benefits equally from the imposition of nonnegligent risks and should share equally in the costs of the accidents that issue from those risks. When accidental losses meet criteria of insurability, this is what enterprise
123. See, e.g., Epstein, supra note 76; Priest, Insurance Crisis, supra note 76, at 1534-39. 124. This is one of a “parade of horribles” recounted by Priest. Priest, Insurance Crisis, supra note 76, at 1527, 1578-79. 125. See, e.g., Croley & Hanson, supra note 76 (arguing that enterprise liability is stimulating the rise of mutual insurance companies, which are constructing more homogeneous and therefore more efficient risk pools); Hanson & Logue, supra note 76 (arguing that first-party insurers fail to adjust premiums according to consumption choices and that a negligence regime therefore induces manufacturers to make suboptimal investments in product safety, whereas enterprise liability optimizes manufacturer care and activity levels).

1916

FORDHAM LAW REVIEW

[Vol. 72

liability does. When accidental losses are not insurable, however, strict liability will simply concentrate the costs of nonnegligent accidents on those whose misfortune it is to have their activity inflict injury. Negligence concentrates those costs on a different class of persons—those whose misfortune it is be injured. Strict liability simply substitutes the “risk of liability for the risk of personal loss.”126 Neither distributes the costs of accidents fairly. Either way, the cost of the harm is concentrated, not dispersed across the community that benefits from the freedom to create the kind of risk that has matured into harm in the case at hand. In a world where liability cannot be insured against, either liability or loss may be devastating. If so, negligence liability may be preferable, because it is equally fair and less expensive to operate. Strict liability is probably more expensive to administer than negligence because strict liability requires cranking up the liability system for nonnegligent accidents as well as for negligent ones, with all the administrative costs that this entails.127 The benefit of this extra expense is not a fairer distribution of the costs of nonnegligent accidents—just a different distribution. Holding constant the incidence of injury,128 negligence liability may be to the long run advantage even of those most disadvantaged by a negligence regime, namely, the victims of nonnegligent accidents. They fare better bearing such losses and sharing in the extra resources saved by a system of negligence liability than they do under a system of strict liability. Under strict liability, they stand an equal chance of bearing an equivalent loss (in the form of liability) and forego their share of the extra resources that negligence makes available. In a world of uninsurable risks, then, negligence liability seems preferable for reciprocal risks. The most general argument of this Article has been that fairness finds its natural expression not in the reciprocity of risk criterion, but in enterprise liability. The fair distribution of harm is more important than the fair distribution of risk and—in a world of insurable risk— enterprise liability usually distributes harm more fairly than negligence liability does. The argument assigns pride of place to
126. Fletcher, supra note 4, at 547. 127. On the other hand, strict liability may be cheaper to administer in each case, because it dispenses with determinations of fault. Just how these offsetting tendencies play out is unclear. For purposes of the argument, I shall assume that negligence is cheaper because fewer cases will be brought. 128. If strict liability reduces the incidence of injury, things may be different. Because nonnegligent accidents strike like lightning in a world of uninsurable risks, and because victims and injurers are the same people when risks are reciprocal, there is some reason to doubt that strict liability will induce greater safety. Strict liability and negligence will leave the costs of nonnegligent accidents on the same people: strict liability in those persons’ capacities as injurers, negligence in their capacities as victims.

2004]

THE LAW OF ACCIDENTS

1917

considerations of fairness in our thinking about the design of tort liability; they are placed in the center of our “deliberative field.”129 The argument of the next and final part of the Article is that considerations of fairness are not the only considerations operating within that field. Other considerations, normative or practical, may lead to the displacement of the enterprise liability generally favored by considerations of fairness. The “all things considered” best liability regime for any particular context will depend on the interaction of general considerations of fairness with any special normative and practical considerations applicable to that context. One consequence of this is that substantial domains of fault liability will persist even if we make the ideal of fairness the principal ideal of our tort accident law, and even if that ideal generally favors enterprise liability. C. Competing Normative Considerations In some cases, competing normative considerations (perhaps in conjunction with practical problems in the construction of attribution rules) defeat the common law institutionalization of enterprise liability. Property rights are an interesting case in point, in part because they can cut both for and against enterprise liability. In certain circumstances, the presence of property rights increases the attractiveness of enterprise liability and facilitates its administration.130 Real property rights increase the attractiveness of enterprise liability when accidents arise from the overflow of one landowner’s activities onto another’s. Ownership of real property confers special freedom of action within the property’s boundaries, and sharpens the boundaries between zones of activity. Within the boundaries of their properties, owners and occupiers are free to build reservoirs and keep wild boars, even if these activities impose abnormally great risks of injury.131 When boars run wild and reservoirs burst, however, owners and occupiers are justly subject to enterprise liability. The special freedom conferred by property rights includes the freedom to subject one’s own property to unusually great risks. Overflowing the boundaries of one’s property and imposing the cost of one’s activities on strangers, by contrast, is the very essence of unfairness. By helping to locate the boundary between the
129. For the idea of a “deliberative field” within which all considerations of practical reason must be unified, see Barbara Herman, The Practice of Moral Judgment 152, 182-83, 196-202 (1993). 130. See William K. Jones, Strict Liability for Hazardous Enterprise, 92 Colum. L. Rev. 1705, 1729, 1757, 1779 (1992) (emphasizing how strict liability can protect various zones of activity, including ones defined by property rights, from intrusion). 131. See Marshall v. Ranne, 511 S.W.2d 255 (Tex. 1974) (holding defendant strictly liable for injuries inflicted by his vicious hog when it escaped from his property and injured his neighbor); Rylands v. Fletcher, 3 L.R.-E. & I. App. 330 (H.L. 1868) (appeal taken from Ex.) (holding Rylands strictly liable for harm to Fletcher’s mines inflicted by the escape of water from his reservoir).

1918

FORDHAM LAW REVIEW

[Vol. 72

permissible use of one’s own resources and the impermissible appropriation of another’s resources, the presence of property rights increases the attractiveness of enterprise liability. By sharpening the often elusive boundary between enterprises, property rights also facilitate the practical administration of enterprise liability. In Bushey, for example, the plaintiff’s ownership of the dry dock plays a critical role in identifying the boundaries between its enterprise and the Coast Guard’s.132 The plaintiff’s and the defendant’s enterprises are intimately and beneficially intertwined in Bushey. Remove the plaintiff’s property rights from the scene and the boundary between injurer and victim enterprises becomes much more difficult to establish. Seaman Lane’s trespass disappears. Property rights tend to diminish the attractiveness of enterprise liability, however, when victims suffer injury in the course of entering onto injurers’ land. The risks to which entrants on others’ land are exposed do not arise out of the voluntary agency of injurers in the straightforward way that normal accidents among strangers do. Highway accidents involving the abnormal risks of gasoline tankers, for instance, arise out of a voluntary decision made by those who own such vehicles to expose other drivers to this abnormal risk; a decision made in pursuit of profit. Voluntary exposure of others to risk in pursuit of one’s own ends is a normal and morally significant feature of those accidents to which enterprise liability most readily attaches.133 Landowner liability, by contrast, emerges out of the victim’s entry onto the owner’s or occupier’s property. This affects the initial distribution of burden and benefit and the foreseeability of the victim’s presence at the scene of the injurer’s activity. The entry of the victim onto the scene of the injurer’s activity also makes it more difficult, both conceptually and practically, to locate the boundary between the injurer’s and the victim’s enterprises. Entrants onto land—unlike victims on sidewalks or persons standing in the doorways of their own homes134—are not legally
132. Ira S. Bushey & Sons, Inc. v. United States, 398 F.2d 167, 172 (2d Cir. 1968). 133. See, e.g., Rylands, 3 L.R.-E. & I. App. at 342. The Defendants, in order to effect an object of their own, brought on to their land . . . a large accumulated mass of water, and stored it up in a reservoir. The consequence of this was damage to the Plaintiff, and for that damage, however skillfully [sic] and carefully the accumulation was made, the Defendants . . . were certainly responsible. Id. (Lord Cranworth, J., concurring) (emphasis added); see Clare Dalton, Losing History: Tort Liability in the Nineteenth Century and the Case of Rylands v. Fletcher (unpublished manuscript, on file with the author) (setting out the moral foundations of “activity based liability”). 134. See Shipley v. Fifty Assocs., 106 Mass. 194, 199 (1870) (addressing a claim brought by a plaintiff who walked on a public sidewalk and was struck by falling ice and snow that had accumulated on defendant’s peaked roof); Tuchkashinsky v. Lehigh & W. Coal Co., 49 A. 308 (Pa. 1901) (addressing claim brought by a plaintiff who was standing in the doorway of her father’s house, 700 feet from defendant’s

2004]

THE LAW OF ACCIDENTS

1919

entitled to be where they are absent special authorization from the owner or occupier of the land. The circumstances of their entry affect both the distribution of burden and benefit between injurer and victim, and the extent to which the potential victim’s presence should be anticipated and her safety ensured. Felony trespassers, for instance, enter the injurer’s land without permission, seeking to appropriate a benefit to themselves and to reap that benefit by inflicting injury upon the owner or occupier of the land. This relationship of burden and benefit is enough to suspend the imposition not just of enterprise liability, but of ordinary negligence liability as well. Burden and benefit are arranged differently—they are presumptively mutual—when the potential victim’s presence on the property is authorized. The authorization of the potential victim’s presence also makes that presence expected, not just foreseeable. By authorizing a potential victim to enter the property, the owner or occupier of land assumes an affirmative obligation towards him.135 These are all morally significant features of the relationship between potential injurer and potential victim, and they bear upon—though they do not determine—the design of tort duties. The present state of landowner liability to entrants onto land, indeed, shows how the presence of property rights and relationships can affect the articulation of tort duties in a way that presses against enterprise liability, for both practical and normative reasons. Enterprise liability has never gained a foothold here. Negligence liability vies with an older regime of variable duty drawn from property law, a regime that distinguishes among entrants and the duties owed to them on the basis of their status, where gradations of status roughly reflect the extent of the entrant’s permission to enter and the degree of benefit the entrant confers on the occupier. Enterprise liability faces two daunting challenges. On the one hand, there is a practical problem of attribution. The victim’s entrance onto the injurer’s land makes the agencies of injurer and victim difficult to disentangle. On the other hand, the victim’s entry onto the injurer’s land raises normative questions about the distribution of burden and benefit. The normative implications of the victim’s entry are implicated in a range of cases. They come into play when potential victims enter onto land with criminal intentions; when they enter seeking commercial benefit; and when they enter in pursuit of particular risky recreational experiences. (In this circumstance, landowner liability takes on the mine, and was harmed by the concussion from a blast caused when lightning ignited explosives stored at the mine); Bolton v. Stone, 1951 A.C. 850 (appeal taken from C.A.) (addressing a claim brought by a plaintiff who was struck by a cricket ball that escaped from a nearby cricket field while he was walking on the street). 135. See Francis H. Bohlen, The Basis of Affirmative Obligation in the Law of Tort, 53 Am. L. Reg. 209, 273, 337 (1905).

1920

FORDHAM LAW REVIEW

[Vol. 72

essential feature of “primary assumption of risk” cases.136) The difficulty of disentangling injurer and victim activity is practically significant when that interpenetration of activities makes it difficult to attribute harm to one or the other enterprise. Other things equal, the attractiveness of negligence liability increases in step with the difficulty of disentangling victim and injurer activity. Fault criteria can attribute harms to one party or another in circumstances where enterprise liability rules are unable to do so. There is, moreover, a wide range of circumstances in which practical problems of attribution and competing normative considerations may, separately or jointly, rebut the thrust of even a common law predisposed to enterprise liability. (It is worth recalling that even the vicarious liability of employers for the torts of their employees—the oldest common law source of enterprise liability— itself tends to incorporate a substantial element of negligence. It attributes the torts of employees to a firm, and those torts are most often negligence.) The set of possible circumstances and legal norms that might affect the articulation of tort norms is open-ended. We cannot say in advance what fairness calls for when tort norms interact with a circumstance we have never considered. Consider, very briefly, two further circumstances. Within a community of risk, the special character of an activity can justify retreating from strict liability and, indeed, even relaxing negligence duties. The character of an activity can require a higher than normal level of risk. In such cases, risk is essential to the enjoyment of the activity; reduce the risk and you destroy the activity. Sports are the preeminent example here. Making every injury characteristic of the sport an occasion for liability would undermine professional football. Indeed, taking a traditional negligence approach and making every injury inflicted in violation of the rules an occasion for the imposition of liability would cast a pall over the play of the game. For these activities to flourish, it is necessary to retreat from our normal conceptions of responsibility to prevent and rectify harm done. In tort doctrine, this is the domain of “primary assumption of the risk” and “relaxed duty.”137
136. See infra text accompanying note 137. 137. See, e.g., Crawn v. Campo, 643 A.2d 600 (N.J. 1994) (suspending duty of ordinary care in the recreational sports context, and adopting a recklessness standard); Scott v. Pac. W. Mountain Resort, 834 P.2d 6, 13 (Wash. 1992) (applying the doctrine of “primary assumption of risk,” which relieves prospective injurers of their duty of ordinary care, to skiing). Note, though, that negligence liability wreaks more havoc with such activities than strict liability does. Negligence liability criticizes conduct. Unless its ordinary strictures are relaxed, it tends to criticize the normal conduct of contact sports. Enterprise liability in the form of mandatory loss insurance among those engaged in the sport might well be desirable. Participants in these activities might well prefer receiving compensation for out-of-pocket expenses incurred as a result of injuries received in the ordinary course of such activities to suffering severe injury without any compensation. But the law of torts is not in a

2004]

THE LAW OF ACCIDENTS

1921

To take just one more example, consider product accidents. The paradigm product accident, for example, arises out of a contractual relationship between injurer and victim—the relationship of buyer and seller. Contract norms therefore interact with tort ones, and the significance of the victim’s participation in the enterprise that injured her must be appraised. This need not lead to the displacement of enterprise liability by fault liability; liability for breach of contract has a strict element, and the body of doctrine leading into and developing out of Section 402A of the Second Restatement had a strong enterprise liability cast.138 But it does present enterprise liability with a significant practical challenge: How to go about distinguishing between injurer and victim activities in this setting? Injurers typically participate in the genesis of product accidents by manufacturing, designing, and marketing the instrumentality of injury. Victims participate by purchasing and using that instrumentality. The design of enterprise liability rules must take these facts into account. The “characteristic risks” of choosing and using a particular product must be distinguished from the “characteristic risks” of manufacturing and marketing that product. Under enterprise liability, the former risks belong on the victim who purchases and uses the product, whereas the latter risks belong on the injurer who produces and markets the instrument of injury. Devising adequate rules is a formidable practical challenge. The partial resurgence of negligence conceptions in product warning and design defect law, so that negligence and enterprise liability conceptions now compete in the law,139 may reflect, in part, the difficulties involved in instituting enterprise liability. Even with these limits and exceptions, the phenomenon of enterprise liability is a remarkably important one. We live, as Holmes long ago saw, in a world of vast, organized enterprises and systemic risk—in a “world of activities”—not in the world of “isolated, ungeneralized wrongs” out of which “[o]ur law of torts comes.”140 In our world, enterprise liability generally distributes the costs of accidents more fairly than negligence liability does. When enterprise liability is feasible, we have strong reasons of fairness to favor it over negligence liability. It is a tribute to the power and fertility of John Rawls’s work that it enables us to see more clearly, and comprehend more deeply, these reasons. position to implement this sort of enterprise liability, because it is not in a position to adjust the normal tort measure of damages. 138. George L. Priest, in The Invention of Enterprise Liability: A Critical History of the Intellectual Foundations of Modern Tort Law, 14 J. Legal Stud. 461 (1985), overstates this truth and mistakenly identifies strict liability with absolute liability, but captures vividly the enterprise liability aspirations of Prosser and other “founders” of modern product liability law. The doctrines noted in the previous footnote reflect these aspirations. 139. See Tort and Accident Law, supra note 1, at 951-68, 974-1026. 140. Holmes, supra note 64, at 183.