The Role of the Courts in Risk Management

Richard B. Stewart

Richard B. Stewart is Byrne Professor of Administrative Law and Associate Dean of the Harvard Law School.

My topic is a large one, and I cannot hope to cover it in detail. Instead, I shall offer a sketch of an idealized environmental risk management system and ask how the courts have contributed or might contribute to its realization. I shall focus on three specific contexts in which courts participate in environmental risk management: review of administrative action, particularly of federal regulatory action (or inaction); actions in the first instance for injunctive relief against claimed environmental hazards; and actions for compensatory damages. Although the bulk of experience to date has been in areas other than nuclear power, hazardous waste cleanup, toxic torts and biotechnology, I shall cite occasional examples from these fields to illustrate the extent of the contribution we can expect realistically from the courts in dealing with problems of environmental risk management.

An Ideal Risk Management System

Our institutions, legal and otherwise, ought to be judged by their fitness for making and carrying out wise social choices. In the area of environmental risk management, such choices should, ideally, respond to the following five principles:

1. Subject to resource constraints, institutions should encourage timely and accurate risk and risk reduction assessment. We confront a portfolio of risks to human health and natural systems, generated by industrial processes, products, and their residuals. The risks vary in magnitude according to the potency of the risk mechanism, the severity and extent of exposure, the size of exposed populations, and other factors. The function of risk assessment is to measure the relative as well as absolute magnitudes of the risks contained in the portfolio.

The risks also vary according to the extent to which they can be reduced by a given commitment of societal resources. The extent of reduction depends upon the availability and efficacy of technology for reducing the generation of risks or for avoiding their adverse consequences; it also depends upon opportunity costs. The function of reduction assessment is to determine the magnitude of the resources needed to produce a given reduction in risk.

2. The overall goal of risk management should be the progressive reduction of the risk in the existing portfolio.

Inevitably, there will be disagreement as to how ambitious our ultimate goal should be and how quickly we should attempt to achieve it. Most who have considered the matter would agree that, while zero risk is generally neither feasible nor desirable, our risk reduction goal should be more ambitious than that which would be dictated by any contemporaneous cost-benefit calculation. While the decision as to how far and how fast to reduce is ultimately a political judgment, in an affluent society we should be able to move steadily towards an overall reduction of risk.

Of course, risks differ qualitatively, and cannot be quantified or summed up in any mechanical way. For example, human health risks differ according to whether they are assumed voluntarily, according to the magnitude of individual (as opposed to aggregate) risk, according to whether they are life-threatening, and so forth.¹

The definition of risk portfolio objectives involves decisions concerning the overall risk reduction goals, the rate of such reduction, and the allocation of priorities among different qualitative types of risk. Institutions should promote informed public and governmental consideration and selection [16 ELR 10209] of risk portfolio objectives.

3. Institutions should manage risk to achieve risk portfolio reduction objectives in the most cost-effective manner. To reduce waste and to secure greater and faster progress in achieving risk portfolio objectives, priority shoul be given to those risks that will yield large reductions in risk at relatively low economic and administrative cost. The priorities should be adjusted to take into account the qualitative differences in risks that have been recognized in risk portfolio objectives. Priority risks should not necessarily be controlled to the maximum extent economically and technologically feasible, as this might involve a waste of scarce resources that could be used better elsewhere. The extent of control should be guided by consideration of cost-effectiveness.

4. Institutions should facilitate an appropriate portfolio turnover strategy. The overriding goal of risk reduction implies that the creation of new risks should be prohibited or minimized. This objective, however, should be pursued on a net-balance basis. The creation of a new risk by a new product or process may be acceptable, because it displaces an existing product or process that generates a substantially greater risk. A strategy of minimizing new risks, viewed in isolation from existing risks, may discourage new investment and a turnover of the portfolio that might substantially decrease net risk.

5. A risk portfolio strategy must deal with the question of compensation. Under prevailing tort law, those creating environmental risks ("defendants") are presumptively liable to those injured by reason of exposure to the risk ("plaintiffs") for the injury suffered. Liability theoretically serves the following purposes: deterrence; risk-spreading; and vindication of the right not to suffer harm as a result of involuntary exposure to risk.

The presumption favoring compensation, however, may be overcome where: 1) the transaction of providing defendant-plaintiff compensation are high; 2) alternative incentive systems are available; 3) the gains from risk-spreading are low or alternative means exist to achieve risk-spreading; 4) the risk is voluntarily assumed or could be avoided more readily and cheaply by the plaintiff; 5) insurance against risk can be more cheaply or appropriately provided by plaintiffs or government; or 6) provision of defendant-plaintiff compensation results in serious adverse side-effects (e.g., overdeterrence). Institutions should respond to these considerations in decidinghow and when to provide compensation.

Of course, our world is far from ideal. Information and transaction costs make it difficult for the government to acquire and analyze data, determine social objectives, and achieve the coordinated management necessary to carry out this ideal program. In addition, the need to maintain citizen confidence in government may require that a low-priority but highly-publicized risk be controlled before a more serious but less publicized risk. It seems appropriate, nonetheless, to inquire how far our institutions — including our legal institutions — can encourage and enable us to choose and achieve risk portfolio objectives that accord with the principles sketched above.

Institutional Implications of the Portfolio Approach: The Need for Administrative Management

The basic institutional implication of a portfolio approach to risk management is clear: primary responsibility for managing risk must be given to administrative agencies.

The portfolio approach to risk management requires consistency and coordination in decisionmaking to achieve risk reduction in a rational and cost-effective manner and to ensure that similarly situated, competing generators of risk are treated equally. Administrative agencies (in contrast to both courts and legislatures) are centralized and specialized. Accordingly, they can achieve a greater degree of consistency and coordination and are bettern suited to serve as the frontline mechanism for regulating risk.

The reasons leading to this conclusion are essentially the same as those which led to the creation of administrative commissions to regulate railroad rates in the nineteenth century. Effective regulation was then needed to ensure that the overall revenues generated by hundreds of different rates were sufficient to cover the carriers' costs without providing monopoly profits, and that similarly situated, competing carriers and shippers were treated equally. Since a decentralized judicial system of case-by-case adjudication could not offer the consistency and coordination necessary for effective regulation, administrative agencies were established to accomplish these goals.²

In addition to centralization, administrative management of portfolio risk offers advantages of specialization: it allows for economies of scale in handling similar cases, use of informal, nonforensic fact-finding and decisionmaking procedures, and cultivation of expertise that would be exceedingly difficult or expensive to develop through the courts.

Agencies can also initiate actions on behalf of the collective interests of large numbers of persons exposed to risks that might be small from the perspective of any individual but large when viewed in the aggregate. The judicial system, on the other hand, depends on private initiatives that may not be forthcoming because of the small economic stake of any individual affected by industrial practices. Furthermore, legislatures are often unwilling or unable to initiate action because of increasingly crowded agendas and relatively fixed decisionmaking capacities.

The argument favoring administrative management of portfolio risk is strong. Congress gives primary responsibility to regulatory agencies rather than to the courts. Case-by-case adjudication cannot ensure the achievement of an appropriate overall level of risk reduction, nor can it ensure that the most significant and readily remediable risks are addressed first. Persons similarly situated might end up being exposed to quite different levels of risk, and firms generating similar types of risk might end up facing quite different cleanup obligations.

If the administrative regulatory process were to approximate the ideal risk management strategy described earlier, the role of the courts might be limited to reviewing the decisions of administrators who manage risk. Ideally, one would expect Congress to create one or more administrative agencies with adequate resources and jurisdiction to carry out a portfolio approach to risk management. One would also expect Congress to specify with reasonable clarity the risk [16 ELR 10210] portfolio objectives that the agency should pursue. In this ideal world, the courts' task might be limited to ensuring: 1) that agencies follow procedures that generate adequate risk assessment and reduction assessment information; 2) that agencies base their decisions on findings of fact reasonably supported by evidence; and 3) that agencies adhere to congressional instructions regarding portfolio objectives.

Of course, our world is not this one, and agencies are quite far from achieving any such idealized system of risk management. Congress creates many different agencies and statutory programs for managing the same or similar risks. The risk management objectives set forth in the various statutes are vague, unrealistic, and often inconsistent. Agencies do quite a poor job of risk management. Administrative risk assessment and reduction assessment is often scanty, careless, or worse. Both quantitative and qualitative differences in risk receive little or no explicit attention. Cost-effectiveness concerns are largely ignored. As a result, there are shocking disparities in the treatment of risks even within a single program or agency.³ Trivial risks are regulated at great exense, while far more serious risks reducible at far less cost are ignored. Striking disparities between programs and agencies also exist.⁴ The setting of priorities of risk often seems largely a function of media exposure and political attention and usually bears little relation to sensible risk management policies. As a result, regulatory agencies enjoy an exceptionally low degree of confidence among industry, environmental groups, unions, and the public.

Given our imperfect world, what contributions (if any) can the courts make towards the advancement of sensible risk management principles when reviewing administrative decisions? In addition, what contribution can the courts make when deciding controversies over risk in the first instance? For example, where administrative remedies are unavailable or inadequate, courts may be asked to regulate risk through injunctive remedies. Even where a system of administrative regulation is fully operational, agencies generally lack the power to award compensation. Accordingly, courts may be asked to provide such compensation through liability awards against defendants.

The Court's Role in Reviewing Administrative Management of Risk

Reviewing courts must first determine the directives contained in the relevant statutes and ensure that agencies comply with them. This task is especially difficult in the area of risk management, as many of the statutes rest on false or obsolete premises. For example, many statutes are based on the view that it is possible to distinguish "safe" from "unsafe" pollution or exposure levels when, in fact, no clear threshold exists.⁵ Other statutes appear to suppose that it is possible to eliminate or minimize the existing portfolio by imposing maximum feasible control methods simultaneously on all risks, when, in fact, enormous administrative and economic obstacles block such a course.⁶ Both of these approaches constitute pious fictions in the sense that they ignore the true nature of the risk management problem. They are "pious" in that, presumably, they are well-intentioned; they are fictions in that they deny the trade-offs and portfolio mixes inherent in the risk management problem.

Courts are not at liberty simply to ignore these pious fictions and to substitute their own notions of sound policy. Courts cannot, for example, require that standards be established by weighing costs against benefits where Congress has made clear that standards are to be based on health considerations alone. There is no reason, however, why health-based standard setting should not proceed from a portfolio perspective, particularly in a world without clear thresholds. Such a perspective would not only allow, but would also encourage, administrators to ask themselves the following questions when setting standards for a given risk: (a) how serious is the given risk in relation to other risks; and (b) to what extent will stricter or less stringent regulation of this risk consume administrative and compliance resources that might better be directed against other risks? Courts can thus make a substantial contribution by exposing the true nature of the risk portfolio problem and by dispelling some of the erroneous assumptions upon which current statutes are based.

A portfolio perspective is not necessarily inconsistent with existing statutes. As a linguistic matter, the establishment of standards to "protect the public health" under § 109(b)⁷ may include a comparative assessment of how the limited resources of agencies and the private sector should best be allocated to further the overall goal of risk reduction.

A portfolio perspective can be accommodated even more readily within statutes such as OSHA,⁸ which call for control of risks to the extent that control is "feasible" or that technology is available. It is generally accepted that "feasibility" and technological "availability" must take into account the economic capability of the regulated industry to bear the costs involved. Once it is admitted that there is no bright line between "feasible" and "infeasible," the question becomes one of degree. We must decide just how far out along the cost-control curve we want to move, and what price — including capacity shutdowns and capital and employment losses — we are willing to pay. This question can be resolved intelligently only from a perspective of risk portfolio.

Moreover, if an industry generates more than one risk — as is usually the case — adoption of a portfolio perspective will prevent dissipation of resources on the first risk that happens to be regulated.⁹

The existing statutory scheme does not impose a rigid straitjacket. In most cases (perhaps all) the statutory language or legislative history allows some room for a portfolio perspective to operate. With some creativity on the part of administrators and courts, it should be possible to build [16 ELR 10211] elements of a portfolio approach into the existing statutes. In the past, courts have interpreted statutes creatively to promote environmental values.¹⁰ There is no reason in principle why they should not also do so in the area of risk management. Instead, courts have seized all too often on a myopic literalism that takes Congress's pious fictions at face value.¹¹ Judges may adopt this myopic approach believing it necessary to foreclose executive use of open-ended cost-benefit techniques that could seriously compromise risk reduction goals.¹²

A risk portfolio strategy, however, differs fundamentally from a cost-benefit strategy. Under a portfolio strategy, risk reduction goals are chosen not by cost-benefit analysis but by much broader consideration of relevant social values. In addition, the prioritizing of risks is accomplished, not by an isolated monetary quantification of the costs and benefits of regulating a particular risk, but by a comparison of that risk with other potential candidates in the portfolio. In practice, one can restrict the discretion allowed an agency under a risk portfolio approach, by requiring the agency to justify less than full-fledged regulation of a given risk by showing that such action is consonant with an operative strategy for control of other, more suitable candidates elsewhere in the portfolio. Such a strategy could be embodied in regulations or guidelines.

In lieu of a risk portfolio strategy, current public policy largely reflects a primitive best available technology (BAT) approach: whatever risks happen to rise to the top of the agenda are controlled to the extent that technology is "available" to do so and to the extent that the industry generating the risk can "afford" to do so. This approach appeals to the interests of regulated firms in its predictability, to the interests of agencies in its administrative practicality, and to the interests of environmental groups in its encouragement of progress without shutdowns.

Despite its attractions, the BAT approach has been fostered, and in some cases mandated, by an unimaginative judicial approach to statutory interpretation.¹³ The BAT approach has several serious flaws. It discourages portfolio turnover by penalizing new investment and development of more profitable industries.¹⁴ It also produces inactivity in the standard-setting process. If an agency decides to regulate a risk, the BAT approach may require it to impose quite rigorous controls that consume enormous resources. As a result, the agency must limit sharply the number of substances that it regulates, and the selection of these few substances is often quite arbitrary.¹⁵ As BAT simply ignores the problem of priorities, it provides no assurance that we will devote resources to those risks that will yield the biggest gains in risk reduction for the least investment.¹⁶ It is not surprising, therefore, to encounter extraordinary variations in cost effectiveness in existing programs.

The problem of priorities has two aspects: Overregulation and underregulation. First, an agency may overregulate a low priority risk that should not be regulated at all, or that should be regulated to a far lesser degree. In extreme cases, a court may strike such overregulation as arbitrary without undertaking a comparative assessment of other risks in the portfolio. This is essentially the approach of the Supreme Court plurality in the Benzene case.¹⁷

In most cases, however, a comparative assessment will be necessary. One way to make sure it is performed is to require a regulatory agency to establish — in response to a prima facie showing by industry that a given regulation is grossly cost-ineffective — that the regulation is reasonably cost-effective in relation to other risks in the portfolio for which the agency is responsible.

Judge Leventhal rejected this approach in the first Portland Cement case, concluding correctly that it would give industry boundless opportunities for delay.¹⁸ To the extent, however, that an agency has adopted, through regulations or guidelines, an operative risk portfolio strategy, a regulated firm may properly demand that a particular regulatory initiative be consistent with such a strategy.

An agency may underregulate by failing to control risks that are comparatively substantial and that can be reduced readily and cheaply.¹⁹ The remedy here is the private right of initiation, codified in some statutes in citizen suit provisions. These provisions enable an environmental group or other suitable plaintiff to obtain judicial review of the agency's failure to act.²⁰ There may be sins of omission so gross that a court may properly mandate enforcement without a comparative [16 ELR 10212] assessment of other risks in the portfolio. Such cases, however, will be relatively rare.

A seemingly less instrusive judicial solution is to mandate agency initiation of formal decisional proceedings once a prima facie case for control has been presented. This remedy, pioneered in EDF v. Ruckelshaus (DDT), has the practical effect of dictating agency enforcement priorities.²¹ Without a portfolio perspective, courts' efforts at "action-forcing" can squander resources on low-priority risks.²²

A more appropriate solution would be to require the agency to show that its failure to regulate or to regulate more rigorously in a given case is reasonably consistent with a risk portfolio strategy. This requirement could provide substantial incentives for agencies to formulate and adopt such strategies. While some courts have taken preliminary steps towards instituting such a requirement, the Supreme Court's recent decision in Heckler v. Chaney²³ may take the steam out of these efforts. By making non-reviewable many agency decisions not to initiate controls, the decision cuts off a potential avenue for promoting more cost-effective portfolio strategies.

In practice, the extent to which courts can stimulate better portfolio management across agencies or across different statutory programs within the same agency is extremely limited. Even when a single agency has responsibility for regulating various risks under a single statute, the obstacles are great. In order to promote better portfolio management, courts must apply "hard-look" review not only to risk assessment but also to reduction assessment data, analyses, and decisions. As muchof this assessment is already prepared routinely by agencies in response to OMB review under Executive Order 12291 the additional burden placed on agencies may not be great; the burden placed on judges, however, may be considerable.

A more fundamental objection to the strategy outlined above is that it advocates judicial imposition of a policy framework more appropriately determined by the political branches of government. The problem, however, is that we have been unable to rely on either Congress or EPA to face up to the inevitable trade-offs involved in risk management or to reveal the true nature of the problem for public understanding and debate.

The conspicuous failure of the political branches to develop a coherent or successful strategy for risk management suggests that judicial leadership in this area would be welcome. To date, however, the courts, as much or more than other institutions, have fallen down on the job. They have failed us by taking a rather wooden and deferential approach to risk management. If we look at what the courts have accomplished in antitrust law, labor law, corporation law and administrative law, all of which are statutory-based systems, we see them imposing judicially created policy frameworks. These frameworks would, I believe, be judged on the whole a considerable success.

Regrettably, the courts have declined to assume a similar leadership role in environmental law. If judicial initiatives in risk management are not forthcoming — and such initiatives are unlikely, given the Supreme Court's current hostility to trans-statutory judicial lawmaking,²⁴ — full-scale judicial retreat may be a better alternative than the present practice.

James Landis made a powerful argument that the courts should get out of the business of superintending such agency management decisions as those involved in risk portfolio regulation.²⁵ The ability of adversary proceedings to promote high-quality risk and reduction assessment and to improve risk management is subject to serious doubt. Unless courts are willing to assume a larger responsibility for risk management strategies, they might better withdraw from the field entirely and limit themselves to correcting plain statutory or procedural violations. Such withdrawal would make it clear that one must look elsewhere — to the agencies themselves, to OMB, or to new institutional arrangements — for major improvements in the currently defective system of risk management.²⁶

Judicial Management of Risk Through Injunctive Relief

Courts enjoy wide residual remedial jurisdiction to deal with environmental risk. State courts have common law powers, and federal courts often have jurisdiction to make federal common law. Nonetheless, when a plaintiff brings an action for injunctive relief against a risk, and a fully functioning administrative system is capable of providing such relief, courts will generally decline to entertain the action, invoking principles of statutory preemption or primary jurisdiction. Given the functional imperatives favoring administrative management of risk discussed earlier, this judicial response is both understandable and appropriate.

When the legislature, however, has not yet created a system of administrative management for certain risks, or when the current system is complete, not fully operative, or otherwise inadequate (as with interstate pollution spillovers), courts must decide whether to exert their residual powers despite their previous retreat from the field. Contemporary problems of hazardous waste cleanup and field applications of new biotechnologies exemplify this predicament.

We have already noted the structural factors that make it extraordinarily difficult for courts to develop and implement a portfolio risk management strategy through case-by-case adjudication. Courts have, nonetheless, acted to remedy significant but currently unregulated risks, by requiring adoption of the best available and affordable methods for identifying and controlling risk. This approach — the judicial equivalent of a BAT strategy — was pioneered in air and water pollution control cases²⁷ and has been carried forward [16 ELR 10213] in more recent years²⁸ in the toxic waste and biotechnology cases.

Problems arise, however, in identifying the threshold of risk significance that will trigger a BAT remedy, and in the tendency to apply a full battery of BAT controls once a risk is found significant can lead to overregulation. With increased understanding of risk management, one would hope for greater judicial sophistication in determining the magnitude of risk, identifying the range of possible responses, and selecting, by analogy to risk management in other contexts, an appropriate level of response (sometimes short of full BAT), pending development of a full-fledged administrative scheme that will define and execute a more coordinated and consistent portfolio risk strategy. Given their institutional limitations, the courts' rather primitive BAT response to the individual cases before them is understandable. What is dismaying, however, is the seeming inability of the administrative system to move beyond this quite primitive response, despite its vastly superior institutional capabilities.

Actions for Compensatory Damages

Environmental regulatory agencies generally do not have the authority to award damages to persons injured by reason of exposure to risk. Persons claiming such injury must therefore invoke the residual jurisdiction of the courts when seeking damages from the defendants assertedly responsible for the risk.

As already indicated, courts should presumptively provide compensation in such cases.²⁹ The argument for compensation is especially clear in the case of "signature" injuries that can be linked reliably to a particular chemical or other mechanism to which the plaintiff was exposed. Causation problems may still arise in "indeterminate defendant" cases, where several generators of the risk caused plaintiff's injury and it is not known which of them was responsible for the particular exposure that resulted in plaintiff's injury. This problem, however, can be resolved through techniques of joint or proportional liability.³⁰

As the studies of asbestos litigation indicate, the transaction costs of providing compensation for signature torts through the judicial system are quite high. Rising concern over these costs has impelled a search for alternatives. There are two basic methods to reduce such costs. The first is to reduce legal uncertainty. Many lawyers for both plaintiffs and defendants in environmental tort actions, perceiving trials of scientific and technical issues to be little better than lotteries, have expressed strong interest in informal processes for resolution of factual questions. Parties, especially repeat players (a group that can include plaintiffs' lawyers representing a large number of similar claims) are, nevertheless, often reluctant to negotiate settlements, when the applicable legal rules of liability and contribution are uncertain. Until judicial decisions clarify the law, parties will tend to litigate rather than to settle cases. While there is no easy solution to the problem of legal uncertainty, the creation of a specialized court constitutes one possible approach.

The second method of reducing transaction costs is to achieve economies of scale through specialized processing of large numbers of similar claims (by a single centralized institution). Such specialized processing permits routine disposition of recurrent factual and legal issues, as well as the use of more informal procedures. Specialized processing can be achieved by creating an administrative agency along the model of Social Security Disability of Workmen's Compensation to adjudicate claims, or by encouraging class actions and litigation consolidation procedures in the courts. In the case of signature pollutants, such changes would help to reduce transaction costs dramatically. In the asbestos litigation, only thirty-five to forty percent of each dollar spent by society currently reaches the actual claimant. By contrast, the Social Security claimant receives approximately ninety-seven cents of each dollar spent.

Experience indicates that the administrative alternative is often cheaper. In order to take full advantage of such costs savings, however, one would have to create an agency with general jurisdiction over a wide range of claims, e.g., over all environmental tort claims. An agency with a more specialized jurisdiction would be created only after claims of a particular sort (asbestos claims, for example) had clogged the court system, signalling the existence of a serious processing problem. By then, however, it would already be too late to take adequate advantage of the administrative alternative.

Creation of an agency with general jurisdiction over compensation claims for environmental harms would be extremely controversial, particularly if there were limitations on the size of damage awards. Moreover, a high percentage of such agency decisions would probably be appealed to the courts, undercutting further the cost savings of an administrative system. Given these drawbacks, creation of a standing judicial procedure to encourage specialized processing of claims within or adjacent to the court system may be the best practical solution.

Non-signature environmental torts present additional difficulties that are currently the subject of lively academic and professional debate. Essentially, the problem is that of the "indeterminate plaintiff." The problem arises when the plaintiff has been exposed to numerous "background" sources that create the risk of a particular disease, such as cancer or cardiovascular failure. These background sources may include diet, smoking, stress, etc. The defendant has created an additional source of risk of the same disease (e.g., air pollution) to which plaintiff has also been exposed.³¹ When a person thus exposed contracts the disease, there is no way of knowing which source of risk caused the disease.

Assume, however, that it is possible, on the basis of epidemiological or other studies, to estimate the increased risk above background that the defendant has created. While the estimate may be subject to considerable uncertainty, assume further that the marginal increase in risk contributed by defendant is substantially less than fifty percent of the aggregate risk (background plus defendant's contribution). In such a case, the plaintiff is "indeterminate" in the sense that there is no way of telling whether he or she is among those whose disease was caused by defendant, or among those whose disease was caused by background exposure. All one can say is that the probability is something less than fifty percent that the plaintiff, as a member of a group exposed to [16 ELR 10214] both sources, contracted the disease as a result of exposure to the risk generated by the defendant, and that the probability is something greater than fifty percent that it occurred as a result of exposure to background risk.

The above summary addresses the essence of many environmental tort controversies, where the risk contributed by the defendant is usually far less than fifty percent of the aggregate. A plaintiff who has contracted the disease in question brings suit against the defendant to recover all of the damages attributable to the disease. Under the traditional rule requiring plaintiff to show a greater than fifty percent probability that the injury was caused by defendant, a court, even if willing to permit proof of causation on a probabilistic basis, must grant judgment for defendant.

This rule of preponderance of evidence denies recovery to all plaintiffs, even though a substantial number (though less than fifty percent) contracted their disease as a result of the risk generated by defendant. This automatic denial of recovery undercuts the goals of the tort system. On the other hand, it would obviously be unjust to require defendant to provide full compensation to all plaintiffs. As Professor David Rosenberg and others have argued, the only logical remedy in this situation is proportionate liability.³²

Where a group of plaintiffs has already contracted the disease, the solution is to award each plaintiff damages equal to compensation for the disease, discounted by the probability that plaintiff contracted the disease as a result of exposure to the increased risk generated by defendant.³³

Where a group of plaintiffs has been exposed to risk, but no one has yet contracted the disease, the same principles of proportionate liability could apply. A fund could be created to cover medical surveillance costs and proportionate recoveries to those who eventually contract the disease. The settlement pioneered by Judge Weinstein and Special Master Feinberg in the Agent Orange litigations is one version of this approach.³⁴

Should courts abandon the traditional rule for one of proportionate liability? The question is a close one. Implementation of a proportionate recovery rule would involve substantial transaction costs. In most cases, available scientific knowledge and evidence could not determine even the approximate extent of additional risk over background contributed by the defendant. In a class action, millions of dollars could turn on the determination of the defendant's precise contribution. This uncertainty provides a powerful incentive for heavy investment in discovery and litigation by the parties. Moreover, such uncertainty could undermine the perceived fairness and rationality of the ultimate judgment.

The goals of the tort system are somewhat attenuated in the proportionate liability context and may be better served by alternatives to damage actions. The argument for risk-spreading is weakened when only a minor portion of the loss suffered by the plaintiff can be attributed to defendant. Moreover, why should we gear up expensive litigation machinery to spread that portion of the risk attributable to the defendant, while doing nothing to spread the risk attributable to background?³⁵ The ultimate solution may be some form of medical and disability insurance to spread the entire bundle of risks. The argument for rights vindication seems similarly attenuated when the particular injury suffered by plaintiff cannot be attributed to defendant and there is only a less than preponderant statistical probability linking the two.³⁶

While the prospect of proportionate damage awards would have a deterrent effect on risk generators, a regulatory system of risk management would offer an alternative means of providing incentives that at least potentially promises far lower transaction costs and greater risk portfolio rationality.³⁷ Regulatory programs, however, often lag behind the risks they are supposed to manage. They are established only after risks have been created and harm suffered. Damage awards, on the other hand, operate as a general deterrence system that can prevent such harms from occurring in the first place, or at least reduce their incidence. The choice between the preponderance rule and proportionate liability may turn ultimately on an assessment of the need for and efficacy of a system of general deterrence and of the transaction costs involved in operating such a system.³⁸

Some courts have attempted to develop an alternative to the preponderance rule on one hand and proportionate liability on the other. While shunning proportionate liability, they have nonetheless managed to provide recovery — and full recovery at that — for indeterminate plaintiffs, even though the proportion of aggregate risk attributable to defendant is substantially less than fifty percent. They have accomplished this remarkable feat by allowing such plaintiffs to recover on a showing 1) that the risk generated by defendant was a "substantial" factor in causing plaintiffs' injury;³⁹ 2) that defendant created a risk above background; or 3) [16 ELR 10215] some combination of these approaches.⁴⁰ The burden of production and persuasion then shifts to the defendant to show by a preponderance test that defendant did not cause plaintiff's injury; by hypothesis, this is an impossible burden to overcome.⁴¹

These judicial strategems to provide recovery simply disregard the hard reality that one cannot ascertain whether an indeterminate plaintiff's disease was caused by the defendant or by background risk; one can know only aggregate probabilities in the context of group exposures. These strategies proceed, nonetheless, as if it were somehow possible to know the "true" cause in a given case. The fictions upon which these strategems rest are, like those in congressional statutes premised on "safe" levels of pollution, pious fictions worthily motivated.⁴² Following Holmes in the view that the life of the law has been experience rather than logic, we might pretend to believe in these pious fictions if we thought they advanced the goals of the tort system in a useful way. But they do not.

While there may be good reason to provide a system of social insurance to spread the entire aggregate risk (background plus defendant's contribution) of diseases such as cancer, it contravenes both equity and efficiency to impose the entire burden of financing such insurance on the defendant. The alternative strategies described above may deter some risk generators, but requiring a defendant who has caused less than 50 percent of the risk to pay for 100 percent of the damages may result in overdeterrence: we will see very little investment in products and processes that create some additional risk.

The argument for risk spreading becomes weaker, as the risk generated by the defendant becomes proportionately smaller. Why should we spread the 20 percent of the risk generated by defendant and not the other 80 percent? It seems particularly unjust, and in contravention of defendant's rights, to spread the entire 100 percent on the defendant.

Finally, awarding plaintiff an amount greatly in excess of the harm caused by defendant will create incentive for litigation, analogous to those created by antitrust treble damage awards. Judicial strategies which provide such awards are thus an open invitation for lawyers on both sides to generate enormous transaction costs to deal with a small percentage of the total risk.

Some might defend "pious fiction" strategies on the grounds (1) that the ultimate solution to risk management may well be a general social insurance system for all risks — background plus those generated by defendants — with general or special deterrence handled by other means; and (2) that adoption of a "substantial-factor" or burden-shifting approach by the courts will hasten the day when Congress will adopt such a system.

While the insurance system itself may be a desirable goal, judicial adoption of a "substantial factor" or burden-shifting approach would most likely impede, rather than advance, such a goal. Various bills pending in Congress would mandate some variant of these approaches. If the courts were to embrace these approaches, Congress would almost certainly follow suit. Judicial endorsement of the pious fictions would hardly encourage Congress to face the hard reality that we often cannot know the cause of injury in a particular case, and that we must develop radical new approaches to deal effectively with incentives and risk spreading.

Ultimately, we must confront the difficult choice between the preponderance rule and that of proportionate liability. Peter Huber's article⁴³ presents a powerful argument for retaining the existing rule, stating that the risk-shifting and incentive effects of alternative stratagems are problematic and will be purchased at enormously high transaction costs. Huber believes, therefore, that the legal system should not take such an expensive step where only a small percentage of total risk has been generated by a particular defendant.

But there are also powerful arguments in favor of the proportionate liability approach, which makes the nature of the problem emphatically clear. Although this approach may be expensive, and cannot be justified entirely by risk-spreading or compensation criteria, it serves as a general deterrent mechanism to ensure that defendants who generate industrial or product risks will pay a price for doing so, regardless of whether the risk has been incorporated into a regulatory system.

Conclusions

Only a system of administrative risk management is capable of defining and implementing sound risk portfolio strategies. The performance of the current administrative system is, however, quite poor. Courts have done little to stimulate improvement. They should either assume a more vigorous leadership role or withdraw, and leave the task to others.

Pending the development of adequate administrative systems, courts have an important residual responsibility to control unregulated risks and to provide compensation for environmental harms. In dealing with the growing number of compensation claims by "indeterminate" plaintiffs, courts should either retain the current preponderance rule or adopt principles of proportionate liability. In either case, they should reject "substantial-factor" and burden-shifting stratagems, which obscure the nature of the risk portfolio problem.

1. See generally, Committee on Risk and Decisionmaking, National Research Council, Risk and Decisionmaking: Perspectives and Research (1982).

2. The imperatives of consistency and coordination persuaded the Court in Texas & P.R.R. v. Abilene Cotton Oil Co., 204 U.S. 426 (1907) to invent the doctrine of primary jurisdiction. The Court held that an unreasonable rate claim filed in court by a shipper should be decided in the first instance by the Interstate Commerce Commission (ICC), notwithstanding an express provision in the Interstate Commerce Act saving common law remedies.

In theory, a legislature might be able to achieve coordination and consistency in setting individual rates, but in practice the Granger state legislatures that attempted the task soon delegated it to administrative bodies.

3. See, e.g., VISCUSI, RISK BY CHOICE, Ch. 2 (1983) (OSHA health and safety standards); Haigh, Harrison & Nichols, Benefit-Cost Analysis of Environmental Regulation: Case Studies of Hazardous Air Pollutants, 8 HARV. L. REV. 395 (1984) (Proposed regulations under Clean Air Act § 112, 42 U.S.C. § 7412, ELR STAT. 42215); Putnam, Hayes & Bartlett, Inc., Analysis of Cost Per Health Risk Avoided in Selection EPA Standards and Regulations (Report to EPA, 1984) (toxic air pollutants, drinking water contaminants, pesticides). See also Harrison & Nichols, Benefit Based Flexibility in Environmental Regulation (Energy and Environmental Policy Center, Kennedy School of Government, Harvard University, Discussion Paper E-83-06, 1983) (documenting gains in cost-effectiveness feasible in controlling various air and water pollutants and noise by varying the level of controls in relation to benefits achieved).

4. Consider, for example, the differences in the stringency of controls imposed and the level of residual health risks tolerated in the regulation of nuclear electric generating plants and of coal-fired electric generating plants.

5. See, e.g., Clean Air Act §§ 109, 112, 42 U.S.C. §§ 7409, 7412, ELR STAT. 42209, 42215.

6. See, e.g., Occupational Safety & Health Act § 6(b)(5), 29 U.S.C. § 656(b)(5).

7. 42 U.S.C. § 7409, ELR STAT. 42209.

8. 29 U.S.C. § 650.

9. See Industrial Union Dept. AFL-CIO v. American Petroleum Institute, 448 U.S. 607 (1980) (Powell, J., concurring).

10. See, e.g., Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 1 ELR 20110 (1971); Sierra Club v. Ruckelshaus, 344 F. Supp. 253, 2 ELR 20262 (D.D.C. 1972), aff'd, 2 ELR 20656, 4 ERC 1815 (D.C. Cir.), Aff'd by an equally divided Court, 421 U.S. 541, 3 ELR 20684 (1973).

11. See, e.g., American Textile Manufacturers Institute v. Donovan, 452 U.S. 490, 11 ELR 20736 (1981); American Petroleum Institute v. Costle, 665 F.2d 1176, 11 ELR 20916 (D.C. Cir. 1981); Lead Industries Ass'n v. EPA, 647 F.2d 1130, 10 ELR 20643 (D.C. Cir. 1980).

12. See American Textile Mfrs. Institute, supra, involving OSHA standards regulating exposure to cotton dust, which had been promulgated under the Carter Administration. When the Reagan Administration assumed office, it asked the Court to remand the standard to OSHA. Perhaps fearing that the Administration would weaken the standard through application of cost-benefit considerations, the Court refused, construed the OSH Act to mandate installation of all technologically and economically feasible controls once a significant health risk had been established, and upheld the standards.

13. Courts sometimes mandate a BAT approach through a wooden reading of statutes. See American Textile Mfrs. Institute, supra. Or, they may achieve the same result indirectly by excluding questions of compliance feasibility from the standard-setting "frontend" of regulation, only to stay enforcement of unattainable standards at the "back end" stage of compliance. See R. MELNICK, REGULATION AND THE COURTS (1983) (Clean Air Act NAAQS).

14. See generally, Stewart, Regulation, Innovation, and Administrative Law: A Conceptual Framework, 53 CALIF. L. REV. 1339 (1980).

Even the generation of information on "worst-case" scenarios — a matter of controversy in the implementation of the National Environmental Policy Act — may create bias against new and therefore uncertain technologies, as compared with those whose adverse effects are better known and are, on average (even taking into account risk-aversion) more serious. See Zeckhauser, New Technologies and the Valuation of Life (Energy and Environmental Policy Center, Kennedy School of Government, Harvard University, Discussion Paper E-84-10, 1984).

15. See, L. LAVE, Introduction, QUANTITATIVE RISK ASSESSMENT IN REGULATION (1982).

16. See Note, Deciding What to Regulate: Priority-Setting at OSHA, 2 VA. J. NAT. RES. L. 87 (1982).

17. See Industrial Union Dept. AFL-CIO v. American Petroleum Institute, 448 U.S. 607, 11 ELR 20736 (1980). For an analysis of the overregulation involved in this case, see Latin, The Significance of Toxic Health Risks: An Essay on Legal Decisionmaking Under Uncertainty, 10 ECOLOGY L.Q. 339 (1982); Note, Deciding What to Regulate, supra note 16.

18. Portland Cement Ass'n. v. Ruckleshaus, 486 F.2d 375, 3 ELR 20642 (D.C. Cir. 1973).

19. For many years, EPA underregulated the hazards generated by leaded gasoline, see U.S. EPA COSTS AND BENEFITS OF REDUCING LEAD IN GASOLINE, FINAL REGULATORY IMPACT ANALYSIS (Feb. 1985), while squandering its resources on far more trivial matters. This diversion of resources was by no means the fault of EPA: Congress, the media, and environmental groups must share the responsibility.

20. See, e.g., New York v. Gorsuch, 554 F. Supp. 1060, 13 ELR 20748, 18 ERC 1575 (S.D.N.Y.) (action under Clean Air Act § 304 to mandate EPA promulgation of standard for airborne arsenic). See generally, Stewart and Sunstein, Public Programs and Private Rights, 95 HARV. L. REV. 1193 (1982).

21. EDF. v. Rucklehaus, 439 F.2d 584, 1 ELR 20059 (D.C. Cir. 1971) (DDT). See also Public Citizen Health Research Group v. Auchter, 702 F.2d 1150, 13 ELR 20877 (D.C. Cir. 1983) (declining to require OSHA to issue Emergency Temporary Standard (ETS) for ethylene oxide but ordering OSHA to commence rulemaking on a permanent standard on an expedited basis). cf. UAW v. Donovan, 590 F. Supp. 747 (D.D.C. 1984) (remanding, for further explanation, OSHA refusal to issue ETS for formaldehyde).

22. For example, Sierra Club v. Gorsuch, 551 F. Supp. 785, 13 ELR 20231, 18 ERC 1549 (N.D. Cal.) mandated EPA promulgation of standards for ambient radionuclides. The risks to be regulated under this standard are, relatively speaking, quite trivial. The entire exercise represents a serious misallocation of administrative resources.

23. In many instances, the only available point of leverage will be the environmental impact statement requirement in the National Environmental Policy Act — a requirement from which many environmental regulatory functions are exempt. 105 S. Ct. 1649, 15 ELR 20335 (1985).

24. See Note, Intent, Clear Statements and the Common Law: Statutory Interpretation in the Supreme Court, 95 HARV. L. REV. 892 (1982).

25. LANDIS, THE ADMINISTRATIVE PROCESS (1938).

26. For discussion of various alternatives, some of which contemplate a substantial continuing judicial role, see Yellin, Science, Technology, and Administrative Government: Institutional Designs for Environmental Decisionmaking, 92 YALE. L.J. 1300 (1983).

27. See, e.g., Renken v. Harvey Aluminum, Inc., 226 F. Supp. 169 (D. Or. 1963); Reserve Mining Co. v. EPA, 514 F.2d 492, 5 ELR 20596 (8th Cir. 1975).

28. See, e.g., United States v. Vertac Chemical Corp., 489 F. Supp. 870, 10 ELR 20709 (E.D. Ark. 1980).

29. I put to one side the question of punitive damages, noting only that the structural considerations favoring agency management of portfolio risk presumptively disfavor judicial imposition of punitive damages as a risk management tool. But see Silkwood v. Kerr-McGee Co., 464 U.S. 238, 1 ELR 20077 (1984).

30. See, e.g., Sindell v. Abbott Labs, 26 Cal. 3rd 588, 163 Cal. Rptr. 132, 607 P.2d 924 (1980).

31. There may be multiple defendants creating risks above background, raising the problem of the "indeterminate defendant," discussed above; this possibility, however, does not alter the analysis which follows.

32. See Rosenberg, The Causal Connection in Mass Exposure Cases: A "Public Law" Vision of the Tort System, 97 HARV. ENV. L. REV. 849, 855 (1984). See also, Trauberman, Statutory Reform of Toxic Torts, 7 HARV. ENV. L. REV. 177, 190 (1983); Note, Inapplicability of Traditional Tort Analysis to Environmental Risks, 35 STAN. L. REV. 575, 585 (1982); Ginsberg & Weiss, Common Law Liability for Toxic Torts: A Phantom Remedy, 9 HOFSTRA L. REV. 859, 930 (1981).

33. For example, if the risk generated by defendant is 20 percent of the aggregate, and full compensation for the disease would be $500,000, each person contracting the disease would be awarded $100,000 against the defendant.

34. In re Agent Orange Product Liability Litigation, 597 F. Supp. 740 (E.D.N.Y. 1984). The basic question about this case is why the class was certified in the first place if the plaintiff's case was as weak as Judge Weinstein's opinion indicates.

35. General purpose insurance and compensation programs, both private and governmental, are often available, however.

36. But see, Rosenberg, supra note 32. Involuntary exposure to risk may itself be viewed as an invasion of an individual's rights. The Supreme Court, however, has declined to follow this approach. See Duke Power Co. v. Carolina Environmental Study Group, 438 U.S. 59, 8 ELR 20545 (1978); Metropolitan Edison Co. v. People Against Nuclear Energy, 460 U.S. 766, 13 ELR 20515 (1983).

37. Since a system of proportionate liability would levy monetary incentives case-by-case on the basis of highly uncertain and variable estimates of defendant's contribution to aggregate risk in particular circumstances, it is doubtful that such a system would achieve the risk portfolio objectives described earlier in this paper.

38. The transaction costs in a system of proportionate liability might be greatly reduced, if its operation were entrusted to a specialized administrative agency, as described earlier in this paper, rather than to the courts. Such an agency could also deal in a more informed and consistent fashion, with the uncertainties involved in estimating defendants' contribution to aggregate risk and in promoting risk portfolio objectives. As noted above, however, there are a variety of practical obstacles to creating such an administrative system.

39. See Allen v. U.S., 588 F. Supp. 247 (D. Utah 1984). The opinion in this case provides no meaningful guidance on how to determine substantiality; determination appears to be a highly subjective judicial reaction to different factual situations. Where the probability that plaintiff's injury was caused by defendant is less than 50 percent, plaintiff may recover total damages if the judge finds a "substantial relation" between the injury and the exposure generated by defendant, unless the defendant meets the burden of showing that the exposure did not cause the particular injury in the particular case.

40. See Plommer v. U.S., 580 F.2d 72 (3d Cir. 1978); Wetherhill v. University of Chicago, 565 F. Supp. 1553 (N.D. Ill. 1983); City of North Glenn v. Chevron USA Inc., No. 81-C-44 (D.Col., April 21, 1982). Some courts have also provided recovery for indeterminate plaintiffs by allowing recovery for the emotional distress caused by exposure to risk.

41. See, e.g., Grasso v. B.F. Goodrich Co., No. 78-156 (D. N.J. 1981).

42. These "pious fictions" may reflect deep cultural and social ambivalence about environmental risk. See G. CALABRESI & P. BOBBIT; TRAGIC CHOICES (1979); M. DOUGLAS & A. WILDOUSKY, RISK AND CULTURE (1982).

43. See Huber, Safety and the Second Best: The Hazards of Public Risk Management in the Courts, 85 COLUM. L. REV. 277 (1985).