Panel Discussion

HENRY NICKEL, Esq., Hunton & Williams, Washington, D.C.

WILLIAM A. ANDERSON II, Esq., Bracewell & Patterson, Washington, D.C.

KENNETH R. FEINBERG, Esq., Kaye, Scholer, Fierman, Hayes & Handler, Washington, D.C.

NICHOLAS C. YOST, Esq., Member Standing Committee on Environmental Law; Center for Law in the Public Interest, Washington, D.C.

HENRY NICKEL: The objective of Professor Stewart's idealized risk management system is simply to do more good than harm. That is not a very radical notion; unfortunately, our statutes do not currently operate to ensure this result in every case.

To determine whether a particular course of action will produce more good than harm, agencies must look at the economic expenditures involved and at the entire spectrum of risks that will be created by a decision to proceed or not to proceed in a particular manner.

In the case of coal-fired power plants, the Environmental Protection Agency's (EPA's) recent rulemaking on radionuclide emissions provides a good example of a risk that EPA determined did not justify regulation. The agency estimated the risk to the maximum exposed individual at one in two million. The population at large (240 million) faced an estimated risk of three deaths annually. Application of maximum controls would cost an estimated $3 billion per year and would reduce the number of deaths from three to two and one-half. The impact of this expenditure on local economies would have health consequences, as would the construction and other activities required to install controls. The health risks created would be ten to twenty times greater than the health benefits that would be achieved by further regulation.

I agree with Professor Stewart that an agency can implement a portfolio approach under current law, if it simply chooses to do so. The Clean Air Act provides a good example. The purpose of that Act is "to protect and enhance air quality so as to promote the productive capacity of the population." In other words, the statute's purpose is not the absolute protection of the public health but protection of the public health in a way that produces good. This purpose, I think, provides the basis for the kind of risk management system postulated by Professor Stewart.

The problem with having courts take the lead in the area of risk management is that courts must defer to agencies' constructions of their own statutes. In the Chevron decision¹ — the case involving EPA's construction of "source" under the nonattainment provisions of the Clean Air Act — the Supreme Court made it clear that the Courts of Appeals must defer to reasonable interpretations by federal agencies. Interestingly, recent Courts of Appeals cases interpreting Chevron suggest that, even though the court cannot substitute its own interpretation of the law for that of the agency, it can require the agency to show that, in choosing a particular interpretation, it considered all of the major policies underlying the statute.² This approach, I believe, may provide a way for the courts to lead the agencies in a better direction on the question of risk management.

Finally, I would like to mention an issue that Professor Stewart did not discuss in his paper. I am not sure that the current problems in risk management should be blamed entirely on the courts; it is Congress that created them. We have in our environmental statutes one regulatory requirement after another in which Congress imposed statutory deadlines that are totally infeasible. These requirements create a series of defaults by agencies. In virtually every provision of the Clean Air Act, one can find some duty that the agency has failed to perform.

Citizen suit provisions permit individuals to go into district court to enforce these duties that have been breached. This means that people bringing citizen suits are exerting a tremendous influence on the regulatory priorities of the agencies. The courts, in setting rulemaking schedules, are affecting the agencies' ability to perform their responsibilities in a sound way. For example, EPA's rulemaking on radionuclides was supervised by the court, and the agency was under a tight deadline. I believe that the explanation for the final rules would have been better had the agency been given more time. In fact, in that particular litigation, the court went so far as to require the agency to establish standards for source categories that EPA had already concluded posed no significant risk. EPA responded by setting standards believed to represent current emissions. In other words, it set standards that did nothing. Citizen suit provisions can therefore interfere with the ability of agencies to perform the type of risk management strategies described by Professor Stewart.

WILLIAM A. ANDERSON, II: The risk portfolio approach advocated by Professor Stewart is an extremely valuable contribution to our thinking on risk assessment. I have been doing my own risk assessment, and would like to address two critical issues: quantitative risk assessment and the range of uncertainty.

First, quantitative risk assessment may be worse than useless. For courts, decisionmakers, and lawyers, the numbers may be all too alluring. They trick us into believing that scientists know with certainty a truth that they have only assumed. Quantitative risk assessment may be no more than guesswork masquerading in scientific garb.

In the end, quantitative risk assessment may generate numbers that simply disguise the underlying uncertainty. Scientists have an obligation to report — indeed to underscore — the range of uncertainty in their risk assessment. Perhaps we should have a statute that requires a statement of the uncertainty range to accompany the risk analysis, as an environmental impact statement (EIS) is required to accompany a proposal through consideration by federal agencies.

Second, I believe that the degree of conservatism in a risk assessment is necessarily a function of the range of uncertainty. The greater the degree of uncertainty, the more conservative the analysis is apt to be.

Conservatism, whether quantitative or visceral, becomes our hedge against uncertainty. Earlier today, Dr. Silbergeld [16 ELR 10217] said that we apply a double standard to known versus unknown risks. The inescapable consequence is that risk assessment and risk management decisions will be biased against innovation and new products. For example, in the University of California case involving genetically engineered bacteria to retard the formation of frost, this bias was directed against the introduction of biotechnology.

There is a danger that risk assessment may be counterproductive. This danger is especially acute in areas such as biotechnology, where the uncertainty is vast, and we have very little experience upon which to rely.

KENNETH R. FEINBERG: Let me begin by saying that I will certainly satisfy the nineteenth-century historian Macaulay's definition of a lawyer — somebody who, in ten minutes, can delve into the very depths of the superficial. That is all the time I have been allotted.

Ineffective regulation is not, relatively speaking, the most serious problem that we face. The more serious problem institutionally is risk that is not causally connected to the alleged injury. That is the Agent Orange problem in its purest form, and that is the problem that confronts the Congress, the courts, and our administrative agencies.

Ineffective regulation is a serious problem, but it is a political problem. The more serious problem highlighted by Agent Orange is how institutions react to risks that are not known or are at least perceived to be less serious than they might subsequently become.

There is a small group of cases in which the courts have no business being the final arbiter of a solution. These are cases where the pervasive exposure in mass tort cases is very, very large, the number of claimants is huge, and the transaction costs are enormous. Most importantly, these are cases where the causal link between the exposure and the claimed disease is weak, nonexistent, or, as Judge Weinstein says, "dubious at best." The causal connection, if it exists at all, is latent — ten years, twenty years, even thirty years may elapse between exposure and manifestation of disease.

In this small group of cases, courts should get out of the business of compensating injury. For the following reasons, the courts are not an effective institution for handling these cases.

First, although Judge Weinstein worked out a very effective solution in the Agent Orange case, every case will not go to Judge Weinstein or to the handful of master jurists who can effectively handle these problems.

Second, the Agent Orange pretrial discovery period dragged on for almost seven years. Pretrial preparation ended on the eve of trial, and the case was settled on the morning of trial. To go through seven years of preparation leading up to a trial suggests that something is very wrong with the ability of the judicial system to deal with these types of cases.

Third, some have forgotten that the government played no part whatsoever in the Agent Orange case. The government absolutely refused to participate, notwithstanding its role in the controversy. The government has also refused to participate in the asbestos litigation. The government realizes — and perhaps I might have the same attitude if I worked for the government — that many similar cases are pending on the horizon, and it is a serious question as to whether the government can afford to become a guarantor of last resort or even a party participant in these cases.

Fourth, as Professor Stewart points out, even if the courts are going to play the game in this limited type of mass tort litigation, they will play in a very uncertain way. At worst, the courts will twist legal doctrine; at best, they will be very, very creative. One need only look at the case law that has developed over the last three or four years to deal with the indeterminate defendant, the indeterminate plaintiff, alternative and proportionate liability, the burden of proof and statutes of limitation and repose.

If the courts are ineffective in handling this small group of mass tort cases, what, if anything, would be a better institutional approach?

Ironically, Congress is ill-equipped to deal with this type of risk. Congress has decided that when it comes to such risk, the victim ought to recover. The real question is, who are the "financially responsible parties"? This question has, I think, largely been lost on industry, which is lobbying in favor of the Kasten Bill, a vehicle designed to reform product liability law. The question is not whether the victim will recover — Congress and the courts have increasingly decided "yes" — but who shall pay.

What, then, is the better approach? There are two schools of thought on this issue. The first is the anticipatory approach, which holds that we must come up with a blanket solution, akin to national health insurance, to spread the risk across society in mass tort cases. This would seem to be the only solution in cases involving absence of causation. Cost, history, and ideology constitute the major obstacles to such an approach. The anticipatory approach will probably not work in the current political climate, but it is interesting to see certain major companies, who fifteen years ago led the charge against Senator Kennedy's National Health Insurance program, today drafting such programs as possible alternatives to what they see coming around the corner.

The second school of thought I call the reactive approach. It involves various ways in which new approaches can be developed once the crisis crystallizes. This approach is illness and industry specific.

One can react to a crisis in two ways. One is to centralize a national disaster court having its own special rules of evidence and procedures. We have a railroad reorganization court, an energy court, a foreign intelligence court; perhaps we ought to have a national disaster court, whose jurisdiction would be triggered by a specified dollar loss, number of cases, or whatever.

A second way to react, which is presently under consideration by Congress, albeit in its earliest stages, is the development of an alternative compensation system to tort. Under this system, a person who contracts a certain disease will be paid for economic loss — period. There will be no pot of gold and no attorneys fees. And who pays? The defendants pay. I suggest, however, that this should be a voluntary system. If companies — including indeterminate defendants — do not want to participate, they should not be required to do so. The plaintiffs are also at liberty to forego this route, but if they do so, no strict liability in tort will apply. They will have to show negligence.

In conclusion, I agree with Professor Stewart. I am wary of courts staying in this business. I am also wary of congressional involvement. After all, if one looks at the Black Lung program, one sees thast more people have been compensated for black lung than ever worked in a mine. The problem is one of statutory presumptions, grounded in the notion that the victim shall recover. I am not questioning that the victim should recover. But combined with problems of causation [16 ELR 10218] and latency, it leads one to be suspicious of whether Congress is the appropriate vehicle to solve our most serious institutional problem: nonsignature risk.

NICHOLAS C. YOST: I was disturbed by Bill Anderson's assertion that risk assessment is biased against innovation. Courts can continue — as the Ninth Circuit, in particular, has done under NEPA — to perform worst-case analysis and to insist that agencies look at ranges of environmental consequences, including the most severe ones.

PROFESSOR RICHARD B. STEWART: I would distinguish two issues. One is the new risk/old risk controversy, and the other is the bias against risks associated with greater uncertainty. These tend to be but are not necessarily congruent.

I would hope that the risk portfolio approach would ameliorate the old risk/new risk problem by articulating the trade-offs. One will never comp ltely avoid a certain bias, but it would be desirable, for instance, if a new plant that is better controlled or a new product that is less environmentally damaging could replace a more polluting or risky way of meeting the same social need.

The question of whether there is or should be a bias against more uncertain risks really goes to the problem of worst-case analysis. The less we know about a risk, the greater the danger of overemphasizing that uncertainty. This tendency becomes obvious, for example, if one compares our worries about the worst-case nuclear power scenario with what we know happens in fossil fuel combustion. On the other hand, it can be argued that we ought to be risk-averse, and if catastrophic accidents occur, we ought to penalize the more uncertain technology.

I think it comes down to a social value judgment. I might be happy with worst-case analysis if we also performed best-case and average-case analysis. The focus on worst-case analysis begs the question of the social choice that must be made.

PARTICIPANT: Ken Feinberg's remarks seemed to underscore the poverty of the tort system. I think we followed the tort model in establishing the Superfund, and now we are seeing that the transaction costs are enormous, that the cleanups are being delayed, and that some better risk-spreading scheme needs to be worked out.

STEWART: The alternative, as Ken put it, is to discard as an old-fashioned hangover from the private law system the notion that compensation and incentives ought to be linked together in the same institutional mechanism. In the modern world, we must separate these two goals and deal with incentives through some sort of regulation. For the incentive effect, I would, wherever possible, have economic-based incentives, transferable pollution permits, and risk markets set and managed by the government; I would handle the compensation risk-spreading problem through a social insurance scheme.

DR. ELLEN K. SILBERGELD: I want to comment on the remarks about causation and epidemiology. There is a tendency to dichotomize these matters in ways that confuse the issue. Consider, for instance, the distinctions drawn earlier betwee proportionate liability and the "substantial-factor" approach. I think there is a temptation to use epidemiology improperly to develop proportionality arguments. One could almost define this as the "weatherman" problem. The weatherman says that there is a ten or twenty percent chance of rain, but actually it either rains or not. An event either occurs, or it does not occur, although there are predisposing factors that allow one to make variable predictions.

Medically, one can speak about contributing factors to disease, and attempt to apportion them out. When it comes to managing disease in a preventive fashion, however, one does actually work in a proportional way. One identifies a proportional factor and deals with it. Management of cardiovascular disease is a primary example.

I would urge you not to dismiss too quickly the substasntial-factor approach identified by Judge Jenkins. Medically, it has a great tradition.

PARTICIPANT: Of course, let us talk about the defendant's risk and the background risk as the two contributing ingredients. It might be that in particular cases the defendant's risk and the background risk would interact to create the harm. We would still have to know in the aggregate the increase in the total number of cases that resulted from the addition of the defendant's risk to the background risk. And that question brings us back to the same problem: we cannot know the cause in the individual case.

Whatever the physiological situation, if the background risk is ten and the defendant's contribution raises it to twelve, we might say that it was a factor in every case, but I do not think we would say that defendant ought to pay the full cost of the twelve cancers.

SILBERGELD: We still come down to the weatherman paradox. In epidemiology, which rests upon a population basis, one may make those statements; in the individual case the person either has or does not have the disease.

PARTICIPANT: But we are dealing with cases where people did get the disease. The question becomes who is responsible.

SILBERGELD: Ken Feinberg made a distinction between signature and nonsignature risk. I would suggest that the difference is not so clear. What appears to us today as a signature relationship depends upon the relationship having become clear over a period of time. In dealing with Agent Orange, we are now at a midpoint in gathering medical and scientific information; perhaps we have not yet seen the signature, if there is going to be one.

On the other hand, there are probably very few signature relationships between agents and disease. As you know, asbestos is associated with many medical conditions other than mesothelioma. The same holds true for vinyl chloride and angiosarcoma. Sometimes we tend to censor a large amount of medical information in order to develop signature relationships that may not be real.

FEINBERG: While an epidemiologist may talk about nonsignature risks becoming signature risks over a period of time, the law looks at things a bit differently. One runs into the problem of statutes of limitation and repose and also into the problem of deterrence and incentive. I have not yet heard a good argument as to how 25,000 asbestos lawsuits can deter 25 or 35 companies that have been out of the business for 25 years. This lack of congruity between science and the law — and it is the fault of the lawyers, not the scientists — is where the tension lies.

[16 ELR 10219]

SILBERGELD: I think that the asbestos experience has had a significant effect. Compare it, if you will, with the bischloromethyl ether story, where, for a long time and for a variety of reasons, the information associating exposure with a serious risk of lung cancer was concealed. The asbestos experience has changed how one deals with the uncertainties of risk in the present that may be expressed in very large terms at a later time.

MICHAEL LAST: I would like to invite further comment on the delay in progress that might result from risk management. This calls to mind the questions raised by the asbestos litigation: the question of the failure to warn, the utility of the product, and the issue of whether users should have been alerted to known risks so that they would have had the option to choose other less efficient, but also less risky, products. The appeal of Professor Stewart's portfolio approach is that it allows one to perform cross-evaluations. Mr. Anderson, how would that observation apply to your experience with biotechnology?

ANDERSON II: Frost is a problem that costs the United States agricultural community something on the order of $3 billion a year in lost products. A couple of years of hard freeze can devastate the Florida citrus industry. Scientists have discovered that (ethophitic) bacteria, which secrete an ice-nucleating substance, cause much of the frost damage in the temperature range of five to seven degrees above what plants could otherwise sustain without damage.

Researchers at the University of California at Berkeley experimented by chemically inducing genetic changes to these little organisms. They also conducted laboratory and growth-chamber experiments that actually (rather than chemically) altered the genetic structure of the bacteria. The theory was to create ice-free or ice-minus bacteria and to put those bacteria on the leaves of plants to forestall the formation of frost by having these modified bacteria colonize specific plant parts to the exclusion of their natural counterparts.

The experiment itself was subject to review and approval by the National Institutes of Health (NIH), since the University received NIH funds. NIH approved the experiment and was promptly sued for failure to prepare an EIS or otherwise comply with NEPA.

One of the alleged environmental "horribles" — and this refers back to the worst-case analysis — was the threat of worldwide catastrophic draught and famine, because the genetically engineered mutants might overrun their natural counterparts, which conceivably (although no one knows for certain) play a role in major-scale atmospheric precipitation processes. The natural ice-nucleating bacteria would be displaced nd unable to rise up into the clouds; we would therefore have cloudy days forever and no rain.

The worst-case analysis is not very helpful to a decisionmaker or very informative to a public concerned about the future. The researchers at Berkeley were trying to do something that many believed would have fewer consequences than conventional frost-control techniques.

STEWART: The real problem is not the EIS but the need for an inflationary impact statement. What will happen to the federal budget, if we have $3 billion more agricultural products each year?

ANDERSON II: Professor Stewart, you poked a little fun at the worst-case analysis. We have heard your worst-case, best-case, and average-case analysis. As many of you know, the Council on Environmental Quality is now looking into the possibility to changing its regulation dealing with worst-case analysis. Would you change the regulation? If so, how?

STEWART: I do not oppose the notion of worst-case analysis, but I think it must be given reasonable content so that it does not just look at totally catastrophic improbabilities.

Worst-case analysis ought to be balanced by an average-case analysis that looks to the expected value or the average value, as well as to the tails of the probability distribution. There is a danger in focussing on one end of the tail without looking at probability distribution as a whole.

ANDERSON II: Mr. Yost, how do you reconcile the notion of an upper plausible limit with what you characterize as a worst-case analysis? Worst case is catastrophe, but it is extremely improbable.

YOST: I am not sure you are talking about different concepts there. You are using different words to talk about something that is not so improbable as to be impossible, and that responsible scientific opinion says may happen. One must look at that scenario, among others.

PARTICIPANT: Over the last 18 months or so, we have been inundated with tort claims under the Federal Tort Claims Act brought by individuals who have no manifestation of injury. The complaints allege exposure to anything from lead to toxins at such places as acid pits. Does risk assessment play any role where the major harm alleged — beyond nausea, dizziness and other minor symptoms — is the fear of contracting something? The exposures are so recent that sufficient time has not yet elapsed for any type of natural injury to manifest itself, but complaints are nevertheless being asserted at this point. For other reasons, none of the complaints has gone forward as yet.

PARTICIPANT: I do not have any view on whether this would be an actionable tort, but it does seem that the risk portfolio approach would be useful here, if only to compare the risk estimate with respect to the particular exposure with other risks that the complainants are assuming daily. This comparison can help determine whether the complainant's fear is reasonable in light of what the risk assessment numbers would show.

FEINBERG: If one did perform such an assessment, what would we hope to achieve? The government is now spending $150 million on Agent Orange research, and I will be very surprised if any concrete conclusions or causal connections come out of it. We seem to place too much faith in the scientific community.

1. Chevron U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S. 837, 14 ELR 20507 (1984).

2. See, e.g., Rettig v. Pension Benefit Guaranty Corp., 744 F.2d 133 (D.C. Cir. 1984).