The Temporal Dimension in Environmental Law

Lisa Heinzerling

Professor of Law, Georgetown University Law Center. This Article is adapted from Lisa Heinzerling, Environmental Law and the Present Future, 87 GEO. L.J. 2025 (1999) (reprinted, as one of the best environmental law articles of 1999, in 31 LAND USE & ENVTL. L. REV. (2000)). I am grateful to Marguerite McLamb for excellent research assistance.

[31 ELR 11055]

Pollution of the air, water, and land increases the risk that human beings will fall ill and prematurely die. Laws restricting this pollution begin to reduce the risk of illness and death as soon as they are implemented. Often, however, the people who would have died in the absence of regulation would not have done so for many years, and thus laws restricting harmful pollution may not reduce the rate of premature human mortality for a long time. This may be because the disease that would have killed people has a long latency period, or because it would have taken many years to amass sufficient exposures to cause death, or because the people exposed to a persistent chemical did not even exist at the time the regulation was imposed. Whatever the reason, it is often the case that the deaths prevented by regulation are deaths that would have otherwise occurred in the remote future. For this reason, it has become commonplace to assume that the benefits of life-saving environmental regulation occur, for the most part, at a large temporal distance from the regulatory activities that reduce risk.

For many, this focus on the future is one of the great triumphs of environmental law. The fact that we have constructed a vast regulatory apparatus aimed at preventing harms in the distant future, harms even to people who do not yet exist, provides a heavy counterweight to claims that individual and collective actions are dominated by shortsightedness, selfishness, and parochialism. Attention to the future also encourages, perhaps even necessitates, the kind of mindset and lifestyle that many environmentalists embrace: frugal and simple, and humble in the face of uncertainty about what the future holds—"conservative" in the old-fashioned sense of the term.

For others, the future orientation of environmental law is cause for a radical restructuring of our regulatory priorities. For them, it is troubling that we should attend so closely to the future when the present, too, presses upon us its urgent demands. If only we put the future in its proper place, demoting it in importance relative to the present, we would see that many of the things we now try to avoid—like radioactive waste and hazardous air pollution—do not make much of a difference to our lives, and that some of the things we sometimes ignore—like bicycle safety—could greatly improve our lives in the here and now. Putting the future in its proper place means, in operational terms, the use of accounting methods, such as discounting, that systematically decrease the importance of future harms relative to present ones.

Despite the importance of the temporal dimension to these competing views of environmental law, there has been little examination of the actual temporal distribution of the benefits of environmental law. That environmental law looks predominantly to the future appears to have been accepted as a truism by both sides of the debate.

In this Article I offer a new conception of the temporal dimension in environmental law. My specific focus will be on the prevention of human disease and death through environmental law. I will argue that, in reality, the beneficial consequences of life-saving environmental regulation do not occur within a single time frame, either the present or the future, one to the exclusion of the other. Instead, such regulation produces benefits beginning the very moment it takes effect, up until the time the last person helped by the regulation would otherwise have drawn her last breath. Thus the good human consequences of environmental law—ranging from reducing risk to preventing disease to forestalling death—are arrayed along a continuum stretching from the immediate present to the distant future.

Equally important, the present and the future are not so much sequential as interactive, and their interaction works both forwards and backwards: events today shape events tomorrow, and perceptions about what the future is likely to hold affect present well-being. Moreover, this interaction sometimes makes remote human health consequences as problematic as immediate ones, or even more so. Hazards that will materialize in physical harm only after a long period of delay, or even in future generations, seem to arouse unique fears in the general public. There is thus often something like an inverse relationship between anxiety and immediacy, which makes the very futurity of a physical harm more troublesome right now.

The interaction between present and future also works in the other direction; that is, present events affect future events. Many human diseases, most prominently cancer, are characterized by a long interval, or latency period, between a harmful event (such as exposure to a hazardous substance) and the clinical appearance of disease. The long latency of many of the diseases prevented by environmental regulation is, in fact, one reason why such regulation has been thought to attend mostly to the remote rather than immediate future. But these diseases have typically begun an inexorable course toward physical suffering, and perhaps death, long before clinical symptoms appear. To focus on the time lag between exposure and clinical detection—the events which, for scientific purposes, have marked the beginning and end of the latency period—is to ignore the fact that, for regulatory purposes, the moment when a disease is detected comes too late, as it is by then impossible to prevent it. In this Article, therefore, I argue in favor of a new understanding of latency in the regulatory context, an understanding which would terminate the "regulatory" latency period at the moments when risk is reduced and disease initiated, rather than at the moment when an existing disease becomes detectable.

[31 ELR 11056]

My analysis has important implications for current methods for evaluating the wisdom of environmental regulation. Specifically, my analysis greatly undermines the case in favor of the most important analytical method for taking time into account in calculating the benefits of life-saving regulation. This accounting method—the practice of discounting future lives saved—has in recent years been endorsed by the federal government. In particular, the Office of Management and Budget (OMB) has dictated that the future benefits of regulations, including future human lives saved, be discounted at a fixed rate for every year that passes before the benefits accrue.¹ Although OMB's basic position on the matter of discounting has remained the same for many years,² the U.S. Environmental Protection Agency (EPA) has sometimes resisted this accounting technique.³ Even now, its position on discounting seems to remain unsettled.⁴ Discounting can have an enormous effect on the perceived wisdom of environmental regulations. Indeed, economic analyses employing discounting are, perhaps more than anything else, responsible for the common perception that environmental regulation costs too much while saving too few.⁵

The temporal analysis I offer in this Article severely undermines the case in favor of discounting as a method of accounting for the temporal dimension in environmental law. Cost-benefit analyses using this method have, to date, wrongly assumed that the life-saving benefits of environmental regulation occur in the quite remote future. In addition, discounting proceeds from the mistaken premise that one can isolate a specific moment when the benefits of regulation suddenly appear. It also implicitly embraces a reactive, rather than preventive, approach to regulation, and a bizarre metaphysics which holds that an illness is not prevented, nor a death averted, at the moment when it is avoided, but at the moment when physical hardships otherwise would have become patent. If the temporal analysis I offer in this Article is correct, this accounting method cannot stand.⁶

In this Article, I discuss only the direct effects of pollution on human health. I do not discuss indirect effects on human health caused by the degradation and destruction of ecosystems, nor effects on the ecosystems themselves. Thus the analysis that follows severely understates the adverse effects of pollution, and thus also the good consequences of environmental protection. Although I do not pursue the point in detail here, my conviction is that the temporal dynamic I describe in this Article applies with equal force to the effects of pollution (and regulation) on ecosystems. That is, the timeline over which these effects occur is also lengthy, continuous, and interactive. And, in that context, too, a perception of an expansive, even indefinite future—for a vista, a species, or the planet itself—is an important part of our present well-being.

The Temporal Dimension of Environmental Problems

In this part, I discuss the temporal sequence of the benefits of environmental law, that is, the temporal range over which these benefits accrue. I find this range to be vast, stretching from the immediate present to the distant future, and even into generations beyond our own. I also find that present fears of future harms can profoundly affect present psychological and social well-being, and that present physical conditions can inexorably lead to diseases that will become manifest only in the remote future. Present conceptions of the future, in other words, affect the present, and present physical events affect the future. My discussion also implies the appropriate timing of regulation itself: if we wish to reduce the kinds of harms I am about to describe, we must begin now. That many environmental risks are gradual, cumulative, and persistent means that we cannot wait to act until disease or death is upon us.

Risk

Regulation reducing the risk of human disease and death produces benefits from the moment the risk is reduced. One need not wait until the time when the adverse health condition (the risk of which has been reduced) would otherwise have led to material harm in order to declare that regulatory benefits have begun to accrue. Reducing risk is itself a benefit, separate and apart from the prevention of illness and death. This is both because a reduction in risk carries with it a decrease in dread and its debilitating effects on individuals and communities, and because the reduction in risk reflects an immediate reduction in expected loss.

Dread

It is by now well known that the kinds of risks reduced by environmental regulation have a singular ability to strike fear in the hearts of ordinary citizens. This fear often does not correlate very well, if at all, with numerical probabilities of illness and death.⁷ To name one famous example, laypeople tend to view nuclear power as a much riskier activity than medical X-rays, even though expert risk assessors estimate that the numerical probability of illness and death is higher for the latter than for the former activity.⁸ [31 ELR 11057] The risk perceptions of experts, too, are affected by factors in addition to statistical probabilities of harm, but less so, perhaps, than those of lay citizens.⁹

The discrepancy between lay and expert perceptions of risk, and more specifically, between lay perceptions of risk and numerical probabilities of physical harm, has inspired large efforts to understand what might explain the divergence. Many researchers have concluded that at least part of the divergence between lay and expert, or nonstatistical and statistical, views of risk arises from laypeople's tendency to take a wider set of considerations into account in judging risk. The list of these considerations is by now familiar to any student of environmental policy: the controllability, familiarity, immediacy, diffuseness, voluntariness, equity, reversibility, and naturalness of the hazard—or the converse of these characteristics—all appear to play a large role in shaping lay perceptions of risk.¹⁰ Laypeople also appear to care a great deal about whether a hazard threatens only this generation, or also future generations, and appear to perceive the latter as riskier than the former.¹¹ Finally, in making judgments about risk, lay citizens tend to fold in the benefits they perceive to flow from the substance or activity in question.¹²

Thus, it appears that citizens' anxieties about a particular substance or activity do not depend on the immediacy of the physical harm that might be done by it. Indeed, in many cases there appears to be an inverse relationship between anxiety and immediacy. To perceive more risk from hazards that pose threats to future generations than to ones that threaten this generation alone, and to sometimes worry more about latent hazards than immediate ones, is obviously to reserve a special dread for the remote threat of harm. Less obviously, considerations such as equity, controllability, knowability, and reversibility also have a large temporal dimension. A long passage of time between the imposition of a risk and the manifestation of physical harm makes inequity more likely, insofar as those who benefited from the risk are less likely to be around to suffer the consequences of having imposed it. In addition, it is harder to control and to know a threat whose consequences cannot be perceived until many years after one acts.¹³ One's ability to engage in trial and error—a time-honored technique for controlling and learning about a hazard—is severely diminished by a large temporal gap between the trial and the error.¹⁴ It is also more difficult to reverse a threat that has been many years in the making,¹⁵ and that arises—as many environmental hazards do—from durable agents that, once unleashed, persist in the environment and in living tissue for many years.¹⁶

There is also a temporal element in citizens' tendency to consider the benefits of an activity in judging its riskiness. Put simply, latent hazards pack no thrill. The chance, for example, that exposure to a chemical will lead to a diagnosis of cancer one-quarter century hence just cannot compete with the high-adventure, high-adrenaline possibility of dying immediately in a spectacular skiing, boating, or motorcycling accident. Part of the reason why people enjoy activities like skiing and motorcycling is that they are risky, and immediately so.¹⁷ If all goes well, one's payoff is immediate: one returns home safe and sound the same day, having experienced at once the thrill of danger and the thrill of mastering it.¹⁸ Almost simultaneously with experiencing the risk itself, one experiences (again, if all goes well) the certainty of having survived it intact.¹⁹ With latent risks, one must wait years for such assurance.²⁰

This lack of early assurance distinguishes even the involuntary immediate threat (such as a hurricane or flood) from the latent one, and helps to explain why a special anxiety might be reserved for hazards whose physical consequences are remote. As the sociologist Kai Erickson has written:

[31 ELR 11058]

We generally use the word "disaster" in everyday conversation to refer to a distinct event that interrupts the accustomed flow of everyday life. "Disasters" seem to adhere to Aristotle's rules of drama. They have "a beginning and a middle and an end." …

An alarm sounds the beginning. It is a signal to retreat, to take to storm cellars, to move to higher ground, to crouch in the shelter of whatever cover presents itself. A period of destruction then follows that may take no more than a brief, shattering moment or may last many days. Sooner or later, though, the disaster comes to an exhausted close. The floodwaters recede, the smoke clears, the winds abate, the bombers leave, and an all clear is sounded either literally or figuratively….

Toxic disasters, however, violate all the rules of plot. Some of them have clearly defined beginnings …; others begin long years before anyone senses that something is wrong …. But they never end. Invisible contaminants remain a part of the surroundings, absorbed into the grain of the landscape, the tissues of the body, and worst of all, the genetic material of the survivors. An all clear is never sounded. The book of accounts is never closed.²¹

"Chronic disasters" is the name Erickson gives to the cumulative, insidious, and gradual harm that is characteristic of so many environmental problems.²² That the term at first seems oxymoronic—as if an emergency could last a lifetime—just shows how large the gap is between explicit intuitions about time and risk (surely, one thinks, perceptions of risk must decrease with the time it takes for a physical harm to occur), and the widespread, probably mostly unconscious amplification of perceived riskiness with time.²³

Latent hazards provoke another kind of psychological response as well. Latency is, simply, dormancy, and dormancy in this context refers to a condition in which a harmful agent, or the beginning of disease itself, is present but invisible. Latency thus creates a sense of contamination, of slow and invisible poison. Many people reserve a special dread for this kind of hazard. Again, Erickson's work is illuminating. Toxic substances, he writes,

invert the process by which disasters normally inflict harm. They do not charge in from outside and batter like a gust of wind or a wall of water. They slink in without warning, do no immediate damage so far as one can tell, and begin their deadly work from within—the very embodiment, it would seem, of stealth and treachery…. Toxic poisons provoke a special dread because they contaminate, because they are undetectable and uncanny and so can deceive the body's alarm systems, and because they can become absorbed into the very tissues of the body and crouch there for years, even generations, before doing their deadly work.²⁴

In interviews with people who lived near the nuclear reactor at Three Mile Island when it experienced the nation's worst nuclear power accident to date, Erickson discovered a widespread sense of the kind of contamination he describes: people feared that radiation from the power plant had infiltrated their bodies, their genes, their houses, yards, and gardens.²⁵ Simple events—like "grandchildren romping in [one's] backyard"²⁶—had become laced with dread.

The special anxieties associated with long-lived hazards can have large effects on the lives of individuals and communities. Individuals who have been exposed to substances whose physical effects will probably not become manifest for years, perhaps decades, have reported a wide range of adverse psychological responses to these exposures, including anxiety and anguish about their future health, depression, and physical conditions linked to their emotional distress, such as fatigue and insomnia, headaches, diarrhea, and muscle pain.²⁷ These anxieties can also provoke hormonal and immunological changes that cause or exacerbate many physical illnesses.²⁸ Although many of these responses have taken place in the context of heightened cancer risk, a similar set of responses has been reported by people who have been exposed to a risk of other illnesses whose physical effects are not immediately manifest. Indeed, a whole new branch of tort law has become concerned with the claims of persons who suffer heightened risk of a disease but who do not yet have physical symptoms of disease²⁹; this class of cases is concerned, almost by definition, with latent potential harms.³⁰

Certainly, anxieties and anxiety-based disorders are not limited to the context of latent harms. Anxiety-based tort claims, in fact, first developed in cases in which people suffered mental distress as a consequence of acute near-misses, such as barely escaping being run over by a train.³¹ There is a difference between these two classes of cases, however, and it has to do with the reasons why latent harms might provoke special anxieties in the first place: whereas a person involved [31 ELR 11059] in a near-miss with a train or a car knows immediately that she is safe, the person exposed to substances whose physical consequences become manifest only after a period of years cannot find the peace of mind that comes with the assurance of safety. Where there is no early test available to detect the presence (or absence) of disease in a person exposed to a disease-bearing agent, only time will tell whether the person will become sick.³² In this way, the emerging legal distinction between acute and chronic risks, and the awarding of recovery in tort for fear engendered by the former but not the latter,³³ has it exactly backwards, as it is precisely the person who does not yet know she is safe who has the most reason to be fearful. A veteran exposed to radiation during atomic testing described the awful uncertainty engendered by latent risk in this way: "The worst would be better than this."³⁴

Studies of communities that have lived through exposure to long-lived risk have also revealed profound and adverse effects on the communities themselves, beyond the individual reactions I have described. Erickson puts the point succinctly:

The experience of trauma, at its worst, can mean not only a loss of confidence in the self but a loss of confidence in the scaffolding of family and community, in the structures of human government, in the larger logics by which humankind lives, and in the ways of nature itself.³⁵

Within families and among neighbors, long-lived threats have often produced schisms and disputes that did not exist before the threatening exposures occurred. Typical problems include family disputes over whether to move away from the area where the exposures occurred or are occurring, and disputes among neighbors over whether to join legal or political efforts to combat the risks.³⁶ In addition, many researchers have found that one of the most dramatic effects of long-lived environmental threats is a loss of trust. Indeed, virtually all case studies of "contaminated communities," as Michael Edelstein calls them,³⁷ have found a generalized loss of trust in society's institutions.³⁸ This mistrust extends, most obviously, to the entities directly responsible for the threat, but it also reaches the local, state, and federal government entities deemed responsible for reacting to the threat. Indeed, in some cases, citizens have lost more faith in the government than in the polluter.³⁹ This loss of trust can have severe effects on citizens' relationship with their government and thus, in a participatory government like ours, on the functioning of government itself.⁴⁰ And, once lost, trust is hard to restore.⁴¹

The loss of trust experienced as a result of environmental contamination is connected to the long-term nature of environmental threats. The longevity of these threats increases the uncertainty associated with them, as years may pass before the exposed community learns the full consequences of its exposure. This uncertainty puts the government and other institutions involved in disclosing and explaining the threat in the position of wanting to say something about the threat, without having very much that they can say. The dilemma is well illustrated by the following statement by EPA regarding dioxin exposures in Missouri:

Dioxin in Missouri may present one of the greatest environmental problems in the history of the United States. Conversely, it may not.⁴²

In the absence of information about the actual physical effects of pollution—which, with long-term threats, may not be available for many years, if ever—people are forced to come to their own conclusions about the risks they face. The "nonempirical belief systems," as Henry Vyner has called them, that are formed in the absence of empirical information may consist either of denying any threat, or fearing the worst.⁴³ Government often embraces the former perspective,⁴⁴ citizens the latter. This divergence creates fertile space for mistrust.

Of course, communities can be destroyed by short-lived threats, too. Indeed, Erickson himself has produced one of the most famous accounts of the devastating effects an acute disaster—like a flood—can have on a community's sense of itself.⁴⁵ But latent threats are special in this sense: they can convert what would otherwise be a discrete, diffuse kind of harm—one death here, another there—into a catastrophe that tears the web of a whole community. When an exposed community will not know for many years whether anyone will fall ill or die as a result of their exposure, when, indeed, they may never know whether the illnesses and deaths they [31 ELR 11060] experience came from the exposure or from something else,⁴⁶ the whole community becomes involved in the threat of death—even if, ultimately, only a handful of illnesses and deaths will reasonably be attributed to the exposure they fear. In this way, there is a large temporal dimension to yet another of the qualitative features of risk which worry so many citizens: a long temporal lag between exposure and physical effects can transform a diffuse and individual harm into a collective harm, a disaster. Prolonged enough, risk itself becomes the disaster.

The social and psychological reactions I have described have been consistently reported by people who have lived through exposure to long-lived environmental threats.⁴⁷ These include the famous events at Three Mile Island, Love Canal, Times Beach, and Woburn, and less well-known toxic episodes in places like Hardeman County, Tennessee, and Legler, New Jersey. One might say, in fact, that the set of reactions I have described is so strikingly similar from place to place, and so intimately tied to the special nature of toxic exposures, that these reactions comprise a kind of syndrome.⁴⁸ Perhaps the most vivid expression of the clustering of the rather disparate reactions I have described, and their association with toxic exposures, is the "demoralization scale" developed by Bruce Dohrenwend—who led the task force on behavioral and mental health effects as part of the President's Commission on the Accident at Three Mile Island—and his colleagues.⁴⁹ The demoralization scale, a measure of "nonspecific stress,"⁵⁰ includes feelings of sadness, depression, loneliness, and anxiety; nervousness; restlessness; sour stomach; poor appetite; cold sweats; headaches; generalized physical ailments; and helplessness, hopelessness, uselessness, and failure.⁵¹ Populations exposed to various toxic episodes have been found to score "high" on the demoralization scale as compared to other populations, including the clients of mental health centers.⁵² Demoralization is, in short, a compact way of describing a large part of the cluster of social and psychological responses common to toxic exposures.

The prevalence of these psychological effects on individuals, and of their concomitant harms to communities, has two important implications for the temporal dimension of environmental law. First, these effects show the profound interaction between present conceptions of future prospects and present well-being. The account I have given here of the effects of dread on individuals and communities begins to create doubts about the possibility of isolating the benefits of environmental regulation within a single time frame, either the present or the future. Second, the social and psychological effects of dread mean that the benefits of environmental regulation that avoids these effects begin accruing immediately upon a reduction in risk. I do not mean to suggest that all or even most of the benefits of regulation occur immediately upon implementation, nor do I mean even to suggest that all of the benefits specifically relating to a reduction in dread occur immediately. I mean only to argue that some of these benefits begin to accrue as soon as a regulation is in place, and that they continue to accumulate throughout the life of the regulation. In this way, the risk-related benefits of environmental regulation are very much like the problems environmental regulation attacks: they are cumulative, gradual, and (positively) stealthy.

Expected Loss

A reduction in risk is a reduction in the estimated probability of loss, or a reduction in expected loss. Expected loss is a commodity distinct from the tangible loss (of life, of money, and so forth) itself. Thus environmental regulation produces a benefit both at the moment when it reduces a risk and at the moment when the relevant loss—here, an illness or death—would otherwise have occurred.⁵³

Probabilities of harm matter to people, and they matter long before the date on which a person subject to a hazard may fall ill or die as a result of it. Indeed, the entire current literature on the monetary value of a human life takes this as given: this literature seeks to identify, not the value a person places on life itself, but the value she places on small increases (or reductions) in risk to life.⁵⁴ The empirical evidence purporting to find, for example, that workers accept higher wages in return for accepting small increases in risk makes no sense unless risk itself is a kind of commodity, separate and apart from life itself, as no one is suggesting that workers would accept a finite sum of money in exchange for certain death.⁵⁵ While there is much reason to question the numerical valuations found in the wage premium studies that have been done to date,⁵⁶ there is little reason [31 ELR 11061] to question the basic premise that people conceive increased risk (or decreased risk) as a real cost (or benefit) to themselves. This premise is supported by other kinds of empirical evidence as well. It often happens, for example, that the property values of homes surrounding sources of environmental hazards, such as toxic waste sites, decrease as a result of these hazards, and researchers have found that decreases in value correspond to perceptions of increased risk.⁵⁷ Risk is, in short, sufficiently concrete, in and of itself, to have a value in the marketplace. Indeed, this insight is the basic foundation of the institution of insurance.⁵⁸

Other common behaviors also suggest that risk itself is a consequential event in people's lives. In many cases, people choose where to live,⁵⁹ where to work,⁶⁰ and even where to play⁶¹ based on the risks they face. Sometimes risk even affects the decision whether to have children.⁶² Risk also influences what people eat,⁶³ what they drink,⁶⁴ and how and where they bathe.⁶⁵ An increase in risk, in short, constrains options along a broad spectrum of human choice, from home to work to family to daily habits of living.

All of this suggests that, for many, probably most, people, risk is more than an abstraction, or if it is an abstraction, it is an abstraction with concrete consequences for their lives. It does not seem to me useful or important to decide which of these it is. The important point is that the imposition of a risk often brings with it compensating or averting behavior (such as a demand for higher wages, or refusal to move into a neighborhood, or a reluctance to go outside) that shows that risk itself is of palpable significance in people's lives, long before the date on which the physical consequences of the risk may make themselves manifest.⁶⁶ Thus a regulation reducing the kind of risk that people would rather avoid brings with it benefits beginning the moment it is in place.

Disease

Thus far I have focused on events apart from the development of physical disease arising out of environmental contamination. To be sure, many of the stress-related disorders described in the preceding section have physical manifestations and consequences. But the physical effects I have mentioned so far are the result of anxiety that follows from toxic exposures, rather than being the direct result of exposure itself. Now I would like to turn to the wide array of physical conditions and ailments that result directly from exposure to environmental agents. As we shall see, these adverse health states span a vast range, from skin rashes to cancer. Moreover, even within the context of a single disease—such as a single kind of cancer—one finds many different physical events that long precede the appearance of overt symptoms of disease.

These facts have large implications for the temporal distribution of environmentally related disease. First, many of the adverse health conditions caused by environmental agents appear, and appear overtly, either immediately or within a short interval after exposure. It is only by ignoring these health conditions—and the benefits of alleviating or preventing them through regulation—that one can say that environmental law mostly prevents remote future harms. Second, although the overt physical symptoms of some environmentally related diseases—most prominently, some forms of cancer—follow exposure by many years, the process of disease is often in place, and inexorable, long before detectable disease appears. That is, the period of disease induction—the period between a potentially harmful exposure and the initiation of disease⁶⁷—may be completed many years before the period of latency—the period in which disease is clinically undetectable⁶⁸—is over. I will argue that the periods of increased risk and disease induction are the ones relevant for purposes of identifying the temporal distribution of the benefits of environmental regulation, and that the latency period, which has figured so prominently in standard accounts of this temporal distribution, is unhelpful for this purpose. But first I will briefly describe the range of health effects that have been associated with environmental agents, and give a sense of the length of the time span represented by this range of harms.

Health Effects

A large number of adverse health conditions have been linked to exposure to harmful environmental [31 ELR 11062] agents.⁶⁹ These include rashes, blemishes, and other skin disorders; nausea; headaches; vision loss; decreased lung function; disorders of the respiratory, neurological, endocrine, and gastrointestinal systems; hematological disorders such as anemia; reproductive dysfunctions including decreased reproductive capacity and higher rates of miscarriage; birth defects; impaired cognitive functioning and development; chromosomal abnormalities; and virtually all varieties of cancer, including leukemia, and cancers of the lung, colon, pancreas, brain, breast, skin, and genito-urinary and lymphatic systems.

Given this impressive list, it is more than a little surprising that common wisdom has it that environmental law has been directed mainly at the prevention of cancer.⁷⁰ This impression cannot be attributed to the agencies' public statements about their rules. In describing the beneficial consequences of their rules, the federal agencies concerned with human health frequently, indeed almost invariably, discuss more than cancer cases avoided.⁷¹ And there has been more than enough media coverage of the non-cancer health effects of environmental agents, including most recently the attention devoted to toxic substances' role in disrupting the human endocrine system,⁷² to counter the impression that the whole environmental project is about preventing cancer.

The widespread misperception that cancer prevention is the dominant purpose of environmental regulation mostly stems, I believe, from two facts: first, cancer risk is, for a variety of reasons, more easily quantified than most other kinds of environmentally related health risks⁷³; second, quantified risks have come to dominate the national discussion about environmental policy.⁷⁴ Most academic and political accounts of the benefits of environmental regulation fixate exclusively on cancer prevention because cancer cases avoided are often the only benefits we can quantify.⁷⁵ But this in no way means that cancer is the only disease avoided by environmental law.

Of course, many of the health effects I have mentioned are not fatal, and in this way they may be deemed less serious than a disease like cancer which, without medical intervention, so often leads to death. Moreover, the public appears to fear cancer more than most other diseases, and this, too, may make cancer a more serious threat to human well-being.⁷⁶ But rashes, nausea, headaches, and shortness of breath also have large effects on the quality of life. In addition, many chronic, low-level health effects—various skin disorders, stomach complaints, headaches, breathing troubles, fatigue, and so forth—are vague enough, and common enough, that they are not easily attributed to exposure to harmful agents in the environment.⁷⁷ Very often, in fact, exposed citizens who have suffered these symptoms either have not even considered bringing them to the attention of a medical professional⁷⁸ or, if they have done this, have been told that the cause of their troubles cannot be determined.⁷⁹ The very generality of these symptoms, in other words, creates an uncertainty about their causation that itself can exacerbate anxiety.⁸⁰ Equally important, the chronic, relatively low-level effects of pollution often serve as early warning signs of even more serious troubles to come.⁸¹

In short, preventing the chronic, low-level effects of environmental contamination is an important benefit of environmental law, even though such effects are not fatal and their causation is difficult to pinpoint. To be sure, not every environmental regulation prevents every one of the adverse health conditions I have mentioned. But most environmental regulations prevent at least several of these ailments, and thus their human-health benefits range far beyond the prevention of cancer.⁸² Most important forpresent purposes, the ample range of illnesses prevented by environmental [31 ELR 11063] regulations also means that the benefits of environmental regulations are distributed over a wide temporal range.

In many cases, the health effects of pollution occur roughly contemporaneously with exposure to the pollution. Skin disorders, breathing troubles, headaches, and the like often follow soon after exposure occurs.⁸³ Indeed, as I have said, such symptoms often provide the first clues that something is amiss in the environment. Adverse responses to in utero exposure to harmful agents also often fall in this early range of consequences; overt birth defects due to in utero exposure to harmful agents, for example, inevitably appear within months of the fetus' exposure.⁸⁴ These early effects are typically followed by a range of chronic conditions that result from the steady accumulation of harmful exposures; health conditions such as anemia and various neurological disorders would fall in this intermediate temporal range. Last to appear are generally the cancers and other health conditions whose symptoms do not become overt until years after the process of disease has been initiated. In these cases, a disease may have begun its deadly course long before anyone can detect it. In the next section, I explain this process of disease induction in more detail, with particular reference to cancer, and I argue that the period in which a disease is initiated is the one that should be relevant to regulators trying to prevent that disease. At the same time, I argue that the latency period, as it has been defined in the medical sciences, is not helpful in evaluating the wisdom of life-saving regulations.

Induction and Latency

I have observed that, temporally speaking, the physical consequences of the many varieties of environmentally related disease together occupy a lengthy continuum. In this section I will explain that, even within the context of a single disease such as cancer, this temporal dynamic holds true. That is, the physical events relating to the development of cancer typically stretch over a long period, from the first exposure to a substance bearing the potential to transform a normal cell into a cancerous one, on until the development of overt symptoms of disease, including death. I will argue that the moments in this time line that should most concern regulators are the moments of increased risk and disease initiation, and that the latency period, which has so dominated current perceptions of the temporal dimension of environmental regulation, is of little importance to the regulatory process.

I must first define some terms. The "induction period" is the interval during which a process of disease is initiated. For cancer, this is the period in which a cell progresses from a normal state to a cancerous state, following exposure to a harmful stimulus. It is the period in which all of the component causes which lead to cancer fall into place.⁸⁵ The "latency period," on the other hand, is conventionally defined as the interval between the first exposure to a harmful stimulus and the appearance of clinically detectable disease. This definition is used both by scientists studying human disease⁸⁶ and by regulators attempting to describe the temporal distribution of environmentally related disease.⁸⁷ Finally, what might be called the "true" latency period is the interval during which the first cancerous cell (later cells) are present but undetectable. That is, it is the interval following disease induction and preceding disease detection—the time during which a process of disease is under-way but invisible.⁸⁸

Cancer is the most prevalent example of a disease that is initiated long before detectable physical symptoms appear.⁸⁹ As currently conceived, carcinogenesis is a multistage process in which a single normal cell must pass through several "irreversible, heritable, mutation-like event[s]," in a certain order, on the way to becoming a cancerous cell.⁹⁰ Every cancerous tumor, in turn, develops from a single abnormal cell.⁹¹ But there is often a long period between the time a cell becomes cancerous and the time the cancer can be detected.

In the case of breast cancer, for example, it is estimated that a cancer cell requires on average approximately 10 years to develop into a growth sizable enough to be picked up by a mammogram.⁹² By that time, the growth may already have spread to other areas of the body and, as a consequence, may be impossible to control.⁹³ Moreover, by the time the cancer has developed to the point where the cancer itself causes overt physical symptoms like nausea or headaches, it will often be past the point of remedy.⁹⁴ The onset of disease, in other words, may precede by many years the [31 ELR 11064] capacity to detect it.⁹⁵ Barry Castleman has explained this point vividly by describing the progression of asbestos-related respiratory disease:

The disease process would not become evident for the first few years of exposure no matter how intense the exposure was. Yet slowly but surely, the lung scarring would develop as the mineral fibers accumulated in the lungs and had time to provoke the characteristic response. Moreover, by the time the disease became evident, cessation of exposure could not halt the inexorable progress of the disease caused by the durable fibers already trapped in the lung tissues.⁹⁶

At each stage of the induction period for cancer, the regulator who reduces exposures to harmful agents reduces the risk that a normal cell will experience all of the changes required to become a cancerous one. Each of these changes has quite a small probability of occurring, and of taking permanent hold in the cell. The more a cell is exposed to an agent capable of causing these transformations, the more likely it is that carcinogenesis will occur. This helps to explain cancer's rather lengthy causal interval: in most cases, it takes time to accumulate exposures sufficient to cause the disease.⁹⁷ But science is not yet (and may never be) so precise that it is able to identify the very exposure that causes harm and, indeed, the etiology of cancer and other diseases often implicates not just one but multiple exposures.⁹⁸ Thus the most that can be said is that each exposure incrementally increases one's riskof developing a disease. The end of the induction period is marked by the moment when—if our skills of subclinical and clinical detection were sufficiently acute—we would see that a person has a disease, even if she has not experienced overt physical symptoms of a disease. Here, too, current science is not able to identify the moment when this happens.

For the regulator interested in preventing human illness, the crucial period in the development of disease should be the period during which the process of disease is initiated, that is, the induction period. It is only during this period that the regulator has a chance actually to prevent the disease, rather than merely react to it once it has developed. In my view, a regulator has succeeded in her work of reducing risk as soon as she prevents an exposure which could cause one of the stages of carcinogenesis in a cell, and she has succeeded in her work of preventing disease as soon as she prevents the disease from being initiated. To say this does not depend on our ability to identify the precise moment when these events occur. For purposes of describing the temporal range of the benefits of environmental regulation, it is enough to say that, for many diseases, including cancer, the moments of risk reduction and disease prevention long precede our ability to detect the disease itself. The moments of risk reduction and disease prevention are not, however, the ones important to standard latency analysis. For that reason I conclude that the concept of latency—which has dominated conventional understandings of the temporal dimension of environmental law—is not useful in the regulatory context.

The standard definition of the latency period for cancer is the period "between exposure to the carcinogenic stimulus and appearance of the clinically diagnosable cancer."⁹⁹ Thus this period embraces both the induction and "true" latency periods, as defined above. As my discussion of the induction period implies, however, the induction period is a time of important and destructive activity; it is the time during which a normal cell passes through the stages requisite to becoming a cancerous cell. To include this interval within the umbrella of "latency" falsely implies that it is a period of dormancy, even inactivity.

The problems go deeper than this. Latency analysis is, in the most general sense, a way of identifying the temporal interval between the exposure or other event of concern and the development of the health endpoint one deems relevant. For many diseases, analysts can choose from a variety of health endpoints in deciding which event marks the end of the latency period. In the case of cancer, for example, this endpoint may include any abnormal or adverse condition ranging from any one of the initial, cellular stages of carcinogenesis, to "precancerous" lesions, to a single malignant cell, to detectable aberrant growth, to detectable unbridled growth, to physical symptoms such as pain or nausea, to diagnosis, to death. Depending on which of these events one chooses to mark the end of the latency period, the latency period will vary enormously for a single kind of cancer.¹⁰⁰ Yet, as I shall explain, the choice of endpoint depends, not on science, but on the purposes of identifying the end of the latency period in a specific context. In medical science, the presence of clinically detectable disease is used to mark the end of the latency period because the purposes latency analysis serves in medicine and epidemiology require the ability to identify the time when a process of disease has assuredly [31 ELR 11065] begun. In the regulatory context, on the other hand, the moment when clinically detectable disease appears after, sometimes long after, the period in which regulation tries to do its work—that is, the period during which a disease is still preventable.

Latency analysis serves two basic purposes in medical science. First, identification of the latency period allows assurance, following exposure to a potentially harmful agent, that a particular individual will not contract a disease from that exposure. This assurance may simply permit personal peace of mind; given the latency period for lung cancer from cigarette smoking, for example, a person who quit smoking 25 years ago may rest assured that she now has a much lower risk of contracting lung cancer from smoking than does a person who did not quit.¹⁰¹ The knowledge that the latency period for a disease has passed also helps to determine the appropriateness of public health measures such as surveillance and screening. For example, since the incubation period for diphtheria is rarely more than one week,¹⁰² a person who has been in close contact with a person who has diphtheria need not be kept under close medical surveillance for more than one week following the last contact.¹⁰³ Likewise, given that most lung cancers from asbestos appear within 15-24 years following first exposure,¹⁰⁴ a person who has not developed lung cancer within approximately 25 years following his first exposure will not likely be advised to undergo the heightened cancer screening frequently recommended for persons who have been exposed to asbestos. Both of these uses of latency are, obviously, backward-looking in the sense that they provide information about the likely consequences of exposures that have happened in the past.

Medical science's second purpose in identifying latency periods is to trace the origins of human disease. The field of epidemiology uses the concept of latency to identify which time period is relevant to a particular study.¹⁰⁵ This will help to identify both the exposures and the adverse outcomes that will be included in the study.¹⁰⁶ For example, in order to determine whether a cluster of lung cancer cases in a particular workplace is attributable to workplace arsenic emissions, one must identify the levels of arsenic to which the workers were exposed. In order to do this, one must first figure out which arsenic exposures are relevant to the analysis. Only exposures that might plausibly have played a role in the development of the observed lung cancers are relevant. Latency analysis is necessary to identify these exposures. If, for example, the mean latency period for lung cancer due to occupational exposures to asbestos is 30-35 years,¹⁰⁷ then exposures that occurred much more than 35 years before the development of lung cancer did not likely cause the lung cancer. Latency is, then, an input to epidemiological analysis, one which helps to identify the possible causal agents of observed disease.¹⁰⁸ This purpose, too, is backward-looking, insofar as it aims to determine the etiology of already-existing human disease. It bears emphasizing that epidemiological analysis is not possible without making some assumption about latency; thus, despite potential misgivings about the variability and uncertainty surrounding latency analysis, one cannot do without it unless one wants to abandon epidemiological analysis itself.

Achieving these two purposes—assurances of safety and causal analysis—requires the capacity to identify a moment when the process of disease has assuredly begun. This requires some means of detecting the disease. Thus, it should come as no surprise that medical science has traditionally marked the end of the latency period with the development of clinically detectable disease.¹⁰⁹ Subclinical symptoms that may not lead to disease, such as "precancerous" lesions (including dysplasia of the cervical epithelium, tested for in pap smears),¹¹⁰ do not, on this view, qualify as the end of the latency period because they do not necessarily show the presence of disease. On the other hand, nondetectable conditions that will ultimately and inexorably lead to a clinically identifiable disease—such as the presence of a single cancerous cell—can also be of no help in serving the purposes medical science asks latency to serve, since those purposes depend on the ability to detect the presence of disease. Thus, given the purposes of latency analysis in medical science, the development of clinically detectable disease must be regarded as the relevant endpoint.¹¹¹

[31 ELR 11066]

Even in the regulatory context, latency analysis can help to elucidate the origins of the human diseases that concern regulators. But here its role ends. Specifically, identifying the moment when disease is clinically detectable does not help in identifying the moment when the benefits of risk regulation begin. Rather than attempting to determine, after the fact of exposure, whether the exposure caused the disease of interest, regulatory interventions attempt to prevent, before exposure, the development of disease. Thus the "latency period" for exposures to hazardous substances ends, from the regulatory point of view, as soon as the risk is reduced and the disease prevented.¹¹² From a preventive perspective, the moment when a disease becomes clinically detectable is simply beside the point; it comes past the time when prevention is possible.

Moreover, adopting the endpoint chosen by medical science (that is, the presence of clinically detectable disease) to mark the end of the "regulatory" latency period—and thus the beginning of the accrual of regulatory benefits—leads to perverse results in the regulatory context. Suppose that a regulator is considering regulating one of two substances, A or B. Suppose that each possible regulation will reduce exposures to each substance beginning on the very same date. Now suppose that the disease caused by substance A, malignant melanoma, a cancer of the skin, is detectable at an earlier stage, and thus more readily curable, than the disease caused by substance B, pancreatic cancer. It happens that the late detection of pancreatic cancer renders the disease hard, often impossible, to treat, and death often follows soon after a diagnosis of this disease.¹¹³ By measuring the end of the latency period—and thus the beginning of the production of regulatory benefits—from the presence of clinically detectable disease, the regulator would be pressed to regard the prevention of malignant melanoma as producing regulatory benefits sooner than the prevention of pancreatic cancer just by virtue of the relative delay in detecting pancreatic cancer. But this same delay makes pancreatic cancer harder to treat, and thus also seems to make the disease more, rather than less, important to prevent through regulatory measures.¹¹⁴ The same peculiar logic would justify treating cancer promoters different from cancer initiators insofar as promoters do their work relatively late in the process of carcinogenesis. Yet treating two carcinogens differently for regulatory purposes, depending on the stage of carcinogenesis at which they operate, would make no sense.¹¹⁵

Latency analysis is, in short, an essential concept for purposes of elucidating the causes of human disease. It is not, however, a useful concept in the regulatory context. The latency period for cancer, for example, varies by decades according to the health endpoint one regards as marking the end of the latency period. The choice among health end-points is not a scientific one, but a normative one, dependent on the purposes one seeks to achieve. Where, as in environmental law, those purposes are preventive, the endpoints of relevance are the increase in risk and the initiation of disease. Thus, just as the preventive purpose of environmental regulation was critical in arguing the relevance of the induction period to regulatory analysis, this purpose shows the ir-relevance of latency in this analysis.

Death

Many of the diseases prevented by environmental regulation will, without medical intervention, ultimately culminate in death.¹¹⁶ The prevention of death itself is a benefit of this regulation, separate and apart from the benefits of preventing risk and disease. The timing of death in the absence of regulation varies from disease to disease, and even from individual to individual. In many cases, it is a function of the state of advancement of the methods of clinical detection and treatment of disease. The sooner a disease is detected, the more effective the treatment usually can be. And, obviously, the more effective the treatment, the greater the chance of forestalling death. In any case, the time of death, in the absence of regulation, typically follows, sometimes by a long time, both the imposition of a risk and the development of a disease. Thus the recognition of the prevention of death as a separate regulatory benefit further expands the time line during which regulatory benefits accrue, but in this case, expands it into the future rather than toward the present.

A discussion about the timing of deaths from environmental exposures is, at least in part, a discussion about the loss of life expectancy that these exposures cause. Describing the loss in life expectancy caused by environmental [31 ELR 11067] agents is complicated because there is no neutral baseline for making the requisite comparison between the life expectancies of affected and nonaffected populations. What is the "normal" life span against which to compare the life span experienced in a population exposed to a pollutant of concern? Is it the average life expectancy of all of us? Of women? Of men? People of color? Healthy people? Sick people? Smokers? Drinkers? Even if one decides that the best approach is to identify the life expectancy of a group of people similarly situated (except for the harmful exposure in question) to the targeted population, and use that as the baseline for comparison, one will no doubt be required to conclude that certain determinants of life expectancy cannot, or should not, be incorporated in the analysis. One reason that some factors might be excluded from the analysis is that they may be endogenous to the problem under study. For example, the poor economic circumstances of a population surrounding a hazardous waste facility might imply a shortened life expectancy, and yet these economic circumstances may themselves be due, in part, to the hazardous waste facility itself. The simple point is that there will often be no obvious or easy way to identify the appropriate baseline.

Recognizing the prevention of death itself (apart from the reduction in risk and the prevention of disease) as a separate regulatory benefit expands the distribution of the benefits of environmental regulation into the more distant future. Identifying the specific moment when death would occur in the absence of regulation is, however, extremely complicated. Here, too, efforts at temporal precision must fail.

Future Generations

So far I have been discussing the temporal distribution of the benefits of environmental regulation as they occur within a single lifetime. Thus the temporal range I have described generally spans no more than, and in some cases considerably less than, a single generation.¹¹⁷ But many environmental regulations extend their reach beyond a single lifetime and generation. They do this in two different ways: first, by reducing exposure to persistent substances that can cause harm over more than one generation, and second, by changing the habits and perhaps even the culture of succeeding generations.¹¹⁸

Persistence

Many harmful agents take a long time to become harmless. These are, typically, agents that resist degradation by biophysical, chemical, or biological means.¹¹⁹ Radioactive substances are perhaps the most dramatic example of such a long-lived threat. The radium in existing tailings piles from uranium mills, for example, will take several hundred thousand years to decay to about 10% of current levels.¹²⁰ It would take about three million years for spent fuel from nuclear power plants to decay to a level where its toxicity matched that of the original uranium ore.¹²¹ And in planning for the disposal of the most radioactive of our radioactive wastes, EPA has dictated that the disposal site must be one that will remain undisturbed for at least 10,000 years.¹²²

Other persistent contaminants include chlorinated organic compounds such as polychlorinated biphenyls (PCBs), dichlorodiphenyltrichloroethane (DDT), chlordane, dieldrin, and dioxin.¹²³ These can persist in the environment, and in human tissue, for many years. The half life of chlordane in soil, for example, is approximately 1 year, and the half life of dioxin in soil is 10-12 years.¹²⁴ As much as 50% of the DDT applied to soil remains there 10-15 years following application.¹²⁵ Inorganic compounds, including heavy metals such as mercury, are also long-lived.¹²⁶ While the precise definition of "persistence" remains largely political and practical rather than scientific,¹²⁷ it nonetheless remains true that today's use and disposal of radioactive substances, chlorinated organic compounds, and heavy metals will continue to pose threats to human health for many decades, in some cases centuries, to come.

Indeed, if anything, these threats may grow with time. The longer persistent hazardous substances are allowed to remain in the environment, the more opportunity they have to escape into areas where they will do the greatest harm. A [31 ELR 11068] landfill loaded with PCBs may, today, pose little risk to the surrounding population if the PCBs are contained within the landfill; but the longer they remain there, the more chance they have to seep out of the landfill and into the surrounding soil and groundwater.¹²⁸ Similarly, the more persistent a substance, the more likely it is to survive long-range transport, and to be in a position to do harm at a place far remote from the place where it originated—and, perhaps, do greater harm because the people living in the place where the substance ends up probably do not know it is there.¹²⁹ Thus persistence has contributed to "the presence of compounds such as PCBs all over the world, even in regions where they have never been used. [Persistent organic pollutants] are ubiquitous. They have been measured on every continent, at sites representing every major climatic zone and geographic sector throughout the word."¹³⁰ Persistence, in other words, has a spatial as well as a temporal dimension, as it is a substance's durability which will largely determine how far that substance can travel without degrading.

Exposure to hazardous substances, and the absorption of them into human tissue, can also pose a risk to the next generation in utero. Persistent substances may still be around to do their harm when the person originally exposed becomes pregnant. For example, children born up to seven years after their mothers had ingested PCB-contaminated rice oil experienced symptoms consistent with the effects of PCBs, including hyperpigmentation, deformed nails and natal teeth, and cognitive and behavioral problems.¹³¹ Similarly, children born to mothers who had consumed slightly higher than average amounts of Great Lakes fish contaminated with PCBs and other persistent hazardous substances were found, on average, to have IQs six points lower than children born to mothers not so exposed.¹³²

Persistent substances may also still be around to do their harm when the person originally exposed to them begins to nurse her baby:

During breast feeding, human infants are exposed to higher concentrations of [persistent chemicals] than at any subsequent time in their lives. In just six months of breast feeding, a baby in the United States and Europe gets the maximum recommended lifetime dose of dioxin, which rides through the food web like PCBs and DDT. The same breast feeding baby gets five times the allowable daily level of PCBs set by international health standards for a 150-pound adult.¹³³

There seems to be little that can be done, in the short term, to reduce the amount ofpersistent chemicals passed on to infants through breastfeeding. In one study involving Dutch women, researchers found that low-fat diets did not reduce dioxin concentrations in breast milk. To make a difference, the researchers concluded, the dietary change would have to occur "years before the mother becomes pregnant."¹³⁴

Thus many environmental contaminants persist in air, water, land, human tissue, and even human milk for many years, and in this way can extend their harmful reach into generations beyond the one responsible for the contamination. Regulation reducing exposures to these substances can provide benefits stretching into the distant future.

Habits

Habits can be as persistent as PCBs. An action taken now may create a kind of dependency—on a particular production process, or product, or habit of living—that will make it ever more difficult to change the current course. The status quo achieves a kind of presumption or priority simply because it is the status quo. In this way, too, our actions now can reach into the future, and even into the next generation.

A production process may become entrenched for several reasons. For one thing, a company may have invested so much money in the current process that it is infeasible to change it. Converting the automobile manufacturing process to one that would produce the lower polluting "stratified charge" engine, for example, would require major and expensive retooling of car manufacturing plants.¹³⁵ The adoption of a particular production process by one company, moreover, may have a snowball effect, insofar as a process that has already been developed, tested, and used has more value to each subsequent user than a new or experimental process has. Production processes may, in other words, exhibit some of the features of "network effects," as their value may increase with the number of users.¹³⁶ Production processes may also become entrenched due to mere inertia and failure of imagination. Although some economists are skeptical that unexploited opportunities for cost-saving can exist in a competitive market,¹³⁷ numerous studies of the actual costs of regulatory programs have found that companies were actually able to save money as a result of regulation; regulation forced them to find new ways of producing goods and services which were superior [31 ELR 11069] to their previous practices.¹³⁸ Such research shows that some production processes may remain in place not due to the economic advantage they offer over other processes, but due to rather less rational tendencies on the part of company managers.

Products themselves may also become entrenched due to sunk costs, network effects, and simple inertia. Consumers may be unwilling to switch from incandescent to compact fluorescent lightbulbs, for example, if their household light fixtures and appliances are not compatible with compact fluorescents, and if they do not want to invest in new fixtures and appliances. Network effects may play a role here, too: products may increase in value with an increase in the number of users, as an increase in users will imply an increase in the ancillary services and facilities necessary to use the product. For gasoline-powered cars, for example, an increase in consumers implies an increase in gas stations, highways, parking facilities, and so forth, all of which increase the value of the product to its consumer. (Conversely, a small number of users may imply a lower value. One challenge which has faced manufacturers of electric cars, for example, is that there may be insufficient demand for the cars to produce a reasonably convenient number of battery-charging stations.) Finally, inertia may again be a powerful influence. Once a consumer has become accustomed to using a particular product—lighter fluid, or pesticides, or a riding lawn mower—she may become so accustomed to using the product that it will be difficult for her to change to a different one (or none at all), even if, had she not become familiar with the product in the first place, she would not have felt deprived. This is a particular instance of the general phenomenon of "status quo bias," which appears to play a large role in individual behavior.¹³⁹

In these ways, habits formed now, with respect to products and production processes, may influence behavior for many years to come. Whether such habits will influence behavior even into succeeding generations will depend on many factors, including the longevity of the products and processes as to which costs have already been incurred, the magnitude of network effects, and the force of inertia. One kind of habit, however, will reliably cast a shadow into the next generation, and that can best be described as the habit with regard to habits. Given the bias toward the status quo, the initial embrace of a habit assumes critical and lasting importance. If one becomes accustomed, for example, to having one's rooms warmer in the winter than one would tolerate them in the summer, or to driving everywhere rather than walking anywhere, or to running the dishwasher after every meal whether it is full or not, then these habits are likely to persist even if one could have done without them if one had always done without them. One's approach to forming such habits is, moreover, the kind of habit—ultimately shaping a lifestyle and, indeed, a value system—which one is likely to pass on to one's children, and thus it is the kind of habit likely to extend its reach into succeeding generations. Environmental regulation influencing these habits, and habits about habits, thus produces benefits reaching into the remote future.

Conclusion

The discussion in this part has illuminated three aspects of the temporal dimension of environmental law. First, it has shown that the benefits of environmental regulation extend over a vast temporal range, from the first reduction in risk, to the prevention of disease, to the delaying of death, both in this generation and succeeding generations. Second, the present and future events relevant to environmental law are not just sequential, but interactive. Present fears of future harms can have profound effects on present psychological and social well-being. Likewise, present physical conditions can inexorably lead to the development of a disease that will become manifest only many years hence. Thus the future reaches into the present, and the present into the future.

Finally, the discussion so far has made clear that if we are to reduce the harms I have described, we must begin now. The gradual, cumulative nature of many environmental risks means that we cannot wait to act until disease or death is upon us. Moreover, the persistent nature of many environmental threats makes it impossible to reduce the risk from them by acting later. And the formation of habits now will entrench practices that have future effects. What we do now, in other words, may make people sick, or even kill them, many years from now, but by the time that day nears it will be impossible to do anything to prevent it.

Regulatory Implications

I have argued that the benefits of environmental regulation cannot be isolated in time, but instead appear continuously over a large temporal interval. I have also argued that there is an interaction between present and future events which also renders any precise temporal placement of these benefits problematic. The practical implication of this analysis, for regulatory policy, is to cast doubt on the analytical technique of discounting human lives saved in the future. This practice is at odds with my account of the temporal dimension of environmental law.

Discounting is the calculation of the present value of a future benefit or cost. The calculation of the present value of a benefit (cost) is accomplished by applying a fixed discount rate to the benefit (cost) one expects eventually to receive (incur), over the period of time one must wait before one receives the benefit or incurs the cost.¹⁴⁰ The temporal analysis I have offered in this Article undermines both the practicability of discounting and its theoretical underpinnings.

[31 ELR 11070]

In practical terms, discounting cannot be done unless one knows how long one must wait before receiving the relevant benefit or incurring the relevant cost. As a consequence, the practice also requires quite precise identification of what the relevant benefit or cost is, and when it occurs. As we have seen, however, precision with accuracy is elusive in the environmental context. The benefits of environmental law are so multifarious, and extend over such a large and fluid temporal period, that the freeze-frame approach necessitated by discounting will sacrifice much by way of an accurate portrayal of the full range of benefits provided by environmental law. As usual, however, precision without accuracy is within easy reach.

For the most part, discounting in the context of life-saving environmental regulation has been accomplished to date by assuming that environmental regulation confers one benefit (the saving of quantified human lives) which accrues at one moment (the moment when a life-threatening illness would otherwise have become manifest).¹⁴¹ This approach ignores the many health-related benefits of environmental law that either cannot be quantified or do not lead to death. It also ignores the many benefits flowing from a reduction in risk and dread. Moreover, this approach embraces an awkward metaphysics by assuming that the moment when "life-saving" occurs is the moment when a life-threatening illness would otherwise have become patent, rather than the moment when a chain of events which would otherwise culminate in death is not allowed to begin. It is strange to claim that a life is saved, or death prevented, at the former moment rather than the latter. A simple example illustrates my point.

The symptoms of the deadly disease kuru most often appear many years after exposure to the disease-bearing agent. Indeed, kuru and other forms of Creutzfeldt-Jakob disease have come to be known as "slow viruses" because of the long period between exposure and symptoms.¹⁴² Imagine that you are a researcher studying kuru, and you are about to inject yourself with a substance which you mistakenly believe is healthy human blood, but is actually blood infected with kuru. Imagine further that, just before you insert the needle into your vein, a fellow researcher in your lab shouts "Stop! That blood's infected with kuru!" Would you regard your colleague as having saved your life when she warned you about the blood—or two decades later, when you otherwise would have learned that you had the deadly disease carried by the blood? I think it would be an unusual person who postponed her relief, and her gratitude, for the moment when she otherwise would have become manifestly ill rather than feeling them in the moment when the course of events that would have led to death was prevented from beginning. It is equally strange to conclude that the life-saving benefits of environmental regulation are not conferred until the moment illness would have become manifest in the absence of regulation.

Thus the discounters in the environmental context have erred both by fixating on only one of the many benefits of environmental law, and by picking the wrong moment from which to discount this one benefit. They might try to rehabilitate the practice of discounting by saying that they have picked this particular moment out of necessity. As I have noted, the relevant science has not yet advanced to the point where it can identify the precise moment when a long-developing disease (like cancer) is initiated. Thus, if one wants to discount, one cannot use the moment of disease initiation in order to do so, because discounting requires a quite precise definition of the moment when benefits accrue. Discounters might therefore argue that just as epidemiology, as a matter of necessity, uses initial exposure and clinical detectability to mark the beginning and end of the latency period, regulatory analysis similarly is forced to use an identifiable moment (like clinical detectability) to mark the moment when regulatory benefits accrue. But we need to mark this moment precisely only if we decide to discount. My argument is that one reason we should decide not to discount is that we cannot mark this moment precisely and accurately at the same time.

There is, however, one moment we can identify with both precision and accuracy, and that is the moment when risk is reduced. Risk begins to be reduced as soon as exposures to harmful environmental agents are reduced. Thus one might argue that discounting, if it is to be done at all, should be done only over the interval, if any, between the time when costs for risk reduction are expended and the time when risk reduction actually begins.¹⁴³ This approach would have the further virtue of being consistent with the current approach to valuing the very goods being discounted, which are the benefits of life-saving regulation. This approach asks how much individuals would be willing to pay (or to accept) in order to avoid (or suffer) a small increase in the probability of death.¹⁴⁴ This amount is then divided by the probability of harm in order to produce the value of a single "statistical life."¹⁴⁵ But the commodity being valued is the small reduction (or increase) in risk, not life itself.

The government's current approaches to valuation and discounting are inconsistent with each other. When the government discounts, it assumes that the moment when life-saving environmental regulation confers a benefit is the moment when physical illness would otherwise have become patent, not the moment when risk is reduced. When the government attaches a monetary value to the benefits of life-saving regulation, however, it assumes that the benefit being conferred is a reduction in risk.

Discounting the life-saving benefits of environmental law could be made both practicable, and consistent with current methods for valuing human life, by discounting from the date of the reduction in risk accomplished by environmental regulation rather than from the date of clinical detection in an unregulated environment. But even with this adjustment, discounting remains problematic in light of the analysis I have offered in this Article. For one thing, I have not singled out the reduction in risk as the only benefit of [31 ELR 11071] life-saving environmental regulations; the prevention of illness and death are also important benefits. Thus any analysis which isolates a single benefit, and discounts from the date on which that benefit accrues, will inadequately capture the full range of benefits conferred by environmental law.

Equally important, the analysis I have offered here casts considerable doubt on the theoretical case in favor of any kind of discounting. Two arguments have been especially prominent in making this case. The first sounds in preferences: most people, the argument goes, prefer postponed harm to immediate harm, and discounting simply reflects this common preference.¹⁴⁶ The second sounds in opportunities: if we do not discount future harms, we will miss opportunities to achieve the same good results with less money, or even more good results with the same amount of money.¹⁴⁷ I discuss each of these arguments in turn.

The argument from preferences is inconsistent with two of the lines of analysis I have offered in this Article. First, the assumption (usually rather casually and dubiously made) that people systematically prefer remote to immediate risks is in considerable tension with the evidence revealing that citizens reserve a special dread for hazards posing latent risks, and risks to future generations. Second, the assumption that people prefer remote harms is of little relevance to the question whether to discount the benefits of life-saving environmental regulation unless the harms against which such regulation protects us are indeed remote. As I have explained, however, many of the harms addressed by environmental law are not remote, but are instead immediate or near-term. Evidence that people in fact discount remote harms is not pertinent to evaluating the wisdom of preventing immediate harms.¹⁴⁸

The argument from opportunity costs is also undermined by the temporal analysis I have offered here. This argument is typically illustrated by the following kind of example:

Suppose … policy A saves [5] expected lives now and policy B saves [5] expected lives [10] years from now. Each policy costs $ 15 million…. Policy A is clearly more attractive because it saves [5] lives now for a cost at the time the lives are saved of $ 15 million. In contrast, policy B saves the [5] lives a decade from now, but the cost at the time the lives are saved is the $ 15 million invested this year plus any accrued interest over the next decade on this investment. Therefore, the cost per life saved during the year that lives are being saved will be greater for policy B.¹⁴⁹

Thus, to calculate the opportunity cost of life-saving interventions, one may either appreciate present-value estimates of costs to the date on which the benefits they produce will accrue, or discount estimates of benefits back to date on which the costs of producing them will be incurred; both techniques have the same operative effect.¹⁵⁰

Here, too, however, the assumption has been that the life-saving benefits of environmental regulation are realized only on the date when, but for the regulation, physical illness would otherwise have been detected. As I have explained, however, where regulation prevents the initiation of an inexorable course of fatal illness, we should conceive of life-saving as occurring at the moment of prevention. Thus the argument from opportunity costs does not justify discounting from the moment disease is detected.

More fundamentally, the argument from opportunity costs rests on the mistaken premise that opportunities to engage in life-saving at a future date—specifically, on the date when the illnesses to be prevented by regulation would otherwise have become manifest—will be equivalent to those we enjoy today. Indeed, the argument from opportunity costs has spawned an argument from absurdity that depends quite explicitly on an assumption that future life-saving opportunities in the future will be equivalent to today's. The argument from absurdity is this: unless we discount the future benefits of life-saving regulation, we will postpone all life-saving indefinitely because we will always be able to argue that the pot of money we might spend today will only grow with time and therefore can be put to greater use, the further into the future our regulation is delayed.¹⁵¹ But this assumes that we can achieve the same regulatory purposes by postponing regulation as by acting now. As I have argued here, however, this is not the case. For the kinds of harms prevented by environmental law—cumulative, insidious, persistent, and irreversible—postponing regulation until the moment when illness would otherwise become manifest amounts to an admission of defeat.¹⁵² The argument from opportunity costs ignores this aspect of the temporal dimension of environmental law.

Conclusion

I have tried to provide as completely as possible a picture of the temporal distribution of the beneficial consequences of environmental law. As we have seen, the picture that emerges is of a lengthy continuum, stretching from now until centuries from now. The benefits of environmental law begin with the reduction in risk, which, I have argued, is a benefit separate and apart from the prevention of illness or death. A large part of my discussion on this point was devoted to exploring the adverse psychological and social consequences of risk on individuals and communities—a phenomenon to which I have attached the shorthand label "dread." The reduction in risk of disease and death is not, however, the only consequence that may follow immediately upon the implementation of environmental regulation. Many adverse health effects—including nonfatal, acute effects such as headaches, rashes, and nausea—occur roughly contemporaneously with exposure to harmful environmental agents. In many cases, environmental regulation thus causes not only an immediate reduction in risk, but also an immediate reduction in physical illness. Next in this temporal [31 ELR 11072] sequence come the chronic adverse health conditions associated with cumulative exposures to hazardous substances, including various hematological, respiratory, and neurological disorders. Last to appear, almost by definition, are the long-latency diseases such as cancer. All of these conditions may appear not only in the present generation, but also in future generations, as a result of our actions today. Thus the temporal chain of the benefits of environmental regulation that reduces human exposures to agents posing a threat to human health is indeed long, stretching from the first implementation of the regulation, on into the distant future.

Throughout my discussion, however, I have stressed that the present and future are not only sequential, but interactive, and that this interaction sometimes works in a way that makes remote human health consequences as troublesome as immediate ones, and perhaps even more so. Indeed, as I have suggested, there often exists something like an inverse relationship between anxiety and immediacy. The interaction between present and future also works forward as well as backwards. Many human diseases, including cancer, exhibit a long interval between a harmful exposure and the clinical appearance of disease. But such diseases typically set their victims on an inexorable course toward illness and death long before clinical symptoms emerge. Thus the present can determine the future before we know it.

My analysis shows that it is a mistake to conceive of the good consequences of environmental regulation as existing within a single time frame, either the present or the future. This analysis has important implications for regulatory policy. In particular, my account of the temporal dimension in environmental law casts a shadow over the analytic technique that has come to dominate the government's account of the relationship between environmental law and time: the discounting of future lives saved. This technique underlies much of the analysis concluding that environmental law costs too much and achieves too little. But discounting hinges on a narrow, indeed single-minded, definition of regulatory benefits, and a narrow and static view of the timing of those benefits. These premises are at odds with the diverse, lengthy, continuous, and interactive time line of regulatory benefits I have sketched here.

1. See Benefit-Cost Analysis of Federal Programs; Guidelines and Discounts, 57 Fed. Reg. 53519, 53522-23 (Nov. 10, 1992).

2. OMB in its early years recommended a discount rate of 10%, see OMB, EXECUTIVE OFFICE OF THE PRESIDENT, CIRCULAR A-94, at 4 (1972), but recently changed its recommended discount rate to 7%. See Benefit-Cost Analysis of Federal Programs; Guidelines and Discounts, supra note 1, at 53522-23.

3. For an account of the controversy between OMB and EPA over whether to discount the life-saving benefits of EPA's 1989 rule banning asbestos, see HOUSE SUBCOMM. ON OVERSIGHT & INVESTIGATIONS, 99TH CONG., EPA's ASBESTOS REGULATIONS: REPORT ON A CASE STUDY ON OMB INTERFERENCE IN AGENCY RULEMAKING (Comm. Print 1985).

4. While EPA has, in its general guidelines on economic analysis, embraced discounting in the intragenerational context, its position on discounting for latency remains unclear. Compare U.S. EPA, GUIDELINES FOR PREPARING ECONOMIC ANALYSIS §§ 6.3 to 6.4 (2000) (endorsing discounting in the intragenerational context) with U.S. EPA, ARSENIC IN DRINKING WATER RULE ECONOMIC ANALYSIS 5-27 to 5-28 (2000), available at http://www.epa.gov/safewater/ars/econ_analysis.pdf (last visited June 28, 2001) (declining to discount for latency in primary economic analysis due to uncertainty concerning length of latency period for arsenicrelated cancers).

5. See generally Lisa Heinzerling, Regulatory Costs of Mythic Proportions, 107 YALE L.J. 1981 (1998) [hereinafter Heinzerling, Regulatory Costs of Mythic Proportions].

6. For further discussion and criticism of discounting, see Lisa Heinzerling, Discounting Our Future, 34 LAND & WATER L. REV. 39 (1999); Lisa Heinzerling, Discounting Life, 108 YALE L.J. 1911 (1999).

7. Paul Slovic, Perception of Risk, 236 SCIENCE 280, 282 (1987) [hereinafter Slovic, Perception of Risk].

8. Id. at 281; Paul Slovic et al., Facts and Fears; Understanding Perceived Risk, in SOCIAL RISK ASSESSMENT: HOW SAFE IS SAFE ENOUGH? 192 (R.C. Schwing & W.A. Alberts Jr. eds., 1980).

9. J.J.J. Christensen-Szalanski et al., Effects of Expertise and Experience on Risk Judgments, 68 J. APPLIED PSYCHOL. 278 (1983), discussed in Lola L. Lopes, Risk Perception and the Perceived Public, in THE SOCIAL RESPONSE TO ENVIRONMENTAL RISK: POLICY FORMULATION IN AN AGE OF UNCERTAINTY 57, 61-62 (Daniel W. Bromley & Kathleen Segerson eds., 1992) [hereinafter THE SOCIAL RESPONSE TO ENVIRONMENTAL RISK: POLICY FORMULATION IN AN AGE OF UNCERTAINTY]; Paul Slovic, Trust, Emotion, Sex, Politics, and Science: Surveying the Risk Assessment Battlefield, 1997 U. CHI. LEGAL F. 59, 68-69, 83-87 [hereinafter Slovic, Trust, Emotion, Sex, Politics, and Science].

10. Slovic, Perception of Risk, supra note 7, at 283; WILLIAM W. LOWRANCE, OF ACCEPTABLE RISK: SCIENCE AND THE DETERMINATION OF SAFETY 87, at fig. 3-1 (1976); Robin Gregory & Robert Mendelsohn, Perceived Risk, Dread, and Benefits. 13 RISK ANALYSIS 259 (1993) (using regression techniques to identify major explanatory variables for risk perceptions reported by laypeople in other studies).

11. Gregory & Mendelsohn, supra note 10, at 261.

12. Id. at 262; Slovic, Trust, Emotion, Sex, Politics, and Science, supra note 9, at 81 & nn.65-67.

13. HENRY M. VYNER, INVISIBLE TRAUMA: THE PSYCHOSOCIAL EFFECTS OF INVISIBLE ENVIRONMENTAL CONTAMINANTS 14-18 (1988).

14. Clayton P. Gillette & James E. Krier, Risks, Courts, and Agencies, 138 U. PA. L. REV. 1096 (1990). See also Laura M. Davidson et al., Toxic Exposure and Chronic Stress at Three Mile Island, in 6 ADVANCES IN ENVIRONMENTAL PSYCHOLOGY: EXPOSURE TO HAZARDOUS SUBSTANCES: PSYCHOLOGICAL PARAMETERS 35, 44 (Allen H. Lebovits et al. eds., 1986) [hereinafter ADVANCES IN ENVIRONMENTAL PSYCHOLOGY] (uncertainty about future consequences of past exposures to harmful agents may increase perceptions of loss of control).

15. See Michael Gaffney & Bernard Altshuler, Public Health Implications of Carcinogenic Exposure Under the Multistage Model, 124 AM. J. EPIDEMIOLOGY 1021, 1029 (1986).

16. LOWRANCE, supra note 10, at 93-94.

17. As Ralph Keyes has written:

Only an actual masochist enjoys danger as such. Yet we all enjoy its by-products: alertness, intensity, and a sense of elation once danger has past [sic]. With its faster pulse, shortness of breath, and copious perspiration, the body responds to moderate stress much as it does to physical exercise…. As with exercise, short doses of tolerable stress are essential for keeping body and spirit tuned…. But it must be emphasized that this means occasional stress at tolerable levels. There is little good to be said for even occasional panic, constant phobias, or nagging anxiety.

RALPH KEYES, CHANCING IT: WHY WE TAKE RISKS 35 (1985).

18. Lola L. Lopes, Between Hope and Fear: The Psychology of Risk, 20 ADVANCES IN EXPERIMENTAL SOC. PSYCHOL. 255, 288 (Leonard Berkowitz ed., 1987) (quoting KEYES, supra note 17, at 115); KEYES, supra note 17, at 41, 62. See also Paula Horvath & Marvin Zuckerman, Sensation Seeking, Risk Appraisal, and Risky Behavior, 14 PERS. INDIVIDUAL DIFFERENCES 41, 41 (1993).

19. KEYES, supra note 17, at 65 (thrill-seekers report that one of the main rewards of risk-taking is "the calm that follows" the danger); cf. W. KIP VISCUSI, RATIONAL RISK POLICY 11-12 (1998) [hereinafter VISCUSI, RATIONAL RISK POLICY] (people willing to pay "certainty premium" to be assured of zero risk, over and above amount they would be willing to pay for similar reductions in risk that achieve nonzero levels of risk); see also Lopes, supra note 18, at 278 (discussing "seemingly special status of certainty in risky choice").

20. See, e.g., VYNER, supra note 13, at 55-57.

21. KAI ERICKSON, A NEW SPECIES OF TROUBLE: THE HUMAN EXPERIENCE OF MODERN DISASTERS 147-48 (1994) (quoting ARISTOTLE, THE POETICS 29-31 (W. Hamilton Fry trans., Harvard Univ. Press 1932)) [hereinafter ERICKSON, A NEW SPECIES OF TROUBLE]. See also MICHAEL R. EDELSTEIN, CONTAMINATED COMMUNITIES: THE SOCIAL AND PSYCHOLOGICAL IMPACTS OF RESIDENTIAL TOXIC EXPOSURE 9 (1988) ("A sense of finality is elusive for the toxic victim, in part because toxic disasters lack a 'low point' from which things would be expected to improve." (citations omitted)).

22. ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 21-22.

23. For other accounts sounding the same theme, see, for example, EDELSTEIN, supra note 21; PHIL BROWN & EDWIN J. MIKKELSEN, NO SAFE PLACE: TOXIC WASTE, LEUKEMIA, AND COMMUNITY ACTION (1990).

24. Kai Erickson, Toxic Reckoning: Business Faces a New Kind of Fear, HARV. BUS. REV., Jan./Feb. 1990, at 118, 122 [hereinafter Erickson, Toxic Reckoning].

25. Id. at 123-24.

26. Id. at 124.

27. For discussion and a complete list of sources, see Lisa Heinzerling, Environmental Law and the Present Future, 87 GEO. L.J. 2025, 2034-35 (1999) [hereinafter Heinzerling, Environmental Law and the Present Future].

28. Andrew Baum, Disasters, Natural & Otherwise, PSYCHOL, TODAY, Apr. 1988, at 56, 60; see also Andrew Baum et al., Emotional, Behavioral, and Physiological Effects of Chronic Stress at Three Mile Island, inREADINGS IN SOCIAL PSYCHOLOGY: GENERAL, CLASSIC, AND CONTEMPORARY SELECTIONS (Wayne A. Lesko ed., 1991) (documenting the role of the Three Mile Island accident in increasing levels of stress—as indicated by emotional, behavioral, and physiological measures—in the population surrounding the nuclear power plant); Nancy Levit, Ethereal Torts, 61 GEO. WASH. L. REV. 136, 184 (1992) (noting effect of hopelessness on levels of depression and suicide, incidence and spread of cancer, and the immune system).

29. For compilations and discussion of the relevant cases, see Mary Donovan, Is the Injury Requirement Obsolete in a Claim for Fear of Future Consequences?, 41 UCLA L. REV. 1337, 1369-80 (1994); Adam P. Rosen, Emotional Distress Damages in Toxic Tort Litigation: The Move Toward Foreseeability, 3 VILL. ENVTL. L.J. 113 (1992).

30. Donovan, supra note 29, at 1341-42; Bill C. Wells, The Grin Without the Cat: Claims for Damages From Toxic Exposure Without Present Injury, 18 WM. & MARY J. ENVTL. L. 285, 310 (1994).

31. See, e.g., Mack v. South-Bound R.R., 52 S.C. 323 (1898); see also cases cited in Metro-North Commuter R.R. v. Buckley, 521 U.S. 424, 430-31 (1997).

32. Cf. Faya v. Almaraz, 620 A.2d 327, 337 (Md. 1993) (plaintiffs' recovery for mental distress arising from exposure to the acquired immune deficiency syndrome (AIDS) virus limited to period during which she had not yet received test results indicating she was human immunodeficiency virus (HIV)-negative); Kerins v. Hartley, 21 Cal. Rptr. 2d 621, 632 (Ct. App. 1993) (same).

33. See, e.g., Metro-North, 521 U.S. at 430 (disallowing tort recovery for mental distress caused by "simple physical contact with a substance that might cause a disease at a substantially later time").

34. VYNER, supra note 13, at 57.

35. ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 242.

36. See, e.g., Margaret S. Gibbs, Psychopathological Consequences of Exposure to Toxins in the Water Supply, in ADVANCES IN ENVIRONMENTAL PSYCHOLOGY, supra note 14, at 52; Adeline G. Levine & Russell A. Stone, Threats to People and What They Value, in ADVANCES IN ENVIRONMENTAL PSYCHOLOGY, supra note 14, at 125; ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 122-28.

37. EDELSTEIN, supra note 21.

38. See, e.g., Bruce P. Dohrenwend, Psyhological Implications of Nuclear Accidents: The Case of Three Mile Island, 59 BULL. N.Y. ACAD. MED. 1060 (1983); BROWN & MIKKELSEN, supra note 23, at 118-20; EDELSTEIN, supra note 21, at 70-82; Kenneth M. Bachrach & Alex J. Zautra, Assessing the Impact of Hazardous Waste Facilities: Psychology, Politics, and Environmental Impact Statements, in ADVANCES IN ENVIRONMENTAL PSYCHOLOGY, supra note 14, at 84; Gibbs, supra note 36, at 52; ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 129-33; Levine & Stone, supra note 36, at 127; M.R. Fowlkes & P.T. Miller, LOVE CANAL: THE SOCIAL CONSTRUCTION OF DISASTER (1982).

39. See, e.g., EDELSTEIN, supra note 21, at 80-81 (describing citizens' reactions to events at Love Canal).

40. See, e.g., Paul Slovic, Perceived Risk, Trust, and Democracy. 13 RISK ANALYSIS 675 (1993).

41. Id. at 677; see also Paul Slovic et al., Perceived Risk, Trust, and the Politics of Nuclear Waste, 254 SCIENCE 1603, 1606 (1991); Dohrenwend, supra note 38, at 1071-72 (finding high levels of distrust among residents living near the Three Mile Island nuclear reactor, stemming from accident there; distrust did not dissipate after the accident, as mental distress did).

42. EDELSTEIN, supra note 21, at 76 (quoting ST. LOUIS POST-DISPATCH, Nov. 14, 1983).

43. See generally VYNER, supra note 13.

44. See, e.g., Lee Clarke, Politics and Bias in Risk Assessment, 25 SOC. SCI. J. 155, 161 (1988) (government officials often defend risky systems even more adamantly after a disaster has occurred); VYNER, supra note 13, at 182-88 (invisibility of environmental threats allows government to deny their seriousness).

45. KAI ERICKSON, EVERYTHING IN ITS PATH: DESTRUCTION OF COMMUNITY IN THE BUFFALO CREEK FLOOD (1976).

46. Many environmental contaminants pose the problem of what Henry Vyner calls "etiological invisibility," that is, the cause of a particular health effect cannot be isolated. Ionizing radiation, for example, leaves no "marker" that distinguishes leukemias caused by it from leukemias caused by something else. VYNER, supra note 13, at 15-16.

47. For an excellent summary of these effects, see Michael B. Gerrard, Fear and Loathing in the Siting of Hazardous and Radioactive Waste Facilities: A Comprehensive Approach to a Misperceived Crisis, 68 TUL. L. REV. 1047, 1137-46 (1994).

48. Cf. VYNER, supra note 13, at 121-40 (describing features of what author calls "radiation response syndrome," a collection of symptoms found among veterans exposed to radiation during testing of atomic bombs).

49. See Bruce P. Dohrenwend et al., Stress in the Community: A Report to the President's Commission on the Accident at Three Mile Island, 365 ANNALS N.Y. ACAD. SCI. 159 (1981).

50. See Dohrenwend, supra note 38, at 1067.

51. See Jeffrey S. Markowitz & Elane M. Gutterman, Predictors of Psychological Distress in the Community Following Two Toxic Chemical Incidents, in ADVANCES IN ENVIRONMENTAL PSYCHOLOGY, supra note 14, at 95, tbl. 6.2.

52. See, e.g., id. at 103; Bachrach & Zautra, supra note 38, at 79, 83-85. As Dohrenwend puts it: "High scores on this symptom scale are in some ways analogous to body temperature: when it goes up, one knows something is wrong." Dohrenwend, supra note 38, at 1067-69.

53. See also Kenneth J. Arrow, Behavior Under Uncertainty and Its Implications for Policy, in FOUNDATIONS OF UTILITY AND RISK THEORY WITH APPLICATIONS 19, 21 (Bernt P. Stigum & Fred Wenstop eds., 1983) [hereinafter FOUNDATIONS OF UTILITY AND RISK THEORY WITH APPLICATIONS]:

The risk of a disutility is itself a cost and a proper subject for measurement along with the direct costs of the usual resource-using type. Similarly, a reduction in risk is to be counted as a benefit. This is true even if individuals are risk-neutral, since we would still want to count the expected value of the risk; in the presence of risk aversion, there is still an additional cost or benefit, as the case may be.

54. See, e.g., VISCUSI, RATIONAL RISK POLICY, supra note 19, at 45.

55. See, e.g., JOHN M. MENDELOFF, THE DILEMMA OF TOXIC SUBSTANCE REGULATION: HOW OVERREGULATION CAUSES UNDERREGULATION AT OSHA 27 (1988).

56. The critiques by the late economist Don Shakow are particularly illuminating and incisive. See, e.g., Don Shakow, Market Mechanisms for Compensating Hazardous Work: A Critical Analysis, in EQUITY ISSUES IN RADIOACTIVE WASTE MANAGEMENT (Roger E. Kasperson ed., 1983); Julie Graham & Don Shakow, Hazard Pay for Workers: Risk and Reward, 23 ENV'T 14 (1981); Julie Graham, Don Shakow, & Christopher Cyr, Risk Compensation—in Theory and in Practice, 25 ENVIRONMENT 14 (1983).

57. See, e.g., Howard Kunreuther & Douglas Easterling, Gaining Acceptance for Noxious Facilities With Economic Incentives, in THE SOCIAL RESPONSE TO ENVIRONMENTAL RISK: POLICY FORMULATION IN AN AGE OF UNCERTAINTY supra note 9, at 155. See also ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 116 n.* (reporting that one family's home had been declared to have "no value" due to leaking underground storage tank).

58. See, e.g., WARREN FREEDMAN, 1 FREEDMAN'S RICHARDS ON INSURANCE 12 (6th ed. 1990).

59. See, e.g., Valerie Preston et al., Adjustment to Natural and Technological Hazards: A Study of an Urban Community, 15 ENV'T & BEHAV. 143, 160-61 (1983) (finding that 12% of residents of a community exposed to various technological risks had attempted to relocate in response to the risks, and that one-third of the residents had considered relocating).

60. See, e.g., W. KIP VISCUSI, RISK BY CHOICE: REGULATING HEALTH AND SAFETY IN THE WORKPLACE 67 (1983) [hereinafter VISCUSI, RISK BY CHOICE] (finding that "workers' risk perceptions had a powerful influence on their intentions [and propensities] to quit").

61. See, e.g., ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 156.

62. See, e.g., BROWN & MIKKELSEN, supra note 23, at 95; VYNER, supra note 13, at 48; Donald G. Unger et al., Living Near a Hazardous Waste Facility: Coping With Individual and Family Distress, 62 AM. J. ORTHOPSYCHIATRY 55, 57 (1992).

63. See, e.g., ROBERT V. PERCIVAL ET AL., ENVIRONMENTAL REGULATION 440 (3d ed. 2000) (reporting "rapid decline in apple consumption" following television report on the risks of Alar, a growth regulator used on apples); ANASTASIA M. SHKILNYK, A POISON STRONGER THAN LOVE: THE DESTRUCTION OF AN OJIBWA COMMUNITY 202-03 (1985) (describing effects of mercury contamination of river on community's consumption of fish).

64. For a detailed account of the effect on daily life of being advised not to drink or use one's tap water, see EDELSTEIN, supra note 21, at 34-37.

65. See, e.g., Potter v. Firestone Tire & Rubber Co., 6 Cal. 4th 965, 978 n.2 (1993); EDELSTEIN, supra note 21, at 29, 36.

66. See also Robin Pope, The Pre-Outcome Period and the Utility of Gambling, in FOUNDATIONS OF UTILITY AND RISK THEORY WITH APPLICATIONS, supra note 53, at 137, 158 (discussing the "tangible costs of uncertainty"); MARK R. GREENE & JAMES S. TRIESCHMANN, RISK AND INSURANCE 3, 35 (7th ed. 1988) (discussing economic costs of risk, and economic benefits of insurance).

67. Kenneth J. Rothman, Induction and Latent Periods, 114 AM. J. EPIDEMIOLOGY 253, 253 (1981).

68. Id. at 254.

69. For discussion and a complete list of sources, see Heinzerling, Environmental Law and the Present Future, supra note 27, at 2047-49.

70. See, e.g., STEPHEN BREYER, BREAKING THE VICIOUS CIRCLE: TOWARD EFFECTIVE RISK REGULATION 6 (1993) (cancer is "the engine that drives much of health risk regulation"); John F. Morrall III, A Review of the Record, REGULATION, Nov./Dec. 1986, at 32.

71. See Heinzerling, Regulatory Costs of Mythic Proportions, supra note 5, at nn. 108, 136, 151, 165, 272, 331, 354, 382 and accompanying text.

72. For a discussion of the scientific evidence of non-cancer health effects from hormone-disrupting chemicals, see THEO COLBORN ET AL., OUR STOLEN FUTURE (1996).

73. See, e.g., OFFICE OF POLICY ANALYSIS, U.S. EPA, UNFINISHED BUSINESS: A COMPARATIVE ASSESSMENT OF ENVIRONMENTAL PROBLEMS 98 (1987).

74. See generally Lisa Heinzerling, Reductionist Regulatory Reform, 8 FORDHAM ENVTL. L.J. 459 (1997) [hereinafter Heinzerling, Reductionist Regulatory Reform].

75. See, e.g., Heinzerling, Regulatory Costs of Mythic Proportions, supra note 5, at 2060-61.

76. One study has found that people place a negative value on death from cancer that is about three times higher than the negative value of instant death. George Tolley et al., State-of-the-Art Health Values, in VALUING HEALTH FOR POLICY: AN ECONOMIC APPROACH 341 (George Tolley et al. eds., 1994).

77. See, e.g., SHKILNYK, supra note 63, at 192-97 (describing low-level effects of mercury poisoning); ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 36 (on subtlety of low-level mercury poisoning in a poor Ojibwa community: "It is almost as if mercury poison mimics and ridicules the suffering of the already damaged."); Allen H. Lebovits et al., The Case of Asbestos-Exposed Workers, in ADVANCES IN ENVIRONMENTAL PSYCHOLOGY, supra note 14, at 5 (describing "vague physical complaints" associated with polybrominated biphenyls (PBBs)); G.G. Brown & R. Nixon, Exposure to Polybrominated Biphenyls: Some Effects on Personality and Cognitive Functioning, 242 JAMA 523 (1979) (same).

78. See, e.g., Levine & Stone, supra note 36, at 117 (at Love Canal, health problems included those "of a chronic, low level nature, such as excess fatigue, mild skin rashes, headaches, and allergies, the sorts of problems that people try to care for themselves").

79. See, e.g., VYNER, supra note 13, at 147-66, Vyner has captured the problem well, in discussing veterans exposed to radiation during testing of atomic bombs: "In contrast to the hypochondriac, who appears to be a person in search of an illness, the atomic veteran is a person in search of meaning for a disease that he has already developed." Id. at 132 (emphasis in original). See also Robert R.M. Verchick, In a Greener Voice: Feminist Theory and Environmental Justice, 19 HARV. WOMEN'S L.J. 23, 47-48 (1996):

Scientific experts and government bureaucrats, far removed from a contaminated site, are sometimes slow to link health problems or property damage to environmental contamination. In contrast, residents of polluted communities, who every day tend the gardens, do the laundry, and care for their children, are much more likely to notice the first clues of an environmental threat.

80. Vyner terms this kind of uncertainty "diagnostic ambiguity," consisting of "the absence of a diagnostic, or conceptual, category in the presence of apparent somatic symptoms." VYNER, supra note 13, at 16-17.

81. See, e.g., EDELSTEIN, supra note 21, at 28, 34 (describing early warning signs that something was amiss in Legler, New Jersey; the "water was funny" and had the "worst smell," people developed "itchy, dry skin" and rashes and other skin irritations).

82. And this is all without any consideration of the ecological benefits of environmental law. For a discussion of the slighting of ecological benefits in accountings of the benefits of environmental law, see Heinzerling, Reductionist Regulatory Reform, supra note 74.

83. See, e.g., OFFICE OF TECH. ASSESSMENT, HABITABILITY OF THE LOVE CANAL AREA: AN ANALYSIS OF THE TECHNICAL BASIS FOR THE DECISION ON THE HABITABILITY OF THE EMERGENCY DECLARATION AREA-A TECHNICAL MEMORANDUM, OTA-TM-M-13 45 (1983).

84. Due to the persistence of many environmental contaminants within human tissue, however, birth defects and other problems resulting from in utero exposure to these harmful agents may follow by several years the mother's exposure to them.

85. Cf. Rothman, supra note 67, at 254 (discussing induction period generally).

86. See, e.g., id.

87. National Emission Standards for Hazardous Air Pollutants; Benzene Emissions From Chemical Manufacturing Process Vents, Individual Solvent Use, Benzene Waste Operations, Benzene Transfer Operations, and Gasoline Marketing Systems, 55 Fed. Reg. 8292 (Mar. 7, 1990).

88. Rothman uses the term "latency period" to apply only to the interval between disease initiation and detection, and uses the term "empirical induction period" for the period between the first exposure to a harmful stimulus and detection. Rothman, supra note 67, at 255.

89. This is not the only example of such a disease. The symptoms of mercury poisoning, sometimes known as "Minamata disease" after the Japanese fishing village whose fish had been poisoned by mercury during the 1950s and 1960s, include numbness, tunnel vision, impairment of hearing, speech, balance, coordination, and memory. See SHKILNYK, supra note 63, at 185. These symptoms of "acute" mercury poisoning have been known to appear as late as seven years following mercury exposure. See id. at 186 & n.9.

90. Richard Peto et al., Cancer and Ageing in Mice and Men, 32 BRIT. J. CANCER 411, 412 (1975); see also Richard Doll, The Age Distribution of Cancer: Implications for Models of Carcinogenesis, 134 J. ROYAL STAT. SOC'Y 133, 149 (1971); P. Armitage, Multistage Models of Carcinogenesis, 63 ENVTL. HEALTH PERSP. 195 (1985).

91. JOHN CAIRNS, CANCER; SCIENCE AND SOCIETY 15 (1978).

92. Sandra A. Norman et al., Cancer Incidence in a Group of Workers Potentially Exposed to Ethylene Oxide, 24 INT'L J. EPIDEMIOLOGY 276, 282 (1995); Ferris M. Hall, Screening Mammography—Problems on the Horizon, 314 NEW ENG. J. MED. 53, 54 (1986).

93. For a sobering assessment of the current state of cancer treatment, see John C. Bailar III & Heather L. Gornik, Cancer Undefeated, 336 NEW ENG. J. MED. 1569 (1997).

94. Jerry B. Buchanan et al., Tumor Growth, Doubling Times, and the Inability of the Radiologist to Diagnose Certain Cancers, 21 RADIOLOGIC CLINICS N. AM. 115, 119-20 (1983). Certainly, at least, removing cancerous growth becomes harder—"more difficult, more costly, more disfiguring, and less effective"—the further the cancer has progressed. Id. at 120.

95. A rare example of a type of cancer detectable at an extremely early stage is cancer of the placenta, which leads to the production of the hormone chorionic gonadotrophin, which can be detected in the blood. CAIRNS, supra note 91, at 161.

96. BARRY I. CASTLEMAN, ASBESTOS: MEDICAL AND LEGAL ASPECTS 39 (4th ed. 1996).

97. This also helps to explain another puzzle in cancer etiology, which is why so many cancers occur in relatively old age. There appears to have developed something of a consensus that duration of exposure, rather than age per se, explains this demographic pattern; duration of exposure is highly correlated with age. For an early article making this point, see Peto et al., supra note 90.

98. David F. Goldsmith, Calculating Cancer Latency Using Data From a Nested Case-Control Study of Prostatic Cancer, 40 J. CHRONIC DISEASES 119S, 122S (1987); Harvey Checkoway et al., Latency Analysis in Occupational Epidemiology, 45 ARCHIVES ENVTL. HEALTH 95, 96 (1990).

99. OSHA, Occupational Exposure to Butadiene 1, 3, 55 Fed. Reg. 32736 (proposed Aug. 10, 1990) [hereinafter OSHA, Butadiene].

100. The latency periods for different kinds of cancers vary for other reasons as well. First, latency periods vary according to type of cancer. Although the latency period for most cancers appears to hover in the 15- to 30-year range, see Richard B. Hayes & Paolo Vineis, Time Dependency in Human Cancer, 75 TUMORI 189, 189 (1989), there are important exceptions. Clinically detectable leukemia, for example, has appeared in persons exposed to benzene within a median of three years following exposure. OSHA, Butadiene, supra note 99. Mesothelioma, on the other hand, follows exposure to the causative agent—mostly asbestos—by, on average, 35 to 40 years. Irving J. Selikoff et al., Latency of Asbestos Disease Among Insulation Workers in the United States and Canada, 46 CANCER 2736, 2740 (1980); B.T. Westerfield, Asbestos-Related Lung Disease, 85 S. MED. J. 616, 619 (1992). The latency period of cancer can also vary according to the causative agent implicated in the development of the disease. Hayes & Vineis, supra, at 189. The median latency period for radiation-induced leukemia, for example, is different from the median latency period for leukemia brought on by benzene exposure. OSHA, Butadiene, supra note 99. Latency periods have even been found to vary among different epidemiological studies of the same type of exposure and the same type of cancer, suggesting that some of the variability in latency periods for cancer is the result simply of differences in study design.

101. See U.S. DEPARTMENT OF HEALTH & HUMAN SERVS., THE HEALTH BENEFITS OF SMOKING CESSATION: A REPORT OF THE SURGEON GENERAL 135 (1990).

102. See HARRISON'S PRINCIPLES OF INTERNAL MEDICINE 909 (Eugene Braunwald et al. eds., 15th ed. 2001).

103. See id. at 912.

104. Selikoff et al., supra note 100, at 2740.

105. Haroutune K. Armenian & Abraham M. Lilienfeld, The Distribution of Incubation Periods of Neoplastic Diseases, 99 AM. J. EPIDEMIOLOGY 92, 98 (1974); S. Cobb et al., On the Estimation of the Incubation Period in Malignant Disease: The Brief Exposure Case, Leukemia, 9 J. CHRONIC DISEASES 385 (1959).

106. Norman et al., supra note 92, at 283.

107. Selikoff et al., supra note 100, at 2736.

108. Checkoway et al., supra note 98, at 95. The length of the "true" latency period, on the other hand—that is, the period between disease initiation and clinical detection—is dependent on nothing more than the present state of advancement in methods of detection and treatment. If these methods advanced far enough, the true latency period could in theory be reduced almost to zero. Rothman, supra note 67, at 254. True latency—the period in which a disease is present, but undetectable—is, in other words, a function more of the state of medical science than of biology. True latency tells us nothing about the etiology of human disease.

109. See, e.g., Rothman, supra note 67, at 254.

110. For general discussion, see CAIRNS, supra note 91, at 147-52.

111. Even so, the determination of the latency period for cancer is fraught with uncertainty. For example, the choice between two-stage and multistage models of carcinogenesis has been found to influence the length of the latency period for cancer. Duncan C. Thomas, Statistical Methods for Analyzing Effects of Temporal Patterns of Exposure on Cancer Risks, 9 SCANDINAVIAN J. WORK ENV'T HEALTH 353, 361-62 (1983). The length of the cancer latency period is also greatly affected by whether the substance in question acts in the early stages of carcinogenesis (or acts as what is often called an "initiator"), or in the late stages (or acts as a "promoter"). Anthony P. Polednak, Bone Cancer Among Female Radium Dial Workers: Latency Periods and Incidence Rates by Time After Exposure, 60 J. NAT'L CANCER INST. 77, 81 (1978); Checkoway et al., supra note 98, at 99. The choice between two-stage and multistage models of carcinogenesis and the identification of particular substances as initiators or promoters are among the most hotly debated issues in cancer research today. Insofar as it depends on an assumption about the mechanism of carcinogenesis, the estimation of latency periods is filled with uncertainty. For overviews of this complex subject, see CAIRNS, supra note 91, at 91-97; Polednak, supra, at 354; A. Sivak, Cocarcinogenesis, 560 BIOCHIMICA ET BIOPHYSICA ACTA 67 (1979). Seminal articles on the subject of two-stage and multistage models of cancer include P. Armitage & R. Doll, A Two-Stage Theory of Carcinogenesis in Relation to the Age Distribution of Human Cancer, 11 BRIT. J. CANCER 11 (1957); Doll, supra note 90, at 133; Suresh H. Moolgavkar & David J. Venzon, Two-Event Models for Carcinogenesis: Incidence Curves for Childhood and Adult Tumors, 47 MATHEMATICAL BIOSCIENCES 55 (1979). It is also important to realize that even the latency periods that have become standard referents in the literature—such as the 35- to 40-year latency period for mesothelioma due to asbestos exposure—represent average or median values for latency; many people will contract the relevant disease from the relevant cause within a shorter, and some within a longer, period. Peto has cautioned, for example, that the "mean latency" of a given cancer is "strongly dependent on the pattern of deaths from other causes" and thus should be used "extremely cautiously, if at all, in describing the results of an experiment." Richard Peto, Guidelines on the Analysis of Tumour Rates and Death Rates in Experimental Animals, 29 BRIT. J. CANCER 101 (1974). Likewise, latency periods observed in epidemiological studies heavily depend on study subjects' rates of survival from competing causes of mortality. Goldsmith, supra note 98, at 121S.

All this suggests that the determination of the latency period for any given disease is a highly fact-specific, research-intensive, variable, and uncertain enterprise—thus rendering efforts to specify the exact temporal location of the benefits of life-saving regulatory measures, based on latency, an unwieldy and uncertain endeavor. Thus, although it has become commonplace in the legal literature to refer to "the long latency period" of cancer, this blanket formulation masks both the variability and uncertainty associated with the derivation of latency periods for this disease. The latency period for some cancers, such as leukemia, is quite short; even cancers exhibiting longer latency periods may, in certain individuals, strike sooner than the average or median latency period would predict. Thus, unqualified references to "the latency period" for a particular disease often oversimplify what is in fact an extremely complicated matter.

112. Use of Cost-Benefit Analysis by Regulatory Agencies: Joint Hearings Before the Subcomm. on Oversight and Investigations and the Subcomm. on Consumer Protection and Finance of the House Comm. on Interstate and Foreign Commerce, 96th Cong. 82 (1979) (statement of Nicholas A. Ashford) ("The 'benefit' of removing a person now from risk of future damage, which is irreversible, inevitable, and non-arrestable once the risk exposure occurs, can be considered to be a present benefit—and quantified, for example, as the benefit of removing those presently at risk from future harm.").

113. CANCER: PRINCIPLES AND PRACTICE OF ONCOLOGY 852 (Vincent T. DeVita et al. eds., 4th ed. 1993) ("Cancer of the pancreas is a highly metastatic disease. Most patients present with disease advanced beyond the scope of potentially curative treatment.").

114. Id. at 160.

115. This issue arose rather obliquely in the recent rulemaking regarding arsenic in drinking water, in which EPA noted that arsenic likely acts as a promoter of bladder cancer—which means that the latency period for arsenic-related bladder cancer may be relatively short. See U.S. EPA, ARSENIC IN DRINKING WATER RULE ECONOMIC ANALYSIS, supra note 4, at 5-28.

116. See, e.g., U.S. EPA, Air Contaminants, 57 Fed. Reg. 26002 (to be codified at 29 C.F.R. pts. 1910, 1915, 1917, 1918, 1926, 1928) (proposed June 12, 1992) (discussing cancer).

117. The exception is harm to a fetus accomplished through simultaneous maternal and in utero exposure to hazardous substances; here, the temporal interval between exposure and harm still occurs within a single lifetime (the baby's), but the harmful event involves two generations.

118. A third way in which regulations may benefit multiple generations is by reducing exposures to substances that cause genetic mutations that appear in succeeding generations. This effect is probably quite limited. In order to affect succeeding generations, a mutation must be of cells in the germ line; that is, it must affect the egg, or sperm, or their precursors. See, e.g., CAIRNS, supra note 91, at 97. Such mutations are quite rare; radiation is one of the few harmful stimuli known to produce such mutations. See id. at 97-98.

119. United Nations Environment Programme, Note by the Secretariat. Preparation of an International Legally Binding Instrument for Implementing International Action on Certain Persistent Organic Pollutants (1998).

120. See U.S. EPA, Environmental Standards for Uranium and Thorium Mill Tailings at Licensed Commercial Processing Sites, 48 Fed. Reg. 45926, 45927-28 (Oct. 7, 1983).

121. See Dan M. Berkovitz, Pariahs and Prophets: Nuclear Energy, Global Warming, and Intergenerational Justice, 17 COLUM. J. ENVTL. L. 245, 256 (1992).

122. See U.S. EPA, Environmental Standards for the Management and Disposal of Spent Nuclear Fuel, High-Level and Transuranic Radioactive Wastes, 47 Fed. Reg. 58196, 58199 (Dec. 28, 1982). This requirement has led the U.S. Department of Energy to "spend several million dollars designing a 'keep out' sign … that would be effective for 10,000 years and recognizable by any future earthling." Gerrard, supra note 47, at 1133. Regarding the perils of efforts to predict human events this far into the future, Erickson has written:

The most mature and accurate scientific report we can issue, it seems to me, would conclude: We do not know, we cannot know, and we dare not act as though we do know. To speak thus is not to introduce a note of drama into what can otherwise seem a dry and technical business but to assess the matter in as rational and unemotional a way as language permits. Things will change drastically over the next few hundred years—never mind the next few thousand—and in ways that we cannot, by definition, foresee. To think otherwise is unrealistic, unscientific, and more than a little crazy.

ERICKSON, A NEW SPECIES OF TROUBLE, supra note 21, at 224.

123. See persistent organic pollutants (POPs) background document located at http://irptc.unep.ch (last modified July 16, 2001).

124. See id. §§ 6.2 & 6.5.

125. See id. § 6.3.

126. See, e.g., SHKILNYK, supra note 63, at 184-85 (describing persistence of mercury contamination).

127. U.S. government officials involved in the negotiations regarding the treaty to ban or restrict POPs, for example, have concluded that "there are no scientific 'bright lines' for determining POPs screening or listing criteria." See Bruce D. Rodan et al., International Action on Persistent Organic Pollutants (POPS); Developing Science-Based Screening Criteria 2 (unpublished and undated manuscript; copy on file with author).

128. Cf. Shinsuke Tanabe, PCB Problems in the Future: Foresight From Current Knowledge, 50 ENVTL. POLLUTION 5, 10 (1988) (arguing that threats from PCBs may increase in the future because most of the PCBs in existence today are still in use, and have not yet escaped into the environment).

129. A dramatic example of the spatial effects of persistence comes from researchers' efforts to find a "control" group to compare to populations exposed to persistent chemicals; because of the remote and pristine nature of the high Arctic, they hoped to find such a group in the indigenous peoples living there. Their hope proved forlorn: breastfeeding babies in the high Arctic take in seven times more PCBs than average babies in the United States and Canada. See COLBORN ET AL., supra note 72, at 107.

130. See supra note 123, at § 1.

131. Id. § 6.10.

132. See Joseph L. Jacobson & Sandra W. Jacobson, Intellectual Impairment in Children Exposed to Polychlorinated Biphenyls in Utero, 335 NEW ENG. J. MED. 783, 786 (1996)

133. COLBORN ET AL., supra note 72, at 106-07; see also Linda S. Birnbaum, The Mechanism of Dioxin Toxicity: Relationship to Risk Assessment, 102 ENVTL. HEALTH PERSP. 157, 163 (1994) (supp. 9) (daily exposure of nursing infants to dioxin and related chemicals may be 10 to 20 times greater than exposure of general population).

134. The same study cited research showing "subtle abnormalities" due to contaminated milk in 10% of newborns in the Netherlands. See J.G. Koppe, Nutrition and Breast-Feeding, 61 EUR. J. OBSTETRICS GYNECOLOGY REPROD. BIOLOGY 73 (July 1995).

135. See Barry Commoner, Failure of the Environmental Effort, 18 ELR 10195(June 1988).

136. For general discussion, see, e.g., Michael L. Katz & Carl Shapiro, Network Externalities, Competition, and Compatibility, 75 AM. ECON. REV. 424 (1985); Michael Klausner, Corporations, Corporate Law, and Networks of Contracts, 81 VA. L. REV. 757 (1995); Mark A. Lemley & David McGowan, Legal Implications of Network Economic Effects, 86 CAL. L. REV. 479 (1998); S.J. Liebowitz & Stephen E. Margolis, Network Externality: An Uncommon Tragedy, 8 J. ECON. PERSP. 133 (1994).

137. Robert W. Hahn, Regulatory Reform: What Do the Government's Numbers Tell Us?, in RISKS, COSTS, AND LIVES SAVED: GETTING BETTER RESULTS FROM REGULATION 208, 226 (Robert W. Hahn ed., 1996).

138. See OFFICE OF TECH. ASSESSMENT, U.S. CONGRESS, GAUGING CONTROL TECHNOLOGY AND REGULATORY IMPACTS IN OCCUPATIONAL SAFETY AND HEALTH — AN APPRAISAL OF OSHA's ANALYTIC APPROACH, OTA-ENV-635 (1995) (finding, in retrospective studies of actual costs of complying with rules of the Occupational Safety and Health Administration, that actual costs were significantly lower than predicted costs because of technological innovation); Michael E. Porter & Claas van der Linde, Green and Competitive: Ending the Stalemate, HARV. BUS. REV., Sept./Oct. 1995, at 121 (finding cost savings due to regulation); Dallas Burtraw et al., The Costs and Benefits of Reducing Acid Rain (Resources for the Future Discussion Paper No. 97-31-REV, 1997) (finding that actual costs of federal acid rain program have been far lower than predicted costs); Richard D. Morgenstern et al., The Cost of Environmental Protection (Resources for the Future Discussion Paper No. 98-36, 1998) (finding cost savings due to regulation).

139. For general discussion, see, e.g., William Samuelson & Richard Zeckhauser, Status Quo Bias in Decision Making, 1 J. RISK & UNCERTAINTY 7 (1988).

140. For a basic explanation of discounting, see WILLIAM J. BAUMOL & ALAN S. BLINDER, ECONOMICS: PRINCIPLES AND POLICY 386-87 (8th ed. 2001).

141. OMB has explicitly stated that agencies should discount life-saving benefits by considering the latency period associated with the diseases being prevented by regulation. REPORT OF INTERAGENCY GROUP CHAIRED BY A MEMBER OF THE COUNCIL OF ECONOMIC ADVISORS, ECONOMIC ANALYSIS OF FEDERAL REGULATIONS UNDER EXECUTIVE ORDER 12866 (1996).

142. For general discussion of these diseases, also known as transmissible spongiform encephalopathies, see, e.g., J.D. Fratkin & A.G. Smith, Slow Virus Infections, SURV. OPHTHALMOLOGY, Jan./Feb. 1977, at 356; E.H. Lennette, Classic Slow Virus Diseases, 11 BULL. PAN AM. HEALTH ORG. 157 (1977).

143. See Heinzerling, Regulatory Costs of Mythic Proportions, supra note 5, at 2054.

144. See, e.g., VISCUSI, RATIONAL RISK POLICY, supra note 19, at 45.

145. See id. at 46. The same result can be achieved by multiplying the probability of harm by the number of people necessary to produce one death, given the probability of harm. See id.

146. See, e.g., MENDELOFF, supra note 55, at 48; DEREK PARFIT, REASONS AND PERSONS 480 (1984); Daniel A. Farber & Paul A. Hemmersbaugh, The Shadow of the Future: Discount Rates, Later Generations, and the Environment, 46 VAND. L. REV. 267, 286 (1993).

147. See, e.g., Farber & Hemmersbaugh, supra note 146, at 280, 287.

148. In addition, the empirical evidence on people's actual preferences with respect to remote harms is quite inconclusive. See Heinzerling, Regulatory Costs of Mythic Proportions, supra note 5, at 2046-48.

149. W. Kip Viscusi, Equivalent Frames of Reference for Judging Risk Regulation Policies, 3 N.Y.U. ENVTL. L.J. 431, 436-37 (1994).

150. See id. at 437 n.16.

151. See, e.g., OMB, REGULATORY PROGRAM OF THE UNITED STATES GOVERNMENT, APRIL 1, 1990-MARCH 31, 1991, at 40 (1991).

152. Cf. W. KIP VISCUSI, FATAL TRADEOFFS: PUBLIC AND PRIVATE RESPONSIBILITIES FOR RISK 31 (1992) ("Although in many cases we can simply let future generations spend more later, for irreversible effects on the environment or policies with long-term effects, decisions must be made now.").