Risk and Trust: The Role of Regulatory Agencies

Thomas O. McGarity

Thomas O. McGarity is Professor of Law at the University of Texas Law School, Austin. Texas.

Life is a risky business. In this complex technological society, we simply cannot go about our daily activities without exposing ourselves and others to risks to our health and to our shared environment. Risks lurk in our diets, in our jobs, and in our recreational activities. Fortunately, we can reduce risks by amending our conduct and by implementing risk-reduction technologies. These efforts, however, claim resources, in both time and money. Individually and collectively, we face hard choices about how much risk is acceptable and how much risk should be avoided.

William Ruckelshaus, former Administrator of the Environmental Protection Agency (EPA), has suggested that we bifurcate our examination of scientific and technological risk into two separate inquiries: risk assessment and risk management.¹ A recent publication of the National Academy of Sciences (NAS) also draws this distinction.² Risk assessment has to do with the nature and degree of risk that an activity poses to humans and the environment; risk management has to do with how an individual or society deals with risks after they have been assessed. While I do not agree with all the implications that Mr. Ruckelshaus draws from this distinction, I think that the division is a useful one for purposes of analysis.

Assessing Risks

Past societal attempts to deal with risks have bred a budding risk assessment profession. Professional risk assessors gather data, conduct studies on surrogate systems, and develop mathematical representations of reality (models). They then apply these models to the available information to derive quantitative or qualitative estimates of the risks that various activities pose to humans and the environment. The professional risk assessor usually expresses his or her assessments as statistics or ratios that an agency, company, or individual can use when deciding how to manage those risks.

Most risk assessments, however, are not undertaken by professional risk assessors. The vast bulk of risk assessments are performed by ordinary individuals on an ad hoc basis as they go about their daily lives. When professional risk assessments are available, individuals may rely on them. Yet, because professional risk assessments are not available for most risks (and probably never will be), we do the best we can with the information we have and with our own meager analytical capabilities.

Most lay people assess risks in an unsophisticated and impressionistic fashion. We do not think of risks as statistical combinations of probabilities and consequences. Indeed, most of us would probably prefer not to think about risks at all. We tend to view risks in terms of "either-or." Either an activity is safe, or it is dangerous. Either a chemical is a poison, or it is not. Either a chemical causes cancer, or it does not. We want stark risk assessments that tell us clearly that we should avoid (or prohibit) an activity, or that we may proceed without worry.

Lay people do not have a very keen comparative perspective on risk. When asked how the risks of smoking a cigarette compared to the risks of climbing a mountain, most of us react viscerally, basing our opinions largely upon our evaluations of the comparable worth of the two enterprises, without any resort to statistics or analysis. We are inclined, however, to distinguish sharply between the risks that we impose upon ourselves and the risks that others impose upon us. It is perfectly plausible, therefore, for an avid amateur motorcar racer to be a vigorous opponent of nuclear power.

Attitudes play a very large role in individual risk assessments. With little information about risks beyond ordinary experience, and with even less time and inclination to analyze the information that is available, most of us manage risks on the basis of fleeting impressions, biases, and prejudices. Individuals tend to underestimate the risks of activities that yield high benefits to them and to overestimate the risks of activities that yield no benefits. I suspect that most individuals also tend to underestimate the risks that they impose on others. This is all to suggest that individual risk assessment is not an especially rational or objective undertaking.

Although professional risk assessments can be a good deal more rational and objective than lay assessments, even they have important limitations. A professional risk assessor rarely has enough information to produce a confident qualitative assessment of the risks that an enterprise poses to humans and to the environment during its normal, planned operations. Some "trans-scientific" questions, such as the shape of the carcinogenesis dose-response curve at low exposure levels, may never be answerable in a definitive way.

In addition to judging the quality of existing information, professional risk assessors must draw inferences from available data (e.g., that a chemical appears to cause cancer in laboratory animals) in order to answer questions where data are poor or lacking (e.g., whether the chemical causes cancer in humans). The models used by professional risk assessors to make quantitative predictions are filled with highly debatable assumptions about the state of nature. The recent NAS Report on risk assessment, for example, contains a listing of such assumptions that goes on for more than three pages.³

If this were not bad enough, professional risk assessments these days must also deal with "abnormal" risks of human error and institutional negligence of the sort that characterizes the accidents at Three Mile Island and Bhopal. Even more difficult to assess are the risks of intentional malfeasance, such as sabotage or the Tylenol poisonings of a few years ago. The professional risk assessor can drive himself to distraction attempting to attach some quantitative dimension [16 ELR 10199] to these kinds of risks. The bottom line for most risk assessments is that they are influenced heavily by the risk assessor's "professional judgment."

While "professional judgment" may reflect a purely scientific belief about how the universe operates, it often represents no more than the professional risk assessor's beliefs about the best regulatory policy to apply. In any event, when the scientists who provide the information and judgments that form the underlying basis for risk assessments disagree on a fact, inference, or assumption, the risk assessor or, more appropriately, the policymaker, must resolve the disagreement in an outcome-oriented fashion, based on his or her view of the most appropriate regulatory policy.⁴ The recent NAS Report, for example, characterizes the policy component of risk assessment as "risk assessment policy."⁵ The suggestion that risk assessment requires policy input raises the question of what policy, or, more importantly, whose policy should apply.

Managing Risks

Individual Risk Management

Just as the vast bulk of risk assessment is performed by lay persons, most risks are managed by individuals as they go about their daily business. Once aware of the nature and magnitude of a risk, the individual must decide how that risk will affect his or her conduct. Sometimes a change in conduct (e.g., a change in diet) will reduce individual risks. Sometimes it will reduce risks to both the individual and others (driving carefully), and sometimes it will reduce risks to the individual while raising risks to others (packing a .357 magnum). Sometimes a change in conduct will reduce risks to others while leaving the individualunaffected (screening susceptible workers out of the workplace), and sometimes an extraordinary individual will increase risks to him- or herself to reduce risks to others (a daring rescue of a child from a burning building).

An individual risk management decision may be based upon one or several risk assessments. Most are based upon lay assessments by the individual risk manager. Occasionally, the individual may have access to a professional risk assessor through a newspaper clipping, an owner's manual, a product label, or a book on health or safety. In addition, there are a few professional risk assessors (for instance, Consumers Union and personal physicians) who make their services available to individual risk managers for a fee. Not surprisingly, professionals who serve individual risk managers go to great lengths to inspire public confidence in their assessments. The perception of objectivity is a very valuable asset to these professionals. If, for example, Jack Anderson revealed that articles in Consumer Reports were being written by the manufacturers of the products evaluated, or even that the articles relied heavily upon data or models supplied by manufacturers, the subscription list would no doubt decline dramatically.

Collective Risk Management

While individuals can assess and manage most day-to-day risks, we also depend very heavily upon collective risk managers. Collective risk managers are, for the most part, institutions established by society to assess and manage risks for all of us. In this society, three institutions — common law courts, legislative bodies, and regulatory agencies — are primarily responsible for collective risk management. Professors Stewart and Green will have more to say about courts and legislatures. I will focus on regulatory agencies. Before moving to a discussion of agencies, however, it is important to analyze why we have given these institutions the responsibility for assessing and managing risks. Why is individual risk assessment and management not enough?

First, individuals simply do not have the time or the expertise to assess and manage most risks that they encounter in their daily lives. A consumer cannot inspect every box of cake mix for possible EDB contamination. Assuming the label told the consumer that the mix contained one part-per-billion of EDB, the consumer would still know little about the risks of eating it. Even if the consumer could read in Consumer Reports that a person eating cake mix containing one part-per-billion of EDB faced a one-in-one-million chance of contracting cancer over the next seventy years, he or she could not make a sophisticated decision about whether the cheaper EDB-laced cake mix was worth the added risk of contracting cancer. Indeed, for many relatively minor risks, it is not clear that consumers even want to have an opportunity to assess and manage risks. Many, if not most, consumers would prefer to delegate those unattractive tasks to some trustworthy public institution.

Second, the individual who is exposed to risk is often powerless to manage that risk. Even if an individual has access to a professional risk assessment, the risks may be attributable to the conduct of others over whom the individual has no control. The individual can, of course, manage risks to himself to the extent that he can affect the conduct of others. One obvious way to affect the conduct of another is to threatenan even greater risk if he persists. While this primitive form of individual risk management is not unheard of, we generally prefer a more civilized form of risk management.

One civilized way to influence the conduct of another is to enter into a contract with him. This solution, however, is unattractive for several reasons. First, society is simply too complicated to allow the thousands of contractual arrangements that would be needed each day to enable one to proceed with one's business without encountering unacceptable externally imposed risks.⁶ Moreover, the contract solution allows the wealthy to impose risks upon the poor with impunity, because the poor cannot afford to contract for risk reduction. While some disparity in the allocation of risks in society is probably as inevitable as disparties in wealth (the wealthy, after all, can better afford water purifiers, pesticide-free food, and Mercedes with airbags), there comes a point at which society is unwilling to allow the powerful to profit from the harm they impose upon others and upon the environment. There is probably some room for private risk management through contract, as, for example, the labor [16 ELR 10200] contract between a circus performer and her employer, but we would expect this to be an acceptable solution for only a small portion of externally imposed risks.

For reasons of fairness, efficiency, and practicality, we are forced to conclude, therefore, that we cannot rely upon our own efforts to assess and manage risks. If we are to achieve an acceptable degree of risk in society, we must create institutions capable of collectively assessing and managing risks.

The courts are viable institutions for assessing and managing many risks, and I will largely defer to Professor Stewart's discussion of this question. I shall pause only long enough to suggest some reasons why the courts cannot assume complete responsibility for the task.

First, although it is possible to secure injunctive relief against threatened harm, courts generally act only after some harm has already occurred. Second, a large portion of the risks that are in need of collective management are chronic risks which, when they manifest themselves in human or environmental harm, do not leave footprints. It is very difficult in an after-the-fact judicial proceeding to establish to a court's satisfaction a causal connection between the conduct that posed the risk and the manifestation of that risk in the plaintiff's individual harm. Third, judges and juries are no better risk assessors than other lay persons. A trial affords the parties an opportunity to educate the judge and jury on the complexities of risk assessment, but these adversarial efforts are likely to leave the judge and jury even more confused than when they entered the courtroom. Fourth, the judiciary was not designed to be a manager of much of anything. We should therefore not expect courts to be effective risk managers.

Finally, the courts are generally unaccountable to the public. If we are to place our trust in a collective institution to assess and manage risks, we want that institution to be accountable. In the federal system, judges are appointed for life. Although courts can be reversed through the enactment of statutes, they are not particularly accountable institutions. We have already seen how both risk assessment and risk management are policy-laden enterprises. When we ask the courts to undertake collective risk assessment and risk management, we are asking them to be policymakers and policy implementers. In this democratic society, we usually insist that policymakers and policy implementers remain accountable to the public.

Professor Green will address the subject of Congress as collective risk assessor and manager. I will therefore observe only that Congress also has little expertise in risk assessment. While Congress is very much accountable to the public, it cannot be an effective risk manager. As an institution, it is simply too busy to address the details of the complex patterns of conduct and technologies that must be regulated if risks are to be managed effectively. When Congress addresses risks, it does so in very unsophisticated ways; witness, for example, the Delaney Clause and the Saccharin Amendments to that Clause. While sophistication is not always a virtue — especially in areas where uncertainties preclude sophisticated collective responses to social problems — we can expect a collective institution to parse issues with sharper knives than those available to Congress. I also suspect that by and large Congress would in many cases be unwilling to make the difficult judgments required by public risk management policy.

Regulatory agencies are capable of acquiring expertise in risk assessment, and they can afford to focus attention on risk management problems. Unlike the courts, agencies can address risks before they have caused any harm and without the necessity of drawing clear causal connections between general risks and particular manifestations of harm. Like Congress, agencies are accountable to the public, although it is fair to say that bureaucracies are not as accountable as legislatures. Independent agencies are almost as unaccountable as courts. While regulatory agencies may be an imperfect institution for collective risk assessment and management, they may be the best institutional alternative that we have.

Risk Regulation and Public Trust

To a large extent, the entire collective risk assessment and risk management enterprise rests upon public trust in the institution or institutions charged with that enterprise. We individuals may be unwilling to rely upon our own lay risk assessments to manage many of the risks of modern society, but we cannot trust those who would subject us to risk to do so for our benefit. On the other hand, we still want the freedom to impose some risks upon ourselves and others. And we are willing to have others impose risks upon us so that society can enjoy the benefits of their risk-producing enterprises. We do not, in other words, demand perfect safety (nor could we in a complex society). We simply want a trustworthy institution to determine the nature and extent of the risks we face and to ensure that we are not exposed to unacceptable risks.

Regulatory agencies are prime candidates for undertaking the tasksof collective risk assessment and risk management, provided they are truly agents for those of us whom the statutes are designed to protect. If the duty of regulatory agencies is to protect the public from unacceptable risks, the relationship must be one of trust. In addition, since both risk assessment and risk management must be informed by public policy, we all have a strong interest in the content of the policies and in the effectiveness with which agencies implement those policies.

It is fair to say, however, that in recent years the regulatory agencies have lost a large measure of the public trust that they enjoyed during the 1970s. During the late 1970s, the public learned that the information base forming the foundation for many previous regulatory decisions was seriously flawed. For example, three officials of a company that performed animal testing for several pesticide and drug firms were convicted in 1983 for falsifying test results.⁷ The investigation of the Industrial Biotest Laboratories (IBT) undermined the confidence that agency scientists and the public could place in that data. Nevertheless, the drugs and pesticides remained on the market. Although EPA became aware of the IBT situation in the late 1970s, as of 1983 it had assigned only one full-time professional to audit the performance of all pesticide testing laboratories for evidence of additional fraud.⁸ In addition, it approved emergency exemptions from the normal pesticide registration process for fifty pesticides whose registrations were supported by tests conducted by IBT.⁹

In addition, during the first part of the Reagan Administration, budget cuts severely weakened the capacity of most [16 ELR 10201] agencies to establish an effective data base for risk assessments where the regulated industry was not required to provide the data base. Lacking adequate data on the health and environmental effects of products and technologies, the agencies could not manage risks, because they could not begin to assess them. Furthermore, the new EPA Administrator moved several career employees out of the risk assessment and risk management process.¹⁰

Numerous reports of closed-door meetings between agency officials and representatives of regulated industries to discuss regulatory strategies also weakened public trust in EPA's risk management capacities. EPA also initiated a series of "decision conferences" in which EPA and industry technical staff were invited to resolve technical questions concerning the registration of pesticides outside of the public view;¹¹ industry even went so far as to draft documents to be published by the agency in the Federal Register. Such behind-the-scenes interaction between the regulatory agency and the regulated industry severely strained society's trust in the agency. These activities, combined with numerous "sweet-heart deals" between regulated industries and agency enforcement staff¹² and private promises not to enforce existing regulations,¹³ made the EPA appear to be an agent of the regulated interests rather than of the general public.¹⁴

Less dramatic, but perhaps more disturbing in the long run, is the prominent role that the Office of Management and Budget (OMB) has come to play in risk assessment and risk management. That agency's intervention into EPA's approach to asbestos regulation is the latest in a long line of policy management decisions that lead inevitably to the conclusion that OMB has become the government's most important policy management institution and risk manager for health and environmental regulation.¹⁵ While OMB certainly does not win all of the big battles with the agencies, it wins enough to retain an important influence over all important risk management decisions of the federal agencies.

To a large degree, the desk officers and policy analysts in OMB who make these essential risk management decisions are unaccountable to the public. Very little of OMB's input is in writing. Most of its management is conducted over the telephone prior to the time that the proposed agency actions become public. Rarely are the Director of OMB or the President called to account for off-the-record policy management decisions. For example, while the discount rate that OMB requires agencies to use in calculating the benefits of their actions is clearly a policy question about which expertise has almost nothing to say, the rate is also a matter of intense public interest when used in the cost-benefit analyses upon which agency decisions may rest. Until very recently, however, there has been no public debate about the high ten percent discount rate upon which OMB insists.¹⁶

OMB often becomes heavily involved in the details of agency risk assessments, although its technical staff includes no scientists or engineers with expertise in these particular matters.¹⁷ Finally, it has become increasingly clear that in many cases OMB review merely serves as an ex parte opportunity for the regulated industries to have policy and factual input into proposed agency decisions, before they are published in the Federal Register.¹⁸ Such off-the-record risk assessment and policymaking, while certainly not unusual in Washington, D.C., cannot inspire trust in the ultimate output.

Restoring and Maintaining Public Trust in Regulatory Agencies

We may now be at a low point of public trust in federal agencies as assessors and managers of health and environmental risks. Indeed, until recently, the level of public trust was so low that the legitimacy of the entire enterprise was in question. Not surprisingly, individuals and groups who had previously placed their trust in agencies began to seek out other institutional solutions. The massive increase in "toxic tort" cases in state and federal courts indicates a possible return to the courts. The enactment of numerous community "right-to-know" statutes, heavy sales of health and diet books, and the success of "whole foods" stores suggest that individuals, no longer trusting the professional risk assessors in the agencies, may be returning to individual risk assessment and management.

While the return to some individual responsibility for health and safety management is probably not a bad thing, it cannot be the ultimate solution to the current legitimacy crisis in the regulatory agencies. Once the "easy" toxic tort cases — those in which the causal connection between risk and harm is relatively clear — have been litigated, the rest will fade away, unless tort doctrine changes significantly. While individuals should be encouraged to spend more time thinking about their own safety, they still lack the expertise to make sophisticated risk assessment and management decisions. Moreover, they are still powerless to make others modify their conduct toreduce risks, short of ugly confrontations, which, by the way, we are witnessing more and more [16 ELR 10202] often these days. Finally, to insist that individuals manage their own risks when they have so little power to do so invites cynicism and disillusionment.

I believe that we should not lose all faith in regulatory agencies as assessors and managers of risks in this society. At the same time, I am convinced that if they are to regain their legitimacy, the agencies must implement some changes. These changes need not represent drastic departures from the "interest representation" model that has dominated health and environmental decisionmaking since the early 1970s.

Indeed, many of the necessary changes have already been made. The blase attitude of many former agency heads toward their statutory duties was probably the single most important contributor to the loss of trust in the agencies during the early Reagan Administration. Just as high-level personnel during the Carter Administration were perceived as untrustworthy by those who stood to benefit from the products and services of the regulated industries, so too the extreme positions of the initial Reagan appointees were perceived as a breach of faith by the beneficiaries of health and environmental regulation. Had Mr. Ruckelshaus remained with EPA, the agency would no doubt have suffered crises of confidence, but these crises would likely not have reached the severity of the crisis of 1983. The general purge of the early Reagan-appointed leadership from the major health and environmental agencies during the years 1983-1985 will probably help restore public trust. Much damage has been done, however, and a full restoration will take time.

Is it necessary, then, to go further and purge the agencies of the scientists who were appointed at the beginning of the Reagan era? Should the agencies prepare a new "hit list" of the "wrong-thinking" scientists on the various advisory committees and replace them with scientists whose technical judgments more closely represent the new policy attitudes that are more sympathetic to the environment? These questions raise the broader and much-debated question of "adversarial science."

Every trial lawyer knows that there are scientists for hire. And there are scientists whose attitudes about scientific issues relevant to regulation are sufficiently crystallized that their expert evaluations are entirely predictable. An outcome-oriented agency can appoint enough of these scientists to its staff and advisory committees to ensure equally predictable regulatory results. It seems foolish, however, to make agency policy through periodic purges of agency staff and advisory committees.

The "policymaking-through-purge" approach cynically manipulates expertise to shield policymakers from responsibility for the implications of their policies. When an agency head stacks a scientific advisory committee with persons on one side of a scientific debate, he or she is simply basing a policy choice upon the predictable outcome of the group's deliberations. But this policy is made behind the scenes. The agency head appears to defer to the "experts," who are, after all, better trained to resolve these questions. In reality, however, the policy choice is made at the time the panel is chosen. The choice appears to outsiders to be a technical one, and the agency head thereby avoids political accountability.

Rather than purge agency staff and advisory committees, the agencies should attemptto inform themselves of all legitimate points of view in a scientific dispute. When the scientists disagree, the decisionmaker should explain why the agency used one assumption or adopted a particular inference, rather than another assumption or inference. Only in this way can the agency decisionmakers be made accountable for the policy judgments that necessarily inform regulatory decisionmaking in the health and environmental area.

Administrator Ruckelshaus has suggested that one solution to the problem of "adversary science" is to separate risk assessment from risk management. He writes:

[R]isk assessment at EPA must be based not only on scientific evidence and scientific consensus. Nothing will erode public confidence faster than the suspicion that policy considerations have been allowed to influence the assessment of risk.¹⁹

While in an ideal world, risk assessment would be an entirely scientific enterprise, divorced from "messy" policy considerations, the real world cannot accommodate such a clean distinction. Policy considerations pervade all health and environmental risk assessments and will continue to do so in the foreseeable future. If Administrator Rucklshaus had practiced what he preached, I venture to say that EPA would never have completed a single risk assessment, because the agency could not have achieved a "scientific consensus" on all of the crucial scientific components of that assessment. If the agency is not to remain paralyzed, it must do the best risk assessments it can with the best evidence available, whether or not the evidence is "scientific" in the sense of being acceptable for publication in a scientific journal, and whether or not it can command a scientific consensus.

When scientists fail to reach consensus, someone must fill the gaps with policy considerations. If the agency decision-makers do not, the experts will. This is demonstrated by a revealing passage in an article published by two prominent cancer biostatisticians in Cancer Research. After suggesting that the most appropriate dose-response curve for carcinogenesis is a "log-probit" curve with a slope of one, the authors explain that a slope of less than one would be inappropriate, because it would result in too many products being taken off the market.²⁰ This rationale is obviously not scientific; nor is it some special brand of "risk assessment policy," comprehensible only to professional risk assessors. It is raw, result-oriented, garden-variety policy of the sort that motivates Congressment to write statutes. To establish a separate office to prepare risk assessments based entirely upon expert judgment without any exterior policy guidance would simply delegate policymaking authority to agency experts.

Some industry groups and Congressmen have suggested that risk assessment for all of the federal government should be performed by some separate institutional entity composed of scientists who would be isolated from the pulls and tugs of policymaking.²¹ In my mind, this proposal is fundamentally misconceived. Far from separating the science from the policy, this proposal would embroil the independent risk assessment panel in public controversy. There would be, of course an enormous battle over whom to appoint to the panel. As policymakers or policies changed, one would expect periodic purges. The panel would, of necessity, become a policymaking insitution, but it would lack guidance as to what policy should guide the risk assessment.

The recent NAS Report observes that risk assessment has important policy components and recommends against the establishment of an independent government-wide risk [16 ELR 10203] assessment institution. The Committee, however, takes the position that the policy component of risk assessment is somehow different from ordinary public policy. Labeling this special policy component "risk assessment policy," the Committee suggests that this brand of policy has to do with how "conservative" the decisionmaker wants to be. Frankly, I have difficulty distinguishing the policy that determines the "conservativeness" of the model one uses from ordinary public policy.

Ultimately, the source of either policy must be the legislative process. If Congress tells the agency to err on the side of safety, agency risk assessors are not free to refrain from using conservative assumptions. Likewise, if Congress instructs the agency to emphasize the benefits of products or technologies, the agency risk assessors are not free to use conservative assumptions. Of course, Congress is rarely explicit about this question, but congressional will can often be discerned from other aspects of the agency's statute. When no policy is discernible at all — as, for example, with a pure risk-benefit balancing statute — there is probably room for administrative policymaking on risk assessment issues, but the policymaking will be no different than any other kind of administrative policymaking.

The NAS Report suggests further that uniform risk assessment "inference guidelines" should be developed for the entire federal government. The panel defined "inference guidelines" as "an explicit statement of a predetermined choice among alternative methods (inference options) that might be used to infer human risk from data that are not fully adequate or are not drawn directly from human experience."²² Inference guidelines can address issues ranging from the shape of the dose-response curve at low doses to whether benign and malignant tumors should be combined in evaluating laboratory animal studies. These guidelines would not be prepared by a separate panel of scientists; rather, they would be assembled by an interagency group, such as the now-defunct Interagency Regulatory Liaison Group.

This recommendation dovetails rather nicely with another idea of former Administrator Ruckelshaus; namely, that Congress should enact a common statutory formula for all risks. In a scheme combining the two ideas, Congress would enact a risk assessment statute empowering some entity to promulgate risk assessment guidelines in accordance with explicit congressional policy guidance.

This proposal has several attractive features. First, it recognizes that risk assessment has important policy components, and it allows our national policymaker (Congress) to establish that policy. The great risk assessment policy debate would play itself out before the public, not behind closed doors. All interested parties would have the opportunity to participate.

Second, the proposal would ensure uniformity in risk assessment within a given agency and throughout the federal government. An agency would not be able to use different sets of inferences, assumptions, and models for different chemicals. In addition, the federal government would not be able to use one set of inference guidelines to protect workers and another set to protect consumers. Since agency risk assessments would be more predictable, industry could use the uniform inference options to decide what chemicals to produce and how to evaluate them.

Third, uniform inference guidelines would enhance the efficiency of government decisionmaking by allowing agencies to limit chemical-by-chemical debate on issues relevant to the inference options determined by the uniform inference guidelines. While the guidelines would have to allow some flexibility for case-by-case evaluation, many of the most important and controversial issues would be debated and decided in the initial public proceeding concerning the broad government-wide inference guidelines.

The proposal also has some important disadvantages. First, explicit policy guidance from Congress may be difficult to arrange. Congress may not be anxious to come to an explicit conclusion about whether risk assessments should or should not "err on the side of safety" for all governmental programs. Such a policy statement would have important implications for many large government programs, ranging across several jealous committees in both houses. Moreover, the issue is by nature abstract and a bit boring. The bill could not be entitled the "Clean Air Act" or the "Bhopal Bill"; it would have to be the "How Do We Decide What is Dirty Bill?" Short of some galvanizing event, the nature of which I cannot even imagine, I am confident that such a bill would not make it to the floor of either house.

Second, even if such a bill were enacted, it would be subject to interpretation. It would be difficult to formulate the kind of phrase that would give risk assessors general guidance on how to draw inferences, make assumptions and choose models. Whatever language Congress uses would be subject to interpretation by the entity charged with drafting uniform inference guidelines. This fact suggests that the entiry should include a lawyer or judge with experience in interpreting statutes. Moreover, since interpreters of statutes are as much policymakers as statutory draftsmen, the composition of this body will be of immense importance to many interests. The selection process will be the subject of intense lobbying, and the legitimacy of the entity and its product will be open to question if it appears that its composition has been stacked.

Third, while consistency across products in a single regulatory program is certainly desirable (the political clout of a manufacturer should not determine the assumptions and inferences applicable to its product), uniformity across regulatory programs is not necessarily a good thing. We may want to apply more cautious assumptions in some regulatory programs than in others. For example, it may be desirable to use less cautious assumptions in a program for licensing new lifesaving drugs than in a program for permitting new hazardous waste facilities. To a limited extent, Congress can resolve this difficulty by articulating different risk management policies for the two programs, but this would be a very subtle undertaking, perhaps too subtle for Congress. The more likely result is a "lowest common denominator" risk assessment policy that confounds administrative attempts to apply varying risk management policies.

Fourth, it is not clear that establishing a separate entity to promulgate uniform risk assessment guidelines will enhance public understanding of the policy components of risk assessment and the distinction, if any, between risk assessment policy and risk management policy. If conducted in a proper fashion, the large "shoot-out" hearing over government-wide inference guidelines could be educational for the public, in much the same way that the lengthy hearings over the Occupational Safety and Health Administration's (OSHA's) generic carcinogen policy were educational. Whether the hearing will have an educational effect will depend largely on the composition of the entity and upon the nature of the hearings. If the entity is composed entirely of scientists of a single persuasion and if participation at the [16 ELR 10204] hearings is limited, the proceeding will have little educational value. The effect may be enhanced if wide-open public hearings are conducted before a panel containing representatives of many scientific points of view, representatives of disciplines outside the hard sciences (such as political scientists, lawyers, and economists), and representatives of the affected interest groups. While one purpose of the hearing should be to state a set of generally applicable principles to guide agencies in the future, two equally important purposes should be to acknowledge the uncertainties that hinder this effort and to state forthrightly the policy basis for the assumptions and inferences that must fill the factual gaps.

Finally, the entity that promulgates the inference guidelines should be an independent policymaking body, cut in the mold of the independent agencies. Only in this way can we be confident that the guidelines will reflect statutory policy, and not the ad hoc policy preferences of a particular Administration. The panel that makes risk assessment policy should not be a foil for the OMB; nor should it be a smokescreen for behind-the-scenes negotiations of the real power brokers.

In the final analysis, while the process of preparing uniform inference guidelines for the entire federal government might be a useful enterprise, I doubt very much if the NAS Panel's idea has much practical vitality. The divisions are still too great and the trust levels too low. The idea may have worked at the end of the 1970s as a natural culmination of several individual agency attempts to articulate their own carcinogen policies, but it has little chance of success in these more polarized times, even if for some reason it has worked its way to the top of Congress's agenda. Perhaps the best way to initiate the process of reconciliation and to restore turst is to ensure full participation in the scientific decisionmaking entities within the individual agencies.

By "full participation," I mean more than simply allowing everyone to submit comments in individual rulemaking proceedings; I mean that outsiders should participate in the actual decisionmaking process. Given that agency scientific advisory committees must make and apply policy, in addition to exercising scientific judgment, members of these committees should be chosen with an eye toward the interest groups that will be affected by their recommendations. For example, OSHA's scientific advisory committees should contain labor scientists as well as industry scientists; the agency should attempt to secure a balance of scientists from all interest groups. I call this idea "pluralistic scientific policymaking."

I would even go further and suggest that the agency should attempt to assemble scientific consensus-building forums, in the spirit of several recent "negotiated rulemaking" efforts. These forums should contain a balanced composition of scientists and professionals from all relevant disciplines to address issues such as the protocols that should guide future health and safety studies, the strengths and weaknesses of existing studies relevant to regulatory questions, the inferences to be drawn from existing studies and the assumptions that should provide the basis for risk assessment models. These groups, of course, would be unable to reach consensus on all issues, but I am confident that some questions will command a consensus or near-consensus of well-meaning participants from a broad spectrum of disciplines and interest groups.

Recent indications that the Office of Science and Technology Policy is moving in this direction in its attempt to forge a national carcinogen risk assessment policy are encouraging.²³ Even if consensus is possible only on a fairly narrow range of issues, the effort itself should reduce the distrust and suspicion that currently pervade public policymaking for health and safety regulation.

Health and environmental regulation will always have an adversarial component. Regulation has winners and losers, and both groups have an intense interest in the process of collective risk assessment and risk management. Yet the process does not have to foster the degree of acrimony and mistrust that has characterized it in recent years. Once we recognize that risk assessment is more than a matter of applying scientific expertise to clearly established facts and that risk management must reflect a balanced view of contending social forces, we can begin to build institutions capable of "pluralistic scientific policymaking" in a spirit of trust and common purpose.

1. Ruckelshaus, Science, Risk and Public Policy, 221 SCIENCE 1026 (1983)

2. National Academy of Sciences, National Research Council, Risk Assessment in the Federal Government: Managing the Process (1983) [cited hereinafter as NAS Risk Assessment Report].

3. NAS Risk Assessment Report, supra note 2, at 29-33.

4. For a more detailed discussion of this point, see McGarity, Substantive and Procedural Discretion in Administrative Resolution of Science Policy Questions: Regulating Carcinogens in EPA and OSHA, 67 GEO. L.J. 729 (1979).

5. NAS Risk Assessment Report, supra note 2, at 3.

6. Another objection to dealing with externally imposed risks through contracts is the problem of the "free rider." To the extent that the risk reduction technologies installed by a person or company to reduce the risks to the individual or individuals who contracted for the risk reduction also protect individuals who are not parties to the contract, the free riders have obtained a benefit without paying for it. Economists tell us that this is inefficient and will, in the long run, reduce the viability of the contract solution, because everyone will attempt to be a free rider and no one will be willing to enter into the contract.

7. Washington Post, Oct. 22, 1983, at A7, col. 4. For a general account of the IBT affair, see, J. LASH, A SEASON OF SPOILS 192-93 (1984).

8. Staff Report on Regulatory Procedures and Public Health Issues in the EPA's Office of Pesticide Programs, reprinted in Hearings on EPA Pesticide Regulatory Program Study before the Subcommittee on Department Operations, Research, and Foreign Agriculture of the House Committee on Agriculture, 97th Cong., 2d Sess. 85 (1982).

9. Washington Post, July 28, 1983, at A3, col.1.

10. See, More Cuts in EPA Research Threaten its Regulatory Goals, Critics Warn, NAT'L J., April 10, 1982; Will EPA's Budget Cuts Make it More Efficient or Less Efficient? NAT J., August 15, 1981; No One Knows for Sure if Pollution Control Programs Are Really Working, NAT'L J., March 23, 1985, at 643. See also, ENVT. REPT. (BNA) 762 (Sept. 14, 1984).

11. See CHEMICAL REGULATION REP. (BNA) 134 (Sept. 17, 1982); id, Feb. 5, 1982, at 1165; Pesticide and Toxic Chemical News, Oct. 6, 1982. An environmental group challenged these behind-the-scenes decisionmaking initiatives under the Federal Advisory Committee Act and obtained an agreement from the Rucklshaus EPA to change its decisionmaking process. See Washington Post, September 21, 1984, at A19, col. 3. n12 See Washington Post, February 10, 1983, at A1, col. 2 (Settlement of Superfund action against the Seymour Recycling Corp. hazardous waste disposal facility in Indiana).

13. See ENV'T. REP. (BNA) 1639 (Apr. 16, 1982).

14. The criticism here would, of course, be the same if the closed-door sessions and the sweetheart deals had been between the agency and representatives of an environmental group. As an agent of the public, the agency is obliged to assess and manage risks in accordance with the policies of its statute in the public interest. As mentioned earlier, the public has an interest in the benefits of risk-producing technologies. Therefore, a one-sided approach is not likely to inspire trust, even if the beneficiary is a self-proclaimed "public interest group."

15. Arguing that the costs of an asbestos ban and phase-down far out-weighed the discounted benefits of extending several thousand lives for a few years, OMB objected to the agency effort and demanded that EPA refer asbestos regulatory matters to OSHA and CPSC. Inside EPA, Sept. 28, 1984 at 1; Inside EPA, Apr. 6, 1984, at 1. EPA complied with the OMB demands, but after a Congressional committee began to investigate the incident, EPA reversed field and halted its plan to refer asbestos actions to the other two agencies. ENV'T. REP. (BNA) (Mar. 15, 1985).

16. See generally, New York Times, Apr. 17, 1985 at B28, col. 1. With a 10 percent discount rate, a benefit of one dollar 50 years from now will be worth slightly less than a penny in today's cost-benefit calculation.

17. See Inside EPA, Mar. 22, 1985, at 1 (OMB objection to EPA National Ambient Air Quality Standard for nitrogen dioxide on the ground that the agency used a "deterministic" approach to assessing the risks of exposure to that substance, rather than a new "three-year statistical" approach).

18. See Washington Post, Sept. 28, 1983, at A1, col. 2 (testimony of former EPA chief of staff, John Daniel).

19. Ruckelshaus, supra note 1, at 1027-1028.

20. Mantel and Schneiderman, Estimating "Safe" Levels, A Hazardous Undertakin, 35 CANCER RESEARCH 1379, 1382 (1975).

21. See H.R. 638, 97th Cong., 1st Sess (1981).

22. NAS Risk Assessment Report, supra note 2, at 4.

23. See Pesticide and Toxic Chemical News, Apr. 17, 1985, at 22.