Risk, Death and Harm: The Normative Foundations of Risk Regulation

Is death a harm? Is the risk of death a harm? These questions lie at the foundations of risk regulation. Agencies that regulate threats to human life, such as the EPA, OSHA, the FDA, the CPSC, or NHTSA, invariably assume that premature death is a first-party harm - a welfare setback to the person who dies - and often assume that being at risk of death is a distinct and additional first-party harm. If these assumptions are untrue, the myriad statutes and regulations that govern risky activities should be radically overhauled, since the third-party benefits of preventing premature death and the risk of premature death are often too small to justify the large compliance costs that these laws create.In this Article, I consider the harmfulness of death, and of the risk of death, in a philosophically rigorous way. The analysis is complicated, since a variety of plausible theories of welfare have been proposed, and since risk too is a multifaceted concept. A given person P's "risk" of death might be risk in a Bayesian sense (some person's subjective probability that P will die), or risk in the frequentist sense (the objective frequency with which persons like P die prematurely as a result of the kind of threat to which P is exposed). These two conceptions of risk are very different, yet too often are not distinguished in legal or policy-analytic writing about risk. As for the harmfulness of death: this raises knotty philosophical problems, problems that have prompted some contemporary philosophers to deny that the dying person is worse off than she would have been had she continued to live.I ultimately conclude that death is a first-person welfare setback - common sense is vindicated here, I argue - as is risk in the Bayesian sense, but that risk in the frequentist sense is not. This conclusion has implications for a range of regulatory practices - specifically, for cost-benefit analysis, risk-risk analysis, the interpretation of statutes that create health or safety thresholds, environmental justice policy, and comparative risk analysis - and also for tort and criminal law. These implications are explored, at length, in the final section of the Article. In particular: the widespread use of frequentist risk measures as a determinant of regulatory choice is misguided. EPA, OSHA, FDA and other federal and state agencies typically determine how stringently to regulate some toxin by looking (at least in part) to the frequentist risk imposed by the toxin on the maximally exposed, highly exposed, or representative individual. Similarly, environmental justice analysis is often keyed to the distribution of frequentist risks. And some propose that regulatory priority-setting (so-called comparative risk assessment) also take into consideration frequentist risk. This regulatory focus on frequentist risk was encouraged by the Supreme Court's seminal decision in the "Benzene" case (Industrial Union Dept v. American Petroleum Institute, 1980), and is endorsed by the risk assessment community. But the practice has no normative basis, and should be abandoned. Similarly, risk-imposition in the frequentist sense should be neither tortious nor criminal - at least if harmfulness is a precondition for liability in these domains, as it may well be.