Comparative risk analysis: limitations and opportunities

Journal of Economic Issues, June, 1996 by James A. Swaney

Comparative Risk Analysis

Comparative risk analysis (CRA) is a policy application of risk analysis designed to provide a framework for systematically evaluating risk reduction programs and budgets for the purpose of ascertaining whether different approaches or different priorities might go further or faster to produce environmental improvements. In short, CRA attempts to look at the big picture to see if we are getting the "most bang for the (risk management) buck." Initially, CRA used the opinions of risk analysts and other scientists to compare and rank risks. More recently, some comparative risk projects have produced risk rankings that also incorporate public opinion. The United States Environmental Protection Agency introduced CRA with its 1987 report, Unfinished Business: A Comparative Assessment of Environmental Problems [EPA 1987]. The objective of Unfinished Business was to employ "sound science" to identify manageable but neglected risks and refocus policy toward them.(1) Thirty-one environmental problems were examined for their effects on human health, ecology, and welfare, and an integrated risk ranking was produced (Appendix 1). Some high profile risks, such as hazardous waste sites, were ranked fairly low, and some low profile risks, such as indoor air pollution, were ranked fairly high.(2) The basic message of Unfinished Business was that new regulatory initiatives were needed to address some higher ranked risks that had received relatively little regulatory attention. About a year later, two nationwide Roper polls produced a citizen ranking of a similar list of problems, exposing a gulf between public and expert opinion in several areas, including hazardous waste, indoor air pollution, and global warming. The "consensus interpretation" of Unfinished Business and the Roper polls, respectively, is that regulatory resource allocation could be improved and that better communication of scientific opinion to the public is needed both to facilitate low-cost risk reduction strategies and to build political support for new regulatory initiatives. These points are well taken, yet risk analysts will have to admit that "we do not have (and may never have) good methods for comparing different types of risk" [Davies 1995, 8]. Specific risks, and by extension classes of risks, are difficult to compare because "risk assessment is still a relatively crude science, and, depending on which methodological assumptions are used, its results may vary a hundredfold or more" [Davies 1995, 8]. But even if these measurement hurdles could be cleared, three fundamental, interrelated problems inhibit the comparison of different types of risk: the values problem, the dimensions problem, and the rationality problem.

The values problem lurks in the definition of risk analysis. The terms "hazard," "harm," "unwelcome outcome," and "adverse effect" all imply value judgments. As risk analysis has expanded from its initial focus on human mortality to other harms or adverse effects, its value foundations have become apparent. When outcomes are considered adverse by some but not by others, or when a hazard produces multiple bad outcomes and people disagree on the "relative unhappiness" of those outcomes, or when risk reduction alternatives involve tradeoffs between different types of bad outcomes, the definition and severity of adverse effects will be contentious. For example, how should risks to mother and unborn child be weighed in a problem pregnancy? How does one presume to compare risks to individual liberties, risks to doctors' professional rights and responsibilities, risks to community standards of decency and morality, and risks to a woman's life? How should harm to ecosystems, human health, and quality of life in the future be compared to employment benefits today? Granted, these questions involve tradeoffs. But for questions such as these, benefit-cost analysis (BCA) is a high-risk activity. Hurt to the world's poor, to future generations, and to the land community (species, ecosystems, global commons) are shortchanged or ignored in BCA. Even if the intra- and inter-generational welfare issues were solved, BCA would still fall short because the values it measures are suspect. Increasingly, "value-added" by the market satisfies, or does not, the preferences produced there. Even if consumer preferences were not of suspect quality, there is more to life, and immeasurably more to good community life, than individual satisfactions from consumer goods.

The dimensions problem arises from the fact that risk assessment measures only two dimensions of risk, probability and severity. But risk is in fact multidimensional, with perceptions of risk influenced by voluntariness, dread, familiarity, controllability, potential for catastrophe, and other dimensions. Since there is no metric for measuring or combining these diverse risk dimensions, there is no way to systematically compare risks. Paul Slovic and colleagues have developed a simple model that groups the various dimensions of risk into two categories, dreaded risk and unknown risk [Slovic 1987]. Dreaded risk includes catastrophic potential, controllability, and voluntariness. Nuclear war is high dread; power mowers are low dread. Unknown risk includes threats with delayed or unobservable consequences. DNA technology is unknown; automobile accidents are known. Likelihood and severity appear to have little influence on public concern and desire for regulation relative to these other dimensions of risk. For example, both DNA technology and radioactive waste are high dread and unknown, and, although experts rank both as "low risk," the public wants both strictly regulated. Yet the public does not want regulation of downhill skiing or home swimming pools, two high-risk activities that are low dread and known. While better communication could inform citizens and reduce public apprehension, "dreaded" and "unknown" in Slovic's model are headings for categories that include diverse risk dimensions. Should public policy address citizen concerns, or should it "stick to the facts" and follow the advice of those risk professionals who claim that their science provides the valid basis for risk management? Slovic argues that public risk perceptions are legitimate and that wisdom from the public, as well as from scientific risk assessment, needs to be incorporated into the policy process [Slovic 1990, 14].