Uncertainty

Abstract

Many of the pressing policy issues facing us today require confronting the unknown and making difficult choices in the face of limited information. Economists distinguish between "uncertainty" (where the likelihood of the peril is nonquantifiable) and "risk" (where the likelihood is quantifiable). Uncertainty is particularly pernicious in situations in which catastrophic outcomes are possible, but conventional decision tools are not equipped to cope with these potentially disastrous results. This Article focuses on situations in which uncertainty, particularly about catastrophic outcomes, is a dominant factor. The Article describes new analytic tools for assessing potential catastrophic outcomes and applies them to some key policy issues: controlling greenhouse gases, adapting to unavoidable climate change, regulating nanotechnology, dealing with long-lived nuclear waste, and controlling financial instability. More specifically, economic modeling and policy analysis are often based on the assumption that extreme harms are highly unlikely, in the technical sense that the "tail" of the probability distributions is "thin"-in other words, that it approaches rapidly to zero. Thin tails allow extreme risks to be given relatively little weight. A growing body of research, however, focuses on the possibility of fat tails, which are common in systems with feedback between different components. As it turns out, fat tails and uncertainty often go together. Economic theories of "ambiguity" deal at a more general level with situations in which multiple plausible models of reality confront a decision maker. Ambiguity theories are useful in considering systems with fat tails and in other situations in which the probabilities are simply difficult to quantify. The Article considers both the policy implications of fat tails and the use of ambiguity theories such as α-maxmin.