Parameterizing the Precautionary Principle
In Cass Sunstein's Worst Case Scenarios and Risk and Reason, and Richard Posner's Catastrophe, the authors grapple with how and when policymakers ought to fashion regulation and incentives to manage risks--especially catastrophic ones. Though mundane in the details, proper environmental, economic, and safety regulations have enormous welfare effects: deciding whether the risks from global warming justify cap-and-trade, or whether the risk from arsenic in drinking water justifies a maximum mandated level of 5 micrograms per liter or 20 micrograms per liter, for example, are extremely technical questions in which real lives are at stake. Posner and Sunstein generally endorse cost-benefit analysis as a means of settling these questions. But alternative decision-making paradigms exist. One that carries particular sway in environmental circles is the so-called precautionary principle. This principle is stated in an extremely vague form that renders it useless and contradictory, which Posner and Sunstein recognize. In several off-hand comments, Sunstein tempers his rejection, but he does not give a thorough explanation of how the precautionary principle would practically be applied. I think, however, that it can be quantitatively specified in such a way that it can be used in traditional cost-benefit analysis.
The first problem with the precautionary principle is its general vagueness. Stating that society ought to "err on the side of caution" with respect to catastrophic risks is useless because the relevant question is how much society should err on the side of caution. Certain risks are too improbable to spend time worrying about, and the precautionary principle does not give any indication of what risks we should care about. More devastating, however, is the fact that the precautionary principle is ultimately contradictory because there can be risks to inaction as well as action. Banning GMO foods to avert the risk of ecological interference creates the risks of food shortages and environmental damage from increased farming due to decreased agricultural productivity. The precautionary principle is useless in this situation. Worse yet, it is prone to rampant abuse, as every party to a regulation could claim to be justified by the precautionary principle. There is a significant literature on the other problems of the precautionary principle, so I will not dwell on this matter. But the point is that it is generally deficient.
It is noteworthy that the precautionary principle appears to be very high regard in environmental circles. This is probably because the spirit of the precautionary principle captures a very intuitively appealing notion: risk aversion. Most people are risk averse: this is, after all, why insurance exists. Risk aversion is also not form of cognitive bias. Even with perfect information, people will usually be somewhat risk-averse. Given that a democratic government ought to somewhat reflect the interests and desires of its constituents, incorporating society's "average" level of risk aversion into cost-benefit analysis (CBA) seems reasonable.
From this standpoint, one can view a regulation that eliminates one risk as insurance. From the usual two-state model, a person's willingness to pay for this insurance is simply the difference between his wealth in the better outcome and his wealth at the point at which his utility function equals the expectation value of his utility from the two outcomes. For a given degree of risk-aversion, willingness to pay is thus some function of the probability P and magnitude L of the possible loss, say W=W(P,L). In a similar way that standard CBA uses wage premia in risky professions to infer the value of a statistical life, one could use information from how much people pay for insurance to estimate the function W(P,L). In a standard cost-benefit analysis one could then allow the costs of acceptable regulation to exceed the benefits by N*W(P,L/N) in a population of size N.
Granted, this process would be highly uncertain in the same sense that valuing a statistical life is uncertain: values cluster around $6-7 million, but range from $0.7 million to $15 million. Calculating the function W would also be extremely difficult because risk-aversion is very different for qualitatively different risks. Without delving into the complete details, we can say that any quantitative attempt to address risk-aversion in a social setting, however, is better than the vague handwaving of the usual precautionary principle. Moreover, this approach solves the problem that occurs when action and inaction each carry risks. Given the probability and monetized magnitude of each, one can calculate society's "willingness to pay" for averting each risk and then add this onto the costs and benefits from the usual CBA. Given how tenuous many assumptions in CBA are anyway, this proposition is probably not unreasonable.
0 TrackBacks
Listed below are links to blogs that reference this entry: Parameterizing the Precautionary Principle.
TrackBack URL for this entry: http://www.appmagic.com/cgi-bin/MovableType/mt-tb.cgi/76
This all presumes that people can people have some sort of decision calculus that uniformly deals with low probability-catastrophic events.
My intuition is that people don't have any such thing.