Prospects for utilitarian decision analysis

Decision analysis evaluates options in terms of their expected utility, the expected amount of good. Many regulatory decisions require estimates of utility, which is not the same as monetary outcomes or other objectively measurable quantities. I review some of the problems of utility measurement for consequences (hence “utilitarian”) and suggest possible solutions. Contingent valuation (CV), a widely used method for obtaining money equivalents, is usually insensitive to the quantity of the good being evaluated, but this problem might be avoided by using per-unit pricing and/or by requiring responses on both attributes (money and the good in question). Valuation of consequences can also be distorted by provision of information about how the consequences are achieved. I report a study showing how valuation can also be improved by stripping away information other than the consequences of interest; most subjects did not think that removal of the extra information was unfair. Conjoint analysis, another method in which respondents rate objects that vary in different attributes such as cost and benefit, seems promising because it is often sensitive to quantity. I report a study suggesting that it can suffer from the problem that people limit their attention to a few attributes, possibly even one. I conclude with a discussion of the role of utility measurement in regulation.