Cost-Benefit Analysis: Why Relative Economic Position Does Not Matter

Abstract

The current debate over cost-benefit concerns in agencies' evaluations of government regulations is not so much whether to consider costs and benefits at all but rather what belongs in the estimated costs and benefits themselves. Overlaid is the long-standing concern that the distribution of costs and benefits needs some consideration in policy evaluations. In a recent article in the University of Chicago Law Review, Robert Frank and Cass Sunstein proposed a relatively simple method for adding distributional concerns to policy evaluation that enlarges the typically constructed estimates of the individual's willingness to pay for safer jobs or safer products. One might pay more for safety if it were the result of a government regulation that mandated greater safety across-the-board. The reason, Frank and Sunstein argue, for enlarging current estimates is that someone who takes a safer job or buys a safer product gives up wages or pays a higher price, which then moves him or her down in the ladder of income left over to buy other things. Alternatively, a worker who is given a safer job via a government regulation has no relative income consequences because all affected workers have lower pay. We show that when considering the core of the Frank and Sunstein proposal carefully one concludes that current regulatory evaluations should be left alone because there is no reason to believe that relative positional effects are important either to personal decisions in general or to currently constructed cost-benefit calculations of government regulations in particular. One of the practical problems with trying to consider relative position of income and consumption when estimating willingness to pay is that there is no unique way to ascertain from a statistical model the person's actual social reference group. A researcher must specify ex ante a reference group and then net out the behavioral effects of a possibly incorrectly attributed reference group's behavior on the individual. There is no well-established result from survey data for a typical person's economic reference group. Moreover, the econometric literature generally finds that reference group or social interaction effects are small and easily ignored, perhaps because the relative positional effects of workplace or product safety offset possible reference group effects on residual income (income net of the implicit cost of the extra product or job safety). It is also the case that Frank and Sunstein's recommended increase in the value of willingness to pay for safety used in current regulatory evaluations is already considered. Regulatory evaluations often include a pessimistic and an optimistic value of likely benefits, and Frank and Sunstein's suggested revised value of willingness to pay is still below the optimistic case that carefully formulated cost-benefit studies use. It is easy to show that almost doubling the estimated value of a statistical life would have an inconsequential effect on the economic desirability of a broad set of regulatory policies. Finally, we argue that the most important refinements one could make in the area of regulatory evaluation would be for agencies involved to adhere more to the framework of what is generally considered a carefully done cost-benefit study, and for agencies to make greater actual use of appropriately done cost-benefit studies when recommending regulations.