Regulators must measure welfare using transparent methods before determining the policies themselves.
Once again, I appreciate Professor Viscusi responding to my essay. I believe he has a decent grasp of my position. I would, however, like to point to several areas of misunderstanding that remain and make one important point for clarification.
First, the position I am advocating for is perfectly consistent with basing benefit values on the willingness to pay (WTP) of citizens. I simply choose to take the WTP of present and future people into account, rather than to consider some generations and not others or to weight the WTP of people in an unequal way.
The value of a statistical life (VSL) is based on an assumption that marginal WTP values observed in the marketplace are efficient. That assumption is untenable; although consumers and workers may optimize their own utility across their lifetimes, they are not optimizing utility across generations. The VSL gives inefficient recommendations as a result (as do many other estimates of policy benefits, as Viscusi correctly notes), and market failures are exacerbated when policy is guided by the VSL.
Furthermore, I disagree with Viscusi’s contention that this is a recipe for more-dangerous workplaces or “polluted neighborhoods with hazardous waste exposures.” Viscusi’s own research demonstrates why a cost-of-death approach to valuing lives would likely reduce risk, while basing policy on the VSL will increase it over time.
Regulatory policy directed by the VSL may well reduce risk temporarily—assuming, of course, that regulations are well-designed enough to work. But it generally does so at the expense of capital investment and growth, which means fewer resources to devote to health, safety, and the environment in the future. We enjoy short-run benefits while our successors have fewer resources to address those risks most pertinent to their own lives.
We should not limit ourselves to a choice between their safety and our own, however. Policy can aim to increase efficiency over the long term and make people better off today according to their own values. Systematically overriding consumer choices is not required, but we do need to think carefully about what kinds of institutional arrangements produce such a balancing of present and future interests.
Finally, I would like to take a moment to explain why Viscusi is wrong that “benefit-cost analyses of regulatory policies recognize the preferences of both current and future generations without shortchanging either group.”
I believe Viscusi makes this claim because he misunderstands the role of the discount rate in cost-benefit analysis (CBA). He has stated in several places that he views the social discount rate as representing something like an underlying rate of productivity growth in the economy. I can hardly criticize him for taking this view. Indeed, at one time I did too.
It is true that the underlying productivity growth rate he describes is represented by a discount rate in the special case of CBA, in which all benefits and costs are financial. In that special case, all benefits and costs are growing in value at the same rate. But in more general cases where some of the benefits and costs of policy are nonpecuniary—which is the case with regulations that address mortality risks—this productivity growth rate is represented by a shadow price, not a discount rate. Why? Because benefits and costs grow or depreciate at different rates, and these differences need to be accounted for separately.
This may seem like a minor technical detail, but it is not. The discount rate for social regulations diminishes the value of benefits and costs that occur in the future. That is its role. Furthermore, the productivity of capital that Viscusi describes is consistently not accounted for in the regulatory impact analyses for social regulations, because no shadow price is applied to the value of capital investment.
For these two reasons, I do not share Viscusi’s confidence that CBA treats all generations equitably. The lives of future citizens receive less weight in analysis than do our own, and compounding returns from induced or displaced capital investment go systematically overlooked.
An additional advantage of my approach is that it is entirely consistent with the Kaldor-Hicks framework that underlies cost-benefit analysis. That criterion requires that benefits and costs receive the same weight irrespective of who receives them. The unequal treatment of benefits and costs Viscusi endorses does not meet this requirement of distributional insensitivity, and therefore is inconsistent with CBA measuring efficiency.
Some tradeoffs we face may make us uncomfortable. I believe that is a key reason for the VSL’s popularity. It obscures uncomfortable truths, while reliably reaching fashionable policy conclusions. But we are economists. It is our job to face reality head on. That means settling on a meaningful measure of welfare for CBA that makes tradeoffs transparent, and then letting the chips fall where they may with respect to the policy implications. It is time to put economic science back in its proper order.
This response is part of a seven-part series, entitled The Value of the Value of a Statistical Life.