Incommensurability and cost-benefit analysis.

• Author: Adler, Matthew
• Position: Symposium: Law and Incommensurability

INTRODUCTION

Cost-benefit analysis is a flourishing practice, desperately in need of a justification. By "cost-benefit analysis," or "CBA," I mean the monetized version: the version where the various benefits and costs of a governmental project are reduced to monetary sums and then aggregated, with total money costs subtracted from total money benefits, so that the project ultimately is assigned a single net money amount, positive or negative, the sign of which is in turn taken by the analyst as a significant (if not conclusive) indicator that the project should or should not be approved.(1) This type of analysis has become widely used.(2) In 1981, President Reagan issued an executive order enjoining federal agencies that "[r]egulatory action ... not be undertaken unless the potential benefits to society for the regulation outweigh the potential costs to society,"(3) and requiring the submission of reports detailing the costs and benefits of large projects to a presidential oversight agency, the Office of Management and Budget (OMB).(4) In 1993, President Clinton replaced the Reagan order with a new one, but retained the central requirement that a project's benefits outweigh its costs--"[e]ach agency shall ... propose or adopt a regulation only upon a reasoned determination that the benefits of the intended regulation justify its costs"(5)--and the central mechanism of OMB review.(6) It bears noting that neither the Reagan nor the Clinton orders expressly requires the monetization of costs and benefits. But the orders certainly can be read to license and encourage monetization,(7) and, over the last several decades, federal agencies have indeed, to a significant extent, relied upon monetized CBA as a test by which to approve or reject projects. And yet--this is the desperate part--CBA has been roundly criticized by most scholars who have looked seriously at its normative grounding. The prevailing view--not just among the moral philosophers(8) and legal scholars(9) who have written about the moral justification for CBA and its cousin, the Kaldor-Hicks test,(10) but also among the welfare economists who have done so(11)--seems to be that these tests are morally indefensible. There is now a truly startling disjunction between, on the one hand, the practices of government agencies and (more broadly) the applied economists and public-policy analysts who have generated a vast body of work detailing the monetized costs and benefits of regulatory measures,(12) and, on the other hand, the skepticism about the moral status of CBA that prevails in the scholarly literature.

Is the skepticism warranted? Let me distinguish between CBA as a criterion of moral rightness or goodness and CBA as a morally justified decision procedure.(13) The thrust of the scholarly criticism has been that CBA is not the former. It is not the case that, just because a project's monetized benefits are larger than its monetized costs, there is moral reason to favor the project. And this criticism is plausible. It is difficult to see how CBA marks out something morally important, even prima facie, about government projects or, more broadly, about options for choice. Consider two possible defenses of CBA as a moral criterion, which might be called welfare-dependent and welfare-independent. A welfare-dependent defense (which, as far as I know, no one has seriously pursued)(14) tries to show how CBA is equivalent to a moral criterion of overall well-being. On this defense, a project increases overall well-being--which, after all, surely is a morally significant if not conclusive feature of projects--just in case it satisfies CBA. But why would that be the case? Standardly, a project's monetized benefits are defined, within CBA, as the cumulative amount of money that the "winners" from the project, those whose welfare it improves, would be willing to pay in return for securing it; conversely, a project's monetized costs typically are defined as the cumulative amount of money that the "losers" would require as compensation for accepting it.(15) The welfare-dependent defense of CBA as a moral criterion says, thus, that a project improves overall well-being if and only if the cumulative amount that the winners would pay exceeds the cumulative amount that the losers would require. If this is true, then the following can never obtain.

Trivial Interpersonal Benefits for Rich Winners

The winners from the project are very rich. They own large stocks of money

and other resources and would be willing to pay, in dollars, large

sums to secure a project that, truly, has only a trifling effect on

their own well-being. The losers, by contrast, are quite poor. The

project has a significant negative effect on their well-being, but

they would require only a moderate sum in compensation. Because of

their poverty, the project losers can use moderate amounts of money

to purchase improvements in the conditions of their own lives (and

in overall well-being) that are even greater than the negative effects

on them wrought by the project. So the project passes the CBA test

but, if implemented (without any payment to the losers), decreases

overall well-being.

A welfare-dependent defense of CBA entails, implausibly, that this scenario is impossible. If, for example, the construction of a particulate-emitting device produces a delightful reddish tinge to sunsets that the rich view from afar, at the cost of considerable pulmonary discomfort to the poor neighbors who live nearby, it is surely true or at least possible that the project decreases overall well-being even though the project's monetary benefits are greater than its monetary costs.(16)

The alternative defense of CBA, as a criterion of moral rightness or goodness, is a welfare-independent defense. On this defense, CBA marks out a moral feature which is not equivalent to a project's contribution to overall well-being, but is morally important nonetheless. Historically, at least within welfare economics, this has been the standard way to think about CBA. The view has been that CBA is normatively defensible just insofar as it tracks a Kaldor-Hicks test. Kaldor-Hicks, in turn, originated in an intellectual milieu in which it was thought that interpersonal welfare comparisons were impossible.(17) The effect on "overall well-being" of a project that improved the well-being of some, while making others worse off, was thought indeterminate. Rather, and alternately, it would count as a reason in favor of the project that the winners could afford to compensate the losers, making at least one person better off and no one worse off. This is what the Kaldor-Hicks test asks: Is there a costless redistribution from winners to losers such that, if performed together with the project, the project- plus-redistribution would constitute a true Pareto improvement? Yet the Kaldor- Hicks test is vulnerable to the same kind of objection routinely leveled against proposed moral criteria that incorporate the merely hypothetical rather than actual consent of affected persons. A person has the power to change her moral position by promising, consenting, and undertaking other such performative utterances; but a necessary condition for this kind of change is that the utterance actually be performed.(18) The fact that, hypothetically, the rich winners from a project that decreases overall well-being could afford to compensate the poor losers counts as no moral reason whatsoever in favor of the project.

In short, CBA is indefensible as a criterion of moral rightness or goodness--or so I will assume here. This is not to say, however, that CBA is indefensible tout court We are considering, after all, a government practice: a decision procedure by which (an increasing number of) federal agencies and other government institutions assess projects. At bottom, the question whether government institutions should employ this decision procedure is a moral one. But that question, in turn, is not answered by saying that the relevant criteria of moral rightness or goodness are not incorporated directly into the practice. It could be the case that--by virtue of the epistemic limitations of agencies in determining what morality requires, plus their tendency, more or less strong, to pursue nonmoral aims--morality warrants a legal requirement that agencies employ a decision procedure more algorithmic, or verifiable, than the direct application of moral criteria.(19)

This is what Professor Eric Posner and I have argued in an unpublished work on CBA.(20) In particular, we have suggested that CBA is most plausibly defended as a decision procedure by which to implement the moral criterion of overall well-being. We claim that whether a project increases or decreases overall well-being is a morally relevant, if not conclusive, feature of that project.(21) And CBA will, in some (not all) contexts, be both reasonably congruent with the welfare criterion and significantly easier to implement than that criterion itself In particular, CBA is reasonably congruent with overall well-being where money is equally productive of well-being as between winners and losers. Sometimes it is not, as in the case of the reddish sunset, but sometimes it is, and--before scholars conclude that the widespread and increasing practice of monetized CBA is simply misconceived--CBA should be evaluated in this light, as a welfarist decision procedure. "Welfarist" here just means "relative to a criterion of overall well-being."(22) Why is monetized CBA a more plausible welfarist decision procedure than one that scales costs and benefits in terms of some other item, such as roses, or corned beef, or liberty? Money, unlike roses or corned beef, is a primary good, that is, it is broadly (instrumentally) advantageous for well-being;(23) and, unlike liberty, has an existing public scale in the sense that many constituents of welfare, namely, marketed goods and services, already are scaled...