Rationality or Rationalism? The Positive and Normative Flaws of Cost-Benefit Analysis, Part 3

Here’s the third (and final) installment of my Houston Law Review piece on Rationality or Rationalism? The Positive and Normative Flaws of Cost-Benefit Analysis. (Click here for the first installment and here for the second installment. See the printed version, 48 Hous. L. Rev. 79 (2011), or the SSRN version, for all the footnotes.)

* * *

Finally, there’s the normative point, which has already been extensively covered by critics of utilitarianism. Deciding whose costs and benefits count is normatively loaded—this is why Cass Sunstein got into trouble for his concern with the interests of animals. So is deciding how costs and benefits should count—this is why the question of intergenerational discounting is so contentious. So is deciding what is a cost and what is a benefit. When a regulation results in a reduction in suicides, it’s common to count this reduction as a benefit; Dean Revesz does so, for instance, when discussing the benefits of auto emission limitations and a similar claim is often made in the context of gun control. But calling suicide prevention a benefit is (or should be) ethically controversial both from a “consumer sovereignty” perspective and from a libertarian perspective that sees suicide as a choice like any other.

[Footnote (citations omitted): Revesz defends a concern for suicide prevention on the grounds that “[a] person’s choice to undertake suicidal behavior may not be a reflection of her true self and her self-inflicted death could be an act that she would, in calmer and clearer moments, recoil at.” Because suicidal thoughts may be connected to mental illness, a suicidal person’s “rational autonomy” might be compromised, so that “the decision to engage in suicidal behavior [might] not [be] a reflection of one’s considered values or aims.” But this sort of reasoning is precisely what should be morally controversial. To count suicide reduction as a benefit means that one is privileging some hypothetical “true self” that has “considered values [and] aims,” rather than the currently existing self that has actual values and aims. The idea that the cost–benefit analyst can identify a “norm,” deviations from which are considered a “mental illness” and can justify overcoming an actor’s revealed preference, clearly shows the interventionist (and possibly illiberal) potential of cost–benefit analysis.]

But the biggest problem isn’t particular choices about how to implement the utilitarian basis of cost–benefit analysis; it’s the utilitarian basis itself. Dean Revesz answers the critique that cost–benefit analysis inappropriately commodifies human life or wilderness areas, correctly pointing out that assigning a price to life shouldn’t be understood to diminish its inherent worth. Pets, homes, and wedding rings are, after all, traded in markets. But this is still a bit too glib: the purpose of formal cost–benefit analysis isn’t just to assign a number, but to do so in order to see whether that number can be trumped by a bigger number on the other side.

Dean Revesz’s phrasing makes the moral collectivism clear: cost–benefit analysis “allows us to spend money to the point at which the last dollar spent buys one dollar of risk reduction. If we spend beyond that point, we will pay more than we receive. But if we spend any less, we forego risk reductions that are socially desirable.” I’ve added the emphasis to show how seamlessly one slips from the we who pay and receive to the we who receive and forego. But these are different groups, and there is neither a common pocketbook nor a common valuing mind. Coase notwithstanding, there is no social cost. Cost–benefit analysis legitimizes a social welfare function that treats it as unobjectionable to, without compensation, take a book from Derek, who values it at $2, and give it to Amartya, who values it at $3. But, as Ronald Dworkin asks, is the world after the transfer “in any respect superior” to the world before the transfer?

I do not mean whether the gain in wealth is overridden by the cost in justice, or in equal treatment, or in anything else, but whether the gain in wealth is, considered in itself, any gain at all. I should say, and I think most people would agree, that [the world after the transfer] is not better in any respect.

[Footnote to Dworkin blockquote: Dworkin may be pushing this point too far. Perhaps a single such transfer doesn’t make the world better off in any way, but two such transfers, one from Amartya to Derek and one from Derek to Amartya, could make both parties better off, and this would make the world better off in some way, at least to someone who thought that Pareto improvements make the world better off, other things being equal. If these two transfers make the world better off, then surely one can say that each individual transfer plays a role in making the world better off provided it’s embedded in some structure of reciprocal transfers. But Dworkin is still on to something: there’s generally no guarantee that the reciprocal transfers will take place; there’s nothing in the logic of cost–benefit analysis that guarantees that this regulation will be counterbalanced by other regulations that benefit today’s harmed parties; and today’s political losers are likely to be people without much political power, who are likely to continue to be political losers tomorrow.]

Commodification, here, isn’t the primary problem. I think commodification is great and encourage its more widespread use in the law, whether for sexual services, human organs, or human life. A basic tenet of libertarianism is self-ownership, and what’s more fundamental to ownership than transferability? Thus, preventing people from agreeing to suffer a risk of death (on any basis they see fit, including for a price) is morally indefensible. But while I favor treating most rights as commodities that can be alienated, traded, and sold, I would also insist that these commodities be treated as the property of their owners and not of society. Cost–benefit analysis asks us not only to commodify but also to transfer the right to the commodity to the highest valuer.

The Political Case: Can Cost–Benefit Analysis Be Defended Instrumentally?

Of course, even if cost–benefit analysis is theoretically incoherent, doesn’t generally provide the right answers, and can’t be justified ethically on its own terms, it might still be a good idea. But the instrumental defense of cost–benefit analysis has pitfalls of its own. I’ll examine three instrumental defenses: one from procedural rationality, one from substantive efficiency, and another from substantive morality.

First, the procedural defense. Maybe cost–benefit analysis disciplines thinking and makes it more transparent. Indeed, this is, in Dean Revesz’s view, what makes formal cost–benefit analysis better than its “crude,” intuitive counterpart, which doesn’t count costs of the politically weak and overweights benefits of the politically strong. Formal cost–benefit analysis “reveals the distortions of politics” and makes it easier for voters to, if necessary, “throw[] the bums out.”

Of course, Dean Revesz believes that the result of a neutral cost–benefit analysis will also be “more accurate,” so perhaps his transparency argument hinges on truth here. But conceivably, even those who don’t believe cost–benefit analysis is theoretically defensible could still defend it as an improvement over the undisciplined, ad hoc approach. “Just as the environmental effects of government actions must be accounted for in an environmental impact statement,” says Dean Revesz, “the economic effects should be acknowledged as well.”

This is all well and good, provided that cost–benefit analysis really does make regulatory costs and benefits more transparent. But the critics of cost–benefit analysis make a strong case that the rhetoric of cost–benefit analysis may, in fact, obfuscate and make things less transparent—providing a mere “veneer of technical precision to regulatory judgments and augment[ing] the political case for action.” This is especially true if the methods are hard to understand or if they’re not very determinate and thus subject to political manipulation. Straightforward democratic deliberation based on moral values, by contrast, can be extremely transparent.

Second, the argument from substantive efficiency. Suppose one believes there’s inefficiently much regulation and that cost–benefit analysis is antiregulatory. One could then support cost–benefit analysis as a mechanism that, whatever its faults, at least puts a thumb on the scale in the direction of restoring the efficient amount of regulation.

But for this argument to work, at a minimum, there must be inefficiently much regulation. There may be good reasons to think this is correct. For instance, self-aggrandizing administrators may want to expand their power. Or, even without any self-aggrandizement motive, risk regulators may face perverse incentives if they’re “punished when they fail to avert a risk that becomes manifest,” whereas “the costs of their rules are born[e] by regulated firms and the general public.” But there are also good theoretical reasons to think the opposite. The agency-aggrandizing hypothesis, for instance, is iffy; and agency administrators’ incentives may be reversed—in the direction of inaction—when the costs fall on a concentrated industry while the benefits are dispersed, or when the costs are borne today but the benefits accrue far in the future, as some argue is the case for global warming. One may also argue that environmental concerns will always be slighted in policy deliberation because industry is always more concentrated and, therefore, better able to organize than the polluted public. (Note, though, that both environmental and industry groups are subject to the collective action problem, and environmental organizations have organized successfully and exerted influence on policy).

Finally, the argument from substantive morality. One might think that the amount of regulation, while perhaps efficient, is nonetheless excessive for other reasons. Suppose one is a natural-rights libertarian. An efficient regulatory system may succeed in regulating all sorts of “externalities,” but not all externalities are equal. A system of rights might tell us that some externalities, like killing someone through pollution, can legitimately be addressed by government while others, like my objections to someone else’s religious beliefs, can’t. One might, of course, argue that religious freedom, or freedom of speech, or the like, is in fact efficient, and that if a cost–benefit analysis doesn’t yield that result, it just means that the analyst has insufficiently weighted the relevant liberty interests or has imperfectly understood the negative consequences of violating the rights. This may well be true, but it won’t satisfy the philosopher: a natural-rights perspective is more than just “utilitarianism properly done,” even if the two approaches would usually ((almost?) always?) yield the same result. Under a natural-rights perspective, a right can’t be violated even if everyone would agree that it would be efficient to do so. One might also argue that certain pleasures, like the pleasure of suppressing someone’s religious beliefs, shouldn’t be counted in a cost–benefit analysis. But this would be contrary to cost–benefit’s utilitarian logic; more particularly, it would imply a violation of the Pareto principle.

Add to this the natural-rights libertarian’s emphasis on economic rights as deserving of as much protection as freedom of speech or religion, and it’s not surprising that, for the natural-rights libertarian, (possibly) efficient but (nonetheless) immoral regulation is everywhere. This includes workplace safety regulation, consumer product safety regulation, safety and effectiveness regulation of pharmaceuticals, and other areas where exposure is voluntary and product characteristics should be determined by contract—as well as various noncontractual areas where people impose risk on themselves, like seat belt or motorcycle helmet mandates. (I don’t mean to categorically rule out regulation of these areas: there may be fraud at work; risks may affect third parties; or, for some reason or other, an apparently voluntary party may be incapable of consenting. But at least the natural-rights libertarian can categorically rule out considering the effect of risk on the adequately informed, consenting, sane adult as a rationale for regulation.) In this context, an antiregulatory cost–benefit analysis can serve a morally valuable function, gumming up the works and generally making it harder to enact regulations, even if the actual numbers it produces are indefensible.

But there are two problems with this approach. First, it’s blunt. The regulatory state today covers many areas that would probably be off limits in a libertarian state, but it also covers some areas that a libertarian wouldn’t categorically object to. For instance, air pollution regulation, provided it regulates true health risks in a cost-effective manner, may well be justifiable, even on a libertarian view. An antiregulatory cost–benefit analysis would thus unfortunately block both permissible and impermissible regulations.

Second, this strategy requires that cost–benefit analysis actually be antiregulatory. Here, Dean Revesz performs a valuable service in showing that this isn’t necessarily true. If he’s right—and I see no reason to doubt his conclusions as a general matter—even the instrumental use of cost–benefit analysis by natural-rights libertarians may be a bad idea.

Dean Revesz isn’t the first to argue that “neutral” economics can be proregulation—many libertarians have done so with dismay, stressing that neoclassical welfare economics can be a recipe for economic intervention. Still, libertarians have tended to go along with cost–benefit analysis; and indeed, it’s been politically useful, even in Democratic administrations. If Dean Revesz’s insights catch on in the practice of cost–benefit analysis, libertarians might have to rethink this marriage of convenience.

Dean Revesz makes the argument to environmentalists that their moral rhetoric, while politically attractive, should be merged with economic arguments because “[t]he same story can be told in many ways.” Why, indeed, shouldn’t a moral environmentalist participate in cost–benefit analysis, always appropriately beginning his remarks with: “I believe cost–benefit analysis is immoral. But since you’re doing it anyway, here are some arguments that cut in favor of environmental regulation . . .”?

This is possible. But some ways of telling the story may, in the long run, undermine one’s case. Surely one shouldn’t always participate in every available political process to pursue one’s goals. Dissidents in countless illegitimate regimes have argued that participation in indefensible processes, even though it can seem to serve one’s goals in the short run, ultimately gives moral legitimacy to the oppressor. Closer to home, it’s not always easy for political organizations or their supporters to keep conflicting messages in their head at the same time; we’re not all trained as lawyers in arguing in the alternative. Those who oppose the utilitarian basis of cost–benefit analysis might do well by participating—but they might also do well by dissenting from the sidelines, or at least (if they still participate) by not losing sight of, by always stressing, and perhaps, in this case, by rediscovering their principled dissent.