Harold Leventhal Talk: Cost-Benefit Analysis Colloquy: Squaring the Vicious Circle

Cost-Benefit Analysis Colloquy: Squaring the Vicious Circle
By Stephen F. Williams*

Introduction
It's a pleasure to give a talk honoring Judge Harold Leventhal. My subject today—government control of environmental risk—is one on which Judge Leventhal greatly distinguished himself ¹, and one which could surely use his analytic powers. But here I would like to pursue issues raised by a similar figure, Justice Breyer, in his well-known Holmes Lectures, published as Breaking the Vicious Circle². I will explore an intellectual issue that I think must be at least resolved in general terms, if the reforms he proposed are to have any chance of success: whether cost is to be seriously considered in deciding how far government should go in reducing environmental risk.

• Justice Breyer's Breaking the Vicious Circle
  Many of the anomalies that Justice Breyer addressed came to his attention in a First Circuit case, United States v. Ottati & Goss, Inc.³, which has become the poster child for agency absurdity in risk regulation. One of the parties had cleaned up a toxic waste dump well enough that dirt–eating children could safely eat dirt there for seventy days a year; but it had resisted the EPA's demands that it spend an extra $9.3 million to make the dirt safe for children who might eat dirt there for 245 days a year. The area was a childless swamp, with no residences and no prospect of residential development, and at least half of one of the types of pollutants would likely disappear on its own by 2000 ⁴. The First Circuit agreed with the district court that there was no need for the extra clean–up. Ottati & Goss, it turns out, is not the only instance of the EPA's surprising zeal in the protection of non-existent dirt–eating children. In Leather Industries of America, Inc. v. EPA ⁵, our court considered EPA risk assessments based on an assumption that children would consume sewage sludge used as fertilizer in highway medians, roadside cemeteries, industrial parks, and golf courses, and that they would do so every day for up to five years. ⁶

  Relying on these examples and others, Justice Breyer found a general mismatch in priorities: some are corrected at extravagant cost, while others are neglected or attract far less investment per statistical life saved. He included a table that lists regulations by cost per life purportedly saved, including twenty–one regulations with costs per life of over $10 million and one with costs per life reaching $5.7 trillion.⁷ One commentator has criticized this list because it includes rules rejected by the EPA or reviewing courts, makes somewhat arbitrary cost and risk adjustments, and fails to make clear the contestable value judgments that underlie the discounted cost of human lives.⁸ For our purposes, it is enough to note that there is wide agreement about the inconsistency and arbitrariness of many of the assumptions underlying risk regulation.

  Justice Breyer identified three primary problems with risk regulation as currently practiced. First, agencies are afflicted with "tunnel vision," ⁹ pursuing their mandates without perspective. As a result one branch of the EPA may stamp out a small risk of water pollution, at the expense of generating a greater air pollution risk. Second, there is no central coordination, either in Congress or elsewhere.¹⁰ Third, the various agencies pursue different substantive mandates and different methodologies for calculating risk.¹¹

  He offers three explanations for these problems. First, inaccurate public perceptions lead policymakers and their constituents to give a high priority to correction of relatively small or even nonexistent risks.¹² Second, the institutional structure of Congress promotes inconsistency.¹³ Different committees originate legislation, and all respond to the daily news, a source that focuses on the dramatic and wholly overlooks the nuances of risk. No institution operates to counteract the centrifugal tendencies. Third, there is considerable inherent uncertainty in assessing environmental risks.¹⁴

  A brief digression on the measurement of environmental risk is in order. Agencies such as the EPA—driven by the inability to secure meaningful epidemiological data—rely on rodent tests, inferring human responses at very low doses from rodent responses at very high doses. The method of extrapolation almost invariably assumes a linear dose-response curve. But rodents and humans differ in ways that appear relevant to the likelihood of carcinogenicity.¹⁵ And even if a linear dose–response curve were accurate within the relevant range of plausible human exposures, experimental doses—the highest that can be administered without killing the rodents from sheer toxicity—appear to work quite differently. Huge doses of substances that are not themselves genotoxic—i.e., DNA injuring—are nonetheless accompanied by toxicity, cell death, and cell replacement. These conditions are favorable for tumor growth. But at doses at which cellular death does not occur, these substances do not promote tumor growth.¹⁶ And because the majority of chemicals neither are genotoxic nor give rise to genotoxic substances when metabolized, linear extrapolation is often unsound.¹⁷ An analogy illuminates the problem. Suppose that 1000 fires in a city in a 24-hour period cause 10,000 deaths; would it be proper to infer that one fire will cause 10 deaths? The answer seems clearly "No." A thousand fires would overwhelm a city fire department, greatly increasing the risk from each fire. Without such a catastrophe, the fire department could readily respond and the likely losses per fire would be far fewer.

  Justice Breyer's proposed solution was the creation of a group of highly skilled civil servants, at home in worlds of science, law, policy, and politics, modeled after the French Conseil d'Etat: Renaissance men and women—rather like Justice Breyer—who would coordinate regulatory decisions. This "centralized administrative group" would take the form of a highly beefed up version of the regulatory analysis office in the Office of Management and Budget (OMB).¹⁸

  Although Justice Breyer does not explicitly advocate the use of cost–benefit analysis (CBA), it is certainly implicit in his message that his central group would engage in some form of expert ranking of risks; in doing so it would compare various hazards, at least paying serious attention to the costs of removing them. Presumably the quest for sound priorities would take at least two forms: the office would (1) encourage agencies to give priority to regulations that promised a favorable net return; and (2) call for a legislative remedy when it found potential subjects of sensible regulation not covered by any agency mandate (or agencies with mandates covering no promising field).¹⁹ In pursuit of the first objective, the group would be empowered "to create priorities within as well as among programs"²⁰ in particular agencies and to transfer resources between different agencies to overcome resource misallocation. ²¹ And as for the second, Justice Breyer anticipated that the central administrative group's technical successes would lead congressional committees to ask its advice in drafting legislation. ²²

  In a review of Breyer's book, I noted the comment of a friend that "Steve thinks the solution to most problems is to turn them over to a room full of people as smart as he is." ²³ I questioned whether such a group would be able to overcome the various political forces tending to distort any sort of government risk control program.

  While I still think that that analysis is sound, it strikes me that the feasibility of Justice Breyer's solution could be greatly enhanced if there were a clear consensus behind his background policy assumptions about the relevance of costs. Justice Breyer clearly believes that cost matters. If cost were irrelevant, there would be no absurdity in making a childless tract safe for children who might eat dirt 245 days a year—or, for that matter, 365 days a year. Justice Breyer's proposals would accomplish little unless regulators seriously considered costs, not only in setting priorities but in deciding on levels of stringency. Yet, it is common to find strong claims that any consideration of cost is wrong, perhaps immoral. ²⁴

• Considering Costs, and the Common Objections
  I shall try here simply to explore a possible consensus for taking cost into account. ²⁵ A good starting place is a statement of Ben Franklin, who, writing to a friend who was perplexed by a difficult decision, explained his own approach:

When those difficult cases occur, they are difficult, chiefly because while we have them under consideration, all the reasons pro and con are not present to the mind at the same time… To get over this, my way is to divide half a sheet of paper by a line into two columns; writing over the one Pro, and over the other Con. Then, during three or four days consideration, I put down under the different heads short hints of the different motives, that at different times occur to me, for or against the meas-ure. When I have thus got them all together in one view, I endeavor to estimate their respective weights… And, though the weight of reasons cannot be taken with the precision of algebraic quantities, yet when each is thus considered, separately and comparatively, and the whole lies before me, I think I can judge better, and am less liable to make a rash step, and in fact I have found great advantage from this kind of equation, in what may be called moral or prudential algebra. ²⁶

The soundness of weighing competing considerations seems generally accepted. How can that acceptance coexist with vehement denunciations of any consideration of cost in the control of environmental risk? In this speech I'll offer some thoughts.

But first a few caveats. I will not try to address formal cost-benefit analysis as conventionally defined, with its understanding that benefits and costs must be computed by reference to the affected parties' willingness to pay or willingness to accept compensation.²⁷ The "cost consideration" of which I will speak is more general, and as a result is too vaguely conceived to operate as a universal tool for regulation of environmental risk. Thus, I will not be concerned with academic critiques of cost-benefit analysis that depend on the conventional assumptions, such as those arising from the Scitovsky Paradox or the relation of ordinal versus cardinal utilities. ²⁸

Second, by the same token I will not address a recent argument that despite such defects, the use of cost–benefit analysis is justified by the discipline it can impose on public decisionmaking. ²⁹

Third, whatever the merits of my position here, it is not inconsistent with—and I intend no comment on—Congress making a judgment that an agency should for a particular pollutant or class of pollutants attain some specific standard without regard to cost. Thus, for example, I do not address Professor Sunstein's suggestion that in some contexts a "feasibility" standard may as a practical matter accomplish reasonable congressional goals better than CBA.³⁰

My concern here is what could be no more than the first step toward the sort of prioritizing analysis advocated by Justice Breyer—to address attacks on any consideration of cost as a constraint on the pursuit of stringency in risk regulation. Thus I state the attacks and endeavor to supply some answers.

A. It's Wrong to Consider Cost, Because Health is More Important than Wealth
It is true, of course, that health is a prerequisite to much enjoyment of wealth, and life is virtually an absolute prerequisite (if we disregard the ability to leave wealth to descendants or others).

The problem here is the failure to evaluate choices at the margin. It doesn't follow from the primacy of survival that any increment in chances for good health, taking the form of a reduction of one specified risk, is worth any sacrifice in wealth. People take business trips by plane and car that carry some clear risk of being killed; more generally, people go to work, accepting the risks of the commute (as against those of languishing at home), in exchange for all the advantages, both in income and personal satisfaction, that come from productive work. People enjoy risky recreation. And even in going to low-risk recreation, they cheerfully run the risks of driving and flying. Just as we would not expect much of a person who for fear of risk never got out of bed in the morning, so we can hardly expect much of a society that stultifies itself in the name of health.

A variant on this is the claim of incommensurability—the claim that matters such as health belong to a completely different dimension than matters of wealth, so that there exists no way to make trade–offs between the two. But so formulated the claim is belied by the everyday individual choices of the sort discussed above.³¹

B. Consideration of Cost is Wrong Where Regulation Addresses Risks that are Imposed on People Involuntarily
There is some ambiguity in the concept of involuntariness. What if someone takes a job with risks known to him, and receives a compensating pay differential? Is that involuntary? What if someone lives downwind of a coal-fired power plant, and enjoys lower rent as a result, but might get purer air elsewhere, at a higher rent? Cass Sunstein's response to the issue of "involuntariness" is to say that it may be reasonable to make special adjustment for risks that are especially costly to avoid.³²

• But I think there are some additional answers (perhaps implicit in Sun-stein's analysis). So I will assume that we can identify some risks as truly involuntary-ones that at least for some individuals are so costly to avoid that we would think it unjust to put a person to the choice. That inclination can be resisted, I believe, on the ground that a prudential approach to risk control has favorable net pay-offs even for those exposed to "involuntary" risk.

  First case. For some specific risks, the benefits of suffering exposure to the risk, even for those most at risk, exceed the probable costs. Suppose I live under a transcontinental flight path. There is a risk-minuscule, but not zero—that a plane will crash into my house while I or my loved ones are inside. I would submit that such a person is better off exposed to the risk, but compensated not only by lower land costs but also by his share of the benefits of a world allowing air transport (rather than forbidding it on grounds of risk exposure): his own travel, and—for one who never flies—lower costs and greater efficiencies for the businesses whose goods and services he uses, not to mention government officials whose mobility and effectiveness may be enhanced.

  Second case. For some risks, the persons most exposed may be subject to expected costs greater than the expected benefits. Suppose, for example, that I live downwind of dust stirred up by mining and transportation of coal—to a degree not matched, for me, by the economies in the cost of electric power (or whatever products the coal is used for) that result from the absence of greater stringency. But under a regime that considers costs, the person most exposed to a specific risk of this sort is simultaneously the beneficiary of cost savings that result from the state's refraining from total eradication of the risks implicit in thousands of other activities. He might be better off in a specific context if the state set out to eliminate the risks local to him, but a system of universal risk eradication would subject him to very high costs. It seems unlikely that there will be any net losers in a system aimed at eliminating only the risks that can be removed at a cost lower than the resulting benefits.

  Of course, it's important to think of these things ex ante and not ex post. Presumably a person killed by a falling plane could never be described as a net beneficiary of that flight. But at the moment we adopt a rule allowing flight, despite its perils to involuntary victims, he is the prospective recipient of a probability distribution of benefits from a system that allows aviation to proceed, and the prospective recipient of a probability distribution of disasters. I would submit—pursuing the airplane example—that all of us, including those who never fly, are in this sense net beneficiaries of the rule allowing airplane flight. And even when we find a citizen like the coal mine neighbor who is adversely affected (ex ante) by a specific class of risks, he seems almost certain to become a net winner from a system of risk control that in general balances cost against benefit.

  But, of course, if it can be shown that there are specific cases of persons involuntarily exposed to a risk such that they are net ex ante losers from a general system in which cost is balanced against benefit (vis-à-vis a general system of more aggressive risk control), Professor Sunstein's proposal to make an exception is a solution. ³³

  Theories relying on involuntariness, along with those of incommensurability, can be reformulated as theories of rights—an argument that people have a moral right to be free of involuntary risky exposures. But the source of any such right seems obscure. People who by their lives demonstrate that they make trade–offs between specific health risks and competing advantages—and this appears true of everyone not sunk in Oblomovian immobility—seem ill-positioned to advocate any such "right." The qualification suggested above-more stringent standards to protect against risks suffered by persons who lose on net from a scheme where stringency of regulation is generally constrained by cost—seems generally adequate to protect the broadest right one could reasonably support.

  It is true that even a normal, active person chooses the health risks he will take, whereas issues of risk regulation are (by definition) resolved collectively. But this is hardly an argument for constructing a right to cost–indifferent regulation, which would simply constitute another collective resolution of the issue, equally denying him any choice. Indeed, risk regulation occurs primarily in areas where individual choices are hard or impossible to implement; there is no way to allow individual trade-offs between cost and ambient air quality. Given the general practice of individuals, a right to perfectly clean air seems unlikely to be a close surrogate for the choices individuals would actually make.

  Professor Adler's discussion of my analysis is puzzling.³⁴ He notes that a universal rule favoring the interests of redheads would fairly clearly be more advantageous to redheads than a general CBA rule. ³⁵ But, acknowledging a few of the myriad objections to the pro-redhead rule, ³⁶ he moves to a broader objection to ex ante evaluation of rules, noting that a rule, policy or choice rationally chosen as preferable may turn out to have worse effects on the chooser (or a portion of the subject population) than a rationally rejected rule, policy or choice; Eric may rationally choose to cross a bridge and yet fall to his doom when his crossing triggers the bridge's collapse. ³⁷ Of course a choice may have such results; the difference between ex ante and ex post is that the ex post world involves the outcomes from the choices made earlier. But Professor Adler's analysis (on this issue) no more shows a deficiency in CBA than the occasional auto accident shows that citizens who drive, after comparing the relative costs and benefits of driving and not driving, have chosen irrationally or unwisely.

• C. Distorting Effects of Wealth
  In classical cost-benefit analysis, an effect on a rich person may get what seems like improper weight compared with effects on a poor man. This differential may be morally offensive. I vividly recall the outrage of a friend who had studied the cost-benefit analysis of a proposed third London airport, in which estimates of the benefits of speedier travel reflected the demand of relatively wealthy travelers, while the low values placed on the noise and disruption of facilitating that travel similarly reflected the demand—a function in part of wealth—of the poor working stiffs in the flight path.

• One answer of course is that life in a market economy is subject to the same skew: a rich man can hire a poor man to clear his land of hazardous materials. But of course in that transaction the poor man is paid the market rate for his risk–taking; whereas our system of environmental regulation gives no explicit compensation to those bearing the relatively larger residual risks.

• The answer seems again to lie in a focus on systems. A system of risk–control measures that generally pass a rough cost-benefit test seems likely to leave poor individuals far better off than one in which government pursues zero risk regardless of cost. Thus, a CBA system as a whole should provide implicit in-kind compensation.

• Moreover, the net advantage of regulation guided by CBA, compared with cost-indifferent risk control, may flow more to the poor than to the rich; increases in the price of goods and services fall more heavily on the poor. Even so, of course, it remains possible that the advantage of CBA–guided regulation in absolute value favors the rich—they may gain more from the choice of CBA over cost-indifference than do the poor. Thus CBA, though favorable to all income classes, might increase income inequality. But if there is such an increase in inequality, and if it is troubling on ethical or political grounds, it can be offset by side payments.

• Similarly, suppose that in some cases wealth discrepancies have the effect of justifying a proposed regulation that is less stringent than what would emerge if some uniform wealth assumption were imputed to all affected parties, and also that this diminished stringency is believed offensive. The principle that one should reject restrictions whose costs exceed their benefits does not require that cost and benefit be measured exclusively by affected parties' willingness to pay (or to accept) on the basis of their existing incomes. To be sure, adjustments may be arbitrary, but formulae can be devised that roughly accommodate the concern.
  
  D. The Value of Life and the Discounting of Life
  Valuing human life is at best a discomforting exercise. All the more so when the values are calculated by methods that give controlling effect to willingness to pay or willingness to accept risk in exchange for compensation—both functions of wealth. As suggested above, adjustments for wealth can be used to palliate concern on that score.

• But for my limited purposes the broad issue of our queasiness over valuing life can be left up in the air. Whatever its resolution, there remains—once the principle of considering cost is accepted—a value that must be assigned human life in the name of preserving human life. Where regulatory requirements or attendant price increases reduce access to risk-reducing goods such as safe (and weighty) cars, or to more nutritious and healthy foods, or to health services themselves, to a degree that adds more danger than the risk-reduction effort eliminates, ³⁸ clearly cost-disregarding risk control has gone too far.

• In regard to discounting for futurity, Professor Revesz has argued that conventional cost–benefit justifications do not reach cases that involve losses to be suffered by future generations, which he sees as posing distributional issues different from those within a generation.³⁹ But for my limited purposes we may be agnostic on his argument, as he embraces intergenerational opportunity cost as a proper constraint: "Even if the objective were to transfer resources to a future generation, it might nonetheless be preferable to leave the problem unattended if alternative investments would yield a higher rate of return." ⁴⁰
  
  E. Precautionary Principle
  There are a variety of formulations of this principle, all loosely purporting to be applications of a bunch of appealing adages: "better safe than sorry," "an ounce of prevention is worth a pound of cure," or "look before you leap."
  
  Some formulations are perfectly sound—simply variations on the truth that decisions must be made before certain scientific knowledge is obtained. But that hardly captures the flavor of the "principle." Here is a quotation from an enthusiast—by no means the most starry-eyed:

• 
  [The precautionary principle] requires reduction and prevention of environmental impacts irrespective of the existence of risks…[T]he crucial point is that environmental impacts are reduced or prevented even before the threshold of risks is reached. This means that precautionary action must be taken to ensure that the loading capac-ity of the environment is not exhausted, and it also requires action even if risks are not yet certain but only probable, or, even less, not excluded. ⁴¹

First, the adage that it's better to be safe than sorry assumes that the speaker will pursue safety only up to a reasonable point, not that he'll become paranoid and refuse to get out of bed. That a stitch in time saves nine hardly proves we should take a hundred thousand stitches to save nine. And five tons of prevention are not necessarily worth a pound of cure. In some areas application of the principle would be transparently unhealthy. In the FDA drug approval process, for example, applicants can never demonstrate a drug's safety and efficacy with certainty; imposing such a requirement would mean doing without drugs altogether.

• Second, the strongest argument for the precautionary principle—that some risks are so huge that they absolutely must not be run—becomes meaningless unless we are ready to consider probabilities. Consider genetic food modifications. The argument against them often takes the form of saying that they may be like The Andromeda Strain—once loose, they may extinguish life on earth.

• The trouble is that when such scenarios are unmoored from probabilities, they can be matched by equally frightening counter-scenarios. Failure to use genetic modifications could lead to food shortages, leading to popular desperation, to a seizure of power by someone of Hitler's ilk, and ultimately to nuclear war, bringing on nuclear winter. Or it could lead to increased use of the rain forest, causing the extinction of a species that could produce the antidote to a disease that would otherwise devastate humanity. Unless you look at the probabilities on each side, doomsday scenarios are meaningless. ⁴²

• More broadly, the precautionary principle seems to reflect so much anxiety about novelty that it invites governmental policies that would freeze society and disable the most powerful engine of response to changed conditions: the existence of decentralized capacity to research and develop innovations, on at least such scale as markets will support. Such a freeze would, I suspect, do no more good for us than an approach paralleling China's decision in the 15th century to eliminate its fleet and to circle the wagons around the status quo of its native culture, then effectively dominant over all known alternatives.⁴³
  
  F. Polluters Will Cook the Books
  Obviously in a command-and-control system polluters have every incentive to submit inflated cost estimates. But in such a system those who make their living selling pollution control equipment have an offsetting interest in minimizing cost estimates. Furthermore, what do we pay agency officials for? If polluters can easily and systematically bamboozle them, we had better rethink the system more fundamentally.
  
  G. Numbers Win—and Unfairly
  "All in all, there can be little doubt that numerical precision is often mistaken for accuracy and certainty." ⁴⁴ There is obviously some truth in this observation of Professor Heinzerling. But the argument reminds me of an anecdote that Lévi-Strauss tells in Tristes Tropiques. He talks of visiting some tribe and handing a book or paper to a tribal chieftain. The chieftain, completely illiterate, starts pretending to comprehend the writing, and uses it to spout authoritatively to his tribesmen. Lévi–Strauss observes that the chieftain has instinctively grasped that one function of writing, an important function, is to lay claim to power and prestige.
  
  The chieftain's impulse was right, and Professor Heinzerling is right in a sense: some people do get unduly mesmerized by numbers. But if you accept the basic Ben Franklin preference for net benefit, then you must in some way consider costs and compare them with benefits; that's the only way you can get to net benefit. Sure, there is a risk that some unsophisticated people will be fooled by numbers—as indeed they will be fooled by all kinds of rhetoric and demagoguery. But that risk does not make out a case for dispensing with cost considerations any more than Lévi-Strauss's story makes out a case for illiteracy.
  
  So, to return to Justice Breyer: I've been trying simply to explore the intellectual feasibility of the sort of consensus that his proposals implicitly require. Ben Franklin's "prudential algebra" may not be a bad place to start.

*Stephen Williams is a judge on the United States Court of Appeals of the District of Columbia Circuit. This is the revised text of an address before the District of Columbia Bar at its Harold Leventhal Lecture Series, held in Washington, DC on June 13, 2000.