to set the public agenda and to coopt public rescurces for the sake of undertaking a costly, complex analysis of existing rules.

This could be.a significant burden on agencies and hence on taxpayers. If people in the private sector can ask all agencies to undertake CBA under the new standards of S. 343, they will be able to use taxpayer resources for an analysis that might well serve simply to produce waste and delay.

To be sure, many agencies might respond to petitions simply by repeating their previous CBA's. But since S. 343 creates somewhat different standards from those set out in previous executive orders, some of those CBA's would have to be redone. Circumstances may well have changed in the interim. In light of the fact that S. 343 would apply to many hundreds and perhaps thousands of rules, the paperwork burden could be enormous.

A particular difficulty here is raised by the prospect of judicial review. If courts can review denial of petitions to review existing rules, there will be high litigation costs. This is especially true since section 625 would require an agency to decide whether "reasonable questions exist" about the CBA judgment, with judicial review of agency decisions on this point. In many cases, "reasonable questions" do "exist," and hence judicial involvement and management might well be common. Thus section 625 seems to me likely to fail cost-benefit analysis. It would require agencies to provide an analysis that they have probably already undertaken, and therefore it would do little good; but it would give affected persons a new and valuable weapon by which to drown administrators in paperwork.

I suggest that if section 625 is to be retained, it should not be judicially enforceable, and the executive branch should be given the authority to set priorities by reevaluating existing rules in accordance with its own assessment of whether they impose excessive costs. Of course Congress might well enact new legislation to correct old statutes, or old regulations, that fail cost-benefit analysis.

B. What is CBA? The Problem of Criteria Some form of CBA makes a great deal of sense for most regulatory decisions; but the second problem with S. 343 is that far more thought should be given to the definition of CBA in section 622. As it stands, the definition does not allow enough scope for values that are hard to quantify or monetize, and it is in some ways too open-ended.

More particularly, some regulatory statutes have social goals that fit uncertainly with the cost-benefit criteria now set out in section 622. For a measure with the enormous scope of S. 343, some nonmonetizable values should be acknowledged as relevant. To be sure, it is important to be precise and quantitative when this is possible, but sometimes this is not possible. Moreover, Congress might attempt to give more guidance for valuing life and health. To explain these complex points, I will have to deal with some

fairly technical material in a brief space; I end with some simple suggestions.

1. Background. Let me offer some general remarks by way of background. Cost-benefit analysis sounds simple and scientific, but actually it is a protean idea, one that is based on values, and not only on science. If CBA means an assessment of the advantages and disadvantages of action, no one could object to it. If it means that agencies should balance advantages and disadvantages, the problem with CBA is that it is vacuous or empty. Agencies can interpret it however they wish.

But CBA often means something more concrete, more technical, and more controversial: an effort to turn all consequences of government action into dollars or dollar equivalents, and to use "private willingness to pay" as the way of assessing both benefits and costs. On this view, for example, the benefit of clean air, or occupational safety, or helping the handicapped, or protecting endangered species, or reducing racial discrimination, would be measured by asking how much people would be willing to pay for these goods. On this view, we would measure the value of human life, or of endangered species, by seeing how much people would be willing to pay to avoid death or extinction. Economists have tried, for example, to measure the value of life by looking to labor market evidence, concluding that $5 million should be the reasonable "upper bound" on expenditures per life saved. See W. Kip Viscusi, Fatal Tradeoffs 265 (1993). There are advantages to efforts of this sort, but they raise many questions as well.

Presidents Reagan, Bush, and Clinton have been attentive to both the goals and limits of this more technical understanding of CBA Quantification and even monetization can be desirable as a way of simplifying the analysis and given officials real, "hard" figures with which to work. But sometimes the goods at stake cannot easily be monetized, and CBA has a spuriously scientific caste. This is partly because of enormous problems of measurement: With what evidence can we measure the value of greater visibility near the Grand Canyon, in light of the fact that survey evidence has enormous problems? For many regulations, CBA is inevitably based on controversial judgments about values, not on pure science. There is also a serious conceptual problem in claiming that all goods are properly valued along the same basic dimension of dollars or dollar equivalents.

The various goods at stake in regulation -- cleaner air, saved lives, reduced health risks, disemployment effects, inflationary effects, and sc forth are quite diverse. They differ not just in quantity but also in quality. They are valued in different ways, and efforts to reduce these goods to a single scale, especially the single scale of dollars, can impair rather than improve analysis. See Richard Pildes & Cass R. Sunstein, Reinventing the Regulatory State, 59 University of Chicago Law Review 1 (1995); Elizabeth Anderson, Value in Ethics and Economics (1992).

To say this is not at all to deny that government must make tradeoffs among different goods. Even if goods cannot all be understood on a single scale, or in terms of dollars, officials must often choose among them. Any choices should be reasonable, and based on reasons. What I am suggesting is that the process of choice can be harmed, not helped, by translating everything into the single metric of dollars at least if we are unaware of the limitations of the translation. It may therefore be best for agencies to try to do a technical CBA, using dollars or dollar equivalents, but then to look at the effects of regulation in a more qualitative way, and to allow the outcome of the CBA to be adjusted accordingly. (More details are offered below.) The decisional criteria of the bill should acknowledge this possibility.

The "willingness to pay" criterion has similar problems; I outline them briefly here. Sometimes people may be "willing to pay" for something that agencies should not count at all; malevolent people may be willing to pay to discriminate on the basis of race, to see a species die out, to see people be hurt. Often the question for a democratic society is not whether people are "willing to pay" for something, but instead whether they can offer good reasons for their position. Sometimes there is a large disparity between what people would be willing to pay for a good and how much they would willing to accept to sell the same good; the notion of CBA does not tell us which question to ask. See Richard Thaler, Quasi-Rational Economics (1992). This appears to be true for many of the goods at stake in regulation. See id.

Sometimes, moreover, people, in their capacity as citizens assessing government action, support outcomes that would diverge from what they would support through private willingness to pay, and it is by no means clear that the market judgment should be preferred to the democratic judgment. People may, for example, urge a ban on racial discrimination, or call for support for educational programming for children, even if their "private willingness to pay" for these goods is relatively small. Some statutes of course ban cost-benefit balancing, perhaps out of fear that agencies will "balance away" interests that are especially important. See, for example, the Endangered Species Act; the Occupational Safety and Health Act; the Delaney Clause. (Some of these "absolutist" provisions should undoubtedly be changed to allow consideration of cost; I do not discuss that controversial matter here. S. 323 would not affect laws that forbid CBA to be used as a basis for decision.*)

*Section 623(b) allows an agency to decline to make CBA the basis for decision when "explicit statutory language" so suggests. This provision may make sense, but it is important to point out that it is unprecedented. "Explicit statutory language" is not the only source of statutory meaning, courts look, of course, at statutory structure and history as well as applicable canons of construction. It would probably be best to change the "explicit statutory language" phrase to say more simply "except when such statute prohibits the consideration "If Congress would like to amend existing statutes forbidding CBA -and it should do so in several cases - the best route is to examine this issue in the context of the particular statute.

It is important, morever, to emphasize that good CBA recognizes that people care not simply about aggregate amount of lives lived, but also about a range of factors involving the nature of the particular risk. For most people, among the most salient contextual features are: (1) the catastrophic nature of the risk; (2) whether the risk is uncontrollable; (3) whether the risk involves irretrievable or permanent losses; (4) whether the risk is voluntarily incurred; (5) how equitably distributed the danger is or how concentrated on identiñable, innocent, or traditionally disadvantaged victims; (6) how well understood the risk in question is; (7) whether the risk would be faced by future generations; and (8) how familiar the risk is. An assessment of these variables is part of a good CBA. See generally Pildes & Sunstein, supra.

2. Suggested changes in section 625. Any required CBA for diverse regulatory agencies should reflect these points. It should recognize, more clearly than it now does, the limits of understanding everything in quantitative or monetary terms. Any CBA should, moreover, be accompanied by a more disaggregated and more qualitative description of the consequences of government action, so that Congress and the public can obtain a fuller picture than the crude and misleadingly precise "bottom line" of the CBA.

The statement and definition of CBA in section 622 should therefore be rethought, so as to emphasize that agencies can take account of considerations that Americans consider important -- including the eight factors listed above --as part of the cost-beneñt judgment. This is not at all to deny that it is important to be precise and quantitative when agencies can be precise and quantitative. It is only to say that any "bottom line" is likely to involve judgments about values, not about science, and Congress and the public should see what those judgments are.

A particular issue here,not taken up in S. 323, is how to value life or health. It may well make sense to set a benchmark standard of, say, $7 million per life saved as the maximum amount, with permission for an agency to select a higher number if the agency can explain that special circumstances call for that higher number. Without a figure per life saved, agencies effectively have discretion to weight costs and benefits however they wish. It is of course troublesome to assign dollar values for life, partly for the reasons I have sketched; but since tradeoffs are inevitable, it may be best for Congress to set out some guidelines governing expenditures, without pretending to say how much a life is "really worth." (I would be happy to offer more details about possible changes if the Subcommittee would be interested. I am restricting myself to very general suggestions.)

C. Judicial review. The third problem is that S. 343 would significantly confuse the law of judicial review of administrative action, and even worse, it would simultaneously threaten to increase judicial control of government in a quite undemocratic way.

35-770 96-4

1. Section 628 appears to be an effort to codify and slightly to adjust the standards governing judicial review of administrative interpretations of law. See Chevron USA v. NRDC, 467 US 837 (1984). I do not believe, however, that section 628, as written, would much change what courts are already doing; and because it is so long and complex, its principal effect would be to increasing the level of uncertainty in the law. In short, section 628 would not make things better; would mostly restate current practice; but would force lawyers and judges to spend their resources in coming to terms with a set of confusing new instructions. I suggest that section 628 should be deleted on the ground that it would fail CBA.

=

2. Section 624 would subject a range of new administrative decisions to judicial review. Presidents Reagan, Bush, and Clinton all decided to insulate CBA from the judiciary - on the theory that judicial review would produce delay, confusion, and error, especially in light of the judges' lack of democratic accountability or factfinding competence. This was a reasonable decision. At the very least, it certainly does not make sense to subject the CBA to judicial review when the agency's decision does not, under the relevant statute, depend on CBA.

I suggest that section 624 might be deleted, and that most or all agency decisions under section 622 might be insulated from judicial review. If judicial review of the substance of the CBA is to be made available -- and this would be a reasonable judgment too section 624 should be simplified to say that when an agency's decision turns on CBA under the applicable law, courts shall review that decision under the normal standards set out in 5. USC section 706. This seems to be the real intention of section 624(d), and that intention should be stated much more clearly.

3. As I have noted, judicial review of petitions to engage in costbenefit analysis could create enormous problems. At most, courts should review denials of such petitions for arbitrariness under the existing legal standard.

D. Unauthorized rules. Section 627 has a laudable purpose: To prevent the federal government from doing things that it may not do. But in its current form, section 627 is exceptionally confusing, and in all likelihood, it would accomplish far more harm than good.

The basic problem is that the section appears to be based on a misunderstanding of regulatory law. Agencies are generally given the power to issue "necessary or appropriate" rules to implement their statutory authority. Many of these rules involve internal agency management. Many of them are "implementing" rules that enforce the statute pursuant to a legislative mandate that does not specify any statutory minima. Many of them involve allocating licenses or holding auctions (consider the Federal Communications Commission) that are not, strictly speaking, legally required.