May 5, 2003 Office of Information and Regulatory Affairs NEOB, Room 10235 1725 17th Street, N.W. Washington, D.C. 20503 Re: OMB’s Draft Guidelines for the Conduct of Regulatory Analysis To Whom it May Concern: OIRA’s proposed guidelines place cost-benefit analysis at the heart of regulatory decision-making, which we believe is deeply misguided. Cost-benefit can be a useful tool, but it should not replace good judgement, common sense, and congressional mandates for regulatory decision-making, which frequently place human life as the preeminent concern, ahead of costs. OIRA’s guidelines (and perhaps equally important, OIRA’s commitment to cost-benefit analysis as the ideal decision-making tool) threaten to do all these things, wrongly making cost-benefit analysis determinative. The phrase “cost-benefit analysis” conjures the image of even-handed, dispassionate decision-making. Yet in the regulatory context, such analysis demands the monetization of benefits, such as the saving of human life. As we discuss below, the analytical methods underlying monetization are value laden, and in the case of OIRA’s proposed guidelines, would decidedly tilt decision-making in favor of inaction. In the process, transparency is sacrificed as artificial, make-believe dollars mask real-life choices and real-life benefits that should be the focus of public debate. Presenting Costs and Benefits in Plain English Above all, regulatory analysis should be useful to decision-makers and the public. OIRA’s proposed guidelines emphasize monetization of costs and benefits, as discussed in detail below. But we believe they should also address in greater detail the presentation of costs and benefits in layman’s terms, or plain English, absent monetization. In our experience, agencies frequently do not present this information in a clear and concise way in a single place. Information on benefits is often scattered about in supporting documents, and not effectively summarized up front in the regulatory proposal. OIRA should direct agencies to state expected benefits clearly in a non-technical way. What is the nature of the problem being addressed? What are the expected health, safety, or environmental benefits? How many lives are expected to be saved, and over what period of time? How much disease or injury is expected to be avoided? What are the ecological benefits of action? In many cases, agencies may not be able to generate specific answers to these questions, which should not preclude action (see the discussion of “Emerging Risk” in our comments on OIRA’s report on the costs and benefits of federal regulation). Agencies should clearly explain any uncertainty associated with such estimates – identifying the basis for their conclusions – and qualitatively describe other likely benefits that cannot be quantified or for which quantitative estimates are not available. Agencies might say they already do this, but the presentation is frequently lacking and confusing to say the least. OIRA needs to press agencies for greater clarity in describing benefits, keeping in mind the public and decision-makers who may lack the time and technical expertise to unravel all the information that flows from a rulemaking. As for costs, agencies should clearly state what expenditures are expected, and by whom? Rather than simply give an aggregate amount, agencies should discuss how costs are spread out over affected parties, including any costs that might be passed on to the public. For private business, agencies should also report costs as a percentage of revenue if possible. This places estimates in a more useful context for decision-makers. Again, in some cases, agencies might say they do this, but frequently this information is not presented in a clear, concise way in a single place. Monetization & “Net Benefits” Decision-Making Frequently, benefits prove extremely difficult or even impossible to monetize, which can skew cost-benefit analysis to favor inaction. For instance, EPA recently proposed a rule to protect the trillions of fish and aquatic organisms that are sucked up and killed each year by power plants, which use rivers, estuaries, and oceans to cool their systems. In performing its cost-benefit analysis, EPA did not monetize losses of invertebrate species, such as lobsters, crabs, and shrimp, as well as endangered or threatened species, nor did it consider the interrelationships of the species affected. Rather, EPA’s estimate was based exclusively on the commercial value of the fish that would have been caught had they not already been killed by power plants. This accounts for less than 20 percent of the total fish killed by cooling systems. EPA acknowledged the problems with its analysis, and used the non-monetized benefits to argue for a relatively protective standard, which it submitted to OIRA for review on Sept. 10, 2001. During its review, however, OIRA forced EPA to adopt a less protective option (published in the Federal Register on April 9, 2002) that showed fewer benefits, but greater “net benefits” by EPA’s estimates, according to agency documents. This meant the qualitative benefits – because they could not be monetized – were essentially ignored. OIRA’s proposed guidelines seem to enshrine this dismissive treatment of non-quantifiable factors: “Non-quantifiable benefits or costs may be important in tipping an analysis one way or the other, but you should not use non-quantifiables as ‘trump cards,’ especially in cases where the measured net benefits overwhelmingly favor a particular alternative.” OIRA should be clearer about exactly what this means. What is meant by “tipping” or “trump cards”? When specifically can non-quantifiables come into play? OIRA seems to be saying that an agency must show “net benefits” before non-quantifiables can even be considered – meaning that agencies better monetize if they want it counted. Non-quantifiables can then be used to “tip” one regulatory option over another, as long as they both show “net benefits” that are relatively similar. Again, the problem here is that agencies are frequently unable to monetize a wide range of benefits. If you have identified a significant problem, the fact of that significant problem – and the urgency to address it – may take precedent over measuring its precise contours, which can take an enormous amount of time and resources, and may be impossible in any case. Consider EPA’s proposed rule to control runoff from construction and development (published in the Federal Register June 24, 2002), which is the largest source of pollution in coastal waters and estuaries in the United States. In its original submission to OIRA, EPA pointed out that it was unable to monetize what it considered substantial benefits, including effects on natural habitat, benefits to human health, and impacts of many storm water pollutants, such as lead, zinc, herbicides and pesticides, as well as oils and grease. Yet OIRA acted as if these benefits didn’t exist since they weren’t monetized, and according to agency documents, forced EPA to remove any reference to permanent controls for after construction, which was the most significant part of the agency’s submission. We believe OIRA has a burden to explain why such non-monetized benefits should not be considered. These likely benefits don’t just disappear because the agency was unable to monetize. They are still there. OIRA is simply choosing to ignore them. The question is, why? Does OIRA believe that non-quantifiable benefits are not worth considering without first showing monetized “net benefits”? Or in other words, does OIRA believe that non-quantifiables cannot be used to demonstrate a showing of net benefits? If this were the case, it would be a troubling departure from past practice. The current guidelines, for instance, state, “we recognize that monetizing some of the effects of regulations is difficult, if not impossible.” OIRA needs a more substantial discussion of non-quantifiables to clarify its views, and should acknowledge that non-quantifiable factors are frequently crucial in the consideration of benefits. Statutory Requirements Congress has long recognized the deficiencies of cost-benefit analysis. Major health and safety statutes, such as the Clean Air Act and the Occupational Safety and Health Act, prohibit its use to determine a standard. Instead, most environmental regulation is “technology based,” requiring the best available methods for controlling pollution. This approach has allowed environmental protection to move forward, avoiding the trap of paralysis by analysis. Similarly, for OSHA, the health and safety of workers is supposed to be the preeminent concern. In these cases, Congress was clearly more interested in solving what it saw as serious health and safety problems than in the costs that might be required of responsible parties. OIRA may disagree with this reasoning, but it is the law and must be followed. Unfortunately, OIRA’s proposed guidelines seem designed to skirt these congressional directives, and wherever possible base decisions on “net benefits” determinations (monetized benefits minus costs). Ostensibly, this decision-making framework puts cost considerations on an even par with benefits. Yet given severe limitations in monetizing benefits (as discussed above), costs are in fact given considerably more weight than benefits, which would turn the Clean Air Act and the OSH Act, among others, on their heads. In fact, both EPA proposals discussed above were supposed to be based on the best technology available. Yet this did not stop OIRA from forcing changes based on “net benefits” calculations – in apparent violation of the law. Likewise, OIRA’s guidelines give short shrift to statutory mandates, which must guide agency decision-making. OIRA should make clear that agencies should base rulemaking decisions on their statutory requirements where they exist. Under such circumstances, the analysis performed under OIRA’s cost-benefit guidelines should be purely informational, and not determinative. To pretend otherwise is to flout the law. Yet considering the overall thrust of the guidelines, OIRA clearly has more than information in mind. Indeed, in its limited comments on statutory requirements, OIRA seems to be feeling around for wiggle room: “If your regulatory intervention results from a statutory or judicial directive, you should describe the specific authority for your action, the extent of discretion available to you, and the regulatory instruments you might use.” Later, OIRA writes, “You should also discuss the statutory requirements that affect the selection of regulatory approaches. If legal constraints prevent the selection of a regulatory action that best satisfies the philosophy and principles of Executive Order 12866, you should identify these constraints and estimate their opportunity cost.” OIRA should forthrightly acknowledge that Congress has determined that cost-benefit analysis is not the appropriate decision-making tool for most health, safety, and environmental protections. Discounting Lives Saved in the Future OIRA’s guidelines ask that agencies use two separate discount rates of 7 percent and 3 percent. OIRA should also ask that agencies provide estimates without any discounting of lives saved in the future, which would be simple to do and require no new resources. This would allow the public and policy-makers to observe the implications of discounting, and decide for themselves whether it is appropriate in a particular case. Indeed, there are many who reject the practice of discounting human life, including OMB Watch, for reasons best described by Georgetown Law Professor Lisa Heinzerling and Tufts economist Frank Ackerman in their recent booklet, “Pricing the Priceless.” We would be more interested in analysis that did not include it. We understand that OIRA believes otherwise, and that can be stated. But this should not deter a fuller presentation of information, which in the process, can further the interest of democratic decision-making. This is, after all, presumably the reason for OIRA’s proposed guidelines. Cost-Effectiveness Analysis for Health & Safety Standards The current guidelines say that agencies may use cost-effectiveness analysis (the ratio of costs to units of benefits, such as number of lives saved) in place of a “net benefits” analysis if they have difficulty monetizing. The new proposed guidelines, on the other hand, require both types of analyses for all major health and safety rules. Cost-effectiveness analysis avoids some of the problems of monetization of benefits, but nonetheless, it too can lead to skewed and timid decision-making. For example, a cost-effectiveness analysis that looks at costs relative to the number of lives saved would miss a whole slew of other significant benefits, such as non-fatal disease or injury, effects on ecosystems, and equity considerations. Moreover, the least protective regulatory alternatives are frequently estimated to be the most cost-effective. This is because additional levels of protection are forecast to require increasingly demanding and more costly methods. Forcing decisions based on a cost-effectiveness test may lead an agency to inappropriately choose a less protective alternative – because it is the most “cost-effective.” For these reasons, cost-effectiveness analysis is inadequate as the basis for regulatory decision-making. Like cost-benefit analysis, it should be seen as a tool, not the foundation, for regulatory decision-making – especially given the explicit, and frequently contradictory, instructions of many underlying statutes. OIRA should acknowledge this and discuss the limitations of cost-effectiveness analysis. In addition, OIRA’s proposed guidelines require agencies to incorporate discounting in cost-effectiveness analysis, which makes it appear less cost-effective to save lives in the future. As Heinzerling explained in testimony on March 11 before the House Subcommittee on Regulatory Affairs, “Suppose EPA proposed a regulation that would save 100 people from a type of cancer that has a latency period of 20 years ... Through the ‘magic’ of discounting at OMB's preferred rate of 7 percent, these 100 lives would be converted to 25.84 lives.” This is worse than discounting the dollar value of a “statistical life” because it reports expected benefits in a way that appears factually inaccurate. Under this scenario, three-quarters of the future lives saved are simply swept away. Yet that doesn’t change the fact that these “statistical” people will still die if action is not taken. To avoid confusion, OIRA should not require agencies to discount for cost-effectiveness analysis. Of course, the latency period can and should be identified, but OIRA should aim to make analysis understandable to the public and decision-makers. At best, discounting units of benefits makes such analysis less transparent, and at worst, it is deeply misleading. Use of “Life Years” in Evaluating Fatality Benefits Agencies commonly base benefit estimates on the “value of a statistical life” (VSL), drawn from the number of lives expected to be saved by regulatory action. However, OIRA has recently promoted the use of “value of statistical life years” (VSLY), which looks at the number of life years saved as opposed to the number of lives. OIRA’s proposed guidelines do not direct agencies to use VSL or VSLY. Instead, they say, “agencies should consider providing estimates of both VSL and VSLY, while recognizing the developing states of knowledge in this area.” The current guidelines also notes VSLY as a way to measure fatality benefits, but adds, “You should keep in mind that regulations with greater numbers of life-years extended are not necessarily better than regulations with fewer numbers of life-years extended. Longevity may be only one of a number of relevant considerations pertaining to the rule.” At the very least, OIRA should retain this note of caution in its new guidelines. Reliance on VSLY, in effect, rations our regulatory protections based on age, elevating life-saving measures aimed at the young over those that primarily protect the elderly, who have fewer life years remaining. Yet in reality, nothing is stopping the government from protecting both young and old; the tradeoff is unnecessary. Moreover, there is no good evidence that the elderly value their lives any less than the young, or are any less willing to pay for regulatory benefits – no matter how many life years they may have remaining. The same can be said for society as a whole, which generally recognizes a special obligation to our seniors. In this way, VSLY distorts society’s values, skewing against protections that primarily benefit the elderly. Additional Analysis for Billion Dollar Rules OIRA raises the bar for rules with “economic effects that exceed more than $1 billion per year,” requiring agencies to provide “a formal probabilistic analysis of the relevant uncertainties.” It is unclear why OIRA decided to propose this new, seemingly arbitrary threshold, along with this more demanding analysis of uncertainty. Why $1 billion? Is there a problem that’s being addressed? What are the benefits of such analysis? How will it improve decision-making? OIRA should not be pressing increasing levels of analytical rigor for its own sake. At some point the need to reach a decision and take action should take precedent, especially when lives are at stake. Such analysis could prove extremely resource intensive, causing further delays in an already ossified rulemaking process – which OIRA’s proposal would willingly tolerate. Health and safety could suffer as a result. We do not believe this additional analysis will improve decision-making in any real way, and considering its potential burdens, the requirement should be discarded. OIRA’s Request for Underlying Data & Comparison of Rulemakings In the president’s budget submission to Congress last year, the administration stresses the importance of “league tables” for setting regulatory priorities across federal agencies. These tables are intended to compare the costs and benefits of one type of regulation, such as auto safety, to another, such as environmental protection. In presenting a sample league table, the administration implies that we should contract efforts at environmental protection (e.g., health standards) because safety regulation (e.g., addressing accidents) is more cost-effective and produces greater “net benefits.” Yet this presents a false choice (leaving aside whether it is true), forcing an unnecessary tradeoff between one protection and another. In reality, with our $9 trillion economy, we can do both – and we do. There is no fixed budget for risk reduction. Perhaps OIRA has league tables in mind when it says, “It is difficult for OMB to draw meaningful cost-effectiveness comparisons between rulemakings that employ different cost-effectiveness measurements. As a result, agencies should provide OMB with the underlying data, including mortality and morbidity data, the age distribution of the affected population, and the severity and duration of disease conditions or trauma, so that OMB can make apples-to-apples comparisons between rulemakings that employ different measures.” OIRA should be more explicit about what it means here. For what purpose is OIRA making these comparisons? In what ways would OIRA manipulate underlying data? OIRA’s guidelines ask agencies for a number of different calculations (e.g., VSL, VSLY, two separate discount rates, cost-effectiveness analysis). What are OIRA’s preferences in doing such comparisons? Potential Technological Innovations by Regulated Entities Cost considerations are inherently easier to monetize than benefits. For example, they may involve purchases of new equipment or the hiring of additional personnel. Yet ironically, this does not mean cost estimates are any more accurate. Frequently, regulated entities are able to drive down compliance costs over time through technological advances or “learning by doing,” which are not typically predicted by cost-benefit analysis. As a result, agency cost estimates often prove overblown in the real world. In examining estimated costs next to actual costs for 13 major rules, economists Eban Goodstein and Hart Hodges found estimated costs were at least double the actual costs for all but one. For instance, EPA estimated in 1990 that acid rain controls would cost electrical utilities about $750 per ton of sulfur dioxide emissions; yet the actual cost turned out to be less than $100 per ton, billions of dollars less than what was initially anticipated. To its credit, OIRA seems to indicate that agencies should incorporate likely adaptive responses. If the purpose is to produce more accurate analysis, it seems obvious that agencies should incorporate what we know about the effects of technological advances and learning-by-doing. Specifically, OIRA’s proposed guidelines instruct, “Estimates of costs should be based on credible changes in technology over time,” adding, “regulatory performance standards and incentive-based policies may lead to cost-saving innovations that should be taken into account.” This is helpful, but OIRA should provide further discussion of what it has in mind, and perhaps present a few examples, illustrating what it expects agencies to do. Generally, agencies have not tried to predict or factor in likely technological advances. In fact, in discussing this matter in a previous report to Congress, OIRA explicitly discouraged this: “Although there are important cases of overestimating costs because of technological progress and learning-by-doing over time reduced expected costs, it is not clear that agencies should compensate for this tendency by reducing cost estimates.” Given its seeming change in mind, and the fact that this would be new for the agencies, OIRA needs to be more specific, which can help ensure this important point won’t just be ignored. Sincerely, Reece Rushing Policy Analyst, OMB Watch