Baixe o app para aproveitar ainda mais
Prévia do material em texto
LIST OF CONTRIBUTORS USA Yoram Bauman W ashington State Department of Health, Seattle, WA, USA Carol Bray U .S. Government Accountability Office, Neil Bruce D epartment of Economics, University of Washington, Seattle, WA, USA Jared C. Carbone D epartment of Economics, Williams College, Williamstown, MA, USA Daniel H. Cole I ndiana University School of Law-Indianapolis, IN, USA Scott Farrow U MBC/Department of Economics, Baltimore, MD, USA Aaron Finkle D epartment of Economics, University of Washington, Seattle, WA, USA Victor B. Flatt A .L. O’ Quinn Chair in Environmental Law, University of Houston Law Center, Houston, TX, USA Lawrence H. Goulder D epartment of Economics, Stanford University, Stanford, CA, USA Benjamin Groom D epartment of Economics, SOAS, London, UK Tim Guinane U .S. Government Accountability Office, USA Robert Halvorsen D epartment of Economics, University of Washington, Seattle, WA, USA ix LIST OF CONTRIBUTORSx Arnold C. Harberger D epartment of Economics, University of California, Los Angeles, CA, USA Robert Haveman D epartment of Economics, University of Wisconsin-Madison, Madison, WI, USA Richard E. Just U niversity of Maryland, Agricultural and Resource Economics, College Park, MD, USA Jack L. Knetsch P rofessor Emeritus, Department of Economics, Simon Fraser University, Burnaby, British Columbia, Canada Andreas Kontoleon D epartment of Land Economy, University of Cambridge, Cambridge, UK Mark L. Plummer N ational Marine Fisheries Service, Seattle, WA, USA Theodore M. Porter D epartment of History, University of California, Los Angeles, CA, USA Gordon C. Rausser U niversity of California Berkeley, Agricultural and Resource Economics, CA, USA Frederick Rossi S outhern University at Shreveport, Shreveport, LA, USA Andrew Schmitz F ood and Resource Economics Department, University of Florida, Gainesville, FL, USA Troy G. Schmitz M orrison School of Agribusiness and Resource Management, Arizona State University, Mesa, AZ, USA V. Kerry Smith D epartment of Economics, Arizona State University, Tempe, AZ, USA List of Contributors xi Tim Swanson D epartment of Economics and Faculty of Law, University College London, London, UK Richard O. Zerbe Jr. S chool of Public Affairs, University of Washington, Seattle, WA, USA ACKNOWLEDGMENTS These chapters were produced with support from the Daniel J. Evans School of Public Affairs at the University of Washington, the Law School at the University of Washington, the Philosophy Department of the University of Washington, and The John D. and Catherine T. MacArthur Foundation. We thank them for their support. xiii INTRODUCTION TO BENEFIT–COST VOLUME In the earlier part of the twentieth century, cost–benefit (CBA) or benefit– cost analysis was used as a vehicle by Congress to curtail its wasteful spending, by using the Army Corp of Engineers to examine Congressional projects using CBA. Theodore Porter here examines the rise of the use of CBA in historical context and finds that the Corp was highly successful in reducing wasteful spending. Regardless of the present day effectiveness of the Corps, CBA currently provides valuable service. To appreciate this one need look no further than the effect Arnold Harberger’s work and students have had in less developed countries, and at the several hundred useful evaluations of social programs produced over the last several years. Finally, one can look, criticisms of Ackerman and Heinzerling notwithstanding, at many of the analyses of environmental programs. Nevertheless, problems remain. Robert Haveman shows how the effect of the absence of well-accepted and enforced benefit–cost standards leads to spurious analyses that result in public waste. Haveman examines the benefit–cost analysis performed by the City of Chicago for the FAA. The Chicago analysis results in positive net benefits of about 2.2 billion; Haveman’s analysis indicates a net social loss of around 4 billion. There are as yet no uniform standards for CBA analysis even within the Federal Government. Daniel Cole notes the limited success of OMB and EPA in improving methodological consistency. He suggests that an independent group of economists, policy analysts and legal scholars, under the auspices of the National Science Foundation, take on the job of establishing revisable ‘‘best practice’’ standards. Somewhat similarly, Victor Flatt argues that the courts can recognize the proper standard that governs what an agency can do. The courts can then set a standard for BCA that should be used when an agency is engaged in its application. Such an approach is consistent with the non-delegation doctrine, which suggests agencies should not be engaging in political manipulation in performing such analyses; political considerations are for Congress. Taken together, then, these two articles spell out an approach to increasing rigor and consistency. The article by Carol Bray, xv INTRODUCTION TO BENEFIT–COST VOLUMExvi Scott Farrow and Tim Guinane also falls within the purview comptemplated by these first two articles. The authors identify and review current uses of CBA measures within the Federal Government, and then discuss the opportunities for their expanded use and for the possibility of creating a set of uniform standards, which they call ‘‘generally accepted analytic principles.’’ The article in this volume by Larry Goulder considers another expansion of CBA application, a broader domain. Goulder advances well-reasoned arguments for a proposal by Zerbe to include the willingness to pay for outcomes by third parties. Goulder notes that, particularly in an international setting, third party effects might be significant enough to justify the costs associated with estimating them. This is consistent with the article by Richard Zerbe, Yoram Bauman and Aaron Finkle. They show that the objections made by economists to the inclusion of moral sentiments have little weight on either technical or utility grounds. There is an increasing interest in considering general equilibrium (many markets) rather than partial equilibrium (one or few markets) outcomes in CBA use. Arnold Harberger notes that he and Glenn Jenkins find that partial equilibrium analysis in analyzing the effects of trade are valid only when (1) the demand and supply for the good in question are not substantially affected by the way in which the project funds are raised, and (2) when there are no distortions (taxes and tariffs) involved in the raising of these funds or are taken into account elsewhere. Otherwise a general equilibrium analysis is required. This is not dissimilar from the results of V. Kerry Smith and Jared Carbone who show the importance of taking into account general equilibrium effects for the estimate of the efficiency costs of taxes in a distorted economy. They consider environmental amenities and find it is necessary to consider both market and non-market effects in conjunction. It turns out that even when such amenities are less that three percent of virtual income, the error from ignoring feedback effects between market and non- market effects can be very large. Stressing again the importance of general equilibrium outcomes, Richard Just and Gordon Rausser examine the difference between partial equilibrium and general results for vertical market structures. In particular they examine predatory selling and bidding. Their results demonstrate that the widely used partial equilibrium analysis is not robust for assessment of effects of monopoly or monopsony, whether or not predation occurs. In general monopoly and monopsony turn out to have asymmetrical effects Introduction to Benefit–Cost Volume xvii and the Supreme Court requirements of Brook Group for a finding of predation do not appear to accord with the general equilibrium results. Neil Bruce and Robert Halvorsen make another expansionary argument. In place of the usual ‘‘value of statistical life’’ (VSL), they develop the notion of ‘‘adding expected life years.’’ Such an approach can accommodate theeconomic result that an older person may evaluate an added year of life more than a young person and can obviate the tortured logic to which the VSL approach has been subject. They spell out the research that needs to be done to establish the use of ‘‘adding expected life years.’’ Andrew Schmitz, Frederick Rossi and Troy G. Schmitz examine the effects of trade-distorting agricultural policies. They address the question of what difference it makes to the results when producers respond to the government’s target price (coupling) when making production decisions or instead make decisions based on the loan rate (decoupling). The negative welfare effects are much less if the production decision is decoupled. More research is needed to determine how producers form production decisions and if there are policy changes that would support decoupling. Timothy Swanson, Ben Groom and Andreas Kontoleon assess the effect of information on absolute and relative ranking of the value of remote mountain lakes. Three levels of information were provided. The first two levels increased aggregate WTP only, but the last level also creates differences between the ranking of four remote mountain lakes. The third level of information, but not the first two, increased the congruence between the respondent’s valuations and that of experts. Thus this study also provides information about the level of information needed to get reasonable convergence. Information of course is the determinant to how much can be done with CBA. Mark Plummer describes how useful results can be provided by a hybrid of benefit–cost and cost-effectiveness analysis combined with a hypothetical valuation experiment. Market’s example of choosing critical habitat designation under the Endangered Species Act shows how a useful analysis can be made under data poor circumstances. The example provided is exemplary and valuable. Richard O. Zerbe Jr. Editor ‘BEST PRACTICE’ STANDARDS FOR REGULATORY BENEFIT–COST ANALYSIS Daniel H. Cole ABSTRACT Government agencies have endeavored, with limited success, to improve the methodological consistency of regulatory benefit–cost analysis (BCA). This paper recommends that an independent cohort of economists, policy analysts and legal scholars take on that task. Independently established ‘‘best practices’’ would have four positive effects: (1) they would render BCAs more regular in form and format and, thus, more readily assessable and replicable by social scientists; (2) improved consistency might marginally reduce political opposition to BCA as a policy tool; (3) politically-motivated, inter-agency methodological disputes might be avoided; and (4) an independent set of ‘‘best practices’’ would provide a sound, independent basis for judicial review of agency BCAs. 1. INTRODUCTION Benefit–cost analysis (BCA)1 is an inherently controversial practice, especially in the realm of regulatory policy. Like the Kaldor–Hicks (K-H) efficiency criteria upon which it is based, regulatory BCA yields results that Research in Law and Economics, Volume 23, 1–47 Copyright r 2007 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0193-5895/doi:10.1016/S0193-5895(07)23001-5 1 dx.doi.org/10.1016/S0193-5895(07)23001-5.3d DANIEL H. COLE2 are, at best, ambiguous with respect to social welfare and, at worst, subject to manipulation for political ends. As a consequence, the results of all BCAs are contestable. Unfortunately, no practical alternative exists for predicting and measuring the social-welfare outcomes of regulatory policies. In theory, the Pareto criterion yields unambiguous results, but it is ruled out in practice by its strict requirement of voluntary consent of all affected parties, which presumes actual compensation of losers at their subjective valuations of their losses. These strict conditions are met only by the negligibly small and uninteresting set of voluntary market exchanges with no significant negative externalities. In the realm of regulatory policy, resource allocations never are based on the unanimous consent of all affected parties; and the losers from regulatory policies never are compensated for losses based on their subjective valuations. So, the strict Pareto conditions cannot be met. Consequently, society is left with the ambiguous outcomes of imperfect and manipulable BCAs based on the K-H criteria (or marginally improved versions of K-H offered, for example, by Zerbe, 2001).2 The ambiguities and manipulability that render BCA generally contro- versial are exacerbated in the realm of regulatory policy by: long time horizons and uncertain social discount rates; the absence of primary or secondary markets for many environmental goods, which makes any prices assigned to them inherently contestable; and the virtual impossibility of assigning universally acceptable values to human lives. For these reasons, critics such as Ackerman and Heinzerling (2004) reject BCA entirely as a tool of regulatory policy making.3 Even if regulatory BCA never can be rendered completely uncontroversial, it can be made substantially less controversial than it is now simply by improving its methodological consistency. Even an admittedly imperfect methodology, applied consistently in accordance with standard criteria widely accepted among scholars and practitioners, should be less objection- able than various, inconsistently applied methodologies. This paper will demonstrate that the methodological consistency of regulatory BCA has been improving, though more in theory than in practice, over the last decade or so, and provide a few discrete suggestions for additional improvements. The ultimate goal – though not the goal of this particular paper – is to standardize a set of ‘‘best practices’’ for regulatory BCA. That goal, unfortunately, is unlikely to be achieved by the primary users of BCAs – the regulatory agencies themselves – because those agencies are subject to divergent political pressures that inevitably would raise questions about the motives behind any set of standards they might establish ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 3 and attempt to enforce. For reasons explained in Section 2, a set of ‘‘best practices’’ established by the Environmental Protection Agency (EPA) almost certainly would differ in important respects from a set of ‘‘best practices’’ established by the President’s Office of Management and Budget (OMB). The differential political biases of different agencies with different missions would likely undermine the benefits to be gained from adopting a set of ‘‘best practices.’’ Indeed, the EPA (2000) and OMB (2003) both have recently published ‘‘best practice’’ guidelines, which diverge in important respects. Multiple sets of ‘‘best practices’’ are better than none only to the extent that multiple sets of ‘‘best practice’’ standards might provide a starting point for conciliation and consolidation of a single set of standards. This paper recommends that an independent and heterodox cohort of economists, legal scholars, policy analysts, and decision theorists be appointed under auspices of an independent organization such as the National Academy of Sciences, American Economics Association, or a newly minted ‘‘Society for Benefit–Cost Analysis,’’ to derive a set of ‘‘best practices’’ for regulatory BCA. The hope is not that such a group would be able to derive completely objective and neutral standards for BCA; that would a pipe dream. BCA contains too many subjective elements, including social discount rates and valuations of non-market goods and services, ever to be completely objective and neutral. However, an independent cohort of experts is less likely to be swayed by the immediate political concerns that motivate the agencies that produce and utilize BCA. Independent economists, legal scholars and policy analysts presumably would have less of a political stake than OMB or EPA officials, for example, in the selection of a rate (or rangeof rates) for discounting future costs and benefits or pricing various mortality and morbidity effects. Assuming a cohort of experts could establish a legitimate and useful set of ‘‘best practices’’ for BCA,4 their product would not constitute a legal standard directly applicable to government agencies but a social–scientific standard or norm that agencies might choose to adopt and reviewing courts might choose to recognize. If some government agency based a policy decision on a BCA that did not conform to the ‘‘best practices,’’ without a satisfactory excuse or explanation, a reviewing court might reject the BCA and possibly the substantive policy it supported. In sum, the adoption of methodological ‘‘best practices’’ for regulatory BCA could have four positive effects: (1) it would render BCAs more regular in form and format and, thus, more readily assessable and replicable by social scientists; (2) it presumably would reduce, at the margins, political opposition to BCA as a policy tool; (3) it might reduce (though not DANIEL H. COLE4 eliminate) politically-motivated, inter-agency methodological disputes; and (4) it would provide a sound, consistent and independent basis for judicial review of agency BCAs, which is an increasingly important consideration given the proliferation of regulatory BCAs subject to review under state and federal administrative procedure statutes. This paper proceeds as follows: Section 2 offers evidence of improvements in the consistency of BCA methodology over the last decade or so. Many of those improvements have come from efforts to build consensus among government agencies, including the EPA and the OMB. However, Section 3 explains why more improvements are necessary (or, at least, desirable), but unlikely to come from the agencies themselves because of their divergent political and ideological commitments. Thus, this paper recommends that the American Economic Association, the Association of Environment and Resource Economists, the National Academy of Sciences, or a new ‘‘Society for Benefit–Cost Analysis’’ create a task force charged with developing a set of ‘‘best practice’’ standards for regulatory BCA. Section 4 then offers a tentative but necessarily incomplete list of issues such a task force might attempt to resolve in order to improve methodological consistency. The paper concludes by reiterating the benefits to be gained from the independent establishment of ‘‘best practice’’ standards for regulatory BCA. 2. THE STRUGGLE FOR METHODOLOGICAL CONSISTENCY IN BCA Regulatory BCA is a young and still developing method for predicting ex ante or assessing ex post the social welfare effects of public policies. Prior to the twentieth century, when federal government regulation of economic activity was a rare phenomenon, the lack of a regulatory BCA tool mattered little.5 But with the rise of the welfare/administrative state during the twentieth century – a rise enabled by theories of Pigovian welfare economics (see Adler & Posner, 1999, p. 169) – the total of amount of regulation at all levels of government increased dramatically. Still, before the 1970s government agencies only rarely attempted to quantify the costs and benefits of their burgeoning regulatory programs, let alone predict costs and benefits prior to policy implementation. Consequently, the government and the public had no way of knowing whether those regulatory programs were creating more social good than harm. ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 5 This problem called for a legislative fix that was not forthcoming. Until the mid-1990s, Congress displayed little interest in assessing formally the economic/welfare effects of regulatory programs.6 Consequently, nearly all efforts to assess the costs and benefits of agency regulations came from the Office of the President, which has the authority to regulate by Executive Order the activities of agencies within the Executive Branch of the federal government.7 2.1. 1970–1992: BCA under Reagan and Bush I BCA (often referred to within the federal government as ‘‘regulatory impact analysis’’) first emerged as a regular decision process tool at the beginning of the 1970s,8 when Presidents Nixon, Carter, and Ford issued successive executive orders calling for economic analyses of regulations and inter- agency review of major rules (see Morgenstern, 1997, p. 10). During this period, the President’s OMB emerged as ‘‘‘the lobby for economic efficiency’’’ in government (Morgenstern, 1997, p. 10, quoting Schultze, 1977). In the 1980s, the Reagan Administration fashioned a more significant role for BCA as a ‘‘decision rule’’ – a tool that, by itself, would determine whether or not a regulation should be promulgated (Morrison, 1998, p. 1333, n. 2). In 1982, President Reagan issued Executive Order (E.O.) 12,291 (46 Fed. Reg. 1193), which required that the ‘‘potential benefits’’ of all ‘‘major’’ federal regulatory actions (defined as those with economic effects of at least $100 million, significant employment effects, or significant price effects) had to ‘‘outweigh the potential cost to society.’’ The ostensible goal of the E.O. was to ‘‘maximize the net benefits to society’’ from regulation. In addition, the Order required comparative cost-effectiveness studies of alternative means of achieving regulatory goals: ‘‘Among alternative approaches to any given regulatory objective, the alternative involving the least net cost to society shall be chosen.’’ Finally, the Reagan E.O. placed the Director of the President’s OMB in charge of implementing the E.O., and compelled executive branch agencies to submit their regulatory impact analyses to OMB for review prior to final publication of new agency rules. Four years later, President Reagan issued another Executive Order, 12,498, which supplemented E.O. 12,291 by requiring regulatory agencies to submit ‘‘annual regulatory plans’’ for OMB review. According to Hahn and Sunstein (2002, p. 1506), E.O. 12,498 ‘‘increase[d] the authority of agency heads [political appointees] over their staffs [career bureaucrats], by exposing proposals to top-level review at an early stage... . [I]t also increased the DANIEL H. COLE6 authority of OMB, by allowing OMB supervision over basic plans, and by making it hard for agencies to proceed without OMB preclearance.’’9 Despite E.Os. 12,291 and 12,498, government agencies continued to dispute the proper role of, and approach to, BCA especially when the Executive Orders appeared to conflict with statutory mandates calling for regulations based not on economic costs and benefits but on the protection of public health (see Morgenstern, 1997, pp. 10–11). The decade from 1982 to 1991 (from Reagan through Bush I) was characterized by increasing OMB efforts (under authority granted by the two Reagan Executive Orders) to control regulatory processes and substantial regulatory agency resistance to OMB control. Despite E.O. 12,291’s seemingly clear mandate that agencies base their regulatory decisions on a calculation of costs and benefits, widespread disagreement persisted about the extent to which regulatory decisions should depend on BCAs. Analysts also disagreed about the appropriate methodology for conducting BCAs. Much of this disagreement, predictably, swirled around the inherently subjective elements of BCA, including discount rates and human-life valuations. At the time President Reagan issued his two Executive Orders on BCA, OMB was operating under Circular No. A-94, ‘‘Discount Rates to be Used in Evaluating Time-Distributed Costs and Benefits’’ (March 27, 1972). That Circular called for the use of a 10 percent real (i.e., inflation-adjusted) discount rate for assessing federal regulations. Interestingly, OMB did not revise that Circular in light of the Reagan Executive Orders until nearly the end of the George H. W. Bush Administration in November 1992. For a full decade all major federal regulations technically were required to provide net benefits undera BCA using a 10 percent discount rate. Not surprisingly, the regulatory agencies often ignored OMB’s Circular and utilized lower discount rates, which justified more regulation. Finally, on November 12, 1992 the OMB of the Bush I Administration revised Circular A-94 (57 Fed. Reg. 53519). Among the revisions, OMB reduced the base-case discount rate from 10 to 7 percent ostensibly because the later rate ‘‘approximate[d] the marginal pretax rate of return on an average investment in the private sector in recent years.’’ The guidance did not attempt to defend or explain why historical returns on private investments constituted an appropriate measure for regulations designed especially to deal with pollution and other environmental externalities associated with market failures.10 Nor did it explain why alternatives such as the ‘‘shadow price of capital’’ or the real return on long-term government debt were inappropriate.11 Still, the move from a 10 percent discount rate to a 7 percent discount rate ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 7 constituted a significant move toward moderation and conciliation with the regulatory agencies. 2.2. 1993–2001: BCA under Clinton Shortly after taking office in 1993, the Clinton Administration replaced the two Reagan era Executive Orders with one of its own. According to Hahn and Sunstein (2002, p. 1507), President Clinton’s E.O. 12,866 ‘‘endorsed the essential features’’ of the Reagan orders. This is superficially true in that E.O. 12,866 continued to require agencies to perform regulatory BCAs and submit those BCAs for OMB review. But E.O. 12,866 differed in fundamental respects from the Reagan orders. First and foremost, it demoted BCA from a ‘‘decision rule’’ to a requisite source of information for decision making. Unlike President Reagan’s E.O. 12,291, which expressly prohibited regulatory measures unless the potential benefits outweighed the potential costs, President Clinton’s E.O. 12,866 merely required agencies to prepare BCAs to the extent consistent with their statutory mandates; it did not require the agencies to base their regulatory decisions exclusively (or even primarily) on the outcomes of BCAs. In addition, E.O. 12,866 mandated that agencies consider not just economic efficiency but also ‘‘distributive impacts ... and equity’’ when choosing among alternative regulatory approaches. As it did during the Reagan Administration, the OMB dragged its feet in revising its guidelines pursuant to President Clinton’s new Executive Order on BCAs. When OMB finally did so in 1996, those revisions – referred to for the first time as ‘‘best practices’’ for BCA – included a more elaborate, thoughtful, and detailed explanation of BCA standards. OMB maintained its ‘‘base-case’’ 7 percent discount rate, but noted for the first time that ‘‘[m]odern research in economic theory has established a preferred model for discounting, sometimes referred to as the shadow price approach.’’ That approach, according to OMB, ‘‘is viewed as being approximated by the real return to a safe asset, such as Government debt.’’ Normally, the return on such an investment would be substantially lower than the 7 percent rate, which the OMBmaintained as its ‘‘best practice’’ standard. Indeed, the OMB stopped short of endorsing the shadow price method as the preferred way of selecting a discount rate; and it required any agency preferring to use that method to ‘‘consult with OMB prior to doing so.’’ The 1996 revisions to OMB’s guidelines for regulatory analyses also paid more attention than had previous versions to: issues of risk and uncertainty DANIEL H. COLE8 in BCA, including the use of sensitivity analysis; various methods for valuing human life, among other non-market goods, through ‘‘value of statistical life’’ and ‘‘value of statistical life-years extended’’ analyses; and stressed the importance of making all BCA assumptions transparent. In these and other respects, the 1996 revised OMB guidelines reflected recent developments in the theoretical and academic literature on BCA. A close reading of the 1996 revised guidelines suggests that OMB was becoming more professional and ostensibly less political in its oversight of federal regulatory processes. This evolution at OMB did not, however, portend greater consistency in regulatory agencies’ BCAs. Agencies, including the EPA, continued to use widely varying discount rates with little or no explanation or justification. Between 1992 and 1998, federal agencies utilized discount rates ranging between 3 and 10 percent for assessing regulations with time horizons of 20 years or less, and rates between 3 and 7 percent for longer-term regulations (with time horizons in excess of 30 years) (Morrison, 1998, Tables 1 and 2).12 The EPA itself relied on various discount rates, with little explanation of its choice in any particular case. For example, the agency employed a 3-percent discount rate when it sought to regulate lead-based paint, but 7- and 10-percent rates, respectively, for assessing proposed regulations of drinking water and locomotive emissions (Morrison, 1998, p. 1337). A 1997 report by the General Accounting Office (GAO, 1997) criticized the EPA’s economic analyses for their lack of transparency about assumptions. Despite these and other evident problems of implementation,13 OMB concluded in its 1998 report to Congress that ‘‘the overall picture remains one of slow but steady progress towards the Best Practices standards’’ established in its 1996 revised guidance (OMB, 1998, p. 83). It is not at all clear, however, that the OMB’s 1996 revised standards actually constituted ‘‘best practices’’ in the first place. In October 1999, Professor (now Dean) Richard Revesz of the New York University School of Law testified before the Senate on the use of BCA under the Clean Air Act.14 He cited several respects in which the OMB guidelines remained deficient. In particular, he criticized the OMB’s preferred 7 percent discount rate for being ‘‘a great deal higher than rates supported by economic theory,’’ resulting in the undervaluation of the benefits of environmental regulations. He argued that a 2–3 percent discount rate would be more appropriate, and noted that the General Accounting Office and Congressional Budget Office already used 2–3 percent rates in their own economic assessments. In addition, Revesz argued that OMB’s approach to valuing risk and human lives solely on the basis of wage premia in labor markets was faulty because it did not take into account the involuntary nature of many environmental risks (in contrast to ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 9 voluntary risk acceptance by workers in exchange for additional income). In addition, Revesz complained that workers who accept risky jobs are not representative of the population as a whole; and he noted the particularly painful nature of carcinogenic deaths. More generally, Revesz criticized OMB procedures for turning BCA ‘‘into an anti-regulatory tool, rather than into a tool to make regulation more rational.’’ He noted, for example, that the OMB only required BCA for regulatory impositions; it did not require BCA when regulations were relaxed or repealed, even if deregulation might lead to net social welfare losses. He also criticized OMB for its own lack of transparency, including its occasional failure to disclose its contacts with groups that might be interested in the outcome of regulatory proceedings.15 Professor Revesz’s criticisms of OMB are supported by examples of direct OMB interference in regulatory decision-making processes. The most famous example might be the mid-1980s fight between OMB and EPA over the latter’s BCA for proposed bans on certain uses of asbestos in the workplace (on which see, for example, Menell & Stewart, 1994, pp. 103–116). In that conflict, the OMB sought to impose its own estimation of benefits – $1 million per cancer case avoided – and, whereas EPA sought to discount costsand benefits from the time of workplace exposure to asbestos, the OMB argued that the agency should discount only from the time of disease manifestation (potentially decades after exposure). Moreover, there was evidence that OMB pressured EPA ‘‘behind closed doors’’ to accept OMB’s approach to discounting. Whether one sides with EPA or OMB on these issues – the House Subcommittee on Energy and Commerce sided squarely with the EPA (see Menell & Stewart, 1994, pp. 106–113) – the conflict itself exemplified the absence of clear and agreed-upon ‘‘best practices’’ and the lack of transparency in regulatory BCA.16 At the end of the 1990s, neither the OMB nor the regulatory agencies appeared truly committed to a methodologically consistent BCA tool. This observation should not surprise anyone familiar with the literature on positive political-economy. Applying that literature to the case of regulatory BCA, Eric Posner (2001, p. 1141) argues that the very purpose of requiring regulatory agencies to perform economic analyses is not to increase the efficiency of regulation but to ‘‘ensure that elected officials maintain power over agency regulation.’’ On his view, the question is not whether regulatory BCA is methodologically consistent or biased one way or another; rather the question is whether it is effective at maintaining political control over regulatory agencies.17 Toward the end of the Clinton Administration, in September 2000, the EPA published a long (179 pages) and highly detailed set of Guidelines for DANIEL H. COLE10 Preparing Economic Analyses. These Guidelines, which were based in part on an earlier EPA publication on BCA dating from the early 1980s (EPA, 1983) as well as OMB’s 1996 ‘‘Best Practice’’ guidelines, were peer reviewed by the Environmental Economics Advisory Committee (EEAC) of EPA’s Science Advisory Board.18 The EPA’s 2000 Guidelines addressed all aspects of economic analysis of environmental policy from the setting of goals to the determination of the best mechanisms for achieving those goals (i.e., environmental instrument choice) to ex post assessments of implemented policies. The EPA developed a tripartite framework under E.O. 12,866 for economic analysis of environmental regulations, including (1) net social welfare assessments utilizing BCA, (2) assessments of policy winners and losers using economic impact analysis (EIA), and (3) analysis of policy consequences for disadvantaged sub-populations using ‘‘equity assessments.’’ Each of these functions depends critically on a clear and consistent specification of the regulatory ‘‘baseline,’’ which is the situation at the time a new policy is promulgated and implemented. The 2000 EPA Guidelines called for the agency to make predictions on the expected effects of a new policy from baseline. Such predications are, of course, highly dependent on assumptions, the information available to the agency at the time a new policy is promulgated, and issues of risk and uncertainty. In accordance with the OMB’s 1996 ‘‘Best Practice’’ standards, EPA’s 2000 Guidelines emphasized the importance of dealing forthrightly with uncertainty by focusing on expected values of costs and benefits, clearly disclosing assumptions, and subjecting those assumptions to sensitivity analyses. Also, following Arrow and Fischer (1974), the EPA Guidelines noted the significance of ‘‘quasi-option’’ values for potentially ‘‘irreversible decisions.’’ From the perspective of those interested in the evolution of regulatory BCA, the most interesting chapters of EPA’s 2000 Guidelines are those that deal with social discounting and the evaluation of environmental (including public health) benefits. In its 2000 Guidelines, EPA explicitly noted that ‘‘choosing the discount rate has been one of the most contentious and controversial aspects of EPA’s economic analyses of environmental policies’’ (EPA, 2000, p. 33). At the same time, however, the agency was aware that ‘‘the effects on net benefits of alternative assumptions made for measuring and valuing uncertain effects of environmental policies can overwhelm the effects of changes in the discount rate’’ (EPA, 2000, p. 33). EPA’s 2000 Guidelines reviewed the ever-growing literature on social discounting, and like OMB’s 1996 revised guidelines, the agency noted ‘‘widespread support’’ for the consumption rate of interest/shadow price of capital (CRI/SPC) method of discounting in intergenerational contexts ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 11 (EPA, 2000, pp. 40, 43). But unlike the OMB’s 1996 revised guidelines, in which the OMB maintained its preferred 7 percent discount rate based on historical returns on private investments, EPA’s 2000 Guidelines actually sought to establish BCA procedures on CRI/SPC. While noting that this method can be very expensive to employ (EPA, 2000, p. 48, n. 20),19 EPA concluded that relatively useful, practical measures exist: ‘‘What is offered in the empirical literature for choosing a social discount rate focuses on estimating the consumption rate of interest at which individuals translate consumption through time with reasonable certainty. . . . For this, historical rates of return, post-tax and after inflation, on ‘safe’ assets, such as U.S. Treasury securities, are normally used, typically resulting in rates in the range of one to three percent’’ (EPA, 2000, p. 47). Thus, EPA concluded that its economic analyses should use a discount rate of 2–3 percent (EPA, 2000, p. 48). However, because the OMB continued to mandate a 7 percent discount rate in its 1996 revised guidance, the EPA felt compelled to present cost and benefit estimates using alternative 2–3 percent and 7 percent discount rates (EPA, 2000, p. 48). Implicit throughout EPA’s discussion of the literature on social discounting is a sense that OMB’s preferred 7 percent discount is inappropriately high. Finally, it is worth noting that although EPA’s 2000 Guidelines explicitly rejected use of a zero discount rate, the agency considers it appropriate for BCAs to present streams (but not summations) of non-discounted costs and benefits over time. Chapter 7 of EPA’s 2000 Guidelines addressed the equally important and contentious issue of evaluating non-market effects of environmental policies, including the value of human lives saved. The Guidelines discussed the wide array of market and non-market costs and benefits, and various approaches – all of them more of less defective – to valuing non-market benefits, such as contingent valuation and hedonic pricing. The EPA reviewed 26 articles, including both labor market (wage-risk) and contingent valuation studies, estimating the value of a statistical life. Those articles established a range of values (in 1997 dollars) from $0.7 million (Kneisner & Leeth, 1991) to $16.3 million (Garen, 1988; EPA, 2000, p. 89, Exhibit 7-3). EPA selected the mean of the range for its assumed value of a statistical human life: $4.8 million in 1990 dollars. Adjusting for inflation, the value would increase to $6.1 million in 1999 dollars (EPA, 2000, p. 90). In addition to valuing lives saved, the EPA also had to estimate other values not entirely encompassed in market prices, including the morbidity benefits of regulations, i.e., the avoided costs of non-fatal illnesses, and ecological benefits. The agency did not, however, adopt any specific values with respect to such effects, as it did in the case of human-life valuations. DANIEL H. COLE12 After addressing issues of social welfare calculation and distributional effects (i.e., equity issues), the EPA concluded its 2000 Guidelines with some generally applicable rules governing agency BCAs. It stressed that: all aspects of economic assessments should be presented clearly and transparently; important data sources and references should be cited, along with their assumptions and justifications; uncertainties should be highlighted by use of upper- and lower-bounded ranges of expected values; policy outcomes should be monetizedto the fullest extent feasible; unquantifiable effects should be highlighted so that they are not ignored in policy decisions based (in part) on BCAs; results of all distinct parts of the economic assessment – BCA, EIA, cost-effectiveness analysis, equity effects, should be clearly presented (EPA, 2000, pp. 175–178). Finally, the EPA stressed the limits of economic analyses for regulatory decision-making: ‘‘The primary purpose of conducting economic analysis is to provide policy makers and others with detailed information on [a] wide variety of consequences of environmental policies ... . Determining which regulatory options are best even on the restrictive terms of economic efficiency, however, often is made difficult by uncertainties in data and by the presence of benefits and costs that can only be qualitatively assessed. Thus, even if the criterion of economic efficiency were the sole guide to policy decisions, social benefits and cost estimates alone would not be sufficient to define the best policies’’ (EPA, 2000, p. 178). In concluding that BCA should not be the sole basis for environmental policy making, the EPA’s 2000 Guidelines were fully consistent with E.O. 12,866 (but not with its predecessor E.O. 12,291). EPA’s 2000 Guidelines ostensibly constituted a major step in the evolution and maturation of economic analyses at that agency. No longer, presumably, would various EPA assessments rely, without explanation, on various discount rates and human life valuations seemingly chosen at random. The agency had committed itself, on paper at least, to follow certain standards consistently, even if those standards were contestable and not fully consistent with OMB guidelines. 2.3. 2001 to the Present: Competing ‘‘Best Practice’’ Standards and the Political Manipulation of BCA under Bush II Since 2000, the evolution of regulatory BCA within the federal government has been a mixed bag of further methodological refinements and continued, even increased, politicization. The further refinements are mostly evident in the OMB’s Circular A-4 (in OMB, 2003). The increased politicization is ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 13 apparent in recent cases where BCAs have been blatantly manipulated to achieve or avoid certain outcomes. According to OMB, the purpose of its 2003 revised Circular A-4 on ‘‘Regulatory Analysis’’ was to ‘‘refine’’ OMB’s ‘‘best practices’’ guidelines of 1996. Like EPA’s 2000 Guidelines, the OMB’s 2003 Circular A-4 was peer reviewed by a distinguished group of social scientists.20 Interestingly, the OMB’s newest guidelines are less than one-third the length of EPA’s Guidelines, and whereas the EPA’s Guidelines discussed and cited literally dozens of works in the academic literature on BCA, the OMB’s 2003 Circular A-4 cites only a handful of articles. In many respects, however, the OMB’s 2003 document and EPA’s 2000 document express similar sentiments about the importance of consistency, clarity, and transparency in BCA processes. Somewhat surprisingly, OMB’s 2003 version of Circular A-4 expresses less faith in BCA as a decision tool than did the EPA’s 2000 Guidelines. For example, OMB notes that ‘‘you cannot conduct a good regulatory analysis according to a formula’’ because different regulatory circumstances require ‘‘different emphasis in the analysis’’ and all BCAs require ‘‘competent professional judgment’’ (OMB, 2003, p. 3). The 2003 OMB Circular A-4 also states that ‘‘[w]hen important benefits and costs cannot expressed in monetary units, BCA is less useful, and it can even be misleading, because the calculation of net benefits in such cases does not provide a full evaluation of all relevant benefits and costs’’ (OMB, 2003, p. 10). With respect to cost- effectiveness studies, the OMB denies any effort to impose on regulatory agencies a certain approach; instead it suggests that agencies – experiment with ‘‘multiple measures of effectiveness that offer different insights and perspectives’’ (although those measures should all be clearly articulated and explained) (OMB, 2003, p. 13). And in accordance with E.O. 12,866, OMB’s 2003 Circular A-4 stresses the importance of distributional effects. Much of the OMB’s discussion of BCA in 2003 Circular A-4 is uncontroversial and thoroughly consistent with EPA’s 2000 Guidelines. Few of the standards OMB articulates actually relate to the more controversial aspects of BCA, e.g., the selection of a discount rate and the valuation of non-market goods, including human life. In its revised Circular A-4, OMB acknowledges that some costs and benefits, including ecological benefits, can be difficult to quantify. In dealing with such difficult-to-quantify costs, it calls on agencies to discuss the ‘‘strengths and limitations of the qualitative information’’ (OMB, 2003, p. 27). It even allows that ‘‘unquantified benefits’’ might, from time to time, ‘‘affect a policy choice.’’ When that happens, agency BCAs should ‘‘provide a clear explanation for the rationale behind the choice’’ (OMB, 2003, p. 27). DANIEL H. COLE14 In a section of the 2003 revised Circular A-4 on the assessment of fatality risks, the OMB addresses the contentious issue of choosing a method for assessing the value of statistical human lives. First, the agency discusses the ‘‘value of statistical life’’ (VSL) method, which has been the predominant approach since the 1980s. Whereas EPA (2000, pp. 89–90) discerned a range of VSL estimates in the literature from about $1 million to over $16 million, OMB’s 2003 Circular A-4 indicates a range of $1 million to $10 million (OMB, 2003, p. 30). Assuming that the mean of the range is generally acceptable, OMB’s preferred VSL would be very close to the EPA’s selected mean of $4.8 million (EPA, 2000, pp. 89–90).21 Importantly, the OMB rejects, at least for the time being, the notion that VSL might be adjusted based on the age of members of the affected population (OMB, 2003, p. 30). This decision to reject age-adjusted VSL estimates reflects a concession to public opinion. OMB previously had been widely criticized for attaching lower values to the lives of elderly Americans (see, e.g., Los Angeles Times, April 30, 2003, p. 1). On the other hand, OMB’s 2003 Circular A-4 embraces a second approach to valuing fatality risks: ‘‘value of statistical life-years extended’’ (VSLY). This approach, in contrast to VSL, does account for discrepancies in age among the members of affected populations. However, OMB recommends that agencies adopt larger VSLY estimates for older citizens because they face larger overall health risks from all causes and they may have greater accumulated savings to spend on health care (OMB, 2003, p. 30). Still, by its very nature the VSLY method discriminates against older citizens because they have fewer life years remaining. Recognizing this problem, OMB cautions agencies that ‘‘regulations with greater numbers of life-years extended are not necessarily better than regulations with fewer numbers of life-years extended’’ (OMB, 2003, p. 30). Ultimately, OMB recommends that agencies provide estimates of both VSL and VSLY in their BCAs (OMB, 2003, p. 30). Between 1996 and 2003 the OMB marginally refined its position on the social rate of discount. It maintained the same 7 percent discount rate as the ‘‘base-case for regulatory analysis’’ (OMB, 2003, p. 33), but added that regulatory agencies ‘‘should provide estimates using both 3 percent and 7 percent’’ discount rates (OMB, 2003, p. 24). The 3 percent rate is OMB’s estimate of the ‘‘social rate of time preference.’’ In its 1996 Circular A-94, OMB discussed this method of discounting, but did not make any recommendations based upon it. Apparently, by 2003 OMB had become convinced that such a recommendation was appropriate. Finally, Circular A-4 recommends using other discount rates ‘‘to show sensitivity to the estimates of the discount rate assumption’’ (OMB, 2003, p. 24). Thus, OMB’s 2003 ‘Best Practice’ Standards for Regulatory Benefit–CostAnalysis 15 Circular A-4 is in complete agreement with EPA’s 2000 Guidelines about the use of discount rates in regulatory economic analyses, with the caveat that the EPA’s standards include a 7 percent discount rate only because OMB requires it. When OMB opened its 2003 Circular A-94 for public comment, one commentator argued that OMB should simply abandon the 7 percent discount rate, but the OMB rejected this recommendation on the grounds that a lower discount rate (such as 3 percent) would ‘‘not be appropriate for regulations that had a strong displacing effect on capital investment’’ (OMB, 2003, p. 176). Once again, the OMB did not consider the propriety of basing discount rates for environmental BCA on private investment markets that themselves do not account for the environmental risks they generate. OMB’s 2003 revised Circular A-4 also addresses the problem of ‘‘intergenerational discounting,’’ that is, discounting over long time periods. While recognizing that some scholars oppose discounting the utility of future generations, the revised Circular offers two positive reasons for intergenerational discounting: (1) the expectation that future generations ‘‘will be wealthier and thus will value a marginal dollar of benefits and costs by less than those alive today’’; and (2) the longer the time horizon of the policy being analyzed, the greater the uncertainty concerning the correct discount rate (OMB, 2003, p. 36). Based on its reading in the literature on intergenerational discounting, the OMB recommends that agencies adopt a third discount rate, below 3 percent but still positive, for assessing intergenerational costs and benefits (OMB, 2003, p. 36). Finally, OMB’s Circular A-4 addresses ‘‘other key considerations’’ in BCA, including the treatment of technological change over time, and the treatment of uncertainty. First, in recognition that technologies change in response to market forces as well as regulatory requirements, OMB recommends that agencies ‘‘should assess the likely technology changes that would have occurred in the absence of regulatory action (technology baseline)’’ (OMB, 2003, p. 37). Otherwise, agencies are likely to overstate the benefits of regulatory requirements.22 With respect to uncertainty, OMB suggests that agencies respond by conducting additional research prior to rulemaking, especially in cases of irreversible or large up-front investments, unless they can show that the cost of delay is likely to exceed the value of any additional information (OMB, 2003, p. 39). OMB also recognizes the growing literature on ‘‘real options’’ methods of incorporating uncertainty into BCA (OMB, 2003, p. 39).23 Between EPA’s 2000 Guidelines and OMB’s 2003 Circular A-4, it appears that the two agencies are growing closer to agreement on a set of standards for regulatory BCA.24 Significant differences remain, and the next section of DANIEL H. COLE16 this paper explains why it remains unlikely that EPA and OMB ever will reach complete consensus on a set of ‘‘best practices’’ for BCA. In any event, the improvements to BCA policies since 2000 have not been matched by improvements in the consistency and adequacy of individual BCAs, which in some cases at least continue to be highly influenced by partisan politics. Political manipulation is plainly evident, for example, in the case of EPA’s 2005 BCA for the Bush Administration’s ‘‘Clear Skies’’ initiative, which was extensively criticized in a recent Congressional Research Service (CRS) Report to Congress (McCarthy & Parker, 2005).25 The ostensible goal of Clear Skies was to deal comprehensively with air pollution problems from the electric utility industry. In 2003, that industry was responsible for 72 percent of all sulfur dioxide emissions, 24 percent of nitrogen oxide emissions, and 41 percent of carbon dioxide emissions, and more than 40 percent of all mercury emissions in the United States. Power plant emissions of sulfur dioxide and nitrogen oxides have been trending downwards in recent years thanks mainly to the acid rain program of the 1990 Clean Air Act Amendments. However, utilities have long complained about the ‘‘complexity’’ of the ‘‘multilayered and interlocking pathwork of controls’’ applied to them (McCarthy & Parker, 2005, p. 2). As noted in the CRS Report (McCarthy & Parker, 2005, pp. 2–3), a more simplified and uniform approach has been evolving for several years within the EPA under existing statutory mandates. However, the Bush Administration and Congress are both advocating new legislation that would regulate utility emissions of major air pollutants in a more comprehensive and integrated way. The Bush Administration supports a group of bills known collectively as ‘‘Clear Skies,’’ which would require a 70 percent reduction in SO2 and NOX emissions by 2018, although actual attainment would likely be delayed until 2026 or later because of the legislation’s generous ‘‘banking’’ provisions. Two alternative legislative proposals, one sponsored by Senator James M. Jeffords (I-Vt.) and the other by Senator Thomas R. Carper (D-Del.), would also permit banking and trading of allowances, but would require greater overall emissions reductions on shorter deadlines. In addition, and unlike Clear Skies, the Jeffords and Carper bills, would impose regulatory controls to reduce utility emissions of carbon dioxide in order to mitigate climate change. On October 27, 2005, EPA published a BCA (comprised of 45 separate documents26) which purported to compare the various legislative proposals to control air pollution emissions from power plants. As a baseline for its BCA, EPA assumed unrealistically that in the absence of new legislation neither EPA nor the states would impose additional regulatory controls on power plant emissions. This assumption was flat out contradicted by three ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 17 newly minted EPA rules regulating power plant emissions of sulfur dioxide, nitrogen oxides, and mercury. EPA’s final BCA for Clear Skies did not even mention those new rules. As the CRS noted in criticizing EPA’s regulatory baseline assumptions, ‘‘[c]ontrolling air pollution is a moving target and ... it is important that any analysis work from updated baseline projections and assumptions when possible’’ (McCarthy & Parker, 2005, p. 5). One conclusion reached by EPA’s BCA on Clear Skies, and confirmed by the CRS Report, was that Clear Skies would cost less than the alternative legislative proposals. According to the CRS, this conclusion was unsurpris- ing because Clear Skies ‘‘has less stringent requirements and later deadlines’’ (McCarthy & Parker, 2005, p. 5). On the other hand, Clear Skies would yield fewer overall benefits than the other legislative proposals. According to the EPA’s own estimates, the Clear Skies bill would provide $6 billion in annual benefits in 2010, compared to $51 billion in annual benefits for Senator Carper’s bill and $83 billion in annual benefits under Senator Jefford’s bill. ‘‘The higher benefits for the Carper and Jeffords bills reflect the fact that Clear Skies’ required pollution caps are less stringent, and the implementa- tion schedule is more relaxed’’ (McCarthy & Parker, 2005, p. 9). The incremental benefits of Clear Skies would be even lower, if EPA’s baseline assumptions were changed to incorporate EPA’s recently promul- gated rules on sulfur dioxide, nitrogen oxide, and mercury emissions. According to the CRS Report to Congress (McCarthy & Parker, 2005, p. 9), incorporating those regulations into baseline would reduce Clear Skies’ incremental benefits (above baseline) to 10 percent in 2010 and only 2 percent in 2020. In sum, the net benefits of Senator Jefford’s bill ‘‘far exceed those of Clear Skies’’ and Senator Carper’s bill (McCarthy & Parker, 2005, p. 11). However, the social welfare advantages of Senator Jefford’s proposal were minimized in EPA’s BCA by the unrealistic assumption of a baseline that (a) excluded recently promulgatedrules and (b) presumed that, in the absence of new legislation, no new regulations would be promulgated. In addition to its unrealistic baseline assumptions, EPA’s BCA for Clear Skies made no attempt to monetize environmental benefits, which disadvantaged the Jeffords and Carper proposals significantly because their more stringent emissions requirements were predicted to lead to greater environmental benefits than the Bush Administration’s Clear Skies initia- tive.27 Worse still, the BCA did not model the health effects of regulating mercury emissions. According to another CRS Report to Congress (McCarthy, 2005), health benefits from EPA’s mercury regulations could range from ‘‘a few million dollars per year to several billion dollars per year’’ (McCarthy & Parker, 2005, p. 15). Omitting these benefits from the Clear DANIEL H. COLE18 Skies BCA favored the Bush Administration’s proposal over alternative proposals that would impose more stringent caps on mercury emissions. Similarly, the EPA’s BCA for Clear Skies did not attempt to monetize the benefits of reductions in carbon dioxide emissions. Finally, the EPA’s Clear Skies BCA unreasonably assumed that the price elasticity for electricity and natural gas would be zero and that power plants were subject to short-term construction constraints. Both of those dubious assumptions served to make the Bush Administration’s Clear Skies initiative more attractive and Senator Jefford’s bill in particular less attractive. The CRS concluded that ‘‘EPA’s benefit analysis is limited and incomplete, which works to the disadvantage of alternatives to Clear Skies that include more stringent standards’’ (McCarthy & Parker, 2005, p. 16). A Washington Post reporter was somewhat more pointed in her conclusion: ‘‘The Bush Administration skewed its analysis of pending legislation on air pollution to favor its bill over two competing proposals.’’ The EPA argued in response that the CRS ‘‘ignores and misinterprets our analysis’’ (Washington Post, May 11, 2006). Interestingly, OMB has been silent about EPA’s BCA for Clear Skies,28 which again raises the question of whether OMB review is designed to maximize regulatory efficiency or simply to minimize regulation. In other cases, the OMB has reviewed agency BCAs in ways that appear (at least) to be politicized. Despite regular OMB denials that it is an anti- regulatory agency, the empirical evidence suggests otherwise. According to a recent review by the legal scholar David Driesen (2006, pp. 364–380) of 25 environmental, health and safety regulations ‘‘significantly affected’’ by OMB review between June of 2001 and July of 2002, 24 were significantly weakened and none was strengthened.29 Now this may just be evidence that environmental, health and safety agencies always over-regulate and never under-regulate, but that hardly seems likely.30 More likely, as Driesen (2006) concludes, OMB review is not nearly as neutral as OMB officials and supporters claim.31 Another recent study by Laura J. Lowenstein and Richard L. Revesz (Nov. 2004) found that OMB, in reviewing EPA BCAs, regularly substituted its own cost and benefit valuations based on ‘‘questionable techniques that inappropriately lower[ed] the value assigned to human lives.’’ The Bush EPA has embraced the Value of Statistical Life-Years Saved (VSLY) approach, instead of the traditional Value of Statistical Life Saved (VSL) approach to valuing human lives. The VSLY approach (as already noted) reduces the expected benefits of regulation by assigning lower values to the lives of older, unhealthy, and disabled Americans. Pursuant to its newly adopted VSLY approach, the EPA derived an age-based adjustment factor that reduced VSL ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 19 estimates by 37 percent for individuals aged 70 and over. That age adjustment factor was based on a single study conducted in the United Kingdom in 1989, even though more recent and comprehensive American studies were available. In addition, EPA ‘‘encouraged the use of a substantially-reduced initial VLS estimate from which life-year values were derived.’’ Obviously, the lower the initial VSL, discounted according to estimated life-years remaining, the lower the VSLY-based benefits of regulation will be. Finally, EPA assumed that the reduction in life expectancy associated with exposure to particulate matter (dust and soot) is only five years, regardless of the individual’s age at time of death. Significantly, EPA did not explain any of these ‘‘radical’’ departures from prior practice. Lowestein and Revesz (Nov. 2004) conclude that ‘‘the current administration’s approaches to valuing environmental benefits are theoretically unjustified and have a profound anti-regulatory impact.’’32 Even though EPA and OMB have come closer to agreement about basic principles for BCA over the past several years, the practice of BCA at both the EPA and the OMB remains substantially politicized and methodologi- cally inconsistent. In 2003, law professor William Funk noted that OMB estimates of the costs and benefits of major rules still sometimes differ from regulatory agency estimates ‘‘by more than an order of magnitude.’’33 For example, Funk noted that the U.S. Department of Agriculture’s estimate of the costs and benefits of a new labeling rule for meat and poultry were $218–272 million and $1.75 billion. OMB’s estimates of the costs and benefits of that same rule were $25–32 million and $205 million. Both agencies amortized costs and benefits over a 20-year period using a constant 7 percent discount rate. What, then, explains the discrepancies? As Professor Funk notes, if a ‘‘principle function of quantitative cost-benefit analysis is to improve the transparency of analysis while improving public understanding of the costs and benefits of regulation,’’ that function is ‘‘thwarted ... when radically divergent numbers are offered without explanation of the reasons for the difference.’’ Whatever progress may have made toward achieving consensus in principles and practices of BCA over the past two decades, it is clear that a great deal more remains to be done. The fact that EPA and OMB have continued to manipulate and politicize BCAs should not make us too pessimistic, however, about the utility of BCAs for decision making. Even though EPA’s BCA for the Clear Skies initiative was politically skewed and methodologically flawed, at least it improved the transparency of EPA’s decision-making process. The value of this increased transparency should not be underestimated. The BCA for ‘‘Clear Skies’’ provided the ammunition for the Congressional Research Service’s powerful critique (see McCarthy & Parker, 2005), which helped DANIEL H. COLE20 doom the initiative in Congress.34 Likewise, the transparency of OMB’s regulatory review process gave Professors Lowenstein, Revesz, and Funk the grounds to criticize that agency’s review procedures. 3. WHERE DOES BCA GO FROM HERE? A NEW EXECUTIVE ORDER OR AN INDEPENDENT, NON-GOVERNMENTAL TASK FORCE? In order to correct the continuing problems of regulatory BCA, Robert Hahn and Cass Sunstein (2002) recommend that President Bush issue a new executive order to add ‘‘greater depth and width’’ to BCA by instituting eight specific recommendations: First, the new executive order should ‘‘explicitly requir[e] agency compliance with OMB guidelines for regulatory analysis.’’ According to Hahn and Sunstein (2002, p. 1494), ‘‘regulatory compliance with [OMB] guidelines would significantly increase the rationality and coherence of the regulatory process.’’35 Second, to reduce the actual or perceived anti-regulatory bias of OMB’s BCA process, Hahn and Sunstein (2002, pp. 1494–1495) would create a mechanism whereby OMB might issue ‘‘prompt letters’’ to ‘‘spur regulation in those cases where it will do more good than harm.’’36 Third, agency BCAs should, to the extent consistent with enabling statutes, consider substitute risks that might be created by regulations. Atthe same time, agencies should avoid regulating de minimis risks. Fourth, when an agency promulgates a regulation that fails a strict BCA (of quantified costs and benefits), it should provide its rationale for acting. For example, a statute might require the agency to act regardless of the outcome of a BCA, or the agency may base its decision on important and thoroughly explained qualitative data that are not quantifiable. According to Hahn and Sunstein (2002, p. 1496), such explanations would contribute to improved accountability and transparency. So too would their fifth recommendation that agencies should make the underlying assumptions of their BCAs explicit, so that ‘‘interested parties inside and outside of the government can understand how the results were obtained, and perform their own analysis of the issue if they so choose’’ (Hahn & Sunstein, 2002, p. 1496). Sixth, each year agencies should generate backward-looking ‘‘regulatory retrospectives’’ as well as forward-looking ‘‘regulatory plans.’’ The purpose of the annual retrospectives would be to facilitate OMB’s task of preparing its annual reports to Congress on executive branch regulations. The purpose of the annual plan would be to ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 21 facilitate OMB’s early participation in the BCA. Seventh, Hahn and Sunstein (2002, p. 1496) recommend that the new executive order extend BCA requirements to ‘‘independent’’ regulatory agencies. While recognizing that such agencies never have been subject to executive orders in the past, Hahn and Sunstein (2002, pp. 1531–1537) argue that such orders might lawfully be extended to them.37 Eighth and finally, Hahn and Sunstein (2002, p. 1497) recommend that the new executive order authorize limited judicial review of agency BCAs. In effect, BCAs would come within the ambit of the Administrative Procedures Act (APA) (5 U.S.C. y 553), so that a reviewing court could invalidate a regulation based on a defective BCA as ‘‘arbitrary and capricious.’’38 Several of Hahn and Sunstein’s (2002) recommendations are appealing (and are discussed infra). However, their recommendations are unresponsive to several of the methodological problems identified in the Section 2 of this paper. One reason for this is that Hahn and Sunstein’s recommendations were not motivated primarily by the authors’ perception of methodological problems in BCA. Rather, Hahn and Sunstein (2002) were concerned first and foremost about a perceived lack of compliance with presumably well- established BCA principles, resulting in regulations that ‘‘seem to do more harm than good’’ (Hahn & Sunstein, 2002, p. 1490). The empirical basis for this finding was an earlier book of Hahn’s et al. (2000, pp. 15–19), which examined 24 regulations and found that only 9 would pass a BCA (see Hahn & Sunstein, 2002, pp. 1490–1491). Thus, despite their recommendation concerning regulatory ‘‘prompt letters,’’ Hahn and Sunstein’s immediate concern was inefficient over-regulation. In addition, Hahn and Sunstein’s recommendations (particularly the first and sixth recommendations) suggest that the problems with BCA in the federal government lie primarily with the regulatory agencies and less (if at all) with OMB. Importantly, their recommendations (except that relating to OMB ‘‘prompt letters’’) would not significantly alter current OMB principles or practices relating to BCA. Perhaps Hahn and Sunstein believe that OMB already has succeeded in deriving a set of ‘‘best practices’’ for BCA with which regulatory agencies simply should be required to comply. If so, I would suggest they are mistaken for reasons outlined in Section 2.339 To the extent that OMB’s own BCA principles and practice require revision, Hahn and Sunstein’s (2002) recommendations are mostly immaterial, and the first recommendation, which would require regulatory agencies to comply with OMB standards no matter how unreasonable those standards might be, could well prove harmful. The history of regulatory BCA in the federal government, recounted in Section 2, suggests that the situation is rather more complicated than Hahn DANIEL H. COLE22 and Sunstein’s (2002) simple story of agency under-compliance with presumably well-founded OMB procedures. In reality, that history has been a mixed bag of progress and politics on the part of all institutional players, including the OMB. Were we to predict the future based on the past, we might expect some additional tinkering with BCA policies at both OMB and EPA. The two agencies might come somewhat closer to consensus on methodological principles. But because of their divergent institutional missions and political predilections it seems highly unlikely that they will ever agree in principle, let alone in practice, to a single, useful set of ‘‘best practices’’ for CBA. In some cases – particularly those in which the political stakes are high – economic analyses likely would be politicized and manipulated by one institution or another to achieve politically preferred outcomes. For that reason, this paper recommends that the task of establishing ‘‘best practices’’ for environmental BCA should be taken out of the hands of the government agencies that prepare and review them. Instead, those standards should be set by a cohort of independent economists, legal scholars, and policy analysts convened under the auspices of a quasi-governmental organization, such as the National Academy of Sciences, or a non- governmental organization, such as the American Economics Association or a new Society for Benefit–Cost Analysis.40 The goal would not be to establish a single, ‘‘pure’’ set of ‘‘neutral and objective’’ ‘‘best practices;’’ BCA simply contains too many subjective elements to ever claim the mantle of objectivity. Nor should we suppose that individual members of any group assigned to draft a set of ‘‘best practices’’ would come to the task without their own predispositions and biases. However, if the group is sufficiently large and representative of differing viewpoints, many of those predisposi- tions and biases could well wash out. Meanwhile, the group’s independence from the immediate political concerns of government agencies would allow it to take the time to more carefully review and discuss the existing literature on BCA. There is reason to believe that any set of revisable ‘‘best practices’’ established by such a non-governmental group is likely to less politicized and more legitimate than any set of ‘‘best practices’’ the OMB might seek to impose on the EPA and other regulatory agencies. Moreover, it makes sense that the social–scientific community should play a central role in defining what counts as ‘‘best practices’’ in regulatory BCA, given that BCA is supposed to be a social–scientific process. Society does not rely on government agencies to determine ‘‘best practices’’ for medical or dental procedures, constructing and conducting experimental economic studies, or performing empirical legal studies. Instead, society relies on more ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 23 or less formal cohorts of medical, natural, and social scientists to develop ‘‘best practices.’’ In law, permanent non-governmental organizations such as the American Law Institute and the National Conference of Commis- sioners on Uniform State Laws meet annually to review and revise, respectively, the Restatements of Law and the Uniform Commercial Code. It hardly seems radical to suggest that independent social scientists (including many from outside the Washington beltway) should play a lead role in developing, reviewing, and revising ‘‘best practices’’ for the social– scientific tool known as BCA. Fortunately, an independent BCA task force would have a head start on the process of setting ‘‘best practices’’ thanks to the efforts of economists and legal scholars who already have identified – independently of pressing regulatory and political concerns – the proper role of, approaches to, and limitationsof BCA. The next section discusses some of the existing literature on which the BCA ‘‘best practices’’ task force might rely; and it rehearses some recent contributions to that literature relating to the more contentious (because subjective) elements of BCA, including human life valuations and social discount rates. 4. HOW AN INDEPENDENT TASK FORCE MIGHT APPROACH THE PROBLEM OF DERIVING ‘‘BEST PRACTICES’’ FOR REGULATORY BCA 4.1. First Determine What is Not Controversial in BCA The logical starting point for any task force seeking to derive a set of ‘‘best practice’’ standards for regulatory BCA is to separate out those aspects of BCA on which consensus already exists from the more controversial aspects that require greater research and discussion. Several basic and generally agreed-upon principles of BCA were set out in a short article by Arrow et al. (1996) in the journal Science: First, BCA is a more or less useful tool of policy analysis, although in cases of great uncertainty a BCA may not be able to draw firm conclusions about social welfare effects. Second, policy-makers should be allowed and encouraged to use BCA.41 Third, BCAs should be required for all major regulatory decisions. Fourth, agency actions should not be bound by BCAs, even where BCAs indicate substantial net social costs, because ‘‘factors other than aggregate economic benefits and costs, such as equity within and across generations, may be important in some DANIEL H. COLE24 decisions.’’ Fifth, benefits and costs should be quantified to the extent feasible, and best estimates should be presented along with a description of uncertainties. Importantly, Arrow et al. (1996, p. 222) note that quantitative factors should not be allowed to dominate important qualitative factors in policy-making. Indeed, they support the notion that agencies might build ‘‘margin[s] of safety’’ into their regulations. However, all qualitative judgments in BCAs should be explicit. Sixth, Arrow et al. (1996) note that the more external review a BCA receives, the better it is likely to be. Thus, they support both peer review and OMB review of regulatory BCAs. They also support retrospective analyses to determine post hoc the quality of BCA predictions of future costs and benefits. Seventh, regulatory BCAs should be based on a ‘‘core set of economic assumptions,’’ including about discount rates, human life valuations, and health improvements, for the sake of consistency and comparability. To this end, ‘‘[a] single agency should establish a set of default values for typical benefits and costs and should develop a standard format for presenting results’’ (Arrow et al., 1996, p. 222). In other words, a set of ‘‘best’’ or ‘‘standard’’ practices should be adopted. As part of this recommendation, Arrow et al. (1996, p. 222) note the importance of discounting future benefits. But, they suggest, the discount rate should be ‘‘based on how individuals trade off current for future consumption.’’ This approach would almost certainly favor a lower rate than the OMB’s current base-case rate of 7 percent. Given the difficulty (even impossibility) of identifying a single, correct discount rate, Arrow et al. (1996, p. 222) recommend using a range of discount rates, noting that the same range should be used consistently in all regulatory BCAs. Eighth and finally, the authors note that even though overall efficiency is an important goal, agency economic analyses should pay close attention to the distributional consequences of their policies. They caution, however, that environmental (among other public health and safety) regulations ‘‘are neither effective nor efficient tools for achieving redistributional goals’’ (Arrow et al., 1996, p. 222). Arrow and his colleagues conclude by reiterating that BCA is not a panacea for policy-making: ‘‘formal benefit-cost analysis should not be viewed as either necessary or sufficient for designing sensible public policy.’’ It can, however, ‘‘provide an exceptionally useful framework for consistently organizing disparate information, and in this way, it can greatly improve the process and, hence, the outcome of policy analysis’’ (Arrow et al., 1996, p. 222). Few readers (other than those who deny any useful role for BCA42) would find any of Arrow et al.’s assertions to be controversial. OMB might disagree with the recommendation that discount rates should be based solely on ‘Best Practice’ Standards for Regulatory Benefit–Cost Analysis 25 individual’s decisions about trading off between present and future consumption, without any consideration for the displacement effects on private investment. Otherwise, there seems very little about which either the OMB, the EPA, or any other producer or consumer of BCAs would complain.43 Arrow et al. (1996) articulated generally acceptable principles or standards for producing quality economic analyses. They are not, however, sufficient. The chief problem is that the standards enunciated by Arrow et al. (1996) are very general while, as always, the devil is in the details. We might all agree that regulatory BCAs should always employ a range of discount rates for converting future costs and benefits into present prices, but just what should those rates be? We would probably all agree that agencies should establish a ‘‘core set of assumptions’’ for valuing for non-market goods, such as human lives and ecological goods, and apply them consistently from one BCA to the next, but just what should those core assumptions be? Arrow et al. (1996) provide a useful starting point for independent efforts to derive ‘‘best practices’’ for regulatory BCA, but most of the heavy lifting remains to be done. 4.2. The Hard Part: Achieving Consensus on the Devilish Details In a way, Arrow et al. (1996) created a misleading impression that BCA is less complicated and controversial than it really is. Even the aspects of economic analysis that appear to be purely mechanical, such as identifying the policy alternatives and listing their impacts, are not as straightforward as they might appear.44 Sometimes, there seem to be an almost unlimited number of policy options for dealing with a particular social or environmental problem. Consequently, policy analysts are forced to draw more or less arbitrary lines between those alternatives that are addressed in a regulatory BCA and those that are not. Likewise, analysts must draw lines in deciding which effects of different regulatory alternatives (including the ‘‘do nothing’’ alternative) are considered (or not considered) in a BCA. For example, a decision to locate a new road entails various direct and indirect, primary, secondary, and even tertiary effects. Among the primary effects are the direct costs for labor and machinery, and lost opportunities for using the land to be covered by the road for alternative purposes. Once built, the road will have various secondary impacts, as its very existence spurs development of adjacent lands, particularly at major intersections. Presumably, those indirect and secondary impacts should be considered as part of the BCA for the road project. But what about tertiary effects, for example, if the DANIEL H. COLE26 increased development resulting from road construction displaces economic development from some other region of the state? It probably would not be good enough simply to specify, for example, that policy-makers should consider all ‘‘foreseeable and significant’’ impacts, because what is ‘‘foreseeable and significant’’ often is in the eye of the beholder. Meanwhile, decisions about which alternative policies, and which impacts of those policies, to consider in a BCA are always likely to be a mere subset of all the conceivable alternatives and impacts. More or less arbitrary lines must be drawn, and just where those lines are drawn can affect the outcome of a BCA. Such problems pale in comparison, however, with the inherently more controversial aspects of economic analyses: valuing human lives (and other non-market goods)
Compartilhar