Tobacco Documents Online

BENEFIT-COST ANALYSIS OF ENVIRONMENTAL REGULATION: CASE STUDIES OF HAZARDOUS AIR POLLUTANTS John A. Haigh* David Harrison, Jr.** Albert L. Nichols*** Regulating toxic chemicals is highly controversial, yet it promises to be a major task confronting an industrial society. Increasing attention to toxic substances reflects in part recent growth in the number and quantity of man-made chemicals. As controls over the conventional pollutants take effect, toxic substances move to center stage in the political arena. This increased attention also stems from the fact that many of the statutes and regulatory procedures developed for the conventional pollutants are ill-suited to the new substances. The Administrator of the Environmental Protection Agency, William Ruckelshaus, has urged Congress to reconsider the present statutory framework for regulating toxic air pollutants.' EPA may shift its regula- tory strategy from the identification of specific control technology and ' B.A. 1976 Grinnell College; M.P.P. 1982 John F. Kennedy School of Government, Harvard University. Presently at Temple, Barker & Sloane, a privatc consulting firm in Lexington. Massachusetts. Previously employed by the Energy and Environmental Policy Center at the John F. Kennedy School of Government. Harvard University. °° A.B. 1967 Harvard University; M.S. 1968 London School of Economics and Political Science; M.A. 1971 Harvard University; Ph.D. 1974 Harvard University. Currently an Associate Professor at the John F. Kennedy School of Government, Harvard University. Formerly Senior Staff Economist. President's Council of Economic Advisers. •'• A.B. 1973 Stanford University; M.P.P. 1975 John F. Kennedy School of Govern- ment, Harvard University; Ph.D. 1981 Harvard University. Currently an Associate Profes- sor at the John F. Kennedy School of Government on leave serving as Director of the Economic Analysis Division in the Office of Policy, Planning, and Evaluation, U.S. Envi- ronmental Protection Agency. Financial support for this article provided by the U.S. Environmental Protection Agency under C'ooperative Agreement CR-809802-01-0 with the Energy and Environmental Policy Center, John F. Kennedy School of Government, Harvard University; Harvard School of Public Health, Interdisciplinary Programs in Health, U.S. Environmental Pro- tection Agency Grant #CR-807809; and The Alfred P. Sloan Foundation Program in En- vironmental Health. Although the research described in this report has been funded by the agencies listed above, it has not been subjected to their peer and administrative reviews, may not reflect the views of those agencies, and no official endorsement should be inferred. I. See Statement by W. Ruckelshaus, Administrator of the EPA. Before the Sub- comm. on Oversight & Investigations of the House Comm. on Energy & Commerce 10- I I(Nov. 7. 1983) (listing the specific problems experienced in implementing section 112) thereinafter cited as Statement by Ruckelshaus]. C909V1;Szoz

396 Harvard Environmental Law Review [Vol. 8:395 evaluation of the industry's ability to afford contruls= to a strategy that weighs the trade-offs between control costs and risk reduction.' This article evaluatec alternative met_hndc nf infruratino ho,neFit_n..F! considerations into the regulation of toxic substances. The use of benefit- cost considerations in this context is highly controversial and widely debated. The debate, however, has incorporated little or no reference to specific decisions made by environmental policy makers.a Proponents of benefit-cost analysis point to the general virtues of explicit evaluation of benefits and costs. Critics, on the other hand, stress the philosophical difficulties involved in making judgments about life and death5 or the practical difficulty of estimating the costs and benefits of control.6 These broad debates do not consider what is at stake in particular circumstances and, indeed, whether those who assess the scientific evidence very dif- ferently might find much common ground in actual regulatory decisions. This article attempts to fill that gap by considering three toxic pollutants - benzene, coke oven emissions, and acrylonitrile. All three pollutants are currently considered targets for control under section 112 of the Clean Air Act.' This article focuses on the ideas that benefit-cost principles can help to identify regulatory alternatives and that benefit-cost analysis can yield widely accepted policy recommendations despite large uncertainties in many parameter estimates. Critics caricature benefit-cost analysis as a mindless toting up of costs and benefits, but benefit-cost principles are -2. Id. at 20. 3. See W. Ruckelshaus, Administrator of the EPA, Science, Risk and Policy 10 ( June 22, 1973) (speech to the National Academy of Sciences). See also W. Ruckelshaus, Ad- ministrator of the EPA, Risk in a Free Society (Feb. 18, 1984) Ispeech at Princeton University); speech by J. Cannon, EPA Asst. Administrator for Air and Radiation, to the Natural Resources Law Section of the American Bar Association (Mar. 10. 1984). 4. See. e.g.. Crandall, The Use ofCost-Benefit Analysis in Regulatory Decisions, in MANAGEMENT OF AssESSED RIsK FOR CARCINOGENS 99-107 (W. Nicholson ed. 1981) (defending the general applicability of benefit-cost analysis tu regulatory decisionmaking); Harrison. Cosl-Benefit Analysis and the Regulation of Environmental Carcinogens, in MANA4F.MENT or ASSESSED RISK FOR CARCINOGENS 109-22 (W. Nicholson ed. 1981) (evaluating the advantages of using benefit-cost principles in regulating carcinogens); Ash- ford, Alrnnatives to Cost-Benefit Analysis in Regulatory Decisions, in MANAGFMENT OF ASSESSED RISK FOR CARCINOGENS 129-37 (W. Nicholson ed. 1981) (discussing the gencral iimilations of bencfit-cost analysis in regulatory decisionmaking). 5. See, e.g.. S. KELMAN, WHAT PRICE INCENTIVES? 27-88 (1981) (summarizing the ethical concerns involved in using the market for pollution control); Kelman. Cost-Benefit Analysis and Environmental, Sqfety, and Health Regulation: Ethical and Philosophical Considerations, in COST-BENEFIT ANALYSIS AND ENVIRONMENTAL REGULATIONS: PUI.I- ncs, ETHICS, AND METHODS 137-54 (D. Swartzman, R. Likoff & K. Croke cds. 1982). 6. See. e.g.. Ashford, supra note 4, at 129-37. 7. 42 U.S.C. 17412 (Supp. V 1981). Although this article provides background in- formation on the provisions and history of section 112 to place the specific case studies discussed in context, its analysis is not restricted to regulatory alternatives permitted by the current statute. Thus, some of the alternatives that it considers might require statutory chan-c, 19841 Hazardous Air Pollutants 397 more properly viewed as a framework for exploring opportunities to :nrrPace 1nMlt{~ ;~I ~t~~~ b~cilcllts Gi rcullLC unnecessary costs. The icruciai concept is marginaiism. (;iven an existing regulation, benefit-cost pnalysis identifies marginal changes that increase benefits more than costs, or decrease costs more than benefits.8 Critics argue that the data on benefits and costs of regulatory alter- natives are simply too uncertain to use risk assessment or benefit-colt results in policymaking.9 In some cases, however, all plausible estimaues of the parameters lead to the same policy recommendation. Thus, Ihe results in such cases remain robust with respect to uncertainty. Two of the three case studies evaluated in this paper fall in this category.10 Uncertainty, therefore, should not serve to dismiss out-of-hand benefit- cost analysis in environmental regulation. The first Part of this article discusses section 112 of the Clean Air Act, which provides the framework for regulating the three case-study pollutants." Part II presents the three case studies and includes an anal- ysis of regulatory alternatives for the three pollutants.'2 The next Part . summarizes the uncertainties in calculating regulatory benefits, and r.he sffect of those uncertainties on policy recommendations." Finally, Part I V outlines the overall conclusions derived from examining the case >tudies." L. REGULATORY CONTROLS A. Sec•tion 112 of the Clean Air Act Seetion 112 provides the statutory authority for regulating "hazar4- )us" air pollutants emitted from stationary sources." That section reflects he need to regulate hazardous pollutants outside the complex framework 8. The technology-bascd standards that the EPA has promulgated provide a basis for valuating the benelits and costs of those standards and for a detailed investigation of tgulatory alternatives. 9. See, e.g., Hurter, Tolley & Fabian, Benefit-Cost Anal,vsis and the Common Sense Environmental Polic•y, in COST-BENEFIT ANALYSIS AND ENVIRONMENTAL REOULA- ONS: POLITICS, ETHICS, AND METHODS 92-99 (D. Swartzman. R. LikofT& K. Croke eds. 82j (discussing the potential sources of uncertainty in comparing the bcncfits and costs environmental programs). 10. See infra text accompanying note 109. See also lqjra fable I. 11. See infra notes 15-56 and accompanying text. 12. See irt/'ra notes 57-140 and accompanying text. A more detailed analysis of these e studies has been presented in an earlier manuscript. Haigh, Harrison & Nichols, nefits Assessment and Environmental Regulation: Case Studies of Hazardous Air Pol- anis (July 1983) (unpublished manuscript available upon request from the authors). 13. See iq/'ra notes 141-220 and accompanying text. 14. See iq fra notes 221-228 and accompanying text. 15. 42 U.S.C. 4 7412 (Supp. V 1981). Vq09Vqszoz

39e Harvard Estvironr7e7.al Law IDevietv [Vol. 8:395 19841 Hrszarcloazs Air Pollutants 399 of ambient standards, state implementation plans, and new source per- formance standards established for the more ubiquitous "criteria" pollut- ants." The Act defines a hazardous air pollutant as one "to which no ambient air auality standard is applicable and which in the Judgment of `f'e A.t `-a°.:° ^iiavietvs to, 'aa pivuu 11..iio_a _ wrrrL L__L F.v io.oa.uaauwc :.`.vv, or vva mll rlr4y reasonably be anticipated to result in mortality or an increase in serious irreversible, or incapacitating reversible, illness."" Section 112 requires the EPA Administratow to establish a list of hazardous air pollutants and, within 180 days of listing a substance, to set emission standards for sources "at the level which . . . provides an ample margin of safety to protect the public health."'® The language of section 112 emerged as a compromise from the House-Senate conference committee on the Clean Air Act amendments of 1970.19 The House bill proposed basing national emission standards for hazardous air pollutants on technological and economic feasibility.20 In contrast, Senator Edmund Muskie and his supporters in the Senate 16. (tlt is urgcnt that Congress adopt new clean air legislation which will make possible the more expeditious imposition of sptcific emission standards both for mobile and stationary sources and the effective enforcement of such standards by both State and Federal agencies .... Therefore, particuiar attention must be given to new stationary sources which are known to be either particulary large-scale polluters or where the pollutants are extraha:ardous. H.R. REP. No. 1146, 91st Cong., 2d Sess. 5, reprinted in 1970 U.S. CODE CONG. & AD. NEws 5356, 5360-61. 17. 42 U.S.C. § 7412 (Supp. V 1981). 18. Section 112(b) provides that: ItNA) The administrator shall, within 90 days after December 31, 1970, publish (and shall from time to time thereafter revise) a list which includes each hazardous air pollutant for which he intends to establish an emission standard under this section. (B) Within 180 days after the inclusion of any air pollutant in such list, the Administrator shall publish proposed regulations establishing emission standards for such pollutant together with a notice of a public hearing within thirty days. Not later than 180 days after such publication, the Administrator shall prescribe an emission standard for such pollutant. unless he finds. on the basis of information presented at such hearings, that such pollutant clearly is not a hazardous air pollutant. The Administrator shall establish any such standard at the level which in his judgment provides an ample margin of safety to protect the public health from such hazardous air pollutant. Id. 4 7412(b). 19. H.R. REP. No, 1783, 91st Cong., 2d Sess. 10-12. 45-47, reprinted in 1970 U.S. CODE CONG. & AD. NEWS 5356, 5378-79, 20. The relevant section provided that: (a) For the purpose of preventing the occurrence of significant new air pollution problems arising from or associated with any class of new stationary sources which, because of the nature or amount of emissions therefrom, may contribute substantially to endangerment of the public health or welfare, the Secretary shall from time to time by regulation, giving appropriate con- sideration to technological and economic feasibGty, establish standards with respect to such emissions . . . . (b) Such emission standards shall provide that - (1) If such emissions are extremely hazardous to health, no new source of such emissions shall be constructed or operated, except where (and subject to such conditions as he deems favored a zero-discharge requirement, which would have applied to fewer pollutants than the House bill.21 The final language of the section, how- ever, refers neither to technological feasibility nor to zero discharges.22 This suggests that, while the conferenr=e Cnmmittee exn...-...,..... ....s.,.,. m.t t.m..r.r r. i.s7il- si[leratinnc in dPt_° r sta-.~a--~s- aivrl a lyt eaitil ".+° .~~, at ~, eApc~;t. ir protection to require the absolute elimination of all hazardous emissions. i B. Dilemmas in Implementation EPA's regulatory activity under section 112 over the past thirteen years has been modest.23 Emission standards have been promulgated for necessary and appropriate) the Secretary nutkes a specific exemption with respect to such construction or uperation. (2) In the case of other emissions, any new source of such emissions shall be designed and equipped to prevent and control such emissions to the fullest extent compatible with the available technology and economic feasibility, as determined by the Secretary. H.R. REP. No. 1146, 91st Cong., 2d Sess. 35 (1970). 21. Bonine, The Evolution of 'Technology-Forcing' in the Clean Air Act, (Monograph No. 2116 ENV'T REP. (BNA) 7 (July 25, 1975). The Senate report indicated its determination "that existing sources of pollutants should meet the standard of the law or he closed down, and in addition, that new sources should be controlled to the maximum extent possible to prevent atmospheric emissions." S. REP. No. 1196, 91st Cong., 2d Sess. 2-3 (1970). Later, however, the report says that "liJn writing a relatively restrictive definition of hazardous agents, the Committee recognized that a total prohibition on emissions is a step that ought to be taken only where a danger to health, as defined, exists." Id. at 20. The bill provided in part that: (a) (I) The Secretary shall, within ninety days after the enactment of this section and from time to time thereafter, publish in the Federal Register a list of those air pollution agents or combination of such agents which available materiat evidence indicates are hazardous to the healfh of persons and which shall be subject to a prohibition or emission standard established under this section. (2) Within one hundred and eighty days after the publication of such list, or revision thereof, the Secretary, in accordance with section 553 of tille 5 of thc United States Code, shall publish a proposed prohibition and a notice of a public hearing within thirty days. As soon as possible after such hearing. but not later than six months after such publication, the Secretary shall promulgate such prohibition, unless, based upon a preponderance of evidence adduced at such hearing. he finds within such period and publishes his finding - (Al that such agent is not hazardous to the health of persons; or (B) that a departure from such prohibition for stationary sources will not be hazardous to the health of persons. (3) If the Secretary finds under paragraph (2HA) of this subsection that such agent is not hazardous to the health of persons, he shall immediately publish an emissions standard in qccordance with the procedures established under section 114 of this Act. (4) If the Secretary finds under paragraph (2xB) of this subsection that a depnrlure from such prohibition for any stationary source will not be hazardous to the health of persons, he shall immediately promulgate an emission standard for such agent or combination of agents from any such stationary source to protect the health of persons. Id. at 95-96. 22. See supra note 18. 23. See generally Doniger, Federal Regulation of Vinyl Chloride: A Short Course in Law and Policy of Toxic Substances Control, 7 Ecot_ocv L.Q. 497, 565-85 (1978); Currie, Direct Federal Regulation of Stationary Sources Under the Clean Air Act, 128 U. PA. L. REV. 1389 (1980) (generally discussing regulatory activity under section 112). S9o9t,~~zo.z

400 Harvard Environnrental Law Review [Vol. 8:395 only four substances: beryllium, asbestos, mercury, and vinyl chloride.2' The EPA has listed three additional substances: benzene, radionuclides, .d nd' argu c srse:.ic.u Both EPA and the environmental groups monitoring the agency's actions under section 112 have concentrated on pollutants suspected of causing cancer.26 Thes focus on carcinogens creates a dilemma for the agency because many scientists believe that there are no thresholds for carcinogens - no exposure levels short of zero that are risk free.27 "I'hus, a strict interpretation:of section 112's requirement to provide "an ample margin of safety" would require zero-discharge standards, tantamount to banning the listed substances. Such a strict interpretation of section 112 could be impractical. Many substances subject to regulation under section 112 are important indus- trial chemicals. Zero-discharge limitations on these substances would lead to numerous plant closures and the loss to consumers of many valuable products.2® Consequently, EPA has avoided a strict interpreta- tion of section 112 and instead has proposed standards requiring the degree of control achievable with the "best available technology" (BAT).29 Standards promulgated by the EPA for asbestos and vinyl chlo- ride illustrate the agency's dilemma and its eventual decision to base control requirements on technological feasibility. In 1971, EPA proposed standards for asbestos because of its link to a form of cancer known as asbestosis.10 Public comments on the proposed standards revealed no scientific doubt about asbestos hazards, but also stressed the importance of asbestos to the economy." Although the EPA maintained that the• final standard "was not based on economic considerations"`= and that "the overriding considerations are health ef- fects,"" the preamble to the standard acknowledged the dilemma: 24. 40 C.F.R. §§ 61.20-34, 61.50-.55 (1979) (promulgating emissioh standards for asbestos, beryllium and mercury): 40 C.F.R. §§ 61.613-.71 (1979) (promulgating emission standards for vinyl chloride). 25. 42 Fed. Reg. 29,332 (1977) (listing benzene as a hazardous air pollutant); 44 Fed. Reg. 76.738 (1979) (listing radionuclides as a hazardous air pollulant); 48 Fed. Reg. 33.112 11983) (listing inorganic arsenic as a hazardous air pollutant). 26. See Statement by W. Ruckelshaus, Administrator of the EPA, Before the Sub- comm. on Health & the Env't of the House Comm. on Energy & Commerce 10 (Mar. 29, 19841 Mazardous Air Pollutants 401 EPA considered the possibility of banning production, processing, and use of asbestos or banning all emissions ... into the atmosphere, but reiected these approaches .... Either aoproach would recult irn the pe•~t,a:,(:or o( many activities which are extremely important; moreover, the available evidence relating to the health hazards of asbestos does not suggest that such prohibition is necessary to protect public health.'° , The effect of this dilemma on EPA action is indicated by the fact thiat the agency did not even adopt this compromise standard until 1973 (well beyond the 180-day limit), and then only after a court order." ' The language of the vinyl chloride standard, promulgated in October 1976,'6 provides an even clearer indication of the adoption of a technol- ogy-based approach. In the proposed regulation, EPA interpreted section 112 as allowing it to set standards: that require emission reduction to the lowest level achievable by use of th,! best available control technology in cases involving apparent non-threshold pollutants, where complete emission prohibitions would result in wide- spread industry closure and EPA has determined that the cost of such closure would be grossly disproportionate to the benefits of removing the risk that would remain after imposition of the best available control - technology." Thus, although section 112 mentions only health effects, and a literal reading might require that all emissions of non-threshold pollutants be banned, the EPA developed an accommodation that bases control on technological feasibility. EI'A did not identify guidelines for listing substances under section 112 in its standards for asbestos or vinyl chloride. Asbestos and vinyl chlor(de presented clear cases of proven carcinogens, but over fifty other substances are identified only as potentially hazardous air pollutants." In contrast, many toxic water pollutants were listed (and a schedule for developing regulations established) in 1976 as part of a consent decree with the Natural Resources Defense Council." Environmental groups became dissatisfied with the slow pace at which the agency was listing substances and promulgating standards under section 112.1 In November 1977, the Environmental Defense Fund 1984) thereinafter cited as 1984 Statement by Ruckelshaus). 27 I d l U 448 U S 1980 S ri i D A ri P l i 607 624 . ust a . ). ee n n on ept. v. me etro eum nst., . ( can , 28. 1984 Statement by Ruckelshaus, supra note 26, at 13. 34. 38 Fed. Reg. 8820, 8822 (1973). 29. As discussed in more detail below, a "generic" policy proposed in 1979 would 35. Id. have formalized the agency's implicit policy of requiring, at a minimum, BAT controls for 36. 40 C.F.R. §§ 61.60-.71 (1976). sources emitting pollutants listed under section 112. See iry/'ra text accompanying notes 43- 37. 40 Fed. Reg. 59,534 (1975). 50. 38. 44 Fed. Reg. 58,642, 58.643 (1979). 30, 40 C.F.R. §§ 61.20-.25 (1971). 39. Natural Resources Defense Council v. Train, 8 Env't Rep. Cas. (BNA) 2120 31. 38 Fed. Reg. 8820, 8822 (1973). (D.D. C. 1 976). 32. Id. 33. 40 C.F.R. §4 61.20-.25 (1971). 40. See, e.g.. Doniger. supra note 23, at 565-85 (discussing the politics underlying EPA's promulgation of a vinyl chloride standard). ~9m9Vssza~~

402 Harvard Environmental Law Review [Vol. 8:395 19841 Hazardous Air Pollutants 403 (EDF) filed a petition requesting that EPA establish the terms of the vinyl chloride agreement as a generic approach to the regulation of all carcin- og_ens." Finally, in October 1979, EPA proposed a cancer policy entitled "roiicies and Procedures for idcniifyirtg, Assessing, aild negiiiatii~E; i.ir- borne Substances Posing a Risk of Cancer."'Z Although the proposed poliCy was never promulgated, a review of its provisions provides an indication of the procedures that evolved over the first decade of section 112's existence. C. Cancer Policy The most important features of the EPA's proposed "cancer policy" involved the criteria for listing substances and the criteria for setting standards for source categories.•1 The proposal established a relatively low hurdle for listing; EPA would list any substance having a high prob- ability of carcinogenicity unless there was no evidence of a significant threat of ambient exposure from emissions by stationary sources." Upon listing, a set of generic regulations including maintenance, storage, and "housekeeping" requirements would immediately apply to sources emit- ting the substance.'S For each listed substance, the EPA would prepare detailed estimates of health effects and •use those estimates to set priorities to develop emission standards for individual source categories posing the most im- minent threat to the public health." The emission standards would, at a minimum, require BAT controls. The procedures for determining BAT do not involve risk asSessment. Quantitative risk estimates would, how- ever, be employed in the standard-setting process if they showed that the residual risk after BAT controls was "unreasonable." In such a case, EPA would impose tighter controls.47 41. See Doniger, supra note 23, at 584. 42. 44 Fed. Reg. 58,642 (1979). The proposal was part of a larger effort by the Carter administration to develop regulatory policies for carcinogens. A controversial cancer policy proposed by the Occupational Safety and Health Administration (OSHA) preceded the EPA document, see 45 Fed. Reg. 5002 (1980). In addition, the heads of the four major regulatory agencies dealing with carcinogens had formed the Interagency Regulatory Liai- son Group. That group had a mandate to develop a greater scientific consensus on cancer risk assessment procedures. Id. at 58,647. Finally, in 1979 EPA was developing regulations on benzene emissions under section 112 to be used as a prototype for the procedure the agency was elaborating in its generic policy. Indeed, when the White House Regulatory Analysis Review Group selected the EPA cancer policy for review, the agency suggested that the group use benzene as an indicator of how the policy would be implemented. See Nichols, The Regulation of Airborne Benzene. in INCENTIVES FOR ENVIRONMENTAL PRO- TErrtoN 148 (T. Schelling ed. 1983). 43. See 44 Fed. Reg. 58,642 (1979). 44. Id. at 58,654. 45. Id. at 58.648. See also 44 Fed. Reg. 58,662-70 (1979). 46. 44 Fed. Reg. 58,642, 58,654 (1979). 47. Id. In sum, EPA's record in implementing section 112 has consisted of much study and little regulation. The proposed cancer policy did create a methodology that would have allowed vastly greater listings, but would aiso have severely iirniied EPA discrr;iion in setting specinc standards for listed substances." In the last several years, the EPA has continued to analyze potential section 112 pollutants, but has not listed any pew substances, nor proposed new standards for substances previously listed, nor promulgated standards proposed earlier.'s The following statement made by David Patrick, the chief of the Pollutant Assessment Branch in the Office of Air Quality Planning and Standards at EPA, illustrates the concerns of the agency: All have perceived that a literal interpretation of section 112 would not preclude open-ended control requirements or the possibility of zero emis- sion goals, regardless of the control costs. Given this potential and the apparent lack of flexibility regarding the removal of substances from the list of hazardous pollutants or the exclusion of source categories from control requirements, the Agency has also been reluctant to list pollutants as hazardous without some reasonable assurance that subsequent regula- tions would convey health benefits that are not grossly disproportionate to the costs of control.w D. Recent Congressional Debate In the current debate on reauthorization of the Clean Air Act, en- vironmental groups have criticized EPA's review process as "slow and repetitive."S' The Environmental Defense Fund has urged Congress to: (I) adopt a generic method for listing airborne carcinogens; (2) list the thirty-seven substances now under study; and (3) require that EPA -de- velop a systematic regulatory approach that includes literature reviews, periodic reports, and time limits for action.S2 In contrast, the Chemical Manufacturers Association (CMA) advocates modifying section 112 to allow EPA to regulate only those substances that pose a significant risk to health and to consider social, technical, energy, and economic con- sequences in setting standards.53 Finally, EPA Administrator Ruckels- haus advocates a regulatory strategy that is based on the balancing of 48. See supra text accompanying notes 45-47. See also Harrison, supra note 4, at 112-13. 49. See [14 Curr. Dev.) ENV'T REP. (BNA) t 109-11. But see infra notes 216-220 and accompanying text (discussing the recent developments in regulation under Section 112). 50. See D. Patrick, Air Toxics: Regulation and Research 3 (Apr. 6, 1982) (speech presented at the Air Pollution Control Association (APCA) Conference, Houston, Tex.). See also Harrison, supra note 44, at 112-13 (critiquing EPA's proposed cancer policy). 51. D. Doniger, Statement on Behalf of the National Clean Air Coalition, Before the Subcomm. on Oversight & Investigations of the House Comm. on Energy & Commerce 10 (Nov. 7, 1983). 52. Doniger, supra note 23, at 579-84. 53. (1I Curr. Dev.) ENV'T REP. (DNA) 1026 (1981). 490~VS~92:oz

404 Harvard Environmental Law Review (Vol. 8:395 many factors including the nature of the risk posed by a substance and the cost of eliminating or minimizing it.' The eventual result of this debate over Sectinn It? eAn-t y et '`e detetmined. Thus far, however, sentiment in the House seems to favor swifter, more aggressive regulation of airborne carcinogens. In August 0982', the House Energy and Commerce Committee voted in favor of an amendment requiring that, in each of the next four years, EPA review twenty-five percent of the thirty-seven substances discussed earlier.ss The amendment would create a presumption In favor of listing; each of the thirty-seven substances would be listed automatically unless EPA deter- mined that it was not hazardous,'6 If this provision, or a similar one, is enacted, the pace of regulation under section 112 shoukd reach substan- tially higher levels than ever before. II. THE CASE STUDIES A. Steps in Estimating Benefits These studies estimate the benefits of pollution control standards by tracing the links from tmissions to exposure to risk, The purposes of the analysis are either to estimate the dollar value that affected parties place on the reduced risk or to use the risk estimates to calculate the implicit cost per statistical life saved. The steps used, presented schematically in Figure I, apply in assessing the benefits of controlling virtually any dangerous pollutant. The following discussion provides a general over- view of the calculations associated with each step in the context of regulating airborne carcinogens. The change in emissions due to regulation is the most straightforward of (he calculations that produce benefit estimates." For each plant, the Exposure , Risk Dollar Valuation 54. 1984 Statement by Ruckelshaus supra note 26, at 14. 55. 113 Curr. Dcv.) ENV'T REP. (BNA) 491 (1982). 56. Id. 57. Bur see infra notes 141-51 and accompanying text (discussing the uncertainties inherent in (his analysis). 19841 Ilazardous Air Pollutants 405 EPA estimates the emissions with and without controls in place.s' The differenec betwae^ (hese two esiiniidies equais the emissions redne.tin.; attoib:;table tv ilic reguiation imposed. Emissions reduction estimates are converted into more meaningful estimates of exposure reductions by calculating an "exposure factor" for individual plants.'y The exposure factor indicates the amount of exposure caused by a unit of emissions from a particular source.6" E3oth the dits- persion pattern of emissions and the population pattern in the area sur- rounding the plant contribute to calculating this factor. In many cases, EPA estimates emissions dispersion using a "model plant."61 For a given level of emissions, the dispersion model uses me- teorological data to generate estimates of average annual pollutant con- centrations at various distances from the source. The estimated con:.en- trations are then combined with plant-specific population data to estimate total exposure levels for a given level of emissions. Exposure levels are expressed in terms of "µg/m'-person-years," which is simply the average annual concentration (in micrograms per cubic meter) multiplied by the number of people exposed and the period of exposure.°' This summary measure of exposure provides sufficient information to predict total risk under certain conditions."' Dividing the exposure level by the total level of emissions gives the exposure factor, expressed in terms of µg/m'-person-years per kilogram emitted. Reduced exposure is translated into reduced risk using the unit risk factor for the particular pollutant. A unit risk factor represents the risk of cancer posed by exposure to one unit of a substance - measured as the risk of cancer per µg/m'-person-year.°' Eash of the three case studies used unit risk estimates prepared by EPA's Carcinogen Assessment Group (CAG). The CAG unit risk estimate meastires the increased probability of cancer resulting from exposure to I µg/m' for a lifetime.*" This figure divided by seventy equals the risJc of 58. See Ofnce of Air Quality Planning & Standards, U.S. Env(I. Protection Agcncy, Benzene Emissions from Maleic Anhydride Industry - Background Information for Pro- posed Standards, Table 1-5 (Feb. 1980 draft) thereinafter cited as Benzene Emia•ions Background Informationl. 59, If a plant with an exposure factor of 0.6 µg/m'-person-years/kg reduces its emis- sions by I million kilograms, for example, exposure falls by 0.6(1,000.000) = 600,000 µg/ m'-person-years. 60. See Nichols, supra note 42, at 187-88. 61. See. e.g., Benzene Emissions Background tnformation, supra note 58. at E 8. 62. Thus, for example, 1000 people exposed, on average, to 10 µg/m' for one year generate 10,000 µg/m'-person-years of exposure, as do 10.000 people exposed to I µg/m'. 63. Such risk is independent of how total exposure is distributed across the popul:1tion if risk is proportional to exposure. See iq/Ya notes 164-171 and accompanying text. 64. The risk of getting cancer obviously varies with the carcinogenicity of thc sub- stance. See ir(/'ra notes 164-183 and accompanying text (discussing the difficulties oi ex- trapolating from low to high doses). 65. CAG considers a lifetime to be seventy years; hence, in this study the CAG's estimated exposure factor is dividcd by seventy to obtain an annual estimate. See, e.g., Carcinogen Assessment Group. Office of Health & Envtl. Assessment, U.S. Envtl. Pro-

4cm, alarvard Environmental Law Review [Vol. 8:395 19841 Hazardous Air Pollutants 407 cancer per µg/m'-person-year. In applying epidemiological data, the CAG employs a procedure that assumes that risk remains proportional to dose at low levels of exposure.' Over the past decade or two, a substantial literature has accumulated on the iccue of ya_Iuino rrdtoMinna in ~g4V t.. u' :,47 .°wnp,iusis agree that t; e aff„':rp~iste critcrivn is "wiiiingness to pay."°" The principle is a simple one: an individual values each benefit just as much as the amount he would be willing to pay to secure it. Inferences drawn jrom actual behavior provide the best estimates of willingness to pay. Many studies have estimated willingness to pay for reduced risks to life based on the wage premiums associated with occu- pational risks.69 Bailey has reviewed several empirical studies,, adjusting them for consistency.70 His estimate covers a_ range of $170,000 to $715,000 per life saved,*with an intermediate estimate of $360,000 in 1978 dollars, or approximately $500,000 in 1982 dollars." Other studies, how- ever, have estimated much higher wage premiums for occupational risks, with the highest estimates in excess of $5 million per life saved in 1982 dollars.72 Thus, the published estimates from wage studies range from several hundred thousand dollars to several million dollars per statistical life saved. Many of the calcuiations in this article forgo the final step of placing a dollar value on lives saved and presenting a single net benefit result. However, estimates of the reductions in lives saved and the implicit cost per statistical life saved are presented. These results are then compared tection Agency, Carcinogen•Assessment Group's Final Report on Population Risk to Am- bient Benzene Exposures 12 (1977) Ihereinafter cited as Final EPA Benzene Assessment I. 66. :d. at 2. 67. See, e.g., Zeckhauser, Procedures for Valuing L(f'e, 23 Pt1n. PoL'Y 419 (1975); Graham & Vaupel, Value of a Ljfr: What Djfjerence Does it Make7, I RISK ANALYSIS 89 (1981). 68. See Schelling, The Life You Save May Be Your Own, in PROBLEMS IN PUBLIC EXPENDITURE ANALYSIS 127, 142-58 (S. Chase ed. 1968). Schetling is generally credited with being the first to argue that willingness to pay for risk reduction is the appropriate conceptual approach to valuing "life saving." A slightly different formulation, which should yield virtually identical results when dealing with small risks, is to ask how much money an individual would have to receive to forgo the benefit. The technical terms for these two measures are "compensating variation" (CV) and "equivalent variation" (EV). In general, when discussing risk reductions, EV (how much money an individual would have to receive to be willing to go without the risk reduction) will exceed CV because of income effects. For small changes in risk, however, the differ- ences between the two measures will be negligible. 69. See Thaler & Rosen. The Value of Saving a Ljfe: Evidence from the Labor Market, in HOUSEHOLD PRODUCTION AND CONSUMPTION 265-301 (N. Terleckyj ed. 1976); G. Blomquist, Valuation of Life: Implications of Automobile Seat Belt Use (1977) (Ph.D. dissertation, University of Chicago); A. Dillingham, The Injury Risk Structure of Occu- pations and Wages (1979) (Ph.D. dissertation, Cornell University). 70. See M. BAILEY REDUCINO RISKS TO LIFE it app. 35-45, 52-66 (19go). , 71. ld. at app. 66 (Bailey's estimates are based on Thaler & Rosen, supra note 69; j supra note 69; and Dillingham Blomquist supra note 69), , , I 72. See Viscusi, Labor Market Valuations of Ljfe and Limb: Empirical Evidence and ` Policy Implicarions, 26 Pu®. POL'v 359 (1978). ~ with reasonable estimates of the value of this risk reduction to determine if the regulation is likely to pass a benefit-cost test. R. T he Case Studie.r Benzene, coke oven emissions, and acrylonitrile are all high-prio)ity section 112 pollutants. Benzene has been listed formally" and regulati®ns have been proposed," and recently re-proposed, for several source cat- egories.'s Coke oven emissions and acrylonitrile are included in a list of thirty-seven substances the EPA is currently evaluating.76 The health risks of and control options for these pollutants are well documented." Although the following case studies use a common underlying method- ology to estimate the benefits of controls for all three pollutants, the empirical details of the methodology vary considerably with each pollutant. This section presents the results of benefit-cost analysis in each of the three case studies. The next sections suggest two approaches as alternatives to uniform BAT standards: (I) modification of the uniform standards to increase net benefits and (2) differential standards based on exposure levels around individual plants.78 1. Maleic Anhydride (Benzene) Case Study" Maleic anhydride plants emit benzene, a major industrial chemical used in making nylon, plastics, insecticides and polyurethane foams,7O A 1977 study by the National Institute of Occupational Safety and Health showed an abnormally high incidence of leukemia in workers exposed to benzene while employed at two plants in the rubber industry." Following this study, the EPA listed benzene under section 112."= - 73. 42 Fed. Rcg. 29,332 (1977). 74. 45 Fed. Reg. 26.660 (1980). 75. 49 Fed. Reg. 8386 (1984). 76. D. Patrick, supra note 50, app. on Section 112--The Process and Status. 77, See, e.g.. Office of Health & Envtl. Assessment, U.S. Envtl. Protection Agency, Health Assessment Document for Acrylonitrilc (Mar. 1982) (draft) Ihereinafter cited as Acrylonitrile Assessment Document]. 78. See infra notes 114-40 and accompanying text. 79. Maleic anhydride plants convert benzene into maleic anhydride - a crystalline cyclic acid anhydride used chiefly in manufacturing resins and modified drying oils. The primary source of data for this case study is Benzene Emissions Background Information, supra note 58. For additional sources, see Nichols, supra note 42. The analysis is based on data available to EPA when it proposed the standard for maleic anhydride plants in April 1980. Since then, however, several new developments have led EPA to propose the withdrawal of the proposed benzenc control standards. See infra text accompanying notes 216-18. 80. See S. Mara & S. Lee, Assessment of Human Exposure to Atmospheric Benzene 21 (May 1978) (report prepared by SRI International for U.S. Envti. Protection Agency) lhereinafter cited as Human Exposure to Benzenel. 81. See Infante, Leukemia in Benzene Workers, 2 LANCET 76 (July 9, 1977). See also Nichols, supra note 42, at 149-50 (summarizing the studies of benzcne's health effectti). 82. 42 Fed. Reg. 29,332 (1977). After listing thc pollutant, EPA commissioned studies Gqoqv!-~~_S_-Sz®z

408 Harvard Environmental Law Review (Vol. 8:395 In April 1980, almost three years after listing benzene, EPA proposed an emission standard for maleic anhydride plants that use benzene as a fPn„rictnr4 8~ The BAT ctn,n,dnrrt called a i ,'- , r an aiissivi~s icUU(:LIUtI of roughiy nineiy-seven percent from uncontroiied levels.5° A majority of the plants, however, already had installed controls of ninety percent or better; probably in response to state regulations directed at hydrocarbons or the hope that the benzene recovered would pay for the controls.eS As a result, the proposed B~kT standard was expected to reduce full-capacity emissions by less than ninety percent, from 5.6 million kilograms per year to just under 0.5 million kilograms per year.'6 The costs of implementing the proposed standard were estimated at $2.6 million per year in 1982 dollars.'7 These costs are quite affordable to the maleic anhydride industry, whose total sales grossed $142 million in 1979."r The cost estimates are meaningless in isolation, however; they can be judged appropriately only in relation to the benefits they secure. As estimated, the proposed regulations would have reduced exposure by 3.6 million µg/m'-person-years and saved 0.4 lives annually."" 2. Coke Oven Emissions Case Study9O Coke, produced by distilling coal in ovens, is essential to the pro- duction of iron and steel. In 1978, U.S. plants produced approximately of benzene emissions. See PEDCo Environmental, Inc.. Atmospheric Benzene Emisssions (Oct. 1977) (report submitted to U.S. EPA) (EPA-450/3-77-029) [hereinafter cited as At- mospheric Benzene EmissionS); S. Mara & S. Lee. Human Exposures to Atmospheric Benzene (Oct. 1977) (report prepared by Stanford Research Institute for U.S. EPA): Human Exposure to Benzene, supra note 80. These studies provided a rough idea of the relative amounts of pollution contributed by different types of sources. See also Nichols, supra note 42. 83. 45 Fed. Reg. 26,660 (1980). EPA developed an emission standard for maleic anhydride plants first, because more than half of all estimated emissions from chemical manufacturing plants came from the eight plants that used benzene to produce malcic anhydride. See Atmospheric Benzene Emissions. supra note 82. Table 1-2. 84. The standard limited existing plants to 0.3 kg of txnzene entitled per 100 kg of benzene input. 45 Fed. Reg. 26,669 (1980). 85. See Benzene Emissions Background tnformation, supra note 58, Table I-5. 86. 45 Fed. Reg. 26,66(/, 26,661 (1980). 87. Id. at 26,666. See also Benzene Emissions Background tnformation, supra note 58. Fur the two plants that had 90% controls, however, the cost estimates assume that they would need all-new control equipment; no credit is given for possible adaptation of existing controls. All of the cost estimates arc for carbon absorption controls, which the EPA estimates indicated would be the lowesl-cost control technique (including a credit for bcnzene recovered), and all assume 100% capacity utilization. 88. Facts and Figures for the Chemical Industry, CHEMICAL AND ENGINEERING NEws 26, 31 (June 13. 1982). The costs estimates included credits for the bcnzene recovered. 89. See Haigh, Harrison & Nichots, supra note 12, at 25-28. 90. The primary sources for the coke oven emission case study arc: Emission Stan- dards & Eng'g Div., Office of Air Quality Planning & Standards, U.S. Envtl. Protection Agency, Preamble and Regulation for Coke Oven Emissions from By-Product Coke Oven Charging, Door Leaks, and Topside Leaks on Wet-Coal Charged Batteries I (Mar. 1981) 19841 Hazardous Air Pollutants 409 44 billion kilograms of coke.91 Epidemiological studies of coke-oven workers show that emissions from the coking process increa_sed the ;isks ~ ~.-- vg I uug, irachea, bronchus, kidneV_ and nrncrate cµ nccos4? AltltOnlsh the toxic elements include-gases and respirable particulate matter, most at- tention has focused on the polycyclic organic matter (POM) contained in coal far particulates.yB Coke oven emissions are released from numerous fugitive sottrt±es, including leaks and imperfections in the ovens. Charging emissions occur when coal is added to the ovens at the~ beginning of the coking process. Door leaks are the result of imperfect fits between the ovens and the doors through which the finished coke is later removed. Finally, imp,:rfect seals on the lids and offtakes on the tops of the ovens create topside leaks.w If the EPA listed coke oven emissions under section 112, the Agency would probably specify standards similar to the following as BAT: twelve percent of doors visibly leaking; three percent of lids visibly leaking and six percent of offtake systems visibly leaking; and sixteen seconds ol' visible emissions for each charging.9y EPA estimates suggest that only thirty-seven of the fifty-four identified coke plants would have to increase control efforts to meet these standards (and some of those plants already meet one or two of the three potential BAT standards).' EPA estimates annual control costs for those plants at $24.5 million.y' Plant-specific emission estimates indicate that coke oven emissions would fall by 289,000 kg/year and exposure would fall by approxitr<ately (draft) (Research Triangle Park, N.C.) [hereinafter cited as 1981 EPA Draft Coke Oven Regulatitml: Office of Air Quality Planning & Standards, U.S. Envtl. Protection Ai{ency, Coke bvcn Emissions from By-Product Coke Oven Charging, Door Lcaks. and Topside Leaks on Wet-Coal Charged Batteries - Background Information for Proposed Standards (Juty 1981) (draft) (Research Triangle Park, N.C.) Ihereinaftcr cited as 1981 nackFround Informationl; Carcinogen Assessment Group, Office of Health and Envtl. Asses•ment, U.S. Envtl. Protection Agency, Carcinogen Assessment of Coke Oven Emission /Feb. 1982) (draft) (EPA-600/6-82-003) [hereinaRer cited as EPA Coke Oven Assessmeml: and Research Triangle Institute, Cost Estimates of Meeting the Potential EPA Regulation Affecting Coke Oven Emissions from By-Product Coke Oven Charging, Door Leaks. and Topside Leaks on Wet-Coal Charged Batteries (Apr. 1983) (computer printout) [heremafter cited as 1983 Research Triangle Cost Estimatcl. 91. See 1981 Background Information, supra note 90, at 3-2. 92. See, e.g.. EPA Coke Oven Assessment, supra note 90, at 108-12. 93. Id. at 54-63: 94. 1981 EPA Draft Coke Oven Regulation, supra note 90, at 4. 95. Id. at 4-S. 96. A detailed breakdown of the status of individual plants is not available. The cost data supplied by the Research Triangle Institute, the primary EPA contractor for the coke oven analyses, includes positive entries only for those plants that arc expected to require controls if standards are promulgated. Personal communication from Phillip Cooley of Rescarch Triangle Institute (Aug. 1983). 97. 1983 Research Triangle Cost Estimate, supra note 90. EPA's emission and cost estimates are stated in terms of 1982 dollars and assume current compliance with existing state and OSHA regulations. Id. 0409VSRZ®Z

410 Harvard Environmental Law Review (Vol. 8:395 19841 Hazardous Air Pollutants 411 819,000 µg/m3-person-years if the above BAT standards were imposed." Coke oven emissions are very potent carcinogens; this relatively slight reduction in exposure would save an estimated 10.6 lives each year.9' i• A~rv~nnilr:/o ('.. Acrylonitrile is an important industrial feedstock, employed primar- ily in the production of chemicals used to make a wide range of common products including rugs, clothing, plastic pipes, and automobile hoses.101 Almost a billion kilograms of acrylonitrile were produced in 1981.102 Extensive evidence indicating acrylonitrile's carcinogenicity exists.10' Specifically, epidemiological studies have associated acrylonitrile with respiratory cancers.104 While EPA has neither listed acrylonitrile nor proposed specific regulations, EPA contractors have identified available control options that could reduce emissions by at least ninety-five percent from uncon- trolled levels.10S All thirty existing plants, however, already have imple- mented some type of controls. Thus, potential BAT standards would only cut annual emissions from 3.6 million kilograms to 0.5 million kilograms, a reduction of slightly less than eighty-seven percent.10R Uniform controls 98. Haigh, Harrison & Nichols, supra note 12, at 32-34. 99. EPA Coke Oven Assessment, supra note 90, at 144-63. See also infra Table I. 100. The acrylonitrile case study relied on data assembled from several sources, including Click & Moore, Emission, Process and Control Technology Study of the ABS/ SAN Acrylic Fiber, and NBR Industries (Apr. 1979) (report prepared by Pullman Kellogg for the Office of Air Quality Planning & Standards, U.S. EPA, contract 68-02-2619): Key & Hobbs, Acrylonitrile (Nov.' 1980) (report prepared by IT Enviroscience for the Office of Air Quality Planning & Standards, U.S. EPA); Energy & Envtl. Analysis, Inc.. Source Category Survey for the Acrylonitrilc Industry (July 1981) (draft report prepared for the Office of Air Quality Planning & Standards, U.S. EPA, under contract 68-02-3061); Radian Corporation, Locating and Estimating Air Emissions from Sources of Acrylonitrile (Dec. 1982) (draft report prepared for Office of Air Quality Planning & Standards, U.S. EPA); Carcinogen Assessment Group, Office of Health & Envtl. Assessment. U.S. Envtl. Pro- tection Agency, The Carcinogen Assessment Group's Carcinogen Assessment of Acrylo- nitrile (Feb. 1982) (draft) (hereinafter cited as EPA Acrylonitrile Assessment); B. Suta, Assessment of Human Exposure to Atmospheric Acrylonitrile (Aug. 1979) (report prepared by SRI Int'l for U.S. EPA) (hereinafter cited as 1979 Assessment of Exposure to Acrylo- nitrilel; B. Suta, Revised Assessment of Human Exposure to Atmospheric Acrylonitrile Using Industry Supplied Emission Estimates (1982) (report prepared by SRI Int'l for U.S. EPA); and personal correspondence from B. Suta (Aug. 1982) (data on exposure to acry- lonitrile emissions) (hereinafter cited as Suta Data on Acrylonitrilel. 101. Energy and Envtl. Analysis, Inc., supra note 100, at 3-1. 102. Facts and Figures for the Chemical Industry, CHEMICAL AND ENGINEERING NEws 30, 37 (June 14, 1982), 103. EPA identified three epidemiological studies; seven lifetime laboratory studies with rats; several mutagenicity studies with bacteria, Drosophila (fruit flies), and rodents; chromosomal studies of humans; and numerous metabo6c studies. Carcinogen Assessment Group, Office of Health & Envtl. Assessment, U.S. Envtl. Protection Agency, Health Assessment Document for Acryolnitrile 101 (1982). 104. See EPA Acrylonitrile Assessment, supra note 100, at 1, 63-67. 105. See Key & Hobbs, supra note 100, ch. V, at 1-4, ch. VII, at 1-3 (discussing such control systems). 106. These calculations of emission reductions are based on "current" emissions in would create an estimated annual expense of almost $29 million in 1982 dollars.107 Reduced exposure to acrylonitrile, just over 450,000 µg/m'- person-years, would avoid only one case of cancer every five years 1,0,2 l' veJ " pc- i}/edl J.t_M '.. C. Analysis oJthe Best Available Technology Standards 1 1 Table I summarizes the results of the BAT standards analyzed. Controls on coke oven emissions produce much greater health benefits than do controls on the emissions of benzene or acrylonitrile. BAT con- trols on coke ovens would result in almost eleven fewer cases of cancer each year, compared to reductions of 0.4 cancer deaths for maleic an- hydride benzene controls and 0.2 cancer deaths for acrylonitrile standards. The final line of Table I presents the most relevant figure in mea- suring the cost-effectiveness of the three control standards - the value placed on saving a life that is necessary to justify incurring control costs. To justify acrylonitrile controls on benefit-cost grounds, the value of a statistical life would have to be at least $144 million, an implausible figure from virtually any perspective.1O/ The cost-effectiveness figure for ben- zene. $6.5 million, also is larger than the range of plausible estimates. Controls on coke oven emissions are the most attractive of the three BAT options. To justify the coke oven emissions standards on benefit- cost grounds, the value of a life saved must be equal to or greater than $2.3 million. That value does fall within the range ol, the published benefit estimates. Nevertheless, all three BAT options would fail a conventional benefit-cost test based upon a value of $1 million per life saved. Table 2 indicates two principal reasons why the cost-effectiveness of control varies so greatly among the pollutants. First, the carcinogenic potency of coke oven emissions is much greater than for acrylonitrile or for benzene.10 Second, coke oven emissions affect many more people than do the other pollutants. Fugitive coke emissions occur at ground U.S. Envtl. Protection Agency, Summary of Acrylonitrile Emission Estimates and Pro- duction Capacities (Jan. 1983) (draft) (tables provided by R. Crume, Office of Air Quality Planning & Standards), and on model-plant controlled emissions in Key & Hobbs, supra note 100, ch. V, at 1-4 for AN monomer and in Click & Moore• supra note 100, at 61-64. See also Haigh. Harrison & Niehols, supra note 12, at 36-38. • 107. The control costs are estimated from model plant data in Key & Hobbs, supra note 100, at Table VI-2, and new plant data in Energy and Envtl. Analysis, Inc., supra note 100, at Table 5-5 for AN monomer and in Click & Moore, supra note 100, at Table 6-1, for the other categories. All costs have been updated to 1982 dollars using the GNP implicit price deflator. See also Haigh. Harrison & Nichols, supra note 12, at 36-38 and Table 2.11. 108. See Haigh, Harrison & Nichols, supra note 12, at 36-41. Exposure factors were estimated using dispersion modeling results and plant-specific population data provided in Suta Data on Acrylonitrile, supra note 100. 109. See supra notes 67-72 and accompanying text. 110. See infra Table 2 (indicating carcinogenic potency with unit risk factors). See also supra notes 64-66 and accompanying text. 1/C.a09~SszOz

412 Harvard Environmental Law Review Table 1: Benefits and Costs of BAT Standards Vol. 8:395 Benzene" Coke Ovens i.ci-yio- nitrile Annual Costs and Benefits Control Cost ($1000) 2,577 24,511 28,988 Number of plants 0 37 31 Reduced Emissions (1000 kg) 5,059 289 3,112 Reduced Exposure (1000 µg/ m'-person-yrs)° • 3,646 819 455 Lives Saved` 0.4 10.6 0.2 Cost-Effectiveness Emissions (S/kg) 0.51 84.8 9.3 Exposure (S/µg/m'-yr) 0.71 29.9 63.7 Lives saved (SI million/life) 6.5 2.3 144. Notes: a. Estimates are based upon the 1980 pproposed standard for maleic anhydride ptants, b. Exposure reductions are calculated by aggregating the concentration changes for people at different distances from each plant. For example, if 1000 people have their exposure reduced by 10 micrograma per cubic meter (µgW) in a given year, exposure would be reduced by 10,000 µg/m'-person-years. c. Lives saved are calculated by multiplying the exposure reduction by a unit risk factor that measures the increased probability of contracting cancer as a result of exposure to I µg/m' for one year. For example, if exposure is reduced by 1000,000 µg/m'-4er-years for a carcinogen that increases the risk of cancer by 1.5 x 10-4 for each µg/m -per-year, a total of 15 statistical lives would be saved. (Note: this article assumes that all cancer cases result in premature death.) level rather than from stacks, and coke plants tend to be located closer to large population concentrations."' As a result, a kilogram of coke oven emissions causes three times the exposure that a kilogram of ben- zene emitted from maleic anhydride plants does and over seventeen times the exposure that a kilogram of acrylonitrile does.12 Because of these two factors, a reduction of one kilogram in coke oven emissions produces a risk reduction roughly 500 times greater than for either of the other cases."' Together, tables I and 2 indicate that concentrating only on the cost per kilogram of emission reduction provides a misleading measure of the relative attractiveness of the three BAT standards. A kilogram of coke oven emissions is much more costly to control than a kilogram of either acrylonitrile or benzene. The marginal benefit of controlling coke oven emissions is so much larger, however, that coke ovens are far more cost- effective objects of regulation. This comparison gives the most compel- I 11. See irtfra Table 2 (comparing population figures across the three case studies). 112. Id. (comparing average exposure factors across the three case-study pollutants). 113. Id. (comparing risk per kilogram of emissions across the three case-study pollutr 1984) Hazardous Air Pollutants 413 i Table 2: Risk and Exposure Information for the Three Cases Coke Oven Benzene' Emissions Acrylonitrile Unit risk factor (deaths/ 1.3 x µg/mD-yr)b I.1 X 10-7 10-s 4.4 x 10-' Total population exposed` 8,080,000 25,948,0(ED 8,457,D0® Population within I km 27,550 90,193 7,138t Average exposure factor (µg/m'-person-yrs/kg)° 0.721 ' 2.83 0.1q(, 3.7 x Risk per kg of emissions 7.9 X 10-a 10-5 6.4 x 10 " Notes: a. Estimates are based upon the 1980 proposed standard for malcic anhydride plants. b. See footnote c, Table I. c. Population within 20 km of all plants. d. The exposure factor is calculated by dividing the reduced exposure by the reduced emissions. For example, the calculation for coke oven emissions is: 819,000 µg/m'-person- years divided by 289,000 kg, which equals 2.83. ling reason for formally evaluating the benefits of toxics control, It is impossible to target controls where they provide the greatest health benefits without considering relative carcinogenicity and relative expo- sure factors. D. Analysis of Alternate Standards Benefit-cost criteria assist policymakers in evaluating regulatory al- ternatives beyond uniform BAT standards as well. This section analyzes two alternatives for each pollutant: (1) a relaxed uniform standard; and (2) a set of differential standards that would be more stringent for plants located in more densely populated areas than for plants that cause less exposure. Choosing the appropriate degree of control is a common issue in pollution regulation. "' Controls should be tightened as long as the mar- ginal benefits exceed the marginal costs. Negative net benefits at one control level do not imply that regulation is undesirable at all levels, because a less stringent alternative may provide positive net benefits. Pollution control regulations can also be targeted to specific fi,-ms."' The EPA and other regulatory agencies typically develop regulations for 114. E. STOKEY & R. ZECKNAUSER, A PRIMER Foa PoLIcY ANALYSIS I39-4:: (1978). 115. See Harrison & Nichols, Benefit-Based Flexibility in Environmental Regulation (Apr. 1983) (1)iscussion Paper Series. Kennedy School of Govemment, Harvard Un-versity) (discussing the general advantages of these differential standards and an cvaluation of potential obstacles). The potential policy considerations that might arise in imposing Jif- ferent standards on different plants, including equal protection issues and problems arising from regulations that encourage businesses to locate new plants in less populated but generally more pristine areas, lie beyond the scope of this article. . Za0~Vr1!z0Z