Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix Setting National Standards for Inorganic Arsenic Emissions from Primary Copper Smellers: A Case Study RALPH A. LUKEN Section 112 of the Clean Air Act requires the Environmental Protection Agency (EPA) to establish emission standards for hazardous air pollutants that protect public health with an "ample margin of safety." In interpreting the language for the purposes of regulatory development, EPA does not consider the word "safety" to imply a total absence of risk. Many activities involve some risk but are not considered "unsafe." In EPAs view, standards under Section 112 should protect against significant public health risks. In setting a Section 112 standard, EPA identifies sources of pollution that may pose significant lisle, determines the current and planned levels of control at those sources, and assesses the health risks associated with those levels. If a source is judged to pose a significant risk EPA selects a level of control that, in its judgment, reduces the health risks to the greatest extent that can reasonably be expected, after considering the uncertainties in the risk analysis, the residual lisle that remain after the application of the pollution control technology, the costs of further control, and the societal and other environmental impacts of the regulation. This process is referred to as risk assessment and risk management. Policy analysts and decision makers long have struggled with how best to apply economic methods and assumptions when analyzing and controlling risk. On the one hand, the limitations of economic assumptions and Ralph "Skip" Luken is chief of the Economic Studies Branch, Economic and Regulatory Analysis Division, in EPA:s Office of Policy, Planning, and Evaluation. The views expressed in this case study are those of the author and not necessarily those of EPA. 209
210 APPENDIX analysis have been widely discussed and have in fact somewhat restricted the application of this approach to such areas as risk management. On the other hand, the necessity of recognizing the risk choices associated with and the trade-offs of regulatory options argues for some role to be played by economics in analysis and decision making. Most debate over the use of economic principles and methods in risk assessment and risk management eventually focuses on the underlying assumptions (usually implicit) that relate to rational behavior, ethics, public choice, and time preference. At this level, the issues concern notions of equity, social values, philosophical presuppositions, and the social contract between government and its citizenry. This case study is intended as an opportunity for the reader to apply the concepts of risk assessment and risk management to arrive at a decision about what constitutes an adequate level of health protection from one source of hazardous air pollution. In this case, the hazardous air pollutant is inorganic arsenic, and the source is uncontrolled fugitive emissions from primary copper smelters that process copper ore containing arsenic as an impurity. The primary copper smelting industry in the United States uses pyro- metallurgical processes to extract copper from sulfide copper ores that contain arsenic as an impurity. At the 15 primary copper smelters operating in the United States in 1983, the average arsenic content of copper ore ranged from 0.0004 to 4.0 weight percent. The average arsenic content of the ore was well below 0.5 weight percent at the majority of smelters; only the Tacoma smelter processed ore with more than 1 percent arsenic. The 14 low-arsenic copper smelters are the subject of this case study. Arsenic emissions from these smelters total 730 megagrams per year. Pro- cess operations, which emit about 530 megagrams per year, are already controlled to the extent technically possible. Uncontrolled fugitive sources account for the remaining 200 megagrams of arsenic emissions per year. The regulatory question is as follows: How many primary copper smelters should be required to control fugitive inorganic arsenic emissions? On June 5, 1980, EPA published a Federal Register notice (U.S. Envi- ronmental Protection Agency, 1980) listing inorganic arsenic as a hazardous air pollutant under Section 112 of the Clean Air Act. On July 11, 1983, EPA proposed standards for inorganic arsenic emissions from the 14 low-arsenic primary copper smelters as well as from the single high-arsenic copper smelter (U.S. Environmental Protection Agency, 1983c). On August 4, 1986, EPA issued a final standard for inorganic arsenic emissions from the 14 low-arsenic primary copper smelters and withheld further action on the high-arsenic copper smelter because it had ceased operation (IJ.S. Environmental Protection Agency, 1986b).
RALPH A. LIKEN 211 The reader is encouraged to make his or her own regulatory decision, given the same information that was available to EPA in mid-1986, before reviewing the actual EPA rule making for inorganic arsenic. The next section presents background risk assessment information as a scientific estimate of health risk. The following section presents back- ground risk management information composed primarily of engineering and economic descriptions of the consequences of controlling emissions at the 14 smelters. The final section highlights some of the important factors to consider in determining the level of pollution control that would protect the public against significant health risks with an ample margin of safety. RISK ASSESSMENT: QUANTIFYING CANCER RISKS The quantitative estimates of public cancer risk presented in this section are based on (a) EPA:s linear nonthreshold model, which numerically relates the degree of exposure to airborne inorganic arsenic to the risk of getting lung cancer; and (b) EPA:s Human Exposure Model, which expresses numerically the degree of public exposure to ambient air concentrations of inorganic arsenic from the 14 copper smelters. This section describes these models and the assumptions used to assess cancer risks and presents EPA:s quantitative estimates of individual and population risks. It also discusses uncertainties in the risk characterization that should be considered before preparing estimates of individual and population risks. Estimated Dose Response The numerical constant that defines the exposure (dose)/risk (response) relationship in the linear nonthreshold model is called the unit risk factor. For an air pollutant, the unit risk factor is the excess cancer risk associated with an individual's lifetime of exposure (70 years) to an average concen- tration of 1 microgram per cubic meter (1 ~g/m3) of the pollutant in the air. For inorganic arsenic, the unit risk factor is based on EPAs analysis of five data sets of the latest smelter worker epidemiological data collected by four researchers at two smelters (%ble 1~. ~ establish a single point estimate, EPA obtained the geometric mean for the data sets within distinct exposed populations and took the final estimate to be the geometric mean of those values. Based on this analysis, EPA used a 0.00429 ~g/m3 unit risk factor in assessing the health impact of inorganic arsenic.
212 TABLE 1 Combined Unit Risk Estimates for Absolute-Risk Linear Models APPENDIX Geometric Final Estimated Exposure Mean Unit Unit Risk for Source Study Unit Risk Risk Both Smelters Anaconda Brown & Chu 1.25 x 10 3 2.56 x 10 3 smelter Lee Feldstein 2.80 x 10 3 Higgins et al. 4.90 x 10 3 -3 4.29 x 10 3 ASARCO Enterline & 6.81x10 3 7.19x 10 smelter Marsh 7.60x10 SOURCE: U.S. Environmental Protection Agency (1986a). Estimated Public Exposure EPA applied its Human Exposure Model (HEM) to the 14 smelters to produce quantitative expressions of public exposure to ambient air con- centrations. In addition, EPA carried out more site-specific dispersion modeling at E1 Paso, Texas, and Douglas, Arizona, to evaluate the ef- fects of terrain and buoyancy at fugitive emissions on airborne arsenic concentrations. Table 2 lists, on a plant-by-plant basis, the total number of people in- cluded in the exposure analysis. "Any risk" is the number of people exposed to emissions from the specified source, as calculated by HEM. "Maximum risk" is the number of people exposed to the maximum individual risk from the specified source, as calculated by HEM. Estimated Individual and Population Risks By combining numerical expressions of public exposure with the unit risk factor, two types of numerical expressions of public cancer risks are produced. The first, called individual risk, relates to the person or persons who are thought to live in the area of highest concentration as estimated by the computer model. Individual risk is expressed as "maximum individual risk." As used here, the word "maximum" does not mean the greatest possible risk of cancer to the public but is based only on the maximum annual average estimated exposure. The second expression of risk, called population risk, is a summation of all the risks to people estimated to be living within the vicinity (usually within 50 kilometers) of a source. The population risk is expressed as incidences of cancer among all of the exposed population after 70 years of exposure; for convenience, it is often divided by 70 and expressed as cancer incidences per year.
RALPH A. LIKEN TABLE 2 Number of People Exposed to Emissions 213 Plant Total Number of People Exposed Any Riska Maximum Risks Distance (km) from Source ASARCO-El Paso ASARCO-Hayden Kennecott-Hayden Kennecott-Hurley Kennecott-McGill Kennecott-Garfield Phelps Dodge-Morenci Phelps Dodge-Douglas Phelps Doge-Ajo Phelps Dodge-Hidalgo Copper Range-White Pine Magma-San Manuel Inspiration-Miami Tennessee Copper Copperhill 493,000 46,800 46,800 26,300 7,350 810,000 25,500 31,000 6,600 2,560 16,900 211,000 35,700 164,000 1.0 0.3 0.3 0.3 0.3 5.0 2.0 0.2 909 1 2.4 0.2 0.4 05 . . . bA 50-kilometer radius was used for the analysis. People exposed within the distance specified in the next column. SOURCE: U.S. Environmental Protection Agency (1986a). Table 3 summarizes the maximum individual risk and the annual in- cidence for baseline and pollution control scenarios. The baseline level of risk is that resulting from the level of emissions after applying in-place controls or those controls that are required to comply with current state or federal regulations but before applying best available technology (BAT) controls. BAT controls would result in additional reductions of emissions by placing secondary hoods on converter operations. The converter control scenario level of risk is that of the remaining risks after installing BAT controls at all 14 plants. Uncertainties in Risk Characterization Exposure for an Entire Lifetime There are several basic assumptions implicit in the exposure methodol- ogy: (1) that all exposure occurs at people's residences, (2) that people stay at the same location for 70 years, (3) that the ambient air concentrations and the emissions that cause these concentrations persist for 70 years and
- cn o u .c uq ou ~ - u) u, ~; ~q ~n ct - ~: .= - - ~: .~ ~s .s .q o 4 - ~ o o o u ~ - m ~: o .~ O X ~ R' y O 0 ~ UO X _1 ~ O m x - V' 214 ~o ~C~ ~ o~ ~ ~ om ~ - - ~ oo ~ ~ - ~ ~= ~ ~= o oQ ooo o oo o oo ooo o oooooo o oo oooo oo oo ~ ~ ~ o o o ~ ~ ~ o. ~ oo. ooo oooo o o~ oo ~ ~ oo mo o° Oo 0° O o° o o o o o o~ oo ', oo Q ~ ~ Q ~°=o~oS~o~80.80g oooooooooooooo o~ ~o mo o C) C~o ~ o o~ ~ o o o ~ o -4 ~ ~ cr, ~ 0 ~ ~ ~ da O `0 ~ ~ omoooo ~o o. 0 0 0 ~ 0 oo0 oo .· r S ,0 ~S ~ ~ ~ ~ r _r ~ ~ ~U} S ~' ~ S S S ~
~ on ~ JO to to to to to o. o. JO to Con ~ to ~ to O O Q 0. O. ~ O O O \0 d- A Q ggg O O O O ~ O O O O ~ to . . . O ~ ~ .~ l? ~ ~ Of E ' E .m o ~ 4 - ~ o ·= ~ ~ _ ~ o ~ o 4 - Cat ~t 14 4 - ~ o ° og C~ . oi ~ o .a b~ ~ ~n C) ~o 4 - ~·4 - (Q ~.U, O O ,,C, o 4" ~ . ~ o. ·~= o o ~ .~ ~ ~q ~ ~ ~ := st: ~ ·9 ~ i~ ~ g.~ ~4 - E~ ~4- ~ ,y~ ~ `: o o Ct :~ ~ ~ ~ ~ o -O =, a c a 2 i~lf .'Ae ~ ~ ~ ~ ~ o ~ ·~ ~ 0-- ~ ~ ~ O ~ ~ 3 3 ~ ti ~ ~ $: - · oo os - ¢ f:: o . - CJ P" i~ Q .~ o U) 215
216 APPENDIX (4) that the concentrations are the same Inside and outside a residence. In sum, the exposure methodology assumes that individuals are exposed to inorganic arsenic emissions for their entire lives. Several reviewers of EPA:s methodology have questioned these simpli- fying assumptions, particularly the assumption of 70year resident immobil- ity. If EPA used what to the reviewers is a more reasonable assumption-for example, a 10-year residency in the area-then the maximum lifetime in- dividual risk would decrease by approximately one order of magnitude. This 10-year assumption, however, would not change the annual cancer incidence because this calculation is independent of population mobility. Early Lifetime Exposure Although the estimates derived from the various epidemiological stud- ies are quite consistent, there are a number of uncertainties associated with them. The estimates were made from occupational studies that involved exposures only after employment age was reached. In estimating risks from environmental exposures throughout life, EPA (1984) assumed in the linear nonthreshold model that the increase in the age-specific mortality rates of lung cancer was a function only of cumulative exposures, irrespective of how the exposure had been accumulated. Although this assumption adequately describes all of the data, it may be in error when applied to exposures that begin very early in life. Similarly, it is possible that linear models are inaccurate at low exposures, even though they reasonably describe the epidemiological data. Given greater access to the data from these studies, other dose mea- sures, as well as models other than the linear nonthreshold model, could be studied. Such analyses would indicate whether other approaches are more appropriate than the ones applied here. Use of Census Data The official EPA risk assessment (Table 3) underestimated the max- imum individual risk and cancer incidence at the E1 Paso and Douglas smelters because it did not include the local Mexican population living in border towns and illegally in the United States. In the case of the E1 Paso smelter, the population in neighboring Juarez is approximately the same as in E1 Paso. The maximum individual risk there is similar, 10-3, and the annual cancer incidence ranges from 0.40 to 0.70, which is double the incidence among the U.S. population. These estimates do not include the Mexicans living illegally in the United States because they are not counted by the Census Bureau. In the case of the Douglas smelter, the population of neighboring Agua Prieta is about double the population of Douglas. The
RALPH A. LIKEN 217 maximum individual risk is similar, 10-3, and the annual cancer incidence is 0.04, which is double the incidence among the U.S. population. If this information were incorporated in risk management decisions, it would lower the cost per life saved and increase the economic efficiency of the regulation, particularly at El Paso. Assumption of No Latency Period EPA's risk assessment assumes that there is no latency period between exposure and incidence. Although there is no definitive information on the exact length of the latency period for airborne arsenic, it is greater than zero. Enterline and Marsh (1982) suggest that it may be in the range of 1~19 years because their standardized mortality ratios appear to become significant about 1~19 years after exposed workers have left the plant. If this information were incorporated in risk management decisions, it would decrease the economic efficiency of the regulation, particularly at El Paso. Exclusion of Other Health Effects The unit risk factor used in this case study applies only to lung cancer. Other health effects are possible, however, including skin cancer, hyper- keratosis, peripheral neuropathy, growth retardation and brain dysfunction among children, and increases in adverse birth outcomes. No numerical expressions of risk relevant to these health effects were included in the EPA regulatory analysis. Evaluation of Risk Assessment in preparing a risk assessment as a result of reviewing EPA:s risk estimates and the uncertainties in the estimates, the analyst or decision maker should focus primarily on the baseline maximum individual risk and annual incidence and assume that the proposed converter controls will achieve the EPA estimated reduction in fugitive emissions. If the baseline level of risk is changed, however, the risk remaining after the installation of converter controls should also be changed proportionally. RISK MANAGEMENT: EXAMINING THE CONSEQUENCES After determining the cancer risks from arsenic emissions, the regu- latory decision maker should examine the consequences of requiring BAT
218 APPENDIX controls at the 14 copper smelters. This examination could include the es- timated emission and risk reduction, the remaining risks after BAT control, the costs and economic impacts, cost-effectiveness and economic efficiency, and the remaining public exposure or equity considerations. A standard should be determined for emissions from smelter converters that would require all smelters above a specific arsenic feed rate to install pollution controls (column 3 in Able 4~. If the standard were 100 kilograms per hour (kg/h) or greater for converter operations, no controls would be required; if it were 0.5 kg/in, controls would be required on converter operations at all 14 smelters. Emissions and Risk Reductions The BAT controls would require the installation of secondary hoods on converters. The potential emission reductions from and the estimated annualized costs of the converter secondary controls at each of the existing smelters appear in Able 4. The estimated cost-effectiveness ranges from about $100,000 to $9.7 million per megagram ($/Mg) at the 14 smelters. Applying BAT controls for converter secondary emissions would reduce the range of estimated maximum individual risks from between 1.0 x 10-3 at the Phelps Dodge-Douglas and 2.0 x 1O-s at the Phelps Dodge-Hidalgo smelters (see Able 3~. It would also reduce the estimated annual incidence of lung cancer from between 0.09 (regulating only the plant with the highest incidence, that is, (ASARCO-E1 Paso) to 0.14 (regulating all 14 plants). Remaining Exposure and Risks The remaining exposure and consequent risk are a function of the number of plants that are required to control converter emissions. Applying controls for these emissions at all 14 smelters would change the range of estimated maximum individual risk from between 1.3 x 10-3 and 5.0 x 10-6 to a range of 1.2 x 10-3 and 3.0 x 10-6 (see Table 3~. Applying controls would also reduce the estimated annual incidence of lung cancer from a range of 0.3~0.001 to a range of 0.2~0.0001 Applying controls at none of the plants would leave the remaining risks at the same level as the baseline risks. Costs and Economic Impacts The annualized control costs per smelter range from a low of $379,000 at ASARCO-El Paso to $3,432,000 at Phelps Dodge-Morenci. If BAT controls were applied to all 14 smelters, the total annualized control costs would be approximately $29 million.
RALPH A. LIKEN 219 EPA studied the economic impact of imposing controls on the eight smelters with an arsenic feed rate greater than 1 kg/in. For two of the eight smelters, Kennecott-Hayden and Kennecott-McGill, the control costs were likely to result in permanent closure of the smelters. The analysis also indicated that the economic impact would be minimal only at ASARCO-E1 Paso. An additional factor considered for this smelter was that secondary hoods were scheduled to be installed on all converters to comply with requirements in the Texas state implementation plan for attaining the national ambient air quality standard for lead. Economic Cost-Effectiveness 1b determine whether the standard is cost-effective in terms of number of cancer cases avoided, the analyst must establish a reasonable value for a statistical life saved. The agency's Guidelines for Regulatory Impact Amasses (U.S. Environmental Protection Agency, 1983b) suggest that this value should fall in the range of $4~0,000 to $7 million, with a point estimate of $2 million. This value is supported in part by a recent survey of 130 decisions made by the U.S. government to regulate carcinogens (Travis et al., 1987~. The survey found that the average implicit value of a statistical life saved was approximately $2 million. If $2 million is a reasonable value to consider in a decision to control arsenic, then the figures in Able 5 suggest that not regulating any plant is cost-effective. They show that the cost per case avoided exceeds $2 million at all plants. If the lowest value, $4.2 million at ASARCO-EL Paso, were considered to be a reasonable value to initiate action, a argument could still be made against controlling fugitive emissions from the converter at that plant. An alternative risk assessment scenario at ASARCO-E1 Paso, described in Table 3, shows a smaller reduction in cancer incidence (0.02) and consequently a higher value per case avoided ($18.9 million). These figures exceed the highest value in the Guidelines and the average value implied by past U.S. government regulatory decisions. The scenario assumes that only 3.75 percent, rather than 15 percent, of arsenic emissions escape the primary vent hood. The higher emission factor of 15 percent reflects actual emissions before E1 Paso upgraded its gas management system. The lower emission factor of 3.75 percent is an EPA estimate that was derived from the performance of other hoods for which EPA had data rather than from actual measured values at El Paso after installing the new equipment. The new emission factor is probably lower than 15 percent, which means that the cost per incidence avoided is higher than the lower bound of $4.2 million.
ha O ~ ·~q Cal 4- ca ~ y ~ ·_ ·. - ~ ~ 3 §~ o o V} cn o 4 - o . a Cal . .= C) . o - Ct Cal in .; 4 - .~ o At; A CJ 4 - v to - Ct do o <; En O 220 ·_ - ~ ~ ~ [Q o ~ ~ ~ ° U ~ A ~ o ~ ° ~ o C _ 'in ~ 'C TO' 4 - ·= ~ ~ ~ .O ~ ~ ~ C ~ ~ 4 - V) o o US o o ~ ~ o ~ ~ ~ o ~ o m-~- ~ ~ ~ o~ ~i 0 ~ oo ~ so o~ o~ ~ oo ~ ~ ox 0 ~q C~ o~ ox . . . . . . ~o ~ ~o oo ~ 0 0 ~o os ~ - 4 oo oo ~ o~ ~ o~ so =~ - ooo. 0 00000 U. W l~ i? ~ ~ ~ ~ C C ¢ ¢~m
Q to to to to to lo to to o o o o o o o o o o oo oo o o oo Y crx 0 N N ~ ~ ~ ~ ~ moo moo m~ ~opo ~ ~ ~ ~ r~ ~ ~ oN oN ~ o ~ ~ oN m^~^~^~^~^~^~ oo oo 0 ~ ~ r~ et ~ ~ - - oo oo oo So o ) ~ So ~ ~ ~ ~ - - oooo oo ~ ~ ~ ~ o ox · . . · · · · ~ - - oooo oo x - . ~. - . =. =. =. ~ - - oo oo oo So ~ o u~ o oo oo~880088 ooooooooo - .£ ._ ~ ~·~ ~ C1S °s ~ _ 4- _ ~ O ~ ~ . e a ~ a O ,: a ~ ~ == ~ C ~ ~ ° a ,0 ~ C c ·- 0 o~s ~'`ib~ O ~ ~ ~ ~ m,~ ~ ~ ~53 ° 0 g ~ ~ c t} `, E <~, c~ ~ ~ ~ ~ ug c ° C' ° S C) o ~ C ° c ~ ~ ~ g =0 a :r ~z ~ ' a j~ ~ ~ O ~ .= ~ ~ ~ . ~C ~ c ~ ~ 4~ =~.a ~ p, C, . ~ ~ .C ~ C c ~ ~ c g ~ O .° ~ ~ ~ ~ ~ ~ ~ ~ ~ ~a ce ~ ~ ~ ° ~ ~ ~ ~ ~ 8 - ~ Ct ~ ~ ~ ~ V' 221
Go ~ ~ To hi ~ ~ ~ ~ Do 1 1 1 1 lo/ Go Go ~ ox ~ ~ ~ oS oS an Go ~ oN ~ oN of ~ on ~4 ~ ~ ~ ~ ~ ~ ID ~ ~ \0 ~ 0\ ~ ~ ~ 00 0 cimmoom - = =`ou~iooo~oo~c~`o 1 1 1 1 Jo ~ ~Hi) An ~ ~ ~ ~ ~1 ~ ~ ~ - o Do oN By o I: . - .e ~ - ~n - 2 I: o _' d-) - ~S 4 - d) d) 4 - d) A: it day Pa .t I} 4 - ~A ¢ ~ m ~ _~ ~0 _~ ~0 ' '` d~ ~n ~ ~ O ¢ 0d ~ ! ~ :m ~: ~: ~o ~o t-e ~U; .~ ~ ~ ~ ~ Ct o ~ ~ ~ :~aL ~;~= ~I~i~! a a 8 a aaag Y BI - ~ ~ ~Y 8 ~ e = ~ ~ ~ L! ~ O. O. 00 ~ ~ ~ ~ ~ ~ ~ ~ ~ C~ ~ C~ O ~ O 1 1 1 0 ~ ~ ~ r~ O U ~O1 ~) ~ \0 C~ 0\ 00 0 ~i ~ ~ oo ~ ~ ~so iD ~ ~ 00 0\ 00 ~ C~ \0 1 1 1 ~ ~m mm~~ - 0\ 00 ~ ~ o' o ~ ~ ~ ~ <~ ~ O. r~ ~ o. ~ a00 ~ a0~a 00 ~ ~ ~00 ~0 ~D ~0\ ~0~0~~ O ~ ~ ~ ~ O ~ ~ 0 8 08 o o o o 000 0 00 . ...... 0000000 0000000000 222 d. ~: D Ct E~ .e d-,, . ~ U o cn
RALPH A. LIKEN 223 One could dispute the claim about the unreasonableness of the cost per case avoided on the grounds that the estimate fails to include the benefits to the Mexican population. Including the Mexican population near the E1 Paso smelter and using a high emission rate reduce the annual incidence by 0.265, as opposed to the 0.09 reduction achieved with controls on fugitive emissions. Consequently' the cost per case avoided is $1.4 million, rather than $4.2 million. Including the Mexican population and using a low emission rate reduce the annual incidence by 0.12, compared with the 0.2 reduction achieved with controls on fugitive emissions. In this instance, the cost per case avoided is $3.2 million, rather than $18.9 million. These revised estimates suggest that the cost per case avoided at E1 Paso is consistent with EPA Guidelines and other U.S. government decisions. However, including the Mexican population in the estimates for the Douglas smelter only brings the cost per case avoided nearer to the high end of the range in the EPA Guidelines. Economic Efficiency One can determine whether a standard is economically efficient by comparing the net present value of the standard's benefits and costs (Table S). As the table shows, with a 10 percent discount rate and no latency period, no control option has a positive net present value (assuming $2 million per statistical life saved). Changing the discount to 4 percent does not significantly change the results. The net present value of all controls is negative. Thus, regulation at all plants would be rejected on the grounds of economic efficiency. Using a lS-year latency period changes the economic efficiency eval- uation only at the El Paso smelter. The magnitude of the change in the negative net present value is greater for this plant because of the greater proportionate reduction in the values for cancer cases avoided. The assump- tion of a lS-year latency does not markedly alter the negative economic efficiency evaluation at the other 13 plants because fewer cases are avoided and because the control costs overwhelm the benefits in the net present value calculations. Assuming a value of $7 million per statistical life saved only partly alters the negative economic efficiency valuation at the E1 Paso smelter. Using the highest reduction in annual cancer incidence-0.09 cases makes the net present value positive for both discount rates with no latency period and for the 4 percent discount rate with a 15-year latency period (not shown in Table S). In all other circumstances at E1 Paso (0.09 incidence/10 percent/lS years and 0.02 incidence/any combination of rates and years) and at all other smelters, the net present value would remain negative, even for a relatively high value per statistical life saved.
224 APPENDIX TABLE! 6 Health and Welfare Benefits from Reducing Lead and Particulate Matter Emissions Number of Reduced Benefits Pollutant Effects/Ton of a ($/ton) of Effect Emission Reduction Emission Reduction Lead 3 Chelation 4.5 fewer eases 16.5 x 1O3 Hypertension 47.0 fewer cases 10.5 x 1O3 Myocardial infarction 0.14 fewer cases 8.8 x 103 Strokes 0.03 fewer cases 1.4 x 10 Mortality 0.13 fewer deaths 260.0 x 1O3 Total for lead 297.2 x 10 Particulate Matter Mortality Morbidity Lost workdays Reduced-activ~ty days Medical expenditures Soiling and material damages Total for particulate matter (national average) 4.3 x 10 5 fewer deaths 0.1 x 10 8.8 fewer sick work loss days 27.9 fewer sick work loss hours 0.7 x 103 1.1 x 1O3 0.3 x 10 0.2 x 103 2.4 x 10 Effects based on county-weighted average for estimates of reduced number of effects per ton of reduced pollution emissions. Range for the particulate matter total benefit value varies between $0 and $300,000/ton for individual counties in the United States. The variation by county for the benefits of reducing lead emissions was not calculated. S()URC:~: U.S. Environmental Protection Agency (1983a, 1985). In making a regulatory decision, one could also examine whether the standard would be economically efficient if the total benefits of emission reductions were compared with the costs. At E1 Paso, installation of BAT controls for fugitive arsenic from converters would also achieve a 6.6- Mg reduction in lead (Pb) and a 30-Mg reduction in particulate matter (PM). Controls at Douglas would reduce PM by 780 Mg. An immediate issue, however, is how to compare and aggregate these health and welfare benefits. In the case of PM, there are reductions in mortality, morbidity, and welfare damage to be considered (Bible 6~. EPAs Guidelines encourage the analyst to monetize these benefits and then add them together to obtain a dollars-per-ton figure. The PM benefit/ton typically ranges from $300/ton to $10,000/ton, depending on the PM sources and exposure patterns for receptors near the sources. Incorporating the PM benefits into an economic analysis would not alter the efficiency evaluation at the Douglas plant if one accepted
RALPH A. LIKEN TABLE 7 Net Present Discount Values for Benefit-Cost Streams of Particulate Matter (PM), Lead, and Arsenic Emission Reductions 225 Present Value Benefitsa Reduced ($ million) Emissions Discount Rates Net Present Value ($ million) Discount Rates Plant/Pollutant (tons) 4% logo To 10% ASARCO-El Paso Lead 6 6 26.7 16.7 PM , 29.5 1.2 0.8 Arsenic Scenano 1 3.7 2.4 1.5 Scenario2 1.0 0.5 0.3 Totalc i Scenanol 30.3 19.0 26.3 15.8 Scenano 2 28.4 17.8 24.4 14.6 Phelps Dodge-Douglas PM d 776 10.5 6.6 Arsenic Scenano 1 3.7 0.4 0.3 Scenano 2 3.7 0.3 0.2 Totale Scenario 1 10.9 6.9 -25.7 -18.2 Scenario 2 10.8 6.8 -25.7 -18.2 Scenano A ~ 0.0 Scenano B > 0.0 bAssumes no latency period for assessing health and welfare effects for PM, lead, and arsenic. ASARCO-EI Paso Scenano 1 assumes that 15 percent of the arsenic emissions escape the primary vent hood; Scenano 2 assumes that 3.75 percent of the arsenic emissions escape. dThe expected benefit/ton value for reductions in PM is $3,000/ton for the El Paso area. Phelps Dodge Scenano 1 nslcs are based on a standard exposure analysis. Scenario 2 risks are based en site-specific analysis (ISCL.T/Valley model). The expected benefit/ton value for reductions in PM is $1,000/ton for the Douglas area. To acheive a net present value greater than zero with a discount rate of 4 percent (Scenano A), the PM benefit/ton value must be at least $3,000/ton. The benefit/ton value given a 10 percent discount rate (Scenano B) must be at least $3,800/ton. the population weighted value for the Douglas area of $1,000/ton (Ibble 7~. However, using a value of $3,800/ton makes the benefits equal to the costs, in contrast to the negative net present value of $2~$36 million for which no PM benefits are considered, or a negative $1~$26 million for the scenario with $1,000/ton PM benefits. Incorporating the PM benefits into an economic analysis of controls at the El Paso plant does not alter the negative efficiency calculation, given the population weighted value for the El Paso area of $3,000/ton, nor any value within the range of $30$10,000/ton. Incorporating the lead benefits,
226 APPENDIX however, yields a positive net present value of $15-$26 million, rather than a negative net present value of $0.~$2.3 million. Equity The regulatory decision maker may also consider whether a standard meets an equity objective rather than an economic efficiency objective. For example, it might be decided that controls should apply in all circumstances in which maximum individual risk equals or exceeds 10-4. In 1986, this risk level existed at three smelters: E1 Paso, Hayden, and Douglas. Alterna- tively, controls might be applied to the degree necessary to reduce all risks to at least 10-5. This goal is not technically possible with the proposed BAT controls, however, because maximum individual risk remains greater than 10-5 even after the installation of BAT controls at all plants. FACTORS TO CONSIDER IN SETTING A STANDARD The risk assessment and risk management information presented in the two previous sections is all that is available to make a regulatory decision. As the regulatory decision maker, you the reader must decide how many of the 14 plants to regulate by expressing a standard that imposes converter control hoods on plants above a specific arsenic feed rate. The following points should receive careful consideration. How Should Health Risk Be Characterized? The quantitative information about health risk applies only to lung cancer. -How would you account for other possible health effects (described in the second section)? The quantitative estimate of lung cancer risk appears to be both an overstatement and understatement of risk, as described earlier. It is an overstatement because it assumes continuous exposure for 70 years and does not take into account the potential latency period between exposure and incidence. It is an understatement because it excludes the exposed Mexican population. (As a practical matter, it should be noted that the Clean Air Act does not authorize extraterritorial jurisdiction and thus cannot be applied to Mexico.) It could be further over- or underestimated depending on how one deals with the uncertainties in the epidemiological studies. What Constitutes a Significant Risk? In determining what constitutes a significant risk, you should consider both maximum individual risks and annual cancer incidences resulting from
RALPH A. LUREN 227 exposure to inorganic arsenic. The current maximum individual risks range from 1.3 x 10-3 to 5.0 x 10-6; the annual incidences range from 0.38 to 0.0001. At some smelters, both the individual and population risks are low and would probably be deemed insignificant. In your decision making, you should determine the level of baseline risk you think is insignificant and thus would not warrant regulation. Your decision is complicated by the lack of a perfect correspondence among plants on individual and population risks. For example, the individ- ual risk exceeds 10-4 at three plants. Yet, the baseline cancer incidence varies by more than an order of magnitude among the three plants: the highest incidence, at E1 Paso, is 0.38; the lowest incidence, at Douglas, is 0.022. Consequently, although the significance of the risk generated by the plants is the same for individual risk, it is very different for annual · - inch once. What Constitutes an Appropriate Balance Between Costs and Risks? One approach that you might consider in addressing this issue is an economic cost-effectiveness cutoff for incidence (cases of disease) avoided. As noted in the last section, the cost per incidence avoided ranges from $4.2 million to $6.39 billion. Is it reasonable to exclude some plants from regulation on the basis of this information? Also, as described in the last section, the cost per incidence avoided at two plants (El Paso and Douglas) changes from the official EPA analysis if it includes the exposed Mexican population. The cost per incidence avoided ranges between $1.4 million and $3.2 million rather than between $4.2 million and $18.9 million at El Paso if the Mexican population is included; the cost per incidence avoided still exceeds $7 million at Douglas if the Mexican population is included in the analysis. Does inclusion of the information change what you think about the necessity for regulation? Another approach that you might consider in addressing the issue of balance between costs and risks is the economic efficiency of the stan- dards. Under most conventional approaches, BAT controls at all plants are economically inefficient because the costs exceed the benefits. Even the use of the preferred EPA approach, which assumes no latency period between exposure and incidence and a low discount rate, does not alter the conclusion that the net present value of controlling any plant is negative. Nevertheless, including the "cocontrol" benefits of reduced lead and particulate matter together with the arsenic benefits alters these calculations at one plant. Incorporating all pollution reduction benefits at the E1 Paso and Douglas smelters suggests that the imposition of BAT controls is economically efficient at the El Paso plant but is still not efficient at the Douglas plant.
228 APPENDIX How Should Single-Decision Criteria Be Explicitly Integrated? You have received information about several decision criteria to assist you in making a regulatory decision. These criteria include estimates of maximum individual risk, annual incidence of cancer, cost per life saved, economic efficiency of controls, and economic impacts on the copper smelter industry, as well as the uncertainties in these estimates. In addition, you know that for the El Paso smelter the state implementation plan would result in a significant reduction of inorganic arsenic emissions even if EPA did not issue a standard for inorganic arsenic. Also, you know that there would be, as described in the previous section, cocontrol of other pollutants with a standard for inorganic arsenic. Can you be explicit about the relative importance of these single-decision criteria and how you combined them, together with the qualitative factors, to arrive at your decision to impose controls at specific plants? Is Any Balance Between Costs and Risks Consistent with EPA's Legislative Mandate? An overriding risk management issue is whether consideration of eco- nomic information is consistent with the requirements of the Clean Air Act. As required by Section 112 (b)(l)B of the act, standards must be set "at the level which in [the administrator's] judgment provides an ample margin of safety to protect the public health" from inorganic arsenic emissions. EPA interprets Section 112 to require a judgment about the degree of control that can be considered amply protective. For nonthreshold pollutants, two choices are available: (1) to set the standards at zero emissions to eliminate the attributable health risks, or (2) to permit some residual risk. In the absence of a specific directive on this choice in Section 112, and in recognition of the drastic economic consequences that could follow from a requirement to eliminate all risk from hazardous air pollutant emissions, EPA believes that it is not the intent of Section 112 to eliminate all risks. Standards that permit some level of residual risk can be considered to provide an ample margin of safety to protect public health. Therefore, EPA maintains that there must be a consideration of costs in regulating hazardous air pollutants. The Natural Resources Defense Council (NRDC) is currently con- testing EPAs interpretation of Section 112. It argues that EPA may not consider nonhealth issues, such as technology, economics, and affordability. Furthermore, NRDC holds that public health should be the sole consider- ation in developing Section 112 standards, with no consideration of such factors as cost and the availability of technology.
RALPH A. LUREN 229 You have an opportunity to address this broader issue. If the EPA position is correct, what information should EPA consider in deciding on a level of pollution control that may present some human health nsk? The EPA position in the arsenic rule making was that the administrator "selects a level of control which, in his judgment, reduces the health risks to the greatest extent that can reasonably be expected, after considering the un- certainties in the analysis, the residual risks remaining after the application of the selected control level, the costs of further control, and the societal and other environmental impacts of the regulation" (U.S. Environmental Protection Agency, 1986b:27958~. The EPA position excludes considera- tion of economic cost-effectiveness and economic efficiency. If the NRDC position is correct, EPA ought not to consider any economic factors. In this situation, EPA should impose BAT controls on all 14 smelters at an annual cost of $29 million to prevent 0.134 annual cancer incidence. The cost per incidence avoided would be $216 million, a figure that does not suggest a reasonable balance between costs and risk reduction. EPA'S ACTUAL REGULATORY DECISION FOR INORGANIC ARSENIC EMISSIONS FROM PRIMARY COPPER SMELTERS On June 5, 1980, EPA published a Federal Register notice (U.S. Envi- ronmental Protection Agency, 1980) listing inorganic arsenic as a hazardous air pollutant under Section 112 of the Clean Air Act. On July 11, 1983, EPA proposed standards (U.S. Environmental Pro- tection Agency, 1983c) for inorganic arsenic emissions from the nation's 14 low-arsenic primary copper smelters, as well as for high-arsenic cop- per smelters and glass manufacturing plants. The proposed standard for low-arsenic primary copper smelters regulated secondary inorganic arsenic emissions from converter operations and from matte and slag-tapping fur- naces. The proposed standards for converter operations applied to smelters with an annual average inorganic arsenic feed rate of 6.5 kg/in or greater. The proposed standards for matte and slag-tapping furnaces applied to smelters with an annual average combined inorganic arsenic process rate of 40 kg/in or greater. (The latter standards were less restrictive than the for- mer because secondary emissions from coverters are Epically 1 to 25 times greater than the combined emissions from both matte and slag-tapping operations.) The proposed standards affected 6 of the existing 14 low-arsenic pri- mary copper smelters. The estimated capital and annualized costs required to meet the standards were approximately $35.3 million and $9.5 million, respectively. ~ ~,
230 APPENDIX Following public comment on the proposed standards, EPA conducted additional analyses to ensure that the final rule was based on the most com- plete and accurate information available. These additional anlyses included revising the emission estimates, the exposure concentration estimates, and the risk assessment, as well as conducting additional cost and economic ~m- pact analyses. The scope of these analyses resulted in considerable changes in the risk assessment and risk management information that was incor- porated in the final rule making of August 4, 1986 (U.S. Environmental Protection Agency, 1986b). Using the revised risk and cost estimates, EPA concluded that for eight copper smelters (Inspiration-Miami; Phelps Dodge-Hidalgo, -Morenci, and -Ajo; Kennecott-Hurley; Tennessee Chemical-Coppe-rhill; Magma-San Manuel; and Copper Range-White Pine), the baseline risk was less than or equal to the risks of the standards it had previously withdrawn and that regulation was not warranted. For five of the six remaining smelters (all but ASARCO-E1 Paso) that were affected by the proposed standard, EPA concluded that the costs were disproportionate to the risk reductions that could be obtained. Furthermore, the economic analysis showed that for two of these five smelters (Kennecott-Hayden and Kennecott-McGill), the control costs were likely to result in the smelters remaining permanently closed. For the remaining facility, ASARCO-E1 Paso, the analysis indicated that risk could be reduced at a reasonable cost. An additional factor considered in the assessment was that secondary hoods were to be installed on all converters at ASARCO-E1 Paso to comply with requirements In the Texas state implementation plan for attainment of the national ambient air quality standard for lead. Because the costs of control in this instance are reasonable and the controls can be implemented now, EPA decided that these controls should be applied only at ASARCO-E1 Paso. As a result, the final standard affects only converter operations at 1 of the 14 smelters and does not apply to matte and slag-tapping operations. It applies only to smelters with annual arsenic feed rates to converters of greater than 75 kg/in. The estimated capital and annualized costs required to meet the final standard are approximately $1.8 million and $380,000, respectively. Because the only high-arsenic copper smelter affected by the ear- lier proposal ceased operation in 1985 (a plant owned and operated by ASARCO in ~coma, Washington), EPA withheld further action on the proposed standard for high-arsenic primary copper smelters. Ad;
RALPH A. LUKEN 231 REFERENCES Brown, C.C., and KC. Chu 1983 Implications of the multi-stage theory of carcinogenesis applied to occupational arsenic exposure. JCNI 70:455-463. Enterline, P.E. and G.M. Mach 1982 Cancer among workem exposed to arsenic and other substances in a copper smelter. American Journal of Epidemiology 116:895-911. Higgins, I.T.T., K.B. Welch, M.S. Oh, et al. 1985 Arsenic Exposure and Respiratory Cancer in a Cohort of 8,044 Anaconda Smelter Workers. Report to the Chemical Manufacturers Association and the Smelter Environmental Research Association. October. Lee -Fe ldstein , ~ 1986 Cumulative exposure to arsenic and its relationship to respiratory cancer among copper smelter employees. Journal of Occupational Medicine 38:196-302. Travis, C.C., S.A. Richter, E.A.C. Crouch, R. Wilson, and E. Klema 1987 Cancer risk management by federal agencies. Environmental Sconce and Tech- nology 21:415-420. U.S. Environmental Protection Agency 1980 National emission standards for hazardous air pollutants: Addition of inorganic arsenic to list of hazardous pollutants. Federal Register 45~110~:37886 37888. 1983a Benefits and Net Benefits Analysis for Altematrre National Ambient Air Q~alky Standards for Particulate Matter. final Regulatory Impact Analysis. Washington, I:).C.: Environmental Protection Agency, Office of Air Quality Planning and Standards. 1983b Guidelines for Performing Regulatory Impact Analysis. EPA-231403. Wash ington, D.C.: Environmental Protection Agency. 1983c National emission standards for hazardous air pollutants: Proposed standards for inorganic arsenic. Federal Re~ster 48~140~:33112-33180. 1984 Health Assessrrurnt Document for Inorganic Arsenic. EPA 600/8-83-021f. 1985 Costs and Benefits of Reducing Lead in Gasoline. Final Regulatory Impact Analysis. EPA-230-05-85 006. Washington, D.C.: Environmental Protection Agency, Office of Policy Analysis. 1986a Inorganic Arsenic Emissions from Primary Copper Smelters and Arsenic Plants- Background Information for Promulgated Standards. EPA-450/3-83-OlOb. Wash ington, D.C.: Environmental Protection Agency. 1986b National emission standards for hazardous air pollutants: Standards for inorganic arsenic. Federal Retester 51~149~:27957-28042.
