Click for next page ( 422


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 421
Table C-2 shows the calculated age-standardized lung-cancer rates by smoking category in the two areas. The difference in lung-cancer rates between the two areas, averaged over the smoking categories, is approximately 74. Assuming that this difference is due totally to general air pollution, which was mainly the result of inefficient burning of coal, we may express these rates approximately in terms of Equation 2, with t taking the value 55, and hence in terms of equivalent U.K. cigarettes. These calculations estimate the effect of the additional BaP air pollution in the urban area as the equivalent of 1.09 U.K. cigarettes. Thus, we estimate BaP-coal-burning at 52.5 ng/m3 = 1.09 U.K. cigarettes or BaP-coal-burning at 48.2 ng/m3 = 1 U.K. cigarette. Therefore, even though Stocks failed to address the issue of lifelong smoking habits satisfactorily, his data suggest a figure for BaP-coal- burning that is not much different from BaP-carbonization. If we use only the data on nonsmokers in Table C-2 to estimate the effect of BaP-coal burning, we f ind that BaP-coal-burning at 128 ng/m3 = 1 U.K. c ;garette . RATES IN NONSMOKERS The study of Stocks38 has been criticized, because he obtained data on many of the lung-cancer patients from relatives after the patients' deaths . This would especial ly tend to exaggerate the lung-cancer rates in the "nonsmokers." Dol 1 sugges ted that a more accurate lung-cancer figure for nonsmokers could be obtained by combining the data on lifelong nonsmokers from the prospective studies of Kahol9 and Hammondl3 in the United States. The combined data (Table C-5) show a lung-cancer mortality rate for nonsmokers roughly 45% of that found for nonsmokers in rural North Hales by Stocks. This is the relevant comparison, because the average BaP concentration in urban air in the United States in 1959 was roughly ~ ng/m --a figure very close to that of rural North Wales in 1954. Dolls showed that Equation 2 provided an excellent fit to the combined nonsmoker data from Kahn1 and Hammond (see Table C-5), and the best fit is obtained with the equivalent number of U.K. cigarettes ~ smoked from birth) set at 0.14. If these lung cancers were due totally to BaP-U. S. pollution, we could conclude BaP-U.S. pollution at 6 ng/m3 = 0.14 U.K. cigarette or BaP-U. S . pollution at 42 ng/m3 = 1 U.K. cigarette . Thi s may be cons idered a reasonable upper 1 imit of the potency o f BaP-U. S. pollution in nonsmokers. C-11

OCR for page 421
REGRES S ION STUD IES A ~nultiple-regression analysis undertaken for the National Research Council Subcommittee on Particulate Polycyclic Organic Matter26 attempted to "explain" the annual lung-cancer death rates (per 100,000) in 1950-1969, Y. in the 48 contenminous states of the United States by the independent variables X1 = cigarette sales per person over 15 yr old (1963), in dollars, and X2 = BaP in air, in ng/m3 (1967-1969~. typical result obtained was Y = 89.4 + 1.44 X1 + 7.05 X2 for white men aged 55-64. The observed average lung-cancer mortality rate for such men for the 48 states was 140.6. There are a number of major problems with this approach, which are discussed at length in the report--in particular, the crudity of both the cigarette-consumption data and the air-pollution figure for a whole state. The regression equations also predict lung-cancer mortality rates in the absence of smoking or air pollution that are much greater than the observed lung-cancer incidence in nonsmokers. For example, DollS gave a figure of 13.9 (compared with the above figure of 89.4) for the lung- cancer mortality rate in this age group on the basis of the combined results of Kahn 9 and Hammond. 1 Other regression studies have similar problems, leaving them useless for quantitative risk assessment. COMPARATIVE CARCINOGENICITY OF DIFFERENT AIR-POLLUTION MIXTURES The available epidemiologic evidence reviewed above suggests that the carcinogenic potencies of various air-pollution mixtures (coal carbonization, coal-burning, and general U.S. pollution) are similar when expressed in terms of the BaP content of the mixtures (Table C-6~. We have no useful epidemiologic data on cases in which the major con- tributor to air pollution has been mobile sources; to estimate the effects of such air pollution, lie must use the results of animal- carcinogenesis studies and short-term mutagenesis assays. This approach was used by Harris;15 Table C-7 shows the assay results he considered. Tables C-8 and C-9 show the relative potencies of the various contributors to air pollution computed from the data in Table C-7. Coke-oven extract is taken as the standard, and the results are expressed on a constant-weight-of-extract basis in Table C-8 and ~ constant-weight-of-BaP basis in Table C-9. For example, with the SENCAR C-12

OCR for page 421
mouse assay, roofing-tar extract is 0.255 (0~535/2.101) times as potent as coke-oven extract on an equal-weight basis and 0.137 [~0.255~478/889~] times as potent as coke-oven extract on a constant-weight-of-BaP basis. Tables C-6 and C-9 may be used together to predict the lung carcinogenicity of exposure to spark-ignition or diesel engine exhaust. Table C-9 sugges ts that exposure to a fixed amount of BaP from a Mustang mixture will be between 0.06 and 2.2 times as carcinogenic as such exposure to coke-oven pollution. The different vehicles tested vary widely in diesel-exha~st extrac t. The results shown in Table C-9 suggest that exposure to a fixed amount of BaP from diesel exhaust will be between O.1 and 89 times as carcinogenic as such exposure to coke-oven pollution. If we consider the L5178Y+ assay as the assay of choice, the predicted lifetime (age 50) lung-cancer risk associated with exposure to air polluted by a 1-ng/m BaP source for mobile-source emission is given in Table C-10. OTHER CANCER SITES Increased rates of cancer at sites other than lung were observed in the study of British gasworkersl and in the study of U.S. coke-oven workers. 2 In the study of British gasworkers, an excess risk was noted for cancer of the bladder (age-ad justed rate per 1,000 of- 0.37 vs. 0.12 expected), for cancer of the skin and scrotum (0.10 vs. 0.00), and for cancer at all other sites combined (2.73 vs. 2.27~. Because the excess risk of cancer of the skin and scrotum is extremely unlikely to be due to inhalation exposure, the maximal excess rate of all cancer except lung cancer that can be attributed to gasworks exposure is 0.71 (3.10 - 2.39~. The comparable figure for lung cancer is 2.12 (3.61 - 1.49~. Lung cancer therefore accounts for at least 75% (2.12/2.83) of the excess cancer associated with this British gasworks pollution. Similar calculations from the study of Redmond et al.32 for men employed 5 yr or more in the most polluted area (topside) of the U.S. coke ovens show that lung cancer accounted for at least 83t ( 17.6/21. 1) of the excess cancer associated with U. S. coke-oven air-pollution exposure. FOOD The estimated daily intake of BaP in food is 160-1,600 ng (see Table 6-25~. No epidemiologic studies are available to permit one to estimate the possible carcinogenic effect of such an intake of BaP, and recourse must be made to animal experiments. The experiment of Neal and Rigdon,~8 referred to in Chapter 4, found that BaP administered to mice in their diet produced forestomach tumors. With the extrapolation procedure used by the National Research Council Safe Drinking Water Committee, it can be calculated that a C-13

OCR for page 421
daily human intake o f 47 ng of BaP would lead to a lifetime risk of 1 in 100,000. With this estimate, we may calculate that the daily intake of 160-1,600 ng of BaP translates into an estimated lifetime cancer risk of 3.4-34 in 100,000. The estimated daily intake of PAHs in food is 10 times the intake of BaP (see Table 6-25), so one would estimate the total lifetime cancer risk associated with exposure to BaP and other PAHs in food at something less than 10 times these figures. C-14

OCR for page 421
TABLE C-1 Lung-Cancer Mortality Ratios for Smokers of High-, Medium-, and Low-"Tar" Cigarettes$ 1960-1972a #ITarl1 Content, mg/cigarette High (30) Medium (22.5) Low {15) aData from Hammond et a1.14 _ _ C-15 Mortality RaLio Males Females 1.0 1.0 0.95 0.80 0.81 0.60

OCR for page 421
TABLE C-2 Lung-Cancer Mortality Rates of Men in Rural (North Wales) and Urban (Liverpool) Areas, 1952-1954, by Past Smoking Habitsa Lung-Cancer Rateb Smoking Category Rural Urban _ Nonsmokers 22 (2) 50 (3) Cigarette-smokers: App. 10 cigarettes/d . 68 (23) 168 (71) App. 20 cigarettes/d 147 (36) 248 (140) App. 35 cigarettes/d 317 (33) 344 (138) aData from Stocks (p. 80~.38 bPer 100,000 per year, standardized for age. Figures in parentheses are numbers of lung-cancer deaths. C-16

OCR for page 421
TABLE C-3 Lung-Cancer Mortality Rates of Men, Aged 35-74, in Japan, by Area Pollution and Smoking Habitsa Lung-Cancer Rateb Low Pollution Smoking Category Intermediate High Pollution Pollution Nonsmokers 11.5 (5) 3.8 (1) 4.9 (1) Exsmokers 26.2 (11) 42.6 (7) 61.7 (7) Cigarette-smokers: 1-14 cigarettes/d 10.6 (9) 14.2 (10) 23.5 (14) 15-24 cigarettes/d 14.7 (18) 19.1 (17) 27.0 (17) 25+ cigarettes/d 36.3 (19) 15.8 (4) 46.4 (9) aReprinted from National Research Council26 (Table 17-26~; data derived from Hitosugi.16 bPer 100,000 per year, standardized for age. numbers of lung-cancer deaths. C-17 Figures in parentheses are

OCR for page 421
TABLE C-4 Smoking Habits and Lung-Cancer Mortality Rates of Bri tish Gasworkers Non- Ex smokers, smokers, OF ~0 Population "Exposed" gasworkers Othe r gasworkers 8.3 10.2 a ~0 Cont inning Smokers, % Pipe Mixed 1-9 10-19 Lung Canc e r Cigarettes/d Mortality Ratea 6.7 4.4 18.1 38.5 13.9 5.8 15.3 5.9 6.2 17.8 35.5 13.4 aPer 100, 000 per year, standardized for age. C-18 3.61 1.49

OCR for page 421
TABLE C- 5 Lung-Cancer Mortali ty in U. S . Nonsmokersa Age, yr 35-44 45-54 55-64 65-74 75-84 Annual Mortality Rate, per 100, 000 2.8 5.8 13.9 25.6 49.4 aReprinted with permission from Doll, 5 based on data from Kahnl9 and Hammond.l3 C-19

OCR for page 421
TABLE C- 6 Estimates of Lifetime (7C yr) Lung-Cancer Risk from Exposure to BaP Source at 1 ng/m3 Study Populat ion Risk, per 100, 000 Li f e t ime Lung-Cance r Gasworkers 43 L iverpoo 1 North Wales Al 1 men Nonsmoke r s Nonsmokers - S3 20 <61 C-20 Re ferenc e Doll _ al.6310 S tocks38 DollS

OCR for page 421
TABLE C-7 Estimates of Potency of Organic Extracts from Various Sources of Air Pollutiona Viral SENCAR Trans- L5178Ye Source Bapb Micec formationd - . + Coke oven478 2. 101 0.859 0. 726 9.963 (gasworks) Roofing tar889 0. 535 2.066 0. 311 9. 55 6 Caterpillar2 0.011 0.039 0.156 0. 049 3304 D Oldsmobile2 O. 156 0.067 0.970 0. 764 350 D Volkswagen26 -- 0.128 2.545 1. 012 Turbo D Mus tang II103 0. 02 7 0. 204 0. 348 0 . 990 302 V-8 ? catalyst aData from Harris.15 bNanogram.s of BaP per mi 11 igram of extract . CTumor initiation in SENCAR mice, papillomas/mouse per milligram of extract at 27 wk. dEnhancement of SA7 viral transformation in Syrian hamster embryo cells, transformations per 2 x 10~ cells per nanogram of extract per milliliter. eL5178Y mouse-lymphoma mutagenesis assay (average mutant colonies/106 survivors per microgram of extract per milliliter) without (-) and with (+) metabolic activation. C-21 -

OCR for page 421
30. Pike, M. C., and B. E. Henderson. Epidemiology of polycyclic hydro- carbons: Quantifying the cancer risk from cigarette smoking and air pollution, pp. 317-334. In H. V. Gelboin and P. O. P. Ts'o, Eds. Polycyclic Hydrocarbons and Cancer. Vol. 3. New York: Academic Press, 1981. 31. Raffle, P. A. B. The health of the worker. Brit. J. Indust. Med. 14:73-80, 1957. 32. Redmond, C. K., A. Ciocco, J. W. Lloyd, and H. W. Rush. Long-term mortality study of steelworkers. VI. Mortality from malignant neo- plasms among coke oven workers. J. Occup. Med. 14:621-629, 1972. 33. Royal College of Physicians of London. Smoking or Health. The Third Report from the Royal College of Physicians of London. London: Pitman Medical Publishing Co. Ltd., 1972. 128 pp. 34. Santodonato , J ., P. Howard , and D. Basu. Heal to and ecological assessment of polyouclear aromas ic hydrocarbons . J. Environ. Pathol. Toxicol. 5:1-364, 1981. 35. Sawicki, E. Airborne carcinogens and allied compounds. Arch. Environ. Health 14:46-53, 1957. 36. Sawicki, E., W. C. Elbert, T. R. Hauser, F. T. Fox, and T. W. Stanley. Benzota~pyrene content of the air of American communities. Amer. Ind. Hyg. Assoc. J. 21:443-451, 1960. 37. Segi, M., M. Kurihara, and T. Matsuyama. Cancer Mortality for Selected Sites in 24 Countries. No. 5 (1964-1965~. Department of Public Health. Sendai, Japan: Tohoku University School of Medicine, 1969. 174 pp. 38. Stocks, P. Cancer in North Wales and Liverpool regions. Supplement to British Empire Cancer Campaign Annual Report, 1957. 39. Stukonis, M. K. Cancer incidence cumulative rates. IARC Internal Technical Report No. 78/002. Lyon, France: International Agency for Research on Cancer, 1978. 40. U.S. Department of Health, Education, and Welfare. Office on Smoking and Health. A Report of the Surgeon General. DHEW PubLication No. (PHS)79-50066. Washington, D.C.: U.S. Department of Health, Education, and Welfare, 1979. 1196 pp. 41. -~aller, R. Trends in lung cancer in London in relation to exposure to diesel fumes, pp. 1085-1099. In HeaIth Effects of Diesel Engine Emissions: Proceedings of an International Symposium. EPA-600/9-80-057b. Cincinnati: U.S. Environmental Protection Agency Office of Research and Development, 1980. 42. Wynder, E. L., and D. Hoffmann. Experimental tobacco carcinogenesis. Science 162:862-871, 1968. 43. Wynder, E. L., and D. Hoffmann. Tobacco and Tobacco Smoke: Studies in Experimental Carcinogenesis. New York: Academic Press, 1957. 730 pp. 44. Wynder, E. J~., K. Mabuchi, and E. J. Beattie. The epidemiology of lung cancer. Recent trends. J.A.M.A. 213:2221-2228, 1970. 45. Wynder, E. T.., and S. O. Stellman. Impact of long-term filter cigarette usage on lung and larynx cancer risk: A case-control study. J. Natl. Cancer Ins t. 62:471-477, 1979. C-28

OCR for page 421
APPENDIX D PUBLIC DECISION-MAKING WITH RESPECT TO ATMOSPHERIC PAH SOURCES AND EMISSIONS Lawrence J . Whi te Among the possible justifications for public decision-making with respect to P~\H sources and emissions would be a finding that PAHs pose an actual or potential (and nontrivial) threat to human health. This appendix uses the cancer-risk estimates developed in Appendix C. It assumes that benzotaipyrene (BaP) can be used as a proxy for PAHs and that human 3 exposure to BaP in the ambient air at an average concentration of 1 ng/m over an entire lifetime has the effect of increasing by 0.02-0.06% the risk of dying prematurely (at or before the age of 70) because of lung cancer. Although the appropriateness of BaP as a surrogate for PAHs in general has been questioned, it has been so used extensively in the past, and much of the available information refers to it as an indicator for exposure to PAHs. The estimates of Appendix C are also based on this application. The focus of this appendix is on the lung-cancer consequences of human exposure to atmospheric sources of PAHs. The rationale for public decision-making with respect to PAH emissions from atmospheric sources is explored first, followed by discussions of the general problems of developing the appropriate decision-making tools, deciding on appropriate levels of control, and choosing appropriate means of implementing the decisions. The principles developed are then applied to PAH emissions of various sources, within the constraints of the limited amount of information that is available. These efforts should be viewed primarily as illustrative and approximate, because the data available are rough and approximate. Complete analysis would require a direct linking of the damage caused by an air pollutant to the sources of its emission. For that, the following would be needed: data on emissions of the pollutant, a model of the pollutant's dispersion and possible transformation or decay during dispersion, estimates of the resulting concentrations in the ambient air, data on human exposure to those concentrations, and a model of the exposure dose-response relationship. Reliable estimates of the costs and consequences of control are also needed. With respect to all these subjects, the relevant data on PAHs are scanty and approximate, and compromises will have to be made. Some estimates may be in error by as much as an order of magnitude. Nevertheless, the results should be informative and point the way toward further appropriate study. RAT IONALE PAH emissions from atmospheric sources are in a category of phenomena that economists have labeled "negative externalities" or "negative spillovers." The designations imply that people are taking actions (e.g. , producing coke, driving vehicles, and burning refuse) that generate, as byproducts or as incidental consequences, uncompensated costs imposed on D-]

OCR for page 421
other parties, outside of a market context; i.e., the PAR emissions pro- duced incidentally by these activities ultimately have potentially un- favorable health consequences for others. In such situations, persons who are motivated largely by the prospect of private gain (or, in the case of firms, private profit) are unlikely to take corrective action. Without incentives for corrective action, too much of the activity will occur, and too little effort will be devoted to reducing the costs imposed on others. An externality is an indication of a market failure;3 i.e., even an otherwise properly functioning competitive economy will not achieve an optimal allocation of society's resources, because of the distortion introduced by the externality. In a private-enterprise economy, the source of the problem created by an externality can be traced to an ill-defined property rights (neither the emitters of PAHs nor those who are exposed have a well-defined property right to the ambient air and its cleanness) or to the difficulties of enforcing a property right. The latter difficulties are usually due to the "public-goods" aspects of the phenomena; e.g., because an improvement in air quality in a locality will be enjoyed by all, each individual has an incentive to let others make the necessary effort to enforce emissions reductions, and this incentive for "free riding" leads to too little (or no) action. Externalities (especially those involving public-goods aspects) provide a case for possible public intervention in a private-enterprise economy. But whether, in practice, government intervention to correct an externality increases or decreases societal welfare is an empirical question. LEVELS OF CONTROL . Once an externality has been identified and the decision has been made that some kind of corrective action is warranted, further decisions must be made on the extent of corrective action (e.g., the desired degree of reduction in PAH emissions or the amounts of PAHs that will still be allowed to be emitted) and on the specific tools that are to- be used to implement the desired level of control. This section addresses the former issue, leaving the latter for the next section. The control of an externality brings societal benefits: a reduction in the externality costs imposed on others. In the case of PAHs, reductions in PAR emissions that translate into reductions in human exposure to PAHs mean the avoidance of some premature deaths (frequently termed "the saving of lives") and the avoidance of PAH-induced illness. But the achievement of these benefits almost always involves societal costs: individuals and firms must be induced to change their behavior with respect to emissions, engage in less of their desired activities, and incur costs (use real resources) to reduce emissions. Society's resources are scarce--in essence, society does not have limitless resources and cannot achieve all its desired objectives simultaneously, but must choose among them--and any level of externality D-2

OCR for page 421
control involves both societal benefits and societal costs; therefore, decisions concerning levels of control should focus on levels that best use society's scarce resources in trying to maximize societal welfare--i.e., society ought to aim for levels of control that provide the greatest margin of benefits relative to costs. Two main analytic tools have been developed that can aid decision- makers in choosing the appropriate levels of control: cost-effectiveness analysis and cost-benefit analysis. Cost-effectiveness analysis is the more limited of the two. It takes, as a given, a specific societal goal (ob jective)--~.g., a reduction in emissions by X tons of a specific pollutant or the incurring of only up to Y dollars for the reduction of emissions from a specific source of that pollutant. The principle of cost-effectiveness requires a search to identify the least costly way of achieving a reduction in pollutant emissions. If all sources of the pollutant have equal environmental consequences, then the emission source with the lowest marginal (incremental) cost of control should be chosen. For example, if one source has a marginal control cost of t500/ton and another a marginal cost of $3,000/ton, the first should be chosen over the second. The choice of the first will mean that the achievement of emission reduction by X tons will require less resources, or the expenditure of Y dollars will achieve a greater reduction. The formal principle is that, in achieving the goal, the marginal costs of control from all sources ought to be equated. If this principle is violated, then the cost of achieving a given level of overall control could be reduced (or the level of overall control achieved at given costs could be increased) by increasing the stringency of control from the low-marginal-cost sources and decreasing the stringency of control from the high-marginal-cost sources. Cost-effectiveness analysis can be a useful tool for improving the efficiency of individual programs and for comparing the effectiveness of similar programs. But cost-effectiveness analysis cannot be used to answer the ultimate policy questions: "Should X tons or lOX tons of pollutant-emission reduction be the appropriate societal goal?" "Should a cost of Y dollars or 20Y dollars be incurred to achieve emission reduction?" But cost-benefit analysis can provide an analytic basis for making these decisions. There are only a few primary steps in a cost-benefit analysis. The societal benefits and societal costs should be estimated and converted into dollar equivalents (if they are not already in dollars). An interest rate (discount rate) must be used to convert future benefits and costs into present-value equivalents. The projects (or alternative versions of a project, e.g., alternative levels of stringency of required emission reductions) with the highest margins of benefits relative to costs should be the ones chosen. An equivalent principle is that, in choosing among alternative versions of a project (say, alternative levels of emission control stringency), stringency should be adjusted until the marginal benefits of extra stringency are just equal to the marginal costs The basic methods of cost-benefit analysis are, by now3 standard;13~3 D-3

OCR for page 421
controversies rem2gin, however, as to the interest rate that should be used for discounting, whether the income-distribution consequ9ences of pro jects should be considered explicitly in the analysis,] how to incorporate risk and uncertainty into the analysis, and how (and whether) to place dollar values on nonmarket items and concepts. In this last category, a frequent question that arises in the context of cost-benefit analysis applied to projects or programs that have mortality or morbidity consequences (e.g., many pollutant-emission control programs) is how (and whether) to evaluate the benefits of mortality or morbidity reduction. Claims that "a life is priceless" and that "one cannot put a value on a life or on pain and suffering" are often heard. A logical implication of these claims seems to be that cost-benefit analysis is useless in such instances--that for such projects, so long as any mortality reduction ("lives to be saved") or morbidity reduction (reduction in "pain and suffering") can be achieved, a project or program should be pursued (or extra stringency pursued), regardless of costs. This approach to the benefits of reductions in mortality or morbidity does not provide a useful guide for making societal decisions, because the opportunities for achieving reductions in mortality and morbidity are virtually limitless. Additional resources devoted to medical research, medical care, accident prevention, and pollution reduction are likely to yield reductions (albeit possibly small) in mortality and morbidity. Society could use up its entire gross national product by devoting ever-increasing amounts of resources to the pursuit of such reductions. But, in fact, we do not. Through our societal decision-making processes, at some point we desist. For example, in the wake of the Arab oil embargo of 1973, the Congress enacted a law imposing a national highway speed limit of 55 mph. The major goal of the legislation was to reduce American gasoline consumption, but it was soon learned that the 55-mph speed limit had the beneficial side effect of reducing highway mortality. There have been no efforts to reduce the speed limit to, say, 45 mph, although such a reduction would clearly reduce highway mortality even more. Similarly, society does not build pedes trian underpas ses for every busy urban intersection and does not station ambulances near those intersections, despite the reductions in mortality and morbidity that would be achieved. In effect, society has decided that the extra mortality and morbidity reductions are not worth the resources (costs) that would have to be devoted to achieving them; lines have been drawn. Drawing these lines has been a largely implicit process; drawing them explicitly apparently makes many people uneasy. They are reluctant to put a value on mortality or morbidity reductions. But a society that wishes to achieve the best that it can from its scarce resources must understand the uses to which those resources are put and the tradeoffs (the "opportunity costs") involved. A society may well have multiple goals. Nevertheless, an understanding of the tradeoffs is important in pursuing them; and the use of explicit values for mortality and morbidity reductions is necessary for that understanding. Furthermore, the logic of cost-effectiveness D-4

OCR for page 421
argues for the consistency of these values across projects; otherwise, societal resources are allocated in an ineffective way, as apparently has been the case for actual projects and programs involving mortality and morbidity reductions. A good case can be made, then, for using explicit values for mortality and morbidity reductions. There are a number of candidates for establishing the value of mortality reduction (or, alternatively, "the value of a lifers: The expected discounted future earnings of a person. o The life insurance held by a person. The average (or some other summary measure) of the implicit values yielded by other, recent projects or programs involving mortality reduction. o Compensation awarded in trials involving premature deaths. O Estimates of the value that people, in their day-to-day behavior, place on incurring or avoiding risks of premature death. For the purposes of deciding on the appropriate levels of pollution control, Bailey and Freemanl6 (Chapter 4) have reviewed and criticized these measures. The last measure (risk valuation) is most consistent with the market valuations that are the other components of cost-effectiveness and cost-benefit analyses. An important point here is that pollution- reduction programs (and accident-reduction programs) do not have a knowable effect on specific persons' lives; they do not involve before-the-fact specific deaths. Instead, if they are effective at all, they reduce the probabilities or risks of the premature death of exposed persons. After the fact, this reduction in risk must mean a reduction in premature deaths; but before the fact, the programs can be evaluated only in terms of risk. Because the affected persons benefit from the reduction in risk and because virtually all people expose themselves to risks in their day-to-day behavior (whether they acknowledge it or not), the benefit o f the risk reduction should be roughly comparable with the value of the risks that they incur or avoid ~ at the margin) in their day-to-day behavior. In essence, if they are asked, "What would you be willing to pay in return for a reduction in risk?" or "What would you need to receive to compensate you for an increase in risk?" their responses should be roughly consistent with their private behavior. In a market economy, the prices of goods and services reflect (at the margin) a willingness to pay for those goods and services. Public projects, to maximize the societal value that can be achieved from society's resources, should also use willingness-to-pay measures for valuation purposes wherever possible. Accordingly, the risk-valuation approach is consistent for assessing pollution-reduction programs. There are no specific markets in the private sector where one could directly observe a person's willingness to pay for risk reduction. But people do choose to incur or avoid risk, gaining or giving up other things D-5

OCR for page 421
in return, in most aspects of their lives: They choose jobs that have higher or lower risks of accidental death or injury, in return for explicit or indirect wage premiums; they choose to use or not to use seatbelts in automobiles, trading off time and convenience against reduced risk of death or injury in the event of a crash;4 they choose to live in neighborhoods with higher or lower air-pol lutant concentrat ions, trading off housing costs against the extra risks of mortality or morbidity from the pollutants;30 and so on. Economists have been able to provide models of individual behavior and, with actual data and econometric estimation techniques, estimate the implicit value that people have placed on the risks that they have incurred or avoided. For example, to estimate the wage premium that accompanies extra risk, a researcher could collect a sample of wage rates for various occupat ions, the ac tuarial data on accidental deaths for those occupations, and data on the various influences on wage rates (e.g., degree of unionization, amount of education, extent of experience). The econometric techniques allow the researcher to control for the other influences and thus to infer the implicit wage premium that accompanies extra risk. Placing a value on reducing the risk of death is very difficult and controversial. Different values can be assigned. However, the values discussed here fairly represent the research that has been done in this field, and they provide a useful re ference guide for decision-making with regard to pollution control. Studies of the value of risk do not yield identical estimates, but, as Bailey showed, they can be grouped (after appropriate adjustments and corrections) into a range of $170-715 (in 1978 dollars) in annual payment per 0.O31 (i.e., 0.1%) additional annual risk of death. ~ study by Portney O yielded an additional estimate that is in the middle of this range. Freemanl6 argued that the most likely value is $1,000 (in 1978 dollars) per 0.001 additional risk. This same figure was used in the NRC study (pp. 744-245) of the costs of removing chloroform and other trihalomethanes from drinking water.2 Some problems of using these studies and the estimates they yield for evaluating public pollution-control programs should be noted. First, as with the use of any econometric model, one needs to be satisfied that the model has been properly specified and all important influences properly accounted for. Second, the models assume that the persons involved were aware of the risks they were incurring or avoiding. Third, use of the models' estimates for public-policy purposes assumes that the persons in the sample (and hence the estimates of the value of risk) are typical of the general population. If a wage study included only or mostly high-risk occupations, the resulting estimate of the value of risk might be an underestimate of the value that applies to most of the population, since persons with less fear of risk would likely gravitate toward high-risk occupations or housing locations--i.e., self-selection might bias the results. Fourth, people may feel differently about (value differently) risks over which they have more control (e.g., job choice) and risks over D-6

OCR for page 421
which they have less control (e.g., the general level of pollution in the air they breathe). Finally, even if the models' estimates are representa- tive of the general population's valuation of risk, individual persons will have different values of risk and hence different perceptions of, say, the concentrations for which a pollution-control program should aim. Within a locality, however, all persons will have to be exposed to roughly the same po 11utant concentra ~ ions . The last problem is an unresolvable dilemma that is inherent in the public-goods aspects of most pollution prowlers, which cause them to be the proper concern of nonindividualistic, government action in the first place. This dilemma is present for all public goods (e.g. , national defense and local police protection) that people consume generally automatically and equally as part of a community. As Samuelson31 has demonstrated, the proper procedure for deciding on the appropriate level of a public-goods project is to sum the valuations of all affected persons and extend the project to the point at which the sum of the marginal valuations (benefits) equals the marginal cost of the extension--exactly the criterion stated in the discussion of cost-benefit analysis. Despite the possible problems, the range of estimates yielded by the risk-valuation studies does appear to be reasonable when compared with the income of a typical family and the safety-related expenditures it would find worthwhile. One aspect of the risk-valuation estimates is worth emphasizing. If one finds that people appear to be willing to pay $500 her year each to avoid a 0.001 risk of death in a given year, the proper use of this estimate is as follows: Suppose a government pollution-control program can reduce the risk of death in a community of 1 million by a factor of 0.001. Then, because each person, on average, should be willing to value this improvement at about $500 per year, the 1 million people in the community should be willing to pay about $500 million per year for these benefits, and this aggregate value could be compared with the anticipated cost of the program. In essence, the aggregate cost of the benefit is estimated by multiplying the typical person's valuation of the risk reduction by the number of persons involved (reduction in risk per person). In contrast, the value of risk is sometimes extrapolated to a value of avoiding (or, in reality, delaying) a death or "the value of [extending] a life" ; i. e., the $500 per 0.001 risk would be extrapolated to $500,000 as the value of avoiding a death. It is true that, if the government implements the hypothetical program just mentioned, there will be 1,000 fewer deaths per year; and, because the program was valued at $500 mill ion per year, this implies a value of $500, 000 per avoided death. Furtl~er- more, for some purposes, it is sometimes convenient to speak or write in terms of "the value of a life" (or the value of a statistical life). But . there is nothing in tone statistical or conceptual procedures that leads to the conclusion that any person would, could, or should pay $500,000 to avoid a certain death. Rather, before the fact, the government pro ject D-7

OCR for page 421
promises a change in risk, not a change in the certainty of death for any person. People behave toward and implicitly value risk in their everyday life, so risk valuation is the consistent conceptual procedure to use. The discussion thus far has focused entirely on valuing mortality changes. In principle, the same procedures could be applied to valuing changes in morbidity--i.e, willingness-to-pay measures could be inferred from persons' behavior. There do not appear to be any studies that have tried to generate such estimates. Instead, estimates of the medical costs and lost productivity related to illness and accidents are usually used to estimate these societal costs (and hence the societal benefits from their reduction). These estimates may not be too far away from what the appropriate willingness-to-pay measures, if they existed, would indicate, except that the former probably underestimate the latter by excluding the value of avoiding pain and suffering. Finally3 the limitations of cost-effectiveness and cost-benefit analysis must be acknowledged. Knowledge about costs and benefits is never perfect; in some cases, it may be quite imperfect. Risks and uncertainties often pervade analyses. Society has multiple goals. But, in the end, society's resources have to be allocated, and those resources are scarce and have alternative uses. Cost-effectiveness and cost-benefit analysis, imperfect though they may be, can be aids to effective societal decisions making. IMPLEMENTATION - Regardless of the target level of control desired, a number of choices with respect to the implementation of an emission-control program are possible. A useful dichotomy is provided by the division between fiat methods (frequently called "command and control") and methods that rely on the use of economic incentives. At one extreme, after a desired reduction in emissions (or a desired level of remaining emissions) has been ascertained, a central regulatory control agency can attempt to specify to each emitter (or class of emitters) the reduction or allowable emissions that will be required. If the agency wished to minimize the societal cost of achieving the emission reduction, it would try to have complete information about the total and marginal cost schedules for each of the various emitters and allocate reduction or emission appropriately, following the precepts of cost- effectiveness analysis. At the other extreme, the agency could set an effluent fee that would require an emitter to pay a specified amount per unit of the pollutant that was emitted. In the presence of rising marginal costs of control ~ emitters would find it worthwhile to reduce emissions to the point at which the marginal cost per unit of pollutant reduction was equal to the effluent fee. The same knowledge of cost schedules assumed above would allow the D-8 Hi;

OCR for page 421
agency to set an effluent fee that would achieve the same reductions as those achieved by the fiat method. As the previous paragraphs indicated, under conditions of complete certainty, the two methods can achieve the same outcome. But knowledge about the costs of control is rarely complete. With incomplete knowledge, the control agency is likely to make socially costly mistakes by improperly assigning excessive emission reductions to emitters with high marginal costs of control. The effluent-fee system has an important advantage in this respect, in that it allows the high-cost and low-cost emitters to sort themselves out and achieve the lowest overall cost of control through their own behavior. Incomplete knowledge of costs may also lead to effluent-fee schedules that are too high or too low, with consequent emission reductions that are off target. But the schedules can be readjusted by continuing to observe emissions; incorrect assignments under the fiat method may never be corrected, because correct cost-information is not automatically revealed. An alleged advantage of the fiat method is its apparent certainty of outcome. Emitters will be told to reduce their emissions by a specified amount, and that reduction "will" be achieved. The effluent-fee method appears to be more indirect; one has to rely on the cost-reduction consciousness of firms and individuals to recognize that reducing emissions (up to a point) is less costly than paying effluent fees. But experience with pollution-control programs has shown that even the expected certainty of the fiat method often does not materialize.27~44 Many emission- control programs are intended to be "technology-forcing"; they try to set emission standards that are beyond the economical range of current technology, thus attempting to force the development of advanced technology. The ostensible sanctions for failure to meet emission standards are usually severe fines or closure of offending companies. But if the technology appears not to be available, the sanctions are not credible or enforceable. Furthermore, regulators may have difficulty in ascertaining whether the necessary technology is or is not available or economical or whether a good-faith effort has been made to develop the needed technology. As a consequence of these uncertainties, the emitters (especially in an industry with a relatively small number of large firms) have an incentive to slow down their own technology development. Thus, the apparent certainty of success of the fiat programs is not necessarily reflected in actual practice, as the delays in the implementation of many pollution-control programs have revealed. Even if the sanctions behind them are thought to be credible, fiat methods can lead to the development of inefficient techniques. Technologies that are low in cost but that may fall short of the standards are unlikely to be pursued; technologies that can, at low cost, reduce emissions beyond the point set by the standards will be pursued only to the point set by the standards; technologies that are expected to be low in cost but have an uncertain likelihood of probability of success will be D-9 Hi;

OCR for page 421
discarded in favor of high-cost, more certain technologies. An effluent- fee system would not have these inefficiency properties. Another method of control that retains most of the incentive properties of effluent fees, but also has some of the possible quantitative certainty of a fiat system, is a system of marketable emission permits.! Under this system, the central regulatory agency sets a target of maximal total emissions of a given pollutant. It then creates a set of permits equal to this total . The permits are, in essence, a property right in a given amount of emissions. No one is allowed to emit without a permit; thus, each emitter must control emissions down to the point for which permits have been received. The agency could auction off the permits to the highest bidder (thus lodging the property right in clean air initially with the government) ~ or it could initially assign the permits among emitters' or even among the population generally, in some manner (thus initially ass igning the property rights in the manner chosen) . If the permits are auctioned or can be traded, emitters will again sort themselves into an efficient, least-cost pattern, with low-cost emitters choosing to control emissions more and buying relatively fewer permits and high-cost emitters doing the opposite. It is clear that, with appropriately chosen targets (costs and emissions), an effluent-fee system and a marketable-permit system can achieve the same outcome with comparable incentive effects. One difference between them is that the effluent-fee system always implicitly lodges the property right with the government, whereas the marketable-permit system may lodge the property right with the government (if the auction method is used) or in the private sector (if some assignment scheme is used). Another difference is in the identity of the group that bears the risk in the event O2f uncertainty about or variation in emitters' marginal-cost schedules.3 In an effluent-fee system, variation in marginal-cost schedules will mean that variation can be expected in the quantities of emissions; thus, the risk is borne by those who are exposed to the emissions. In a marketable-permit system, variation in marginal-cost schedules will mean variation in the prices paid for the permits; the risk is borne by the emitters. The choice between the two systems on these grounds should be determined by examining the societal cos ts of lodging the risk with one group or the other. If, for example, the health consequences of small variations in emissions could be severe, a marketable-permit scheme would be preferred; if, however, the health consequences of small variations in emissions are not severe and the price variance of permits would cause firms to take relatively costly offsetting actions, the effluent-fee system would be preferred. Even within the context of a fiat system, there are measures that increase the scope of economic incentives and efficiency. For stationary- source emissions, a "bubble" strategy that allows individual firms to trade off pollutant emission from different sources (e.g., different smokestacks) at the same geographic location provides the possibility of reducing the cost of controlling emission by a given amount.23 ~ 25 In essence, an D-10