Some environmental statutes require the U.S. Environmental Protection Agency (EPA) to balance benefits and costs when issuing regulations, while other statutes prohibit such balancing. But do these requirements or prohibitions make a big difference in the regulations that are written? According to a recent study conducted at Resources for the Future, the answer is "no." The study reveals that both benefits and costs appear to have influenced the regulations issued by EPA, regardless of the statutory mandate under which the agency was operating. The study also suggests that the value that EPA implicitly attaches to the prevention of one case of cancer is very high—from $15 million to $45 million. This is much more than individuals appear to be willing to spend to reduce their own risks of death.
Under the various environmental statutes the U.S. Environmental Protection Agency (EPA) administers, the agency is responsible for issuing regulations to protect the public from exposure to pollution. These regulations can include outright bans of certain products—for instance, pesticides and products containing asbestos. They more commonly include limitations on the amount of pollution a factory or vehicle can emit.
Most economists would argue that these regulations should be made, at least in part, on the basis of benefit-cost analyses. That is, they believe that an environmental standard should be set just at that point where the marginal cost of setting a slightly more stringent standard would begin to outweigh the marginal benefit of increased stringency. Congress, however, sometimes limits EPA's ability to engage in such balancing when the agency issues regulations.
For example, under the provisions of the Clean Air Act that pertain to the establishment of maximum permissible air pollution concentrations, EPA cannot take costs into account. When establishing effluent standards under the Clean Water Act, the agency is allowed to consider costs but not benefits. Only two environmental statutes—the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and the Toxic Substances Control Act (TSCA)—actually require that EPA balance the benefits and costs of regulation in setting environmental standards.
Recently, we conducted an after-the-fact analysis of regulatory decisions that EPA has made over the last two decades. Our purpose was to see whether EPA appears to have balanced costs and benefits in issuing its regulations, regardless of whether the law under which the agency is operating directs or prohibits this balancing. In our study, "balancing" is determined by the following question: If we look back at a class of regulations EPA has put in place—for example, emissions standards for toxic air pollutants—do variations in the costs and benefits of all the regulatory options the agency considered at the time help explain the standards selected? If the answer is "yes," we argue that balancing has taken place.
We conclude that EPA has acted as if both costs and benefits influence its selection of regulatory standards; specifically, other factors being equal, a more costly standard is less likely to have been selected than a less costly one, and a standard that saves a greater number of lives is more likely to have been selected than one that saves a smaller number of lives.
Intuitively, however, balancing requires more than just paying some attention to costs and benefits. It also requires that the cost EPA is willing to impose on society to save an additional life be regarded as "reasonable." One way to determine what EPA considers reasonable is to see if there is some threshold value for the cost-per-life-saved above which the agency has been reluctant to issue regulations. (For lack of a better term, we call this threshold value the "value of a life" implied by the regulations.) For each class of regulations that we examined, we calculated the value of a life that was implicit in the regulations.
We conclude that EPA has acted as if both costs and benefits influence its selection of regulatory standards.
We were especially interested in two issues. The first and most important of these is how the value of a life implicit in environmental regulations compares with society's apparent willingness to pay to save lives: Is this value acceptable to American society? The second issue concerns the way in which the implicit value of a life seems to vary across EPA programs and across population groups: for instance, do environmental regulations pertaining to pesticides place a higher value on a life saved than regulations pertaining to hazardous air pollutants? Also, does EPA implicitly attach more weight to saving the life of a worker exposed to pesticides or asbestos on the job than to the life of a consumer exposed to these pollutants?
To answer these questions, we gathered data on EPA-estimated costs and benefits associated with three categories of pollutants that EPA regulates:
- asbestos, sources of which are regulated under the Toxic Substances Control Act;
- all cancer-causing pesticides used on food crops that underwent EPA's Special Review process between 1975 and 1989; and
- all carcinogenic air pollutants for which EPA set National Emissions Standards for Hazardous Air Pollutants between 1975 and 1990.
When we gathered data for each source of these pollutants (each crop in the case of pesticides), we arrived at a total of 39 asbestos regulations, 245 pesticide regulations, and 40 regulations pertaining to four hazardous air pollutants—benzene, inorganic arsenic, radionuclides, and vinyl chloride.
We limited our study to the regulation of carcinogens because quantitative risk data—that is, estimates of the number of lives at risk as a result of exposure to a particular pollutant or product—are available more often for carcinogens than for other substances. The availability of such data for carcinogens implies that the number of lives saved by each of the regulations we examined can be quantified. We also purposely selected some regulations issued under the two statutes (TSCA and FIFRA) that require EPA to balance costs and benefits, as well as those regulations that set emissions standards for hazardous air pollutants under the Clean Air Act, which prohibits such balancing. We included these regulations in our study in order to determine whether the directives given EPA in the enabling legislation made any difference in the way in which the agency appeared to weigh benefits and costs.
It is important to be clear about one thing. We were not privy to EPA's decision-making process for any of the regulations discussed here. What we have done is to look back at the information on benefits and costs that the agency had when it formulated the regulations, to examine the pattern of regulatory decisions, and then—using statistical analysis—to ascertain whether these decisions were consistent with the hypothesis that benefits and costs influence the regulatory outcome, regardless of the statutory mandate.
We turn now to a discussion of the specific regulations.
Asbestos regulations under TSCA
In 1985 EPA announced its intent, under the authority of the Toxic Substances Control Act, to ban the use of asbestos in 39 products. Because TSCA requires EPA to balance benefits and costs, the agency's Notice of Intent to Regulate was followed by a detailed assessment of the health risks associated with exposure to asbestos fibers, as well as the costs that would result from the proposed bans.
Cost-effectiveness of asbestos ban
Well-documented epidemiological evidence indicates that some forms of asbestos are human carcinogens. This evidence is particularly strong for lung cancer, gastrointestinal cancer, and mesothelioma, a cancer of the lung or abdominal lining. Estimating the number of cancer cases associated with a particular asbestos-containing product (for example, brakes lined with asbestos) requires estimates of the potency of asbestos—that is, of the likelihood that an individual will develop cancer as a function of exposure—as well as of the number of people who are exposed to various levels of asbestos.
In the analysis accompanying its final rule, EPA presented estimates of consumers' and various groups of workers' exposure to each product to be regulated, as well as an estimate of the number of cancer cases then-currently associated with each of these sources of asbestos. It also calculated the number of cancer cases that would be avoided if each product were banned. EPA was able to estimate this number, as well as the cost of the ban, for 31 of the 39 products considered for regulation.
EPA's asbestos regulations imply that the value of a cancer case avoided is $49 million; this value seems high considering that the value of life implicit in workers' occupational choices is about $5 million.
A plotting of the regulatory costs and the number of cancer cases avoided for each of the 31 products for which complete data are available (see figure, p. 7) is consistent with the hypothesis that EPA considered benefits and costs in issuing its asbestos decision. Products for which the cost of the ban was low and the number of lives saved was high (tending toward the lower right-hand corner of figure) were almost always banned, while products for which the cost of the ban was high and the number of lives saved was low (tending toward the upper left-hand corner of figure) were for the most part not banned.
Since avoiding cancer cases is the only benefit of the asbestos ban mentioned by EPA (ecological risks, for example, not being mentioned), it is tempting to infer from the plot that there is a threshold value for a cancer case avoided below which all products were banned. For instance, the rule "ban only those products for which the cost-per-life-saved is less than $10 million" (a rule illustrated by the lower of the two diagonal lines in the figure) would explain many of the bans, but it would yield incorrect predictions for some products. Similarly, the rule "do not ban any product for which the cost-per-life-saved is greater than $100 million" (a rule illustrated by the higher of the two diagonal lines in the figure) would be correct almost all the time, but it would yield incorrect predictions for some asbestos-containing products.
To compute the threshold value of a cancer case avoided that is implied by the asbestos regulations, we estimated statistically the line that maximized the number of regulations correctly predicted by the above-noted rule. We found that the implied threshold value of a cancer case avoided is $49 million (measured in 1989 dollars). (This value would have fallen between the two diagonal lines in the figure.) This value seems high—especially in contrast to estimates of the value of life that are based on individuals' willingness to pay for risk reductions.
Consider, for example, the added compensation that workers require to accept jobs that pose increasingly greater health risks, compensation that provides useful information about individuals' risk-reward trade-offs. Based on dozens of studies, the value of life that seems to be implicit in workers' occupational choices is about $5 million, an amount much lower than the value of life implicit in EPA's regulation of asbestos. While labor market compensation is for risks that are voluntarily borne, it is hard to imagine that the additional premium associated with involuntary risks is $44 million. Not coincidentally, perhaps, EPA's failure to give sufficient weight to the costs of regulation in issuing its asbestos bans was cited by the Fifth Circuit Court of Appeals in the Corrosion Proof Fittings case, which overturned the ban.
Pesticide regulations under FIFRA
Under the Federal Insecticide, Fungicide, and Rodenticide Act, EPA is responsible for ensuring that all pesticides used in the United States have no unreasonable adverse effects on the environment. If EPA suspects that a pesticide poses risks to human health or to ecosystems, the pesticide—or, more accurately, the active ingredients used in the pesticide—is subject to what is known as a Special Review.
A Special Review entails a formal risk-benefit analysis of the pesticide, after which EPA can either ban the pesticide for use on specific crops, restrict the manner in which the pesticide is applied, or allow its continued use without modification. Between 1975, when EPA initiated its first Special Review, and December 1989, Special Reviews of 37 active ingredients were completed. We restricted our analysis to those active ingredients that are suspected human carcinogens.
In considering whether or not to ban a pesticide, EPA examines risks of cancer to consumers of food containing pesticide residues and to persons exposed to the pesticide in the workplace—these are the people who mix the pesticides (mixers) and load them into the dispensing equipment (loaders), as well as those who apply the pesticides (applicators). The agency also examines noncancer health risks, such as risks of miscarriages or of fetal damage. In addition, it considers the adverse effects of the exposure of fish, birds, and mammals to pesticides.
The implicit value of a cancer case avoided among pesticide applicators was $52 million. Risks to mixers and loaders of pesticides and to consumers seemed not to influence EPA's decisions to ban uses of active pesticide ingredients.
Against these risks, EPA balances the benefits of pesticide use—that is, the gains to both farmers and consumers as a result of the increase in agricultural output brought about by pest control. Depending on the relative weight given to these and other factors, EPA might decide that a particular ingredient can no longer be used on a particular crop.
It is tempting to plot the cost of bans against the number of cancer cases avoided for pesticide regulations, as we did for asbestos regulations; however, the resulting diagram would be misleading. Because the avoidance of cancer cases is only one of the benefits of banning a particular use of a pesticide, our inferred threshold value of a cancer case avoided would overstate the value that EPA implicitly attaches to reducing cancer risks. Instead, we estimated a statistical model designed to predict EPA's decisions to cancel (or not cancel) the use of each of the active ingredients in pesticides on each of the food crops for which the ingredients were registered.
Our model, which correctly predicted 87 percent of the 245 decisions EPA made between 1975 and 1989, suggests that EPA considered both the risks and benefits of pesticide use in issuing its pesticide regulations. The benefits of pesticide use were statistically significant and of the expected sign: the higher the benefits of pesticide use, the less likely it was that a pesticide was banned for use on a particular crop. The risks associated with the pesticide were also important in explaining which uses of a pesticide were banned and which were not. Other factors being equal, the higher the risks of cancer to applicators (the group with the highest average exposure to pesticides), the greater the probability that a pesticide was banned. The implicit value of a cancer case avoided among applicators was about $52 million (1989 dollars)—a value remarkably close to the value we found to be implicit in asbestos regulations.
Our analysis was quite surprising in one respect: neither risks to mixers or loaders of pesticides nor dietary risks to consumers seemed to influence EPA's decisions to ban uses of active ingredients in pesticides. One possible explanation for this is that risks to both mixers and loaders and to consumers are lower than risks to applicators, and therefore seen to be a less pressing problem. The median lifetime cancer risk associated with exposures to the pesticide ingredients we examined was 1 in 1,000 for applicators but only 1 in 100 million for consumers of food with pesticide residues.
National emissions standards for hazardous air pollutants
In contrast to regulations issued under TSCA and FIFRA, the National Emissions Standards for Hazardous Air Pollutants were, according to the Clean Air Act of 1970, to be set to protect human health without consideration of costs. As we shall see, however, during the mid-1980s EPA did attempt to consider costs in setting emissions standards for sources of hazardous air pollution. In 1987, the Natural Resources Defense Council successfully sued the agency for making costs a factor in the determination of those standards. As discussed below, the ruling in that case had a pronounced effect on EPA's subsequent setting of standards for air pollution.
Section 112 of the Clean Air Act requires EPA to regulate the so-called toxic air pollutants, substances such as benzene, arsenic, asbestos, and mercury. These pollutants are not as ubiquitous as particulates, sulfur oxides, carbon monoxide, and other pollutants for which EPA is to set ambient air quality standards, but they are nonetheless harmful to human health. According to the Clean Air Act, EPA was required to establish a list of toxic air pollutants and then to set emissions limits for various sources of each pollutant. Between 1970 and 1990, only seven such substances were regulated. Five of these air pollutants are carcinogens, but quantitative risk data are available for only four—vinyl chloride, benzene, inorganic arsenic, and radionuclides. We examined the regulation of these substances.
After 1987, EPA always elected to regulate the source of a hazardous air pollutant if it posed a greater than 1 in 10,000 cancer risk to the maximally exposed individual. If this risk was less than 1 in 10,000, however, then the threshold value of a life saved was the same before and after 1987: $15 million.
In seeking to regulate the various sources of these four pollutants, EPA considered at least one regulatory option that would reduce emissions of each pollutant, as well as the option of no regulation. For each option, it computed cost, the number of associated cancer cases, and the post-regulation risk to the "maximally exposed individual," the individual who receives the greatest dose of a pollutant from a particular source. For most sources of hazardous air pollution, this individual is not exposed to the pollutant in the workplace, but rather lives near the source of the pollutant (for example, the person whose house is nearest to a copper or lead smelter).
To examine the possible trade-off between benefits and costs in the regulation of hazardous air pollution, we estimated a statistical model to explain which regulatory option was chosen for each of the 40 sources of vinyl chloride, benzene, inorganic arsenic, and radionuclides. Our results suggest that EPA's regulatory choices were consistent with the hypothesis that the agency was balancing the cancer risk reductions due to more stringent regulation against the costs this regulation would entail. In technical parlance, when we used all 40 sources of the four air toxics in our study to estimate our model, the coefficients of both the reduced cancer incidence and the regulatory cost were significant. The implicit value of a cancer case avoided—that is, the value that best enabled us to predict EPA's regulatory decisions—was very high—$153 million, to be exact.
These results look somewhat different, however, if we distinguish regulations issued before 1987 from those issued after 1987, when the U.S. Court of Appeals for the District of Columbia ruled that EPA had improperly considered costs in setting emissions standards for toxic air pollutants. In the so-called Vinyl Chloride decision, EPA was directed to consider the costs as well as the technological feasibility of regulatory options only after an "acceptable risk" level had been achieved.
When we modified our analysis to take this decision into account, our results came out quite differently: they implied that a cancer case avoided was valued at approximately $15 million before the 1987 court ruling and at the same amount after that ruling, so long as maximum individual risk was less than 1 in 10,000. After 1987, however, EPA always elected to regulate the source of a hazardous air pollutant if it posed a greater than 1 in 10,000 cancer risk to the maximally exposed individual; in other words, the threshold value of a life was infinite. If this risk was less than 1 in 10,000, however, then the threshold value was the same for the post-1987 regulations as for the pre-1987 regulations: $15 million (1989 dollars).
Surprises and questions
One of the striking findings of our analysis is that, in issuing the asbestos, pesticide, and toxic air pollutant regulations we examined, EPA has been willing to impose substantial costs on consumers and firms in order to save a life. Under each of the two statutes that allow the balancing of benefits and costs, the agency's implicit valuation of a cancer case avoided was in excess of $45 million. Whether members of society would agree with this valuation, which is about ten times greater than individuals implicitly value the risk of death due to occupational hazards, is an important question.
Our findings suggest that EPA has, in the past, put in place more stringent regulations under statutes that require it to balance benefits and costs than it does under a statute that directs it to ignore costs and consider health risks only.
Nevertheless, compensation for risks faced in the workplace is generally for voluntary exposure to immediate risk of death. Exposure to asbestos and pesticides may not be voluntary (even for workers) if people are unaware of the risks they face; this fact may account for the very high implicit value assigned to risk reductions in EPA regulations pertaining to these substances.
It is interesting to note that the value per cancer case avoided that is implicit in regulations pertaining to hazardous air pollutants was about one-third the value implicit in pesticide or asbestos regulations. In a sense, this is quite surprising. Our findings suggest that EPA has, in the past, put in place more stringent regulations under statutes that require it to balance benefits and costs than it does under a statute that directs it to ignore costs and consider health risks only. This does not "prove" that EPA balanced costs and benefits under the Clean Air Act, only that it made decisions that were consistent with the hypothesis that the agency behaved this way.
This in turn raises the question of whether statutes that prohibit consideration of costs in standard setting really make a difference in the regulations that are issued. Our analysis of the setting of the National Emission Standards for Hazardous Air Pollutants suggests that, short of recourse to the courts, prohibitions against consideration of costs may be difficult to enforce. Likewise, Congress may require that the costs of a regulation be balanced against its benefits; but as long as EPA has discretion in the weights it assigns to costs and benefits, the regulations it issues under statutes that allow balancing of benefits and costs may still be very costly.
George L. Van Houtven is a member of the Department of Economics at East Carolina University. Maureen L. Cropper, a senior fellow in the Center for Risk Management at Resources for the Future and a professor of economics at the University of Maryland, is currently on leave as a principal economist at the World Bank. A detailed discussion of the issues in this article can be found in discussion paper CRM93-02, "When Is a Life Too Costly to Save? The Evidence from Environmental Regulations," by Van Houtven and Cropper.
A version of this article appeared in print in the January 1994 issue of Resources magazine.
A version of this article appeared in print in the January 1994 issue of Resources magazine.
