Click for next page ( 108


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 107
6 Environmental Policy Making: Act Now or Wait for More Information? JEFFREY E. HARRIS Environmental policy making is a dynamic process. Rarely do regula- tory agencies make once-and-for-all choices between action and inaction. Instead, they choose, again and again, between degrees of action and wait- ing; making decisions that are based on information scientific, economic, political that changes continually. This dynamic quality of environmental decisions poses serious prob- lems for benefit~ost analysis. 1b evaluate a contemplated regulatory in- tervention, it is no longer enough to compare the intervention's currently estimated benefits and costs. In fact, it is insufficient to assess the whole future stream of expected benefits and costs. Environmental decisions also require estimates of the benefits and costs of regulating in the future as opposed to acting now. If the regulatory agency decides to act now, its experience with implementation may be informative about the costs and benefits of later policy choices, including future rescission of the regulatory action. In deciding to act now, the environmental decision maker thus needs to assess the future benefits and costs of correcting or rescinding policy mistakes. The idea that policy choices are dynamic is harder new. Most public policy decisions-in fact? most individual decisions are dynamic ones. When a public utility commission disapproves a requested rate increase, it contemplates the benefits and costs of approving the increase later on. When stockholders decide not to sell their holdings, they consider the benefits and costs of selling later. The same is true for individuals who are seeking another job or deciding to go on a diet. Jeffrey E. Harris is associate professor, Department of Economics, Massachusetts Institute of Technology, and physician, Primary Care Program, Medical Services, Massachusetts General Hospital. 107

OCR for page 107
108 ENVIRONMENTAL POLICY MAKING Environmental policy, however, is an extreme case of dynamic decision making because regulatory decisions about environmental hazards are rou- tinely made in the face of huge uncertainties uncertainties in estimates of health risks, mechanisms of disease, the extent of exposure, or the costs of risk control. Under such extreme uncertainty, the appearance of even a modicum of new data can swamp the decision maker's prior beliefs concerning the costs and benefits of regulatory intervention. As a result, regulatory action on suspected hazards can be triggered or stifled by the is- suance of preliminary toxicological findings, by false alarms concerning the measurement of environmental contaminants, or by leaks of draft reports of blue-ribbon panels. In the conventional research models, repeated measurements tend to improve the precision of estimates of benefits and costs. With the extreme uncertainties encountered in environmental decisions, however, new research findings can pose unexpected contradictions, thus enhancing rather than reducing uncertainty. My task in this paper is to explore, at least in a preliminary way, these dynamic complications of environmental policy making. My method of analysis is essentially anecdotal; that is, I offer some generalizations and then cite selected case studies for support. The hypotheses put for- ward in this paper need independent and more systematic testing using a representative sample of decisions faced by regulatory agencies. In the next section, I establish the central, paradigmatic problem in the dynamics of environmental decision making that is, the problem of timing. Do we act now, or do we wait for more information? The frequently voiced preference for waiting, I would suggest, is based upon a strong but unstated assumption: environmental policies are irreversible, and interventions by regulatory agencies impose large, sunken costs on private firms and consumers that cannot later be recovered. I then inquire further into the realism of the irreversibility assumption. I find that in many cases, a contemplated environmental policy can grow more irreversible with continued delays. There are two mechanisms for this phenomenon of growing irreversibility. First, an environmental problem in its early phases may be amenable to partially reversible interventions (e.g., restrictions on use or access, product labeling, or pollution fees). If the problem gets worse later on, however, then truly draconian, irreversible actions may be required. Second, regulation is a game between govern- mental agencies and the private sector. The longer the regulatory agency delays action, the more time private agents have to make large, sunken investments in the prevailing technology. If the agency delays too long, the stakes become too high. In a subsequent section, I probe further into the issue of "research." Although a strategy of delay is often coupled with a decision to invest

OCR for page 107
JEFFREY E. HARRIS 109 in new data collection, I suggest that research is just as compatible with regulatory intervention. In fact, some regulatory actions are themselves a form of research because they provide essential information about the benefits and costs of future regulatory decisions. In principle, regulatory action can often be a better investment in knowledge than pure research without intervention. I thus propose that policy makers consider two types of questions when contemplating the benefits and costs of a proposed regulatory action: How irreversible is intervention? How informative is the intervention? In general, my analysis points toward a style of regulation in which agencies take small, incremental regulatory steps at the early stages of a problem. These small steps would be designed to impose minimal sunken investments in compliance and still provide essential information on the uncertain benefits and costs of intervention. IRREVERSIBILITY ANI) THE BIAS TOWARD WAITING All too often, one hears the following refrain from scientists and policy makers: "We do not yet have sufficient information to take regulatory action. We would prefer to wait for better data to come in. We need more research." This bias in favor of waiting and against action has been articulated in many forms. The following examples are illustrative. It may be that a proportion of lung cancers in man are induced by tobacco smoke; at the moment we do not know, but let us be sure of our evidence before we scare our public. (Passey, 1953) Linus, I conclude that in my personal view, given the current information, the banning of saccharin at this point in time is counterproductive, and I believe the ban should not be instituted until or unless some "safer" nonnutrient sugar substitute is available. (Isselbacher, 1977) DES tdiethylstibestrol] could have been taken off the market immedi- ately, without a hearing, if the FDA [Food and Drug Administration] had declared it to be an imminent hazard to health. That is the only statutory basis for immediate withdrawal of a drug from the market without first offering a hearing. The agency went to the National Cancer Institute [NCI] on this issue, and the NCI said that, in its judgment, DES was not an imminent hazard. The government's own scientists concluded that the risk was not of that magnitude. Therefore, there was no legal basis for taking that action. (Huts, 1977) EPA [Environmental Protection Agency] did not immediately suspend these uses [of ethylene dibromide as a grain and fruit fumigant] despite the carcinogenic potential because EPA management did not believe enough was known at the time about the risks from residues on food,

OCR for page 107
110 ENVIRONMENTAL POLICY MAKING the risks from substitute fumigants, or the risks from leaving crops and foodstuffs unprotected.... It decided to await the results of studies then in progress. (Russell and Gruber, 1987) Each of these statements is a variant on the same basic theme: im- mediate action may be too costly in comparison to waiting. In Passey's view, the costs arose from scaring the public. For Isselbacher, the cost would be the absence of an alternative to saccharin. In the case described by Hutt, it was too costly to bypass standard regulatory procedure and ban diethylstilbestrol without a hearing. Russell and Gruber's discussion of ethylene dibromide suggested several types of costs, including the risks of substitutes for ethylene dibromide (EDB). All of the examples contain an implicit benefit-cost calculation. The benefits of a determination that smoking causes lung cancer, Passey argued, did not outweigh the costs of "scaring" the public. The cancer risks of saccharin, Isselbacher contended, were outweighed by its benefits as a nonnutritive sweetener. There is more to each of these examples, however, than a one-time benefit-cost analysis. In each case, the decision to act or wait recurred. In analyzing the benefits and costs of action and inaction, each writer needed to consider how such benefits and costs might change over time. The benefits and costs of action were really the benefits and costs of acting immediately as opposed to acting later. Thus, Hutt's description does not imply that DES carried no dan- ger but rather that, in NCI's opinion, the danger was insufficient to act immediately. Isselbacher likewise did not deny saccharin's cancer-causing potential. Instead, he urged action later, once a substitute was available. EPA did not deny the carcinogenicity of EDB. Instead, the agency believed there were insufficient data for immediate suspension of use of the fumigant chemical. This dynamic view of the decision-making process begs some hard questions: What prevented FDA from banning DES immediately in 1971? If subsequent evidence proved contradictory, the ban could have been modified or lifted. What prevented EPA from immediately suspending the use of EDB as a fumigant? Again, if subsequent data had shown extremely low residues in foodstuffs, the ban could have been modified. What prevented the medical community (and manufacturers of cigarettes) from warning the public immediately in 1953 (and even earlier) of the serious, legitimate evidence that cigarette smoking may cause lung cancer? If further research had shown otherwise, a superseding statement of opinion could have been issued. Implicit in these examples is the assumption that an action taken now cannot be rescinded-or, more precisely, that undoing an action is quite

OCR for page 107
JEFFREY E. HARRIS 111 costly. Thus, implicit in Passey's argument is the contention that it would be quitely costly for the public to recover from a false alarm about smoking and cancer. Implicit in Hutt's description is that the act of bypassing the normal hearing process on DES would have been a costly administrative and political error. In these instances, an unstated assumption of irreversibility creates a bias toward waiting. The concept of irreversibility of decisions has not been considered in the literature on environmental policy making. Yet economists have made a number of attempts to spell out its consequences, especially in recent theoretical work in financial economics (Henry, 1974; Cukierman, 1980; Roberts and Weitzman, 1981; Baldwin, 1982; Bernanke, 1983; McDonald and Siegel, 1986; Maid and Pindyck, 1987~. In the economic models, a decision maker is assumed to be contin- uously faced with three types of choices: (1) invest in, (2) proceed with, or (3) abandon a hypothetical project. Investing, on the one hand, is a noncommittal action. It may accelerate the arrival of new information about a project's benefits and costs, but the project's ultimate fate remains undecided. On the other hand, the decisions to proceed with or to drop the project are assumed to be irreversible. The assumption of irreversibility has a number of simple consequences in the economic models. In particular, conventional, static benefit-cost analysis is rendered misleading (Maid and Pindyck, 1987~. Even if the expected benefits of a project exceed its expected costs at a particular point in time, the decision to proceed may be unwarranted. Instead, the decision criteria should be modified to take into account the benefits and costs of waiting for more information. The modified decision rule is to take action only when expected benefits exceed costs by a fixed, predetermined amount. (Strictly speaking, this rule is applicable only when the stochastic process that generates new information is stationary; see, for example, Roberts and Weitzman [1981~.) Put differently, the expected net benefit of the project has to exceed an "option value" of waiting for more information. These stylized, economic models of the wait-or-act decision have gen- eral application. The financial decision to proceed with or abandon a project is analogous to the public policy decision to approve or disapprove, let us say, a new drug application or cleanup technology. The financial decision to invest parallels the regulatory decision to send the drug or technology back for more study. The critical issue in applying the economic models, however, is the validity of their assumption of irreversibility. It is counterproductive to jump to label an environmental regulation as irreversible until the sunken costs that must be expended to comply with the regulation are actually measured.

OCR for page 107
112 ENVIRONMENTAL POLICY MAKING In conducting such an empirical inquiry, what is needed is a typology of sunken costs. As a preliminary scheme, I shall suggest three classes: (1) producer compliance costs, (2) consumer compliance costs, and (3) credibility costs. The first two categories reflect responses by producers and consumers, respectively, to environmental policy decisions. Thus, banning saccharin might result in a permanent and costly shutdown in saccharin- producing facilities. Prohibiting the use of DES as a livestock fattening agent might result in permanent and costly changes in the consumer diet. Credibility costs, the third category, arise because policy decisions are interdependent. Consumers' and producers' responses to environmental policies depend on the credibility of the policy-making entity. If the FDA banned saccharin or DES immediately and if the action turned out to be mistaken, then the agency's ability to enforce subsequent regulatory actions might be destroyed. Still, we need to ask for hard evidence to ensure that capital in the saccharin industry was, indeed, nontransferable. We need to inquire whether consumers could go back to leaner meats if and when DES were reintroduced. We also need to ask whether the credibility costs of policy mistakes in reality all argue in favor of waiting. WAITING AND SUNKEN COSTS The argument in favor of regulatory delay, we have seen, hinges critically on the proposition that government intervention may impose irreversible, sunken costs on private agents. In this section, I suggest that the irreversibility argument can be turned upside down: waiting can have equally irreversible consequences. When a potential environmental hazard is first recognized, its control may be amenable to~partially reversible interventions (e.g., restrictions on access or use, product warning or labeling, pollution fees). If the hazard later becomes quite large, however, then such small-scale interventions may be ineffective, and only large-scale, irreversible interventions may be worth considering. Thus, the regulator who waits for more data runs the risk that only the most extreme, irreversible measures will be available in the end. Acid rain and toxic waste disposal may be good examples of the problem of increasingly narrow regulatory choices. It is no accident of nature that the costs of effective intervention grow larger when regulatory agencies delay action. Private economic agents, especially business firms, have an incentive to make intervention costly. The longer the regulatory authority waits, the more "breathing time" firms may have to commit themselves to the suspect technology.

OCR for page 107
JEFFREY E. HARRIS 113 Diesel Emissions Since the l950s, the condensates from diesel fuel-burning engines have been known to cause cancer in laboratory animals. These particulate emis- sions are further known to contain carcinogenic polyaromatic hydrocarbons. Yet, there has been little sound epidemiological evidence available on the cancer risks of workers exposed to such emissions. In the late 1970s, in the face of increasing pressures for fuel economy, American automobile manufacturers announced plans to convert 25 percent of their light-duty passenger car fleet from gasoline-to diesel fuel-burning engines. The result of such a conversion would have been an increase in population exposures to particulate emissions by an estimated factor of 1,000. The auto makers' proposal stimulated new research into the combus- tion process and the physical chemistry of the particulate matter contained in diesel and other emissions. By 1979, EPA scientists determined that the organic solvent extracts of diesel Articulates were highly mutagenic in the Ames mutagenicity assay. Directly mutagenic nitroaromatic compounds were identified as the likely culprits. EPA lauched a major research program that included laboratory testing of fossil fuel combustion products. The carcinogenicity and mutagenicity of diesel and other emissions were confirmed in multiple laboratory models. Mathematical extrapolations suggested a small individual risk of cancer, but the estimated number of exposed persons was quite large. There was renewed interest in epidemiological studies of exposed workers but very little hard evidence available on the effects of emissions on humans. A study of London transport workers was negative, but was of sufficiently low power that some lung cancer risk from diesel emissions could not be excluded (Harris, 1983~. A scientific panel of the National Research Council could do no more than reiterate the substantial existing uncertainty about the health risks of the proposed diesel technology (National Research Council, 1981~. Moreover, although the biological data base gradually became more refined, the uncertainty about population exposures grew. Changes in the relative prices of diesel and gasoline fuels, as well as unanticipated changes in consumer preferences, made the large-scale introduction of diesel passenger cars less likely. What is more, there were continued uncertainties about the feasibility of effective, low-cost particulate control technologies for disel engines. In the face of all of these uncertainties, EPA proposed immediate particulate emission standards for diesel cars (at a level of 0.6 gram per mile). This action hardly settled the issue, for it remained unclear whether the proposed standards should remain in effect or whether they should be tightened in the future. At the time, a stricter standard (0.2 gram per

OCR for page 107
114 ENVIRONMENTAL POLICY MAKING mile) was contemplated. Even if particulate standards were to be tightened, however, the agency still needed to know when to impose them. By the early 1980s, EPA could reasonably conclude that diesel emis- sions had at least the potential to cause cancer in humans. With virtually no solid epidemiological evidence, however, the agency could not draw definite conclusions about the extent of human cancer risk. From a purely scientific standpoint, the prudent decision was to wait for the results of newly commissioned epidemiological studies. Concrete results from such studies were expected within five years. EPA's decision was not as simple as it might appear, however. The planned conversion to a diesel-driven auto fleet would require a major investment in a new engine technology. Auto makers could not simply modify the existing production technology for gasoline-burning engines. If diesels were to constitute as much as 18 percent of new car sales by 1990, investments on the order of $3-$4 billion would be required. Moreover, it was unclear whether auto makers might later be able to convert the diesel technology to the production of gasoline-burning engines. As the National Research Council reported, 'based on the current state of knowledge, an irrevocable decision by the EPA . . . could run a danger of costly mistakes" (National Research Council, 1982~. Anyway what did the agency really expect from the additional planned research? EPA could reasonably conjecture that by 1985, retrospective studies of workers exposed to diesel emissions might show an elevated risk of lung cancer. Such studies might bolster the case for regulation of diesel particulates. Still, the results of high-dose exposures in the workplace could not be simply extrapolated to low-dose ambient exposures from tail pipe emissions. Moreover, detailed laboratory studies of the composition and biological action of diesel particulate emissions still might not settle a key, lingering question: Did the apparently unique nitroaromatic constituents in the particulate extracts make diesel fumes a uniquely dangerous species of emissions? What made EPAs regulatory dilemma so acute was not the laboratory discovery that diesel emissions were mutagenic, and not the paucity of direct, human evidence, but the announced intention of manufacturers to sink billions into a new diesel technology. In fact, the agency was engaged in a prototypical regulatory game with the car makers. The longer EPA waited for new information, the further down the diesel road the car makers would be. The investment in diesel technology would not be instantaneous but gradual over a period of a decade or more. By the time EPA had sufficient information to satisfy the blue-ribbon scientific panels, the industry might have invested so much in diesel technology as to make tight emission controls too costly.

OCR for page 107
JEFFREY E. HARRIS 115 In this regulatory game, both EPA and the car makers knew the dilemma the agency might soon face. Hence, car makers had a strong incentive to accelerate their investments in diesel technology; that is, to build up their sunken costs as rapidly as possible. While EPA and some auto companies were conducting their own biological research, information on the likely pace of such research was common knowledge. On the other hand, the car makers possessed far more information on the irreversibility of investments in diesel production technology. In fact, EPA: s lack of expertise in this area was perhaps its central difficulty in reaching a regulatory - aeclslon. In the end, EPA stuck with its proposed emission controls, if only to avoid more drastic interventions later. As it turned out, however, the anticipated major demand for diesel cars never materialized, and the agency bought more time to wait for new data. Cyanazine 1b obtain registration for a pesticide under the Federal Insecticide, Fungicide, and Rodenticide Act or FIFING (7 U.S.C. 136 et seq.), an applicant for registration must demonstrate, among other things, that the pesticide performs its intended function without causing "any unreasonable risk to man or the environment, taking into account the economic, social, and environmental costs and benefits of the use of any pesticide" (Section 2[bb]~. EPA, the enforcing agency for the act, interprets this standard to require "a finding that the benefits of the use of the pesticide exceed the risks of use, when the pesticide is used in compliance with the terms and conditions of registration or in accordance with widespread and commonly recognized practice" (U.S. Environmental Protection Agency, 1988:795~. If at any time EPA should determine that this benefit-cost standard has been violated, then the administrator may modify the conditions of registration or cancel the registration entirely. In April 1985, EPA initiated a "special review" of all pesticide products containing the active ingredient cyanazine (U.S. Environmental Protection Agency, 1985~. The review (formerly called the "Rebuttable Presumption Against Registration" or RPAR process) was instigated following the re- cent finding that cyanazine produced teratogenic and fetotoxic effects in laboratory animals. EPA proposed that a warning be added to the pesticide label concerning c~ranazine's potential to cause birth defects in laboratory animals. Moreover, because the main route of occupational exposure was through skin contact, the product label was to specify that cyanazine's use was restricted to certified applicators or to persons under their supervision. EPA was also concerned about groundwater contamination from agri- cultural uses of cyanazine. Preliminary monitoring studies had identified

OCR for page 107
i 116 ENVIRONMENTAL POLICY MAKING residues of cyanazine in a small percentage of sample wells from five states. Although most positive samples showed cyanazine concentrations of 0.2 part per billion (ppb), a small percentage showed levels close to 1 ppb. The agency thus noted: Cyanazine has the potential to move (leach) through the soil and contam- inate ground water which may be used as drinking water. Cyanazine has been found in surface and ground water as a result of agricultural use. The Agency does not have the data necessary to assess the health risks associated with consuming drinking water which has been contaminated with cyanazine. (U.S. Environmental Protection Agency, 1985:14151) Accordingly, the agency imposed labeling requirements that advised users not to apply cyanazine to highly permeable soils or to areas in which the water table was close to the surface. It also required registrants to conduct groundwater and surface water monitoring studies. In a January 1987 review, the agency proposed a number of additional requirements for cyanazine registration, including the use of protective gloves, closed loading systems, and chemical-resistant aprons. The pesticide label was to include statements regarding the cleaning of protective gloves and separate laundering of protective clothing. In addition, the label was to state that cyanazine was classified for restrictive use because it Alas caused birth defects in laboratory animals and has been found in ground water" (U.S. Environmental Protection Agency, 1987a:589~. By early 1988, however, new data suggested that cyanazine was not as serious a threat to groundwater as had been supposed. In particular, further sampling from 200 wells in hydrogeologically vulnerable areas revealed no detectable residues. The agency thus lifted its prior restriction on the spraying of cyanazine in areas in which the water table was high or the soil was highly permeable. C7 As a result of newly generated monitoring data and the previously avail- able data, the Agency no longer believes that cyanazine has significant ground water contamination potential. Therefore, EPA no longer be- lieves that ground water contamination should be a reason for classifying cyanazine for Restricted Use. Therefore, all cyanazine labels will include a statement that cyanazine products have been classified for Restricted Use only because cyanazine has caused birth defects in laboratory ani- mals. However, because some instances of contamination were reported in the earlier studies, the Agency believes the ground water advisory statement should remain on the label. (U.S. Environmental Protection Agency, 1988:795) In the case of cyanazine, EPA altered its position several times as new evidence accumulated on the pesticide's potential toxicity and the routes of environmental exposure. The agency in fact reversed itself on the issue

OCR for page 107
JEFFREY E. HARRIS 117 of groundwater contamination. The only clear effect of these multiple regulatory changes, however, was to alter the contents of the pesticide's warning label. Ninety-six percent of the cyanazine produced in the United States was used as a herbicide on corn. About 3 percent was used on cotton, and less than 1 percent was used on sorghum and wheat. About 1~16 percent of the total U.S. corn acreage was treated with cyanazine in 1982. Several close substitutes for cyanazine were readily available, and there was little evidence that switching to these substitutes would be costly. EPA was thus in a position to make a series of incremental changes in its regulation of cyanazine use without imposing large sunken costs on the private sector. Users of cyanazine were required to make investments in closed loading systems and protective equipment, but none of these investments was specific to a single chemical. Producers of cyanazine were required to reissue warning labels. In the absence of an outright ban on the use of cyanazine, however, the question of irreversible, cyanazine-specific investments did not arise. Ethylene Dibromide Table 1 traces scientific developments concerning ethylene dibromide (EDB) from 1910 to 1976. EDB was first used by producers of lead antiknock compounds for gasoline in the 1920s. By the late 1940s and early l950s, the compound was widely employed as a fumigant of imported fruits and vegetables, grain, storage silos, and grain-milling machinery. Data on EDB's acute and subacute toxicity go back to the early 20th century. The evidence on EDB arose from reports of accidental human exposure and from studies of ingestion, inhalation, and decimal exposure in various laboratory animals. By the mid-1960s, additional reports appeared on EDB's reproductive toxicity in farm animals. Still, residues of EDB remained essentially undetectable in the food supply. In the early 1970s, two developments-the linking of EDB to muta- genicity and carcinogenicity and the improvement of the technology for detecting EDB brought increased attention to and concern about the compound. In 1971, EDB was found to be a direct-acting mutagen in the Ames mutation assay. By 1974, the chemical's genotoxicity had been confirmed in other experimental systems. At this time, scientists were in- creasingly interested in the possible role of genotoxic events in the genesis of cancer. The finding that EDB was a mutagen stimulated whole-animal carcinogenicity studies by the National Cancer Institute (NCI). NCI's preliminary results showed that EDB was carcinogenic when it was directly instilled into the stomachs of rodents. ~ be sure, there was concern that the NCI results were somehow artifactual because the

OCR for page 107
JEFFREY E. HARRIS TABLE 2 Continued 123 Year Scientific Developments Regulatory Developments reported levels in flour range from nondetectable to 4.2 ppm. 1982 Publication (in March) of the final results of the NCI inhalation study in rats and mice: EDB is found to be carcinogenic. EPA scientists are notified (in June) that three wells in Seminole County, Georgia, are contaminated with EDB levels as high as 100 ppb. SRI International publishes a NIOSH- commissioned risk assessment based on NCI and NIOSH inhalation studies in rats and mice (June); chronic exposure to 130 ppb is predicted to yield 4-26 percent lifetime human cancer risk. CDFA (June 2) revises estimates of EDB residues in fumigated citrus fruits up to 210-880 ppb. Wade and Sakura report two acute lethal reactions among workers exposed to EDB. 1983 The National Toxicology Program reports that inhalation of EDB (10~0 ppm) in Fisher 344 rats produced testicular degeneration. An EPA- commissioned study of groundwater contamination by CDFA issues its preliminary report (Spring), finding EDB at concentrations between 0.1 and 31 ppb in the soil at depths greater than 20 feet, moving down to groundwater. A follow- up report (June) reveals groundwater levels between 0.02-5 ppb in 16 counties in 4 OSHA interprets the Supreme Court ruling as permitting mathematical risk assessment in support of agency regulations (Federal Register, April 9~. Cal/OSHA's emergency standard of 15 ppb is rejected by California Office of Administrative Law; California adopts as a permanent regulation a standard of 130 ppb. EPA issues Position Document no. 4 (September 27), with revisions in its mathematical risk assessment methodology. EPA issues an emergency suspension of its soil fumigation with EDB; it gives notice (September 28) of intent to cancel registration of EDB as a grain and fruit fumigant under the "unreasonable hazard" standard of FIFRA. EDB use in fumigation is to be eliminated by 1986. The state of Florida issues emergency regulations restricting EDB in

OCR for page 107
124 TABLE 2 Continued ENVIRONMENTAL POLICY MAKING . . . Year Scientific Developments Regulatory Developments states. A new EPA risk analysis is issued as part of Position Document no. 4. The original one-hit model of Position Document no. 2/3 is modified to include "Weibull timing.H The estimated average EDB content of grains is revised upward markedly to 31 ppb. CAG's new estimate of lifetime cancer risk from a dietary burden of EDB is 3.3 per 1,000, based on lifetime consumption of current levels of EDB in grain products. 1984 Grocery Manufacturers of America (GMA), modifying the Rains and Holder (1981) detection methodology, find that 79 percent of ready-to-eat, grain-derived products contain EDB levels below 1 ppb; GMA also reports on the disappearance of EDB through cooking raw grain products. Environ Corporation, under GMA sponsorship, issues (January 20) risk assessment of exposures to EDB residues in consumable grain products, based on NCI oral gavage assay and assumptions of no further grain fumigation and of the depletion of EDB in grain stores by 1986; the upper limit of lifetime cancer is estimated to be 1 in 4 million. Temple, Barker & Sloane, Inc., and Economic Perspectives, Inc. issue an economic analysis of the impacts of immediate removal of EDB from the food supply, if 50-60 percent of stored uncooked grain products to 1 ppb (level of detection). EPA announces (February 3) immediate suspension of further use of EDB in the production of grain products and recommends guidelines to states for acceptable levels of EDB in foods, including 900 ppb in raw grain products, 150 ppb in processed products requiring further cooking, and 30 ppb in ready-to-eat foods. The Massachusetts Department of Public Health recommends (February 6) emergency regulation at 10 ppb for all food products, with transition in 30 days to 1 ppb. ("The Department's position is that the only safe level of exposure to a carcinogen is one that is zero or near zero. The Department therefore believes that it is appropriate to move rapidly to levels of EDB in food of less than 1 ppb.")

OCR for page 107
JEFFREY E. HARRIS TABLE 2 Continued 125 Year Scientific Developments Regulatory Developments grains and 67 percent of grain products were immediately restricted from any use, they conclude, grain prices would nearly double, with consumer expenditure increases of $35 billion and grocery manufacturer losses of $2.8 billion in inventories. EPAs suspensions of the use of EDB in 1983 and 1984 were not the first regulatory actions taken with respect to the pesticide. Nor did 1983 see the first instance of damning evidence on EDB. The question arises: What exactly happened between 1977 and 1983? By 1977 the International Agency for Research on Cancer had already classified EDB as an animal carcinogen and mutagen. A review by the National Institute on Occupational Safety and Health (NIOSH) noted that EDB was able to interact chemically with deoxyribonucleic acid (DNA), the basic genetic material. Still, EDB had thus far been found to be car- cinogenic in only one incomplete animal experiment. Moreover, attempts to identify elevated cancer rates among EDB-exposed workers were unsuc- cesful. If EDB in fact posed a cancer threat at low doses, the magnitude of the cancer risk remained uncertain. In the face of this uncertainty, the Occupational Safety and Health Administration (OSHA) proposed a tightening of its EDB exposure stan- dard for workers. EPA, in parallel, began a special RPAR review under FIFED The linchpin of EPAls regulatory analysis was a risk assessment, performed by its Carcinogen Assessment Group (CAG). CAG's initial risk assessment proved to be problematic. The initial dosages of EDB in the NCI oral gavage study-on which CAG relied _ . ~ . . ... ~ . . proved to be too toxic, so the dosage schedule had to be reduced in the middle of the experiment. This changing dosage schedule complicated CAG's attempts to extrapolate from high-dose to low-dose effects and risks. The CAG analysis also predicted a substantial cancer risk from long-term EDB exposures at the levels seen among chemical workers; limited surveys of EDB-exposed workers, however, showed no evidence of a significant cancer increase.

OCR for page 107
126 ENVIRONMENTAL POLICY MAKING Yet by 1979, additional laboratory studies had confirmed EDB's car- cinogenicity. The chemical caused cancers by skin painting in mice, and a NIOSH-sponsored study showed cancers by inhalation in rats. By 1980, EDB was found to be carcinogenic in a separate NCI-sponsored inhalation study of rats and mice. In that year, the American Conference of Gov- ernmental Industrial Hygienists also classified EDB as a suspect human carcinogen. EPA:s special RPAR review continued in 1980. An internal study estimated the probable residue level for EDB in wheat bread made from fumigated grain to be less than 0.1 ppb, with a realistic worst-case residue of 31 ppb. Based on such exposure estimates and extrapolating from the original NCI oral Savage experiment in rodents, CAG projected a 0.03 percent increased lifetime cancer risk owing to the dietary burden of EDB. The agency proposed cancellation of EDB's use as a fumigant of stored grains, milling machinery, and fruits and vegetables by mid-1983. It also ordered studies of potential groundwater contamination. By 1981, new measurements of EDB residues in fruit and grain prod- ucts suggested that previous estimates had been misleading. One study found EDB residues of 36 ppb in biscuits. Another found 57 ppb in the edible portions of fumigated fruits. Concurrently, OSHA proposed fur- ther tightening of the occupational standard for EDB exposure; California imposed a temporary emergency occupational standard. By 1982, EDB levels as high as 100 ppb had been found in three wells in Georgia. The California Department of Food and Agriculture (CFDA) estimated that fumigated citrus fruits contained EDB residues of up to 21() 880 ppb. By spring 1983, CFDA had found EDB concentrations of 0.1-31 ppb at depths greater than 20 feet. By June 1983, EDB had been detected at levels of 0.02-5 ppb in 16 counties. EPA moved in September 1983 to suspend soil fumigation immediately. Based on the new exposure data, as well as a reanalysis of the NCI oral gavage experiment, CAG revised the estimated lifetime risk from dietary EDB to 0.3 percent. In February 1984, the agency suspended further use of EDB in the production of grain products, although it did not order an immediate ban on the sale of all EDB-containing products. Instead, it issued recommended guidelines to the states for acceptable levels of EDB in currently marketed foods. Why did EPA wait six years (from its initial review in 1977 until its emergency suspension in 1983) to take action on EDB? The evidence of EDB's toxicity was long-standing: its mutagenicity was established in 1971, and its carcinogenicity was reported by 1977. Although the initial NCI study required confirmation, independent findings of carcinogenicity were available by 1979. Initially, EDB was thought to be virtually undetectable in the food supply; yet contrary evidence was available by 1981. Groundwater

OCR for page 107
JEFFREY E. HARRIS 127 contamination was an issue as early as 1980, when EPA commissioned a study by CDF~ Residues were found in wells as early as June 1982. Perhaps it is unfair to juxtapose EPAs regulation of cyanazine during 1985-1988 with the agency's drawn-out response to EDB during 1977-1984. By the mid-1980s, the agency had improved its handling of procedural and notification burdens built into FIFRA, which was enacted in 1972. Still, the cyanazine case shows the agency moving quickly in incremental, reversible steps to establish warning labels and restrictions on use. In the case of EDB, the agency essentially found itself having to ban the pesticide late in the game, years after other federal and state agencies had moved on the problem. Had EPA accelerated the information-gathering process, especially in the measurement of food residues and groundwater contamination, less extreme measures might have been necessary. By 1984, the sunken investment in EDB had become enormous: $29 billion in grain stocks and $4.3 billion in manufacturer and retail inventories of grain products and baked goods. It was likely that between 50 and 60 percent of stored grains and grain products contained detectable levels of EDB. Commingling of grains during storage, transport, and manufacture raised the possibility that nearly all such products had detectable levels of the chemical Temple, Barker and Sloan, Inc., and Economic Perspectives, Inc., 1984~. Immediate removal of EDB-containing foods would have been quite costly. In the end, EPA chose an intermediate course: suspension of use of the compound without confiscation of existing stocks of potentially EDB-contaminated food. REGULATION AS RESEARCH Scientists and policy makers may recommend delaying regulatory ac- tion until they see the results of current research. Yet the need to perform more research does not preclude concurrent regulatory intervention. EPA imposed a groundwater advisory on pyanazine's label even as it sought further testing of pesticide residues. The agency imposed a standard on particulate emissions from diesel-powered cars even as it awaited the results of epidemiological studies on diesel workers. Although EPA did not restrict EDB until 1983, earlier action should not have barred further toxicological and exposure studies. In fact, there is no clear dividing line between regulatory intervention and research. The reason is that knowledge can be gained from the experience of regulatory intervention. In some instances, the best way to assess the benefits and costs of regulation is to regulate and see what happens. By contrast, further delay may bring little or nothing in the way of new information.

OCR for page 107
128 ENVIRONMENTAL POLICY MAKING The nation's experience with environmental controls may provide the best source of information and sometimes the only source of informa- tion-on the costs of complying with even stricter controls. At issue here is whether the public or private sectors are best suited to perform the necessary research on new control technologies. When the development of new controls entails highly specialized or proprietary knowledge, it may be impractical for regulatory agencies to fund public research into cleanup technologies. Instead, the most effective way to instigate the necessary research is to impose environmental controls, thus changing the incentives of private firms. Conversely, experience with regulatory controls may be the best or only means of assessing the benefits of environmental regulation. Laboratory experiments can measure small-scale individual effects, whereas environ- mental controls operate on a large scale. Thus, laboratory experiments and meteorologic modeling can offer only imprecise gauges of the aggregate effect on acid rain of curbing sulfur oxide emissions. Measurement of in- dividual tail pipe emissions, in combination with dispersion modeling, may be inadequate to predict the aggregate effect of installing auto pollution control devices. The main point is that small-scale "micro" models and experiments may be inadequate to understand or predict the "macro" consequences of large-scale policy interventions (Harris, 1985~. At best, basic research and data acquisition can only disentangle individual mechanisms; they cannot by themselves show the interaction of multiple mechanisms of environmental damage and multiple routes of toxic exposure. The only way to assess such large-scale effects is by natural experiments; that is, by regulatory intervention. CHLO RO FLU O RO CARB O NS In 1974, Molina and Rowland proposed that long-lived, stable chlo- rofluorocarbons (CFCs) could slowly migrate to the stratosphere, where they would release chlorine following contact with high levels of radiation. The resultant free chlorine could in turn act as a catalyst to break apart ozone molecules. Thus, CFCs might be steadily depleting the stratospheric layer of ozone, the shield that stops ultraviolet-B radiation from penetrating to the earth's surface. The ozone depletion hypothesis was taken seriously by the scientific community, and early work on the topic Includes a 1976 report by the National Academy of Sciences. In 1977, Congress amended the Clean Air Act (42 U.S.C. 7457ib]) and authorized EPA's administrator to issue regulations for controlling substances or activities "which in his judgment may reasonably be anticipated to affect the stratosphere, especially ozone

OCR for page 107
JEFFREY E. HARRIS 129 in the stratosphere, if such effect in the stratosphere may reasonably be anticipated to endanger public health or welfare. Such regulations shall take into account the feasibility and the costs of achieving such control." The statutory language permitted EPA to act in the face of scientific uncertainty (U.S. Environmental Protection Agency, 1987b). In 1978, EPA and the U.S. Food and Drug Administration moved to ban the use of CFCs as aerosol propellants in all but "essential applications." During the early 1970s, aerosol propellants constituted about 50 percent of total CFC use in the United States. Thereafter, CFC use in propellants declined markedly. Largely in response to a series of National Research Council studies in the late 1970s, in 1980, EPA issued an Advance Notice of Proposed Rulemaking under the Clean Air Act. The notice proposed that the production of certain CFCs be frozen and suggested the possible use of marketable permits to allocate CF~ production among various industries. In the early 1980s, however, new data and models suggested that many other factors contributed to ozone depletion in the stratosphere. Carbon dioxide and methane, two atmospheric gases that have been increasing in concentration in recent years, appeared to buffer the ozone-depleting effects of CFCs. Moreover, although CFCs continued to be used as foam-blowing agents, refrigerants, and solvents, the decline in CFC aerosol propellant use resulted in a leveling off of worldwide CFC production. Beginning in about 1983, the demand for nonaerosol uses of CFCs accelerated. Total production expanded to such a point that it now exceeds 1974 levels. Levels of CFC-ll (primarily used as a foam-blowing agent) and CFC-12 (primarily used as a refrigerant) are now rising at 5 percent annually, while CFC-113 (mainly used as a solvent for electronics and metal cleaning) has risen an estimated 10 percent annually. Moreover, there have been increases in demand for certain brominated compounds that are also thought to deplete stratospheric ozone (e.g., Halon-1211, which is used in specialized firefighting applications). These changes have been paralleled by continued increases in carbon dioxide and methane. In 1985, the World Meteorological Organization (WMO) conducted a review of all ground- and satellite-based atmospheric ozone measurements to date. WMO concluded that ozone levels in the upper atmosphere had in fact decreased by 0.2-0.3 percent annually during the 1970s. Moreover, these decreases were offset by increases in ozone in the lower atmosphere, so that the total "column" ozone had remained unchanged. In May 1985, however, Farman, Gardiner, and Shanklin reported that ozone levels in Antarctica, which were measured during the months of September to November, had declined by 40 percent since 1957, with most of the decline occurring since the mid-1970s. The discovery of this Antarctic ozone hole was completely unexpected; a 40 percent decline

OCR for page 107
130 ENVIRONMENTAL POLICY MAKING was not predicted by current atmospheric models of ozone depletion. By 1987, additional measurements of a key compound-chlorine monoxide- suggested that anomalous chlorine chemistry may have played a role in the development of the Antarctic hole. Such findings left open the possibility that seasonal declines in ozone above Antarctica were idiosyncratic and not reflective of global chemistry. Still, researchers have yet to determine the exact mechanisms responsible for the high levels of chlorine monoxide in the Antarctic hole and whether such unknown mechanisms are, indeed, unique to Antarctica. Moreover, recently published evidence (Kerr, 1987) has challenged the conclusion that total column ozone is stable. Ground-based and satellite measurements now suggest a 3-5 percent annual decline during the 1980s. As in the case of the Antarctic ozone hole, these measurements fall outside of the uncertainty bounds computed from current atmospheric models, which predict that column ozone should not have decined by even 1 percent. A review of the newer data has now been instituted by the National Aeronautics and Space Administration and the National Oceanographic and Atmospheric Administration. Why did the models fail to predict the 1987 results? One possibility is that the results are artifactual (e.g., misinterpreted satellite measurements). Another is that the models have failed to consider adequately the solar cycle or volcanic activity. Still, the main problem is that current models, which now include approximately 50 chemical species and simulate over 140 different reactions, may not be able to replicate atmospheric chemistry accurately. Have they failed to predict the limits by which the lower atmosphere can compensate for stratospheric ozone losses? Have they failed to predict the buffering effects of carbon dioxide and methane? Are estimates of the half-lives of certain CFCs (75 years for CFC-ll and 110 years for CFC-12) inaccurate? On September 16, 1987, the United States and 23 other nations signed the Montreal Protocol on Substances That Deplete the Ozone Layer. The agreement set forth a timetable for reducing specified ozone-depleting chemicals, including a freeze on production at 1986 levels, followed by reductions during the 1990s. EPA, in anticipation of U.S. ratification of the Montreal Protocol, has already mandated the reporting of 1986 production, imports, and exports by American firms (U.S. Environmental Protection Agency, 1987b). Formal benefit-cost analysis of CFC regulation is a formidable task. Models are needed to estimate the future decline in stratospheric ozone levels; the possible compensating increase in lower atmospheric ozone lev- els; the potential adverse effects of changes in atmospheric ozone, including increased incidence of skin cancers and cataracts, damage to aquatic or- ganisms, and accelerated weathering of outdoor plastics; and the overall

OCR for page 107
JEFFREY E. HARRIS 131 effects of global warming. In addition, the economic dislocation resulting from restrictions of CFCs and haloes must be determined, including losses in refrigeration, foam production, cleaning of electrical equipment, and firefighting applications. Still other information is also required. Can regulations really wait for better data and models on atmospheric chemistry ozone depletion? What will be the future evolution of such scientific information? Will implementation of CFC and halon controls now provide a critical source of data in understanding the ozone problem? Regulation of CFCs and haloes is hardly an all-or-none proposition. Should the Montreal Protocol go into force, and should the United States ratify it, EPA will be required to implement the 1986 production-level freeze and the planned reductions for the 1990s. The agency currently proposes to use a system of marketable licenses. Production or use charges are also under consideration. It is unlikely that EPA can project the consequences of these proposed regulatory schemes. Accordingly, in choosing which scheme to adopt, the agency needs to ask what near-term interventions are likely to provide information about future regulatory designs. CONCLUDING COMMENTS In environmental decision making, inconclusive scientific evidence is a commonplace occurrence. Still, regulatory agencies continue to make decisions in the face of such uncertainty. In evaluating regulatory choices, it is hardly enough to assess the static benefits and costs of each regulatory option. Instead, regulatory agencies need to solve the problem of timing, which means assessing the benefits and costs of intervening now versus intervening later. ~ attack the problem of timing, I have suggested that regulatory agencies ask two types of questions: Will we be able to take back the regulatory action? Will intervention be informative about future regulatory choices? Environmental regulation takes many forms: requiring private firms to conduct studies or report data, suspending some uses of a chemical while permitting others? mandating or changing warning labels, issuing emergency suspensions, and scheduling phaseouts. In general, my analysis points toward a style of regulation in which agencies take small, incremental regulatory steps at the early stages of a problem. These small steps would be designed to impose minimal sunken investments in compliance, yet provide essential information on the uncertain benefits and costs of intervention. The supporting evidence for the success of this style of regulation, however, has been largely anecdotal. I have cited a few possibly unrepre- sentative examples. 1b assess the results of past environmental decisions

OCR for page 107
132 ENVIRONMENTAL POLICY MAKING and to formulate guides for future choices will require a much wider array of case studies. Still, I see broad application of the idea that environmental decision makers often wait too long to take action in the face of uncertainty. The reasons for delaying action, I suggest, are at best poorly articulated. As- sertions that proof is not yet available, or that intervention will distract attention from more fundamental causes, or that the public will be need- lessly alarmed, should be subject to more careful scrutiny. The refrain that "we need more research before we can act" likewise needs to be questioned. It is unfair to state the problem as "regulation versus research" when the main issues are, instead, the synergies between regulation and research. REFERENCES Baldwin, Carliss Y. 1982 Optimal sequential investment when capital is not readily reversible. Journal of Finance 37:763-782. Bernanke, Ben 1983 Irreversibility, uncertainty, and cyclical investment. Quarters Joumal of Eco- nomics 98:85-106. Cukierman, Alex 1980 The effects of uncertainty on investment under risk neutrality with endogenous information. Joumal of Political Economy 88:462-475. Harris, Jeffrey E. 1983 Diesel emissions and lung cancer. Risk Analysis 3:83-100. 1985 Macro-experiments versus micro-experiments for health policy. Pp. 145-185 in J. Hausman and D. Wise, eds., Social Ex;?enmentation. Chicago: University of Chicago Press. Henry, Claude 1974 Investment decisions under uncertainty: The irreversibility effect. American Economic Review 64:289-322. Hutt, Peter B. 1977 Question and answer session: FDA-diethylstilbestrol panel. Pp. 1675-1682 in H. H. Hiatt, J. D. Watson, and J. A. Winsten, eds., Origins of Human Cancel: Book C, Human Risk Assessment. Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory. Isselbacher, Kurt J. 1977 Statement of Kurt J. Isselbacher, M.D. Pp. 449053 in Propsed Saccharin Ban Oversight. Hearings before the Subcommittee on Health and the Environment, Committee on Interstate and Foreign Commerce, House of Representatives, U.S. Congress, March 21 and 22. Serial Number 95-8. Washington, D.C.: Government Printing Office. Kerr, R. A. 1987 Has stratospheric ozone started to disappear? Science 237:131-132. Maid, Saman, and Robert S. Pindyck 1987 Time to build, option value, and investment decisions. Joumal of Financial Economics 18:7-27. McDonald, Robert, and Daniel Siegel 1986 The value of waiting to invest. Quarter) Joumal of Economics 101:707-727.

OCR for page 107
JEFFREY E. HARRIS 133 National Research Council 1981 Health Effects of Exposure to Diesel Exhaust. The Report of the Health Effects Panel of the Diesel Impacts Study Committee. Washington, D.C.: National Academy Press. 1982 Diesel Cars. Benefits, Risics, and Public Policy. final Report of the Diesel Impacts Study Committee. Washington, D.C.: National Academy Press. Passey, R. D. 1953 Smoking and lung cancer (letter). British Medical Joumal i(Februa~y 14~:399. Roberts, Kevin, and Martin L. Weitzman 1981 Funding criteria for research, development, and exploration projects. Economet- nca 49:1261-1288. Russell, Milton, and Michael Gruber 1987 Risk assessment in environmental policy-making. Science 236:286 295. Ample, Barker & Sloane, Inc., and Economic Perspectives, Inc. 1984 The Econamics of Immediate EDB Removal. Lexington, Mass.: Temple, Barker ~ Sloane. U.S. Environmental Protection Agency 1985 Special review of certain pesticide products; cyanazine. Federal Register 50(Apr. 10):14151. 1987a Preliminary determination to cancel registrations of cyanazine products unless the terms and conditions of the registration are modified; availability of technical support document and draft notice of intent to cancel. Federal Register 52(Jan. 7):589. 1987b Protection of stratospheric ozone; proposed rule. Federal Register 52(Dec. 14):47489. 1988 C~nazine; intent to cancel registrations; denial of applications for registration; conclusion -of special review. Federal Register 53(Jan. 13):795. U.S. Environmental Protection Agency, Office of Policy Analysis 1985 EDB, A Case Study in the Co~nunication of Health Risks. Washington, D.C.: U.S. Environmental Protection Agency.