2
Overview of Ambient-Ozone Standards Development and Benefits Assessment

INTRODUCTION

Ozone has been the subject of extensive research, standard-setting, and air-pollution control activities for over 3 decades in the United States and in other countries. This chapter provides background on the committee’s consideration of the implications of recent findings of associations of ambient ozone concentrations with mortality, including the setting of national ambient air quality standards, the process of implementing the standards, the scientific basis of standard-setting, the use of health studies to quantify expected health effects of changes in air quality, the concepts underlying economic valuation of such health effects, and experience in the United States and elsewhere in applying these concepts to estimate benefits of ozone reduction.

SETTING NATIONAL AMBIENT AIR QUALITY STANDARDS FOR OZONE

Beginning in 1970, the U.S. Clean Air Act (CAA) directed the Environmental Protection Agency (EPA) to consider the best available science bearing on exposure to and effects of several ambient air pollutants that are emitted by a wide array of sources and to set National Ambient Air Quality Standards (NAAQS) for pollutants to which the public was widely exposed. Under the CAA, NAAQS are to be set and reconsidered every 5 y, and the administrator of EPA is to consider setting primary NAAQS to protect the public health and secondary NAAQS to protect the public welfare (for example, buildings, materials, and ecosystems). In setting the primary NAAQS, the administrator is to consider all available science, historically compiled in a so-



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 23
2 Overview of Ambient-Ozone Standards Development and Benefits Assessment INTRODUCTION Ozone has been the subject of extensive research, standard-setting, and air-pollution control activities for over 3 decades in the United States and in other countries. This chapter provides background on the committee’s consid- eration of the implications of recent findings of associations of ambient ozone concentrations with mortality, including the setting of national ambient air qual- ity standards, the process of implementing the standards, the scientific basis of standard-setting, the use of health studies to quantify expected health effects of changes in air quality, the concepts underlying economic valuation of such health effects, and experience in the United States and elsewhere in applying these concepts to estimate benefits of ozone reduction. SETTING NATIONAL AMBIENT AIR QUALITY STANDARDS FOR OZONE Beginning in 1970, the U.S. Clean Air Act (CAA) directed the Environmental Protection Agency (EPA) to consider the best available science bearing on exposure to and effects of several ambient air pollutants that are emitted by a wide array of sources and to set National Ambient Air Quality Standards (NAAQS) for pollutants to which the public was widely exposed. Under the CAA, NAAQS are to be set and reconsidered every 5 y, and the administrator of EPA is to consider setting primary NAAQS to protect the public health and secondary NAAQS to protect the public welfare (for example, buildings, materials, and ecosystems). In setting the primary NAAQS, the administrator is to consider all available science, historically compiled in a so- 23

OCR for page 23
24 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits called criteria document1 prepared by EPA scientists and consultants, and to set standards at levels that are “requisite to protect the public health with an adequate margin of safety.” The act does not include costs of implementation in the criteria for setting the NAAQS, and EPA—with consistent support from the courts—has interpreted that to mean that it may not consider costs in NAAQS decisions (American Trucking Associations, Inc. v. U.S. EPA, D.C. Cir 97-1440 and 97-144). In essence, that has meant that each time a NAAQS is reviewed, the ad- ministrator must weigh the most recent evidence and continuing uncertainties and make a “public-health policy judgment” about whether the newest evidence provides enough certainty about the likelihood and public-health significance of effects above, at, and below the current standard to warrant a determination that the current standard is adequate to protect the public health with an adequate margin of safety or should be lowered or raised. Once that determination is made for a particular pollutant or class of pol- lutants, EPA is expected to make decisions about four aspects of the standards: • The indicator (the pollutant to be monitored and assessed for attain- ment). • The level of the standard. • The averaging time (for example, 1 h, 8 h, 1 d, or 1 y). • The statistical form of the standard (for example, whether the standard will not be met if it is exceeded more than 1% of the time, on the third-highest day each year, or other similar measure). Since the inception of NAAQS, EPA has determined that photochemical- oxidant air pollution, formed when specific chemicals in the air react with light and heat, is of sufficient public-health concern to merit establishment of a primary NAAQS. In implementing that determination, EPA has since 1979 identified ozone, a prominent member of the class of photochemical oxidants (such as nitrogen dioxide), as an indicator for setting the NAAQS and tracking whether areas of the country are in attainment of the standards. Over the last 38 y, as scientific understanding of ozone health effects has evolved, EPA has reviewed and updated, as needed, the primary NAAQS for ozone and other photochemical oxidants in 1971, 1979, 1993, and 1997 (see Table 2-1). As illustrated in Figure 2-1, under the 1997 NAAQS, ozone nonattainment has occurred largely in heavily populated areas east of the Mississippi River, 1 In December 2006, EPA indicated that after the current ozone NAAQS review proc- ess, it would no longer use the historical terminology of criteria document to summarize the science and staff paper to summarize staff risk assessment and recommendations to the administrator. The criteria document would be replaced by an integrated science as- sessment and the staff paper would be replaced by an advanced notice of proposed rule- making (EPA 2007c).

OCR for page 23
TABLE 2-1 History of Primary NAAQS for Ozone and Other Photochemical Oxidantsa Year Indicator Level Averaging Time Form Scientific Rationale 1971 Total photochemical 0.08 ppm 1h Not to be exceeded Principal study cited for final found increased asthma-attack oxidants more than 1 h/y frequency when hourly average reached 125µg/m3 (0.10 ppm); final standard included margin of safety below most likely threshold suggested by this study (Bachmann 2007) 1979 Ozone 0.12 ppm 1h More than 1 d/y with Move to ozone-specific effects; three clinical studies found maximal hourly reduction in pulmonary function or symptoms as having lowest average above effect level in humans at 0.15-0.3 ppm with evidence of lower 0.12 ppm effect levels in animals (Bachmann 2007) 1993 Ozone Determination that no change was necessary 1997 Ozone 0.08 ppm 8h Annual 4th-highest Many new studies over a decade found effects at daily maximal 8-h concentrations below 1979 standard with increased importance average concentration, of 6-8-h exposures; key studies found lung-function averaged over 3 y decrements, respiratory symptoms, increased sensitivity to irritants, indicators of pulmonary inflammation increasing across range of 0.08, 0.1, and 0.12 pm for 6- to 8-h exposures with subjects engaged in intermittent exercise; numerous epidemiologic studies found increased hospital admissions and emergency-room visits for respiratory causes attributed primarily to ozone; proposal highlighted risk assessment for children (nine-city) and for New York City hospital admissions. (Bachmann 2007) 2008 Ozone 0.075 ppm 8h Annual 4th-highest Strengthened clinical database showing effects at 0.080 ppm; daily maximal limited data on potential effects on some subgroups at 0.060 8-h average ppm (6- to 8-h exposure) (Adams 2002, 2003, 2006); concentration, substantially increased epidemiologic evidence on morbidity averaged over 3 y effects (EPA 2008a). Also see Box 2-1. a This table focuses on the primary rather than secondary NAAQS. 25

OCR for page 23
26 FIGURE 2-1 Counties violating 1997 primary 8-h NAAQS for ozone and other photochemical oxidants. Source: EPA 2007d.

OCR for page 23
27 Ambient-Ozone Standards Development and Benefits Assessment BOX 2-1 The Estimation of the “Policy Relevant Background” One step in the process of setting the NAAQS which was not a focus of this report is to assess the amount of risk reduction expected to result from its implementation, an involved and often ambiguous process. It is unlikely that lowering a standard will lead to no ozone concentrations oc- curring that are greater than the set level without ozone concentrations below the level of the standard also being affected. Likewise, raising the level of the standard would not lead quickly to ozone concentrations in- creasing up to the standard at all locations. In order to conduct its risk as- sessment, EPA assesses how ozone at all concentrations will respond to the change in the standard. However, during the NAAQS process the spe- cific rules and controls that will be implemented in response to a change are not known, so neither is it known how a change in the standard will impact air quality. There are a large number of possible paths towards cleaner air and attaining a standard, and as discussed later, their impacts on ozone concentrations at different times of the day and different periods of the year can differ substantially, even if all of the paths suggest that they will lead to attainment. The question is how can EPA assess the health and welfare implications of a revised NAAQS without better knowing the likely air quality response. EPA deals with this ambiguity by prescribing a “Policy Relevant Background” (PRB), and then rolls back current levels towards this back- ground level in such a way that the standard would now be met. In so do- ing, all levels above the PRB are reduced (see EPA [2006a, 2007a] for a more complete description of the definition, calculation and use of PRB). EPA further goes on to define the PRB as the level that the pollutant con- centration would be in the absence of anthropogenic emissions from the US, Mexico and Canada. In the 2006 Ozone Criteria Document, a global chemical model was used to calculate spatially and temporally varying PRBs (EPA 2006a). Removing emissions from Canada and Mexico has been criticized for, among other reasons, providing an unrealistically low estimate for PRB (Brauer et al. 2007). Defining a PRB is unnecessary for establishing a level above which ozone is harmful to human health or for estimating changes due to a spe- cific regulatory action, nor is it needed for quantifying human health re- sponses to short-term ozone levels or monetizing risk changes in response to ozone changes, which are the primary foci of this study. Thus, this Committee did not explore this issue in detail, though the report discusses aspects of PRB in Chapter 3. throughout California, and in major cities in Texas, Arizona, and Colorado. Figure 2-2 shows counties with monitor readings that would violate alternate 8-h ozone standards of 0.070 and 0.075 ppm proposed by EPA in June 2007.

OCR for page 23
28 FIGURE 2-2 Counties with monitors readings that would violate alternate 8-h ozone standards of 0.070 and 0.075 ppm proposed by EPA in June 2007 (on basis of 2003-2005 monitoring data). Source: EPA 2007e.

OCR for page 23
29 Ambient-Ozone Standards Development and Benefits Assessment On March 12, 2008, EPA issued revisions to the ozone primary and sec- ondary NAAQS (EPA 2008a). The level of the primary 8-h standard was low- ered to 0.075 ppm and the secondary standard was set to be the same as the re- vised primary standard. IMPLEMENTING THE NAAQS Once a NAAQS is set, EPA and the states and tribes pursue three basic paths to ensure that the standards are met throughout the country: • First, EPA establishes reference methods for measuring the indicator pollutant, and states and tribes implement monitoring programs to determine whether areas in their jurisdictions are in or out of attainment of the NAAQS. Under the 1997 ozone NAAQS, about 391 counties or parts of counties with a total population of over 36 million are in nonattainment of the standards (Figure 2-1). • Second, once EPA determines that any area is not in attainment of the NAAQS, the state that contains the area must develop a state implementation plan (SIP) that includes the actions it will take to reduce emissions and bring the area into attainment. States that are in attainment must consider actions that will ensure that they maintain that status. If a state fails in those efforts, EPA has the authority to implement its own plan; this has rarely occurred. • Third, to the extent that Congress has authorized EPA to take national or regional action to address emissions that lead to nonattainment, EPA proposes and implements regulations to do so (for example, national motor-vehicle and fuel standards). Beyond those regulatory actions, EPA works with states, tribes, and local authorities to implement an air-quality index system, which assesses the likeli- hood that air quality in a given area on a given day will be near or above the standard. The likelihood of poor air quality in turn sets off broad-based public- information efforts to alert residents, especially such sensitive populations as the elderly or asthmatic, to restrict activity that might increase exposure. In the case of ozone, regulatory actions at state and federal levels have taken many forms and have resulted in controls on emissions of volatile organic compounds and nitrogen oxides—the two primary precursors of ozone and other photochemical oxidants—from fuels, on-road and nonroad engines, electric- power facilities, manufacturing facilities, consumer products (such as paints), and many other sources. A National Research Council analysis of those efforts in 2004 found that although many actions have been taken at federal, regional, and state levels, the vast majority of reductions have come through national and regional multistate actions, rather than individual state actions. An exception is California, which has had the most aggressive air-quality programs in the nation (NRC 2004a).

OCR for page 23
30 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits THE SCIENTIFIC BASIS OF PRIMARY NAAQS FOR OZONE For over 50 y, detailed scientific investigations have examined nearly every aspect of ozone and other photochemical oxidants, and this robust body of research has served as the basis of each succeeding NAAQS determination by EPA. The research has taken several forms: • Atmospheric chemistry to understand the sources of ozone precursors and the formation of ozone and other photochemical oxidants. • Exposure research to characterize pollutant concentrations and human activity patterns that lead to contact with the pollutants over specific periods. Toxicologic research in vitro test systems (such as various human cells and other biologic media) and in vivo experiments (involving a wide array of animal species and animal models of human disease). • Epidemiologic research, including panel, cohort, and broader popula- tion studies of the acute or chronic effects of ozone exposure on morbidity and mortality in children and adults. • Controlled human exposure (or “clinical”) studies of varied healthy and compromised adult volunteers exposed to ozone or ozone combined with other gases. Research findings have contributed to the understanding necessary to es- tablish a NAAQS. In particular, the human studies—epidemiologic and clinical studies—have contributed several key pieces of standard-setting information, including information on the effects of real-world exposure of human beings to contaminants, the smallest exposure that can be demonstrated to have effects in humans, and the degree to which effects are found in sensitive populations. Recent Scientific Conclusions About Risks Posed by Ozone Exposure Overall, extensive toxicologic, epidemiologic, and clinical research on the effects of exposure to ozone has yielded strong evidence of effects on respira- tory end points at or near current regulatory levels, and in a variety of popula- tions. However, clinical studies have shown substantial variability in individual responses; even sensitive populations, such as asthmatics, exhibit a range of responses and nonresponse (EPA 2006a). Although an extensive literature—in toxicology, epidemiology, and controlled human exposure studies—has reported consistent evidence of such effects, there are still uncertainties in our under- standing, for example: • Uncertainty about the relationship between outdoor ozone concentra- tions and personal exposure, especially of persons who spend most of their time indoors or use air conditioning during periods of peak ozone concentrations.

OCR for page 23
31 Ambient-Ozone Standards Development and Benefits Assessment • Uncertainty about the degree to which epidemiologic studies can iden- tify the effects of individual pollutants, such as ozone, that are present in mix- tures of pollutants in the ambient air. • Uncertainty about the degree to which current study designs and data- sets enable understanding of risks resulting from exposures at very low pollutant concentrations and whether there is any threshold for effects. • Uncertainty, despite high-quality studies of long-term effects on lung growth in children (CalEPA 2006), about long-term exposure to ozone as a risk factor for chronic illness and premature death. On the basis of the available evidence, EPA in its most recent criteria document (EPA 2006a, p. 8-77) concluded that “the overall evidence supports a causal relationship between acute ambient ozone exposures and increased respi- ratory morbidity outcomes resulting in increased ED (emergency-department) visits and respiratory hospitalizations during the warm season.” EPA’s most recent staff paper (EPA 2007a, pp. 5-92 and 5-93), summarizing staff recom- mendations on the NAAQS for the administrator) further concludes that there is clear and convincing evidence of causality for lung function decrements in healthy children under moderate exertion for 8-hr average ozone exposures. We also judge that there is strong evidence for a causal relationship between respira- tory symptoms in asthmatic children and ozone exposures and between hospital admissions for respiratory causes and ambient ozone exposures. There is greater uncertainty and somewhat less confidence about the relationship between ozone and non-accidental and cardiorespiratory mortality, although the Criteria Docu- ment’s overall evaluation is that it is highly suggestive that this relationship ex- ists. The strengths and weaknesses of the recent literature on mortality and ozone are explored in greater detail in Chapter 4. Recent reviews of the evidence by the World Health Organization in its establishment of world air-quality guidelines (WHO 2006) and by the California Air Resources Board in its establishment of state air-quality standards (CalEPA 2007) have reached similar conclusions although the latter reviews have given somewhat greater causal weight to the associations with premature mortality. Evolution of the Science of the Health Effects of Ozone and Particulate Matter A broad and deep literature on the health effects of exposure to ozone has developed over the last 35 y and has resulted in a series of actions to set and revise the NAAQS and to reduce emissions of precursors. Similarly, over the last 15 y, much scientific and regulatory activity has been generated by intense interest in the effects of exposure to particulate matter (PM), especially PM with a diameter equal to or less than 2.5 µm (PM2.5) (EPA 2007f). There are useful parallels and differences between the literature on ozone and that on PM2.5, and

OCR for page 23
32 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits they have to some extent resulted in different approaches to the estimation of benefits of reducing exposure and in differences in actions taken. In general terms, the ozone literature developed first out of observations of populations exposed outdoors (for example, panel studies of summer campers) and then from what has become a large set of controlled human exposure (clini- cal) studies. More recently, and to some extent in parallel with the PM literature, ozone research has moved into the realm of broader population epidemiology with a growing number of hospitalization studies and other time-series studies starting in the 1990s. Cohort studies have not been designed to look specifically at ozone and premature mortality, and testing for the statistical significance of ozone-mortality relationships was not feasible due to insufficient variation in estimated long-term exposure to ozone (see Chapter 4). However several studies, especially the Southern California Children’s study, have examined longer-term morbidity effects (Tager et al. 2005; CalEPA 2006). Most cohort studies of PM have examined potential effects of ozone both as a confounder and for its inde- pendent effects, although few have found statistically significant evidence of such effects (see Chapter 4 for a review of the studies). One regulatory result of the relative absence of data on longer-term effects has been a focus on setting only a short-term NAAQS for ozone: a 1-h standard, which after nearly 30 y was replaced by an 8-h standard in 1997. The PM2.5 literature expanded rapidly after findings in the early 1990s of the associations of PM with morbidity and mortality in single-city time-series studies followed by the publication of results of two major cohort studies—the Harvard Six Cities Study (Dockery et al. 1993) and an American Cancer Society (ACS) study (Pope et al. 1995)—that showed much larger associations with longer-term residence in areas with high concentrations of PM2.5. Those results motivated the first major revision of the PM NAAQS in a decade in 1997 with the development of both shorter-term (24-h) and longer-term (annual) standards and a host of aggressive new regulatory actions for vehicles, power plants, and other sources. The controversy over those standards resulted in the launching of a major multidisciplinary research program (over $50 million/y for nearly a dec- ade), which has provided substantial additional information on exposure, toxi- cology, and epidemiology and involved the first use of clinical studies in this context (NRC 2004b). In turn, the new information led to many specific control actions related to such sources as diesel engines (EPA 2007g) and electric- power plants (EPA 2007h) and, in September 2006, an EPA decision to tighten the daily PM standard (but not the annual standard). Those actions were justified largely by cost-benefit analyses based on the substantial potential mortality benefits estimated from the ACS study and later reanalyses and extended analy- ses (Krewski et al. 2000; Pope et al. 2002, 2004). The focus on PM science and regulation over the last decade has had sev- eral ramifications for ozone. The understandably enhanced policy attention to PM resulted in substantially increased funding for and scientific attention to PM research, to some extent at the expense of research on ozone and other pollutants (except as they might be confounders of PM effects). The substantial estimated

OCR for page 23
33 Ambient-Ozone Standards Development and Benefits Assessment mortality benefits of reducing PM have resulted in estimates of health benefits of reducing PM that are much larger than those estimated for reducing ozone and other pollutants; one consequence has been in substantial political support for revisions in regulations that address PM. In the last several years, partly be- cause of the increased attention to mortality, there have been increased efforts on the part of regulators and scientists to understand better whether mortality effects may be attributed to exposure to ozone and, if so, to incorporate them into future standard-setting and benefits assessment. Those considerations caused EPA to fund of studies of the short-term mortality effects of ozone that are a primary focus of the present committee’s review. REGULATORY BENEFITS ASSESSMENT FOR SETTING AND IMPLEMENTING NATIONAL AMBIENT AIR QUALITY STANDARDS Quantitative assessment of the potential health benefits of actions planned to improve air quality has become a central component of regulatory impact analysis in the United States and other countries. As described in Figure 1-1 (see Chapter 1), the process involves estimation of the likely changes in population exposure to pollutants and the resulting changes in health risk as well as the economic valuation of the changes in health risk. Use of Health Studies for Regulatory Benefits Assessments There is a fundamental difference between information needed for regula- tory benefits assessment and information needed to set a protective health stan- dard. Selection of a primary (health-based) standard focuses on the lowest ambi- ent concentration that poses a risk of adverse health effects in the most sensitive population. Assessing benefits requires information for estimating all the reduc- tions in health risks in the entire population that is expected to experience a re- duction in ambient concentrations. The information needs for a comprehensive benefits assessment have led EPA to rely almost entirely on epidemiologic studies to support quantification of health effects of PM and ozone, whereas clinical and toxicologic studies are often important for standard-setting. Epidemiologic studies link health outcomes measured in the general population or in specific cohorts to differences in ambi- ent pollution concentrations at different times or in different locations. Wide- spread monitoring has made large-scale epidemiologic studies of the criteria pollutants feasible in the United States and many other countries. The real world thus becomes the laboratory, and subjects are studied in their normal environ- ments. That has three important advantages for benefits assessment: • Estimates of changes in rates of health outcomes in the population as a result of changes in ambient pollutant concentrations can be directly estimated

OCR for page 23
37 Ambient-Ozone Standards Development and Benefits Assessment BOX 2-3 Definition of Value of a Statistical Life Year in Relation to Value of a Statistical Life, and Numerical Examples The value of a statistical life year (VSLY) is another artificial con- struct, which gives a group of identical people’s aggregate willingness to pay to extend the life of one person in the group by 1 y. In theory, the VSLY could be estimated directly; in practice, it is usually derived from an estimate of the value of a statistical life (VSL). Numerical examples of VSLY: In the simplest case of a zero dis- count rate and a constant VSLY, the VSLY is the VSL divided by the num- ber of life years saved by the reduction in risk (calculated from the remain- ing life expectancy of the people at risk). Thus, using the numerical example for the VSL from Box 2-2, if the average remaining life expectancy for the group with the VSL of $5,000,000 is 40 y, the VSLY equals $125,000 ($5,000,000 divided by 40). In practice, the values for future years are presumed to be discounted at some rate of time preference such that years further in the future are given less weight in current decision- making than years close to the present. For a discount rate of 5%, for ex- ample, a VSL of $5,000,000 implies a larger VSLY of about $291,000 if the VSLY is assumed to be the same for every remaining year. In this exam- ple, the VSL is interpreted as a present value of a stream of equal annual values over a 40-y period discounted at an annual rate of 5%. It is not re- quired that the VSLY be a constant value for every year, although it is often assumed when a VSLY is calculated from a VSL. the reduction in pollution exposure. A calculation of life years saved takes into account the remaining life expectancy of those whose deaths were prevented in a given period. For example, if the reduction in pollution exposure is such that 10 70-y-old people who would otherwise have died next year do not die then and if their average remaining life expectancy is 15 y, 10 lives and a total of 150 life- years are saved. Of course, the specific people whose deaths are prevented can- not be identified, and remaining life expectancy of any specific person is not known, but estimates can be made by using pollution-related risk estimates for age groups and using life tables that give average remaining life expectancies of people of different ages. This is discussed further in Chapter 4. Proponents of using life years saved, rather than lives saved, as the unit of measure for mortality-risk reduction in cost-benefit analyses of environmental regulations have several arguments, including these: • A measure of life years is intuitively appealing because lives are never “saved,” rather, survival curves are shifted (that is, life expectancy is increased). • Life years saved are commonly used in medical-care decision-making and health-care policy analysis.

OCR for page 23
38 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits • An intervention or program that provides a person a larger life expec- tancy increase is intuitively preferable to one that provides a smaller life expec- tancy increase. The relative merits of the alternative methods for measuring and valuing mortality-risk changes are discussed in more detail in Chapters 4 and 5. Role of Cost-Benefit Analysis2 As noted above, the use of economic analysis—of the costs and monetary benefits of an action—is not allowed in the setting of the NAAQS, but such analysis is conducted in three important arenas. The first is a set of retrospective and prospective analyses of the costs and benefits of the CAA; these analyses are required by Section 812 of the 1990 amendments to the act. Two have been conducted: the first looked backward for the period 1970-1990, and the second looked forward for the period 1990-2010 (EPA 1997b, 1999a, 2003a). The other two uses of economic analysis are in response to a series of presidential directives, begun in the Ford administration and most recently codi- fied in Executive Order 12866 during the Clinton era (with later interpretation by the George W, Bush administration’s Office of Management and Budget, for example, in the recent Executive Order 13422). Executive Order 12866 man- dates a regulatory-impact analysis (RIA) of the costs and benefits of any federal agency action expected to have a “significant economic impact” (that is, an eco- nomic impact expected to exceed $100 million). For the air-quality management process, that has two facets: • First, an RIA is prepared for both a proposed and a final setting of each NAAQS. Although the law and recent court decisions (American Trucking As- sociations, Inc. v. U.S. EPA, D.C. Cir 97-1440 and 97-144) do not allow the results of this analysis to be used in setting the NAAQS, the RIA does provide public information about quantifiable costs and benefits of the proposal and in- form public debate. • Second, and perhaps more important, an RIA is prepared for every ma- jor EPA rule designed to reduce emissions of NAAQS pollutants or their precur- sors from mobile, stationary, and other sources. Although the RIA does not have a statutory role in EPA’s setting of the standard, the CAA provisions that author- ize actions to meet a standard normally do include cost among the criteria to be considered, so the cost and benefit analyses conducted in an RIA often are im- portant in determining which actions to reduce emissions may be most effective in improving health at the lowest cost. Furthermore, Sunstein (2002) has argued 2 The discussion in this section contains text that is excerpted or summarized from a document by L. Robinson (2007). The discussion is also informed by another document by L. Robinson (2004).

OCR for page 23
39 Ambient-Ozone Standards Development and Benefits Assessment that a series of judicial decisions has created a default requirement that, “Unless Congress has clearly said otherwise, agencies will be permitted to take costs into account…and expected to balance costs against benefits in issuing regulations.” (Sunstein 2002, p. 31) The U.S. Office of Management and Budget (OMB) gives guidance for conducting RIAs—most recently in 2003 (OMB 2003). EPA has guidance for conducting the analyses that is more specific to the issues related to environ- mental regulation (EPA 2000a). OMB notes (OMB 2003, p. 2) that the motiva- tion for the analyses is “to (1) learn if the benefits of an action are likely to jus- tify the costs or (2) discover which of various possible alternatives would be most cost-effective.” OMB’s guidance acknowledges that there are instances when important beneficial effects or costs of a regulation cannot be quantified or monetized. In such instances, OMB recommends a “threshold” or “break-even” analysis to evaluate the significance of nonquantified beneficial effects or costs. For in- stance, such an analysis would examine what the value of the nonquantified beneficial effects would have to be for the net benefits to be positive. For exam- ple, assume that a pollution-reduction program costs $10 million and has two benefits: benefits to the commercial forest industry that can be quantified and benefits to forest ecosystems that cannot be quantified. If the benefit to the commercial forest industry is estimated at $7 million, the benefit to forest eco- systems would have to be worth more than $3 million for the total benefits to exceed the costs. EPA’s guidance (EPA 2000a) provides more detail on what is to be in- cluded in quantified costs and benefits, how these are defined for cost-benefit analysis, and specific approaches that are used in the literature to quantify the types of costs and benefits that are typically affected by environmental regula- tions. Both guidance documents include the following general recommendations: • Conduct a comprehensive cost-benefit analysis for each regulatory al- ternative being assessed. This includes quantification and monetization of all costs and benefits, including those that might be considered ancillary (i.e., sup- plementary). When important costs or benefits cannot be quantified, a qualita- tive description of these costs or benefits should be included, as well as an as- sessment based on professional judgment of how important such costs and benefits might be. • The methods used and assumptions made should be presented in a transparent way to the reader, so that the policy maker can understand the proc- ess by which the results were obtained. • Uncertainty in the results should be communicated in a quantitative way, if possible, using techniques such as probability distributions, confidence inter- vals, or high, low bounds. The effects of key assumptions on the results should

OCR for page 23
40 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits also be presented, such as with sensitivity analyses that replace key assumptions with other plausible assumptions and demonstrate the effect on the results. • In all cases, the cost-benefit analyses are intended to support decision making, not determine the decision. In some cases there are other statutory goals besides efficiency that are overriding. There are also other considerations such as the distribution of costs and benefits to various groups or sectors of society that should be taken into account. Both of the guidance documents refer readers to standard texts on cost- benefit analysis and summarize the standard concepts that underlie this type of analysis. They note that the goal is to quantify both costs and benefits in terms of “opportunity cost.” For benefits, this means a measure of what those who benefit would be willing to forgo to obtain the specified benefit. This is the defi- nition of WTP, which is essentially a monetary measure of how much better off the benefiting population perceives themselves to be, assuming they have full information about what the benefits are. A fundamental point here is that the monetary measure of the value of the benefit is to be based on the value placed on the benefit by the group or sector receiving. It is not the value held by the policy maker or the experts. This is also why WTP estimates for changes in health are preferred to cost of illness estimates that reflect only medical costs and productivity losses. EPA’s Science Advisory Board, especially the Environmental Economics Advisory Committee, provides reviews and recommendations on drafts of EPA’s guidance document for conducting RIAs, and provides specific advice on cost-benefit analysis. In addition, the Advisory Council on Clean Air Compli- ance Analysis was established under Section 812 of the 1990 Clean Air Act Amendments to provide advice on the cost-benefit analysis of the Clean Air Act that was mandated by Section 812. The Council has provided detailed advice on cost-benefit analysis for air pollution related assessments, and EPA has consid- ered this advice in its RIAs as well as in the assessment of the costs and benefits of the Clean Air Act. In Circular A-4, OMB (2003) discusses the quantification and economic valuation of mortality risk reductions. OMB does not require the use of any spe- cific measure of effectiveness, such as lives saved or life-years saved, but en- courages agencies to report results using multiple metrics that may provide dif- ferent insights and perspectives. The OMB notes that [VSL and VSLY] are subject to continued research and debate and indicates that agencies should describe the limitations of their chosen approach. The Circular reports that the range of VSL esti- mates found in the literature is generally between $1 million and $10 mil- lion; as a result, regulatory agencies generally use values from within this range.

OCR for page 23
41 Ambient-Ozone Standards Development and Benefits Assessment In addition, Circular A-4 discusses options for adjusting VSL estimates to reflect differences between the scenarios addressed in the research litera- ture and the specific regulatory scenarios being assessed…It includes cau- tions on the application of age adjustments [for VSLs] and suggests the use of larger VSLY estimates for older individuals [if VSLY measures are used]…. In some of its regulatory assessments [in 2003 and earlier], the EPA pre- sented sensitivity analyses [using different VSL estimates for different age groups] based on research suggesting that older individuals are willing to pay less for life-saving interventions than younger adults (e.g., Jones-Lee 1989; Jones-Lee et al. 1993). In response, the OMB issued a memorandum advising agencies against adjusting the VSL for age (Graham 2003). This memorandum suggested that more recent research (ultimately published in Alberini et al. 2004a) did not fully support the VSL age adjustment found in earlier studies. It indicated that, when VSLY estimates are used instead of VSL, the yearly values are likely to be higher for senior citizens because “seniors face lar- ger overall health risks from all causes and because they have accumulated savings and liquid assets to expend on protection of their health and safety” (Graham 2003, p. 2)... However, the guidance in this OMB memorandum, which was eventually incorporated into Circular A-4, does not necessarily eliminate the use of different values for younger versus older individuals. When VSLY esti- mates are applied, the total value of a risk reduction is equal to the product of the VSLY estimate and the discounted number of life-years saved. Unless the VSLY estimates for older individuals are large enough to com- pensate for the smaller number of life-years remaining, the use of VSLY estimates will result in lower values for older individuals (Robinson 2007, p. 286-287). ENVIRONMENTAL PROTECTION AGENCY’S APPROACH TO ESTIMATING OZONE MORTALITY IMPACTS AND VALUING MORTALITY RISK REDUCTION3 Several examples show how EPA’s benefits assessments include quanti- fied and monetized ozone health benefits. All of these have included several morbidity endpoints in the primary estimates. Ozone mortality has been quanti- fied as a sensitivity analysis based on available daily time series estimates of the 3 The discussion in this section contains text that is excerpted or summarized from a document by L. Robinson (2007). The discussion is also informed by another document by L. Robinson (2004).

OCR for page 23
42 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits effect of ozone on mortality risk and using the VSL that has been used by EPA for PM mortality. These have included the • 1997 NAAQS RIA (EPA 1997a) • Costs and Benefits of the CAA (EPA 1997b; prospective 1999a) • Tier 2 Motor Vehicle Emission Standards RIA (1999b) • Clean Air Interstate Rule (CAIR) RIA (EPA 2005a) and • Proposed Small Engines Emission Rule (EPA 2007i). EPA has prepared a RIA to accompany their proposal for revising the NAAQS for ozone (EPA 2007b). In that analysis, EPA has included calculations of reductions in mortality associated with reductions in ozone concentrations using alternative approaches and has presented all the results side-by-side. One calculation was based on the results of a multi-city study by Bell et al. (2004). This study provided the basis for the “primary” estimates of mortality reductions in EPA’s earlier risk assessment. A second set of calculations was based on the three meta-analyses of ozone related mortality studies. These results gave mor- tality reduction estimates about 4 to 5 times larger than the estimates based on Bell et al. (2004), and that were relatively close to one another. The third calcu- lation excluded any mortality associated with ozone and counted only the mor- bidity health outcomes. In addition to the health benefits of reductions in ozone, EPA calculated the health benefits of PM2.5 reductions that would result from control strategies expected to be used to meet the ozone standards, which would reduce precursors that cause formation of both ozone and PM2.5. EPA’s current primary approach for economic valuation of mortality risk reductions is a variation on their long-standing approach of using the same VSL for all annual mortality reductions. Starting with the RIA for the Clean Air Inter- state Rule (EPA, 2005a), EPA has been using a central VSL that is a midpoint between results obtained in two meta-analyses of the wage-risk literature (de- scribed in Chapter 5). This is a VSL of $5.5 million in 1999 dollars and at 1990 income levels. When adjusted to expected income levels in 2006, this is about $6.3 (still in 1999 dollars). This VSL reflects the Agency’s estimates of the in- dividuals’ willingness to pay (WTP) for small reductions in the risk of prema- ture death for the population at large. EPA’s current selection of VSL from the literature is similar to, but somewhat lower than the mean VSL EPA used earlier that was based on a review of a number of both revealed preference and stated preference studies (EPA 2000a). In their primary benefits estimates, EPA applies the VSL to all lives saved regardless of the age or health status of the population experiencing the change in mortality risk and regardless of the cause of the mortality risk change. Al- though there is some expectation that willingness to pay for mortality risk reduc- tions may vary with the characteristics of the population affected or with the context of the risk change, EPA has concluded that there is insufficient empiri-

OCR for page 23
43 Ambient-Ozone Standards Development and Benefits Assessment cal information available upon which to base adjustments for these factors at this time. EPA’s SAB (2007) has agreed with this conclusion regarding the state of the currently available literature. Following advice from the SAB, EPA adjusts its base VSL estimates for expected future real income growth and for any ex- pected time lags between change in long-term pollution exposures and mortality risk reductions (often referred to as cessation lags). EPA included several differ- ent sensitivity analyses of the effects of age adjustments (adjusting VSL and/or applying VSLY estimates) in several of its reports prior to the development of OMB’s Circular A-4, which cautions against making age adjustments to the VSL (Robinson 2007). These sensitivity analyses were included, for example, in the retrospective and prospective assessments of the costs and benefits of the Clean Air Act (EPA 1997b, 1999a) and in the RIAs for the heavy-duty diesel rule (EPA 2000b) and the Clear Skies legislation proposal (EPA 2003b, 2006b). “For example, for the heavy-duty diesel rule (EPA 2000c), the EPA used VSL age adjustments based on Jones-Lee (1989) and Jones-Lee et al. (1993) in sensitivity analysis, which reduced its primary benefits estimate by 10 or 40 percent, depending on the adjustment factor applied. In a sen- sitivity analysis for regulations addressing emissions from large spark ig- nition engines (EPA 2002), the agency used a more complicated approach that reflected initial results from the work of Alberini et al. (2004) as well as the adjustment factor from Jones-Lee (1989)” (Robinson 2007, pp. 290- 291). In this case, EPA derived different estimates of VSLY for younger and older age groups from selected VSL estimates. The result was a higher VSLY for the older age group. They applied these VSLY estimates of life-years saved for each age group. The net effect was a lower value for risk reduction for the older age group because the smaller remaining life expectancy more than offset the higher VSLY. As part of the process of updating their Guidelines for economic analysis, EPA asked SAB to consider again the question of whether monetary values for mortality risk reduction in cost-benefit analyses should be adjusted for differ- ences in the life expectancy of those at risk. The SAB (2007) concluded that economic theory provides indeterminate results about how remaining life expec- tancy may affect a person’s WTP for mortality risk reduction and that this ques- tion must be addressed empirically. They further concluded that results in the empirical literature are also insufficient at this time to provide a basis for mak- ing quantitative adjustments to VSL according to the age or remaining life ex- pectancy of the population at risk. In addition, they concluded that the empirical literature provides no support for the assumption implicit in the use of a constant VSLY that WTP for mortality risk reduction is proportional to remaining life expectancy. They recommended continued use of a constant VSL, with encour- agement to EPA to fund more research to address the questions of how VSL varies with age and other population and risk characteristics.

OCR for page 23
44 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits OVERVIEW OF OTHER APPROACHES Other Federal Agencies Other federal agencies do not deal with ozone mortality in their regulatory analyses, but they often deal with other kinds of mortality risk. Other agencies promulgate fewer economically significant rules that re- quire valuing the risk of premature mortality. Between October 2003 and September 2005, four agencies (in addition to the EPA) prepared final rules with quantified health and safety benefits that were reviewed by the OMB (OMB 2005, 2006). These agencies included the Food and Drug Administration (FDA) and the Centers for Medicare and Medicaid Ser- vices (CMS) in the Department of Health and Human Services (HHS), as well as the National Highway Traffic Safety Administration (NHTSA) and the Federal Motor Carrier Safety Administration (FMCSA) in the Depart- ment of Transportation (DOT). An earlier review, covering the period be- tween January 2000 and June 2004, reported similar patterns in agency promulgation of major health and safety rules (Robinson 2004)” (Robin- son 2007, p. 293). Agencies of the Department of Health and Human Services (Food and Drug Administration and Centers for Medicare & Medicaid Services) The FDA does not provide formal internal guidance for economic analysis, but it applies a similar approach across many of its rules. For premature mortality, the agency often uses a VSL estimate of $5million, without specifying a dollar year, and occasionally provides alternative estimates using higher or lower values [see, for example, 68 Fed. Reg. 41434 [2003], 69 Fed. Reg. 9120 [2004]; 70 Fed. Reg. 33997 [2005]]. This estimate is roughly in the middle of the $1 million to $10 million range cited in Cir- cular A-4 (OMB 2003). The FDA rarely adjusts its VSL estimates for sce- nario differences, although it has addressed cessation lag (e.g., in its trans- fat rule, FDA 2003), and added the cost of cancer treatment ($25,000) and an adjustment for psychological factors ($5,000) to the VSL for a rule on X-rays [(70 Fed. Reg. 33997 [2005]]. Thus, while its base VSL estimates are similar to those used by the EPA, the values ultimately applied by the FDA may be quite different because of the income growth and other ad- justments made by the EPA. A few FDA analyses have presented alterna- tive estimates of the value of mortality risk reductions using VSLY as well as VSL estimates [e.g., 68 Fed. Reg. 41434 [2003]] (Robinson 2007, p. 293).

OCR for page 23
45 Ambient-Ozone Standards Development and Benefits Assessment FDA sometimes uses VSLY estimates to quantify monetary values for changes in morbidity following the methods developed by Mauskopf and French (1991); these estimates are derived from VSL estimates in the literature assum- ing constant VSLY for remaining (discounted) life expectancy. Another HHS agency, CMS, develops few economically significant rules with health and safety impacts; most of its programs involve transfers (e.g., from taxpayers to Medicare and Medicaid recipients) and hence are not subject to the OMB requirements for regulatory analysis. In its immuniza- tion rule (70 Fed. Reg. 58835 [2005]), CMS applies the same VSL esti- mate as FDA ($5 million), noting that it is roughly the mid-point of the range of values suggested by OMB (Robinson 2007, p. 294). Agencies of Department of Transportation (National Highway Traffic Safety Administration and Federal Motor Carrier Safety Administration) Both the NHTSA and the FMCSA rely on the DOT guidance for their base VSL estimates. In contrast to EPA and the HHS agencies, the DOT agen- cies primarily address injury-related accidental deaths rather than deaths from illness. The DOT currently recommends the use of a $3.0 million VSL—noting that this value is imprecise and should be used as “a guide for thoughtful decision-making” (DOT 2002, p. 1). Its approach is based largely on the results of Miller (1990), with adjustments for inflation and newer studies. Miller’s 1990 estimates vary from those used by the EPA because he ap- plies different criteria to determine which studies to include, and adjusts the results to address certain limitations of the studies. The DOT indicates that it continues to review the literature and consider whether changes to this value are needed (DOT 2002) (Robinson 2007, p. 294). California The state of California conducted a benefits assessment during the process of making their latest revisions to the state ambient ozone standard in 2004. They selected central, low and high estimates of the percentage change in mor- tality association with daily fluctuations in ambient ozone concentrations, based on available daily time-series studies. The entire selected range was above zero, implying a 100 percent probability of a nonzero effect of ozone on mortality. Their results were not used to help set the standard, but to assess what the health benefits to the California public would be if the alternative ambient standards were met compared to current concentrations. In a subsequent publication, Ostro et al. (2006) included a monetary valuation for the ozone-related mortality cases, using EPA’s selected estimate of the VSL.

OCR for page 23
46 Ambient Ozone and Mortality: Estimating Risk-Reduction Benefits Canada The Canadian government has also used benefits assessments to determine the quantity and economic valuation of health effects associated with air pollu- tion in Canada and for programs to reduce air pollution. Judek et al. (2005) es- timated the effects of all air pollutants on mortality rates in Canada, including estimates of the effects of ozone on mortality. The concentration-response func- tion for ozone was drawn from a daily time-series study of 12 Canadian cities (Burnett et al. 2004), using a multi-pollutant model the authors estimated for Health Canada (not reported in the published paper) (Personal communication with D. Stieb, August 8, 2007). The concentration response function for ozone mortality used was about 0.084% (standard error of 0.014%) change in mortality per ppb change in daily high-hour ozone. For monetary valuation of air pollution related mortality, Health Canada has been using a range of VSL estimates based on the WTP literature in Canada and the United States. These VSL estimates have been adjusted downward somewhat based on results from a relatively early stated preference study (Jones-Lee et al. 1985) that showed a lower WTP for mortality risk reduction for older subjects. VSL estimates from the working age population were thus adjusted downward to reflect the age distribution of non- accidental causes of mortality that are associated with air pollution. Health Can- ada is currently planning to update their procedures for selecting values for mor- tality risk reductions. Europe There have been several efforts in Europe to estimate both the mortality impacts of exposure to ozone, and to value those impacts. These have included the ongoing work of the European Union’s ExternE program, as well as the July 2007 Clean Air Strategy of the United Kingdom (DEFRA 2007). While the gen- eral approach for estimating impacts and benefits is very similar to that followed in the United States (as described in Figure 1-1 in Chapter 1), there are three areas where these efforts have differed from normal EPA practice to date: • First, these analyses have fairly unambiguously accepted from the epi- demiologic literature that there are robust associations of mortality with ozone exposure and thus have used European time-series studies of these effects to estimate health impacts of current exposure, and the potential reduction in those impacts from regulatory actions. • Second, these analyses have generally included an analysis of life-years lost, in the case of the UK including estimates of both life years and numbers of deaths (DEFRA 2007), while in the case of ExternE, rejecting the estimation of numbers of deaths and relying solely on life years lost (Bickel and Friedrich 2005).

OCR for page 23
47 Ambient-Ozone Standards Development and Benefits Assessment • Third, the economic valuation estimates used in this ExternE analysis are from recent stated preference studies conducted in France, Italy, and United Kingdom (Alberini et al., 2006b). Their selected central estimate of EUR 50,000 per life year was derived from the median WTP value for the 5 in 1,000 mortal- ity risk change over 10 years (median VSL was about EUR1.1 million). The ExternE authors converted this 10-year mortality risk change to its equivalent in increased life expectancy for each age/gender cohort and calculated the VSLY implicit in the WTP responses. THE MAJOR QUESTIONS As this overview has suggested, there is a broad and deep literature docu- menting a range of health effects from exposure to ozone, and this has had an important role in the setting and implementation of NAAQS for ozone for over 35 years. In considering whether and how the recent studies of ozone and mor- tality should be incorporated in estimating benefits going forward, the Commit- tee focused on two primary areas of questions: • The Robustness of the Ozone Mortality Studies. The estimation from epidemiologic evidence of health impacts, and the economic value of those im- pacts, requires confidence, as noted above, that ozone exposures can adequately be separated from other exposures in the epidemiology studies and that the epi- demiology has adequately controlled for possible confounding factors. Although a number of others in Europe and California have made the judgment that these studies are adequate, the Committee was charged to make an independent de- termination on these questions. These questions are addressed in detail by the Committee in Chapters 3 and 4. • The Appropriate Methods for Estimating the Value of Potential Ozone Mortality Benefits. If the epidemiology studies provide adequate evidence for the estimation of mortality impacts, there are important questions about the ap- propriate methods for estimating the value of such impacts that are described initially above, and are addressed in detail in Chapter 5.