Framing the Analysis
The estimates of health benefits depend critically on the choices made in framing the analysis (what will and will not be included) at the beginning of the process. The most important of these choices are (1) the regulatory options to consider, (2) the health effects to evaluate, (3) the time frame for the analysis, including the years in which benefits are evaluated, and (4) the assumptions to make about conditions with and without the regulation implemented. The assumptions influence the benefits by determining the size of the emissions reductions attributed to the regulation and by determining the size, income, and health status of the population that will benefit from the air pollution regulation. This chapter discusses how EPA has dealt with each of these sets of decisions and uses examples from the four EPA benefits analyses reviewed by the committee and summarized in Chapter 2 of this report.
REGULATORY OPTIONS EVALUATED
In three of the analyses examined by the committee, EPA focused on evaluating a single set of regulatory options: (1) end-of-tailpipe emissions controls for passenger vehicles and reduction of the sulfur content of gasoline in the Tier 2 emissions standards (EPA 1999a); (2) measures to make heavy-duty engines less polluting and reduction of sulfur content of diesel fuel in the heavy-duty (HD) engine and diesel fuel rule (EPA 2000);
and (3) a set of measures that would achieve the goals of Titles I-V of the 1990 Clean Air Act Amendments (CAAA) in the prospective analysis of the Clean Air Act (EPA 1999b)—hereafter referred to as the prospective analysis. For each of the first two rules, only a single package of phased-in changes in capital equipment and fuel composition was evaluated. Alternative types of controls or different schedules for phasing in the controls were not considered. The prospective analysis estimated the benefits and costs of the first five titles of the 1990 CAAA combined and did not attempt to disaggregate benefits by title.
In the analysis of the particulate matter (PM) and ozone National Ambient Air Quality Standards (NAAQS) (EPA 1997), the agency considered three regulatory alternatives that were combinations of the following annual average and 24-hr standards for PM2.5: (1) 16 µg/m3 and 65µg/m3, (2) 15 µg/m3 and 65µg/m3, and (3) 15 µg/m3 and 50 µg/m3. Similarly, the maximum number of annual exceedences allowed under the proposed 1-hr ozone standard varied from 3 to 4 to 5. These options were compared, assuming partial attainment of each option.
In general, EPA’s approach does not satisfy Office of Management and Budget (OMB 1996, 2000) guidance on benefits analysis. The OMB guidelines include consideration of a range of levels for the standard and different time schedules for compliance, as well as a variety of qualitatively different market interventions, such as information measures, market-based approaches, performance-based standards, and different requirements for different segments of the regulated population. When a regulatory action represents a package of different provisions, such as the various titles of the 1990 CAAA, OMB suggests that the parts of the package be assessed separately to the extent feasible. Specifically, OMB (1996) makes the following statements:
If the proposed regulation is composed of a number of distinct provisions, it is important to evaluate the benefits and costs of the different provisions separately. The interaction effects between separate provisions (such that the existence of one provision affects the benefits or costs arising from another provision) may complicate the analysis but does not eliminate the need to examine provisions separately. In such a case, the desirability of a specific provision may be appraised by determining the net benefits of the proposed regulation with and without the provision in question. Where the
number of provisions is large and interaction effects are pervasive, it is obviously impractical to analyze all possible combinations of provisions in this way. Some judgment must be used to select the most significant or suspect provisions for such analysis.
For the HD engine and diesel-fuel rule, there are clearly good reasons why some changes should not be considered in isolation from other changes. For example, changes in end-of-pipe pollution-control equipment, such as particle filters and regeneration systems, should not be considered without changes in fuel composition. However, there is no obvious reason why the effects of the fuel changes without the equipment changes or with equipment changes implemented at different periods could not have been evaluated for their effects over time.
In the case of the PM and ozone NAAQS, it would be valuable to know how much benefits and costs increase as the ambient air-quality standard for PM is tightened. In other words, how do benefits and costs change as the PM2.5 standard moves from an annual average of 20 µg/m3 to 15 µg/m3? In the case of the 1990 CAAA, over 80% of the total cost of Titles I-V is associated with Titles I and II alone. Although the costs are reported separately for Titles I and II, it would be useful to know whether the estimated benefits of these two titles exceed their estimated costs.
In agreement with and extending the OMB guidance, the committee believes that EPA should seek to represent a realistic range of regulatory choices guided by expert opinion and technical feasibility. The agency should, at the beginning of each analysis, discuss the range of choices and the preliminary analyses that were conducted to exclude certain options from the formal analysis. This approach would strengthen analyses that currently appear to serve the purpose of justifying the agency’s chosen regulatory option without comparing that option with other feasible possibilities.
A related issue concerns assumptions made about compliance with air pollution regulations. As indicated in Chapter 1, current EPA Office of Air Quality, Planning and Standards guidance calls for analysts to assume full compliance with regulatory requirements when estimating the costs and benefits of regulations. The committee believes that this recommended approach should be changed, because decision-makers and the public should be given the likely results of different regulatory choices as accurately as possible. Assuming perfect compliance may often result in overestimation
of the health benefits and costs likely to result from a new regulation. Incorporating alternative assumptions about compliance into a sensitivity or uncertainty analysis would more completely convey the full range of potential benefits.
Furthermore, assuming perfect compliance is likely to result in the agency’s neglecting the important issue of the relative cost and effectiveness of alternative implementation and enforcement measures. In the absence of a comparative analysis of implementation, decision-makers will not be able to compare regulatory options that are likely to differ in the ease and reliability of implementation. For example, EPA enforcement of regulatory requirements that change the emissions characteristics of newly marketed engines may be relatively straightforward and inexpensive compared with enforcement of requirements that operators maintain engines at a specific standard.
Notable exceptions to the above criticisms include the compliance assumptions made in the prospective analysis and the HD engine and diesel-fuel analysis. For the prospective analysis, EPA did not assume perfect compliance with proposed regulations but assumed stationary sources would achieve only 80% of target reductions for nitrogen oxides (NOx) and volatile organic compounds (VOCs). Furthermore, actual emission rates for mobile sources reflected real-world tampering and other noncompliance issues. For the HD engine and diesel-fuel analysis, EPA analyzed the potential impacts on future emissions of tampering with and inadequately maintaining the proposed HD diesel-control technology. The committee endorses EPA’s stated goal of expanding its current capability to analyze the potential impacts of incomplete compliance with proposed regulations by developing improved data on actual emissions.
SELECTION OF EFFECTS TO EVALUATE
EPA must determine how broadly to define the scope of each analysis. This task includes determining the categories of benefits to evaluate and the extent of examination of secondary or unintended effects of the regulation. Although the evaluation of the direct effects of the regulation on human health is the primary focus of the analyses reviewed, the committee notes that the regulations may also affect human health indirectly. Air pollution
regulations intended to change air pollution levels may also change how fuels are made or how combustion devices are operated. These changes can then affect human health through other pathways. Although outside the strict boundaries of public exposure to air pollution, an analysis of health benefits that ignores these indirect effects may result in a substantial misrepresentation of the actual impact of pollution control measures on society. Therefore, the analyst should seek expert guidance when appropriate and consider such issues as the following:
Can the regulation potentially compromise occupational health? For example, a measure to control VOC emissions from an industry may cause an increase in occupational exposures to toxic substances by reducing ventilation in production areas.
Can the regulation potentially increase pollution in other locations? For example, a policy measure that shifts electricity production toward hydroelectric power plants relative to fossil-fuel power plants might result in substantial increases in cement production and subsequent air pollution consequences in other locations.
Can the regulation potentially cause cross-media effects? For example, use of methyl-t-butyl ether to control air pollution from vehicle emissions resulted in increased water pollution.
Therefore, a health benefits analysis should examine the potential for important impacts outside the narrow boundaries of population exposures to air pollution. It should also contain a discussion on whether such impacts could be important. If they are, guidance on assessing them more completely should be included. The committee recognizes that time and resource constraints may require trade-offs between the number of scenarios considered and the level of detail for each.
As an aside, the committee notes that the examples provided are unintended negative impacts and that there may be unintended positive impacts of air pollution control regulations outside the boundaries of the analysis. For example, air pollution control in other parts of the world may be accelerated due to a demonstration effect or economic pull of control efforts in the United States. However, these effects are typically difficult to predict in advance or even to assess after the fact.
TIME FRAME FOR THE ANALYSIS
EPA’s analysis of the costs of a regulation typically begins in the year the regulation first goes into effect and continues until the regulation is fully implemented. For example, for the HD engine and diesel-fuel rule, costs were computed from 2006 (the year in which proposed engine modifications and other equipment are to be installed in new trucks) to 2030 (the year in which these modifications will be embodied in all trucks in the fleet). Similarly, the prospective analysis computed the costs of implementing the 1990 CAAA from 1990 to 2010 (the years in which selected provisions of the 1990 CAAA are likely to have been fully implemented).
On the other hand, health benefits are typically estimated for only a single year in the future. The analyses for the Tier 2 emissions standards and the HD engine and diesel-fuel rule evaluated benefits only in 2030. The analysis for the PM and ozone NAAQS evaluated health benefits in 2010. In contrast, the prospective analysis evaluated benefits in 2000 and 2010. In the prospective analysis, benefits in intermediate years were interpolated to calculate the present discounted value of benefits from 1990 to 2010.
The years 2030 and 2010 were chosen because the policies under consideration would likely be implemented by these dates. For example, the Tier 2 emissions standards and the HD engine and diesel-fuel rule both involve modifications in new vehicles required before 2010. The policies will not be fully implemented, however, until all vehicles in the fleet contain these modifications.
Evaluating benefits in only a single year in the future has two limitations. First, when the costs of a policy decrease over time and the benefits increase, a comparison of the benefits and costs only in the distant future is highly misleading. The comparison will overstate the benefits achieved in the early years of the policy; however, the committee does not know how great the overstatement would be. This problem arose in the HD engine and diesel-fuel rule in which the costs of the rule are concentrated in the early years of the regulation, in part because of research and development costs. No attempt was made, however, to compute benefits for an intermediate year, such as 2015.
Second, choosing an evaluation point in the distant future, such as 2030, is likely to increase the uncertainty associated with the calculation of benefits and costs. For example, it is highly uncertain what the passenger vehicle fleet will look like in 2030 and how polluting it would be without Tier 2
emissions standards. Unless this uncertainty is accurately reflected in benefit and cost estimates, the analysis will be misleading.
To EPA’s credit, the analysis for the HD engine and diesel-fuel rule acknowledged that focusing on 2030 might be misleading but cites the high cost of evaluating benefits in years before 2030—primarily due to the cost of air-quality modeling—as a reason for its decision to use 2030. Specifically, EPA (2000) made the following statements:
A more appropriate means of capturing the impacts of timing differences in benefits and costs would be to produce a net present value comparison of the costs and benefits over some period of years. Unfortunately, while this is relatively straight-forward for the costs, it is currently not feasible to do a multi-year analysis of the benefits as this would require a significant amount of air quality modeling to capture each year. We did not have the resources for such an extensive analysis.
The high cost of running multiple air-quality scenarios is likewise cited in the following statements by EPA (1999b) as a reason for aggregating Titles I-V in estimating the benefits of the 1990 CAAA:
The estimates in Table 8-3 reflect the difficulty we encountered in reliably disaggregating benefits by CAAA Title or even by pollutant…. These difficulties in separating the effects of individual emissions reductions on the benefits estimates also highlights the need for an integrated air quality modeling system that can more readily analyze multiple scenarios within reasonable time and resource constraints. A tool of this nature could allow us to more reliably and cost-effectively estimate incremental contributions to ambient PM and ozone concentration reductions.
In presentations before the Science Advisory Board (M. Cropper, University of Maryland, personal communication, June 6, 2002), EPA staff also cited the high cost of air-quality modeling as a reason for not quantifying the uncertainty in emissions estimates and carrying this uncertainty forward in estimating avoided cases of morbidity and mortality.
The committee believes, however, that EPA should make every effort to estimate health benefits associated with reductions in air pollution at
reasonably frequent intervals, such as every 5 years, over the regulatory time frame, including the period of implementation and the expected period of expression of all significant health effects. EPA should modify air-quality models used in translating predicted emissions into predicted levels of ambient air quality to reduce the resources required for air-quality modeling. This change is necessary if EPA is to evaluate multiple regulatory alternatives and if it is to evaluate each alternative at reasonable time intervals, such as every 5 years. The ability to evaluate the ambient air quality associated with more emissions scenarios is also essential if the uncertainty inherent in emissions estimates is to be carried through to estimating avoided cases of mortality and morbidity. The committee notes that emissions and ambient air quality with and without the regulation are treated as certain in the EPA analyses reviewed by the committee. EPA also treats costs as certain.
Because some important evaluation methods, particularly net present-value calculations, require annual estimates of benefits (and costs), full benefits estimates should be accompanied by presentations of benefits, using an appropriate and clearly described interpolation method, for intervening years. The committee notes that the additional precision provided by running all the models for intervening years is unlikely to be worth the effort, given the overall uncertainties in benefits estimation.
Finally, the health benefits of reducing emissions in a single year might not occur solely in that year but might occur in subsequent years because of physiological and other lags. The analyses should carefully state and document the lag relationships between pollution reductions and health improvements that have been used (see Chapter 4).
DESCRIPTION OF CONDITIONS WITH AND WITHOUT THE REGULATION
To estimate the benefits of an air pollution regulation, EPA predicts future conditions with and without the regulation enacted. Two sets of predictions are especially relevant to calculating the health benefits of the regulation. The first describes emissions by sector in the absence of the regulation and emissions by sector after the regulation is imposed. The second set of predictions relates to the population affected by the changes in air quality—the number of people (by age, gender, and location) living in
the United States and the disease and death rates in this population. This section addresses how EPA makes and reports these predictions. Chapter 4 discusses how emissions predictions are translated into ambient pollution concentrations and how the change in ambient concentrations, together with population and baseline rates of disease and death, are used to calculate avoided cases of morbidity and mortality.
In all four analyses reviewed by the committee, EPA predicts emissions for all major source categories of the criteria pollutants: industrial point sources, utilities, nonroad engines and vehicles, motor vehicles, and area sources for one or more future years, such as 2010 or 2030. These predictions are made without the regulation analyzed in the study (designated the regulatory baseline) and with the regulation. The complexity of the models used to predict emissions for electric utilities (the integrated planning model [IPM] developed by ICF, Inc.) and for motor vehicles (MOBILE5 and MOBILE6) is such that only the emissions predicted by these models are summarized in the appendixes to the regulatory impact assessments (RIAs). The models are described in other documents (EPA 2002a,b).1
Two issues regarding emissions predictions particularly concern the committee. The first issue is how the emissions estimates with and without the regulation are reported. The documents reviewed here fail to give the reader information on what drives the emissions estimates and make it difficult to judge the plausibility of the estimates. In most sectors, emissions are the product of the level of an activity (such as fuel consumed by electric utilities or miles traveled by motor vehicles) multiplied by the amount of
IPM is a linear programming model that describes electricity demand, generation, transmission, and distribution for all plants in the U.S. electric power market. See http://www.epa.gov/capi/ for further details. The MOBILE models use data on the U.S. vehicle fleet to estimate emissions from motor vehicles. The vehicle fleet is characterized by the total number of vehicles in operation within certain categories, their age distribution and fuel type (gasoline or diesel), and their annual mileage rates by age and fuel type. This information, together with estimates of emissions factors, is used to calculate total fleet emissions. See http://www.epa.gov/otaq/mobile.htm for further details.
pollution generated per unit of activity (such as pounds of sulfur dioxide [SO2] per millions of British thermal units [mmBtus] or grams of NOx per mile traveled). The assumptions about activity levels and the pollution intensity, both with and without the regulation, can be made explicit even though it is not possible to describe in detail all of the assumptions underlying these numbers. The committee emphasizes that readers might find it easier to judge the plausibility of the estimates if they were expressed as percentages or if they were compared to historical trends. For example, what percent change in vehicle miles traveled is implicit in emissions estimates for 2030 compared with current levels? What is the percent reduction in pollution intensity estimated to be achieved by a regulation?
The second issue concerns the deterministic nature of the models used to predict emissions. Both IPM and EPA’s mobile-source emissions models fail to incorporate any uncertainty in their emissions predictions. In general, any variable that is likely to have a substantial impact on mortality and morbidity and to have considerable uncertainty should be a candidate for a formal uncertainty analysis. Predictions of activity levels 20 years in the future, such as percent of light-duty trucks using diesel fuel, fall in this category.
The calculation of emissions predictions, the ways in which the information should be presented, and the relevance of uncertainty to the analysis are discussed in the following sections, using as examples emissions predictions for electric utilities and emissions predictions for motor vehicles.
Emissions Predictions for Electric Utilities
In the prospective analysis, EPA predicts SO2 emissions for electric utilities in 2010 with and without regulatory action. In each case, total SO2 emissions are the product of the fuel consumption (measured in mmBtu) and the pollution intensity (the number of pounds of SO2 per mmBtu produced) for each electricity-generating unit, summed over all units. Equation 1 depicts this calculation.
Total SO2 Emissions = • •(mmBtu)i × (SO2/mmBtu)i, (1)
where i denotes a generating unit.
The total SO2 emissions from electric utilities in 2010 are predicted to be 18 million tons without the 1990 CAAA and 9.9 million tons with the 1990 CAAA. However, the analysis gives no information on what accounts for those results. Although this information may be available in technical support documents, additional information about the components of total SO2 emissions could be presented for the two scenarios in the main text as a table listing the national aggregate fuel consumption by category of power plant and the national average pollution intensity by category of power plant. This breakdown of the components of predicted emissions could also be supplemented with historical information on aggregate fuel consumption and average pollution intensity by class of power plant. This information would allow the reader to compare actual values with agency predictions. This table would indicate the extent to which the predicted reduction in SO2 emissions attributed to the 1990 CAAA was the result of an average reduction in fuel consumption or pollution intensity. This information should be supplemented with a measure indicating the extent to which emissions reductions are predicted to result from shifting electricity production from dirtier to cleaner units as a result of the 1990 CAAA.
Supplementing aggregate emissions estimates with the information described above would demonstrate how the predicted reduction in emissions is to be achieved and would highlight important factors to consider in an uncertainty analysis. Suppose, for example, that most of the SO2 reduction is expected to come from a reduction in the average pollution intensity of coal-fired power plants. If this factor drives the results, then it is important to further examine the assumptions underlying pollution intensity with and without the 1990 CAAA. One way to examine the assumptions would be to make the predictions of the IPM model explicit for pollution intensity with and without the 1990 CAAA and to compare those predictions with historical trends in pollution intensity over the period 1980-1995. (The provisions of the 1990 CAAA that affect SO2 emissions from power plants went into effect in 1995.) If the predictions without the CAAA appear to be inconsistent with historical trends, this discrepancy should be explained and formally incorporated into an uncertainty analysis. Other components of emissions that might be subjected to uncertainty analysis are estimates of electricity demand that underlie the total amount of fuel burned by power plants and, for longer time periods, assumptions about the retirement of old plants and the construction of new plants.
Emissions Predictions for Motor Vehicles
All four of the health benefits analyses examined by the committee make predictions about the effects of air pollution regulations on motor-vehicle emissions. The total emissions of a pollutant, such as NOx, from motor vehicles can be written as the sum of average annual NOx emissions for each class of vehicle i times the number of vehicles in that class (ni). Average annual NOx emissions for vehicles in class i are, in turn, the product of NOx emissions per mile (NOx/mile)i times average annual vehicle miles traveled (VMT)i. The overall calculation can be summarized in the following equation:
Total NOx Emissions = • • ni × (NOx/mile)i × VMTi, (2)
where i denotes vehicle class.
EPA’s estimates of the benefits of the Tier 2 emissions standards in 2030 require making assumptions about the relevant categories of vehicles to analyze in 2030 as well as assumptions about each of the three components of Equation 2. It is, however, extremely difficult to understand the key assumptions made about these components or the predictions made for each component at the national level with and without the Tier 2 regulations.
For a reasonable number of classes of vehicles, EPA should present a table showing predicted values of the number of vehicles, emissions per mile for each criteria pollutant, and average VMTs for conditions with and without regulatory action in 2030 at the national level. To put those figures in perspective, a similar table should be constructed showing the values of these variables in the recent past.
Presenting those figures is not sufficient explanation of conditions with and without the Tier 2 emissions standards. The figures should be accompanied in the main text by some explanation of the assumptions that drive the results. For example, if analysts predict a rapid increase in the percent of light-duty trucks powered by diesel, this assumption requires an explanation, especially if it accounts for a large percent of the PM10 emissions in the regulatory baseline and, thus, a large percent of the particulate reductions attributed to the Tier 2 emissions standards.2
As in the case of power-plant emissions, the purpose of describing the various components of total vehicle emissions is to focus attention on the components that have a large impact on emissions with and without regulatory action and on the change in emissions associated with the proposed emissions standards. This information should guide the assessment of uncertainty in the emissions estimates and allow an examination of the possible distribution of values that key components might assume. See Chapter 5 for a description of the procedures for formalizing the uncertainty associated with emissions estimates and other components of the health benefits analysis.
Predictions Regarding Population and Health
The goal of a health benefits analysis associated with a proposed air pollution regulation is to estimate the avoidable risk associated with that regulation—cases of morbidity and mortality that are likely to be avoided if the regulation is implemented. The standard approach to computing avoided cases of morbidity and mortality (assuming a linear concentration-response function) is to multiply the size of the exposed population (Pop) by the baseline incidence of the health effect in question (Yb) in the year, such as 2030, in which benefits are to be evaluated. This calculation yields the predicted baseline number of cases in 2030. The reduction in cases is estimated by multiplying baseline cases by the slope of a concentration-response function (β) that describes the percent reduction in cases per unit of pollutant and by the reduction in ambient pollution associated with the regulation (ΔC). The overall calculation can be approximated by the following equation:
Cases Avoided = β × ΔC × Yb × Pop. (3)
Calculation of avoided cases thus requires estimates of population and baseline disease rates (or death rates) for the years in which benefits are to be evaluated. These estimates are required at the level of geographic disaggregation used in modeling the air quality.
EPA is generally clear about how it projects future population and incidence rates. For the HD engine and diesel-fuel rule, EPA clearly stated that population projections come from the U.S. Census Bureau (EPA 1999c). The methods used to interpolate the population projections for the year of the health benefits analysis (2030) are clearly explained, as are the
methods used to associate county-level data with the grid cells used in the air-quality modeling.
The methods used to estimate incidence for various health outcomes are described in Appendixes B and C of the same document. In many cases, incidence in 2030 is assumed to be identical to that in the late 1990s. For example, the annual county mortality rates from 1994 to 1996 are used to estimate nontrauma mortality rates in 2030. For hospital admissions (by International Classification of Disease code), national incidence in 2030 is assumed to be equal to that in 1994, the most recent year available at the time of the study. For health outcomes that lack national incidence data, incidences are assumed equal to those in the epidemiological studies used to compute the number of avoided cases.
Predicting baseline morbidity and mortality rates 30 years into the future is difficult, because there is much evidence that rates can change significantly over such periods. For example, rates of heart disease, one of the major disease categories affected by ambient air pollution, have been remarkably reduced in the past 30 years. Although it is probably not feasible to project baseline rates for all health outcomes considered in health benefits analyses, EPA should incorporate estimates of future trends in mortality and morbidity for major health outcomes, such as those that make up two-thirds of total deaths or lost life-years, that are being considered. At the least, EPA’s estimates of avoided cases should reflect the uncertainty in these rates. For some outcomes with available data, this uncertainty can be reduced by disaggregating baseline rates and applying them by age groups, because future shifts in age distribution are less uncertain and are projected routinely by widely accepted sources. This approach should be followed whenever possible.
Another source of uncertainty in estimating avoided cases derives from the distinction between attributable and avoidable risk. The β coefficients in Equation 3 come from studies that relate variation in health impacts to variation in air pollution concentrations based on historical data. The result is a measure of the risk attributable to air pollution in the past. Characteristics of the study population that are not explicitly controlled for in the concentration-response function are implicitly reflected in the β coefficients. The extent to which future populations differ from those in the studies will add to the degree of uncertainty associated with estimating the avoided cases.
To illustrate, all the analyses examined by the committee rely on the American Cancer Society (ACS) study (Pope et al. 1995; Krewski et al.
2000) to estimate the impact of changes in PM concentrations on mortality. The estimate of the impact of fine-particle exposure on the nontrauma death rate (with relative risk assumed the same for all age groups) is used to predict avoided cases of mortality. There are many sources of error in applying this coefficient to populations in the year 2030. One error in applying the ACS study occurs because the impact of PM exposure on nontrauma deaths is actually an average of its impact on various causes of death, such as coronary artery disease and lung cancer. To the extent that the distribution of deaths by cause in the U.S. population in 2030 differs from that in the ACS study population, errors will result. Another source of error occurs because the age distribution of the ACS study population may differ from the age distribution of the population in 2030.
To incorporate these considerations into the computation of avoided cases of morbidity and mortality, the predicted characteristics of the population in 2030 must be compared with the characteristics of the populations in the epidemiological studies used to compute avoided cases. Appropriate adjustments should be made if differences are found.
The estimation of health benefits that will result from reducing air pollution depends critically on decisions made at the beginning of the analysis: (1) the regulatory options to consider, (2) the health effects to evaluate, (3) the time frame for the analysis, including the years in which benefits are evaluated, and (4) the assumptions to make about future conditions with and without implementation of the regulation.
A critical step in the preliminary stages of an RIA is the development of a range of regulatory options to evaluate. Fewer regulatory alternatives than would be needed to follow OMB guidelines are presented or appear to be evaluated in recent EPA analyses. The regulatory options should represent the range of choices available.
EPA typically evaluates the costs of the regulatory options examined from the time the regulations are first introduced until they have been fully implemented. By contrast, the benefits of the regulations are often examined for only a single year, usually the year in which the policy will be fully implemented. The comparison of benefits and costs focuses on this one future year rather than comparing the benefits and costs over the period of implementation.
The high cost of air-quality modeling is cited as a major reason for limiting the years in which benefits are evaluated and also as a reason for not calculating the costs and benefits of more regulatory options.
Predictions about emissions with and without the regulations are treated as certain and are presented in terms of total emissions by sector. The components of emissions, such as number of vehicles in a class, average miles traveled per vehicle, and emissions per mile, are seldom presented, and predicted emissions are seldom compared with historical trends to place them in perspective.
Predictions about future population trends and the baseline health of the population are more clearly stated than those for emissions; however, these predictions are treated as certain, even when predictions are made far into the future.
To the extent possible, EPA should estimate the benefits for several regulatory options that represent the full range of choices available to the decision-maker. The regulatory options should include graded levels of stringency requirements and the time schedule for achieving reductions in emissions or exposures. If options are eliminated at an earlier stage, the rationale for doing so should be provided.
EPA should estimate the benefits over the regulatory time period including both the implementation period and the expression period of all important health effects. Because calculating benefits for every future year is resource-intensive and unlikely to show true increases in precision, calculations can be made, for example, every fifth year with simple interpolation techniques applied to estimate benefits for intervening years.
EPA should modify the air-quality models used in translating predicted emissions into predicted levels of ambient air quality to reduce resources required for air-quality modeling. This change is necessary if EPA is to evaluate multiple regulatory alternatives and to evaluate each alternative at reasonable time intervals, such as every 5 years. Evaluation of the ambient air quality associated with more emissions scenarios is also essential if the uncertainty inherent in emissions estimates is to be carried through to the estimation of avoided cases of mortality and morbidity.
The components of emissions estimates (such as number of vehicles in a class, average miles traveled per vehicle, and emissions per mile)
should be presented with and without implementation of the regulation at the national level. This will help readers judge how reasonable these predictions are and will suggest which components of emissions estimates drive the emissions reductions associated with the regulation. Historical trends in these components should also be presented.
The uncertainty in emissions estimates should be quantified and carried through the health benefits analysis to the calculation of avoided cases of mortality and morbidity.
EPA should incorporate estimates of future trends in background mortality and morbidity for the major health outcomes, such as those that make up two-thirds of total deaths or lost life-years, that are under consideration.
EPA should quantify uncertainties with regard to future population distributions and background disease rates. EPA should also summarize what is known about the potential importance of disease interactions and competing risks affecting the health outcomes of primary interest and discuss the possible biases that might be introduced in the final analysis by changes in those factors.
Because a regulation to improve air quality may affect pathways other than air, EPA should determine whether there are likely to be any important indirect impacts of a regulation on human health and the environment. If any such impacts are identified, EPA should include in the analysis a plan to assess them more completely.
EPA (U.S. Environmental Protection Agency). 1997. Regulatory Impact Analyses for the Particulate Matter and Ozone. National Ambient Air Quality Standards (NAAQS) and Proposed Regional Haze Rule. Regulatory Economic Analysis Inventory. A.97.9. Office of Air Quality Planning and Standards, Office of Air and Radiation, U.S. Environmental Protection Agency, Research Triangle Park, NC.
EPA (U.S. Environmental Protection Agency). 1999a. Regulatory Impact Analysis—Control of Air Pollution from New Motor Vehicles: Tier 2 Motor Vehicle Emissions Standards and Gasoline Sulfur Control Requirements. EPA 420-R-99-023. Engine Program and Compliance Division, Office of Mobile Sources, Office of Air and Radiation, U.S. Environmental Protection Agency. December 1999. [Online]. Available: http://www.epa.gov/OMS/regs/ld-hwy/tier-2/frm/ria/r99023.pdf [accessed September 10, 2002].
EPA (U.S. Environmental Protection Agency). 1999b. Final Report to Congress on Benefits and Costs of the Clean Air Act, 1990 to 2010. EPA 410-R-99-001. Office of Air and Radiation, U.S. Environmental Protection Agency. November 1999.
EPA (U.S. Environmental Protection Agency). 1999c. Final Tier 2 Rule: Air Quality Estimation, Selected Health and Welfare Benefits Methods, and Benefit Analysis Results. EPA 420-R-99-032. Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC. December 1999. [Online] Available: http://www.epa.gov/otaq/regs/ld-hwy/tier-2/frm/tsd/r99032.pdf [accessed September 10, 2002].
EPA (U.S. Environmental Protection Agency). 2000. Regulatory Impact Analysis: Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control Requirements. EPA 420-R-00-026. Office of Air and Radiation, U.S. Environmental Protection Agency, Washington, DC. December 2000.
EPA (U.S. Environmental Protection Agency). 2002a. Integrated Planning Model (IPM). Clean Air Power Initiative (CAPI), U.S. Environmental Protection Agency. [Online]. Available: http://www.epa.gov/capi/ [accessed September 10, 2002].
EPA (U.S. Environmental Protection Agency). 2002b. MOBILE Model. Office of Transportation and Air Quality, U.S. Environmental Protection Agency. [Online]. Available: http://www.epa.gov/otaq/mobile.htm [accessed September 10, 2002].
Krewski,D., R.T.Burnett, M.S.Goldberg, K.Hoover, J.Siemiatycki, M.Jerrett, M. Abrahamowicz, and W.H. White. 2000. Reanalysis of the Harvard Six Cities Study and the American Cancer Society Study of Particulate Air Pollution and Mortality, A Special Report of the Institute’s Particle Epidemiology Reanalysis Project. Final Version. Health Effects Institute, Cambridge, MA. July 2000. [Online]. Available: http://www.healtheffects.org/pubs-special.htm [accessed September 10, 2002].
OMB (Office of Management and Budget). 1996. Economic Analysis of Federal Regulations Under Executive Order 12866. Office of Management and Budget, The White House. January 11, 1996. [Online]. Available: http://www.whitehouse.gov/omb/inforeg/riaguide.html [accessed September 10, 2002].
OMB (Office of Management and Budget). 2000. Guidelines to Standardize Measures of Costs and Benefits and the Format of Accounting Statements. Memorandum from Jacob J. Lew, Director, Office of Management and Budget, The White House, for the Heads of Departments and Agencies. M-00-08. March 22, 2000. [Online]. Available: http://www.whitehouse.gov/omb/memoranda/ [accessed September 10, 2002].
Pope, C.A. III, M.J. Thun, M.M. Namboodiri, D.W. Dockery, J.S. Evans, F.E. Speizer, and C.W. Heath. 1995. Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am. J. Respir. Crit. Care Med. 151(3 Pt 1): 669-674.