Industrial Combustion Coordinated Rulemaking (ICCR) Framework for Economic and Benefits Analysis Prepared by the Economic Analysis Work Group as Part of the ICCR FACA Process June 1998 ------- Industrial Combustion Coordinated Rulemaking (ICCR) Framework for Economic and Benefits Analysis June 1998 Prepared by the Economic Analysis Work Group as Part of the ICCR FACA Process ------- PREFACE The U.S. Environmental Protection Agency (EPA) has instituted a FACA process to coordinate rulemaking for industrial-commercial-institutional (ICI) combustion sources. As part of the Industrial Combustion Coordinated Rulemaking (ICCR), the Economic Analysis Work Group was formed to provide economic analytical support for the FACA process. It is responsible for several analysis areas within the ICCR FACA process, including economic impact analysis (market analysis), regulatory impact analysis (cost-benefit analysis), Small Business Regulatory Enforcement Fairness Act (SBREFA) (small business analysis), unfunded mandates analysis, environmental justice, and children's health analysis. These analyses can be an important input into the decisionmaking process and are also required by legislation or executive order. The Economic Analysis Work Group is working closely with the five Source Work Groups to develop the cost and emissions reductions estimates that will serve as the inputs to the economic analysis. One objective of this close coordination is that it will ensure that economic data from other sources can be properly linked to the cost and emissions data. Because of the time constraints associated with the ICCR process, the methodology for the economic and benefits analysis is being developed in parallel with the preliminary estimates of costs and emissions reductions. Preliminary costs and emissions reductions are typically used as inputs to determine the scope of the modeling approach and the industries to be targeted for detailed analysis. Thus, the modeling approach outlined in this document may be modified after reviewing the final cost and emissions reduction estimates. Table 1 presents an overview of the Economic Analysis Work Group's schedule and indicates completed activities and future deliverables. The table does not include completion dates for the economic analysis because they are contingent on receiving cost and emissions data from the Source Work Groups. As specified in Table 4 of the ICCR Organizational Structure and Process document (ICCR, 1997), 6 months have been allocated to conduct the core economic analysis once the Economic Analysis Work Group receives all cost and emissions data. However, because the Source Work Groups may evolve onto different time schedules, the economics and benefits analysis may be conducted in stages as information becomes available from individual Source Work Groups. in ------- Table 1. Economic Analysis Work Group Activities and Deliverables Description of Activity or Deliverable Timeline Representatives from Econ WG meet with Source WGs to Completed discuss data requests for economic and benefits analysis Econ WG meets with Source WG Subgroups Ongoing Econ WG posts methodology document on TTN June 1998 Econ WG presents analysis plan at the CC Meetings July 1998 Econ WG receives final data from Source WG to support Time la economic and benefits analysis Econ WG performs overall economic impact and benefits Six Months Beginning at analysis, considering interactions among source categories Time 1 Econ WG presents preliminary results of economic and Six Months after Time 1 benefits analysis to the CC and Source Work Groups a The Source Work Groups may evolve onto different time schedules. Thus, "Time 1" may be different for each Source Work Group, and the economic and benefit analysis may be conducted in stages (see Section 2.2 for a discussion of the incremental and cumulative analysis associated with individual Source Work Groups). At the July Coordinating Committee meetings in Long Beach, CA, the Economic Analysis Work Group will present an overview of the economic and benefits methodology to the Coordinating Committee and will address any questions or comments the committee may have after reviewing this draft methodology document. iv ------- CONTENTS Section Page 1 Introduction 1-1 1.1 Required Economic and Benefits Analysis for a Significant Regulatory Action 1-1 1.2 Modeling Results Typically Used to Support Required Economic and Benefits Analysis 1-1 1.3 Overview of Typical Economic and Benefits Analysis in Regulatory Development 1-5 2 Scope of Regulation and Information Sources for the ICCR 2-1 2.1 Affected Universe of Sources 2-1 2.2 Pollutants to be Considered 2-2 2.3 Modeling Issues Specific to the ICCR Process 2-2 2.4 Information Sources 2-4 2.5 Linking Data from Source Work Groups 2-4 2.6 Common Analysis Parameters 2-7 3 Economic Analysis Methodology Plan 3-1 3.1 Background on Economic Modeling Approaches 3-1 3.1.1 Modeling Dimension 1: Scope of Economic Decisionmaking 3-2 3.1.2 Modeling Dimension 2: Interaction Between Economic Sectors 3-3 3.1.3 Recommended Approach for Analysis of ICCR Process ... . 3-6 v ------- CONTENTS (CONTINUED) Section Page 3.2 Industry Profiles 3-7 3.3 Modeling Market Components 3-9 3.3.1 Production Process 3-9 3.3.2 Downstream Product Markets 3-11 3.3.3 Upstream Input Markets 3-14 3.4 Operationalizing the Economic Impact Model 3-15 3.4.1 Computer Model 3-17 3.4.2 Aggregating Social Costs 3-21 3.5 Screening Procedures for Determining Scope of Analysis 3-21 4 Benefits Analysis Methodology Plan 4-1 4.1 Theoretical Basis for Benefits Analysis 4-1 4.2 Analysis Framework 4-3 4.2.1 Identify the Primary Pollutants of Concern and Characterize Their Effects 4-8 4.2.2 Select and Characterize Emissions Sources for Emissions, Air Quality, and Exposure Modeling 4-9 4.2.3 Model Emissions for Selected Pollutants 4-10 4.2.4 Select and Model Primary Exposure Pathways 4-10 4.2.5 Estimate Human Health Risks Through the Modeled Exposure Pathways 4-14 4.2.6 Estimate Monetary Benefits 4-18 4.2.7 Characterize and Summarize Benefits 4-20 References R-l vi ------- LIST OF FIGURES Number Page 1-1 Overview of Economic and Benefits Analysis 1-6 2-1 Hypothetical Source Work Group Time Schedules 2-3 2-2 Combine Cost and Emissions Estimates from All Source Work Groups 2-6 3-1 Economic Impact 3-8 3-2 Focus Industry A 3-10 3-3 Market Effects of Regulation-Induced Costs 3-13 3-4 Fuel Market Interactions with Facility-Level Production Decisions 3-14 3-5 Operationalizing the Estimation of Economic Impact 3-16 3-6 Percentage of Establishments by Level of Price Difference Between Residual Fuel Oil and Less Expensive Natural Gas that Would Switch Fuels 3-20 4-1 Conceptual Framework Linking Pollutant Releases to Human Welfare 4-2 4-2 Identification of Damage Pathways for Major Air Pollutants 4-4 4-3 General Procedural Framework for the Quantification and Valuation of Emissions Reductions 4-6 4-4 Physical and Biological Routes of Exposure 4-11 vii ------- LIST OF TABLES Number Page 1-1 Statutes and Their Analysis for a Significant Regulatory Action 1-2 1-2 Executive Orders and Their Analysis Requirements 1-3 1-3 Matrix Linking Statutes and EOs with Typical Impact Metrics 1-4 2-1 Minimal Inputs Required by the Economic Analysis Work Group for Each Subcategory from Source Work Groups 2-4 2-2 Example Format for ICCR Combustion Turbine Model Sources 2-5 3-1 Comparison of Modeling Approaches 3-3 3-2 Market Model Results using Behavioral Models 3-4 3-3 Capacity to Switch from Coal to Alternative Energy Sources by Industry Group, Selected Industries, and Selected Characteristics, 1994 3-19 3-4 Fuel Price Elasticities 3-21 3-5 Total Inputs of Energy for Heat, Power, and Electricity Generation by Industry Group and Selected Industries, 1994 3-23 4-1 Classification of Valuation Methods 4-7 4-2 Baseline Estimated Cancer Risks from Direct Exposure (Inhalation) to HAP Emissions from Electric Utility Steam Generating Units: Local and Long-Range Impacts in 1994 4-17 4-3 Per-Ton Benefit Estimates for Selected Criteria Pollutants (1990 $) 4-20 viii ------- SECTION 1 INTRODUCTION In this section we present a brief overview of the economic and benefits analysis required for a significant regulatory action. The typical modeling results needed to support the analysis are identified in Section 1.2, and an overview of the Economic Analysis Work Group's proposed modeling approach is presented in Section 1.3. 1.1 Required Economic and Benefits Analysis for a Significant Regulatory Action To protect the public's health and welfare the Clean Air Act as amended (CAAA) provides the U.S. Environmental Protection Agency (EPA) with the authority, among other things, to establish ambient air quality standards and to undertake actions designed to reduce the release of pollutants of anthropogenic origin to the atmosphere. The Agency's regulatory development program includes conducting assessments of health and ecological effects, exposure and risk, and economic impacts and benefits of Agency initiatives. Economic and benefits analysis can contribute to informed Agency decisionmaking under the CAAA. Table 1-1 and Table 1-2 list the primary statutes and executive orders (EO) that are relevant for the ICCR process. In addition, a summary of the analysis requirements for each statute and EO are provided. These analysis requirements provide the starting point for developing our economic and benefits analysis methodology. 1.2 Modeling Results Typically Used to Support Required Economic and Benefits Analysis The Clean Air Act, other statutes, and executive orders (EOs) do not typically specify specific methods or impact metrics to be used in analyzing economic impacts or social benefits. Table 1-3 links commonly used impact metrics with the statutes and EOs they support. Section 3 of this report presents the modeling approach for estimating these impact metrics. National-level compliance costs and benefits of emissions reductions provide the basis for cost-benefit analysis of proposed regulations. However, to meet all the 1-1 ------- Table 1-1. Statutes and Their Analysis for a Significant Regulatory Action Statutes Analysis Requirements Clean Air Act Impact analysis cost of compliance inflationary effects small business effects effects on consumers effects on energy users Cost-effectiveness analysis Determine if significant impact on substantial number of small entities. If so initial regulatory flexibility analysis (IRFA) Small Business Advocacy Review Panel Unfunded Mandates Reform Act Budgetary impact analysis for state, local, and tribal governments, including impacts on private sector consider reasonable range of alternative regulations show that adopted regulation is most cost- effective and least burdensome requirements of the statutes and EOs, analysis of proposed regulations needs to be conducted at various levels of aggregation. Market-level models are used to estimate the impact of the regulation on production levels and commodity prices. Changes in production that resulted from adding the control costs can affect the magnitude of the social costs, and changes in price help determine the distribution of social costs between consumers and producers. Market-level models are also used to assess the change in imports and exports associated with a regulation. Facility-level analysis, which is based on profit-maximizing behavior, can be used to evaluate the regulation's impact on plant closures. Facilities may cease to produce a particular product by closing a product line or stopping production altogether by closing the entire facility, which is determined by comparing total revenues and total costs at the facility level. Facility- and company-level impacts are typically analyzed using typical or model facilities. Thus, impact estimates are for representative facilities and not specific entities. Regulatory Flexibility Act (RFA) and SBREFA 1-2 ------- Table 1-2. Executive Orders and Their Analysis Requirements Executive Orders Analysis Requirements EO 12866: Regulatory Planning and Review EO 12875: Enhancing the Intergovernmental Partnership EO 12898: Environmental Justice EO 13045: Children's Healthฎ Assess costs to determine if regulation is "significant" (exceeds $100 million) If significant, assess benefits Consider/evaluate regulatory alternatives Assess impact on state, local, and tribal governments (no threshold, does not include impacts on private sector) Analyze distribution of costs and benefits with respect to minority populations low-income populations Determine if regulatory action is likely to be economically significant and to disproportionately affect children. If so evaluate health or safety effects on children explain why regulation is preferable to others The applicability of EO 13045 to the ICCR is currently being discussed. The benefit methodology is designed to meet any eventual analysis requirements in the area of children's health for the ICCR process. Company-level effects are computed by identifying the ownership of facilities and aggregating the financial effects up to the company level. To satisfy the conditions of the RFA and SBREFA, economic impact analyses include an assessment of small company impacts. Several approaches are available to estimate changes in the financial status of firms affected by regulatory alternatives. The commonly used approach uses ratios computed with financial data. Community-level impacts are used to estimate changes in employment and changes in tax revenues attributable to the proposed regulation. The implementation of the regulation and the associated resource reallocations may change government revenues and costs at the federal, state, and local levels. The costs of program administration may be based on engineering cost analyses or developed by analogy with similar government programs. Changes in tax revenues that may derive from these impacts can also be calculated. 1-3 ------- Table 1-3. Matrix Linking Statutes and Executive Orders with Typical Impact Metrics Impact Metrics Change Compli- in Change in Change in ance Costs Emissions Benefits of Produc- Change Imports and Employ- Emissions Cost/ Change in (exceed Reduc- Emissions tion in Price Exports ment Closures Cost/Sales (regional Benefits Revenue Tax Receipts $100 tions Reduc- (market (market (market (facility (facility (company and source (regional (community (community Requirements million) (tons) tions ($) level) level) level) level) level) level) level) level) level) level) Clean Air Act X X X X X X X X X economic impact analysis (market analysis) EO 12866: XXX Regulatory Planning and Review regulatory impact analysis (cost- benefit analysis) Regulatory XX XX Flexibility Act (RFA) and SBREFA Unfunded Mandates Reform Act X X EO 12898: Environmental Justice X X X X X X X EO 13045: Children's Health1 X X X X X EO 12875: Enhancing the Intergovernmental Partnership X X a The applicability of EO 13045 to the ICCR is currently being discussed. The benefit methodology is designed to meet any eventual analysis requirements in the area of children's health for the ICCR process. ------- 1.3 Overview of Proposed Economic and Benefits Analysis for Regulatory Development Figure 1-1 presents an overview our proposed economic and benefits analysis for the ICCR regulatory development. Because our modeling approach is still evolving as information and data become available, not all the activities represented in Figure 1-1 may be included in our final analysis. However, this figure does reflect the fundamentals of our approach and illustrates basic relationships between input data, analysis, and modeling results. The approach illustrated in this figure is based on established microeconomic theory and incorporates behavioral market models for the economic analysis. In addition, the analysis is conducted at the facility level and results aggregated to obtain national impacts. This approach provides modeling results that support the distributional impact and benefits analysis required by the statutes and EOs presented in Table 1-3. The majority of the Economic Analysis Work Group's efforts will be focused on developing the industry profiles, the economic impact analysis, and the benefits analysis. Industry profiles provide information on the affected entities. Data are typically obtained from a combination of sources, including information gathered by EPA and summarized in previous regulatory support documents; data from publicly available sources; and data from stakeholders, including any affected trade associations. The information is used to identify affected commodities; characterize baseline conditions in affected markets, including prices and quantities, market structure, and international trade; identify and locate producers and consumers of affected commodities; identify and characterize the firms owning the affected facilities; and characterize baseline conditions in the communities where affected producers are located (the populations who will affected by the regulation). Economic impact analysis analyzes the economic impacts of the regulation. The industry profile provides the baseline characterization of market conditions from which the impacts of the regulation are estimated. Frequently an analysis involves using an analytical computer model to simulate the responses of the affected entities to the regulation. The model is designed to be consistent with economic theory and to produce integrated estimates of the responses of affected facilities, firms, and markets. Included are impact estimates for 1-5 ------- Figure 1-1. Overview of Economic and Benefits Analysis 1-6 ------- selected location and company size categories to support impact analysis for small businesses, small communities, and minority and low-income populations. In addition, multiple model specifications can be used to generate policy-making information to support evaluation of different regulatory alternatives. Benefits analysis involves analyzing all the categories of benefits, by first identifying, then quantifying and where possible monetizing, the benefits. The benefits of pollution controls are defined as the increases in human welfare that result from improvements in environmental quality. Assessing the benefits of air pollution controls requires a conceptual framework that specifically links reductions in air pollutant emissions to human welfare enhancements. Such a framework distinguishes the essential components of benefits and ensures that all relevant benefits are accounted for and not double counted. 1-7 ------- SECTION 2 SCOPE OF REGULATION AND INFORMATION SOURCES FOR THE ICCR This section identifies the combustion source categories and potential pollutants to be included in the economic and benefits analysis and discusses modeling and data issues specific to the ICCR. 2.1 Affected Universe of Sources Seven categories of combustion sources are listed for regulatory development under Section 112 (National Emission Standard for Hazardous Air Pollutants) or Section 129 (solid waste combustion) of the Clean Air Act. In addition, existing Section 111 (New Source Performance Standards [NSPS]) regulations affecting some of these source categories are periodically reviewed and revised. The Clean Air Act requires regulations for all of the categories listed below to be promulgated under Sections 112 and/or 129 by November 2000.1 industrial boilers (Sections 112 and 111) commercial-institutional boilers (Sections 112 and 111) process heaters (Sections 112 and 111) industrial-commercial solid waste incinerators (Sections 129 and 111) other solid waste incinerators (Sections 129 and 111) stationary combustion turbines (Sections 112 and 111) stationary internal combustion (Sections 112 and 111) The ICCR has established five Source Work Groups to analyze the above source categories: Boilers Work Group 'See pages 2 through 5 of the ICCR Organizational Structure and Process (1997) for a more detailed discussion of regulatory background. 2-1 ------- Process Heaters Work Group Internal Combustion Engines Work Group Incinerator Work Group Stationary Turbines Work Group. The coordination of these rulemakings will result in more consistent regulations with potentially greater environmental benefits at a lower cost than piecemeal regulations. Not included in the scope of the ICCR analysis are combustion units associated with public utilities and municipal waste combustors (MWCs). These combustion sources are covered under separate regulations. 2.2 Pollutants to be Considered Each Source Work Group is evaluating the magnitude of emissions and other factors to focus their regulatory effort on the most significant pollutants and environmental issues related to their specific sources. Different lists of "pollutants of concern" may be developed separately by each Source Work Group. A preliminary list of pollutants to be considered for regulation as part of the ICCR are hazardous air pollutants (HAPs) listed in Section 112; criteria pollutants regulated under Section 111 NSPS (e.g., S02, NOx, and PM); and pollutants listed in Section 129 (i.e., total and fine PM, opacity, S02, NOx, HC1, CO, lead, cadmium, mercury, and dioxins and furans). 2.3 Modeling Issues Specific to the ICCR Process Each of the five Source Work Groups is developing cost and emissions information independently. Because the Source Work Groups may evolve onto different time schedules as illustrated in Figure 2-1, both incremental and cumulative impacts may be estimated on a flow basis (i.e., in stages as cost and benefits information becomes available). If the economic and benefits analysis is conducted on a flow basis, as information is received from each Source Work Group, we will conduct both incremental and cumulative economic and benefits analyses for each Source Work Group. Incremental impact analysis will be used primarily to assess the efficiency of regulatory alternatives (i.e., emissions 2-2 ------- Time 2 Source WG 1 Source WG 2 Source WG Source WG 4j Source WG 5 Economic Analysis Promulgation -k. ฆ * 6 months Figure 2-1. Hypothetical Source Work Group Time Schedules reductions per control costslbs/$). In contrast, cumulative impact analysis will be more appropriate to assess impacts per facility (such as cost/sales for SBREFA) or when the benefits are not linearly related to emissions (such as for chemical reactions in the atmosphere for the creation of ozone or for nonlinear dose-response functions to assess mortality rates). The Economic Analysis Work Group will analyze small business impacts using the control cost and distributional information provided by the Source Work Groups. The small business impact assessment will be complicated by the fact that the facility-level impacts are being developed separately by the five Source Work Groups. Thus, to determine the full impact on small entities all cost and distributional information from the Source Work Groups must be merged together. For example, for a given entity control costs associated with boilers alone may not be significant compared to the company's sales. However, when costs associated with all five source categories are summed together, total control costs may exceed the cost-to-sales ratio threshold used to identify significant impacts. As a result, if the Source 2-3 ------- Work Groups evolve onto different time schedules, the small business analysis may be incomplete until all Source Work Groups complete their analysis. 2.4 Information Sources The Source Work Groups will supply compliance costs, baseline emissions, changes in emissions, and the distribution of costs and emissions across sources. Table 2-1 lists the minimal inputs required for each subcategory for the economic analysis. Table 2-1. Minimal Inputs Required by the Economic Analysis Work Group for Each Subcategory from Source Work Groups Facility and combustion unit IDs Baseline emissions without a new regulation in place For each control alternative - incremental capital costs - incremental operating/maintenance cost - emission reductions An estimate of the total population of units in the subcategory A statement about the representativeness of the subcategory database compared to the national population of units Note: Baseline emissions, emission reductions, and control system costs can be provided on either an actual unit or model unit basis. In addition, as part of the industry profiles the Economic Analysis Work Group will collect information on and prepare profiles of significantly affected industries. Potential data sources include information gathered by EPA and summarized in regulatory support documents, data available from publicly available sources, and data from stakeholder groups. Currently, the American Furniture Manufacturers Association (AFMA) has volunteered to provide supporting information for the economic analysis. And information from additional affected trade associations will be incorporated into the analysis if it becomes available. 2.5 Linking Data from Source Work Groups The Source Work Groups are developing cost and emissions impacts on a "model source" basis. Table 2-2 presents a draft list of model sources for the Combustion Turbines Work Group. Our plan is to link the model sources from the Source Work Groups to the 2-4 ------- Table 2-2. Example Format for ICCR Combustion Turbine Model Sources Surrogate Model Operating Heat Existing Clean Output Plant Unit Hours Per Recovery Application Fuel MW Ex. Flow Number Size Year (Y/N) (Y/N) (Y/N) Typical Applications Turbine (ISO) (lb/sec) 1 Large 8,000 Y Y Y Existing utility/IPP generating station GE PG 7121 EA 85.40 658.0 1A Large 8,000 Y Y N Landfill operation or residual oil GE PG 7121 EA 85.40 658.0 2 Large 8,000 Y N Y New utility/IPP generating station GE PG 7231 FA 170.00 986.0 2A Large 8,000 Y N N Landfill operation or residual oil GE PG 7231 FA 170.00 986.0 3 Large 8,000 N Y Y Existing utility/IPP generating station GE PG 7231 FA 170.00 986.0 3A Large 8,000 N Y Y Existing utility/IPP station (space constrained) GE PG 7231 FA 170.00 986.0 4 Large 8,000 N N Y New utility/IPP generating station GE PG 7231 FA 170.00 986.0 5 Large 500 N Y Y Existing utility/IPP peaking unit GE PG 7121 EA 85.40 658.0 6 Large 500 N N Y New utility/IPP peaking unit GE PG 7121 EA 85.40 658.0 7 Medium 8,000 Y Y Y Existing industrial site power production GE PG 6561B 39.60 318.0 7A Medium 8,000 Y Y N Landfill operation or residual oil GE PG 6561B 39.60 318.0 8 Medium 8,000 Y N Y New industrial site power production GE PG 6561B 39.60 318.0 8A Medium 8,000 Y N N Landfill operation or residual oil GE PG 6561B 39.60 318.0 9 Medium 8,000 N Y Y Existing pipeline compressor/ind. site-mech. drive GE 5352B 26.00 270.0 10 Medium 8,000 N N Y New pipeline compressor/ind. site-mech. drive GE 5352B 26.00 270.0 11 Medium 500 N Y Y Existing utility/IPP peaking unit GE PG 6561B 39.60 318.0 12 Medium 500 N N Y New utility/IPP peaking unit GE PG 6561B 39.60 318.0 13 Small 8,000 Y Y Y Existing industrial process plant (food, natural gas) Solar Centaur 40 3.50 41.0 13A Small 8,000 Y Y N Landfill operation or residual oil Solar Centaur 40 3.50 41.0 14 Small 8,000 Y N Y New industrial process plant (food, natural gas) Solar Centaur 40 3.50 41.0 14A Small 8,000 Y N N Landfill operation or residual oil Solar Centaur 40 3.50 41.0 15 Small 8,000 N Y Y Existing pipeline compressor Solar Centaur 40 3.50 41.0 15A Small 8,000 N Y Y Offshore platform (physical space constrained) Solar Centaur 40 3.50 41.0 16 Small 8,000 N N Y New pipeline compressor/offshore platform Solar Centaur 40 3.50 41.0 17 Small 200 N Y Y Existing standby power (hospital, university, etc.) Solar Saturn T1500 1.13 14.2 18 Small 200 N N Y New standby power (hospital, university, etc.) Solar Saturn T1500 1.13 14.2 Note: This is a preliminary draft list of model sources supplied to the Economic Analysis Work Group on February 4, 1998. These model sources are subject to change. ------- ICCR Inventory Database. Linking to the Inventory Database will enable us to incorporate the distributional information contained in the database into the analysis. Figure 2-2 illustrates how model sources from different Source Work Groups can be potentially linked to the ICCR Inventory Database. Key parameters that will be used to facilitate the linking are unit capacity, fuel type, operating hours, and source-specific characteristics such as combined-cycle heat recovery for turbines. Engineering Analysis Boiler Source A Boiler SourceB Boiler Sourcex Process Heater Source A Process Heater SourceB Process Heater Sourcex Population Database Com buster ID 005 006 020 021 002 008 009 025 Facility ID SIC Code 011030009 2861 010890104 2911 Economic & Benefits Analysis Cost/Sales Ratio Impacts on Small Businesses Geographic Distribution of Emissions Figure 2-2. Combine Cost and Emissions Estimates from All Source Work Groups Our review of the ICCR Inventory Database indicates that most of the fields in the database have missing values. As part of their analysis, the Source Work Groups are currently "filling in" some of the missing information. For example, based on unit model numbers it is sometimes possible to identify a unit's capacity and/or fuel type. Or, for turbines, operating hours (for which there is typically the most complete information) can be used to determine if a unit is a peaking unit or a stand-by unit, and this information is a good indicator of whether the unit will be a single cycle or a combined cycle (i.e., stand-by units are rarely combined cycle). 2-6 ------- However, we anticipate that a substantial number of units will still have missing data after the Source Work Groups augment the Inventory Database, because in many instances model numbers and capacity are missing and there is simply not enough information on the unit to identify its characteristics. In these instances, we will ask the Source Work Groups to assess if the available information contained in the ICCR database is representative of the actual population of units. Specifically, we need to know, based on the expert judgment from the Source Work Groups, if any fuel types are over- or under-represented in the Inventory Database and if the distribution of unit capacity is representative. In particular, we are interested in the representativeness of smaller units in the Inventory Database, because they will be important in the small business analysis. 2.6 Common Analysis Parameters To support the integration of information from the Source Work Groups, a common baseline projection year and denomination for real dollars will be used for estimating compliance costs. Baseline year of analysis: 2005 Cost data in real dollars: $1998 In addition, common discount rate(s) will be used for estimating market costs, emission reduction benefits, and social benefits for impacts from all Source Work Groups. One commonly used discount rate is 7 percent. However, additional discount rates may be used for sensitivity analysis. 2-7 ------- SECTION 3 ECONOMIC ANALYSIS METHODOLOGY PLAN The economic analysis will include assessing the impact of the proposed regulation at the national, market, company, facility, and community levels. Flexibility will be an important component of our modeling approach. Because of the time constraints associated with the ICCR process, the methodology for the economic and benefits analysis is being developed in parallel with the preliminary estimates of cost and emissions reductions. Ideally the modeling approach would be developed incorporating information on preliminary cost and emissions reductions. Preliminary cost and emissions reduction estimates are typically used as inputs to determine the scope of the modeling approach and the industries to be targeted for detailed analysis. Thus, a flexible modeling approach is particularly important for the analysis to support the ICCR process. This section begins with background information on typical modeling approaches and presents details on the Economic Analysis Work Group's proposed economic impact modeling approach. Section 3.2 describes information to be developed as part of the industry profile. The theoretical structure for modeling market components and the procedure for operationalizing the model are presented in Sections 3.3 and 3.4, respectively. Finally, Section 3.5 outlines the screening processes we will use to select focus industries and to determine if upstream fuel or waste markets should be included in the analysis. 3.1 Background on Economic Modeling Approaches As discussed in Section 1, the economic impact analysis must satisfy, at a minimum, the requirements of the various statutes and executive orders listed in Tables 1-1 and 1-2, respectively. However, the scope of the economic impact analysis typically varies in response to the magnitude and distribution of impacts associated with the proposed regulatory alternatives and with the time and resources available for the analysis. Section 317 of the CAAA states that "the assessment required . . . shall be as extensive as practicable, . . . taking into account the time and resources available to the Environmental Protection Agency and other duties and authorities which [the Agency] is required to carry out." 3-1 ------- In general, the economic impact analysis methodology needs to allow EPA to consider the effect of the different regulatory alternatives. Several types of economic impact modeling approaches have been developed to support regulatory development. These approaches can be viewed as varying along two modeling dimensions: 1. the scope of economic decisionmaking accounted for in the model 2. the scope of interaction between different segments of the economy Each of these dimensions was considered in recommending our approach. The advantages and disadvantages of each are discussed below. 3.1.1 Modeling Dimension 1: Scope of Economic Decisionmaking Models incorporating different levels of economic decisionmaking can generally be categorized as with behavior responses and without behavior responses (accounting approach). Table 3-1 provides a brief comparison of the two approaches. The behavioral approach is grounded in economic theory related to producer and consumer behavior in response to changes in market conditions. In essence, this approach models the expected reallocation of society's resources in response to a regulation. This approach includes examining impacts at both the facility and market levels, as well as consumer impacts and overall changes in social welfare. Table 3-2 indicates the range of modeling results that can be developed based on behavioral response models. The changes in price and production from the market-level impacts are used to estimate the distribution of social costs between consumers and producers. The facility-, company-, and community-level impacts are used to assess the distribution of social benefits and costs that are required by many of the statutes and executive orders, such as the distribution of impacts across company size groups and ethnic and income groups. Facility- and company-level impacts refer to impacts on typical facility or company subgroups. We do not plan to model impacts on specific/actual entities. In contrast, the non-behavioral/accounting approach essentially holds fixed all interaction between facility production and market forces. As a result, a number of important phenomena in an economic impact analysis, such as price, market quantity, consumer, international trade, and net social welfare effects, cannot be adequately addressed using the nonbehavioral approach. These are all important elements of a conceptually sound economic impact analysis. Moreover, omitting these factors can lead to misleading conclusions about economic impacts. For instance, certain characteristics of demand conditions in an industry 3-2 ------- Table 3-1. Comparison of Modeling Approaches EIA With Behavioral Responses Incorporates control costs into production function Includes change in quantity produced Includes change in market price Estimates impacts for affected producers unaffected producers consumers foreign trade EIA Without Behavioral Responses Assumes firm absorbs all control costs Typically uses discounted cash flow analysis to evaluate burden of control costs Includes depreciation schedules and corporate tax implications Does not adjust for changes in market price * Does not adjust for changes in plant production may imply that consumers bear a large impact of the regulatory burden, thereby mitigating the impact on producers' profits and plant closures. This response can only be estimated if the regulation is modeled in a market context. 3.1.2 Modeling Dimension 2: Interaction Between Economic Sectors Because of the large number of markets potentially affected by the ICCR, an issue arises concerning the level of sectoral interaction to model. In the broadest sense, all markets are directly or indirectly linked in the economy, thus, all commodities and markets are to some extent affected by the regulation. For example, the ICCR may indirectly affect the market for potatoes, because the cost of steel incurred to produce farming equipment may increase with the regulation in effect. However, the impact of steel prices on the potato market is expected to be so small that it would be impossible to discern. On the other hand, 3-3 ------- Table 3-2. Market Model Results using Behavioral Models Market-level impacts for selected products Product price changes Production changes Consumption changes Changes in imports and exports Employment Typical facility-level impacts Costs (capital and annual operating) Closures Production Typical company-level impacts Changes in financial viability Financial failure Capital requirements and the cost of capital Community-level impacts Employment Tax receipts Social benefits and costs Environmental impacts Residuals releases by pollutant and medium * Environmental quality by media the impact on the market for steel may be significant and useful to explicitly incorporate into the model. Alternative approaches for modeling interactions between economic sectors can generally be divided in three groups: 3-4 ------- Partial equilibrium model: individual markets are modeled in isolation. General equilibrium model: all sectors of the economy are modeled together. Multiple-market partial equilibrium model: A subset of related markets are modeled together, with intersectoral linkages explicitly specified. Partial Equilibrium Model In a partial equilibrium approach individual markets are modeled in isolation. The only factor affecting the market is the cost of the regulation on facilities in the industry being modeled. Effects on other industries (such as the impact of the change in the price of steel on the cost of producing potatoes) are not included in the analysis because they are assumed to be negligible. Typically, the only factor affecting the market is the cost of the regulation on facilities in the industry being modeled. Under perfect competition, market prices and quantities are determined by the intersection of market supply and demand curves for the commodities. The market supply curve is the sum of all facility supply curves, and a market demand curve is the sum of the demand curves for all demanders of the commodity. Partial equilibrium models focus on intra- and interfirms effects. Control costs directly affect facilities' business decisions, such as facility production levels, inputs and raw material used in the production process, and a facility's decision to continue to operate or shut down. Individual facility-level responses to control costs are then aggregated to the market level to determine the market supply. General Equilibrium Model General equilibrium models are well suited to analyze large-scope environmental policies, such as the ICCR, because they capture welfare and employment effects across all sectors of the economy and specifically model interactions between economic sectors. General equilibrium models operationalize neoclassical microeconomic theory by modeling not only the direct effects of control costs, but also potential input substitution effects, changes in production levels associated with changes in market prices across all sectors, and the associated changes in welfare economywide. The interactions among the different sectors of the economy allow for the estimation of distribution effects between different production sectors and across different consumer groups. However, one of the major limitations to general equilibrium modeling is that markets are typically aggregated at a fairly high level to obtain a manageable number of production sectors for empirical modeling. There is a basic tradeoff in general equilibrium models 3-5 ------- between the institutional realism of the model and its mathematical tractability. Jorgenson and Wilcoxen (1990) have developed one of the larger general equilibrium models, which includes 35 production sectors. This tradeoff leads to a loss in the level of detail in the analysis, which is particularly troublesome when evaluating facility- or community-level impacts required by many of the statutes and executive orders. In addition, few general equilibrium models have the capability to evaluate the regional issues that are an integral part of air quality regulations. An additional disadvantage of general equilibrium modeling is that substantial time and resources are required to develop a new model or tailor an existing model for analyzing regulatory alternatives. Multiple-Market Partial Equilibrium Model To account for the relationships and links between different markets without employing a full general equilibrium model, analysts can use an integrated partial equilibrium model. In instances where separate markets are closely related and there are strong interconnections, there are significant advantages to estimating market adjustments in different markets simultaneously using an integrated market modeling approach. As an intermediate step between a simple, single-market partial equilibrium approach and a full general equilibrium approach, identifying and modeling the most significant subset of market interactions using an integrated partial equilibrium framework provide important information. In effect, the modeling technique is to link two or more standard partial equilibrium models by specifying the interactions between supply functions and then solving for all prices and quantities across all markets simultaneously. The number of linkages is limited only by the resources allocated to the modeling task. 3.1.3 Recommended Approach for Analysis of ICCR Process To conduct the analysis for the ICCR process, we propose to use a market modeling approach that incorporates behavioral responses modeled at the facility level. In addition, we will use a multiple-market partial equilibrium model as described above. Multiple-market partial equilibrium analysis provides a manageable approach to incorporate interactions between markets into the economic impact analysis to accurately estimate the impact of proposed regulations. 3-6 ------- Figure 3-1 presents an overview of the market linkages for our economic impact modeling approach. For illustrative purposes, the model is segmented into upstream markets, industry production processes, and downstream markets. The key intermarket linkages modeled will be the fuel and waste disposal markets' interactions with manufacturing industries significantly affected by the regulation. These linkages will be discussed in detail in Section 3.3. 3.2 Industry Profiles The first step in our modeling approach for the economic impact analysis (as shown in Figure 1-1) is to develop industry profiles. For a selected number of industries (referred to as focus industries), we will prepare a detailed industry profile.1 The industry profiles characterize the baseline against which the regulatory alternatives will be evaluated. The industry profiles will identify and characterize affected entities; define and characterize small entities to prepare to conduct analyses under RFA and SBREFA; define the products produced, including the production process, product attributes, and production methods and costs (supply determinants); identify product characteristics and uses, product users, and possible substitutes (demand determinants); summarize industry organization, including market integration, the structure of affected markets, financial position of firms, and employment; and describe product market characteristics including output prices, relevant price elasticities, domestic and foreign production levels, and domestic and foreign consumption levels. To provide the data used in the economic analysis, the industry profile will identify affected entities and identify the commodities for which markets will be analyzed. In analyzing the economic impacts, we will typically develop model facilities of different sizes or types, reflecting the distribution of affected entities. Because government agencies are likely to be affected by the regulation, the profile includes budget data on the affected 'Section 3.5 describes the screening process that will be used to select the focus industries. 3-7 ------- era c re w M r> ฉ S ฉ a r> Upstream Markets LtJ 00 Fuel Markets Supply Demand Exogenous Endogenous Oil demand oil Gas Coal demand S demand coal Waste Disposal Markets Landfill demand All Other Inputs demand (Q) Production Process /s Downstream Markets Focus Industry A Btu Production Manufacturing Process - Regulatory- Costs Focus Industry B Focus Industry C Remaining Affected Industries Intermediate or Final Product Markets Supply Demand Endogenous Exogenous S supply V product A V Product A Nonaffected Industries ------- agencies. Demographic data will also be included on the communities in which affected facilities are located to identify areas with low-income and minority populations. For selected communities, this information will be used to compare with-regulation conditions with the baseline without-regulation conditions. 3.3 Modeling Market Components A limited number of focus industries will be selected to be explicitly modeled in the economic impact analysis. Because of the parallel development of the economic impact model with the preliminary estimates of cost and emissions reductions, flexibility in the number and selection of focus industries to be included in the analysis is an important aspect of the model. As shown in Figure 3-1, the primary components of our model are individual facility production processes, downstream product markets, and upstream input markets. This subsection discusses each component in turn and then describes the linkages that will be incorporated into the empirical model. 3.3.1 Production Process We begin by modeling the production process for an individual facility as shown in Figure 3-2. In the figure the facility's production process is shown inside the dashed lines. The market for its products (downstream market) is to the right of the dashed box and the inputs to its production process are shown to the left of the dashed box. Inputs are segmented into three categories: fuel to generate Btus, landfill for waste disposal, and all other inputs. The production process is divided into energy production and the manufacturing process. Regulatory costs are modeled as increasing the facility's cost of energy ($/Btus) for heat, power, and electricity generation. In our example, Fuel A is used to generate Btus in the facility's existing units and in the process generates pollutants (p). The pollutants are then treated before emissions (e) are released into the environment. Thus, the cost of the regulation (CA, end of pipe treatment, for example) affects the facility's cost of generating Btus. In our modeling approach, because Btus are an input into the manufacturing process, a change in the price per Btu due to control costs affects the facility's manufacturing process in the same way as a change in fuel prices or as a tax on fuels would. The end effect is that the price of a key input into the manufacturing process has increased ($/Btu) and this price increase will affect the facility's supply decision, as shown in the upward shift in the supply function in Figure 3-2. 3-9 ------- era c re w Tl ฉ r> e s a c LtJ o Fuel Fuel B Equipment Costs Btu Prod- uction Alternative Fuel Unit "V" Waste (no Btus recovered) Btu ($/Btu) Manufacturing Process e All Other ~ Inputs AOI (P) Products e = emissions released to environment p = pollutants generated from process CA = regulatory costs for existing units CB = regulatory costs for alternative fuel units Cj = regulatory costs for incinerators P = fixed price of inputs Q = quantity of final products sold P = price of final product ------- Figure 3-2 also illustrates the facility's fuel switching option. The change in the type of fuel used is referred to as fuel switching. Alternative units burning Fuel B could be purchased. If the alternative units burn a cleaner fuel, the control costs associated with the regulation will be less (CB < CA). If the difference in regulatory costs outweighs the equipment costs of switching, then the facility can lower its cost of Btus used in the manufacturing process by switching from Fuel A to Fuel B. Regulatory costs are modeled as also affecting the facility's waste disposal options by increasing the cost of incineration (C,). In our approach waste disposal is modeled as an input into the manufacturing process, and regulation increases the cost of the disposal option incineration. This increased cost of incineration in turn may increase the facility's demand for landfill services. In summary, the direct costs of the regulation (CA and C,) may induce the facility to modify its production process and change its demand for different fuels. demand for landfill services, and supply of products. For each focus industry selected, we will model the facility's links to these three markets. For all other inputs (AOIs) we will assume that their prices have not significantly changed as a result of the regulation, and we will not explicitly model the markets associated with AOIs. 3.3.2 Downstream Product Markets A partial equilibrium analysis will be conducted for the downstream product market associated with each selected focus industry. The product market modeled will be the first well-defined market downstream from the industry. In many instances this will be a market for intermediate products, as opposed to final products consumed by end users. For example, if the steel industry is selected as a focus industry, we will model the supply and demand for rolled steel, as opposed to modeling the market for automobiles (that use steel as an input). In a perfectly competitive market, the point where supply equals demand determines the market price and quantity. In our analysis, regulation-induced shifts in the supply function in downstream product markets are determined by the relationship between control costs, input prices, production costs, and output rates, as described in the previous section. We assume that the demand for the product is unchanged by the regulation. 3-11 ------- Developing the Supply Function To develop the supply side of the market model, the following steps are necessary for each market to be analyzed: 1. Establish baseline production levels. 2. Assign fixed and variable control costs for each facility. 3. Estimate supply response from each facility. 4. Aggregate supply responses across facilities to get market supply response. Baseline supply conditions are needed to establish the baseline market conditions from which the impacts of the regulation are measured. For each selected industry, information will be collected as part of the industry profile to support the develop of the baseline supply conditions. As shown in Figure 3-3, the market can be viewed as having two separate supply segments, and these segments combine to generate the aggregate market supply function. The two supply groups are suppliers that incur control costs to comply with the regulation (panel a), suppliers that are not required to bear control costs to comply with the regulation (panel b). Figure 3-3 shows the regulation-induced impact on the supply function. By raising marginal costs, the regulation causes an upward shift in the supply function of the producers bearing the control costs from S10 to Sn in panel a. The supply function for the producers not bearing compliance costs remains unchanged by the regulation (S20) in panel b. The combined effect of the regulation-induced changes in supply for the different groups causes the aggregate supply function to shift upward and inward from ST0 to ST, in panel c. This shift in the aggregate supply function leads to a new equilibrium market price and quantity. Developing the Demand Function Even though we assume that the demand for the product is unchanged by the regulation, the baseline demand conditions for the product are important to determining the new equilibrium price and quantity. The intersection of market supply and market demand 3-12 ------- (a) Producers bearing control (b) Producers bearing no (c) Total Market costs (affected) control costs (nonaffected) P0 = market price without regulation Pi = market price with regulation Sio = supply function for affected firms without regulation Sn = supply function for affected firms with regulation Qio quantity sold for affected firms without regulation Qn = quantity sold for affected firms with regulation S20 supply function for nonaffected firms both with and without regulation Q20 quantity sold for nonaffected firms without regulation Q21 quantity sold for nonaffected firms with regulation sT0 total market supply function without regulation ST1 total market supply function with regulation Qto total market quantity sold without regulation Qti total market quantity sold with regulation Figure 3-3. Market Effects of Regulation-Induced Costs curves determines the market price and quantity; thus, the shape of the demand curve influences the change in price and quantity associated with the regulation. The demand function quantifies the change in quantity demanded in response to a change in market price. This is referred to as the elasticity of demand. Depending on industry conditions, demand is modeled as either a single domestic market demand, multiple domestic market demand segments (e.g., consumer and institutional demand), or as some combination of domestic and foreign demand segments. In most cases, demand functions or functional parameters such as demand elasticities can be found in the literature and modified to the current situation. In other cases, estimates are not available from the literature, and a "reasonable" range of elasticity estimates is often 3-13 ------- assigned based on estimates from similar commodities. Special factors may be considered for foreign demand, because export demand is typically modeled as more elastic than domestic demand, taking into account the extent to which non-U.S. products can substitute for U.S. products in the world market. 3.3.3 Upstream Input Markets By explicitly modeling upstream markets that provide inputs to multiple industries, we move away from a strict partial equilibrium analysis. With upstream markets included in the model, changes in production levels in one industry can now affect the production process in other industries through changes in the aggregate demand for common inputs. Our preliminary hypothesis is that the fuel and waste disposal markets will be the primary upstream markets affected by the regulation. As shown in Figure 3-4, compliance costs associated with Btu production are modeled as increasing the price of energy (t $/Btu). This impact on the price of Btus to the facility feeds back to the fuel markets in two ways. The first is through the company's input decisions for the type of fuel it is going to burn to generate Btus for its manufacturing process. Our approach to modeling fuel switching is discussed in Section 3.4.1. Compliance Costs Output Market Figure 3-4. Fuel Market Interactions with Facility-Level Production Decisions 3-14 ------- The second feedback pathway to the fuel markets is through the facility's change in output. The change in facility output is determined by the size of the Btu cost increase (typically variable cost per output), the facility's production function (slope of facility-level supply curve), and the characteristics of the facility's downstream market (other market suppliers and market demander). For example, if consumers' demand for a product is not sensitive to price, then producers can pass the cost of the regulation through to consumers and the facility output will not change. However, if only a small number of facilities in a market are affected, then competition will prevent a facility from raising its prices. One possible feedback pathway we will not model is technical changes in the manufacturing process. For example, if the cost of Btus increases, a facility may use measures to increase manufacturing efficiency or capture waste heat. These facility-level responses are a form of pollution prevention. However, incorporating these responses into the model is beyond the scope of our analysis. 3.4 Operationalizing the Economic Impact Model The production process is linked to the upstream and downstream markets through the supply and demand for commodities as described in the previous section. Compliance costs will affect both the input mix a facility uses in its production process (oil versus natural gas or incineration of waste versus landfill) and the cost of producing its product (cost per unit output increases). Figure 3-5 illustrates the linkages we will use to operationalize the estimation of economic impacts associated with compliance costs. In both the upstream fuel and waste disposal markets, supply is assumed to not be affected by the compliance costs (supply is exogenous; it is determined outside the model) and the demand for different fuel types or waste disposal is determined by aggregating facility-level production decisions (demand is endogenous; it is simulated by the model). Similarly, in the downstream markets, product demand is assumed to not be affected by compliance costs (exogenous demand), and the product supply is determined by aggregating facility-level production decisions (endogenous supply). Adjustments in the facility-level production process and upstream and downstream markets occur simultaneously. A computer model will be used to numerically simulate market adjustments by iterating over commodity prices until equilibrium is reached (i.e., until supply equals demand in all markets being modeled). 3-15 ------- era c re Fuel Markets a Assume Supply Model Demand 3 23 n Oil Gas Coal demand demand 2 demand coal < Production a Final Product Markets /\ A Production Process (Fuel Switching) A Production Levels ------- 3.4.1 Computer Model The computer model comprises a series of computer spreadsheet modules. The modules integrate the engineering inputs and the facility- and market-level adjustment parameters to estimate the regulation's impact on the price and quantity in each market being analyzed. At the heart of the model is a market-clearing algorithm that compares the total quantity supplied to the total quantity demanded for each market commodity. Current prices and production levels are used to calibrate the baseline scenario (without regulation) for the model. Then, the compliance costs associated with the regulation are introduced as a "shock" to the system, and the supply and demand for market commodities are allowed to adjust to account for the increased production costs resulting from the regulation. Using an iterative process, if the supply does not equal demand in all markets, a new set of prices is "called out" and sent back to producers and consumers to "ask" what their supply and demand would be based on these new prices. This technique is referred to as an auctioneer approach because new prices are continually called out until an equilibrium set of prices is determined (i.e., where supply equals demand for all markets). Supply and demand quantities are computed at each price iteration. The market supply is simply the sum of responses from individual suppliers within the market. Included in the iterative process may be an assessment of the plant closure decision. As illustrated in Figure 3-3, after shutdowns are accounted for, production is aggregated across all suppliers to obtain the total supply for each market. The quantity demanded for each market is obtained by using the mathematical specification of the demand function. Modeling Plant Closures Because the profit-maximizing level of production for a facility may actually reflect minimizing losses, it may be in the best interest of the facility to liquidate its assets and shutdown. Plant closures will affect the industry supply. However, the decrease in supply associated with plant closures may be partially compensated for by increases in production by other facilities. One approach to modeling plant closures is to use financial data to assess the impact of the regulation on profitability. However, a major limitation of this approach is that financial data are typically available only at the firm level and not at the facility level (where the closure decision if typically made). Hence, many important factors influencing the closure decision, 3-17 ------- such as the age of a specific facility or facility-level production costs, cannot be included in the analysis. In addition, analyzing individual firms or facilities is very resource intensive. As a result, we will not predict the probability of closure at specific facilities (such as plant X will close but plant Z will not). Alternatively, we will assess the probability of closure for "typical" facilities and use this to adjust the aggregate market supply. This approach accurately provides information on national-level impacts associated with plant closures but is less useful for estimating regional- and community-level impacts. Modeling Fuel Switching Similar to modeling plant closures, we will model fuel switching decisions for "typical" facilities in each industry. Facility-level data on specific equipment and production modification costs associated with fuel switching would be prohibitively expensive and time consuming to develop. As inputs into our analysis, we propose to use secondary data on the population of units with fuel switching capabilities and on the relative changes in fuel prices needed to stimulate fuel switching. The Manufacturing Energy Consumption Survey (MECS) is one potential source of secondary data to model fuel switching. Table 3-3 presents information on the existing capacity of units with the capability to switch from coal to alternative energy sources. The fourth column of Table 3-3 implies that, of the total capacity of coal units in the U.S., 46.1 percent have the capability to switch to alternative energy sources. And, of the switchable units, 69.5 percent (column 6) have the capability to switch to natural gas. Figure 3-6 presents information on establishments' likelihood of switching based on changes in the relative price of competing fuels. As part of the 1994 MECS, approximately 28 percent of establishments indicated that they would switch from fuel oil to natural gas if the relative price difference between the two fuels were to change 5 percent. An additional source of secondary data on fuel switching behavior is shown in Table 3-4, which contains fuel price elasticities developed by the U.S. Department of Energy for the National Energy Modeling System (NEMS). The diagonal elements in the table represent own price elasticities. For example, the table indicates that for steam coal, a 1 percent change in the price of coal will lead to a 0.499 percent decrease in the use of coal. The off diagonal elements are cross price elasticities and indicate fuel switching propensities. For example, for steam coal, the second column indicates that a 1 percent increase in the price of coal will lead to a 0.061 percent increase in the use of natural gas. 3-18 ------- Table 3-3. Capacity to Switch from Coal to Alternative Energy Sources by Industry Group, Selected Industries, and Selected Characteristics, 1994 (estimates in thousands short tons) Coal Alternative Energy Source Short Tons Consumed % of National Total Switch- able % Switch- able % Electricity Receipts % Natural Gas % Distillate Fuel Oil % Residual Fuel Oil % LPG % Other 20 Food and Kindred Products 7,500 13.9 3,330 44.4 9.8 80.2 25.6 24.5 15.6 0.1 24 Lumber W W W W Q W * * * * 25 Furniture and Fixtures 115 0.2 74 64.3 * 66.2 9.5 Q Q 44.6 26 Paper 13,812 25.5 6,198 44.9 10.7 48.5 28.0 52.2 5.4 7.2 28 Chemicals 11,597 21.4 3,789 32.7 4.4 58.6 49.7 30.6 W * 29 Petroleum and Coal Products W W 151 W * 75.5 W 72.2 W * 32 Stone, Clay, and Glass 12,423 22.9 6,990 56.3 * 90.7 24.4 28.3 11.9 14.1 33 Primary Metals 2,327 4.3 1,858 79.8 W 82.9 W 5.9 W W U.S. Total 54,143 100.0 24,943 46.1 6.5 69.5 28.8 33.6 7.7 6.0 Q = Withheld because Relative Standard Error was greater than 50 percent. W = Withheld to avoid disclosing data for individual establishments. * = Estimate less than 0.5. Source: U.S. Department of Energy, Energy Information Administration. December 1997 "1994 Manufacturing Energy Consumption Survey". DOE/EIA-0512(94). Washington, DC: U.S. Department of Energy. ------- 50 %of Establishments That Would Switch Note For remaining 52.3%: 6.3% would not switch due to price 38.7% estimate cannot be provided 7.3% would switch to more expensive alternative 10 -- 0 -I 1 1 1 1 1 1 H 0 5 10 15 20 30 40 50 Percent Level of Price Difference $ Current Alternative Fuel Fuel Current Fuel Figure 3-6. Percentage of Establishments by Level of Price Difference Between Residual Fuel Oil and Less Expensive Natural Gas that Would Switch Fuels Source: U.S. Department of Energy, Energy Information Administration. December 1997. "1994 Manufacturing Energy Consumption Survey." DOE/EIA-0512(94). Washington, DC: U.S. Department of Energy. When using secondary data, analysts must use caution when incorporating this type of fuel switching information into an impact analysis. For example, the MECS data in Figure 3-6 are based on stated intentions for hypothetical price scenarios and are subject to possible bias associated with stated intentions versus actual behavior. In addition, the price elasticities in Table 3-4 include not only behavioral changes associated with fuel switching, but also changes in energy use from market-induced changes in production. However, secondary data on fuel switching can provide useful insights into potential impacts associated with changes in the relative price of fuels. As described in the following section, we will first conduct a screening test to determine if fuel switching resulting from the regulation is likely to be have a "measurable" economic impact. If so, we will incorporate the fuel switching feedback loop (shown in Figure 3-5) into the model. 3-20 ------- Table 3-4. Fuel Price Elasticities Own and Cross Elasticities in 2015 Inputs Electricity Natural Gas Coal Residual Distillate Electricity -0.074 0.092 0.605 0.080 0.017 Natural Gas 0.496 -0.229 1.087 0.346 0.014 Steam Coal 0.021 0.061 -0.499 0.151 0.023 Residual 0.236 0.036 0.650 -0.587 0.012 Distillate 0.247 0.002 0.578 0.044 -0.055 Source: U.S. Department of Energy, Energy Information Administration. January 1998. "Model Documentation Report: Industrial Sector Demand Module of the National Energy Modeling System." DOE/EIA-M064(98). Washington, DC: U.S. Department of Energy. 3.4.2 Aggregating Social Costs After the model has reached equilibrium in all the relevant markets, the changes in price and quantity are used along with engineering cost estimates to calculate total social costs. Engineering costs are categorized in terms of fixed costs and variable costs. Total social costs are the sum of total fixed costs plus total variable costs for all industries. Total fixed costs are determined by weighting the engineering fixed cost estimates ($/unit) by the number of units in facilities that decide to comply with the regulation (i.e., the affected population less plant closure units). Total variable costs are determined by weighting the engineering variable cost estimates ($/output) by the output simulated by the computer model (i.e., baseline output less the change in production resulting from the regulation). Finally, the change in price will be used to allocate social costs between consumers and producers. 3.5 Screening Procedures for Determining Scope of Analysis Because the modeling approach we have outlined in this document has been developed prior to reviewing preliminary estimates of costs and emissions reductions, the focus industries to be selected and final scope analysis are still being determined. Focus industries will be selected on the following screening criteria: high total cost impacts, 3-21 ------- high relative cost impacts (e.g., industry cost-to-sales ratios or $/Btu), potential small business impacts, significant benefits impacts, available information, large variation in the distribution of costs, and high fuel switching potential. It is likely that total industry costs will be closely related to Btu usage and that relative industry cost impacts will be related to Btu intensity per unit output. In addition, the fuel mix industries use to meet their energy requirements will also significantly affect the magnitude of the regulation's impact. Table 3-5 shows the total inputs of energy and the fuel source for selected industry groups. We will also conduct a preliminary screening process to assess potential impacts on the fuel markets and on the waste disposal market when we receive initial cost estimates from the Source Work Groups. This preliminary screening will be used to determine the level of effort appropriate for modeling the fuel markets and waste disposal market. For example, if it is determined that the regulation's average impact on the cost of energy ($/Btu) is less than 1 percent, we may conclude that this change is not large enough to induce significant fuel switching activity. In this instance we would not devote additional resources for a detailed analysis of fuel switching in our model. Alternatively, if the change in the cost of energy resulting from the regulation is large enough to induce fuel switching (significantly affecting at least one fuel type), it will be important to capture these effects in our model. For example, if there is a significant switch from coal to natural gas, the price of natural gas may increase. This price increase would impact units and industries regardless of whether they are "affected" or "nonaffected" by engineering control costs. Although the change in the fuel prices may be small, the aggregate impact on the national economy may by large. A similar screening process will be conducted for waste disposal to determine if the regulation's impact has the potential to significantly affect the facility's disposal decision and, hence, the market for waste disposal. An additional issue to be considered for waste disposal is that these markets are likely to be regional, and selected community-level impacts may be large even if national-level impacts are small. 3-22 ------- Table 3-5. Total Inputs of Energy for Heat, Power, and Electricity Generation by Industry Group and Selected Industries, 1994 (estimates in billion Btus) Fuel type SIC Industry Total Net Electricity Residual Fuel Oil Distillate Fuel Oil Natural Gas LPG Coal Coke and Breeze Other Percent of U.S. Total 20 Food and Kindred Products 1,183 198 30 19 630 6 165 2 134 7.2 24 Lumber 435 68 W 22 48 W W 0 290 2.6 25 Furniture and Fixtures 65 22 * 1 23 1 3 0 14 0.4 26 Paper 2,634 223 173 9 574 5 307 0 1,343 15.9 28 Chemicals 3,273 520 60 13 1,895 W 257 w 521 19.8 29 Petroleum and Coal Products 3,263 121 72 21 W 47 W w 2,181 19.8 32 Stone, Clay, and Glass 945 123 7 23 431 4 274 8 75 5.7 33 Primary Metals 2,568 493 43 12 801 5 52 687 475 15.5 U.S. Total 16,515 2,656 441 152 6,141 99 1,198 703 5,126 100.0 W= Withheld to avoid disclosing data for individual establishments. *= Estimate less than 0.5. Source: U.S. Department of Energy, Energy Information Administration. December 1997 ."1994 Manufacturing Energy Consumption Survey." DOE/EIA-0512(94). Washington, DC: U.S. Department of Energy. ------- SECTION 4 BENEFITS ANALYSIS METHODOLOGY PLAN We will structure our approach on the general conceptual and procedural frameworks described below. We will then identify the areas in which this framework will need to be adapted, tailored, and expanded to address the specific features of the ICCR. 4.1 Theoretical Basis for Benefits Analysis The benefits of pollution controls are defined as the changes in human welfare that result from improvements in environmental quality. Therefore, assessing the benefits of air pollution controls requires a conceptual framework that specifically links reductions in air pollutant emissions to human welfare enhancements. Such a framework serves to distinguish the essential components of benefits and to ensure that all relevant benefits are accounted for and that double counting of benefits is avoided. Figure 4-1 illustrates a simple framework for this purpose. It depicts three fundamental "systems"an environmental system and two human (market production and household) systemsand a number of flows to and from these systems. Pollutant releases are shown as flows of residuals from human activities into the environmental system (i.e., the atmosphere), which disperses and transforms them. The flows from the environmental system to the human systems are defined as environmental "services," which support human life itself, in addition to a variety of production, leisure, and other related human activities. Humans, who reside in the household system, either receive these services directly (e.g., through the air they breathe) or indirectly through the market production system and market exchange (e.g., through purchases of agricultural products whose yields depend in part on air quality). In essence, humans ultimately convert these service flows into human welfare, which is abstractly measured by household or individual "utility." Through this simple framework, pollution controls can be represented as reductions in pollutant releases to the environment. These reductions lead to a change in both the quantity and quality of the flow of environmental services to human systems and, consequently, to changes in utility and human welfare. 4-1 ------- Pollutant Releases ฆ<- Environmental "Services" Market Product Systems Household Systems T Human Welfare Figure 4-1. Conceptual Framework Linking Pollutant Releases to Human Welfare The next conceptual hurdle is to translate changes in human welfare to a monetary measure of value. The most widely accepted measure in the economics profession for valuing changes in utility is individuals' maximum willingness to pay (WTP).1 The use of WTP to measure the value of an improvement in air quality amounts to asking what monetary payment would exactly offset (and thus be inversely equivalent to) the change in utility that an individual experiences with cleaner air. When the WTP for all changed environmental service flows is summed across all affected individuals, it provides a measure of the total aggregated benefits of the improvement in air quality. 'A related concept is willingness to accept (WTA), which refers to the minimum compensation individuals would be willing to accept to forgo a positive change. It is also a benefits measure and may be more appropriate in certain circumstances; however, under certain restrictive assumptions WTP and WTA will be equal. For simplicity we focus on WTP as the conceptual basis for assessing benefits. 4-2 ------- 4.2 Analysis Framework Assessing the benefits of reductions in pollutant emissions requires three fundamental analytical steps: Identify the primary pollutants of concern and the ways in which environmental services are impaired or damaged by emissions of these pollutants. Quantify the measurable physical effects of changes in emissions. Monetize the values associated with the physical effects and related behavioral changes. A thorough identification of damages associated with pollutant emissions recognizes the various ways in which pollutants interact and are transformed in the environment. Figure 4-2 illustrates this for several major categories of air pollutants: volatile organic compounds (VOCs), hazardous air pollutants (HAPs), particulate matter (PM), nitrogen oxides (NOx), and sulfur oxides (SOx). The atmospheric interaction of pollutants creates a myriad of pathways through which emissions ultimately affect air quality and impair the services that humans receive from the environment. These changes in environmental services can be mapped into broad categories of potential damages, such as those described below (adapted from Freeman, 1993): Direct damages to humans: health damages: health damages include primarily the increased morbidity (both acute and chronic) and mortality associated with exposures to harmful substances. visibility and other aesthetic damages: impaired visibility, odors, and other adverse aesthetic effects can result from air pollution. Indirect damages through ecosystems: reduced economic productivity of ecosystems: these damages include the reduced productivity of commercial ecological systems used for agriculture, forestry, and commercial fishing in the areas affected by releases. reduced quality of recreation activities: recreation damages result primarily from the reduced quality of ecological resources used for recreational activities, such as fishing and swimming. 4-3 ------- Sulfur Oxides Particulate Nitrogen (mainly sulfur dioxide) Matter Oxides Hazardous Air Pollutants and Volatile Organic Compounds Human Health Materials Aquatic Ecosystems Vegetation and Visibility Terrestrial Ecosystems Figure 4-2. Identification of Damage Pathways for Major Air Pollutants Source: Adapted from National Acid Precipitation Assessment Program. June 1993. 1992 Report to Congress. p. 25. 4-4 ------- reduced intrinsic/nonuse value: nonuse damages include the lost value individuals associate with preserving, protecting, and improving the quality of resources that is not motivated by their own use of these resources. This category of damages also includes the altruistic and bequest values individuals associate with improvements in the welfare of others. Indirect damages through nonliving systems: reduced productivity of materials and structures: impacts such as soiling corrosion and decay of materials and structures, in effect, reduce their productivity. As shown in Figure 4-3, quantifying the impacts of pollutant emissions requires two general stages of modeling. The first stage involves translating pollutant emissions to air quality by applying air dispersion models. Additional fate and tranport modeling of pollutants can describe how atmospheric concentrations are further transported and accumulate in other media, such as water and biota. Applying these models to both baseline (without regulation) and control (with regulation) emissions levels provides estimates of reductions in ambient pollutant concentrations in air, water, and soil. The second stage involves translating changes in ambient concentrations to changes in physical effects. For example, by applying existing concentration-response functions the effects of reductions in ambient concentrations can be measured as reductions in adverse health risks or increases in crop yields and visibility. Quantification of these physical effects depends on the existence and reliability of fate and transport models, concentration-response functions, and other supporting data such as for emissions, climatic conditions, population distributions, and land uses. Where these models or data are not available, the benefits analysis is limited to a thorough qualitative description of impacts. Assigning monetary value to the physical effects is an extension of the quantification steps, as shown in Figure 4-3. This step requires analysts to estimate total WTP for improvements in environmental quality and/or the related changes in physical effects, by appropriately aggregating individuals' WTP for the changes in question. As shown in Table 4-1, several empirical valuation methods involving primary data collection exist for estimating individuals' WTP. 4-5 ------- Baseline Pollutant Emissions w With-Control Pollutant Emissions 1 r Reduction in Ambient Concentrations 1 f Physical Effects Figure 4-3. General Procedural Framework for the Quantification and Valuation of Emissions Reductions 4-6 ------- Table 4-1. Classification of Valuation Methods Characterization of Valuation Method Types of Assumptions Required for Valuation Method Valuation Method Behavioral Revealed Preference Market exists for product produced with affected air resource services, or Marketed good or service used jointly with air resources affected in production or consumption activities Factor income Hedonic Travel cost Stated Preference Impacts on air resource and/or its services can be described and valued in simulated market using expressed preferences Contingent valuation Conjoint analysis Nonbehavioral Damage Cost Value of affected air resource is at least equal to the cost of remediation Replacement costs Restoration costs Cost of illness Source: Adapted from Smith, V.K., and J.V. Krutilla. 1982. "Toward Formulating the Role of National Resources in Economic Models." In Explorations in Environmental Economics, V.K. Smith and J.V. Krutilla, eds., pp. 1-43. Baltimore, MD: Johns Hopkins Press. Nonbehavioral approaches generally measure the cost of repairing or treating air pollution damage. Assuming that these damage costs are avoidable through improvements in air quality, they are often used to approximate benefits. Although rather straightforward and easy to apply, nonbehavioral approaches are not specifically designed to measure WTP and may therefore only provide rough (and usually lower-bound) approximations of true benefits. Behavioral approaches are preferable on theoretical grounds because they are designed to measure WTP; however, they also suffer from limitations. Revealed preference approaches rely on observed behavior to infer individuals' WTP for environmental (or 4-7 ------- related) improvements, but the data requirements are generally burdensome, and they require assumptions about individuals' perceptions of the improvements. Stated preference methods use surveys to directly elicit individuals' WTP for similar improvements, but it is often difficult to validate whether individuals' stated WTP reflects their true WTP. Resource constraints will effectively preclude the direct application of these methods for this benefits analysis. Therefore, WTP estimates must be derived from existing studies that have either applied these methods directly (in related contexts) or that have used the results of these methods from other studies. In principle, through a process of "benefits transfer," estimated values for specific air quality-related impacts (e.g., health effects), air quality improvements, or even emissions reductions can be taken from existing studies and applied to the present context. For example, reductions in mortality can be valued using information from studies that have applied revealed or stated preference methods to estimate individuals' WTP to reduce their risks of premature death. The accuracy and reliability of value estimates from transfer procedures such as these will depend on the quality of original studies and the correspondence between the original study context to this policy context. Valuation of the quantified effects of the proposed emissions controls therefore depends on the existence and quality of applicable valuation studies and estimates. Where these are not available, the benefits analysis is again limited to a thorough qualitative description of impacts. 4.2.1 Identify the Primary Pollutants of Concern and Characterize Their Effects The first step in the analysis will be to identify the primary pollutants of concern for each of the five combustion source categories and to characterize the damages associated with each of these. Pollutants can be generally categorized as either HAPs, as specified in Section 112 of the Clean Air Act Amendment of 1990, or as conventional criteria pollutants. Initial pollutant screening is being conducted by each of the Source Work Groups, based on the emissions testing data that are being compiled in the ICCR Emissions Database. At this stage, pollutants with very low emissions are being dropped from consideration. Criteria pollutants, including carbon monoxide (CO), NOx, VOCs, S02, PM, and lead, are emitted in large quantities from a variety of sources across the nation. The impacts of these pollutants have been widely studied, and they are associated with a broad range of damages to human health and the environment. EPA has systematically reviewed and summarized the findings of these studies and identified the primary adverse effects of each of these pollutants (EPA, 1997a). For those criteria pollutants that are found to be emitted in 4-8 ------- large quantities from ICCR sources and expected to be significantly reduced by the emissions controls under consideration, we will summarize their primary adverse effects and discuss the key areas of uncertainty associated with these and other potential effects. HAPs are less ubiquitous than the criteria pollutants; however, they are of particular concern because of their potential to cause serious health problems even at relatively low levels of exposure. They may also cause a number of other damages to the environment and human welfare. HAPs can be broadly categorized according to their carcinogenic potential, as well as to their potential to cause noncancer health effects. Known or suspected carcinogens will be given greater priority in the benefits analysis because of the severity of cancer and because these effects are more easily quantified than noncancer effects. We will categorize the HAPs according to their carcinogenic weight-of-evidence and give higher priority to those that are known human carcinogens. Other important considerations in prioritizing HAPs of concern are their persistence in the environment and their potential to bioaccumulate. These properties are particularly important for examining multipathway exposures beyond direct inhalation. We will also prioritize HAPs according to their quantity of emissions prior to ICCR controls and to the expected reduction in emissions as a result of these controls. For those HAPS receiving the highest priority, we will summarize the evidence regarding their potential to cause human health and other damages. 4.2.2 Select and Characterize Emissions Sources for Emissions, Air Quality, and Exposure Modeling Measuring emissions, air quality changes, and exposure impacts will not be feasible for each of the sources covered by the ICCR. To the extent that this type of modeling is conducted for this rule, it will need to be based on a sample of affected facilities. There are three general approaches for doing this. The first, and most thorough, approach would be to select a (stratified) random sample of facilities that is large enough to support statistically valid inferences regarding the universe of affected sources and then model impacts at each of these facilities. The main limitations of this approach are that it will still require modeling a relatively large sample of facilities and that the randomly selected facilities may not be those with the most accurate and complete data required for modeling. The second approach would be to select and characterize a limited number of model facilities for each of the source categories. Rather than being selected at random, these facilities should be "average" or "typical" facilities, chosen to be reasonably representative of a 4-9 ------- large portion of the affected facilities. This limits the number of facilities to be modeled, but it also increases the uncertainty for extrapolating results to the larger universe of facilities. The third approach would be to select only one or two facilities for analysis and to use these as case studies. These case study facilities can be selected to represent worst case scenarios or to reflect more typical conditions; however, the results must be interpreted as illustrative. They do not provide an appropriate basis for quantifying aggregate impacts from the larger universe of affected facilities. Time, data, and other resource constraints for this analysis will effectively preclude the first approach; therefore, any emissions, air quality, or exposure modeling will need to be based on either a model facility approach or a case study approach. These approaches are discussed below for specific applications. An alternative to modeling air quality and exposure is to transfer results from similar studies that have already modeled these processes. This approach is also outlined below. 4.2.3 Model Emissions for Selected Pollutants The ICCR Emissions Database that is being compiled for this rule contains detailed pollutant-specific emissions testing data for a wide variety of emissions sources. These data provide the basis for linking emissions rates (e.g., pounds per year) to facility characteristics (unit type, fuel type, design capacity, operating rate, control device) and for defining model facilities based on these characteristics. Aggregate emissions for all affected sources can then be estimated (under baseline and with-regulation scenarios) by mapping each source in the ICCR Inventory Database to a corresponding model facility. For certain pollutants, estimates of aggregate emissions reductions provide an adequate basis for estimating their associated benefits. That is, using benefits transfer, as described above, these changes can be directly valued using average per-ton values derived from other studies. This is the case, for example, for reductions in most of the criteria pollutant emissions. For this reason, in the subsequent discussions we only consider air quality and exposure modeling for HAPs. 4.2.4 Select and Model Primary Exposure Pathways To the extent that air quality and exposure modeling is conducted for ICCR pollutants, it will be limited to HAPs. For this reason we specifically focus on models for estimating human health risks. As shown in Figure 4-4, human exposures can be roughly divided between direct exposures (through inhalation) and indirect exposures. Indirect 4-10 ------- Figure 4-4. Physical and Biological Routes of Exposure 4-11 ------- exposures can occur through several physical and biological pathways as a result of atmospheric deposition of pollutants to both land and water. Ultimately humans can ingest or absorb harmful substances through the consumption of water and food or through incidental contact. Modeling Direct Inhalation Exposure The Human Exposure Model (HEM) is most commonly used to estimate health risks from HAP emissions. Built around an air dispersion model such as the Industrial Source Complex (ISC) models, HEM estimates ground-level pollutant concentrations at specific points in the vicinity (within 50 km) of an emissions source and the number of individuals exposed at each point. The key inputs to HEM include information regarding pollutant-specific emissions rates, plant configuration (e.g., stack height, stack diameter), local meteorological conditions (e.g., average wind speed), local terrain/topographical conditions (e.g., urban vs. rural), and local population distribution. The HEM results provide the basis for estimating direct cancer and noncancer health risks from inhalation of HAPs to populations surrounding an emissions source. Because HEM is highly automated, it can be relatively easily adapted to model direct exposures under a variety of alternative conditions. A model facility can be specified by selecting a vector of key input values that is reasonably representative of a larger group of facilities. This process can be repeated for other model facility specifications. With an estimate of the number of facilities corresponding to each model facility, the HEM results from the model facility runs can then be used to develop rough estimates of exposures (and therefore risks) for all affected facilities. The level of effort required for this approach will increase substantially with the number of model plants selected, the number of pollutants modeled, and the number of emissions scenarios (baseline and with-regulation) included. Therefore, in selecting the number of model plants, it will be important to trade off the increase in precision against the added cost of including an additional model plant. The number of pollutants can also be limited through an additional screening process, for example, by using preliminary HEM 4-12 ------- results to estimate baseline maximum individual risks (MIR) (described in Section 4.2.5) and excluding pollutants with very low risk from subsequent HEM runs. Modeling Indirect Multipathway Exposures The modeling of indirect exposures builds on the air dispersion component of the direct exposure model and extends it into other media where pollutants are further transported and dispersed. It therefore requires additional layers of modeling and supporting data, which increase substantially with the number of indirect exposure scenarios examined. For example, as shown in Figure 4-4, atmospheric deposition to water can lead to indirect exposure through fish consumption. This requires information on the geographical distribution of surface water in the vicinity of the emissions source, as well as bioaccumulation rates of pollutants in fish tissue, fishing participation rates by anglers, and fish consumption rates. Deposition to land can lead to uptake by plants and animals that are subsequently consumed by humans. Among other things, this requires information on how these plants and animals are locally distributed both before and after they are exposed. The level of effort and data required to comprehensively estimate indirect exposures precludes the use of a model plant approach to estimate aggregate impacts from these exposures. Nevertheless, a more limited case study approach of selected exposure pathways can serve to illustrate the types and magnitude of indirect impacts associated with HAP emissions from combustion units. An important step in conducting such a case study analysis will be to select the pollutants and the emissions sources or model plants to be analyzed. The fundamental criteria for selecting pollutants are their persistence in the environment and potential to bioaccumulate. This will tend to focus attention on inorganic pollutants such as mercury, arsenic, lead, cadmium, and chromium and on dioxins and radioactive pollutants. Case study facilities can be selected by identifying one or two emissions sources with average-to-high emissions of the key pollutants and also selecting sites with reasonably representative characteristics for indirect exposures (e.g., a site with a high proportion of surface water acreage and a typical rural or urban site). The next step will be to select the specific exposure scenarios to be assessed at the case study site(s). One useful approach for illustrating potential impacts through indirect exposures is to select two general categories of exposure scenarios: one representing high- end exposure assumptions and another representing more central-tendency exposure assumptions. For example, the high-end scenario might include a subsistence fisher scenario 4-13 ------- that assumes lifetime exposure and high fish consumption rates. A central-tendency scenario might include shorter-term exposure and more average consumption rates for a recreational angler. An additional step might be to estimate the number of individuals represented by each exposure scenario; however, this is considerably more difficult than estimating populations exposed via inhalation because, in addition to residential location, it requires broad-based information on behaviors (e.g., recreation, food consumption). Because of this uncertainty and because the case study results are less appropriate for drawing inferences for all affected facilities, estimating exposed populations for indirect exposures should be given much less priority than for direct inhalation exposures. The case study analyses of indirect exposures will most likely not provide results that are appropriate for estimating aggregate impacts at all affected facilities; however, they will provide the basis for estimating plausible ranges of individual health risks under alternative scenarios and for identifying the pollutants of primary concern. In addition, the case study sites can provide a framework for examining exposures and risks to sensitive ecological systems. For example, by examining indirect exposures through atmospheric deposition of pollutants to surface water, this provides the basis for modeling exposures and potential risks to aquatic ecosystems as well. Again, such an analysis would only serve to illustrate the potential for ecological damages from combustion sources affected by the proposed rule. 4.2.5 Estimate Human Health Risks Through the Modeled Exposure Pathways Modeling each of the exposure pathways described above will provide the foundation for estimating both the cancer and noncancer risks associated with these exposures. Inhalation Cancer Risks The HEM analysis described above will provide estimates of ambient ground-level concentrations of selected pollutants at various points, generally within 50 km of each model facility. Overlaying these estimates onto Census data and linking populations to each of these points allow the estimation of the number of people exposed at each modeled concentration level. Applying concentration-response functions (i.e., cancer slope factors) to the concentration estimates for the carcinogenic pollutants then allows us to measure and express cancer risk in at least two ways: maximum individual cancer risk (MIR)cancer risk to the individual(s) living at the point with the highest measured concentrations of carcinogens 4-14 ------- cancer incidencethe total number of additional expected cancer cases as a result of the modeled exposures Estimating pollutant-specific MIR under baseline conditions can also screen out pollutants that do not contrbute significantly to risks (e.g., with estimated MIR below 10"6) and, therefore, do not need to be carried over to the analysis of risk reductions with regulation. Both risk measures can be estimated under baseline conditions, as well as under specific emissions control scenarios, thus providing estimates of reductions in cancer risks at the modeled facilities. The estimated levels and reductions in cancer incidence can be extrapolated to all affected facilities using estimates of the number of facilities corresponding to each model facility. Inhalation Noncancer Risks To assess noncancer risks, the HEM results can also be compared to reference concentrations (RfCs) for the noncarcinogenic HAPs. RfCs are considered to be protective thresholds for inhalation exposure and are defined as the estimate of the daily atmospheric concentrations associated with inhalation exposures that are likely to be without deleterious effect during a lifetime. The ratio of a pollutant's estimated concentration to its RfC (i.e., the hazard index [HI]) indicates the noncancer threat from the pollutant. As with cancer risks, the HI can be estimated for the maximally exposed individual. Under baseline conditions these results can be also used to screen out the noncarcinogenic HAPs whose His are estimated to be well below 1 even for the most highly exposed individual. In contrast to cancer risks estimates, however, there is insufficient information to estimate the incidence of noncancer health effects. Nevertheless, the population data can be used to estimate the number of individuals in the vicinity of the model facility that are exposed to levels exceeding the RfCs. Noncancer risks can be estimated under baseline conditions, as well as under specific emissions control scenarios. The estimated reduction in the number of individuals at risk of noncancer effects can be extrapolated to all affected facilities in the same way that cancer risks are aggregated across facilities. Indirect Cancer and Noncancer Risks Using the estimated exposure concentrations from the case study scenarios, indirect risks can be estimated using the same general approach as for inhalation risks. The primary differences are that the estimated concentrations will be in different media than air (e.g., fish tissue), and they will affect humans through different routes, frequencies, and durations of 4-15 ------- exposure. MIR cancer and noncancer risks can be estimated using the high-end exposure assumptions. In principle, the central-tendency exposure scenarios can also be used to estimate cancer incidence and populations at risk of noncancer effects in the vicinity of the case study facilities. However, as discussed above, there is much more uncertainty in estimating populations exposed via indirect pathways, and these results will be much more difficult to extrapolate to all the affected combustion sources. As a result, to depict potential indirect risks from the affected sources, the case study analysis should rely primarily on estimates of individual risk under alternative exposure scenarios. Ecological Risks As discussed above, the case studies for estimating indirect exposure risks may also provide a useful framework and context for examining ecological risks. For example, estimated concentrations of persistent and bioaccumulative toxics in surface water, soils, and biota can be used as a starting point for estimating exceedances of critical ecological benchmarks. As with noncancer effects, it is difficult to extrapolate these impacts beyond the case study area and to assess them in monetary terms; nevertheless, they can be used to illustrate the potential threats to ecological systems from combustion sources. An alternative indicator of ecological risks is the proximity of combustion sources to known locations of threatened and endangered species of plants and animals. We can determine this proximity by merging source location information from the ICCR Inventory Database with threatened and endangered species occurrence data from the Nature Conservancy. Proximity does not necessarily imply threat, but by comparing the number of such species in the vicinity of affected facilities with numbers in other areas of the U.S., it is possible to develop a rough indicator of potential threats to vulnerable species. Alternative Approaches to Assessing Risks Regardless of whether the approaches outlined above for quantitatively assessing cancer, noncancer, and ecological risks prove to be feasible, the analysis will include a thorough qualitative assessment of these risks. For indirect and ecological risks in particular this will include reviewing the evidence linking the primary HAPs of concern to health and ecological effects and discussing the exposure pathways that are likely to be most problematic. In addition, one possible alternative to the modeling approaches discussed above for quantifying risks is to transfer the results from EPA's Study of Hazardous Air Pollutant Emissions from Electric Utility Steam Generating Units (EPA, 1998). Using methods very 4-16 ------- similar to those outlined above, this analysis measured baseline cancer and noncancer risks (direct and indirect) for as many as 67 HAPs from 684 plants in the U.S. For direct inhalation risks, it also measured long-range transport of HAP emissions and risks extending beyond the local vicinity of plants. The results of the direct inhalation cancer risk assessment are shown in Table 4-2. All noncancer risks from inhalation were found to be well below the RfCs for the pollutants of concern. Table 4-2. Baseline Estimated Cancer Risks from Direct Exposure (Inhalation) to HAP Emissions from Electric Utility Steam Generating Units: Local and Long-Range Impacts in 1994 Oil-Fired Plants (137 plants) Coal-Fired Plants (426 plants) Pollutant Nationwide Emissions (tons/yr) Max i in Li in Individual Risk (MIR) Annual Cancer Incidence Nationwide Emissions (tons/yr) Max i in u in Individual Risk (MIR) Annual Cancer Incidence Radionuclides 1 x 1CT5 0.2 Not estimated 0.7 Nickel 320 5 0.2 52 1 x 10"8 0.038 Chromium 3.9 5 x 1CT6 0.02 62 2 x 10"6 0.15 Arsenic 4 1 x 1CT5 0.05 56 3 x 10"6 0.37 Cadmium 1.1 2 x 1CT6 0.006 3.2 3 x 10"7 0.005 All Others 8 x 1CT7 0.006 1 x 10"6 0.028 Total 6 x 1CT5 0.5 4 x 10"6 1.3 Assuming that emissions from the combustion units covered by the ICCR have the same per-unit impacts as those from electric utility units, the cancer estimates in Table 4-2 can be used to infer baseline risks for ICCR sources. Although this assumption may be reasonable, using this approach will require a careful evaluation of the comparability of electric utility and ICCR risks. It will include a comparison of pollutants emitted, plant configurations, and plant locations. To the extent that there are systematic differences in these factors, the analysis will evaluate how these differences are expected to affect the cancer risk estimates for ICCR sources. For example, if ICCR sources are generally located in more densely populated areas, then, other things equal, the per-unit health impacts of their emissions are expected to be greater than those from electric utility units. Using this approach will 4-17 ------- require a careful discussion of the potential biases and uncertainties inherent in such a transfer. One important limitation of this approach is that it will only provide estimates of baseline direct cancer risks. Reductions in cancer risks will need to be inferred directly from percentage reductions in emissions rather than from modeled cancer risks under alternative emissions control scenarios. The indirect risk assessment for electric utility units used four model plants to assess risks from three HAPsarsenic, mercury, and dioxinsunder various exposure assumptions. For arsenic and dioxins, the highest predicted cancer MIRs were in the 10"4 range but most were below 10"5. For mercury, developmental and neurological effects through fish consumption were of primary concern but were not specifically quantified. The applicability of these results to ICCR combustion sources will depend on the representativeness of the four model plants used in the analysis. Using the results of the electric utility analysis will therefore also require a careful comparison of the four model plants with the universe of ICCR combustion sources. 4.2.6 Estimate Monetary Benefits Although emissions reductions from ICCR combustion sources have the potential to improve human welfare in a number of ways, only some of these effects are quantifiable and even fewer can be expressed in dollar terms. As discussed previously, valuing these emissions reductions will require the application of benefits transfer, because conducting an "original" valuation study is beyond the scope of this analysis. Therefore, benefits must be estimated by applying values derived in comparable studies. Two areas in particular present opportunities for transferring benefits estimates. The first area is the estimated reductions in cancer incidence (from direct inhalation exposures), which can be conservatively interpreted as reductions in mortality and can be valued using previously derived estimates of the value of a "statistical life." The second area is the estimated reductions in criteria pollutant emissions (in particular VOCs, PM, and S02), which can be valued using per-ton benefit estimates derived from EPA's analysis of the revised NAAQS for PM and ozone (EPA, 1997b). The Value of Cancer Cases Avoided Based on a large number of economic studies that have examined individuals' WTP to reduce (or WTA compensation to increase) their risk of death, the value of a statistical life is generally assumed to be roughly equal to $5 million (ranging from about $ 1 million to as much as $14 million). The value of cancer incidence reductions from direct inhalation exposures can be approximated using this statistical life value; however, it is important to 4-18 ------- recognize the limitations of this approach. Valuing each avoided cancer case at an average of $5 million may overestimate their value because not all cancers are fatal and because many cancers may only occur after a long latency period (in which case the number of life years saved is less and, thus, the value of statistical life may be lower). Conversely, the dread, pain- and-suffering, and involuntariness associated with cancer risks from HAP exposures may make these risks relatively more valuable to avoid. In contrast to the estimates of cancer incidence, other measures of risk reductions, such as reductions in cancer MIRs and reductions in noncancer risks, will not be quantified in terms that can be valued. The Value of Reduced Criteria Pollutant Emissions In its analysis of the revised NAAQS for PM and ozone, EPA estimated the nationwide emissions reductions of several criteria pollutants that would be necessary to achieve certain ambient standards. The Agency also estimated many of the benefits that would result from achieving the standards. In particular it estimated the monetary value of several categories of avoided health effects, and the value of improved visibility, reduced soiling of households, and improved yields for certain crops. By apportioning these benefits to the different categories of pollutants, it is possible to approximate the benefits per ton of emissions reductions for each one. These estimates are summarized in Table 4-3. Applying these per-unit values to the criteria pollutant emissions reductions resulting from the proposed controls on ICCR sources will provide a rough estimate of their benefits. For both benefits transfer applications discussed above, there are several areas of uncertainty, both in the original benefit estimates and in the transferability of these estimates to the ICCR context. Therefore, the results of the analysis will need to be qualified with a thorough discussion of the sources and potential magnitudes of these uncertainties. 4-19 ------- Table 4-3. Per-Ton Benefit Estimates for Selected Criteria Pollutants (1990 $) Pollutant Lower-Bound Estimate Upper-Bound Estimate VOCs $444 $2,007 PM10 $9,600 $9,800 S02 Eastern U.S. $4,409 $9,764 Western U.S. $3,190 $3,805 Source: U.S. Environmental Protection Agency. 1997b. Memorandum from McKeever, Michele, EPA/ISEG, to Conner, Lisa, EPA/ISEG. November 4. Benefits transfer analysis for pulp and paper. 4.2.7 Characterize and Summarize Benefits The final step in the analysis will be to summarize the findings. One important component of this will be to review the quantified and monetized benefit estimates and to qualitatively assess those benefits that were not quantifiable. It will also discuss the primary areas of uncertainty in the analysis. 4-20 ------- REFERENCES Freeman, A. Myrick III. 1993. The Measurement of Environmental and Resource Values: Theory and Methods. Washington, DC: Resources for the Future. Industrial Combustion Coordinated Rulemaking Federal Advisory Committee. "Industrial Combustion Coordinated Rulemaking: Organizational Structure and Process." June 1997. Revision 2. Jorgenson, Dale W., and Peter J. Wilcoxen. 1990. "Environmental Regulation and U.S. Economic Growth." RAND Journal of Economics 21(2):314-340. National Acid Precipitation Assessment Program. June 1993. 1992 Report to Congress. p. 25. Smith, V.K., and J.V. Krutilla. 1982. "Toward Formulating the Role of National Resources in Economic Models." In Explorations in Environmental Economics, V.K. Smith and J.V. Krutilla, eds., pp. 1-43. Baltimore, MD: Johns Hopkins Press. U.S. Department of Energy, Energy Information Administration. December 1997. "1994 Manufacturing Energy Consumption Survey." DOE/EIA-0512(94). Washington, DC: U.S. Department of Energy. U.S. Department of Energy, Energy Information Administration. January 1998. "Model Documentation Report: Industrial Sector Demand Module of the National Energy Modeling System." DOE/EIA-M064(98). Washington, DC: U.S. Department of Energy. U.S. Environmental Protection Agency. October 1997a. The Benefits and Costs of the Clean Air Act, 1970 to 1990. Research Triangle Park, NC: Office of Air Quality Planning and Standards. U.S. Environmental Protection Agency. 1997b. Memorandum from McKeever, Michele, EPA/ISEG, to Conner, Lisa, EPA/ISEG. November 4. Benefits transfer analysis for pulp and paper. R-l ------- U.S. Environmental Protection Agency. 1997c. Regulatory Impact Analyses (RIA) for the for the Particulate Matter and Ozone National Ambient Air Quality Standards and Proposed Regional Haze Rule. Research Triangle Park, NC: Office of Air Quality Planning and Standards. U.S. Environmental Protection Agency. February 1998. Study of Hazardous Air Pollutant Emissions from Electric Utility Steam Generating UnitsFinal Report to Congress. Vol.1. EPA-453/R-98-004a. Research Triangle Park, NC: Office of Air Quality Planning and Standards. R-2 ------- |