ECONOMY & ENVIRONMENT GOOD PRACTICE NOTE 8 Local Environmental Externalities due to Energy Price Subsidies: A Focus on Air Pollution and Health Santiago Enriquez Bjorn Larsen Ernesto Sánchez-Triana i GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES CONTENTS Acknowledgments v About the Authors v Acronyms and Abbreviations vi Summary 1 1. Introduction 4 2. Methodological Approach to Assessing Local Externalities of Energy Subsidies 8 3. Prioritizing the Analysis 11 Fossil Fuel Consumption Patterns 11 Emissions 11 .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 A Fuel and Technology Perspective A Sector Perspective. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Population Exposure 15 Dispersion of Emissions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Exposure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 4. Energy Consumption Effects of Price Subsidies 19 Choice of Analytical Model 20 Price Elasticities of Energy Demand 21 Substitution among Energy Sources 22 5. Higher Air Emissions from Energy Price Subsidies 24 Motorized Road Transport 24 Residential Sector 26 Industry 27 Electricity 27 6. Population Exposure Assessment 28 Dispersion Modeling 28 Intake Fractions 29 Intake Fraction Applications 31 ii  Proposed Approach 32 Distributed Ground-Level Emissions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Power Plant Emissions Industry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Baseline Outdoor PM2.5 Concentrations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 7. Health Effects 34 Outdoor Air Pollution 34 Mortality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Morbidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Household Air Pollution 36 Mortality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Morbidity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 8. The Value of Health Effects 38 Mortality 38 Morbidity 38 Using the Value of Health Effects to Inform Policy Options 39 9. Air Pollution Health Risk Assessment Tools 40 10. Conclusions 42 Annex 1: Emission Intake Fractions 43 Annex 2: Methodology for Estimating Health Effects 44 Annex 3: Valuing the Health Effects of Energy Subsidies 46 References 48 Endnotes 54 iii GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES BOXES Box 1: Using Income and Price Elasticity to Estimate Gasoline Consumption Reduction from Price Subsidy Removal in Mexico 22 TABLES Table 1: Quantifying the Effects of Energy Price Subsidies on Local Air Pollution and Health 10 Table 2: Energy Consumption Effects of Energy Price Subsidies 20 Table 3: Average Price Elasticities of Energy Demand 21 Table 4: Examples of Substitution among Energy Sources by Sector 23 Table 5: European Union Light-Duty Diesel Vehicle Emission Standards for PM 25 Table 6: European Union Heavy-Duty Diesel Engines Emission Standards for PM 25 Table 7: Population Weighted Mean Intra-Urban Intake Fractions of Distributed Ground-Level Emissions 29 Table 8: Summary of Recommended Intake Fractions 30 Table 9: Recommended PM2.5 Intake Fractions by Region 30 Table 10: Intake Fraction Estimates across 29 Power Plant Sites throughout China 31 Table 11: Premature Deaths and Days of Illness Caused Annually by Household Air Pollution and Their Associated Costs in Selected Jurisdictions 39 Table 12: Key Characteristics of Air Pollution Health Risk Assessment tools 41 FIGURES Figure 1: National and Subnational Level Comparisons—Cost of Environmental Health Effects Caused by Air Pollution 7 Figure 2: Estimated Annual Average PM2.5 33 Figure 3: Relative Risks of Major Health Outcomes Associated with PM2.5 Exposure 35 iv ACKNOWLEDGMENTS ACKNOWLEDGMENTS This is the eighth in the series of 10 good practice notes under the Energy Sector Reform Assessment Framework (ESRAF), an initiative of the Energy Sector Management Assistance Program (ESMAP) of the World Bank. ESRAF proposes a guide to analyzing energy subsidies, the impacts of subsidies and their reforms, and the political context for reform in developing countries. The authors would like to acknowledge the contributions of Dan Biller and Giovanni Ruta, who peer reviewed this good practice note. The authors are thankful to Thomas Flochel, Marianne Fey, and Masami Kojima for their guidance, support, and review of the good practice note. ABOUT THE AUTHORS Santiago Enriquez is an international consultant with more than 18 years of experience in the design, implementation, and evaluation of policies relating to the environment, conservation, and climate change. He has developed analytical work for the World Bank, the United States Agency for International Development, and the Inter-American Development Bank on topics that include mainstreaming of environmental and climate change considerations in key economic sectors, institutional and organizational analyses to strengthen environmental management, and policy-based strategic environmental assessments. From 1998 to 2002, he worked at the International Affairs Unit of Mexico’s Ministry of Environment and Natural Resources. He holds a master’s degree in public policy from the Harvard Kennedy School. Bjorn Larsen is an international development economist and consultant to international and bilateral development agencies, consulting firms, and research institutions with more than 25 years of professional experience. His primary fields of consulting and research are environmental health and natural resource management from over 50 countries in Asia, Central and South America, Europe, Middle East and North Africa, and Sub-Saharan Africa. Fields of expertise include environmental health risk assessment, health valuation, cost-benefit analysis, poverty-environment linkages, child malnutrition and environment linkages, natural resource degradation and valuation, poverty and natural resources, household survey design and administration, and statistical analysis of household survey data. He has worked extensively on indoor air pollution from solid fuels, urban air pollution, water supply, sanitation and hygiene, and child nutrition and health. v GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES Ernesto Sánchez-Triana is the global lead for environmental health and pollution management for the World Bank. He has worked on projects in numerous countries, including Afghanistan, Argentina, Bangladesh, Bhutan, Bolivia, Brazil, Cambodia, Ecuador, India, the Lao People’s Democratic Republic, Mexico, Pakistan, Panama, Paraguay, and Peru. Before joining the World Bank, he worked for the Inter-American Development Bank. He has led the preparation of numerous policy-based programs, investment projects, technical assistance operations, and analytical works. He holds two master’s degrees and a Ph.D. from Stanford University. He has authored numerous publications on pollution management, clean production, environmental economics, energy efficiency, environmental policy, organizational learning, poverty assessment, and green growth. ACRONYMS AND ABBREVIATIONS ALRI acute lower respiratory infection ARB California Air Resources Board CATEF California Air Toxics Emission Factor CMB chemical mass balance CNG compressed natural gas COPD chronic obstructive pulmonary disease CP cardiopulmonary (mortality) DALYs disability-adjusted life-years EMFAC EMission FACtors EPA U.S. Environmental Protection Agency ESRAF Energy Sector Reform Assessment Framework ESMAP Energy Sector Management Assistance Program g gram GBD Global Burden of Disease GDP gross domestic product HAPIT Household Air Pollution Impacts IHD ischemic heart disease km kilometer vi ACRONYMS A ND ABBREVIATIONS kW kilowatt kWh kilowatt-hour LCIA life cycle impact assessment LCV light commercial vehicle LPG liquefied petroleum gas µg microgram m3 cubic meter NH3 ammonia NOx oxide of nitrogen N2O oxide of sulfur OECD Organisation for Economic Co-operation and Development PAF population attributable fraction PIF potential impact fraction PM2.5 particulate matter with a diameter of less than 2.5 microns PM10 particulate matter with a diameter of less than 10 microns ppm parts per million PPP purchasing power parity RR relative risk SBP systolic blood pressure SOx oxides of sulfur UNEP United Nations Environment Programme VSL value of statistical life WHO World Health Organization WTP willingness to pay YLD years of life lost to disability vii 1 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES SUMMARY This note aims to provide an overview and to meet their needs. Traditional use of these guidance on the use of tools to assess the solid fuels (that is, not burning them in stoves environmental and health effects of changes with high combustion efficiency), coupled in the levels of fine particulate matter caused with inadequate ventilation, results in health- by higher consumption of energy due to damaging concentrations of air pollutants subsidized prices at the country level. It also in indoor environments. Price subsidies for provides information to help practitioners gas and electricity can reduce household develop reliable estimates even in the absence air pollution by encouraging households to of data and with limited resources. substitute traditional energy sources with cleaner forms of energy. Price subsidies for The topic of the note is highly complex and natural gas can also reduce coal and oil product involves multiple fields and disciplines. The consumption in the power and industrial note attempts to reduce such complexity sectors, with net reductions in hazardous by breaking the assessment down into local air emissions. Similarly, lower prices several distinct steps, each with its own of automotive LPG and natural gas due to methodologies. The note is intended to serve subsidies can reduce particulate emissions as a source of resources and practical advice when these fuels are substitutes for liquid to guide practitioners along each of these automotive fuels. steps. Consumer price subsidies for energy have Higher consumption of energy arising from indirect effects on pollution, which might be energy subsidies that keep consumer prices either positive or negative, depending on artificially low can have adverse local and a number of factors, including the energy global environmental impacts. An increase sources and the uses they target. Therefore, an in energy consumption can increase local air understanding of the linkages among energy pollution, global greenhouse gas emissions, price subsidies, environmental quality, and water pollution, and soil contamination from health can inform energy subsidy reforms and energy production and use. Energy production identify measures to mitigate the potential and use are a significant source of global negative environmental impacts of subsidy emissions of fine particulate matter, as well as removal. oxides of nitrogen and sulfur, both precursors to fine particulate matter. Price subsidies for While recognizing that the environmental energy can also lead to an increase in energy- effects of the energy sector are broad-ranging, intensive activities and products that can this note focuses on local air pollution and negatively affect the environment (such as health because it is arguably the energy- unsustainable extraction of groundwater and related local environmental externality with increased use of chemical fertilizers). the largest social cost globally. An estimated 6.5 million people died from outdoor ambient However, energy price subsidies can also and household air pollution in 2015 (Cohen have positive environmental effects. Millions and others 2017). Household air pollution of people still rely on solid biomass and coal SUMMARY 2 also contributes to outdoor ambient air The note is confined to cases where there pollution, because pollutants are not confined are no serious shortages of subsidized strictly to rooms where solid fuels are burned energy. Such shortages are pervasive in for cooking and heating. Several analyses some countries and regions, such as in the conducted by the World Bank found that power sector in Sub-Saharan Africa (Kojima ambient air pollution had an average cost of and Trimble 2016). Where shortages are 3.5% of gross domestic product (GDP) in five widespread—so that consumers are forced Asian countries and 2.5% of GDP in six Latin to go without the specific energy source or American countries. Household air pollution else pay much higher prices than the official had a cost that was as high as 3.3% of GDP ones—the methodologies outlined in this note in Apurimac, Peru, and 4.9% of GDP in the are not applicable. Lao People’s Democratic Republic. Higher prices for polluting fuels can help reduce their Defining the priority sector and fuels is consumption, thereby potentially helping to crucial to conduct a useful assessment, given reduce air pollution; conversely higher prices that it will likely be carried out with limited for cleaner fuels could aggravate air pollution. resources and data. In most countries, the adverse health effects of air pollution from While price subsidies to coal have declined energy price subsidies are caused by a few substantially since the 1990s, those to oil fuels and sectors. Identifying these fuels products and natural gas remain in a number and sectors can therefore be a useful first of countries. Electricity tariff subsidies are also step (section 3). Based on current global prevalent in many countries. Such subsidies energy subsidy patterns, the priority sectors can contribute to overconsumption of energy. for most countries from the point of view Where energy is derived from fossil fuels, of public health will likely be industry, heat overconsumption leads to higher air pollution. and power generation, residential, and road transportation. This note proposes a five-step analysis to assess the health effects of energy price Recent meta-analyses of price elasticities of subsidies, focusing on energy demand by type of fuel and energy provide a basis for assessing the effect of price 1 | The effect of consumer price subsidies on subsidies on energy consumption, provided levels and patterns of energy consumption subsidized energy is readily available to all (section 4 of this note) consumers who wish to purchase it. Cross- 2 | Air emissions from energy consumption price elasticities may be applied in sectors and (section 5) to fuels where significant fuel substitution is likely. Using country-specific urban-transport- 3 | Human exposure to air emissions (section environment models would be advantageous, 6) if available, because of the complexity of air emissions from motor vehicles. 4 | Health effects of exposure (section 7) 5 | Monetary valuation of health effects This note focuses the analysis of price subsidies (section 8) on primary and secondary fine particulate matter (PM2.5, atmospheric particulate matter with a diameter of less than 2.5 microns), 3 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES the pollutant with the largest health effects monitoring measurement data and alternative worldwide, and using intake fractions to options for determining ambient PM 2.5 estimate population exposure to PM2.5 from concentrations at the proposed geographic- fossil fuels and solid biomass. This approach demographic scale, as well as approaches to is similar to that of recent global studies of deal with data scarcity. energy price subsidies and taxes. The intake fractions are combined with the relative-risk The proposed method for estimating the functions for major health outcomes of air economic value of mortality caused by air pollution from the Global Burden of Disease pollution follows a recent World Bank report, study to estimate the health effects associated using a cross-country transfer method of the with energy price subsidies. value of statistical life (VSL). In addition, the note proposes methods for incorporating The note proposes three geographic- valuation of increased illness, although demographic scales: urban areas with a morbidity is generally found to constitute a population over 100,000, urban areas with a relatively minor share of the health costs of population less than 100,000, and rural areas. air pollution. The note also discusses the availability of 1. INTRODUCTION 4 1. INTRODUCTION This note provides an overview and guidance on the use of tools to assess the environmental and health effects of price subsidies for energy at the country level. It also provides information to help develop reliable estimates even in the absence of data and with limited resources. Assessing the environmental and health effects of energy price subsidies is highly complex and calls for an interdisciplinary approach. This note discusses available methodologies and provides examples where such an approach has been adopted, with the aim of sharing practical advice to practitioners interested in conducting similar assessments. Good Practice Note 1 defines an energy subsidy as a deliberate policy action by the government that specifically targets electricity, fuels, or district heating and has one or more of the following effects: • Reducing the net cost of energy purchased • Reducing the cost of energy production or delivery • Increasing revenues retained by energy producers and suppliers These include government control of energy prices that are kept artificially low; budgetary transfers to state-owned energy suppliers or tax expenditures granted to energy suppliers to keep costs down to benefit consumers, producers, or both; underpricing of goods and services provided to energy suppliers such as fuels, land, and water; subsidized loans; and shifting of risk burdens, such as assumption of risks through limits on commercial liability. There are several mechanisms through which the subsidies as defined above can affect the local and global environment: • Prices that are artificially low. This is the focus of this note and arguably the most frequently cited case, assumed to increase consumption relative to the counterfactual of no subsidies. Prices may be low because the government sets low prices or price ceilings, restricts exports of the energy in question, or provides producer support (tax expenditures, underpricing of access to land and other goods and services, below-market provision of loans) with the objective of lowering prices. Low prices for clean fuels may have positive effects on the environment, and conversely low prices for polluting fuels are likely to have negative effects. Subsidized fuel inputs to production sectors, including electricity generation and district heating, are also likely to increase consumption compared to the situation with no subsidized inputs. However, as Good Practice Note 1 details, unintended consequences of subsidies that lower the official prices dampen these effects. 5 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES • Energy shortages. Low prices provide Nigeria in box 6 of Good Practice Note strong incentives for diversion of 1 shows. subsidized liquid fuels to ineligible beneficiaries, including out-smuggling. In assessing the impact of subsidized prices on This has led to acute shortages in some consumption, it is critical to take into account countries, suppressing consumption. both the actual prices paid and any limits on Price subsidies also discourage the availability of the subsidized energy. Many, investment because investors fear if not most, studies examining the impact of that reimbursements may be late, subsidy removal do not take these two factors inadequate, or both. Over time, the into account, leading to overestimation of the sector supplying the subsidized energy likely effect of subsidy removal. may decay if subsidized prices are below On the other hand, refineries in disrepair economic opportunity costs, let alone are in no position to produce fuels meeting supply costs. This is one of the drivers stringent specifications designed to protect of chronic power shortages in some public health. As a result, fuel quality may countries, as well as fuel shortages in lag behind those in developed countries by some major oil exporters that are having years or even decades, preventing adoption to import petroleum products at world of advanced exhaust control devices and even prices because their refining sector is deactivating standard three-way catalytic undercapitalized and in disrepair (such converters in spark-ignition engines. as in the Islamic Republic of Iran, Iraq, and Nigeria). In the extreme, if higher • Cash transfers to consumers. Energy prices unsubsidized prices eliminate fuel may not be kept low, but if consumers are shortages, consumption may actually provided with conditional or unconditional increase, rather than decrease, after cash transfers, consumption will be higher subsidy removal (Kojima 2013; Kojima than otherwise. Cash transfers conditional and Trimble 2016). upon energy purchase will increase • Higher prices on the black market. consumption more than unconditional cash Commercial malpractice in the form transfers, which the beneficiaries can use of illegal diversion and out-smuggling for any purpose. This form of subsidy is not creates fuel shortages, which push up considered in this note. prices. There can be a large difference • Shifting of risk burden. Government between official prices and prices assumption of environmental and safety actually paid. In estimating the impact risks, and consumer or resident assumption of subsidy removal on consumption of risks through limits on commercial liability, volume, it is important to use the actual may encourage energy producers to take prices paid, and not official prices, undue risk at the cost of the environment, which can be considerably lower. In resulting in air and water pollution and soil some cases, illegal diversion has been contamination. This form of subsidy is not so widespread and rampant that considered in this note. consumers have ended up paying far more than even unsubsidized prices, as Energy price subsidies can also lead to an the example of subsidized kerosene in increase in activities and products that use 1. INTRODUCTION 6 energy intensively and that can negatively toxics into freshwater, and eutrophication affect the environment (such as unsustainable (Macknick and others 2012). Coal-fired extraction of groundwater and increased plants generate significant quantities of ash use of chemical fertilizers). Lower prices for that, if not managed correctly, can cause automotive fuels encourage higher vehicle environmental impacts such as leachates, use, leading to increased air pollution, storm water discharges, and contamination congestion, and road accidents. By keeping of groundwater and surface water (Hertwich energy prices artificially low, price subsidies and others 2014). Energy systems, including can also deter adoption of cleaner and more power plants and transmission and distribution efficient technologies (Parry and others 2014; lines, affect biodiversity through habitat loss Davis 2016). and fragmentation, which may permanently displace species, alter dispersal patterns, and In the area of household energy, energy price facilitate the introduction of new communities subsidies for gaseous fuels and electricity of species, including invasive species have the opposite effects with positive (Hernandez and others 2014). Frequently environmental effects. Millions of people still cited impacts caused by power plants and rely on traditional use of solid fuels, such as transmission infrastructure on communities wood, straw, crop residues, dung, and coal, range from resettlement to visual pollution and to meet their needs. The use of these fuels, negative effects on lifestyle, cultural values, coupled with inadequate ventilation, results or property (Geissler, Köppel, and Gunther in health-damaging concentrations of air 2013; Saidur and others 2011; Stemmer 2011). pollutants in indoor environments (WHO Other energy-related activities, including 2016). Price subsidies for gas, electricity, mining and accidents such as oil spills, can and district heating can reduce household have profound environmental implications. air pollution by encouraging households to substitute traditional use of these solid While recognizing that the environmental fuels for energy sources that are clean at effects of the energy sector are broad-ranging, the point of delivery (UNEP 2008). Similarly, this note focuses on the health effects caused in industrial, transport, and power sectors, by local air pollution, including both ambient price subsidies for gaseous fuels may reduce air pollution and household air pollution. Of the consumption of more polluting fuels. Natural varying positive, as well as negative, effects gas may substitute coal and oil products in of energy price subsidies on the environment, the power and industrial sectors, with net these health effects may be the largest in reductions in hazardous local air emissions. magnitude. An estimated 6.5 million people If the unit price subsidy is sufficiently large, die each year from air pollution (Cohen automotive LPG and natural gas may substitute and others 2017). This makes air pollution liquid automotive fuels, reducing particulate the fourth largest health risk factor in the emissions in the transport sector. world according to the Global Burden of Disease (GBD) study (GBD 2015 risk factors Energy production and use can have multiple collaborators 2016). Among air pollutants, fine environmental impacts. Electricity generation particulate matter or PM2.5 (particulate matter can affect water quantity and quality through with diameter up to 2.5 micrometers) affects consumption of vast amounts of water for human health the most because they are more cooling and other processes, discharge of toxic and can be breathed more deeply into 7 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES the lungs than other pollutants (Pope and compares these estimates across selected Dockery 2006). Incomplete combustion of national and subnational jurisdictions. fossil fuels and solid biomass is an important source of PM2.5 emissions and is responsible How much lower prices from subsidies increase for a large share of these deaths (IEA 2016). the consumption of the subsidized energy, Other sources of PM2.5 from fuel combustion and the extent to which higher consumption include emissions of oxides of sulfur (SOx) and in turn affects health are difficult to quantify oxides of nitrogen (NOx), which form so-called because of a number of factors. An important secondary (sulfate-based and nitrate-based) point to stress is that emission characteristics particles through chemical reactions in the of fuel combustion is a function of both the atmosphere. About 40% of the deaths are fuel properties and the technical state of the from household air pollution due to a lack equipment burning the fuel. This is particularly of access to clean household energy, clean true in the transport sector, where emission combustion technologies, or both, and 60% characteristics are a much greater function of are caused by outdoor ambient air pollution. the state of the vehicle than the fuel, especially in developing countries. Several analyses conducted by the World Bank found that the economic costs of the health Frequently raised policy questions in the effects caused by ambient and household context of energy price subsidies and air air pollution are significant at the national pollution are who benefits most from the and subnational levels. Per these studies, subsidies and who is affected by pollution. ambient air pollution had an average cost of The first question is discussed in part in 3.5% of gross domestic product (GDP) in five Good Practice Notes 3 and 4, both of which Asian countries and 2.5% of GDP in six Latin focus primarily on price subsidies captured American countries. Household air pollution by households. It is quite difficult to provide had a cost equivalent to 1% of GDP in many guidance to estimate the distributional impacts countries, and as high as 4.9% of GDP in the of outdoor air pollution within a whole country Lao People’s Democratic Republic. Figure 1 because the relevant variables vary widely from one location to another due to factors such as FIGURE 1: National and Subnational Level Comparisons—Cost of Environmental Health Effects Caused by Air Pollution % of GDP 6 Household Air Pollution 5 Outdoor Air Pollution 4 3 2 1 0 Bolivia* Argentina* Peru* Lao PDR* Piauí, Brazil** Hidalgo, Apurimac, Sindh, Mexico** Peru** Pakistan** Note: * = National-level estimate; ** = Subnational-level estimate. Sources: Larsen 2015, 2017a, 2017b; Larsen and Skjelvik 2013a, 2013b, 2014a, 2014b; Sánchez-Triana and others 2015. 2. METHODOLOGICAL APPROACH TO ASSESSING LOCAL EXTERNALITIES OF ENERGY SUBSIDIES 8 climate, topography, and urban development. this type of analysis and that most health This issue is therefore not addressed in this effects associated with energy price subsidies note. In the case of household air pollution, are usually caused by a few fuels and in a rural and poor households that cannot afford small number of sectors. Section 4 focuses modern energy sources are primarily affected, on the linkages between price subsidies and although in some low-income and lower- energy consumption, which are important middle-income countries, even the urban rich to assess what portion of the negative continue to cite solid fuels as their primary externalities caused by air pollution can be cooking fuels in household surveys. Women reasonably associated with the existence and infants face greater risks from indoor air of price subsidies. Section 5 discusses how pollution because they typically spend more energy consumption affects emissions of time near the sources of household air pollution air pollutants. Section 6 focuses on different (Smith and others 2014). methods to estimate human exposure to such pollutants, while section 7 centers on the This note is structured as follows. Section health effects resulting from said exposure. 2 explains the methodological approach to Section 8 describes methods to estimate assess the local externalities of energy price the economic value of the health effects. subsidies. Section 3 provides guidance to Section 9 presents available automated air prioritize the analysis, recognizing that scarce pollution health risk assessment tools. The resources are generally available to conduct note’s conclusions are presented in section 10. 2. METHODOLOGICAL APPROACH TO ASSESSING LOCAL EXTERNALITIES OF ENERGY SUBSIDIES Estimating the health effects of local air The complexity of estimating health effects of pollution arising from energy price subsidies energy price subsidies can be characterized can be a complex task. Complex tasks often call at five levels: for simplifications and approximations, both 1 | The effect of price subsidies on levels and in terms of modeling and data application. patterns of energy consumption (discussed Appreciating some of this complexity can in section 4 of this note) help understand sources and magnitudes of uncertainty in health estimates associated 2 | Air emissions from energy consumption with alternative methodological and data (discussed in section 5) options. In addition, it is useful to differentiate between situations in which simplifications 3 | Human exposure to air emissions and approximations are acceptable and result (discussed in section 6) in relatively small margins of error, and those 4 | Health effects of exposure (discussed in in which the complexity warrants detailed section 7) assessment to provide meaningful estimates of health effects. 5 | Monetary valuation of health effects (discussed in section 8) 9 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES In the absence of information on vehicle where ΔE is change in air emissions (metric stock characteristics, technologies of different tons/year); ΔF is change in fuel consumption combustion engines and boilers, the state of (metric tons/year), for the purpose of this their maintenance and operations (including note incremental fuel consumption that can vehicle driving patterns), and other requisite be attributed to price subsidies; e is fuel data, vastly simplifying assumptions have emission factor (metric ton of pollutant to be made to quantify the impact on emitted/metric ton of fuel consumed); health of changes in fuel consumption. The δD/δE is health effects (for example, deaths relationship between fuel consumption and per year) per metric ton of emissions; and V emissions may be highly nonlinear, as is the is the unit value of health effects. Because relationship between emissions and ambient of nonlinearity, the accuracy of equation 1 concentrations. The nonlinear relationship increases with diminishing changes (that is, as between consumption and emissions is the percentage changes in the parameters in seldom, if ever, taken into account. The the equation approach zero), and conversely extent of simplification means that margins the equation’s inaccuracy increases with of error are certain to be large. Each step— increasing change. Where price subsidies estimation of pollutant emissions from fuel are for electricity or district-heating tariffs, consumption, estimation of changes in the the change in electricity or heat consumption ambient concentrations of the pollutants from is first calculated, and these changes in turn changes in fuel consumption, estimation of need to be traced back to fuel consumption. changes in health parameters in response to This may not be straightforward. For example, changes in ambient pollutant concentrations, for grid electricity, with a handful of exceptions and finally monetization of health damage—is (such as Liberia), it is almost certain that complex, involving large assumptions. the power mix consists of several types of generation sources. How to calculate changes An approach adopted by many is to ignore in fuel consumption in response to changes in the foregoing factors affecting emissions, and electricity consumption is described in greater simply assume a linear relationship between detail in section 4, subsection a. fuel consumption and emissions of pollutants. Such an assumption is valid for carbon To capture the total health effects of price dioxide (in the absence of carbon capture subsidies, equation 1 would have to be estimated and storage), but not for PM2.5, the subject of 1 | For all fuels directly and indirectly affected this note, and other pollutants. Absent a very by the price subsidy to capture the effects large-scale study, however, it would be difficult of substitution among different energy to take account of various factors affecting sources; emissions of harmful pollutants, especially at the country level. If linearity is assumed 2 | For each user of fuel within each sector, at every step of the way, such a simplifying as fuel emission factors generally are user assumption leads to the following equation and sector specific; between a change in fuel consumption and its health effects in monetary terms (B): 3 | For each type of air emissions from fuel combustion; δD δD V, B = ∆E * * V = ∆F * e * * (1) δE δE 2. METHODOLOGICAL APPROACH TO ASSESSING LOCAL EXTERNALITIES OF ENERGY SUBSIDIES 10 4 | For each type of health outcome affected significant constraint, since on-the-ground by air emissions; and monitoring data networks are largely missing or inadequately operated and maintained in 5 | At small geographic scales, as health most developing countries. Another source of effects per metric ton of emissions uncertainty is the emissions for which health vary geographically in relation to effects have not been rigorously established. emission dispersion, ambient pollution concentrations, population density, and Each component of equation 1 is further baseline health conditions. discussed in the following sections to elaborate on the interactions among price Data constraints make the above level of subsidies, fuel consumption, emissions, health detail even for this very limited equation effects, and geographic scales. Table 1 provides practically impossible. Major data constraints an overview of the recommended steps to generally include country- and sector- quantify the environmental health effects of specific fuel emission factors. The dearth of energy price subsidies. local outdoor air pollution data is another TABLE 1: Quantifying the Effects of Energy Price Subsidies on Local Air Pollution and Health Steps Notes 1 Quantify • Focus on fossil fuels, electricity, and district heating. energy price • Quantify the difference between unsubsidized prices and the actual prices paid by subsidies consumers. • Calculate the unit price subsidy for each fuel in each sector, because price elasticities and fuel emission factors vary across fuels and sectors. 2 Estimate Assess the extent of energy shortages. If serious, the procedure below could grossly the impact overestimate energy consumption. If energy shortages are minor, then choose among the of price following tools: subsidies • Apply sector-specific own-price (and cross-price) elasticities of energy demand in partial on energy equilibrium if unit price subsidies are relatively “small”. consumption • For electricity, investigate if there is a power sector model that can be used to estimate which fuels are used more and by how much to meet the incremental power demand from lower electricity tariffs. Do the same for district heating if more than one fuel is used to generate heat. • Apply sector models if price subsidies are concentrated in a few sectors and unit price subsidies are relatively large. • Apply country-specific computable general equilibrium (CGE) models (if available) if subsidies prevail in most sectors and unit price subsidies are very large. • Apply models for road transport sector or motor vehicle fleets if unit price subsidies for automotive fuels are relatively large, because of the complex nature of vehicle emissions. 3 Estimate • Focus on PM2.5 emissions. impacts • Decide whether to include estimation of impacts on secondary PM2.5 (sulfates, nitrates). of energy • Establish fuel- and sector-specific emission factors for fuels and sectors impacted by consumption price subsidies. on emissions • Estimate impacts on emissions in spatial aggregations according to population density, exposure, and data availability. 11 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES Steps Notes 4 Estimate • Establish the prevailing outdoor PM2.5 concentrations in selected spatial aggregations health (using available monitoring data or satellite/chemical transport model estimates). effects of • Choose whether to estimate health effects by using an “intake fraction” approach or by changes in estimating the effect of changes in emissions on outdoor air quality. emissions • Apply generally accepted exposure-response functions (relative-risk functions) for estimating major health effects. 5 Estimate the • Establish a monetary value per unit of health effects, for example, the value of statistical monetary life (VSL) for premature mortality. value of • Multiply the unit monetary value by total health effects. the health effects 3. PRIORITIZING THE ANALYSIS In a majority of countries, the use of a few of fossil fuels, followed by industry. LPG and fuels in a limited number of sectors causes the natural gas use in the residential sector is most significant emissions of air pollutants and important for improving public health because their impacts on health. Where energy price the only affordable substitutes tend to be subsidies increase their consumption, or lead highly polluting solid fuels with severe health to greater use of polluting equipment, such effects of household air pollution. subsidies exacerbate the adverse effects on health. Where the subsidies promote a shift EMISSIONS away from polluting fuels to cleaner fuels, they can improve public health. Identifying As mentioned earlier, the air pollutant these fuels and sectors is a useful first step associated with the largest health effects at that can be carried out by mapping fossil national and global scales is PM2.5. The Global fuel consumption patterns from national or Burden of Disease Study 2015 (GBD 2015 risk subnational energy balances with subsidy factors collaborators 2016) estimated the levels, general patterns of emission intensities, health effects of outdoor ambient PM2.5 and and population exposure by sector. ozone, and reports that PM2.5 accounted for 92% and ozone for 8% of premature deaths FOSSIL FUEL CONSUMPTION (GBD 2015 risk factors collaborators 2016). PATTERNS Therefore, PM 2.5 is the air pollutant that first and foremost needs to be assessed in The power and heating sectors, including relation to energy price subsidies. NOx and combined heat and power, together make up sulfur dioxide (SO2) emissions contribute to the largest consumer of fossil fuels, representing secondary nitrate- and sulfate-based ambient 34% of global fossil fuel consumption in 2015. PM2.5, respectively, formed in the atmosphere Coal in 2015 accounted for 62% of all fossil from these emissions through chemical fuels consumed in the sector (IEA 2017). The reactions. Estimating the effect of subsidies transport sector, which primarily uses diesel on secondary PM2.5 is the next priority if data and gasoline, is the second largest consumer permit and reasonable estimates can be made. 3. PRIORITIZING THE ANALYSIS 12 A Fuel and Technology Perspective full-size buses and large trucks, always run on diesel fuel because diesel vehicles are Emission characteristics of combustion more robust, durable, and fuel-efficient, and depend on both fuel properties and the state conversely small motorcycles2 always use of the technical equipment used to combust spark-ignition engines. This means that the the fuel, including the technology employed. diesel fuel price has to remain significantly This is particularly true with pollutant below that of gasoline for years and more emissions from motorized vehicles, where likely decades before the vehicle fleet becomes it is imperative to treat fuels and vehicles as dominated by diesel-fueled engines, as in India. a joint system. Failure to do so can lead to incorrect assumptions, flawed conclusions, With the phaseout of lead in gasoline, sulfur is and misguided policies. For this reason, air the only automotive fuel property for which fuel pollution from transport is discussed in some alone determines the level of emissions. The detail below.1 level of SOx emissions is directly proportional to the sulfur content. Unlike stationary sources There are two types of automotive engines: burning fossil fuels, where SOx emissions spark ignition and compression ignition. can be controlled using scrubbers and other Vehicles fueled by gasoline, LPG, and natural means, there is no mechanism for reducing gas use spark-ignition engines, and those SOx emissions from vehicles. Sulfur occurs fueled by diesel fuel use compression-ignition naturally in crude oil and consequently is engines. Natural gas is generally in the form of found in both gasoline and diesel fuel unless it compressed natural gas (CNG), but liquefied has been reduced or removed during refining. natural gas (LNG) is used in large carriers LPG contains much less sulfur, and natural gas (large trucks and ships). Converting in-use contains even less. SOx contributes to acid gasoline vehicles to run on CNG is much rain and to the formation of secondary PM2.5. easier than converting in-use diesel vehicles At sulfur levels above about 500 parts per to do so, because the latter involves replacing million (ppm)—a level that is still prevalent in compression-ignition engines with spark- some developing countries—fuel sulfur causes ignition engines. For this reason, most CNG two problems. First, it acts as a poison for vehicles are conversions from in-use gasoline catalysts used in emissions control devices. vehicles. Second, once particulate emission levels are reduced to a fairly low level through vehicle By contrast, conversion from diesel to gasoline technology improvements, the composition or gasoline to diesel does not occur in in-use of PM 2.5 becomes dominated by sulfates vehicles. Instead, vehicle owners switch from rather than carbonaceous materials. The gasoline to diesel and vice versa only at the latter problem was the driver for reducing time of vehicle purchase. As such, gasoline the maximum sulfur level in diesel to 500 and diesel fuel are not substitutes in the short ppm in 1994 in the United States and in 1996 term—slashing the diesel fuel price through a in the European Union. Increasingly stringent large subsidy does not lead to an immediate vehicular emission standards in the subsequent large-scale substitution of diesel fuel for decades have called for correspondingly gasoline in the automotive sector. Further, advanced control devices, which are even the two fuels are never substitutes in certain more susceptible to sulfur poisoning. Today, vehicle categories—large vehicles, such as the sulfur limits on diesel fuel are 10 ppm in 13 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES the European Union and 15 ppm in United cylinder, with locally over-rich zones in the States, and the limits in gasoline are 10 ppm combustion chamber caused by higher fuel and 30 ppm, respectively. By contrast, the injection rates, dirty injectors, and injection limits in many developing countries for diesel nozzle tip wear. Overfueling to increase power fuel remain in the thousands of ppm. However, output, a common phenomenon worldwide, reducing sulfur to 10–30 ppm would be cost- results in higher smoke emissions and effective only if such fuel specifications are somewhat lower fuel economy. Dirty injectors also accompanied by introduction of vehicles are common because injector maintenance with advanced emissions control technology. is costly in terms of actual repair costs and Absent the latter, the extra costs incurred losses stemming from downtime. Adulteration in producing or importing ultralow sulfur with heavier fuels also increases in-cylinder fuels is unlikely to be justified. The variation deposits and fouls injectors. in emission levels as a function of vehicle would be expected to be much greater Among other significant contributors to in developing countries than in advanced particulate emissions historically has been economies with stringent fuel specifications inappropriate quantity and quality of and vehicle emissions standards, and a culture lubricants used in two-stroke engine vehicles of reasonable vehicle maintenance practice. fueled by gasoline. Two-stroke engine gasoline vehicles use gasoline blended with a lubricant. The emissions of all other pollutants— Two-stroke engine vehicles and boats, as carbonaceous PM2.5, NOx, carbon monoxide, well as equipment such as lawn mowers, carcinogens such as benzene, and ozone are common in some countries. As much as precursors such as olefins—depend as much 15–40% of the fuel-air mixture escapes from on the state of the vehicle technology and the engine through the exhaust port. These driving patterns as on fuel properties. Driving “scavenging losses” contain a high level of patterns affect emission levels significantly. unburned gasoline and lubricant. Some of the With the exception of NO x and SO x, the incompletely burned lubricant and heavier emissions of other harmful pollutants are portions of gasoline are emitted as small products of incomplete combustion. oil droplets, which in turn increase visible Combustion can be made more complete “white” smoke and particulate emissions. by supplying plentiful air and increasing These emissions are exacerbated by excessive the combustion temperature, both of addition or poor quality of lubricant. White which increase NOx emissions, presenting a smoke comprises mostly fine oil mist and tradeoff. In general, smooth highway driving soluble hydrocarbons, whereas the black minimizes the emissions of hydrocarbons smoke emitted by diesel vehicles contains a and PM2.5 and increases NOx emissions. By large fraction of graphitic carbon. The health contrast, stop-and-start traffic increases PM2.5 impact of white smoke is not well understood. emissions markedly. Traffic management can Two-stroke engine technology is being phased therefore help reduce particulate emissions out globally, and the relevant question is from transport. Particulate emissions can how widely the in-service two-stroke engines also increase substantially where engines operate, where, and for how much longer. are underpowered or poorly maintained or adjusted. Black diesel smoke results from Three-way catalytic converters have been used inadequate mixing of air and fuel in the for decades to control pollutant emissions 3. PRIORITIZING THE ANALYSIS 14 (other than SOx) in spark-ignition-engine This requires alternative means of reducing vehicles. These converters, when working emissions, which also come at varying costs properly, are extremely effective, although at to fuel economy. Smoking gasoline vehicles the expense of fuel economy (fuel efficiency notwithstanding, diesel-fueled vehicles on is sacrificed to reduce pollutant emissions, average emit much more PM2.5 than vehicles thereby increasing fuel consumption and operating on other fuels. carbon dioxide emissions). However, the catalysts become deactivated over time, not Although gaseous-fuel vehicles should be only from cumulative effects of long-term cleaner, CNG vehicles can be gross emitters exposure to fuel sulfur (although ultralow sulfur of NOx after conversion from diesel to CNG. fuels help), but also from leakage of lubricant in Combustion of lubricants also leads to PM2.5 ill-maintained vehicles and other contaminants emissions from vehicles fueled by gaseous into the fuel. Deactivated catalysts increase fuels. NOx is a product of combustion of air, the emissions of N2O, which is a greenhouse and is produced by all fuels. NOx is a precursor gas that is much more powerful than carbon to ozone formation and to secondary particles. dioxide. More worryingly from the point of For technical reasons, NOx emissions are more view of public health, gasoline vehicles with difficult to control in compression-ignition deactivated catalytic converters can emit engines than in spark-ignition engines. as much PM2.5 as (or even more than) diesel For stationary sources, fuel oil, diesel, and vehicles. A study in Colorado in the 1990s above all coal are significant contributors (Watson and others 1998) suggested that to PM 2.5 emissions. Especially damaging PM2.5 emission factors from gasoline vehicles is combustion of coal and diesel in small in grams (g) per kilometer (km) traveled dispersed sources, such as backup diesel were grossly underestimated because of generation sets—prevalent in many developing the prevalence of highly polluting vehicles countries with acute power shortages—and (Watson and others 1998). A study conducted coal used for cooking and home heating, as in southern California (Durbin and others 1999) in China, Mongolia, South Africa, and Turkey found that some gasoline-fueled passenger (where free coal has been distributed to the cars emit as much as 1.5 g/km, an emission poor). NOx emissions from stationary sources level normally associated with heavy-duty can be reduced using low-NOx burners, but diesel vehicles. Comprising only 1–2% of the control devices are absent in many applications light-duty vehicle fleet, these gross polluters in developing countries. SO2 emissions can be were estimated to contribute as much as high in the absence of flue gas desulfurization, one-third to the total light-duty particulate contributing to secondary PM2.5 formation. emissions. Such a problem is expected to be even more prevalent in developing countries, A Sector Perspective potentially making “smoking” gasoline vehicles account for a disproportionately high share of Although the transportation sector is visible total PM2.5 emissions from road transportation. and may appear as the largest source of PM 2.5 pollution, other sources have been Compression-ignition engines have far better found to be more significant in China and fuel economy because they burn “lean,” India, where more than one third of the with higher air-to-fuel ratio than vehicles world’s population lives. A global partnership equipped with three-way catalytic converters. 15 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES investigating the health effects of air pollution An important nonfossil-fuel source of high PM2.5 has found that combustion of solid fuels emissions is traditional use of solid biomass. accounted for the largest shares of health Biomass is seldom, if ever, subsidized, but is risks in the two countries. In China in 2013, coal available free of cost or at very low prices combustion in stationary sources accounted in many regions, especially in rural areas. for the largest share of population-weighted Substituting LPG and natural gas for these PM2.5 concentrations (and hence premature fuels would reduce emissions markedly, but deaths), constituting 40%. Industrial use of natural gas may not be available (and, barring coal alone accounted for 17%, followed by some parts of Eastern Europe and the former power generation and household use of coal. Soviet Union, is not available in rural areas even By sector, fuel combustion in industry was a in high-income countries), and LPG is typically larger contributor to population-weighted much more costly. An alternative is to use PM2.5 pollution (28%) than household use of electricity, but household use of electricity for solid fuels (coal and solid biomass) at 19% or cooking and heating is rare in many low- and transport emissions at 15% (GBD MAPS WG lower-middle-income countries. 2016). In India in 2015, residential biomass burning contributed to 24% of total exposure, The transportation sector, and specifically followed by coal combustion in industry and in road transportation, is a significant source power generation (7.7 and 7.6%, respectively), of human exposure to PM2.5. Because urban anthropogenic dust (8.9%), open burning of vehicle emissions are emitted near ground agricultural residues (5.5%), transportation level where people live and work, they are (2.1%), and nontransportation use of diesel especially damaging to public health. (1.8%).3 POPULATION EXPOSURE Large stationary sources, such as heat and power generation and large factories, use Population exposure to air emissions from coal, fuel oil, diesel, and natural gas. There are fossil fuel combustion depends largely on usually limits on pollutant emissions, but the two factors: 1) spatial dispersion of emissions, restrictions may be lenient, or monitoring and and 2) population density and distribution in enforcement may be weak. Small stationary the geographic area of emission dispersion. sources burning diesel and coal are also significant sources of exposure where they Source apportionment is an important are numerous. Diesel fuel is frequently used concept. Health effects are based on ambient for backup power generation in countries with concentrations of PM2.5, and policy responses unreliable grid electricity. Coal is used in boilers are driven by what is contributing to the and, where it is cheap, as household energy for elevated concentrations. This is one of the cooking and heating, as in China, South Africa, challenging areas in the science of air pollution and Turkey. Coal used by brick manufacturers and health. The complexity of atmospheric often employ traditional technologies with chemistry and nonlinearity between very high PM2.5 emissions. Small sources tend consumption and emissions and between not to have exhaust emission control devices, emissions and ambient concentrations add making emissions higher than those from to the difficulties. large sources using abatement technology. 3. PRIORITIZING THE ANALYSIS 16 Dispersion of Emissions because evolving technologies of combustion equipment (vehicles, boilers, stoves, Assessing emission dispersion and consequent generators, and so on), their use patterns impacts on air quality requires complementary (driving patterns, steady or intermittent), and, types of tools. Three frequently used importantly, how well they are maintained approaches—emissions inventories, dispersion are unlikely to be captured for lack of data. models, and chemical mass balance (CMB) receptor models—are described below. The state of California in the United States has one of the most comprehensive methodologies Emissions inventories provide a snapshot of to develop air emissions inventories. Emissions the amount of pollutants discharged into from stationary sources are estimated the atmosphere from within a geographic based on the California Air Toxics Emission area (for example, a metropolitan area or Factor (CATEF) database, which contains country) during a specific time period (such approximately 2,000 air toxics emission factors as one year). Emissions inventories include calculated from source test data collected data from multiple sources, which can be through emission measurements in the early classified as follows: 1990s.4 These emission factors are more than two decades old, not having been updated • Stationary or fixed pollution sources, such since 1996. For mobile sources, the California as power plants and factories Air Resources Board (ARB) developed an • Mobile sources, including on-road sources EMission FACtors (EMFAC) model that such as cars, motorcycles, buses, and trucks, calculates emissions inventories by multiplying and off-road sources, including farm and emissions rates with vehicle activity data from construction equipment, trains, and marine all motor vehicles, including passenger cars vehicles to heavy-duty trucks, operating on highways, freeways, and local roads in California. The • Areawide sources comprising emissions most recent version is dated 2014. Similar spread over extensive regions, such as road models are also used to estimate emissions dust, fireplaces, and architectural coatings from off-road vehicles.5 Areawide source • Natural sources, such as wildfires, windblown methods are used to estimate emissions for dust, and emissions from plants and trees approximately 500 categories of emission sources in the emission inventory. The index In general, given the difficulties of obtaining of methodologies by major category includes a direct measurement from all sources, summaries of the methodologies with links to anthropogenic emissions are estimated by the complete methodologies, including fuel using emission factors, or the average rates combustion, waste disposal, cleaning coatings, of emissions of pollutants per unit of activity petroleum production, industrial processes, data for a given sector, which are, in turn, and solvent evaporation.6 obtained from statistics or surveys. Country- specific emission factors provide more reliable The Global Atmospheric Pollution Forum Air results. In the absence of country-specific Pollutant Emission Inventory Manual provides data, default emission factors obtained from a simplified and user-friendly framework for other countries may be used. The use of preparing an emissions inventory that is emission factors introduces large uncertainties, suitable for use in different developing and 17 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES rapidly industrializing countries and that is b. COMPLEX1 is a screening technique compatible with other major international for multiple point sources in complex emissions inventory initiatives. It covers terrains. multiple air pollutants, including PM2.5 and c. Rough Terrain Diffusion Model PM 10. A spreadsheet workbook has been (RTDM3.2) is designed to estimate prepared as a companion to this manual for ground-level concentrations in rough use as an aid and tool in preparing national (or flat) terrain in the vicinity of one or emissions inventories.7 more point sources. Dispersion models are used to understand d. SCREEN3 provides maximum ground- how pollutants travel and disperse in the air, level concentrations for point, area, and can be used to predict concentrations in flare, and volume sources, as well as a downwind location. They complement air concentrations in the cavity zone, and quality monitoring, for example, by estimating concentrations due to inversion breakup air quality in locations where monitoring data and shoreline fumigation. do not have the necessary spatial or temporal e. VALLEY is designed to estimate 24-hour coverage. They are also used to estimate the or annual concentrations resulting from effects of actions such as the operation of new emissions from up to 50 point and area emission sources that do not yet exist or the sources. introduction of emission controls for existing f. VISCREEN calculates the potential sources. Dispersion models can be grouped impact of a plume of specified emissions into three main categories (BCME 2015): for specific transport and dispersion 1 | Screening models are relatively simple conditions. estimation techniques that generally 2 | Refined models incorporate more detailed use preset worst-case scenarios to descriptions of atmospheric processes provide conservative estimates of the with the aim of providing more reliable air quality impact or a specific pollution estimates of the concentration of pollutants source or category. They can be used to in a specific site, including variations in identify sources that do not contribute space and time. However, refined models meaningfully to air pollution and that generally require more detailed and should, therefore, be excluded from more precise input data. Model input consists elaborate, resource-intensive modeling. of geophysical data such as terrain and Listed below are screening models and surface roughness, user-defined receptors, some of the conditions under which their and a sequential, hourly time series of use would be preferred (EPA 2005): meteorological data that are representative a. AERSCREEN will produce estimates of the conditions at the location of the of the worst-case concentrations for source. The U.S. Environmental Protection a single source for time periods of 1, Agency (EPA) has conducted one of the 3, 8, or 24 hours, or one year. Its main most thorough reviews of air quality advantage is that it does not require models that can be used to assess key hourly meteorological data. air pollutants (EPA 2005). Based on the review, it recommends the use of the following two refined models:8 3. PRIORITIZING THE ANALYSIS 18 a. The AERMOD Modeling System dispersion models complement each other. incorporates air dispersion based on CMB helps explain observations that have the turbulence structure of the lowest been made but does not predict ambient part of the atmosphere and is suitable impacts from sources, as do dispersion for surface and elevated sources, and models. Local emissions inventories for both simple and complex terrain. also complement CMB receptor models AEROMOD is designed for short range because documenting the location and dispersion (up to 50 km) and is a magnitude of all sources surrounding steady-state plume model, meaning a receptor enables the identification of that it assumes that emissions from major source types that are likely to have point sources diffuse (that is, move the largest impact on air quality. from areas of high concentration to areas of low concentration) maintaining Data needed to conduct CMB modeling the same distribution of the substance include (a) source categories, (b) chemical over time. composition or profile to be associated with each source category, (c) uncertainty in b. The CALPUFF Modeling System the chemical composition of each source simulates the effects of time- and space- category, (d) chemical composition of the varying meteorological conditions on fallout particles sampled at a receptor, and pollutant transport, transformation, (e) uncertainty in the receptor chemical and removal. CALPUFF can be applied composition. EPA-CMBv8.2, a CMB receptor to long-range transport and complex model, is one of several receptor models that terrain, and is a non–steady state plume has been applied to air quality problems over model, meaning that it assumes that the last two decades. CMB requires profiles the concentration of pollutants changes of potentially contributing sources and the with time. corresponding ambient data from analyzed 3 | CMB receptor models are complementary samples collected at receptor sites.9 models used to estimate the average contribution of specific sources of Exposure pollutant emissions to particulate fallout. Exposure is determined by the number Weather, wind, geography and other of people exposed and ambient pollutant factors affect how pollutants travel, concentrations. The higher the density of disperse, and mix. Therefore, it is generally people exposed and the higher the ambient difficult to establish a direct correlation concentration levels, the greater the exposure between source emission and pollution and hence the greater the health damage. concentrations in the environment. CMB They are both location- and time-specific. receptor models help to overcome this challenge by measuring the concentrations Distributed ground-level emission sources, of different pollutants at a specific location such as road vehicles in urban areas, are and comparing them with the composition among the largest sources of human exposure patterns of emission from different to particulate air pollution. At the opposite sources, which are distinct enough to end of the spectrum are emission sources be identified. CMB receptor models and in thinly populated areas. Generally, for 19 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES outdoor sources of PM2.5 pollution, mobile In terms of fuel characteristics, the higher and stationary sources in urban areas merit the density of the fuel, the higher the sulfur greater attention than sources in peri-urban level and carbonaceous emissions. Coal areas, while sources in rural and other remote has the highest density, followed by fuel oil areas are likely to have the least adverse health (including marine fuel oil), diesel, gasoline, effects because emissions are dispersed over LPG, and finally natural gas. As mentioned large geographic areas that often have low above, fuel characteristics are not the sole population densities. determinant of emission levels. Mobile and large stationary sources tend to be equipped For coastal populations, emissions from ships with emissions control devices, although are an increasing concern. For this reason, standards vary greatly from country to in October 2016, the International Maritime country, and monitoring and enforcement Organization announced that, after a careful vary just as greatly. Operational patterns also review, it had set a global limit for sulfur in fuel affect emission levels significantly. As a result, oil used on board ships of 0.5%—down from emission factors can vary by several orders 3.5% today—from January 1, 2020, for health of magnitude from vehicle to vehicle even and environmental reasons. This dramatic within the same vehicle category, and more reduction in sulfur in fuel oil will reduce air generally from source to source. pollution from sulfate-based PM2.5. 4. ENERGY CONSUMPTION EFFECTS OF PRICE SUBSIDIES Energy price subsidies are intended to lower air pollution and have positive effects on prices changed to consumers to make them public health. Energy price subsidies may also more affordable. If they are implemented and lead to intersectoral or economywide changes operate as designed, such price subsidies in production and consumption, as the would deliver artificially low prices and subsidies affect relative prices in production increase consumption of the subsidized and consumption. For example, energy price energy. In practice, consumption may not subsidies may encourage growth of energy- be as high as what would be expected on intensive industries and a contraction of paper—price subsidies may cause widespread industries that are not energy-intensive. shortages of the subsidized energy, higher prices actually paid, or both. Where these There are several possibilities for supply unintended consequences are largely absence, constraints and adherence to official higher consumption of polluting fuels would subsidized prices: aggravate air pollution, and correspondingly • Energy supply constraints and rationing higher consumption of clean forms of energy lead to a disequilibrium with excess or substituting polluting fuels—subsidized natural unmet energy demand at subsidized prices. gas replacing coal and solid biomass for If available energy is sold at official prices, household energy, for example—would reduce which is typically the case for energy 4. ENERGY CONSUMPTION EFFECTS OF PRICE SUBSIDIES 20 distributed through networks (natural gas, This note focuses on cases where there are no district heating, and grid electricity), energy supply constraints and all consumers are able consumption equals the supply constraint. to purchase energy at the official subsidized If available energy is sold at prices much prices. higher than official prices, which occurs with liquid fuels, then the supply curve CHOICE OF ANALYTICAL MODEL shifts and demand is reduced. In almost all cases, the prices paid follow a distribution For analytical purposes, the consumption curve, from official prices to much higher effects on energy of energy price subsidies prices depending on time, location, and can be categorized as direct and indirect who the purchaser is. For example, the poor effects (table 2). Different analytical tools may have less access to subsidized fuels are available to capture each of these effects, than the better-off or the politically well- with an increasing level of complexity and connected. In both cases, price elasticities data requirements. The level of analysis must, cannot be applied as long as there are therefore, be carefully selected in light of the supply constraints, or reliable information size of energy price subsidies, substitutability on prices actually paid is not available. among different forms of energy, and the importance of the sector in terms of energy • There are no supply constraints on consumption, air emissions, and health subsidized energy and energy consumption effects. Good Practice Notes 3 and 7 provide equals energy demand at subsidized prices. more detailed guidance on the analysis of economywide effects. TABLE 2: Energy Consumption Effects of Energy Price Subsidies Effects of energy subsidies Analytical tools Direct effects Own energy demand Own-price elasticity of energy demand Cross-price elasticities of energy demand Substitution among energy sources Indirect Input/output model, macrostructural model, computable effects Effects on goods and services general equiligrium (CGE) model, dynamic stochastic using subsidized energy as input general equilibrium model, sector-specific model For electricity and district heating, after population growth, economic growth, estimating incremental consumption from evolution of appliances, and other relevant price subsidies, the calculations have to be parameters. traced back to incremental fuel consumption. 2 | Take, or in its absence develop, a least-cost This requires several steps, particularly in power development plan, minimizing the countries with growing demand, which is net present values of costs of investment, the case in almost all developing countries. operation, and unserved energy. This The steps below illustrate how to deal with requires assumptions about options to electricity as an example. expand supply. If incremental demand 1 | Using an econometric model or any is met largely by sources without other suitable model, develop a demand- air pollutant emissions (solar, wind, forecast model based on electricity prices, hydropower, geothermal, or nuclear), the 21 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES impact on local air pollution may be very to incremental power generation, and small. Similarly if the additional supply incremental emissions will also likely be comes from electricity imports, depending correspondingly lower. on the level of the regional impact of air pollution from electricity generation in PRICE ELASTICITIES OF ENERGY the exporting countries, again the impact DEMAND on the importing country may be very small. If more electricity can be supplied There is a large body of empirical estimates by reducing technical losses, incremental of price elasticities of energy demand, and demand may be met without increasing several meta-analyses of these studies have fuel consumption markedly. Lastly, parallel been carried out (Espey 1996 and 1998; Hanly, actions by the utility, such as reducing Dargay, and Goodwin 2002; Graham and commercial losses, would reduce demand, Glaister 2002; Espey and Espey 2004; Brons partially or even fully off-setting incremental and others 2008; Havranek Irsova, and Janda demand from power price subsidies. 2012). 3 | Estimate incremental consumption of fuels A recent paper by Labandeira, Labeaga, and in the power sector based on steps 1 and 2. López-Otero (2016) performs a meta-analysis of papers produced between 1990 and 2014 4 | Estimate emission factors reflecting the with 903 short-term price elasticities and 941 characteristics of the generation fleet long-term price elasticities of energy demand. and calculate incremental emissions. Average elasticities range from -0.2 to -0.26 Emission factors depend on the fuel in the short term and from -0.6 to -0.85 in the type and characteristics, generation and long term for overall energy demand and five abatement technologies, the state of individual energy products. The authors find maintenance and repair, and operational somewhat larger elasticities for residential characteristics, including the load factor and commercial consumers than for industrial (percentage of the installed capacity the consumers and somewhat larger elasticities plant runs). For example, if incremental in developing than in developed countries. consumption comes from increasing the The largest elasticities are for natural gas load factor, fuel efficiency will likely rise, and heating oil, and the smallest is for diesel fuel consumption will not be proportional (table 3). TABLE 3: Average Price Elasticities of Energy Demand Short-term Long-term Generalized least Generalized least Type/model Random-effects panel Random-effects panel squares squares Energy -0.220 -0.224 -0.600 -0.652 Electricity -0.231 -0.209 -0.677 -0.686 Natural gas -0.239 -0.216 -0.736 -0.850 Gasoline -0.249 -0.227 -0.720 -0.715 Diesel -0.213 -0.204 -0.620 -0.595 Heating oil -0.242 -0.259 -0.747 -0.764 Source: Labandeira, Labeaga, and López-Otero 2016. 4. ENERGY CONSUMPTION EFFECTS OF PRICE SUBSIDIES 22 The long-term own-price elasticities of energy There are recent global studies of the cost of demand in table 3 may be used to estimate energy price subsidies that can serve as a the energy consumption effects of removal reference. Davis (2016) uses a long-term own- of price subsidies. The elasticities are valid price elasticity of -0.6 for automotive gasoline for marginal changes in energy prices. Larsen and diesel fuels. Coady and others (2015) use (1994) therefore applied own-price elasticities long-term own-price elasticities of -0.5 for of -0.6 in countries with low subsidy rates oil products and electricity and -0.25 for coal and -0.15 to -0.25 in countries with very high and natural gas, and Parry and others (2014) subsidy rates. Price elasticities can be adjusted use -0.5 for all fuels for estimating the cost as appropriate in light of individual country of post-tax energy price subsidies. Cross-price evidence and subsidy rates. Box 1 provides elasticities, sector models, or CGE models are an example of a study that estimated price not used in these studies. and income elasticities to model the effect of gasoline price subsidy removal in Mexico. SUBSTITUTION AMONG ENERGY SOURCES The effect of energy subsidies on energy demand in a given sector is estimated by The use of cross-price elasticities can be pi -εi important where there is substantial scope ∆qi = -qi [1- ( )piw ], (2) for substitution among energy sources. In motorized road transport, heavily subsidized where q i is consumption of energy i at diesel fuel may substitute gasoline in light- subsidized price pi; piw is unsubsidized price duty vehicles at the time of vehicle renewal. of energy i; and ɛi is a long-term constant In the extreme, diesel fuel may be used even own-price elasticity of demand for energy i. in motorcycles, as in India. Fiscal incentives and large price subsidies for automotive LPG BOX 1: USING INCOME AND PRICE ELASTICITY TO ESTIMATE GASOLINE CONSUMPTION REDUCTION FROM PRICE SUBSIDY REMOVAL IN MEXICO The Government of Mexico (GoM) started subsidizing fuels in 2005. At their highest point, in 2011, they amounted to 150 billion pesos (about US$11 billion). In addition to being costly, subsidies were regressive: about 59% of the total subsidy was transferred to the richest 20% or the population, compared with only 3% to the poorest 20% of the population. Recognizing that subsidies were not an efficient use of public resources, the GoM started in 2010 a consistent, but gradual subsidy phaseout consisting of monthly gasoline price hikes. In 2015, when the subsidy was close to zero, the GoM announced the decision to discontinue the monthly adjustments for gasoline and diesel fuel price subsidies. To estimate the effects of subsidy removals, Montes de Oca and Muñoz-Piña (2016) developed a model combining cointegration techniques and error correction models to estimate the short and long term price and income elasticity of high and low octane gasoline in Mexico. Their econometric model used national data on gasoline consumption and prices, the vehicle stock, GDP, population, employment, vehicle fleet efficiency, and public transportation prices. The analysis found that phasing out of fossil fuels in Mexico resulted in savings of 11 billion liters of gasoline and avoided emissions of 26 million metric tons of CO2. Source: Montes de Oca and Muñoz-Piña 2016. 23 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES and natural gas may promote substitution natural gas, fuel oil, and diesel fuel can be used of gasoline and diesel in most vehicle types, for both baseload and peaking. However, fuel including three-wheelers. The economic oil and diesel fuel are expensive, and are used driver for these substitutions stems from for baseload power generation only when lower fuel costs more than making up for the other options are not available, such as small higher vehicle purchase prices or the cost of island economies with few other options. Over conversion from a liquid to a gaseous fuel. the long term, least-cost, systemwide power More recently, electric vehicles are entering the development planning should determine the market in an increasing number of countries power mix based on a number of parameters, competing with vehicles powered by fossil one of which is the fuel cost. fuels. In the residential sector, fuel substitutability In industry, coal, natural gas, fuel oil, and depends on the purpose of use, income diesel are substitutes over the medium to long level, infrastructure availability (for example, term, and again the fuel price is an important piped natural gas, electricity grid, and determinant of that choice. In the power district heating), and reliability of supply (for sector, the choice depends not only on costs example, grid electricity versus captive diesel but also dispatch characteristics. Coal and generators, reliability of LPG refills) (table 4). nuclear power are for baseload, whereas TABLE 4: Examples of Substitution among Energy Sources by Sector Sector Energy source options Motorized road Gasoline, diesel, CNG, LPG, electricity transport Industry Coal, natural gas, fuel oil, diesel fuel, electricity Electricity Coal, natural gas, fuel oil, diesel fuel, hydropower, biomass, solar, wind, nuclear, geothermal production LPG, natural gas, electricity, kerosene, coal, or biomass for cooking Grid electricity, kerosene, LPG, solar lanterns, solar panels, batteries, candles, diesel generators, or gasoline generators for lighting Residential Electricity, LPG, or natural gas for cooling Electricity, district heating, natural gas, kerosene, LPG, coal, or biomass for heating Grid electricity, solar panels, batteries, diesel generators, or gasoline generators for electric appliances Understanding the potential for substitution The presence of large price subsidies for is one consideration for policy making. From automotive fuels may warrant the use of a the point of view of protecting public health, transport model to estimate effects on fuel the aim is to shift from highly polluting fuels, demand and air emissions. Such models can such as solid biomass, coal, fuel oil, and diesel better incorporate differential effects of price fuel, to cleaner forms of energy at the point subsidies and their removal on transport of delivery (such as grid electricity, district modal choice, vehicle users (by vehicle type, heating, natural gas, LPG, and solar panels). age, and usage), and the vehicle fleet turnover. 5. HIGHER AIR EMISSIONS FROM ENERGY PRICE SUBSIDIES 24 Such models can better capture the changes cited as a useful first-order estimation: 20% in the behavior of vehicle owners with the of vehicles cause 80% of vehicle pollution. most polluting vehicles. The 20/80 rule is often 5. HIGHER AIR EMISSIONS FROM ENERGY PRICE SUBSIDIES Relative volumes of fuel consumed depend, MOTORIZED ROAD TRANSPORT amongst others, on relative prices, to which price subsidies contribute. Emissions of local Where automotive fuels are subsidized, air pollutants (per unit of energy) from fossil estimating PM2.5 air emissions from the road fuels vary greatly in relation to the type of transport sector is important. Combustion of fuel, combustion technology, and emission both gasoline and diesel, especially if emissions control technology. This is the case for primary control devices are elementary or deactivated, PM2.5 as well as NOx and SO2 precursors to contribute to primary particulate formation, secondary PM2.5. while NOx and SO2 emissions contribute to secondary particles. The task of estimating For the reasons cited in section 3 on Dispersion incremental emissions is challenging. Of the of Emissions, it is important to treat fuels three, only SO2 emissions are determined and combustion-equipment technology as solely by fuel characteristics, increasing a joint system. Pollutant emissions per unit linearly with increasing fuel sulfur content. of fuel consumed vary from fuel to fuel and Particulate emissions are increased if lubricant application to application. Mercury emissions is mixed with the fuel through leakage or are specific to coal, and primary PM2.5 largely in two-stroke engine vehicles. Only electric to liquid and solid fuels. The same fuel can vehicles are emission free. Gaseous fuels have emission factors that vary by orders of contribute to NOx emissions, and even to small magnitude depending on the technology of levels of particulate and SO2 emissions. Other the equipment used to combust it, how it is sources of ambient PM2.5 from the sector are operated, and how it has been maintained. resuspended road dust, which is unrelated to Ultra-low-sulfur diesel fuel burning in a well- fuel characteristics, and particulate emissions maintained vehicle with state-of-the-art from breaking and other nonfuel vehicle exhaust control devices can be as clean as sources of PM2.5. As fuel price subsidies affect a natural gas vehicle, but heavily polluting in both total fuel consumption and vehicle usage, an old overloaded truck with a dirty injector all these sources of PM2.5 are affected. and leaking lubricant or an old backup power generator. Similarly, a new coal-fired power Emissions from new diesel vehicles illustrate plant meeting the most recent directive for the above point. PM2.5 emissions from diesel limiting emissions in the European Union for vehicles depend on the vehicle technology, 2021 will be much cleaner than uncontrolled which depends in part on fuel properties, the coal-fired power plants with no control devices most important of which is the level of sulfur. for emissions. The evolution of limits on particulate emissions 25 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES on new vehicles in the European Union over emissions are a function of the driving cycle, the last two decades show how emissions and even a brand new vehicle may emit have declined by about 30-fold (tables 5 more under more aggressive driving cycles and 6). To enable adoption of advanced (characterized by stop-and-start driving with emission control devices, the level of sulfur rapid acceleration). Many developing countries in diesel fuel has correspondingly decreased import secondhand vehicles, and vehicle from 2,000 ppm to 10 ppm. In-use vehicles maintenance practice also tends to be weaker. will have higher emissions, especially those A number of countries still allow sulfur levels with deactivated control devices. Further, in excess of 2,000 ppm. TABLE 5: European Union Light-Duty Diesel Vehicle Emission Standards for PM (g/km) Corresponding Light commercial vehicles (LCV) by weight sulfur level (ppm) class 1–3 Standard* Passenger vehicles LCV (1) LCV (2) LCV (3) Euro 1 (1992/94) 2,000 0.14 0.14 0.19 0.25 Euro 2** (1996/98) 500 0.08 0.08 0.12 0.17 Euro 3 (2000/01) 350 0.05 0.05 0.07 0.10 Euro 4 (2005/06) 50 0.025 0.025 0.04 0.06 Euro 5 (2009/10) 10 0.005a 0.005a 0.005a 0.005a Euro 6 (2014/15) 10 0.005a 0.005a 0.005a 0.005a Notes: Vehicle classes LCV (1) =< 1,305 kg; LCV (2): 1,305–1,760 kg; LCV (3) > 1,760 kg. * The earlier year is for passenger vehicles and LCV (1). The later year is for LCV (2–3). ** Applicable for indirect-injection engines. Slightly less stringent limits apply for direct-injection engines. a: 0.0045 g/km using the particulate measurement program procedure. Source: Adapted from www.dieselnet.com/standards/eu/ld.php. TABLE 6: European Union Heavy-Duty Diesel Engines Emission Standards for PM (g/kWh) Tier Year PM 1992, < 85 kW 0.612 Euro I 1992, > 85 kW 0.36 1996 0.25 Euro II 1998 0.15 Euro III 2000 0.10* Euro IV 2005 0.02 Euro V 2008 0.02 Euro VI 2013 0.01 Note: kW = kilowatts, kWh = kilowatt-hours. * 0.13 g for engines of less than 0.75 cubic decimeters swept volume per cylinder and a rated power speed > 3,000/minute. Source: Adapted from https://www.dieselnet.com/standards/eu/hd.php. 5. HIGHER AIR EMISSIONS FROM ENERGY PRICE SUBSIDIES 26 The simplest approach to estimating the effect nearly 3 million deaths per year (GBD 2015 risk of price subsidies for automotive fuels on PM2.5 factors collaborators 2016). In low- and lower- emissions is to apply an average PM2.5 emission middle-income countries, price subsidies for factor. This can be approximated from a profile LPG, natural gas, or biogas large enough to of the diesel vehicle fleet by type of vehicle shift a significant number of households away (passenger cars, light-duty vehicles, heavy- from traditional use of solid fuels for cooking duty vehicles, buses, and trucks), vehicle and heating will have the largest effects on use by type, the age distribution of vehicles, air pollution and health10 of all energy price and any information available from emission subsidies in the residential sector. However, testing and emission standards. If information the magnitude of price subsidies needed to about the latter is not available, then emission effect fuel switching on a scale that would factors from countries with similar diesel deliver measurable health benefits is beyond vehicle characteristics can be applied. the means of virtually all governments. It should be noted, however, that owners of To the extent that price subsidies lead to different types and age of diesel vehicles will lower use of solid fuels, the benefits depend respond differently to fuel price subsidies. As on the degree of fuel switching, because a result, the effect on PM2.5 emissions from households are known to “stack” fuels—using the sector may differ from that indicated multiple fuels even as they shift away from by the average emission factor. Only more solid fuels—rather than climb up a fuel ladder, sophisticated modeling can capture the as previously thought (Masera, Saatkamp, and direction and size of this difference. Kammen 2000). If they continue to use solid fuels, the benefits of adding gaseous fuels CNG is increasingly used as a motor vehicle may be greatly diminished. Fuel stacking in fuel in many countries. Price subsidies to CNG turn can make the relationship between fuel are likely to be positive or relatively neutral at consumption and PM2.5 concentrations highly worst for local air pollution. In high-income nonlinear. countries, where gasoline vehicles are very clean, the environmental benefits may be very Similarly, subsidies for district heating to make small unless CNG is replacing high emitters, it affordable may shift households away from which is unlikely since high emitters tend to burning coal or solid biomass for heating, be old vehicles and there is no economic again reducing household air pollution. If the case for conversion from gasoline to CNG. In heat generation plant providing heat is fueled developing countries where high emitters are by coal and is located near population centers, much more prevalent, substituting gasoline however, outdoor air pollution may offset for CNG could bring measurable benefits. some of the benefits of reduction in indoor air Substituting CNG for diesel is likely to bring pollution. Price subsidies for electricity that measurable benefits in almost all cases. shift households away from solid fuel use for cooking and heating, or from captive diesel RESIDENTIAL SECTOR generators, would also reduce air pollution and improve health outcomes. A methodology for Household air pollution, which is prevalent estimating health effects of such subsidies is especially among low- and lower-middle- presented in section 8. income households, is estimated to cause 27 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES INDUSTRY and as a result power shortages are common. In such circumstances, price subsidies may Primary PM 2.5 emissions from petroleum merely determine the distribution and size of product consumption in the industry can be the unmet demand. Increased production to calculated from standard emission factors that meet incremental consumption may come do not exhibit the large variations observed from any electricity sources, fossil fuels or among motorized vehicles. Primary PM 2.5 otherwise, depending on what will provide emissions from natural gas are minimal and the marginal supply. If new capacity has to be can be ignored for analytical purposes. Primary added, so-called build-margin grid-emission PM2.5 from coal combustion will depend on factors (emission factors of new generation the industrial sector in which coal is used capacity) may and are likely to be different from and any abatement technology used in it. operating-margin factors (emission factors Any available in-country studies should be of existing generation capacity). If the power used to inform the analysis. In their absence, sector in the country in question does not emission factors from similar sectors in other have build-margin emission factors, options countries can be applied. being proposed by a group of multilateral development banks may be considered.11 The contribution of secondary PM2.5 from oil products, natural gas, and coal may be Existing studies can be used to estimate substantial if industrial fuel consumption is operating-margin PM2.5 emission factors. If large. This may be estimated as a sector share such studies do not exist, emissions can be of ambient secondary PM2.5, with ambient approximated from known fuel and plant secondary PM2.5 estimated from apportionment characteristics, such as the type of fuel studies discussed in the next section. (for example, coal or natural gas) and its characteristics (for example, ash and sulfur ELECTRICITY content of coal); the type of plant (open cycle or combined cycle gas turbine, steam As Good Practice Note 1 explains, the fact that boiler, reciprocating motor, fluidized bed); electricity tariffs do not recover costs does operating characteristics (baseload, peaking, not automatically imply the presence of price load factor); and operation and performance subsidies. To the extent that price subsidies of any particulate, SO2, and NOx emission exist (which is the case if there is underpricing, abatement technology installed. For build- as defined in Good Practice Note 1), lower prices margin emission factors, manufacturers’ would lead to higher electricity consumption in specifications and estimated performance the absence of constraints on supply. However, deterioration may be used. almost all developing countries face rising demand for electricity. In low- and lower- To help the poor consume electricity, middle-income countries in particular, where subsidized connection fees and volume- per capita electricity consumption is much differentiated tariffs may be offered. They lower than that in high-income countries and are effective, provided the lifeline block is the growth rates of electricity consumption kept relatively small to match subsistence- are correspondingly higher, the power level consumption. As Good Practice Note 1 infrastructure has frequently suffered from explains, increasing block tariffs by contrast years and even decades of underinvestment, may benefit the nonpoor disproportionately 6. POPULATION EXPOSURE ASSESSMENT 28 because every consumer, however rich, electricity would displace kerosene for lighting, benefits from the lifeline rate. This is especially but the poor may continue to cook and heat so if the electrification rate is relatively low, with solid fuels. those with access to grid electricity are primarily the better-off in urban areas, and Switching from diesel generators to grid the first block (sold at the lifeline rate) is electricity would decrease local air pollution. relatively large. The same concepts apply The decision to use diesel generators is driven to the other two forms of network energy, by the consumer’s assessment of power supply natural gas and district heating. In all cases, reliability. If price subsidies are contributing targeting requires accurate metering of every significantly to power outages, removing household. This requirement is not met in such subsidies would contribute to increasing many countries, where multiple houses are reliability, but that would take time. Because connected to a single meter, households are poor reliability is caused by a number of billed according to estimated consumption, factors and not just by price subsidies, it may meters are not accurate, or any combination be difficult to disentangle the contribution of of these shortcomings. price subsidies to power outages. Because the poor consume little electricity, Subsidies for solar, wind, and geothermal their impact on the country’s overall electricity power should reduce emissions, although if consumption will likely be very small. In most the subsidies make coal more attractive than low- and lower-middle-income countries, natural as, as has happened in recent years where household use of solid fuels is prevalent, in some European countries, the policy may electricity is not used for cooking, especially backfire and increase coal consumption at the among the poor, and hence the health benefits expense of natural gas, with a net increase in of providing electricity will also likely be small: pollutant emissions. 6. POPULATION EXPOSURE ASSESSMENT Estimating population exposure to air pollution MAPS applies the GBD project methodology due to incremental emissions from energy to estimate health effects.12 The program is price subsidies is arguably the most elaborate, estimating the disease burden due to outdoor data-intensive, and time-consuming task. air pollution from coal burning and other Two broad approaches may be used for this major sources in China (nationally and by purpose: (a) dispersion modeling, and (b) province), India, and Eastern Europe using intake fractions. the GBD framework. DISPERSION MODELING The study in China used the chemical transport model GEOS-Chem and calculated the An example of the use of dispersion modeling contributions of coal combustion, industry to estimate the health effects of air pollution is (noncoal), transportation, domestic biomass the program, “Global Burden of Disease from burning, and open burning to population- Major Air Pollution Sources (GBD MAPS).” GBD weighted ambient PM2.5 concentrations at the 29 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES national and provincial level. Coal combustion the population. The larger the intake fraction, was evaluated from power plants, industry, and the larger the health effects per metric ton domestic coal use.13 The study also estimated of emissions. the health effects of PM2.5 air pollution from various sources, using the health-risk functions Apte and others (2012) estimated the intra- from the GBD project (GBD MAPS WG 2016). urban intake fraction of distributed ground- This approach and model can in principle level emissions of primary pollutants in be applied to assess the health effects of more than 3,600 cities worldwide with a energy price subsidies once changes in population greater than 100,000. Intake energy consumption and emissions from fractions were based on location-specific price subsidies are estimated. geographic, meteorological, and demographic data. Population-weighted intra-urban intake By incorporating population distribution into fractions by country ranged from less than 10 dispersion models, it is possible to estimate to more than 100 ppm, and by major city from the intake fractions (the fractions of emissions less than 5 to more than 250 ppm (or gram inhaled by the population exposed to the per metric ton of emissions). The population- emissions), which can be used to estimate the weighted intake fractions by country are health effects of emissions. Zhou and others reported in the supplementary information (2006) used such modeling to estimate the note of Apte and others (2012). emission intake fractions from power plant sites throughout China, as discussed below. Population-weighted mean intake fractions by region and city size are presented in table 7, INTAKE FRACTIONS which shows that mean intake fractions vary more by city population than by region. The A measure of human exposure is the emission highest mean intake fractions are in South intake fraction, which estimates the fraction and Central Asia, Southeast Asia, East Asia of a metric ton of emissions breathed in by and Pacific, and Sub-Saharan Africa. TABLE 7: Population Weighted Mean Intra-Urban Intake Fractions of Distributed Ground- Level Emissions (ppm) Small cities Medium cities Large cities All cities (100,000–600,000) (600,000–3 million) (> 3 million) (> 100,000) South and Central Asia 15 36 106 55 Southeast Asia 20 46 67 48 East Asia and Pacific 22 49 70 44 Sub-Saharan Africa 18 38 98 43 Latin America 13 32 69 41 North Africa 10 27 57 32 Europe and Japan 10 22 55 30 Western Asia 12 27 41 26 Land-rich developed 7 15 30 20 World 15 35 65 39 Source: Apte and others 2012, supplementary information. 6. POPULATION EXPOSURE ASSESSMENT 30 Humbert and others (2011) summarized the for primary PM2.5 in urban areas is similar to work of an international expert group on the global population-weighted mean intake the integration of human exposure to PM fraction in Apte and others (2012). The intake into life cycle impact assessment (LCIA) fractions in rural areas are about one tenth within the UNEP/SETAC Life Cycle Initiative. of the urban fractions and decline by stack The authors reported recommended intake height. The intake fractions for secondary fractions for primary PM2.5 and secondary PM2.5 are expressed in grams of PM2.5 per PM2.5 from SO2, NOx, and ammonia (NH3) metric ton of precursor emissions, with similar precursors. Recommended global values for recommended values for urban and rural urban and rural areas are reported in table 8. areas. The distributed ground-level intake fraction TABLE 8: Summary of Recommended Intake Fractions (ppm) Height Primary PM2.5 Urban Rural Ground-level emissions 44 3.8 Low stack (25 meters) 15 2.0 High stack (100 meters) 11 1.6 Emission source weighted average* 26 2.6 Precursors Secondary PM2.5 Urban Rural Sulfur dioxide (SO2) 0.99 0.79 Nitrogen oxides (NOX) 0.2 0.17 Ammonia (NH3) 1.7 1.7 * Weighted by typical emission release height (ground, low, and high). Source: Humbert and others 2011. By region, the highest weighted urban intake and Southeast Asia. The highest rural intake fractions of primary PM2.5 are in Latin America fraction is in Southeast Asia (table 9). TABLE 9: Recommended PM2.5 Intake Fractions by Region Urban Rural Generic 26 2.6 United States 15 0.92 Latin America 29 0.75 Europe 18 2.1 Africa and the Middle East 25 1.1 Central Asia 20 1.3 Southeast Asia 29 4.6 Note: Data in this table are weighted by typical emission release height (ground, low, and high). Source: Humbert and others 2011. 31 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES In a seminal study in China, Zhou and sites (table 10). The mean intake fraction is others (2006) selected 29 power plant sites comparable to the estimation for rural areas throughout the country and estimated intake in Southeast Asia in table 9, albeit somewhat fractions at each site. A detailed long-range higher. Intake fractions for secondary PM3 atmospheric dispersion model, CALPUFF, was were also estimated as grams of sulfate or used to model the increase in concentrations nitrate per metric ton of SO2 or NOx emissions. due to emissions from selected power plants. The mean intake fractions are substantially Mean intake fraction of primary PM314 was higher than the global mean recommended 6 ppm, ranging from 1.7 to 12 ppm across by Hubert and others (2011). TABLE 10: Intake Fraction Estimates across 29 Power Plant Sites throughout China (ppm) Mean Minimum Maximum Primary PM3 6.1 1.7 12.0 Sulfate 4.4 0.7 7.3 Nitrate 3.5 0.8 7.1 Source: Zhou and others 2006. Regression models were developed to estimated that health effects due to sulfate- interpret the intake fraction values and allow and nitrate-based secondary PM2.5 were, on for extrapolation to other sites. Explanatory average, 17 and 4 times larger, respectively, variables were meteorological proxies, such than health effects due to primary PM2.5, as climate region and precipitation, and even though Indian coal has a low average population at various distances from the sulfur content of 0.5%. The findings may point sources. Differences in population distribution to the importance of incorporating intake explain a high portion of the differences fractions for secondary PM2.5 when assessing in the intake fractions across sites. The the health effects of electricity price subsidies meteorological regime also had a significant or price subsidies for fuels used in electricity influence on intake fractions (Zhou and others generation.15 2006). Parry and others (2014) applied intake INTAKE FRACTION APPLICATIONS fractions to estimate country-level health effects of primary and secondary PM 2.5 Cropper and others (2012) estimated the emissions from fossil fuels in order to estimate health effects of emissions from coal-fired corrective taxes. The paper by Apte and power plants in India by applying the intake- others (2012) was used for intra-urban intake fraction regression models in Zhou and fractions of distributed ground-level primary others (2006) used in China and adjusting PM2.5 emissions from road transportation and for differences in population distributions residential sources. This was combined with and rainfalls. All coal-fired power plants in the findings of Humbert and others (2011) to India have particulate abatement equipment, estimate country-specific intake fractions for but practically none had sulfur abatement SO2 and NOx. The intake-fraction regressions technology at the time of the study. The study in Zhou and others (2006) were used to 6. POPULATION EXPOSURE ASSESSMENT 32 estimate country-specific intake fractions others (2014), and estimates country-specific from power plants. intake fractions from the intake fraction regressions in Zhou and others (2006). These PROPOSED APPROACH intake fractions can be applied to both coal- and natural gas–fired power plants. This section outlines the proposed approaches to estimating population exposure to air Industry pollution due to incremental emissions from energy price subsidies from different sources. Intake fractions of emissions from fuels consumed by the industrial sector will likely Distributed Ground-Level Emissions fall somewhere between the intake fractions from power plants and distributed ground- The proposed approach for distributed level emissions and will be influenced largely ground-level emissions is to consider three by industrial locations that can be assessed areas: urban areas with a population greater in country assessments of subsidies. Most than 100,000, urban areas with a population cottage industries belong to distributed under 100,000, and rural areas. The country- ground-level missions. specific population-weighted intra-urban intake fractions from Apte and others (2012) is Baseline Outdoor PM2.5 Concentrations the recommended choice for urban areas with a population greater than 100,000. The value Because the relationships between PM2.5 of 3.8 ppm from Humbert and others (2011) is and health outcomes are nonlinear (see next proposed for vehicular emissions in rural areas, section), data on initial or prevailing outdoor including inter-urban vehicle transportation. PM2.5 concentrations are essential to estimate This value may be made country-specific by the health effects of higher fuel consumption adjusting for differences in rural population caused by energy price subsidies. Ambient density. A value between the intake fraction concentrations must be established at the for small cities (100,000–600,000) in Apte selected geographic or demographic scale and others (2012) and the rural value can be to analyze the impacts of energy price applied to urban areas with a population less subsidies. The proposed scale is similar to than 100,000. that discussed in the section on population exposure assessment for distributed ground- For secondary PM2.5, the intake fractions for level emissions: urban areas with a population SO2 and NOx from Humbert and others (2011) greater than 100,000, urban areas with a can be scaled by the country-specific intake population under 100,000, and rural areas. fractions for primary PM2.5 in Apte and others (2012), an approach adopted by Parry and Population-weighted average ambient PM2.5 others (2014). concentrations in areas affected by emissions from power plants may be best approximated Power Plant Emissions using the nationwide population-weighted average ambient PM2.5, as emissions from this The proposed approach for power plant source are dispersed over large geographic emissions follows the procedures used by areas. For emissions from road transport and Cropper and others (2012) and by Parry and other urban ground-level distributed sources, 33 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES the best approximation will generally be The global estimates used by the GBD project urban-population-weighted average ambient are presented below. PM2.5 for the share of fuels consumed in urban areas, and the rural-population-weighted Global estimates of annual PM2.5 concentrations average ambient PM2.5 for the share of fuels at 0.1° × 0.1° spatial resolution for the GBD consumed outside of urban areas. For fuels Study 2013 by the Institute for Health Metrics used by industry, an average of nationwide and Evaluation have recently been published and urban concentrations may well represent (Brauer and others 2016). The estimates were population exposures in areas with industrial produced by combining satellite-based emission sources. estimates, chemical transport model simulations, and ground measurements from Ambient ground-level monitoring 79 different countries. The estimates indicate measurements will rarely be available to that annual PM2.5 concentrations in large parts establish PM2.5 concentrations accurately at of North Africa, Middle East, and Asia exceeded this scale. Assumptions for approximations the World Health Organization’s (WHO) will, therefore, have to be made. One option Interim Target 3 of 35 micrograms per cubic is to apply ambient concentrations estimated meter (µg/m3) (figure 2). However, it is worth from satellite/transport models, an approach highlighting that the accuracy of these global used in the GBD project (Brauer and others estimates is influenced by the availability and 2016; van Donkelaar and others 2015, 2016). calibration of ground-level monitoring For urban areas, the priority will be to measurements of PM2.5, which are relatively assemble as much of available ground-level scarce in many developing countries (van monitoring measurement data as possible. Donkelaar and others 2015). FIGURE 2: Estimated Annual Average PM2.5 (µg/m3) 2013 Annual Average PM2.5 (µg/m3) <10 (meets WHO Guideline) ≥ 10 (WHO Guideline) ≥ 15 (WHO Interim Target 1) ≥ 25 (WHO Interim Target 2) ≥ 35 (WHO Interim Target 3) Source: Brauer and others 2016. 7. HEALTH EFFECTS 34 7. HEALTH EFFECTS This section presents the health effects due fraction (PIF) of health outcomes per change exposure to air pollution. in PM2.5 concentrations. OUTDOOR AIR POLLUTION The PIF is estimated using the relative- risk functions from the GBD project. These This section presents the health effects caused functions are nonlinear (Pope and others by outdoor air pollution. 2009, 2011; Burnett and others 2014; GBD 2015 risk factors collaborators 2016). Thus Mortality the magnitude of the PIF per change in concentrations of PM2.5 is a function of initial The health effects of incremental energy concentration. consumption due to energy price subsidies δD are estimated by the term in equation 1 in The potential impact fraction from a change δE section 2, which represents a change in health in PM2.5 concentrations is effects (δD) from a change in emissions (δE). Xi + Xi-1 Xi + Xi-1 Xi + Xi-1 This term can be expressed as PIF = [ n i=1 PiRR 2 - n i=1 P’iRR 2 ]/[ n i=1 PiRR 2 ], (5) δD PIF =m , (3) where P i and P' i are the percentage of δE δE population exposure before and after a change where m is baseline annual cases of the health in PM2.5 concentrations and RR is the relative outcomes (for example, the total number of risk of health outcomes at PM2.5 concentrations premature deaths) and PIF is the potential at the midpoint of concentrations xi and xi-1 impact fraction of health outcomes associated (see annex 2). with δE, expressed as the percentage change in health outcomes associated with a change The RR from the integrated-exposure- in PM2.5 emissions (see below). By using intake response (IER) function used by the GBD fraction equations (see annex 1), equation 3 Study 2015 are published in GBD 2015 risk becomes factors collaborators (2016)17 (figure 3). The δD m PIF RRs of ischemic heart disease (IHD) and = (4) IF , cerebrovascular disease (stroke) are the δE KP δx smallest for PM2.5 concentrations larger than w h e re i F i s t h e i n t a ke f ra c t i o n o f 30–40 µg/m 3 and the RR of acute lower emissions (ppm); P is exposed population; respiratory infection (ALRI) is the largest at K= Qd * 365 * 10-6 where Qd is the breathing PM2.5 concentrations greater than 40 µg/m3.18 rate of air (m3/day/person); and x is PM2.5 Globally, IHD accounts for 36% of deaths concentrations. Equation 4 says that health from outdoor PM2.5, stroke for 21%, chronic effects per metric ton of changes in PM2.5 obstructive pulmonary disease (COPD) for emissions are a function of the product of 20%, ALRI for 16%, and lung cancer for 7%, the intake fraction and the potential impact according to GBD 2015. 35 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES FIGURE 3: Relative Risks of Major Health Outcomes Associated with PM2.5 Exposure 1.8 1.7 1.6 1.5 1.4 1.3 1.2 1.1 1.0 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Annual average PM2.5 (µg/m )3 Stroke IHD COPD LC ALRI Source: GBD 2015 risk factors collaborators 2016. readily available in most developing countries The RRs are derived from studies of long-term without conducting extensive surveys. exposure to outdoor air PM2.5, secondhand tobacco smoking, exposure to smoke from Monetary valuation of many of these morbidity household solid cooking fuels, and active health outcomes is also a complex task. The tobacco smoking (Burnett and others 2014). studies that have valued both mortality and This provides a risk function that can be morbidity from ambient PM2.5 generally find applied to a wide range of outdoor PM2.5 that mortality counts for about 80% of total concentrations around the world as well as health costs. Thus the substantial efforts to high household air pollution levels of PM2.5 required to accurately estimate and value from the combustion of solid fuels. morbidity from exposure to incremental PM 2.5 concentrations caused by energy The risk functions are nonlinear, with declining price subsidies are not likely to be worth marginal health effects at higher PM 2.5 the resources required for the purpose of concentrations. Thus, the health effects of air improving the overall estimation of the pollution caused by energy price subsidies monetary value of health effects of price greatly depend on initial concentrations of PM2.5. subsidies. Morbidity An alternative, simpler approach is to estimate morbidity by applying the disease burden The empirical research literature presents from morbidity per premature death from a whole set of morbidity health outcomes ambient PM2.5 reported by the GBD project. from ambient PM2.5 and other air pollutants. The disease burden from morbidity is reported The literature most often expresses the risk as “years of life lost to disability” (YLD) and is of these health outcomes as percentage generally in the range 0.5–1.0 YLD per death changes relative to the baseline incidence or according to the GBD project. These years prevalence. However, reliable baselines are not of life lost can be converted to days of illness by multiplying YLD by 365 days per year and 7. HEALTH EFFECTS 36 dividing by the average disability weight for energy may have limited or even undetectable the health outcomes associated with ambient health effects. PM2.5 (typically 0.1–0.2). The estimated days of illness can then be monetized (see the There is little information on the extent of next section). switching to cleaner forms of energy across households. Many households use multiple HOUSEHOLD AIR POLLUTION forms of energy for cooking and heating. And yet most household surveys20 report only the This section presents the health effects caused primary source of energy for cooking, while by household air pollution. many do not ask about heating. Information on quantities consumed is seldom available, Mortality and even when data are collected, they are plagued by inaccuracy. Kerosene, LPG, electricity, and natural gas are alternatives to solid fuels for cooking, which The most optimistic (and also the most are prevalent in low- and lower-middle-income unrealistic) approach is to assume that the countries. Household air pollution is estimated price subsidies lead households to switch to cause nearly 3 million deaths per year (GBD entirely to cleaner forms of energy and stop 2015 risk factors collaborators 2016). Price using solid fuels altogether. This provides subsidies for cleaner alternatives may reduce an upper bound on the benefits of price household air pollution markedly if households subsidies A more reasonable—although still using solid fuels switch substantially or not realistic—approach is to assume that all entirely to these alternatives. Even in such households will start using cleaner forms circumstances, however, if neighbors continue of energy and will also continue to use to burn solid fuels, high outdoor PM 2.5 one or more solid fuels, reducing solid fuel concentrations caused by the neighbors’ consumption by the amount that corresponds activities affect indoor concentrations in the to an overall increase in the consumption dwellings of those using only clean forms of the subsidized energy obtained using of energy, diminishing health benefits. In relevant price elasticities. This scenario does practice, switching is rarely complete or even not envisage an increase in overall household substantial among the poor, who continue to energy use, as found by Masera, Saatkamp, use cheap or free solid biomass or coal for and Kammen (2000). cooking and heating, supplemented by cleaner forms of energy for limited activities. Kerosene In the first case, the first step in estimating the that is pressurized before combustion burns household air pollution effects of subsidies cleanly, but otherwise kerosene combustion is to estimate the percentage change in (as in wick stoves) can be quite polluting, households using solid fuels (Ŝ): although not nearly as much as combustion S1 -S0 L0 - L1 L1 L̂ Ŝ = (6) = =- of solid fuels. Because health effects as a S0 S0 T(1+L̂ ) - L1 function of ambient concentrations of PM2.5 where S is the number of households using are nonlinear and decline slowly with falling solid fuels; L is the number of households using ambient concentrations at relatively high cleaner forms of subsidized energy; T is the levels, partial switching to cleaner forms of total number of households; subscripts “0” and 37 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES “1” denote households in the absence and in the is expressed in the same energy unit as that presence of price subsidies, respectively; and for clean energy.21 L̂ is the percentage change in the number of households using cleaner forms of subsidized The next step is to estimate the change in energy. If households switch to exclusive use household air pollution concentrations from of clean energy, then L̂ is approximately equal the change in solid fuel use. The percentage to the percentage change in the consumption change in concentrations among households of cleaner energy, estimated by the methods using solid fuels may simply be approximated discussed in the previous section. as being equal to ÊS in equation 8. The change in health effects (such as the The change in health effects as a result number of premature deaths per year) is then of changes in household air pollution concentrations among households using solid ŜD1 ∆D = (7) fuels can be estimated using the PIF and 1+Ŝ health-risk functions from the GBD project. where D1 is the number of nationwide cases As the health-risk functions are nonlinear of health outcomes per year associated with with declining marginal health effects, the household air pollution from solid fuels in estimated health effects in the first case will the presence of price subsidies. D1 can be be larger than in the second case. estimated from current patterns of household energy use and health risk methodology from Morbidity the GBD project. As with morbidity due to outdoor PM 2.5 In the second case, it is assumed that all pollution, morbidity from household air households respond to the price subsidies pollution can be estimated by applying the and will use a little less solid fuels and a little disease burden from morbidity per premature more of cleaner energy. Percentage change death from household PM 2.5 air pollution in aggregate solid fuel consumption, ÊS is reported by the GBD project. The disease approximated by burden from morbidity is reported as YLD and ∆ES ∆ECE is generally in the range of 1.0–2.5 YLD per (8) ÊS = =- death according to the GBD project.22 These ES ES years of life lost can be converted to days of where ES is aggregate solid fuel consumption illness by multiplying YLD by 365 days per at prevailing price subsidies for clean forms year and dividing by the average disability of energy; ΔES and ∆ECE are the respective weight for the health outcomes associated changes in the consumption of solid fuels with ambient PM2.5 (for example, 0.1–0.2). and clean energy due to price subsidies, The estimated days of illness can then be with the assumption that ΔES = -∆ECE; and ES monetized (see next section). 8. THE VALUE OF HEALTH EFFECTS 38 8. THE VALUE OF HEALTH EFFECTS This section presents methods to estimate MORBIDITY the value of mortality and morbidity caused by air pollution. The cost of morbidity includes work absenteeism and medical treatment. The willingness to pay to avoid pain and suffering MORTALITY can also be added to this cost, but estimates The predominant measure of the welfare cost are generally not available for most countries. of a premature death used by economists is If a day of illness is valued at 50–100% of the value of statistical life (VSL) (annex 3). average daily wage rates to account for partial Reliable VSL studies are available only from work absenteeism and medical expenses, a minority of countries globally. A common then the cost of one YLD is about 5–10 times approach to estimating VSL in a country GDP per capita.23 that lacks such studies is therefore to use a The cost of morbidity per death from ambient benefit transfer based on meta-analyses of PM2.5 is then 2.5–10 times GDP per capita VSL studies from other countries. Narain and for 0.5–1.0 YLD per death. By contrast, the Sall (2016) presents such a benefit-transfer cost of mortality per death or VSL is about methodology for valuing mortality from air 70 times GDP per capita in lower-middle- pollution, drawing on the empirical literature income countries, per equation 9. The cost of VSL, especially studies on the members of of morbidity is therefore only about 4–14% the Organisation for Economic Co-operation of the cost of mortality. and Development (OECD) (OECD 2012). The proposed benefit transfer function is Studies that have valued both mortality Yc,n and morbidity from ambient PM2.5 pollution (9) ( VSLc,n = VSLOECD * YOECD ∈ ) generally find that morbidity accounts for where VSLc,n is the estimated VSL for country approximately 20% of total health costs. A c in year n; VSLOECD is the average base VSL reasonable approach to valuing morbidity in the sample of OECD countries with VSL from ambient PM2.5 pollution may, therefore, be studies (US$3.83 million); Yc,n is GDP per capita to use a morbidity-cost share of 10–20%, with in country c in year n; YOECD is the average GDP the lower bound being about the midpoint per capita for the sample of OECD countries of the estimate presented above. (US$37,000); and ∈ an income elasticity of The cost of morbidity per death from 1.2 for low- and middle-income countries household PM2.5 air pollution is higher than and 0.8 for high-income countries. All values from ambient PM2.5 pollution, because YLD are in purchasing power parity (PPP) prices. from this pollution is 1.0–2.5 per death. The VSLc,n must, therefore, be converted to local cost of morbidity per death is therefore 5–25 currency using PPP exchange rates, available times GDP per capita. This is 7–35% of the in the World Development Indicators by the cost of mortality, with a midpoint of about World Bank. 20% if the cost per death or VSL is about 70 times GDP per capita. 39 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES A reasonable approach to valuing morbidity as Bolivia, Guatemala, Honduras, Mexico, from household PM 2.5 air pollution may, Nicaragua, and Peru have found indoor air therefore, be to use a morbidity cost share pollution concentrations that far exceed of 20–30%, with the lower bound being in anything that might be considered reasonably the neighborhood of the midpoint of the safe for public health. estimate presented above, and the upper bound reflecting a premium for pain and Exposure to indoor air pollution, particularly suffering associated with illness. These costs of PM2.5, causes several illnesses. They include morbidity can be made specific to a country cardiovascular disease, COPD, and lung by using country-level data on YLD per death, cancer among adults, and ALRI among young wage rates, and VSL. children (Lim and others 2012). Women and children face greater risks from indoor air USING THE VALUE OF HEALTH pollution because they typically spend more EFFECTS TO INFORM POLICY time at home and often near the sources of OPTIONS combustion. Estimating and valuing the health effects Setting aside pain and suffering, there are of air pollution can inform potential economic costs associated with adverse interventions to reduce price subsidies with health effects caused by household air negative environmental and health effects or, pollution. They include medical expenses, alternatively, to evaluate price subsidies with forgone wages, and loss of productivity. positive effects. Added together, these losses can represent a significant economic burden. Table 11 shows Household air pollution is taken here as an calculated premature deaths and days of illness illustration. Billions of people around the attributed to household air pollution and their world do not have access to energy that is associated costs in several Latin American clean at the point of delivery. They use solid countries and subnational jurisdictions, using fuels for cooking and heating, such as coal, the methodologies described in this note. wood, agricultural residues, animal dung, At the national level, the costs imposed by and trash. Burning of solid fuels in traditional household air pollution are equivalent to up to stoves, and often with inadequate ventilation, 1.76% of GDP in Bolivia, and subnationally, they results in high concentrations of air pollutants represent up to 2.88% of GDP in Apurimac, within households. Studies in countries such Peru. TABLE 11: Premature Deaths and Days of Illness Caused Annually by Household Air Pollution and Their Associated Costs in Selected Jurisdictions Country Deaths Days of Illness (million) Cost (% of GDP) Bolivia (2014) 3,082 1.76% Mexico (2013) 12,931 77 0.58% Peru (2012) 6,114 65.5 1.31% 9. AIR POLLUTION HEALTH RISK ASSESSM ENT TOOLS 40 Subnational Jurisdiction Deaths Days of Illness (million) Cost (% of GDP) Piauí, Brazil (2012) 636 3.4 1.17% Hidalgo, Mexico (2012) 504 2.9 1.09% Yucatan Peninsula, Mexico (2013) 538 3.2 0.71% Apurimac, Peru (2012) 212 2.5 2.88% Sources: Larsen and Skjelvik 2013a, 2013b, 2014b; Larsen 2015a, 2015b, 2017b; Sánchez-Triana and others (forthcoming). Replacing solid fuels with clean forms of energy to promote fuel switching. The magnitude of would reduce or eliminate the severe adverse price subsidies needed to effect abandonment health effects of household air pollution. of solid fuel use altogether would be beyond In practice, because the energy choice is the means of any government. Good Practice determined largely by relative prices of energy Note 1 discusses this policy issue in some sources and household income, it is not easy detail. 9. AIR POLLUTION HEALTH RISK ASSESSMENT TOOLS This section presents tools that integrate in most other countries (Anenberg and others pollution emissions and concentration 2016). The Household Air Pollution Impacts data with health and air pollution response Tool (HAPIT) estimates the premature deaths functions to provide estimates of the health and disability-adjusted life-years (DALYs) that impacts caused by exposure to air pollution. would be avoided in any country as a result These tools aim to reduce the complexity of of reductions in exposure to household air estimating the health effects of air pollution pollution concentrations. HAPIT also compares by using automated computer programs to the costs of the intervention implemented to provide relatively simple but reliable estimates. reduce health risks and the resulting benefits.24 A review conducted in 2016 identified Table 12 summarizes key features of available eight existing tools with a global scope— global-scale tools that estimate the health encompassing countries and many cities effects of exposure to PM2.5. While energy price around the world—that assess the impacts of subsidies are generally applied at a national outdoor concentrations of PM2.5 on premature level, in some countries, most of the air quality deaths, several of which also estimate increased problems associated with such price subsidies morbidity. Some of these tools require data might be localized in one or a few large urban from air quality modeling as inputs, while areas. Estimates of health outcomes that others include built-in parameters that can include both deaths and illnesses are likely be used when such data are not available or to provide more accurate projections of the would be costly to obtain. As an example, effects caused by air pollution. However, most of these tools include long-term PM2.5 morbidity estimates are generally less reliable concentration-response relationships from than mortality ones because of differences in studies conducted in the United States, given access to health services, medical procedures, that similar long-term studies are nonexistent and baseline morbidity rates across countries. 41 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES In terms of format, web-based systems are open-source codes is that the algorithms and the most accessible because they can be data sets used for estimation are transparent. run using a freely available Internet browser. A key consideration in tool selection is the Several tools use spreadsheets, with which availability of the inputs needed to run it. Most many potential users are already familiar. models require air quality modeling data to be A few tools require downloading custom entered by the user, but some reduced-form software. Two tools are proprietary and the tools can be used to obtain broad estimates remaining five use open-source codes. Aside by using built-in parameters (Anenberg and from cost savings, the main advantage of others 2016). TABLE 12: Key Characteristics of Air Pollution Health Risk Assessment tools Required user input Tool Spatial resolution Health outcome Format (information source) Web-based, Emissions (primary PM2.5 AirCountsa City level Mortality cases proprietary intake fraction) Mortality and Software Regional, national, Concentration (any AIRQ 2.2b morbidity cases, download, open or city-level concentration input by user) DALYs and YLLs source Mortality and Software Regional, national, Concentration (any BenMAP-CEc morbidity cases, download, open or city-level concentration input by user) DALYs and YLLs source Environmental Mortality and Regional, national, Microsoft Office, Concentration (any Burden of morbidity cases, or city-level open source concentration input by user) Diseased DALYs and YLLs Mortality and Regional, national, Microsoft Office, Concentration (any IOMLIFETe morbidity cases, or city-level open source concentration input by user) DALYs and YLLs Microsoft Office, Emissions (reduced-form LEAP-IBCf National Mortality cases open source chemical transport model) Regional and city Mortality and Microsoft Office, Emissions (regional or urban SIM-Airg level morbidity cases open source atmospheric chemistry model) Regional and Mortality cases, Microsoft Office, Emissions (reduced-form TM5-FASSTh national DALYs and YLLs proprietary chemical transport model) a. http://www.aircounts.com/. b. http://www.euro.who.int/en/health-topics/environment-and-health/air-quality/activities/airq-software-tool-for-health-risk- assessment-of-air-pollution. c. https://www.epa.gov/benmap. d. http://www.euro.who.int/en/health-topics/environment-and-health/pages/evidence-and-data/environmental-burden-of- disease-ebd. e. http://www.iom-world.org/research/iom-research-disciplines/statistical-services/iomlifet/. f. https://www.sei-international.org/low-emissions-development-planning. g. http://www.sim-air.org/. h. http://tm5-fasst.jrc.ec.europa.eu/. Note: YLL = Years of Life Lost. “Reduced form” refers to tools that use built-in parameters instead of inputs from air quality modeling. Source: Authors’ calculations using data from Anenberg and others 2016. 10. CONCLUSIONS 42 10. CONCLUSIONS This note focuses on local air pollution and its own methodologies. The note provides health, which arguably represents the largest guidance on each step and identifies key tools global social cost of the local environmental and methods that are readily available and can externality associated with energy production be used to inform decisions about the potential and use. The pollution here includes both environmental and health effects of removing outdoor air pollution and household air energy price subsidies. It also provides practical pollution from cooking, heating and, to a lesser information to help practitioners develop extent, lighting. An estimated 6.5 million people reliable estimates even in the absence of data die from outdoor ambient and household air and with limited resources. pollution each year, according to the GBD Study 2015. Combustion of fossil fuels and traditional The tools and methods presented in the note biomass fuels is the cause of a large share of can be used to carry out quick assessments these deaths. Annual global price subsidies of the severity of health effects caused by to fossil fuels and electricity in the hundreds air pollution, including from the additional of billions of U.S. dollars exacerbate outdoor pollution caused or reduced by price subsidies. air pollution, while potentially mitigating While these methods can be used in situations household air pollution where cleaner forms with limited local data, they are not meant of energy are subsidized. to substitute more robust assessments of air quality, particularly those based on a reliable This note provides an overview and guidance air quality monitoring network, inventories of on the use of tools that can be applied by mobile and stationary sources, and models experienced practitioners to assess health that explain the contribution of different effects of energy price subsidies at the country sources of pollution, including natural sources. level. Where data exist, the recommended methodologies can also be applied at the Despite the clear and urgent need to better subnational level. Assessing such effects is understand air quality trends to inform air highly complex and involves multiple fields quality management efforts, few cities and and disciplines. The note is limited to cases countries in the developing world have where price subsidies do not cause shortages established well-resourced units in charge of subsidized energy at the official prices— of monitoring air quality based on specialized shortages and high black market prices are equipment, regular maintenance, supplies common with liquid fuels, and while “black of consumables, standardized protocols for market” prices do not affect network energy reading and interpreting data, and quality (electricity, natural gas, and district heating), assurance and quality control procedures. shortages in the form of outages are also Removing energy price subsidies might free common with electricity in many countries. resources that can be used to improve air quality management. Subject to the foregoing limitations, this note helps practitioners by breaking the assessment While removing price subsidies for polluting down into several distinct steps, each with fuels would generally be good for the environment and public health, phasing out GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES rice subsidies for clean forms of energy could have adverse environmental and health effects. Good Practice Note 5 suggests how to mitigate such negative effects of price subsidy reform. The fact that energy price subsidies can have both positive and negative environmental effects underscores the importance of assessing the potential linkages among energy, environment, and health to inform subsidy reform.   ANNEX 1: EMISSION INTAKE FRACTIONS Health effects per metric ton of PM2.5 emissions can be estimated by using location- specific PM2.5 intake fractions. Variations in the social cost per metric ton of PM2.5 emissions are explained mainly by variations in intake fractions, initial PM2.5 ambient concentrations, baseline health conditions, and valuation of health effects. Health effects per ton of changes in PM2.5 emissions in a geographic area are δD PIF H= =m , (A1.1) δE δE where δD is the change in health effects per year (for example, the number of deaths from PM2.5); δE is change in emissions of PM2.5 (metric tons/year); m is the number of baseline annual cases of the health outcomes (such as the total number of deaths); and PIF is the potential impact fraction of health outcomes associated with δE. Solving for H requires a relation between emissions (E) and concentrations (x). The change in the quantity of PM2.5 that a population breathes into the lungs in a year is given by ∂iP = P * Qd * 365 * 10-12 * ∂x, (A1.2) where iP is population intake of PM2.5 (metric tons/year), P is population, Qd is breathing rate of air (m3/day/person), and δx is the change in concentrations of PM2.5 (μg/m3). The change in population intake (metric tons/year) is also given by δiP = δx * iF * 10-6, (A1.3) where iF is the so called intake fraction in ppm, or the fraction of emissions that the population breathes into their lungs.25 Combining equations A1.2 and A1.3 yields ∂E = P * Qd * 365 * 10-6 * iF-1 * ∂x. (A1.4) Equation A1.4 can be rewritten as P δx δE = K , (A1.5) iF 43 ANNEX 2: METHODOLOGY FOR ESTIMATING HEALTH EFFECTS from which can be seen how changes in emissions and concentrations are related to a known population and intake fraction, and K = Qd * 365 * 10-6. Equation A1.1 then becomes m PIF H = iF , (A1.6) KP δx which says that health effects per year per metric ton of changes in PM2.5 emissions are a function of the product of the intake fraction and the potential impact fraction of health outcomes per change in PM2.5 concentrations. The latter is estimated using the integrated-exposure-response (IER) functions from the GBD project, and its magnitude is a function of the initial concentration level.   ANNEX 2: METHODOLOGY FOR ESTIMATING HEALTH EFFECTS Particulate matter (PM) is the air pollutant that globally is associated with the largest health effects. It is a major outdoor ambient air pollutant and a major household air pollutant from the burning of solid fuels for cooking and heating. Health effects of PM exposure include both premature mortality and morbidity. The methodologies to estimate these health effects have evolved as the body of research evidence has increased. OUTDOOR AMBIENT PARTICULATE MATTER AIR POLLUTION Over a decade ago, Pope and others (2002) found an elevated risk of cardiopulmonary (CP) and lung cancer (LC) mortality from long-term exposure to outdoor PM2.5 in a study of a large population of adults age 30 or older in the United States. CP mortality includes mortality from respiratory infections, cardiovascular disease, and chronic respiratory disease. The WHO used the study by Pope and others when estimating global mortality from outdoor air pollution (WHO 2004, 2009). Since then, recent research suggests that the marginal increase in relative risk of mortality from PM2.5 declines with increasing concentrations of PM2.5 (Pope and others 2009, 2011). Pope and others (2009, 2011) derive a shape of the PM2.5 exposure-response curve based on studies of mortality from active cigarette smoking, secondhand cigarette smoking, and outdoor ambient PM2.5 air pollution. HOUSEHOLD PARTICULATE MATTER AIR POLLUTION Combustion of solid fuels for cooking and heating is a major source of household air pollution in many developing countries. Concentrations of PM2.5 often reach several hundred µg/m3 in the kitchen and living and sleeping environments. Combustion of these fuels is therefore associated with an increased risk of several health outcomes, 44 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES such as acute lower respiratory infections (ALRI) in children, and chronic obstructive pulmonary disease (COPD), chronic bronchitis (CB), and lung cancer in adults. The global evidence is summarized in meta-analyses by Desai, Mehta, and Smith (2004), Smith, Mehta, and Feuz (2004), Dherani and others (2008), Po, FitzGerald, and Carlsten (2011), and Kurmi and others (2010). Risks of health outcomes reported in these meta-analyses are generally point estimates of relative risks of health outcomes (with confidence intervals) from the use of fuel wood, other solid biomass fuels,26 and coal relative to the risks from use of electricity, gaseous fuels, or LPG. A randomized intervention trial in Guatemala found that cooking with wood using an improved chimney stove, which greatly reduced PM2.5 exposure, was associated with lower systolic blood pressure (SBP) among adult women compared to SBP among women cooking with wood on open fire (McCracken and others 2007). Baumgartner and others (2011) found that an increase in PM2.5 personal exposure was associated with an increase in SBP among a group of women in rural households using biomass fuels in China. These studies provide some evidence that PM air pollution in the household environment from the combustion of solid fuels contributes to cardiovascular disease. AN INTEGRATED EXPOSURE-RESPONSE FUNCTION The GBD project starts with the findings of Pope and others (2009, 2011) and takes some steps further by deriving an integrated exposure-response (IER) relative-risk function (RR) for health outcome k in age group l associated with exposure to outdoor and indoor PM2.5: RR(x)kl = 1 for x < xcf , (A2.1a) ρkl RR(x)kl = 1 + αkl (1 - e-βkl (x - xcf) ) for x ≥ xcf , (A2.1b) where x is the ambient concentration of PM2.5 in µg/m3, xcf is the critical threshold concentration below which no association is assumed to exist between PM2.5 exposure and assessed health outcomes (theoretical minimum risk-exposure level), α , β, and ρ are the parameters that determine the slope of the IER curve and the relative risks of health effects in relation to PM 2.5 exposure concentrations. The function allows prediction of RR over a very large range of PM2.5 concentrations, with RR(xcf + 1) ~ 1+αβ as β approaches zero, and RR(∞) = 1 + α as the PM2.5 concentrations rising without bound representing the maximum risk (Burnett and others 2014; Shin and others 2013). The health outcomes assessed in the GBD study are ischemic heart disease (IHD), cerebrovascular disease (stroke), lung cancer, COPD, and ALRI (Lim and others 2012; Mehta and others 2013; Smith and others 2014; Forouzanfar and others 2015; GBD 2015 risk factors collaborators 2016). The risk functions for IHD and cerebrovascular disease are age-specific with five-year age intervals from 25 years of age, while singular age-group risk-functions are applied for lung cancer (≥ 25 years), COPD (≥ 25 years), 45 ANNEX 3: VALUING THE HEALTH EFFECTS OF ENERGY SUBSIDIES and ALRI for children and adults in GBD 2013 and 2015. An xcf between 2.4 and 5.9 µg/m3 is applied in the GBD 2015 Project (GBD 2015 risk factors collaborators 2016). The population attributable fraction (PAF) of a specific disease from PM2.5 exposure is calculated by X + Xi-1 X + Xi-1 PAF = i=1 n Pi [RR i (A2.1a) - 1]/( i=1 n Pi [RR i -1]+1), 2 2 where Pi is the share of the population exposed to PM2.5 concentrations in the range xi-1 to xi.27 PAF is calculated for each health outcome k and age group l. The disease burden (D) in terms of annual cases of health outcomes due to PM2.5 exposure is then estimated by D = k=1 l=1 mkl PAFkl , (A2.3) t s where mkl is the total annual number of cases of health outcome k in age group l, and PAFkl is the population attributable fraction of these cases of health outcome k in age group l due to PM2.5 exposure. The potential impact fraction, or the change in PAF, is applied to estimate the change in disease burden from a change in the population exposure distribution, X + Xi-1 X + Xi-1 Xi + Xi-1 PIF = [ i=1 n Pi RR i - i=1 n P’i RR i ]/( i=1 n Pi RR , (A2.4) 2 2 2 where P'i is the population exposure distribution after the intervention.  ANNEX 3: VALUING THE HEALTH EFFECTS OF ENERGY SUBSIDIES The predominant measure of the welfare cost of a premature death used by economists is the value of statistical life (VSL), estimated from individuals’ willingness to pay (WTP) for mortality risk reductions. VSL is calculated based on individuals’ valuation of changes in mortality risk. Everyone in society is constantly facing a certain risk of dying. Examples of such risks are occupational fatality risk, risk of traffic accident fatality, and environmental mortality risks. It has been observed that individuals adjust their behavior and decisions in relation to such risks. For example, individuals demand a higher wage (a wage premium) for a job that involves a higher occupational risk of fatal accident than in other jobs, individuals may purchase safety equipment to reduce the risk of death, and/or individuals and families may be willing to pay a premium or higher rent for properties (land and buildings) in a cleaner and less polluted neighborhood or city. 46 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES Through the observation of individuals’ choices and willingness to pay for reducing mortality risk (or minimum amounts that individuals require to accept a higher mortality risk), it is possible to estimate the value to society of reducing mortality risk, or, equivalently, measure the welfare cost of a particular mortality risk. As an illustration, consider the case where a certain health hazard has a mortality risk of 2.5 per 10,000 persons. This means that 2.5 individuals die from this hazard for every 10,000 individuals exposed. If each individual on average is willing to pay US$40 for eliminating this mortality risk, then every 10,000 individuals are collectively willing to pay US$400,000. Dividing this amount by the risk gives the VSL of US$160,000. Mathematically this can be expressed as VSL = WTPAve * 1/ R , (A3.1) where WTPAve is the average WTP per individual for a mortality-risk reduction of magnitude R. In equation A3.1, R=2.5/10 000 (or R=0.00025) and WTPAve= US$40. Thus, if 10 individuals die from the health risk illustrated above, the cost (C) to society is C = 10 * VSL = 10 * US$0.16 million = US$1.6 million. (A3.2) The main approaches to estimating VSL are through revealed preferences and stated preferences of people’s WTP for a reduction in mortality risk or their willingness to accept an increase in mortality risk. Most of the studies of revealed preferences are hedonic wage studies, which estimate labor market wage differentials associated with differences in occupational mortality risk. Most of the stated preference studies rely on contingent valuation methods, which in various forms ask individuals about their WTP for mortality risk reduction. Studies of VSL have been carried out in many countries and several meta-analyses of these studies have been conducted in the last three decades. Meta-analyses assess characteristics that determine VSL, such as household income, the size of risk reduction, other individual and household characteristics, and often the characteristics of the methodologies used in the original WTP studies. Most of the meta-analyses of VSL are entirely or predominantly based on hedonic wage studies. A meta-analysis prepared for the OECD was, however, exclusively based on stated preference studies, arguably of greater relevance for valuation of mortality risk from environmental factors such as air pollution than hedonic wage studies. These stated preference studies are from a database of more than 1,000 VSL estimates from multiple studies in more than 30 countries, including in developing countries (Lindheim and others 2011; OECD 2012). 47 REFERENCES REFERENCES Anenberg, S. C., A. Belova, J. Brandt, N. Fann, S. Greco, S. Guttikunda, M. E. Heroux, F. Hurley, M. Krzyzanowski, S. Medina, B. Miller, K. Pandey, J. Roos, and R. Van Dingenen. 2016. “Survey of Ambient Air Pollution Health Risk Assessment Tools.” Risk Analysis 36(9):1718–36. DOI: 10.1111/risa.12540. Apte, J., E. Bombrun, J. Marshall, and W. Nazaroff. 2012. “Global Intraurban Intake Fractions for Primary Air Pollutants from Vehicles and Other Distributed Sources.” Environ. Sci. Technol. 46:3415−23. Baumgartner, Jill, James J. Schauer, Majid Ezzati, Lin Lu, Chun Cheng, Jonathan A. Patz, and Leonelo E. Bautista. 2011. “Indoor Air Pollution and Blood Pressure in Adult Women Living in Rural China.” Environmental Health Perspectives 119(10):1390–95. https://www. ncbi.nlm.nih.gov/pmc/articles/PMC3230444/ BCME (British Columbia Ministry of Environment). 2015. British Columbia Air Quality Dispersion Modelling Guideline. Victoria, British Columbia: British Columbia Ministry of Environment. https://www.bcogc.ca/node/13339/download Brauer, M., G. Freedman, J. Frostad, A. van Donkelaar, R. V. Martin, F. Dentener, R. van Dingenen, K. Estep, H. Amini, J. S. Apte, K. Balakrishnan, L. Barregard, D. Broday, V. Feigin, S. Ghosh, P. K. Hopke, L. D. Knibbs, Y. Kokubo, Y. Liu, S. Ma, L. Morawska, J. L. Sangrador, G. Shaddick, H. R. Anderson, T. Vos, M. H. Forouzanfar, R. T. Burnett, and A. Cohen. 2016. “Ambient Air Pollution Exposure Estimation for the Global Burden of Disease 2013.” Environ. Sci. Technol. 50(1):79–88. Brons, M., P. Nijkamp, E. Pels, and P. Rietveld. 2008. “A Meta-Analysis of the Price Elasticity of Gasoline Demand. A SUR Approach.” Energy Economics 30:2105–22. Burnett, Richard T., C. Arden Pope III, Majid Ezzati, Casey Olives, Stephen S. Lim, Sumi Mehta, Hwashin H. Shin, Gitanjali Singh, Bryan Hubbell, Michael Brauer, H. Ross Anderson, Kirk R. Smith, John R. Balmes, Nigel G. Bruce, Haidong Kan, Francine Laden, Annette Prüss-Ustün, Michelle C. Turner, Susan M. Gapstur, W. Ryan Diver, and Aaron Cohen. 2014. “An Integrated Risk Function for Estimating the Global Burden of Disease Attributable to Ambient Fine Particulate Matter Exposure.” Environmental Health Perspectives 122:397–403. Coady, David, Ian W. H. Parry, Louis Sears, and Baoping Shang. 2015. “How Large Are Global Energy Subsidies?” IMF Working Paper 15/105, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/ How-Large-Are-Global-Energy-Subsidies-42940 Cohen, Aaron J., Michael Brauer, Richard Burnett, H. Ross Anderson, Joseph Frostad, Kara Estep, Kalpana Balakrishnan, Bert Brunekreef, Lalit Dandona, Rakhi Dandona, Valery Feigin, Greg Freedman, Bryan Hubbell, Amelia Jobling, Haidong Kan, Luke Knibbs, Yang Liu, Randall Martin, Lidia Morawska, C. Arden Pope III, Hwashin Shin, Kurt Straif, Gavin Shaddick, Matthew Thomas, Rita van Dingenen, Aaron van Donkelaar, Theo Vos, Christopher J. L. Murray, and Mohammad H. Forouzanfar. 2017. “Estimates and 25-Year Trends of the Global Burden of Disease Attributable to Ambient Air Pollution: An Analysis of Data from the Global Burden of Diseases Study 2015.” Lancet 389(10082):1907–18. 48 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES Cropper, Maureen, Shama Gamhkar, Kabir Malik, Alex Limonov, and Ian Patridge. 2012. “The Health Effects of Coal Electricity Generation in India.” RFF Discussion Paper 12–25, Resources for the Future, Washington, DC. https://sites.hks.harvard.edu/m-rcbg/rpp/ RFF-DP-12-25.pdf Davis, L. W. 2016. “The Environmental Cost of Global Fuel Subsidies.” NBER Working Paper 22105, National Bureau of Economic Research, Cambridge, Massachusetts. http://www. nber.org/papers/w22105 Desai, M. A., S. Mehta, and K. Smith. 2004. Indoor Smoke from Solid Fuels: Assessing the Environmental Burden of Disease at National and Local Levels. Environmental Burden of Disease Series No. 4. Geneva: World Health Organization. Dherani, M., D. Pope, M. Mascarenhas, K. Smith, M. Weber, and N. Bruce. 2008. “Indoor Air Pollution from Unprocessed Solid Fuel Use and Pneumonia Risk in Children Aged under Five Years: A Systematic Review and Meta-Analysis.” Bulletin of the World Health Organization 86:390–98. Durbin, T. D., M. R. Smith, J. M. Norbeck, and T. J. Truex. 1999. “Population Density, Particulate Emission Characterization and Impact on the Particulate Inventory of Smoking Vehicles in the South Coast Air Quality Management District.” Journal of Air and Waste Management Association 49:28–38. EPA (Environmental Protection Agency). 2005. “Revision to the Guideline on Air Quality Models: Adoption of a Preferred General Purpose (Flat and Complex Terrain) Dispersion Model and Other Revisions; Final Rule.” Environmental Protection Agency, Washington, DC. https://www3.epa.gov/scram001/guidance/guide/appw_05.pdf Espey, M. 1996. “Explaining the Variation in Elasticity Estimates of Gasoline Demand in the United States: A Meta-Analysis.” Energy Journal 17:49–60. ———. 1998. “Gasoline Demand Revisited: An International Meta-Analysis of Elasticities.” Energy Economics 20:273–95. Espey, J. A., and M. Espey. 2004. “Turning on the Lights: A Meta-Analysis of Residential Electricity Demand Elasticities.” Journal of Agricultural and Applied Economics 36:65–81. GBD 2015 risk factors collaborators. 2016. “Global, Regional, and National Comparative Risk Assessment of 79 Behavioural, Environmental and Occupational, and Metabolic Risks or Clusters of Risks, 1990–2015: A Systematic Analysis for the Global Burden of Disease Study 2015.” Lancet 388:1659–1724. GBD MAPS WG (Global Burden of Disease from Major Air Pollution Sources Working Group). 2016. Burden of Disease Attributable to Coal-Burning and Other Major Sources of Air Pollution in China. Special Report 20, August. Boston: Health Effects Institute. https://www.healtheffects.org/system/files/GBDMAPS-ExecSummaryEnglishFinal.pdf Geissler, Gesa, Johann Köppel, and Pamela Gunther. 2013. “Wind Energy and Environmental Assessments—A Hard Look at Two Forerunners’ Approaches: Germany and the United States.” Renewable Energy 51:71–78. Graham, D., and S. Glaister. 2002. “The Demand for Automobile Fuel: A Survey of Elasticities. Journal of Transport Economics and Policy 36(1):1–26. Gwilliam, Ken, Masami Kojima, and Todd Johnson. 2004. Reducing Air Pollution from Urban Transport. Washington, DC: World Bank. http://www.esmap.org/node/1145 49 REFERENCES Hanly, M., J. Dargay, P. Goodwin. 2002. “Review of Income and Price Elasticities in the Demand for Road Traffic.” ESRC TSU publication 2002/13, Centre for Transport Studies, University College London, London. Havranek, T., Z. Irsova, and K. Janda. 2012. “Demand for Gasoline Is More Price-Inelastic Than Commonly Thought.” Energy Economics 34:201–07. Hernandez, R., S. Easter, M. Murphy-Mariscal, F. Maestre, M. Tavassoli, E. Allen, C. Barrows, J. Belnap, R. Ochoa-Hueso, S. Ravi, and M. Allen. 2014. “Environmental Impacts of Utility- Scale Solar Energy.” Renewable and Sustainable Energy Reviews 29:766–79. Hertwich, E. G., T. Gibon, E. A. Bouman, A. Arvesen, S. Suh, G. A. Heath, J. D. Bergesen, A. Ramirez, M. I. Vega, and L. Shi. 2014. “Integrated Life-Cycle Assessment of Electricity- Supply Scenarios Confirms Global Environmental Benefit of Low-Carbon Technologies.” Proceedings of the National Academy of Science 112(20):6277–82 www.pnas.org/cgi/ doi/10.1073/pnas.1312753111 Humbert, S., J. D. Marshall, S. Shaked, J. V. Spadaro, Y. Nishioka, P. Preiss, T. E. McKone, A. Horvath, and O. Jolliet. 2011. “Intake Fraction for Particulate Matter: Recommendations for Life Cycle Impact Assessment.” Environ. Sci. Technol. 45:4808−16. IEA (International Energy Agency). 2016. Energy and air pollution. World Energy Outlook Special Report. Paris: OECD/IEA. ———. 2017. “World Energy Balances.” IEA World Energy Statistics and Balances (database), Paris. Kojima, Masami. 2013. “Petroleum Product Pricing and Complementary Policies: Experience of 65 Developing Countries since 2009.” Policy Research Working Paper 6393, World Bank, Washington, DC. https://openknowledge.worldbank.org/handle/10986/13201 Kojima, Masami, and Chris Trimble. 2016. Making Power Affordable for Africa and Viable for Its Utilities. Washington, DC: World Bank. https://openknowledge.worldbank.org/ handle/10986/25091OECD Kurmi, O. P., S. Semple, P. Simkhada, W. C. Smith, and J. G. Ayres. 2010. “COPD and Chronic Bronchitis Risk of Indoor Air Pollution from Solid Fuel: A Systematic Review and Meta- Analysis.” 2010. Thorax 65:221–28. Labandeira, X., J. Labeaga, and X. López-Otero. 2016. “A Meta-Analysis on the Price Elasticity of Energy Demand.” EUI Working Paper RSCAS 2016/25, European University Institute, San Domenico di Fiesole, Italy. http://cadmus.eui.eu//handle/1814/40870 Larsen, B. 1994. World Fossil Fuel Subsidies and Global Carbon Emissions in a Model with Interfuel Substitution. World Bank Policy Research Working Paper 1256. World Bank, Washington, DC. ———. 2015a. “An Economic Assessment of Environmental Health Risks in Apurimac, Peru.” Report prepared for the World Bank, Washington, DC. ———. 2015b. “Environmental Health in Mexico: An Economic Assessment of Health Effects and Their Costs.” Report prepared for the World Bank, Washington, DC. ———. 2017a. “Environmental Health Effects and Their Cost in Lao PDR.” Report prepared for the World Bank, Washington, DC. 50 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES ———. 2017b. “Environmental Health in Bolivia: An Economic Assessment of Health Effects and Cost of Ambient Air Pollution—2015 Update.” Report prepared for the World Bank, Washington, DC. Larsen, B., and J. M. Skjelvik. 2013a. “An Economic Assessment of Environmental Health Risks in Peru: An Update 2012.” Report prepared for the World Bank, Washington, DC. ———. 2013b. “Environmental Health in Piauí State of Brazil: An Economic Assessment of Health Effects and Their Costs.” Report prepared for the World Bank, Washington, DC. ———. 2014a. “An Economic Assessment of Environmental Health Risks in Argentina.” Report prepared for the World Bank, Washington, DC. ———. 2014b. “Environmental Health in Hidalgo State of Mexico: An Economic Assessment of Health Effects and Their Costs.” Report prepared for the World Bank, Washington, DC. Lim, Stephen S., and others 2012. “A Comparative Risk Assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010.” Lancet 380:2224–60. Macknick, J., R. Newmark, G. Heath, and K. C. Hallett. 2012. “Operational Water Consumption and Withdrawal Factors for Electricity Generating Technologies: A Review of Existing Literature.” Environmental Research Letters 7 (2012) 045802 (10pp). http:// aquasec.org/wrpg/wp-content/uploads/2011/09/Macknick-etal-2012-Water-electric- generation-coefficients-ERL.pdf Masera, Omar R., Barbara D. Saatkamp, and Daniel M. Kammen. 2000. “From Linear Fuel Switching to Multiple Cooking Strategies: A Critique and Alternative to the Energy Ladder Model.” World Development 28(12):2083–2103. McCracken, J. P., K. R. Smith, A. Díaz, M. A. Mittleman, and J. Schwartz. 2007. “Chimney Stove Intervention to Reduce Long-Term Wood Smoke Exposure Lowers Blood Pressure among Guatemalan Women.” Environmental Health Perspectives 115(7):996–1001. Mehta, S., H. Shin, R. Burnett, T. North, and A. Cohen. 2013. “Ambient Particulate Air Pollution and Acute Lower Respiratory Infections: A Systematic Review and Implications for Estimating the Global Burden of Disease.” Air Qual. Atmos. Health 6:69–83. Montes de Oca, Mariza, and Carlos Muñoz-Piña. 2016. “Fuel Subsidy Reform in Mexico: Environmental and Social Benefits of the Gasoline Subsidy Removal and the Way Forward for Optimal Pricing.” Report prepared for the World Bank, Washington, DC. Narain, Urvashi, and Chris Sall. 2016. “Methodology for Valuing the Health Impacts of Air Pollution: Discussion of Challenges and Proposed Solutions.” World Bank, Washington, DC. https://openknowledge.worldbank.org/handle/10986/24440 OECD (Organisation for Economic Co-operation and Development). 2012. Mortality Risk Valuation in Environment, Health, and Transport Policies. Paris: OECD Publishing. http:// dx.doi.org/10.1787/9789264130807-en Parry, I., D. Heine, E. Lis, and S. Li. 2014. Getting Energy Prices Right: From Principle to Practice. Washington, DC: International Monetary Fund. Po, J. Y. T., J. M. FitzGerald, and C. Carlsten. 2011. “Respiratory Disease Associated with Solid Biomass Fuel Exposure in Rural Women and Children: Systematic Review and Meta- Analysis.” Thorax 66:232–39. 51 REFERENCES Pope III, C. A., R. T. Burnett, D. Krewski, M. Jerrett, Y. Shi, E. E. Calle, and M. J. Thun. 2009. “Cardiovascular Mortality and Exposure to Airborne Fine Particulate Matter and Cigarette Smoke: Shape of the Exposure-Response Relationship.” Circulation 120:941–48. Pope III, C. A., R. T. Burnett, M. J. Thun, E. Calle, D. Krewski, K. Ito, and G. Thurston. 2002. “Lung Cancer, Cardiopulmonary Mortality, and Long-Term Exposure to Fine Particulate Air Pollution.” Journal of the American Medical Association 287:1132–41. Pope III, C. A., R. T. Burnett, M. C. Turner, A. Cohen, D. Krewski, M. Jerrett, S. M. Gapstur, and M. J. Thun. 2011. “Lung Cancer and Cardiovascular Disease Mortality Associated with Ambient Air Pollution and Cigarette Smoke: Shape of the Exposure-Response Relationships.” Environmental Health Perspectives 119(11):1616–21. Pope III, C. A., and D. W. Dockery. 2006. “Health Effects of Fine Particulate Air Pollution: Lines that Connect.” Journal of the Air & Waste Management Association 56:709–42. http://www.noaca.org/pmhealtheffects.pdf Saidur, R., N. Rahim, M. Islam, and K. Solangi. 2011. “Environmental Impact of Wind Energy.” Renewable and Sustainable Energy Reviews 15(5):2423–30. Sánchez-Triana, E., S. Enriquez, J. Afzal, A. Nakagawa, and A. S. Khan. 2014. Cleaning Pakistan’s Air. Policy Options to Address the Cost of Air Pollution. Directions in Development Series. Washington, DC: World Bank. http://documents.worldbank.org/ curated/en/701891468285328404/pdf/890650PUB0Clea00Box385269B00PUBLIC0. pdf. Sánchez-Triana, E., S. Enriquez, B. Larsen, P. Webster, and J. Afzal. 2015. Sustainability and Poverty Alleviation: Confronting Environmental Threats in Sindh, Pakistan. Directions in Development. Washington, DC: World Bank. Sanchez-Triana, E., J. Ruitenbeek, S. Enriquez, and K. Siegmann. Forthcoming. Green and Inclusive Growth in Yucatan Peninsula. Washington, D.C.: World Bank. Shin, Hwashin H., Aaron Cohen, C. Arden Pope III, Majid Ezzati, Stephen S. Lim, Bryan Hubbell, and Richard T. Burnett. 2013. “Critical Issues in Combining Disparate Sources of Information to Estimate the Global Burden of Disease Attributable to Ambient Fine Particulate Matter Exposure.” Working Paper prepared for the Harvard Center for Risk Analysis workshop, “Methods for Research Synthesis: A Cross-Disciplinary Workshop,” Boston, October 3. https://cdn1.sph.harvard.edu/wp-content/uploads/ sites/1273/2013/09/Shin-et-al.-Sept-2013.pdf Smith, Kirk R., Nigel Bruce, Kalpana Balakrishnan, Heather Adair-Rohani, John Balmes, Zoe Chafe, Mukesh Dherani, H. Dean Hosgood, Sumi Mehta, Daniel Pope, Eva Rehfuess, and others in the HAP CRA Risk Expert Group. 2014. “Millions Dead: How Do We Know and What Does It Mean? Methods Used in the Comparative Risk Assessment of Household Air Pollution.” Annu. Rev. Public Health 35:185–206. http://ehsdiv.sph.berkeley.edu/ krsmith/publications/2014/ARPH%20HAP%20CRA.pdf Smith, K., S. Mehta, and M. Feuz. 2004. “Indoor Air Pollution from Household Use of Solid Fuels.” In Majid Ezzati, Alan D, Lopez, Anthony A. Rodgers, and Christopher J. L. Murray (eds.), Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors. Geneva: World Health Organization. Stemmer, O. 2011. “Clearing the Air: A Comparison of Regulatory Frameworks for Siting Wind Farms.” Journal of Energy & Environmental Law 85 (Summer):85–97. 52 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES UNEP (United Nations Environment Programme). 2008. Reforming Energy Subsidies. Opportunities to Contribute to the Climate Change Agenda. Geneva: UNEP. https://unep. ch/etb/publications/Energy%20subsidies/EnergySubsidiesFinalReport.pdf van Donkelaar, A., R. V. Martin, M. Brauer and B. L. Boys. 2015. “Use of Satellite Observations for Long-Term Exposure Assessment of Global Concentrations of Fine Particulate Matter.” Environmental Health Perspectives 123:135–43. van Donkelaar, A., R. V. Martin, M. Brauer, N. C. Hsu, R. A. Kahn, R. C Levy, A. Lyapustin, A. M. Sayer, and D. M Winker. 2016. “Global Estimates of Fine Particulate Matter Using a Combined Geophysical-Statistical Method with Information from Satellites, Models, and Monitors.” Environ. Sci. Technol. 50:3762–72. Watson, John G., Eric Fujita, Judith C. Chow, Barbara Zielinska, L. Willard Richards, William Neff, and David Dietrich. 1998. “Northern Front Range Air Quality Study Final Report.” DRI Document No. 6580-685-8750.1F2, Desert Research Institute, Reno, Nevada. http:// citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.458.3486&rep=rep1&type=pdf WHO (World Health Organization). 2004. Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors. Geneva: WHO. ———.2009. Estimated Deaths and DALYs Attributable to Selected Environmental Risk Factors, by WHO Member States, 2004. Geneva: WHO. ———.2016. World Health Statistics. Geneva: WHO. World Bank. 2013. United Mexican States: Reducing Fuel Subsidies: Public Policy Options. Report No. ACS3784. Washington, DC: World Bank. https://openknowledge.worldbank. org/bitstream/handle/10986/21755/ACS37840P1129500use0only0900A9R29E6. pdf?sequence=1&isAllowed=y Zhou, Y., J. Levy, J. Evans, and J. Hammitt. 2006. “The Influence of Geographic Location on Population Exposure to Emissions from Power Plants throughout China.” Environment International 32(3):365–73. 53 ENDNOTES ENDNOTES 1 This section is based on Gwilliam, Kojima, and Johnson 2004. 2 India has produced large motorcycles fueled by diesel fuel. 3 See https://www.healtheffects.org/publication/gbd-air-pollution-india for the full list of publications and technical details. 4 https://www.arb.ca.gov/ei/catef/catef.htm. 5 https://www.arb.ca.gov/msei/categories.htm. 6 https://www.arb.ca.gov/ei/areasrc/index0.htm. 7 https://www.sei-international.org/rapidc/gapforum/html/emissions-manual.php. 8 https://www3.epa.gov/ttn/scram/dispersion_prefrec.htm#aermod. 9 https://www3.epa.gov/scram001/receptor_cmb.htm. 10 Subsidies to residential coal can have the opposite effect and of similar magnitude, depending on type of stoves and ventilation. Residential coal constituted 4.6% of total residential energy consumption in countries outside of the Organisation for Economic Co-operation and Development in 2015, of which 79% was consumed in China (IEA 2017). 11 The African Development Bank, Asian Development Bank, European Bank for Reconstruction and Development, European Investment Bank, Inter-American Development Bank, and the World Bank Group are working to harmonize the methodology for calculating emission factors for greenhouse gases, and an analogous approach may be used to calculate pollutant emission factors. 12 GBD MAPS is a multiyear collaboration between the Health Effects Institute (HEI), the Institute for Health Metrics and Evaluation (IHME), Tsinghua University, IIT Mumbai, University of British Columbia and other leading academic centers. 13 PM2.5 from power plants using other fossil fuels is not reported. Ninety-seven percent of fossil fuel used for power generation in China in 2015 was coal (IEA 2017). 14 The authors do not report intake fractions for PM2.5. 15 Cropper and others did not report their estimated intake fractions for power sector emissions in India. 16 The PIF per marginal change in PM2.5 concentrations declines as PM2.5 concentrations increase. 17 Supplementary Appendix, Appendix Table 6b, p. 237. 18 RRs for IHD and stroke are population-age weighted and vary across countries in relation to the age structure of IHD and stroke mortality (see annex 2). 19 Can be calculated for each country at http://vizhub.healthdata.org/gbd-compare/. 20 Living Standard Measurement Study, other national household expenditure surveys, Demographic Health Surveys, and Multiple Indicator Cluster Surveys. 54 GOOD PRACTICE NOTE 8: LOCAL ENVIRONMENTAL EXTERNALITIES DUE TO ENERGY PRICE SUBSIDIES 21 That is, a metric ton of solid fuel is adjusted by the difference in energy content between the solid fuel and the cleaner form of energy, taking into account fuel combustion efficiency. The fuel efficiency of a traditional biomass stove is typically 10–15%, and that for a new LPG stove can be as high as 55%. 22 Can be calculated for each country at http://vizhub.healthdata.org/gbd-compare/. 23 This is based on a conversion of a day of illness to YLD using a disability weight of 0.15, and an average annual wage rate equal to GDP per capita. 24 https://hapit.shinyapps.io/HAPIT/. 25 The single compartment intake fraction (ppm) is iF=Qd*P*106/(u*H*√A) where Qd is breathing rate of air (m3/s), P is population, u is wind speed (m/s), H is mixing height (m), and A is the geographic area (m2). 26 Other solid biomass fuels used by households include straw, shrubs, and grass; agricultural crop residues; and animal dung. 27 With a nonlinear RR function, the precision of the calculation of PAF increases as xi-xi-1 approaches zero, or n approaches infinity. 55 Energy Subsidy Reform Assessment Framework LIST OF GOOD PRACTICE NOTES NOTE 1 Identifying and Quantifying Energy Subsidies NOTE 2 Assessing the Fiscal Cost of Subsidies and Fiscal Impact of Reform NOTE 3 Analyzing the Incidence of Consumer Price Subsidies and the Impact of Reform on Households — Quantitative Analysis NOTE 4 Incidence of Price Subsidies on Households, and Distributional Impact of Reform — Qualitative Methods NOTE 5 Assessing the readiness of Social Safety Nets to Mitigate the Impact of Reform NOTE 6 Identifying the Impacts of Higher Energy Prices on Firms and Industrial Competitiveness NOTE 7 Modeling Macroeconomic Impacts and Global externalities NOTE 8 Local Environmental Externalities due to Energy Price Subsidies: A Focus on Air Pollution and Health NOTE 9 Assessing the Political Economy of Energy Subsidies to Support Policy Reform Operations NOTE 10 Designing Communications Campaigns for Energy Subsidy Reform