Policy Research Working Paper 10398 D etox D evelopment : R epurposing E nvironmentally H armful S ubsidies Background Paper Distributional and Health Co-Benefits of Fossil Fuel Subsidy Reforms Evidence from 35 Countries Christoph Klaiber Jun Rentschler Ira Dorband Sustainable Development Practice Group Office of the Chief Economist April 2023 Policy Research Working Paper 10398 Abstract Governments around the world continue to subsidize fossil suggest that across countries, the absolute consumption fuel use, incentivizing unsustainable consumption levels burden of FFS reform on the richest decile would be 13 with consequences for the global climate and human health. times larger than on the lowest-income decile, supporting However, governments have proven reluctant to reform evidence that FFS are an extremely inefficient way of sup- fossil fuel subsidies (FFS). This is mainly due to concerns porting lower-income groups. In relative terms, however, over potential adverse effects on poverty and equity; the pos- the disparity is much smaller, with the richest decile bearing itive effects on air quality and health are often overlooked. a relative consumption burden that is just 1.1 times larger This study offers new insights on the distributional con- than that borne by the lowest-income decile. In terms of sumption incidence of FFS reforms and expected benefits positive health effects, removing FFS in 25 countries could through improved air quality and health outcomes. Using save a total of 360,000 lives by 2035. The magnitude of the the World Bank-International Monetary Fund Climate health effect depends on country-specific factors, such as Policy Assessment Tool, we conduct country-level analyses the size of initial subsidy programs, and the extent to which of a complete removal of domestic FFS, considering 19 these cover the most polluting fuels. FFS reforms can be a countries for the distributional consumption analysis, and first step in improving air quality and reducing the burden 25 countries for the health benefits analysis. Our findings of disease associated with air pollution. This paper is a product of the Office of the Chief Economist, Sustainable Development Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at cklaiber@worldbank.org and jrentschler@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Distributional and health co-benefits of fossil fuel subsidy reforms—evidence from 35 countries Christoph Klaiber, Jun Rentschler, Ira Dorband Keywords: fossil fuel subsidies, air pollution, health JEL codes: H23, Q41, Q53 Topics: Energy Subsidies, Fuel Subsidies, Air Pollution, Acknowledgements: This study is a technical background paper for the World Bank flagship report Detox Development: Repurposing Environmentally Harmful Subsidies. The authors are grateful for valuable feedback and inputs by Daniel Bastidas Cordova, Simon Black, Richard Damania, Ebad Ebadi, Stephane Hallegatte, Dirk Heine, Victor Mylonas, Jason Russ, Paulina Schulz Antipa, Stephen Stretton, Nate Vernon, and Karlygash Zhunussova. 1. Introduction Despite their large social, economic, and environmental externalities, global spending in fossil fuel subsidies (FFS) remains high. In 2021, governments spent an estimated $577 billion in explicit FFS, not accounting for externalities (Parry, Black and Vernon 2021). One of the most common political justifications for FFS is to support low-income households. But empirical evidence shows that, in relative terms, FFS tend to be regressive, particularly in lower-income countries. That is, subsidies benefit richer households relatively more than poorer households as share of income (Arze del Granado et al. 2012; Dorband et al. 2022; Sterner 2012). So, instead of providing such poorly targeted consumption support, fiscal resources may be better used to improve pre-tax distributions of income as well as economic productivity—for example, through education or health sector investments (Franks et al. 2018; Barbiero and Cournède 2013). Given unequal levels of access, infrastructure-related revenue use may further increase the progressivity of FFS reforms (Calderón and Servén 2014; Dorband et al. 2022). By lowering the price of fossil fuels through subsidies, governments incentivize consumption levels that are environmentally and socially unsustainable, as FFS disincentivize investments in energy efficiency and low-carbon technologies. From an environmental standpoint, FFS reform is considered a key instrument for reaching the Paris Agreement’s target of stabilizing global warming at 1.5–2°C above pre-industrial levels (UNFCCC 2015; 2021). From a social perspective, fossil fuel use increases air pollution levels, and thereby adversely affects human health and economic development in the long run (Reis et al. 2022; Sampedro et al. 2020; Rentschler and Bazilian 2017). Fossil fuel combustion is a main source of anthropogenic air pollution, which harms human health (McDuffie et al. 2021). Pollutants include fine particulate matter (PM2.5), nitrous oxides, sulfur dioxide, black carbon, and carbon monoxide (WHO 2021). Long-term exposure to ambient air pollution increases the risk of chronic obstructive pulmonary disease, ischemic heart disease, lung cancer, and stroke, as well as chronic and acute respiratory diseases such as asthma, and is especially harmful for vulnerable populations, including children, the elderly, or populations with poorer health conditions (Cohen et al. 2017; WHO 2018). This study contributes to existing evidence on the socioeconomic effects of FFS reform, providing new insights on the consumption distributional incidence of reform, as well as on the benefits in terms of air quality and health outcomes. We show that FFS mostly fail to improve consumption distributional outcomes while causing harm to human health. Despite their scale, health costs associated specifically with FFS due to increased air pollution have received little attention in the empirical literature. We assess the magnitude of consumption incidence and health effects in 35 countries, using the World Bank- International Monetary Fund (IMF) Climate Policy Assessment Tool (CPAT), a partial equilibrium global model.1 Because of country level differences in resource abundance, socioeconomic characteristics, and overall level of existing FFS programs between countries, it is important to analyze the effects of FFS reform on a country-by-country basis. First, we use CPAT to estimate the distributional impacts of completely removing FFS in 19 countries. In a second step, we assess the potential health benefits of FFS reforms in 25 high-subsidy, high-air pollution countries. CPAT estimates the policy-induced price increases for all fossil fuels and energy consumption reductions across the whole of the economy and on households, compared to a business-as-usual scenario. Indirect price changes of goods and services are estimated based on detailed input-output data 1 We use CPAT version 1.0pre052. CPAT was developed by World Bank and IMF staff. For descriptions of the model and its parameterization, see IMF (2019, Appendix III) and Parry, Mylonas and Vernon (2021). For further underlying rationale, see Heine and Black (2019). 2 across 14 consumption categories. Direct and indirect price changes of consumption baskets are informed by household-level survey data (cf. Dorband et al. 2019). The estimated economy-wide changes in countries’ primary energy consumption entail changes in air pollutant emissions and concentration levels. Based on relative risk functions for specific air pollutants, associated health risks, and diseases resulting from long-term exposure to these pollutants, changes in mortality relative to the baseline are estimated. Our results on consumption incidence show that, across the sample of countries, FFS fail to improve equity in terms of disposable income while harming human health due to elevated levels of air pollution. On average, FFS reforms without revenue recycling would be progressive or distributionally neutral across our sample of 19 low- and middle-income countries. Relative to disposable income, on average, the richest income deciles would face consumption cost increases 1.1 times higher than the poorest. In absolute terms, consumption expenditures would, on average, increase 12.6 times more for the richest income decile than for the poorest. This means that pre-reform, the lion’s share of FFS accrues to the richest income deciles. In terms of health co-benefits, we find that an immediate and complete phaseout of FFS could reduce– though certainly not eliminate—the global burden of disease associated with premature deaths due to air pollution. We estimate that, across 25 low-, middle-, and high-income countries, FFS reforms could help avert nearly 360,000 premature deaths by 2035 —215,000 in the Russian Federation and China alone. Considering country-specific subsidy programs and sectoral structures, we find that in some cases, a complete phaseout of subsidies with no further regulatory actions could lead to substitution effects toward more polluting fuels. For example, as price differentials change, a relative increase of coal consumption compared to liquefied petroleum gas (LPG), as shown in the United States, increases ambient PM2.5 levels and premature deaths. Yet, overall, our findings suggest that phasing out FFS tends to improve air quality and thus health outcomes. The remainder of this paper is structured as follows. Section 2 offers a summary of the related literature, and section 3 outlines the data and methodology we used in this study. Section 4 offers results on distributional impacts of FFS reform, and section 5 discusses the health and air pollution impacts of FFS reforms. Section 6 summarizes our findings and conclusions. 2. Background and related literature Subsidies are typically justified as alleviating poverty, redistributing fiscal resources, or promoting economic development by supporting energy-consuming activities (Rentschler and Bazilian 2018; Strand 2013; Commander 2012). But evidence suggests that FFS rarely achieve those objectives and are generally detrimental to the economic, social, and environmental dimensions of sustainable development (Parry, Black and Vernon 2021; UN 2015). This literature review updates the review in Rentschler and Bazilian (2017), focusing on evidence of the economic and social effects of FFS. In many countries, the primary objective behind FFS policies and maintaining low energy prices is to facilitate economic growth—for example, by conferring an advantage on domestic energy-intensive firms (De Oliveira and Laan 2010). However, energy prices only play a minor role in explaining enterprise performance in low- and middle-income countries, with factors such as skills, infrastructure efficiency, and access to finance being the main drivers of competitiveness (Dethier, Hirn and Straub 2021). FFS also incentivize unsustainable levels of energy use and disincentivize efficiency improvements which, in the longer term, can even reduce private sector competitiveness and growth prospects (Parry, Black and Vernon 2021). In 2017, the magnitude of global fossil fuel consumption subsidies was almost twice as large as those for renewable energies (IRENA 2020). Apart from direct economic and social costs, FFS are associated with 3 negative externalities on the environment, which in turn adversely affect human and economic development. The International Energy Agency identifies FFS reform as a necessary, though on its own not sufficient, measure for stabilizing global warming at 2°C above preindustrial levels (IEA 2019). Figure 1. Global energy sector subsidies in 2017 ene a les uclear Fossil fuels illions Source: IRENA 2020 FFS aggravate fiscal imbalances and reduce aggregate welfare, regardless of whether the country is an oil importer or exporter (Plante 2014). They have particularly adverse effects on the balance of payments in oil-importing countries, exacerbated by fluctuating international energy prices (Gelb 1988; IMF 2013; Parry, Black and Vernon 2021), and can lead to fuel adulteration and smuggling (Rentschler and Hosoe 2022; Victor 2009; IEA 2010; Calvo-Gonzales, Cunha and Trezzi 2015). Evidence from the Arab Republic of Egypt and the Republic of Yemen shows that FFS cause substantial external costs through traffic congestion, local pollution, and associated health impacts, and incentivize unsustainable water resource use in the agricultural sector (Commander 2012; Coady et al. 2017). Subsidies can reduce fiscal resources available for productive public investments—for example, in health, education, or transport infrastructure—more generally. In República Bolivariana de Venezuela, Algeria, and the Islamic Republic of Iran, subsidy expenditure in 2019 exceeded public spending on health and education. Figures 2 and 3 compare public expenditure for FFS with health and education spending in countries with large relative FFS expenditures and subject to data availability. Figure 2. Government spending on FFS and health in selected countries FFS e pen iture ealth e pen iture Share of Source: World Bank staff calculations, based on 2019 data from Parry, Black and Vernon. 2021 and the World Bank BOOST database2 2 https://www.worldbank.org/en/programs/boost-portal 4 Figure 3. Government spending on FFS education in selected countries FFS e pen iture uca on e pen iture Share of Source: World Bank staff calculations, based on 2019 data from Parry, Black and Vernon. 2021 and the World Bank BOOST database3 Much of the public discourse on FFS in developing countries has focused on benefits to the poor (Dube 2003; Gangopadhyay, Ramaswami and Wadhwa 2005; Adam and Lestari 2008; Mourougane 2010; World Bank 2010; Rao 2012; Ruggeri Laderchi, Olivier and Trimble 2013; IEA 2019; Parry, Black and Vernon 2021). Although supporting poorer households is a common political justification for FFS, most energy subsidies are found to be regressive, benefitting richer households more than poorer ones, relative to disposable income. Yet, in absolute terms, consumption cost increases from subsidy removal could be more difficult to stem for poorer income groups (Dorband et al. 2022; Sterner 2012; Arze del Granado, Coady and Gillingham 2012). Arze del Granado, Coady and Gillingham (2012) find that poorer households across 20 developing countries consume a disproportionately small fraction of total energy consumption, and that households in the top income quintile spend almost 20 times more per capita on most energy goods than those in the bottom quintile. Kerosene is the only exception, with consumption broadly evenly distributed across income quintiles. They also find that the bottom income quintile receives on average about 7 percent of the overall subsidy benefit, and the richest quintile almost 43 percent. Explicit fossil fuel subsidies The World Bank defines explicit energy subsidies as “a eli erate policy action y the government that specifically targets electricity, fuels, or heating and that results in one or more of the following effects: • It reduces the net cost of energy purchased. • It reduces the cost of energy production or delivery. • It increases the revenues retained by those engaged in energy production and delivery (energy suppliers).” (Kojima 2018 p.2) Explicit subsidies take various forms but can be broadly categorized as consumer or producer subsidies (Parry, Black and Vernon 2021; Whitley 2013; IEA 2014). Both types effectively reinforce the underpricing of fossil fuels through active government interventions designed to reduce the cost of fuel consumption or production. Consumer subsidies refer to fiscal measures that lower the price of fossil fuel products below their market price—for example, the international market price, or cost-recovery threshold. They 3 https://www.worldbank.org/en/programs/boost-portal 5 generally measure what governments spend on subsidies, and do not reflect the wider societal externalities incurred by people—for example, due to carbon emissions or air pollution. Producer subsidies are more difficult to observe and quantify, as they refer to different kinds of preferential treatment of fossil fuel exploration, extraction, or processing firms, or other energy-intensive companies, industries, or products (Bast et al. 2014; Bast, et al. 2015; GSI 2010). Explicit producer subsidies include grants, low-interest loans, direct payments, such as upstream support for oil exploration, and tax exemptions; in kind producer subsidies include credit subsidies, government guarantees to protect investment, derivatives and subsidies through government procurement (guaranteed contracts), research, and public investment (UNEP 2003; Whitley 2013; OECD 2011).4 The Global Subsidies Initiative estimates producer subsidies for a series of countries, but these vary widely due to data issues (GSI 2012). Overall, annual producer subsidies are thought to range between $80 billion and $285 billion in emerging and developing countries, and to have reached around $444 billion in Group of Twenty (G20) countries in 2014 (Bast et al. 2015; OECD 2015 2013; Whitley 2013). Based on these definitions, there have been several estimates of the global magnitude of subsidies and their implications. The IMF estimates global explicit FFS (consumer and producer) to be around $577 billion in 2021 (Parry, Black and Vernon 2021), while the IEA estimates it at $440 billion (IEA 2022). The range illustrates the differences in scope, definition, and methodology for measuring and aggregating explicit subsidies. But estimates agree that, around the world, substantial financial resources are being used to artificially lower the price of polluting fossil fuels. The IMF’s estimates of e plicit FFS offer a detailed account of the types of fuel supported by the largest subsidy programs (figure 4). Most are for oil, natural gas, and electricity, with a much smaller share going to diesel, gasoline, and coal. Figure 5 presents the 20 largest providers of fossil fuel consumption subsidies in 2020 and the size of the subsidy programs relative to their gross domestic product (GDP). The five largest subsidy providers—Russia, Saudi Arabia, the Islamic Republic of Iran, República Bolivariana de Venezuela, and India—account for $211 billion, almost half of all global explicit subsidies. Figure 4. Explicit global FFS and Energy Price Index real nergy rice In e illions atural gas lectricity ther oil pro ucts iesel asoline oal erosene nergy price in e right han a is Source: World Bank staff calculations, based on data from Parry, Black and Vernon 2021 4 Such in-kind FFS have also been labeled implicit subsidies, but this differs from the IMF’s definition of implicit subsidies, which reserves the term for environmental externalities associated with fossil fuel use. 6 While consumer subsidies are mostly paid in resource-rich developing countries, many developed economies have large producer subsidy schemes. But, because the in-kind component of these subsidies is particularly difficult to identify and measure, few comprehensive studies exist. Bast et al. (2015) estimate that producer subsidies amounted to $444 billion in 2014—including $97 billion in China, $79 billion in Russia, $52 billion in Saudi Arabia, $50 billion in Brazil, $21 billion in the United States, $9 billion in the United Kingdom, and $5 billion in Australia—with producer subsidies taking the form of direct spending and tax breaks, investments by state-owned enterprises, and public finance from majority government-owned banks and financial institutions. The largest share of production subsidies in 2014 was made up of investments by state-owned enterprises, amounting to $286 billion, or 64 percent of all producer subsidies in G20 countries. Figure 5. The 20 largest FFS providers in 2020 real illions lectricity atural gas iesel ther oil pro ucts asoline erosene oal Source: World Bank staff calculations, based on data from Parry, Black and Vernon 2021. Note: data labels refer to GDP share of FFS. Implicit fossil fuel subsidies Even when fossil fuels are not explicitly subsidized, their prices do not fully reflect the vast societal and environmental damages they cause, including air pollution and climate change. Underpricing fossil fuels reinforces and incentivizes the activities that drive these externalities. The most commonly advocated policy measure for addressing this is to impose an externality tax on polluting fuels and activities. Yet failing to impose such a tax has a similar effect to an explicit subsidy. Instead of pricing in societal costs, it makes fossil fuel consumption unsustainably cheap. The IMF refers to this failure to price in externalities as an implicit FFS (Parry, Black and Vernon 2021; Coady et al. 2015, 2017). It is essentially an estimate of the negative externalities associated with fossil fuel consumption, including the social cost of carbon emissions, local pollution, road congestion, and foregone tax revenues. By including negative externalities, an implicit subsidy also includes reduced 7 energy tax rates. For example, the United Kingdom taxes domestic energy use at 5 rather than 20 percent of value added tax, a reduced rate compared with other consumption goods (HM Revenue and Customs 2022). Figure 6. Implicit global FFS and share of global GDP real Share of glo al illions limate change ocal air pollu on ehical e ternali es Foregone revenue Share of glo al right han a is Source: World Bank staff calculations, based on data from Parry, Black and Vernon 2021 Following this definition, explicit subsidies are dwarfed by implicit subsidies. The IMF estimates the cost of FFS at $5.4 trillion in 2020, with local air pollution and global climate change impacts constituting more than 75 percent of total costs (figure 6). This emphasizes that removing conventionally defined—that is, explicit—subsidies alone cannot bring fossil fuel prices to their social optimum. The IMF definition is particularly relevant from an environmental perspective, as it draws attention to the substantial external costs that result from FFS, such as pollution and carbon emissions. Looking at the 20 largest providers of implicit FFS in 2020 (figure 7), it is clear that almost half of the global implicit subsidies stem from China, the United States, Russia, and India. The close correlation between the climate change externalities and health costs of fossil fuels suggests that there is scope to tackle both problems simultaneously. In all 20 countries, the largest societal costs of fossil fuels are associated with climate change and local air pollution. In a business-as-usual scenario, continued economic growth and its associated energy consumption mean that implicit subsidies are expected to increase to over $7 trillion by 2025. 8 Figure 7. The 20 largest providers of implicit FFS in 2020 real real illions illions limate change ocal air pollu on ehical e ternali es Foregone revenue Source: World Bank staff calculations, based on data from Parry, Black and Vernon 2021 Measuring implicit subsidies also sheds light on the vast environmental costs of certain fossil fuels. Coal is estimated to be the leading culprit, with its environmental costs amounting to more than those of gasoline, diesel, and natural gas combined (figure 8). Figure 8. Global implicit FFS, by fuel type real Share of glo al illions oal iesel asoline atural gas ther oil pro ucts lectricity erosene share of glo al right han a is Source: World Bank staff calculations, based on data from Parry, Black and Vernon 2021 3. Methods and data The results presented in this study use the CPAT modelling framework, which enables quantitative analysis at a country level for roughly 190 countries. CPAT contains projections for, among other things, fuel use, greenhouse gas (GHG) emissions per sector, distributional impacts of environmental policies, and pollution effects. Based on historical data for fossil fuel use in the industrial, transport, power, and residential sectors, CPAT provides corresponding projections for input data and GDP. It also uses historical 9 data to generate general assumptions on the income elasticity of demand, the own-price elasticity of demand for electricity and other fuel products, and changes in technology and energy efficiency. The estimated impacts of FFS reform on fuel use, aggregate consumption, and emissions depends on multiple factors AT mo els the policy’s proportionate impacts on future fuel prices in ifferent sectors, using a simplified model of fuel switching within the power generation sector and own-price elasticities for electricity and fuel use in other sectors. Within the CPAT model, fuel demand curves are based on a constant elasticity specification. Next, it compares the policy scenarios to a baseline scenario, to estimate changes and policy impacts. It estimates the baseline model using IEA data on fuel use by country and sector, GDP projections from the latest IMF forecasts, and data on energy taxes, subsidies, and prices by energy product and country compiled from publicly accessible sources. It then projects international energy prices forward, using the average between IMF and IEA projections and choosing assumptions for fuel price responsiveness to be consistent with empirical evidence. Insights from energy models and fuel price elasticities range from -0.5 to -0.8. (World Bank and IMF 2022). For this paper, we simulate an FFS reform consisting of a complete subsidy removal and lump sum redistribution of reform revenues through cash transfers using CPAT version 1.0pre052. We model a complete phase-out of FFS in the absence of any additional global carbon pricing mechanism, such as a carbon tax or emissions trading system. Existing carbon pricing mechanisms are unchanged. For the FFS reform, we phase out the consumer and producer subsidies present in 2021 within one year and across all fossil fuels and use the resulting reform revenues as a lump sum transfer to the poorest 40 percent of the population. To avoid unrealistic substitution effects, we limit new investments in coal power generation capacity after 2019, effectively limiting substitution effects within the model from relatively cleaner fossil fuels, like natural gas, to dirtier ones like coal. However, such fuel switching is still possible to the extent of existing production and generation capacities. The limitation on coal investments prevents model simulations with a drastic increase in power generation capacities from coal; these would contradict stated climate policies and nationally determined contributions. For our analysis, we focus on two sections of CPAT: the distribution and air pollution modules. Distribution methodology in CPAT FFS reforms increase the prices of previously subsidized energy goods, affecting households both directly, through consumption effects from fuel combustion, and indirectly, through price changes in goods and services that use subsidized fuels as an input. Both price effects impact consumers, as higher prices reduce their purchasing power. Based on a global dataset of existing fuel taxes and subsidies (Parry, Black and Vernon 2021), for each fuel-sector pair, CPAT estimates both price changes for eight energy goods and expected economywide emissions reductions if FFS are removed. Based on these outputs and Global Trade Analysis Project multiregional input-output data (Aguiar et al. 2019), we calculate the expected indirect price changes for the domestically produced share of roughly 60 goods and service sectors in CPAT, as imports do not change price if domestic FFS are reformed. Coupled with a detailed representation of median and mean household consumption baskets across deciles (which are informed by country-level household budget surveys), we estimate in CPAT decile-specific consumption incidences of fiscal reform. The model goes beyond the standard carbon pricing approach in the literature (cf. Dorband et al. 2019; Steckel et al. 2021) as it takes into account a detailed representation of observed prices and subsidies, of imported versus domestically produced consumption, and price-induced behavioral adjustments. 10 Consumption effects are expressed as shares of pre-reform disposable income and in absolute per capita monetary terms on a decile level, differentiated by rural, urban, and overall populations. This allows us to differentiate the effects of an FFS reform on populations within and across countries. Air pollution concentrations Policies aimed at reducing fossil fuel consumption and GHG emissions, such as FFS reforms, can also contribute to reducing ambient and household air pollution by reducing the co-emissions of GHGs and local pollutants. This includes pollutants such as black carbon, carbon monoxide, ammonia, sulfur dioxide, and non-methane volatile compounds, which are responsible for forming particulate pollutants, such as PM2.5. The basis for air pollutant concentrations in CPAT is formed by the change of primary energy consumption and several inputs and estimation methods commonly used in the literature, including: • Country-specific energy consumption in time and scenario, by fuel type and sector • Emissions factors from the Greenhouse Gas – Air Pollution Interactions and Synergies (GAINS) model, an integrated assessment model aimed at exploring cost-effective multipollutant emission control strategies • PM2.5 and ozone concentrations for the baseline year • Emission concentration relations based on source receptor matrices from TM5-FASST, a global atmospheric source-receptor model that provides a rapid impact analysis of emission changes on air quality and short-lived climate pollutants • Relative risk functions for exposure to PM2.5 and ozone, which detail disease incidence at different pollution exposure levels • Population projections in time. Using these inputs, the CPAT framework enables us to model the impact of an environmental tax reform— such as phasing out FFS—on particle emissions, and consequently ambient particle concentrations. Equation (1) offers the functional relationship between energy consumption and change in pollution: ∆ = , ∗ ∆ (1) where ΔE is the change of the aggregate level of emissions for pollutant p, e is the pollution rate in country g of pollutant p, and ΔC is the change in energy consumption in country g. The change of emissions of pollutant p can be translated into changes in overall pollution levels, or more generally the ambient particle concentration of pollutant p, denoted by P in equation (2). If emissions of pollutant p, denoted by Ep, increase, so does the ambient particle concentration P as defined by the functional relationship F(Ep), which is determined through the atmospheric chemical transport model: = ( ) (2) By inserting (1) in (2), we see that the level of ambient particle concentration of pollutant P is a direct function of both the pollution rate and consumption level. So, changes in either will directly affect the ambient particle concentration. Reductions in consumption and decreases in the pollution rate—for example, through technological progress—can reduce the ambient particle concentration and vice versa. Health co-benefits Using relative risk curves for particle disease functions, we can translate the changes in ambient pollutant levels into changes in mortality rates (figure 9), allowing us to estimate the health effects of an FFS reform measured in the number of averted deaths. These estimates are relative to a business-as-usual scenario that assumes no change of environmental policies. The underlying assumption is that exposure to lower levels of ambient air pollution translates to observable changes in population mortality, according to the 11 relative risk curves for particle disease functions. CPAT uses reduced-form approximations to estimate emissions, concentration of pollutants, and their health effects. Figure 9 shows relative risk functions associated with increases of the PM2.5 concentration. Figure 9. Exemplary relative risk functions associated with increased PM2.5 concentration Ischemic heart isease average Stro e average lung tracheal an ronchus cancer rela ve ris hronic o struc ve pulmonary isease ia etes mellitus type II o er respiratory infec ons reterm irth M gm Source: GBD 2019 Risk Factor Collaborators 2020 Because CPAT is based on country-level data as much as possible, the estimates and outputs are specific to individual countries. We selected the countries for our analysis according to size of national subsidy program, air pollution intensity, and data availability. In section 4, we compare results for 19 countries to discuss differences in the distributional impacts of FFS reforms, while offering a more detailed focus on China, Mexico, Brazil, and Indonesia, selected because they have both available household-level data and large FFS programs. In section 5, we analyze the health impacts of an FFS reform for 25 of the countries with the largest subsidy programs. To highlight and analyze differences between countries, we focus on Algeria, China, Indonesia, and the Islamic Republic of Iran, selected because they have large FFS programs and face substantial air pollution challenges. 4. Distributional equity: the winners and losers in subsidy reform We assess the distributional impact of FFS removal in our 19 sample countries, estimating the direct and indirect mean consumption effect across income groups. The policy scenario we analyze simulates the phasing out of all explicit FFS, including all coal, natural gas, LPG, gasoline, diesel, and kerosene. Consumption incidences (and health effects) of FFS removal across countries are as heterogeneous as countries’ subsidy programs. And, although levels and distribution of economic impacts vary strongly— depending on overall program size, types of fuels covered, consumption patterns across households, production technologies, and availability of alternative energy sources—common patterns emerge. 12 Distributional consumption effects in relative terms Consumption effects of FFS removal across our 19 sample countries tend to be small as share of consumption, and, on average, neutrally distributed across deciles. Mean additional expenditures for all income groups due to induced consumer price increases would be well below 1 percent of disposable income in the majority of countries analyzed (Appendix C. Relative mean consumption effect in our sample countries (2030) In distributional terms, on average across countries, the richest income deciles would face consumption cost increases that are 1.1 times higher than those faced by the poorest deciles. So, removing FFS would improve consumption equity, or leave it unchanged in most sample countries, regardless of whether revenues are redistributed. Although heterogeneity across countries is large (figure 10), differences in consumption effects between the poorest and richest deciles are small in relative terms. With few exceptions, differences are below 1 percentage point (panel b). Relative to income, the consumption effects of subsidy reform are similar for the poorest and richest households, with some exceptions. In panel a, the blue bars show the ratio between the relative mean consumption effects of the richest and poorest income deciles. The red area represents countries where the poorest income decile loses more relative to their income than the richest decile. This allows us to distinguish three rough categories of countries, where: 1. The richest income decile is more affected in relative terms than the poorest (five countries: Pakistan, Ecuador, Croatia, Rwanda, and Mexico) 2. The relative consumption effect is similar for the poorest and richest deciles (seven countries: Honduras, Türkiye, the Philippines, Bolivia, Vietnam, Chile, and Indonesia) 3. The poorest income decile is more affected than the richest (seven countries: Argentina, the Dominican Republic, Ukraine, Colombia, China, Brazil, and Costa Rica). Overall, these estimates confirm that, across low- and middle-income countries, low-income households tend to spend a smaller share of their disposable incomes on energy-intensive goods than high-income ones (Dorband et al. 2019). Direct effects from energy expenditure tend to account for a similar share of total incidence across income groups. In most countries, the average relative impact of an FFS reform is similar for the poorest and richest deciles. However, irrespective of the distributional effects, sizable consumption incidences of above 2 and up to 10 percent of disposable income for the lowest-income deciles in some countries call for well-planned strategies to provide immediate consumption support to vulnerable groups, such as through compensation and social protection components. Figure 10. Relative mean consumption effects for richest and poorest income deciles in our 19 sample countries a. Ratio for richest to poorest deciles b. Percentage point difference between richest and poorest deciles 13 a istan cua or cua or a istan roa a roa a an a on uras Me ico Me ico on uras an a T r iye T r iye hilippines ietnam olivia hilippines ietnam hile hile ra il In onesia olivia Argen na ominican epu lic ominican epu lic hina raine osta ica olom ia In onesia hina olom ia ra il Argen na osta ica raine a o losses richest to poorest income percen le ercentage points Panel b shows a similar picture, with the percentage point difference of the mean consumption effect between the richest and poorest income deciles close to zero in some countries. Here, we can distinguish three rough categories of countries, where: 1. The relative consumption effect for the richest households is significantly larger than that of the poorest (two countries: Ecuador and Pakistan) 2. The difference in relative effect size between the richest and poorest is small, making up a very small part of incomes in general (14 countries for Croatia, Honduras, Mexico, Rwanda, Türkiye, Vietnam, the Philippines, Chile, Brazil, Bolivia, the Dominican Republic, China, Costa Rica, and Indonesia) 3. The percentage point difference between losses of the poorest and richest income deciles are significant (three countries—Colombia, Argentina, and Ukraine—where the poorest lose 1.8, 3.5, and 4.6 percentage points more of their disposable income, respectively). Looking at the more detailed results for our four case study countries, we can see that in China, Brazil, and Indonesia, the relative mean consumption effect is larger for the lowest-income households than for the highest (figure 11). This reflects that lower-income deciles spend a larger part of their income on energy and energy-intensive goods and therefore tend to be relatively more affected by an FFS removal than richer ones. But in other countries, as illustrated by Mexico in figure 11, the relative consumption effect from an FFS reform is more similar between income deciles and even slightly bigger for the richest . Depending on the structure of FFS and the extent to which domestic production benefits from them, the level of consumption effect is more or less affected by indirect price changes of goods and services. 14 Figure 11. Relative mean consumption effect in our four sample countries (2030) consump on consump on consump on consump on oo s services using fuel as input erosene atural as asoline iesel il oal lectricity Note: Captures the change in disposable income in relative terms (% consumption). The size of relative consumption effects depends heavily on country-specific factors, such as energy price levels, energy consumption patterns, subsidy size, and the nature of subsidies. For example, in Indonesia, the consumption effect of an FFS reform is estimated to range between 1.2–1.4 percent of people’s consumption expenditures, whereas in China, the effect is much smaller, at 0.08–0.019 percent. Our results suggest that, in our four case study countries in figure 11, the Indonesian population would be most affected by FFS reform. In our wider sample of 19 countries (figure 10), the relative mean consumption effect across income deciles of phasing out FFS is smaller than 1 percent for 12 of the 19 countries. This means that the price effects of subsidy reform cause less than 1 percent loss in disposable income, suggesting that the relative cost of FFS reforms can be quite small. Detailed results for the full country sample are presented in appendix C. Distributional consumption effects in absolute terms In absolute terms, high-income households exhibit higher overall expenditure levels than lower-income households. This also holds for energy-intensive products. More affluent households tend to have more energy-intensive lifestyles, as they own larger (and more) cars and electric appliances, and larger homes. Reflecting our findings for relative terms, we find that they accrue the majority of benefit from energy 15 subsidy schemes in absolute terms. So, when energy subsidies are removed, the highest-income households tend to incur the largest consumption burden in absolute monetary terms. Nonetheless, across countries and even within income groups between rural and urban households, differences are large and understanding them is paramount for designing politically and socially acceptable FFS reforms. Figure 12 shows the absolute mean consumption effects of removing existing FFS in Brazil, China, Indonesia, and Mexico by income decile and fuel type, capturing the change in disposable income in local currency units associated with the direct and indirect price increases. Results for the full country sample are presented in appendix A. Figure 12. Absolute mean consumption effect in our four sample countries (2030) ocal currency units ocal currency units ocal currency units ocal currency units oo s services using fuel as input erosene atural as asoline iesel il oal lectricity Note: Captures the change in disposable income in real 2021 local currency units. As with the relative effects, the absolute effect increases in line with higher overall expenditure levels and potentially more energy-intensive consumption patterns; the absolute monetary consumption burden is therefore larger for higher-income households, which have higher baseline spending. This indicates that in the status quo, the high-income deciles profit most from FFS. Depending on the type of FFS that is under reform and the extent to which the subsidy benefits producers in the baseline, there are direct effects from household fuel combustion and indirect effects from increased prices of domestically produced goods and services. Across our four case studies (figure 12), we can see that: • In Mexico, the indirect effects are larger than the direct effects 16 • In Brazil, price changes for household electricity represent the largest (albeit overall very small) source of change in mean consumption • In Indonesia, two-thirds of the consumption incidence is driven by LPG price changes, but indirect price effects from energy-intensive goods and services are also sizable for higher-income deciles • In China, the largest impact on mean consumption stems from increased natural gas prices, while indirect consumption effects are limited. This may occur when households, rather than firms, are the primary users of subsidized energy goods, so the price increase is not passed along value chains. Energy consumption differs across income groups and country regions. In many countries, urban households directly and indirectly consume significantly more energy than rural households, even at the same income level. This reflects the different consumption patterns of energy-intensive goods and services, such as a higher dependence on daily commuting in urban areas. Consequently, urban populations have been documented to benefit more from FFS programs than rural populations (Dorband et al. 2022; Rentschler 2016), which means they also bear a higher burden in absolute terms when subsidies are phased out. This is consistent with anecdotal observations that public opposition to removing energy subsidies has been particularly fierce in more prosperous urban areas, as evidenced, for instance, in the large-scale protests in Lagos and Abuja in response to igeria’s announcement to remove fuel subsidies (Rentschler 2016). Between rural and urban households, fully removing energy subsidies has a consistently larger median absolute consumption effect for urban populations than their rural counterparts, across all income groups in our four case study countries (figure 13). Differences still vary across countries, being more pronounced, for example, in Brazil than Indonesia. Results for the full country sample are presented in appendix B. Figure 13. Absolute mean consumption effect in urban and rural households in our four sample countries (2030) ocal currency units ocal currency units ocal currency units ocal currency units r an ural 17 Note: Captures the change in disposable income in real 2021 local currency units. To compare absolute consumption effects between high- and low-income deciles, Figure 14 shows the ratio of consumption effects between the richest and poorest income deciles across all 19 countries. For example, in Rwanda, households in the richest decile are estimated to lose 80 times more in absolute monetary terms than households in the poorest. This reflects that an a’s poorest househol s often in rural agricultural communities, consume very limited amounts of fossil fuels. Across our sampled countries, the average mean consumption effect was 12.6 times larger for the richest income decile than the poorest. In figure 14, the red line indicates an impact ratio of one between the richest and poorest, or where the simulated subsidy reform has an equal effect on consumption for the richest and poorest, measured in absolute terms. Of our 19 sample countries, only in Costa Rica is the absolute monetary effect of subsidy reform estimated to be nearly equal between the richest and poorest. For 14 countries, the ratio of consumption losses is greater than five. This confirms that the richest benefit most from FFS in absolute terms in virtually all the countries we considered in our study. Figure 14. Ratio of mean consumption effects in absolute terms for richest to poorest income deciles in our 19 sample countries an a on uras cua or a istan Me ico olivia olom ia hile T r iye Argen na hina roa a hilippines ra il In onesia ominican epu lic ietnam raine osta ica a o of consump on e ect 5. Air pollution and health: FFS reforms save lives Besides their immediate economic and social costs, FFS increase particle pollution levels and are associated with severe negative externalities on the environment, damaging human health and economic development in the long run (Rentschler and Bazilian 2017). Through FFS, governments artificially lower 18 the price of fossil fuels, incentivizing overconsumption and eliminating incentives to invest in energy- efficient technologies, modern and efficient electricity infrastructure, and low-carbon energy sources, such as renewables (IEA 2014). According to the IEA (2014), FFS reforms are a necessary, albeit insufficient, instrument for stabilizing global warming at 2°C above preindustrial levels. In this section, we model a complete and immediate removal of all FFS in 25 countries with the largest subsidy programs, to analyze the health co-benefits of FFS reforms. To highlight and illustrate differences between countries, we present more detailed results for four case study countries: Algeria, China, Indonesia, and the Islamic Republic of Iran. The benefits of FFS reform in terms of reduced air pollution and improved health differs between countries, as FFS reforms do not automatically yield large environmental and health benefits in every case. Three factors determine the extent to which energy subsidy reforms can induce the behavioral and technological changes required to reduce the consumption of polluting fuels: 1. Magnitude of explicit FFS: Depending on an FFS scheme’s primary policy objective, the magnitude of subsidy varies. For example, resource-rich, fossil fuel-exporting, low- and middle-income countries tend to have large consumption subsidy programs. When governments spend substantial shares of the public budget to artificially lower fossil fuel prices, it is likely to induce overconsumption and therefore pollution. Relatively small, targeted subsidy programs, on the other hand, have limited impact on overall energy consumption levels. Naturally, the pollution and health benefits of removing subsidies depend on the magnitude of subsidies and the number of people impacted by adverse health effects. 2. Type and relative price of subsidized fuels: The pollution and health benefits of removing subsidies are likely to be larger in countries that subsidize highly polluting fuel types or industries, such as coal for the power sector, or diesel for transportation. In contrast, when subsidy programs target fossil fuels that have relatively low air pollution footprints, such as LPG, the benefits are more limited. Removing subsidies for cleaner fuels can even increase air pollution—for instance, if the relative price of clean cooking fuels like LPG goes up, low-income households may switch to cheaper, more polluting alternatives, such as charcoal or kerosene. Similarly, removing natural gas subsidies in the power sector may make coal more competitive, especially if it is still subsidized. Relative prices between more and less polluting fuel types are crucial determinants of fuel-switching responses to subsidy removal (Rentschler and Kornejew 2017). 3. Responsiveness of energy consumption to prices: In principle, increasing the unit cost of fuels should cause consumers to use less. In practice, they face a wide range of constraints—related to finance, information, technology, capacity, behavioral biases, and so on—that may leave them unable or unwilling to adjust their consumption in response to price changes. In the short term, fuel consumption can be fairly unresponsive to price changes, but in the longer term, people may choose to move closer to their workplace to reduce commuting needs, or benefit from improvements in public transit infrastructure. Because the overall country effects of phasing out FFS depends on existing policies, we summarize the types of subsidies in our four case study countries. We find that, while all four countries have high FFS levels, the types of fuel they subsidize vary, allowing for differentiated insights into the pollution and health effects of FFS reforms. Algeria paid $11.54 billion in explicit FFS subsidies in 2020, equivalent to 8 percent of GDP. Of this, $6.24 billion went to petroleum and other oil products, $3.13 billion to natural gas, and $2.17 billion to electricity (IMF 2022). These were all explicit consumer, rather than producer, subsidies. And although the magnitude of FFS is high, Algeria does not subsidize coal, the most polluting fuel. 19 China maintains the orl ’s sixth largest FFS program. In 2020, explicit subsidies amounted to $15.73 billion, equivalent to only about 0.1 percent of GDP. Of this, $13.69 billion went to electricity consumers, $1.34 billion to natural gas consumers and producers, $0.37 billion to coal producers, and $0.33 billion to petroleum and other oil producers (IMF 2022). Because coal provided percent of hina’s total generated electricity in 2020 (IEA 2022), electricity and coal subsidies are difficult to disentangle. Indonesia‘s overall FFS program amounted to $11.96 billion in 2020, roughly 1.1 percent of GDP. Of this, $5.49 billion went to electricity, $3.44 billion to petroleum and other oil products, $2.85 billion to coal, and $0.17 billion to natural gas producers. These were mostly explicit consumer subsidies, except for natural gas and $0.13 billion which went to petroleum producers (IMF 2022). Overall, Indonesia’s level of direct coal subsidies is high, compared to the other three countries in our key analysis. The Islamic Republic of Iran’ FFS program amounted to around $41.72 billion in 2020, 22 percent of GDP. Of this, $26.51 billion went to electricity and $15.21 billion to petroleum and other oil products. Of the latter, $10.44 billion was for consumer diesel subsidies (IMF 2022). Of our four case study countries, the Islamic Republic of Iran has the highest level of subsidy, in both absolute and relative terms to GDP. But it does not subsidize coal, and scarcely uses coal to generate electricity, so there are no indirect coal subsidies through the electricity market. Effect on air pollution Economic theory suggests that, all else equal, reducing FFS results in reduced fossil fuel consumption and consequently air pollution. In some cases, there are adverse inter-fuel substitution effects that can reverse this effect, such as when FFS reform results in a reduced relative price of coal. The FFS reform simulations presented in this section translate subsidy program removals into fuel price changes. We use estimates of fuel price elasticities from the literature and data on supply chain relationships (input-output tables), to estimate changes in fuel consumption, and then translate these into changes in sectoral emissions of key air pollutants. Figure 15 summarizes the resulting changes in air pollution emissions and concentrations for different sectors and pollutants. The estimated level changes in particle emissions of six air pollutants following a simulated subsidy reform in 2021 (figure 15a) suggests that particle pollution declines as a result of the FFS reforms across our case study countries but varies with the FFS designs and other economic factors. The strongest reduction in particle emissions is estimated to take place in the Islamic Republic of Iran, driven by the large magnitude of its subsidy program. We estimate that FFS reform would lead to a 40 percent fall in black carbon emissions, and similar effects for PM2.5. Algeria would have the second strongest reduction in particle emissions, with black carbon and PM2.5 levels falling by about 35 percent compared to 2019. In China and Indonesia, the effect is less drastic, but still significant, with FFS reform can helping reduce all air pollutants by 2–4 percent in China, and sulfur dioxide and nitrous oxide emissions by 15 and 6 percent, respectively, in Indonesia. Summarizing the reduction in three of the most toxic air pollutants—PM2.5, nitrous oxides, and sulfur oxides—for all four countries, figure 15b shows that FFS reforms could effectively curb air pollution in the Islamic Republic of Iran and Algeria, but would be less effective in China and Indonesia. Results for the full country sample are presented in appendix D 20 Figure 15. Reduction in air pollutants after phasing out FFS in our four case study countries a) By country In e In e In e In e In e M S Note: The index captures pollutant levels before the FFS reform as 100 (in 2019). b) By pollutant, across countries In e In e In e In e Algeria hina In onesia Iran Islamic ep Note: NOx = nitrous oxides; SO2 = sulfur dioxide; BC = black carbon; VOC = volatile organic compounds; CO = carbon monoxide. Because consumers and sectors use different fuels, it is possible to estimate which sectors are responsible for the main reductions in air pollution resulting from subsidy removal. Figure 16 summarizes the sectoral contributions to estimated changes in particle concentrations compared to a no-FFS reform scenario. The largest differences are in Algeria, where annual average ambient PM2.5 concentrations could be reduced by 3 micrograms per cubic meter of air (µg/m3), in large part from changes in the industrial and road transport sectors. In the Islamic Republic of Iran, the changes are slightly smaller, at roughly 2µg/m3, with the largest effect in the transport sector. In China and Indonesia, the effects of phasing out all FFS would reduce ambient PM2.5 concentrations by roughly 0.2µg/m3 and 0.5µg/m3, respectively. In China, this would be mainly in the residential, services, and construction sector; in Indonesia, it is in the transport sector. Although these national average reductions may appear small, these effects would likely be locally 21 concentrated—for example, around major intersections or industrial plants—where they could have substantial impact. Results for the full country sample are presented in Appendix E. Figure 16. Reductions in PM2.5 pollution in our four case study countries, by source gm gm M M gm gm m3 - 5 5 M M - - 5 - 3 3 3 oal o er lants as o er lants ther o er lants oad trans ort esiden al services and construc on ndustries and other ener ood forestry et chan e Note: Graphs use different y-axis scales to account for and visualize the different magnitudes of sectoral impacts. FFS reforms save lives The relationship between particle concentrations of different pollutants and the incidence of disease has been documented extensively in the medical literature. Air pollution can cause multiple severe health conditions, depending on concentration levels and the chemical composition of particulate matter. For example, heightened PM2.5 exposure has been shown to increase the risk of chronic obstructive pulmonary disease, ischemic heart disease, lung cancer, stroke, and chronic and acute respiratory diseases such as asthma (Cohen et al. 2017; WHO 2018). There is also growing evidence of the role of PM2.5 exposure in increasing the risk of type 2 ia etes an neurological iseases such as Al heimer’s (Peters et al. 2019; GBD 2019 Risk Factors Collaborators 2020). The Global Burden of Disease Study 2019 estimates that 2.3 million and 4.5 million premature deaths result from exposure to indoor and outdoor air pollution, respectively (IHME 2021). This means that, by reducing fossil fuel consumption and associated particle emissions, FFS can directly help reduce a country’s air pollution eaths Yet, it can only be considered a first step toward wider actions to reduce air pollution. Based on estimated reductions in particle emissions and ambient concentrations, it is possible to distinguish the health impacts associated with different air pollutants and their originating 22 sectors, and thus identify which health conditions are associated with air pollution from respective sectors for each country. Figure 17 summarizes the estimated cumulative number of deaths avoided if FFS were phased out in 2021, distinguishing deaths associated with different economic sectors according to their respective contribution to air pollution levels. Results for the full country sample are presented in appendix F. Figure 17. Cumulative averted deaths in our four sample countries by 2035 eaths um averte eaths eaths um averte averte um eaths um averte Am ient o one Am ient M ousehol M et These results are complemented by figure 18a, which shows that immediately removing fossil fuels in our larger country sample of 25 countries would avert an estimated 360,000 aggregate deaths between 2022 and 2035 (figure 18a). This represents the marginal contribution of FFS reforms to wider efforts to improve air quality. Our analysis suggests that FFS and the resulting increase in fossil fuel consumption are major contributors to ambient air pollution-related health impacts and deaths, and shows that FFS reform can have a substantial and lasting positive impact on health and society. Figure 18b shows that in three of the 25 countries, FFS contribute to more than 5 percent of premature aggregate deaths caused by air pollution. Among our four case study countries, we estimate the largest health benefits to be in China, which could avoid roughly 104,000 air pollution deaths within 13 years of an FFS reform; this is followed by the Islamic Republic of Iran, with 31,000 deaths, and Indonesia and Algeria, with roughly 24,000 each (figure 17). We should note that air pollution-related deaths may also increase from some sources—for example, removing LPG subsides in Indonesia could cause low-income households to switch to cheaper, more polluting cooking fuels, such as kerosene and charcoal. In most cases, the reduction in deaths is most 23 strongly driven by lower ambient PM2.5 levels, which is responsible for about 62 percent of all air pollution deaths worldwide (Health Effects Institute 2020). Figure 18. Aggregate averted deaths in our 25 sample countries, by 2035 a) Total aggregate averted deaths b) Share of total air pollution deaths ietnam ietnam ene uela ene uela nite States nite States raine raine T r iye T r iye South Africa South Africa Sau i Ara ia Sau i Ara ia ussian Fe era on ussian Fe era on a istan a istan igeria igeria Me ico Me ico Malaysia Malaysia a a hstan a a hstan Ira Ira Iran Islamic ep Iran Islamic ep In onesia In onesia In ia In ia gypt Ara ep gypt Ara ep cua or cua or hina hina ana a ana a ra il ra il angla esh angla esh Angola Angola Algeria Algeria Subsidy reform simulations for our 25 sample countries highlight cases where subsidy reforms can have particularly large health benefits (figure 18), such as Russia and China, with an estimated 111,000 and 104,000 avoided deaths within 13 years of reform, respectively. In Russia, this can be explained by the high level of subsidies for coal, with its toxic sulfur dioxide and PM2.5 emissions. We estimate that FFS reforms in both countries would reduce coal consumption and therefore improve air quality. Overall, the estimated avoided deaths associated with FFS reforms are driven by several factors, including size of the overall population exposed to hazardous air pollution, high subsidy rates for the most polluting fuel types (particularly coal), and avoiding fuel-switching toward more polluting fuels. In high-income countries, which tend to have more stringent air quality regulations, FFS removals may not yield the same high reductions in air pollution-related deaths. For example, reducing fossil fuel consumption may not significantly reduce particle concentrations, if advanced air filtration systems are already reducing the pollution intensity of consumption. Also, high-income countries do not typically maintain major consumer fuel subsidy programs and may therefore be underrepresented. Canada and the United States, for example, have large implicit producer subsidy schemes which are difficult to quantify and tend to be underreported in global FFS databases, such as the ones we use here (IMF 2022). As we have already noted, it is important to note that in a few cases, FFS reform can increase air pollution deaths, if relative prices shift in favor of the most polluting fuels. Of our 25 sample countries, we estimate that two—India and the United States—would experience a net increase in air pollution deaths if no additional measures are taken. Available FFS data suggest that both countries have higher subsidies for 24 natural gas than coal. So, removing FFS would lead a relatively stronger price increase for natural gas and to people partially substituting natural gas with coal, increasing ambient air pollution—including PM2.5, sulfur dioxide, black carbon and nitrous oxides—resulting in higher estimated mortality rates. For India, this is partly due to a lack of detailed data on coal subsidies, which would likely lead to an underestimation of the relative price of coal after FFS reform; in the United States, it is caused by increased use of coal- fired power generation from existing capacity. This is a stark illustration that removing explicit subsidies is not a panacea for reducing air pollution, and these hypothetical examples serve as reminders that FFS reform design must account for the possibility of fuel-switching effects—for example, by facilitating the transition to less polluting fuel types. Besides its mortality and morbidity effects, air pollution reduces cognitive ability and productivity in affected populations. Empirical evidence from brain-training experiments shows that PM2.5 exposure impairs cognitive abilities in adults, particularly among those of prime working age and with low ability (La Nauze and Severini 2021). There is also evidence that short-term exposure to air pollution negatively affects the performance of highly skilled workers, with professional baseball umpires documented to make more incorrect calls when exposed to higher carbon monoxide levels (Archsmith et al. 2018). Evidence on the relationship between air quality and brain health explains the pernicious and hidden impacts on productivity that exacerbate inequalities (Peeples 2020). 6. Conclusion This study assesses the distributional consumption and health impacts of FFS reforms in 35 countries5 using the World Bank-IMF CPAT. Overall, our results on consumption incidence show that, across the sample of countries, FFS fail to improve equity in terms of disposable income while harming human health due to elevated levels of air pollution. Specifically: • The absolute consumption burden of FFS reform on the richest decile is estimated to be 12.6 times larger than on the lowest-income decile, thus supporting existing evidence that FFS are an extremely inefficient way of supporting lower-income groups. This disparity becomes more apparent when we consider how urban and rural populations benefit from FFS. Across all the countries we analyzed, urban populations benefitted significantly more from existing FFS schemes. This also implies that high-income urban households are likely to incur the largest absolute consumption losses from an (uncompensated) fossil fuel subsidy removal – thus making them important stakeholders to consider in the political economy of subsidy reform. • FFS reforms without revenue recycling would be progressive or distributionally neutral across our sample of 19 low- and middle-income countries. This means, in relative terms the disparity is much smaller, with the richest decile bearing a relative consumption burden that is just 1.1 times larger than that borne by the lowest-income decile. • In terms of positive health effects, removing FFS in 25 countries could prevent 360,000 premature deaths by 2035. This is mostly driven by reductions in ambient PM2.5 pollution, which is heavily associated with fossil fuel combustion, especially in the road transport, residential, and construction sectors. The magnitude of the health benefits of subsidy reform depends on country- specific factors, such as the size of initial subsidy programs, and the extent to which these cover the most polluting fuels. One important issue highlighted in this study is that in certain cases a simple removal of FFS can trigger detrimental substitution effects; for instance, when removing natural gas subsidies could cause a shift 5 We model distributional impacts for 19 countries and air pollution impacts in 25 countries. Due to overlap between the two sets of countries, we consider 35 unique countries in this study. 25 to ar more polluting coal riven y changes in relative prices This sho s that accounting for countries’ subsidy structures when designing plans to end FFS is essential to avoid harmful substitution effects. Overall, the study highlights that FFS reforms can be a first step in improving air quality and reducing the burden of disease associated with air pollution. Higher prices for fossil fuels decrease the demand for polluting fuels, which in turn decreases air pollution from fossil fuel combustion. However, the results also illustrate that FFS reform alone is not sufficient to bring fossil fuel prices to their socially optimal level – in addition to externality taxes, complementary measures can help to accelerate the uptake of cleaner more efficient alternative technologies. References Adam, L and Lestari, E. 2008. “Ten Years of Reforms: The Impact of an Increase in the Price of Oil on Welfare.” Journal of Indonesian Social Sciences and Humanities 1: 121–39. Aguiar , A, Chepeliev, M, Corong Mc ougall an van er Mens rugghe “The TA ata ase Version 10. Journal of Global Economic Analysis 4: 1–27. Archsmith, J, Heyes, A and Saberian, S. 2018. “Air Quality and Error Quantity: Pollution and Performance in a High- skilled, Quality-focused Occupation.” Journal of the Association of Environmental and Resource Economists 5(4): 827–63. Arze del Granado, F J, Coady, D and Gillingham, R. 2012. “The Unequal Benefits of Fuel Subsidies: A Review of Evidence for Developing Countries.” World Development 40(11): 2234–48. Barbiero, O and Cournède, B. 2013. New Econometric Estimates of Long-Term Growth Effects of Different Areas of Public Spending. Bast, E, Doukas, A, Pickard, S, Van Der Burg, L and Whitley, S. 2015. Empty Promises: G20 Subsidies to Oil, Gas and Coal Production. Oil Change International. https://odi.org/en/publications/empty-promises-g20-subsidies- to-oil-gas-and-coal-production/ Bast, E, Makhijani, S, Pickard, S and Whitley, S. 2014. The Fossil Fuel Bailout: G20 Subsidies for Oil, Gas, and Coal Exploration. London and Washington DC: ODI & OCI. Calderón, C A, and Servén, L. 2014. Infrastructure, Growth, and Inequality: An Overview. World Bank Policy Research Working Papers 7034. Calvo-Gonzales, O, Cunha, B and Trezzi, R. 2015. When Winners Feel Like Losers – Evidence from an Energy Subsidy Reform. World Bank Policy Research Working Papers 7227. Coady, D, Parry, I, Sears, L and Shang, B. 2015. How Large Are Global Energy Subsidies? Working Paper 15/105. Washington DC: International Monetary Fund. Coady, D, Parry, I, Sears, L and Shang, B. 2017. “ o arge are Global Fossil Fuel Subsidies?” World Development 91: 11–27. Cohen, A J, Brauer, M, Burnett, R, Anderson, H R, Frostad, J, Estep, K., Balakrishnan, K, Brunekreef, B, Dandona, L, Dandona, R, Feigin, V, Freedman, G, Hubbell, B, Jobling, A, Kan, H, Knibbs, L, Liu, Y, Martin, R, Morawska, L, Pope 3rd, C A, Shin, H, Straif, K, Shaddick, G, Thomas, M, van Dingenen, R, van Donkelaar, A, Vos, T, Murray, C J L and Forouzanfar, M H. 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.” The Lancet 389(10082): 1907–18. Commander, S. 2012. A Guide to the Political Economy of Reforming Energy Subsidies. IZA Policy Paper De Oliveira, A and Laan, T. 2010. Lessons Learned from Brazil's Experience with Fossil-Fuel Subsidies and Their Reform. Geneva: International Institute for Sustainable Development. Dethier, J-J irn M an Strau S “Explaining Enterprise Performance in Developing Countries with Business Climate Survey Data ” The World Bank Research Observer 26(2): 258–309. https://doi.org/10.1093/wbro/lkq007. Dube, I. 2003. “Impact of Energy Subsidies on Energy Consumption and Supply in Zimbabwe. Do The Urban Poor Really Benefit?” Energy Policy 31(15): 1635–45. Franks, M, Lessmann, K, Jakob, M, Steckel, J C, and Edenhofer, O. 2018. “Mo ili ing omestic esources for the Agen a via ar on ricing” Nature Sustainability. 1: 350–357. 26 Gangopadhyay, S, Ramaswami, B and Wadhwa, W. 2005. “Reducing Subsidies on Household Fuels in India: How Will it Affect the Poor? Energy Policy 31(2):125–37. GBD 2019 Risk Factors Collaborators. 2020. “Global Burden of 87 Risk Factors in 204 Countries and Territories, 1990–2019: A Systematic Analysis for the Global Burden of Disease Study 2019.” The Lancet 396(10258): 1223–49 Gelb, A. 1988. Oil Windfalls: Blessing or Curse? Washington DC: World Bank. GSI. 2010. A How-To Guide: Measuring Subsidies to Fossil Fuel Producers. Geneva: Global Subsidies Initiative. GSI. 2012. Fossil Fuels – At What Cost? Government Support for Upstream Oil and Gas Activities in Russia. Geneva: Global Subsidies Initiative. Health Effects Institute. 2020. State of Global Air 2020. Special Report. Boston MA: Health Effects Institute. Heine, D and Black, S. 2019. “Benefits eyon limate nvironmental Ta eform in eveloping ountries ” In Fiscal Policies for Development and Climate Action, edited by Miria A. Pigato, 1–56. Washington DC. https://doi.org/10.13140/RG.2.2.13910.88646 HM Revenue and Customs. 2022. “VAT Rates on Different Goods and Services.” https://www.gov.uk/guidance/vat- on-fuel-and-power-notice-70119 IEA. 2010. World Energy Outlook 2010. Paris: International Energy Agency. IEA. 2014. World Energy Outlook 2014. Paris: International Energy Agency. IEA. 2019. World Energy Outlook 2019. Paris: International Energy Agency. IEA. 2022. Energy Subsidies – Tracking the Impact of Fossil-Fuel Subsidies. https://www.iea.org/topics/energy- subsidies. IHME. 2021. Global Burden of Disease Study 2019 (GBD 2019) Results. Seattle: Institute for Health Metrics and Evaluation. https://vizhub.healthdata.org/gbd-results/. IMF. 2013. Case Studies on Energy Subsidy Reform: Lessons and Implications. Washington DC: International Monetary Fund. IMF. 2022. “Fossil Fuel Subsidies by Country and Fuel Database 2021.” https://www.imf.org/en/Topics/climate- change/energy-subsidies. Last accessed: May 26, 2022. IRENA. 2020. Renewable Capacity Statistics 2020 International Renewable Energy Agency. Abu Dhabi: IRENA. Kojima, M. 2018. Identifying and Quantifying Energy Subsidies: Energy Subsidy Reform Assessment Framework (ESRAF) Good Practice Note 1. ESMAP Paper. Washington DC: World Bank Group. La Nauze, A and Severnini, E. 2021. Air Pollution and Adult Cognition: Evidence from Brain Training. Working Paper Series No. 28785. National Bureau of Economic Research (NBER). Mc uffie Martin Spa aro J urnett Smith S J ’ ourke, P, Hammer, M S, van Donkelaar, A, Bindle, L, Shah, V, Jaeglé, L, Luo, G, Yu, F, Adeniran, J A, Lin, J and Brauer, M. 2021. “Source Sector and Fuel Contributions to Ambient PM2. 5 and Attributable Mortality Across Multiple Spatial Scales.” Nature Communications 12(1): 1–12. Mourougane, A. 2010. Phasing Out Energy Subsidies in Indonesia. OECD Economics Department Working Papers 808. OECD. 2011. Fossil Fuel Support: OECD Secretariat Background Report for G20 Meeting of Finance Ministers. Paris: Organisation for Economic Co-operation and Development. OECD. 2013. Inventory of Estimated Budgetary Support and Tax Expenditures for Fossil Fuels 2013. Organisation for Economic Co-operation and Development. http://oecdinsights.org/2013/02/11/fossil-fuel-subsidies- billions-up-in-smoke/. OECD. 2015. OECD Companion to the Inventory of Support Measures for Fossil Fuels 2015 . Paris: Organisation for Economic Co-operation and Development. Parry, I, Black, S and Vernon, N. 2021. Still Not Getting Energy Prices Right: A Global and Country Update of Fossil Fuel Subsidies. Parry, I, Mylonas, V and Vernon, N “Mitigation Policies for The Paris Agreement: An Assessment for G20 Countries ” Journal of the Association of Environmental and Resource Economics 8: 797–823 Peeples, L. 2020. “News Feature: How Air Pollution Threatens Brain Health.” Proceedings of the National Academy of Sciences (PNAS) 117(25): 13856–60. Peters, R, E, N, Peters, J, Booth, A, Mudway, I and Anstey, K. 2019. “Air Pollution and Dementia: A Systematic evie ” Journal of Alzheimer's Disease 70: 145–63. 27 Plante, M. 2014. “The Long-run Macroeconomic Impacts of Fuel Su si ies ” Journal of Development Economics 107: 129–143. Rao, N D. 2012. “Kerosene Subsidies in India: When Energy Policy Fails as Social Policy.” Energy for Sustainable Development 16 (1): 35–43. Rentschler, J. 2016. “Incidence and Impact: The Regional Variation of Poverty Effects due to Fossil Fuel Subsidy Reform.” Energy Policy 96: 491–503. Rentschler, J and Bazilian, M. 2017 “Reforming Fossil Fuel Subsidies: Drivers, Barriers and the State of rogress ” Climate Policy 17(7): 891–914. Rentschler, J and Bazilian, M. 2018. “Principles for Designing Effective Fossil Fuel Subsidy Reforms.” Review of Environmental Economics and Policy 11(1). Rentschler, J and Hosoe, N. 2022. Illicit Schemes: Fossil Fuel Subsidy Reforms and the Role of Tax Evasion and Smuggling Policy Research Working Paper 9907. Washington DC: World Bank. Rentschler, J and Kornejew, M. 2017. “ nergy rice ariation an ompetitiveness Firm-Level Evidence from In onesia ” Energy Economics 67: 242–54. Ruggeri Laderchi, C, Olivier, A and Trimble, C. 2013. Balancing Act: Cutting Energy Subsidies while Protecting Affordability. Policy Research Working Paper 76820. Washington DC: World Bank. Sampedro, J, Smith, S J, Arto, I, González-Eguino, M, Markandya, A, Mulvaney, K M, Pizarro-Irizar, C and van Dingenen, R. 2020. “Health Co-Benefits and Mitigation Costs as per the Paris Agreement Under Different Technological Pathways for Energy Supply“. Environment International 136: 105513. Steckel, J C, Dorband, I I, Montrone, L, Ward, H, Missbach, L, Hafner, F, Jakob, M and Renner, S. 2021. Distributional Impacts of Carbon Pricing in Developing Asia. Nature Sustainability, 4(11), 1005-1014. Sterner, T. 2012. Fuel Taxes and the Poor: The Distributional Effects of Gasoline Taxation and Their Implications for Climate Policy. London: Routledge. Strand, J. 2013. Political Economy Aspects of Fuel Subsidies: A Conceptual Framework. Policy Research Working Paper 6392. Washington DC: World Bank. UN. 2015. Open Working Group Proposal for Sustainable Development Goals. New York: United Nations. UNEP. 2003. Energy Subsidies: Lessons Learnt in Assessing Their Impact and Designing Policy Reforms. Geneva: United Nations Environment Programme. UNFCCC. 2015. Adoption of the Paris Agreement. Proposal by the President. Draft Decision-/CP.21. United Nations Framework Convention on Climate Change. http://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdf UNFCCC. 2021. Annual Report 2021. United Nations Framework Convention on Climate Change. https://unfccc.int/sites/default/files/resource/UNFCCC_Annual_Report_2021.pdf Victor, D. 2009. The Politics of Fossil-Fuel Subsidies. Geneva: Global Subsidies Initiative. Whitley, S. 2013. Time to Change the Game: Fossil Fuel Subsidies and Climate. London: Overseas Development Institute. WHO. 2018. Burden of Disease from Ambient Air Pollution for 2016, version 2. Geneva: World Health Organization. WHO. 2021. WHO Global Air Quality Guidelines: Particulate Matter (PM2.5 and PM10), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide. World Health Organization. World Bank. 2010. Subsidies in the Energy Sector: An Overview. Background Paper for the World Bank Group Energy Sector Strategy. Washington DC: World Bank. World Bank and IMF. 2022. Climate Policy Assessment Tool – Technical notes.. 28 Appendix A. Absolute mean consumption effect in our sample countries (2030) 29 Appendix B. Absolute mean consumption effect in our sample countries, urban vs. rural (2030) 30 Appendix C. Relative mean consumption effect in our sample countries (2030) 31 Appendix D. Reduction in air pollutants in our 25 sample countries after phasing out FFS Note: NOx = nitrous oxides; SO2 = sulfur dioxide; BC = black carbon; VOC = volatile organic compounds; CO = carbon monoxide. Index set to 100 for all pollutants pre FFS reform (2019). 32 Appendix E. Reductions in PM2.5 pollution in our 25 sample countries, by source 33 Appendix F. Cumulative averted deaths in our 25 sample countries, by 2035 34