Policy Research Working Paper 10005 Air Pollution and Poverty PM2.5 Exposure in 211 Countries and Territories Jun Rentschler Nadia Leonova Sustainable Development Chief Economist Office April 2022 Policy Research Working Paper 10005 Abstract Air pollution is one of the leading causes of death world- levels. Moreover, 716 million poor people (living on less wide, especially affecting poorer people who tend to be than $1.90 per day) live in areas with unsafe air pollution. more exposed and vulnerable. This study contributes (i) Around half of them are located in just three countries: updated global exposure estimates for the World Health India, Nigeria, and the Democratic Republic of Congo. Air Organizations’s 2021 revised fine particulate matter pollution levels are particularly high in lower-middle-in- (PM2.5) thresholds, and (ii) estimates of the number of come countries, where economies tend to rely more heavily poor people exposed to unsafe PM2.5 concentrations. It on polluting industries and technologies. The findings are shows that 7.28 billion people, or 94 percent of the world based on high-resolution air pollution and population maps population, are directly exposed to unsafe average annual with global coverage, as well as subnational poverty esti- PM2.5 concentrations. Low- and middle-income countries mates based on harmonized household surveys. account for 80 percent of people exposed to unsafe PM2.5 This paper is a product of the Sustainable Development Chief Economist Office. 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 author may be contacted at 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 Air Pollution and Poverty: PM2.5 Exposure in 211 Countries and Territories Jun Rentschler1, Nadia Leonova1 JEL codes: Q53, Q56, I14, O44 Keywords: air pollution, PM2.5, poverty, exposure, health Acknowledgments: This working paper contributes to the World Bank’s 2022 flagship report on environmentally harmful subsidies. It has benefited from helpful comments, feedback, and inputs by Mattia Amadio, Esteban Balseca, Samira Barzin, Lander Bosch, Richard Damania, Ira Dorband, Xinming Du, Bramka Jafino, Kichan Kim, Christoph Klaiber, Helena Naber, Jason Russ, Melda Salhab, Ernesto Sanchez-Triana, Margaret Triyana, and Esha Zaveri. 1. Introduction Air pollution has wide-ranging and profound impacts on human health and well-being. Poor air quality has been shown to be responsible for over 4 million death each year from outdoor pollutants, another 2.3 million from indoor air pollution, a wide range of cardiovascular, respiratory, and neurological diseases, as well as reductions in cognitive abilities (WHO, 2018). Studies show that the vast majority of the world’s population face unsafe air pollution levels. Yet, our understanding of what exact air pollution levels should be considered unsafe is still evolving. Based on the latest medical evidence the WHO updated its air quality guidelines in 2021, significantly increasing the stringency of its 2005 guidelines. A growing evidence base also highlights that the exposure to and impact of air pollution are not equally distributed. As health and productivity suffer, evidence from the US has shown that air pollution reinforces socio-economic inequalities. Especially poor and marginalized groups are often exposed to higher pollution levels. In addition, these groups also tend to be more vulnerable to the impacts of pollution, since low-paying jobs are more likely to require physical and outdoor labor, meaning that people are faced with heightened exposure. Pollution sources, such as industrial plants or transport corridors, are disproportionately placed in low-income neighborhoods. And as they increase air pollution further, this drives down housing prices, which in turn reinforces their status as low-income neighborhoods. Constraints in terms of the accessibility, availability and quality of health care provision further increase air pollution related mortality among low-income groups. In short, socio-economic marginalization makes people more exposed and vulnerable to air pollution. Yet, apart from substantial evidence for the US, little evidence exists documenting the global scale of poor people’s exposure to harmful air pollution. A better understanding of the interplay between air pollution and poverty could be crucial for several reasons. Studies from high-income countries on the mortality and morbidity associated with air pollution may not be directly transferable to low-income countries and communities, where the nature of occupations and health care differ substantially. The health and productivity implications of unsafe air pollution will also impact the socio-economic development prospects of low- and middle-income countries. This is especially pertinent in low-income countries, which – as this study shows – still tend to have relatively low pollution levels compared to more industrialized middle-income countries. In these countries, it will be important to ensure that development progress does not come hand in hand with intensifying air pollution and the associated adverse effects. Against this context, this study explores the global prevalence of unsafe outdoor air pollution, and how it interacts with poverty. It uses global high-resolution data on ambient air pollution (outdoor concentrations of PM2.5), population distribution, and poverty to provide aggregate exposure estimates (see Section 3.1 on data sources). It also shows that pollution levels are most hazardous in middle-income countries, where economies tend to rely more heavily on polluting industries and technologies. This study does not cover indoor air pollution, which is another pervasive risk to health and well-being, especially in developing countries. Overall, this study contributes to the literature in two ways. It offers the first global population PM2.5 exposure estimates based on the WHO’s recent guidelines, revised in September 2021, with detailed national and subnational estimates. It also offers the first global estimates of PM2.5 exposure and its interaction with national and subnational poverty levels. The rest of this study is organized as follows. Section 2 reviews the existing evidence on air pollution, health impacts and development outcomes. Section 3 summarizes the data sets and analytical methods used in this study. Section 4 presents results, and section 5 concludes. 2 2. Existing evidence Air pollution is pervasive, caused by a wide range of economic activities. Air pollution refers to the heightened concentration of a range of different pollutants, most notably particulate matter (PM), ozone (O3), nitrogen dioxide (NO2), carbon monoxide (CO), and sulfur dioxide (SO2). While also occurring naturally, hazardous concentration levels of these pollutants are predominantly the result of human activities, including transport, fossil fuel combustion, and agriculture. This also means that their occurrence has high spatial correlation. For instance, major urban centers typically have increased concentrations of all of the above-mentioned pollutants. Using remotely sensed carbon dioxide (CO2) measurements from over 1,200 cities in 138 countries, Dasgupta et al (2021) show that emission levels correlate strongly with population size and economic development levels, though the presence of public transport systems (subways) can reduce local emissions. The vast majority of the world’s population faces air pollution levels that are considered unsafe by the WHO. For PM2.5, one of the most pervasive pollutants, the WHO’s 2005 guidelines recommend for average annual concentrations not to exceed 10 μg/m3 (WHO, 2005). Southerland et al (2022) study PM2.5 concentration trends over two decades for over 10,000 urban centers, and show that 86% (2.5 billion inhabitants) of urban inhabitants lived in urban areas that exceeded the WHO’s 2005 guideline threshold of 10 μg/m3 for PM2.5. The WHO’s (2021) Global Air Quality Guidelines further revise the 2005 guideline threshold down to 5 μg/m3, meaning that an even higher percentage of the world’s population is exposed to unsafe concentrations. Exposure to unsafe levels of particle concentration has a wide range of health impacts. The WHO’s revised threshold reflects a growing body of evidence on the wide-ranging global burden of disease associated with air pollution. There is strong evidence on the causal relationship between air pollution – especially PM, O3, NO2, SO2 – and all-cause mortality, as summarized by the WHO (2021). Heightened air pollution exposure has been shown to increase the risk of chronic obstructive pulmonary disease, ischemic heart disease, lung cancer, stroke, as well as chronic and acute respiratory diseases such as asthma (Cohen et al., 2017; WHO, 2018). Besides cardiovascular and respiratory diseases, growing evidence exists on the role of PM2.5 exposure in increasing the risk of type II diabetes and neurological diseases such as Alzheimer’s (Peters et al., 2019; GBD 2019 Risk Factors Collaborators, 2020). Evidence also suggests that air pollution exposure increases neonatal mortality due to low birth weight and short gestation (GBD 2019 Risk Factors Collaborators, 2020; Patel, et al. 2021). Cumulative exposure to air pollution can also result in a range of health issues that lower long-term quality of life (O’Neill, et al. 2003). Estimates of the number of people affected by air pollution related diseases are staggering. The WHO (2018) estimated that ambient (i.e. outdoor) air pollution caused 4.2 million premature deaths in 2016. Of these deaths, 58 percent were due to ischemic heart disease and stroke, 18 percent were due to chronic obstructive pulmonary disease and acute lower respiratory infections, and 6 percent due to lung cancer. Such ambient air pollution is driven mainly by fossil fuel combustion, transport and industry. In addition, the WHO highlights the serious health risks due to indoor air pollution faced by about 2.6 billion people who rely on coal and biomass for cooking and heating their homes. In 2016, the WHO attributed about 3.8 million premature deaths to such indoor air pollution. The 2019 Global Burden of Disease study (GBD) estimates that 4.5 million premature deaths were due to ambient air pollution and another 2.3 million premature deaths due to indoor air pollution. PM2.5 is identified as the pollutant responsible for the vast majority of air pollution deaths. Focusing specifically on PM2.5 exposure in urban areas, Southerland et al (2022) show that average pollution levels have decreased in most regions between 2000 and 2019. Nevertheless, PM2.5 attributable deaths have increased in the same period in all regions except Europe, and North and Latin 3 America – mostly driven by changes in population sizes, shifting age structures, and pre-existing disease prevalence. They estimate that 1.8 million premature deaths in urban areas can be attributed to PM2.5 exposure in 2019. These numbers suggest that meaningful reductions in air pollution deaths can only be achieved through rapid pollution reduction and improved access to affordable high-quality health care. Besides mortality and morbidity effects, air pollution also reduces cognitive ability and productivity of affected populations. Empirical evidence from brain-training experiments shows that PM2.5 exposure impairs cognitive abilities of adults, while these effects are largest for those in prime working age and those with low ability (La Nauze and Severini, 2021). There is even evidence that short-term exposure to air pollution negatively affects the performance of highly skilled workers – professional baseball umpires have been documented to make more incorrect calls when exposed to high levels of carbon monoxide (Archsmith et al 2018). Such evidence on the relationship between air quality and brain health explains how air pollution can negatively impact productivity and exacerbate inequalities (Peeples 2020). Poor people tend to be more exposed and more vulnerable to air pollution. While global estimates of air pollution deaths differ somewhat from study to study, the evidence is in strong agreement that air pollution is one of the leading causes of deaths (Lelieveld et al. 2020). Several factors are compounding so that this burden is disproportionately borne by low- and middle-income countries. Especially in the developing countries of South and East Asia, large populations are located in densely populated urban areas. Less stringent air quality regulations, the prevalence of older polluting machinery and vehicles, subsidized fossil fuels, congested urban transport systems, rapidly developing industrial sectors, and cut-and-burn practices in agriculture are all contributing to heightened concentration levels. In addition, high proportions of physical and outdoor labor mean that more people are faced with heightened exposure. Constraints in terms of the accessibility, availability and quality of health care provision further increase air pollution related mortality in developing countries (Lelieveld, et al. 2020). Evidence on air pollution in India suggests that a one GW increase in coal-fired capacity corresponds to a 14 percent increase in infant mortality rates in districts near versus far from the plant site (Barrows et al 2019). This effect is 2-3 times larger than estimates from the developed world. The burden of air pollution not only falls disproportionately on poorer countries; within countries poorer and more marginalized communities also tend to be exposed to higher pollution levels (Jbaily et al. 2022; Bell et al 2013). This inequality in exposure is well documented in the United States, where data on socio- economic characteristics is available with high spatial disaggregation (Fann et al, 2018; Bell et al, 2013; Colmer et al, 2020; Kioumourtzoglou et al 2016; Bell & Ebisu, 2012; Mikati et al. 2018). Patel et al. (2021) show that in the US, industrial planning policies have disproportionately placed polluting industries in areas with large Black, Hispanic and Asian populations. As a result, these minority groups face systematically higher pollution levels. Tessum et al. (2021) show that the disproportionate exposure to high PM2.5 exposure by racial-ethnic minorities in the US is not limited to industrial emitters. The study documents that the disproportionate pollution exposure is consistent across all major emission categories, across states, urban and rural areas, income levels, and exposure levels. Studying the period from 2004 to 2016, Jbaily et al (2022) show that in addition to ethnic minorities, low-income populations also face systematically higher PM2.5 pollution levels – and these exposure disparities have been increasing over time. Air quality is also documented to influence real estate prices, so that lower-income households tend to locate into cheaper, more polluted neighborhoods (Chay & Greenstone, 2005). Outside the US, evidence on the inequality in air pollution exposure is more limited – often due to a lack of socio-economic data with high spatial disaggregation. Hajat et al. (2015) offer a systematic review of the literature on socio-economic disparities and air pollution exposure, compiling country- and city-level evidence from all regions. They confirm the strong evidence of unequal exposure in the US, but find mixed 4 evidence for European countries. Only a limited number of studies exists for Asian and African countries, though they tend to confirm the presence of inequality in exposure found in the US. In short, there is convincing evidence from the US and other case studies that poor people are disproportionately exposed and vulnerable to air pollution. Yet, there is limited evidence documenting how many poor people globally face unsafe air quality, and how their pollution burden is distributed across and within developing countries. In this context, we contribute (i) the first updated global exposure estimate for the WHO’s 2021 revised air pollution guidelines, and (ii) the first estimates of the number of poor people exposed to unsafe PM2.5 concentrations. 3. Data and methods 3.1. Data sets used in this study This section details the data sets used in this study to calculate global population exposure to high concentrations of air pollution. Air pollution data (PM2.5): In this paper, we study the differentiated exposure to PM2.5 pollution across countries, rather than the cumulative load of all pollutants. Hence, in this study we focus on PM – one of the most common pollutants, and one that is primarily caused by fossil fuel combustion, such as car engines and coal or gas power plants. Airborne PM is commonly categorized by the diameter of particles, as this determines aerial transport and removal processes as well as their impacts within the respiratory tract (WHO, 2021); i.e. PM2.5 for particles of up to 2.5 µm in diameter, and PM10 (up to 10 µm). In this study we focus on PM2.5, which are among the most harmful pollutants as they can pass through the lungs into the bloodstream and affect other organs. Figure 1. PM2.5 concentrations in Southeast Asia Source: authors’ visualization based on van Donkelaar et al (2021) We use the gridded data set of ground-level fine particulate matter (PM2.5, in µm/m3) concentrations provided by van Donkelaar et al (2021). The data set offers both annual and monthly mean concentrations for the years 1998 to 2019 with global coverage and at a resolution of 0.01 degree. The data set is 5 constructed by combining Aerosol Optical Depth (AOD) satellite retrievals from the NASA MODIS, MISR, and SeaWIFS instruments with the GEOS-Chem chemical transport model, and subsequently calibrating to global ground-based observations using a geographically weighted regression. The resolution of 0.01 degree (equivalent to about 1.1 kilometer at the equator) is well suited to capture regional variation in concentrations, though not granular local variations. As a global modeled data set, some uncertainty is to be expected, though sensitivity tests suggest good agreement with ground measurements (van Donkelaar et al, 2021). For more spatially nuanced analyses, for instance at the neighborhood or street level, alternative data based on local measures would be required. Moreover, it should be noted that the chemical composition of PM2.5 particles can differ by pollution source (Thurston, et al 2022). PM2.5 particles associated with the combustion of fossil fuels are more toxic due to higher acidity levels (e.g. sulfuric particulate matter from coal burning). The global PM2.5 can inform on total particle concentration, but not the spatial variation in the chemical composition (i.e. acidity) of particles. Population density: We estimate the location of people using the WorldPop Global High Resolution Population data set (WPGP), produced by the University of Southampton, the World Bank, and other partners. It offers global coverage and is available yearly from 2000 to 2020. While WorldPop provides several data sets (including poverty, demographics, and urban change mapping), we use the population density map (WorldPop-PPP-2020). In a raster format, this data set provides the number of inhabitants per cell, with a resolution of 3 arcseconds, thus specifying the distribution of population. This information is based on administrative or census-based population data, disaggregated to grid cells based on distribution and density of built-up area, which is derived from satellite imagery (Freire et al., 2016, 2020). The choice of population density map is important for estimating people’s exposure to natural hazards. Smith et al. (2019) provide a sensitivity analysis for flood exposure assessments using different population density maps, including WorldPop (3 arcsecond). They show that high-resolution population density maps perform best in capturing local exposure distribution, particularly the High-Resolution Settlement Layer (HRSL) with 1-arcsecond, or ~30 m resolution, produced jointly by Facebook, Columbia University and the World Bank. While HRSL is only available for a limited number of countries, WorldPop is shown to perform better than alternatives with global coverage, such as LandScan data (30-arcsecond, ~900 m resolution) (Bright et al., 2015). Subnational poverty rates: For 1,755 of the 2,227 subnational units, the World Bank’s Global Subnational Poverty Atlas offers several poverty estimates, which are all derived from the latest available Living Standards Measurement Survey (LSMS) for the respective country (World Bank, 2021). Areas where no poverty estimates are available tend to be high-income countries and small island states. For the purpose of this study, the standard World Bank definitions of poverty are used to determine the number of poor people in a given subnational administrative unit. Specifically, poverty is defined by the daily expenditure thresholds of $1.90, $3.20, and $5.50. Administrative boundaries: The definition of national administrative boundaries follows the standard World Bank global administrative map. However, national boundaries are further disaggregated into subnational units for all countries where World Bank household surveys are available with subnational representativeness. These subnational units are typically provinces or states (i.e. admin1) but can also include custom groupings of subnational regions determined by the sampling strategy of household surveys. Overall, this study covers 211 countries, which are disaggregated into 2,183 subnational units. 6 3.2. Methodology and step-wise computational process To estimate the number of people who are exposed to unsafe air pollution levels, this study follows a computational process in four main steps, which are outlined in this section. For further details, refer to Rentschler and Malhab (2020), who offer a detailed description of the stepwise computational process that is replicated in this study. Resample the PM2.5 data: The air pollution map is resampled to ensure that pixels align with the gridded population density map. Define air pollution risk categories: While the air pollution map offers average annual PM2.5 concentration levels along a continuous scale, the values are aggregated into six risk categories as summarized in Table 1. These risk categories are defined in line with the air quality guideline (AQG) level by the WHO (2021). The WHO recommends an annual PM2.5 AQG level of up to 5 µg/m. For countries with PM2.5 concentrations above this threshold, it recommends interim targets at 10, 15, 25, and 35 μg/m3, which correspond to a linearly increasing mortality rate (Table 1). At higher concentrations, the concentration-response function of mortality may no longer be linear. The epidemiological evidence underpinning these AQG thresholds for PM2.5 is discussed in a systematic review by Chen & Hoek (2020). Each 1 degree cell of each country is assigned one of the six risk categories. This is repeated for the world’s landmass of 149 million square kilometers, which implies the processing of about 300 million data points. Table 1. PM2.5 concentration thresholds Risk classification PM2.5 concentration Mortality Safe No/minor risk ≤ 5 μg/m3 baseline Low 5 – 10 μg/m3 4% increased mortality rate Moderate 10 – 15 μg/m3 8% increased mortality rate High 15 – 25 μg/m3 16% increased mortality rate Unsafe Very high 25 - 35 μg/m3 24% increased mortality rate Hazardous over 35 μg/m3 > 24% increased mortality rate Assign air pollution risk categories to population headcounts at the pixel level and aggregate to the administrative unit: As the air pollution and population density maps are converted into the same spatial resolution, each population map cell can be assigned a unique air pollution risk classification – these cells can then be aggregated to the administrative unit level (e.g. province or district level). This allows the calculation of population headcounts for each risk category and for each (sub-)national administrative unit. This process yields an estimate of the number and share of people exposed to no-, low, moderate, high, very high and hazardous air pollution concentrations throughout the year. We offer these exposure headcount maps as global gridded layers with a resolution of 90 meters. In addition, we aggregate them to administrative units, including for each country and subnational unit. These estimates are further aggregated to yield regional and global estimates. Compute the number of poor people exposed to air pollution risk: While poverty estimates are not available at the pixel level, the World Bank’s Global Subnational Poverty Atlas (GSPA) of harmonized household surveys provides them at the subnational level for at least 153 countries. These poverty shares are multiplied with the population headcount that is estimated to be exposed to unsafe air pollution, in order to obtain an estimate of the number of poor people in each administrative unit exposed to air pollution risk. 7 4. Results For each of the countries analyzed, the results are available as raster files with a 1km spatial resolution and as shapefiles with data aggregated to the admin 1 (sub-national), admin 0 (national), regional, and global levels. In this section, we present visualizations of key findings as maps, using a variety of spatial scales, as well as graphs to highlight pertinent insights. 4.1. Global and regional air pollution exposure Our estimates show that globally 7.3 billion people face air pollution levels considered unsafe by the WHO (i.e. annual average concentration over 5 μg/m3; Figure 2); 6.2 billion people are directly exposed to at least moderate level of air pollution risk (i.e. annual average concentration over 10 μg/m3); for 2.8 billion people pollution levels are hazardous – with PM2.5 concentrations over 35 μg/m3 and over 24 percent increased mortality risk relative to safe areas. Only 462 million people world-wide are exposed to PM2.5 that are considered safe by the WHO (lower than 5 μg/m3). In other words, considering a global population of 7.7 billion (World Bank, 2019), approximately 94 percent of the world’s population are exposed to unsafe levels of PM2.5 concentration. Figure 2. Global population headcounts exposed to different levels of air pollution risk By regionally disaggregating global exposure headcounts, it becomes apparent that air pollution risks are particularly prevalent in certain regions. At 2.2 billion people, the East Asia and Pacific (EAP) region has the highest number of people exposed to unsafe PM2.5 concentrations – corresponding to about 95 percent of EAP’s total population. In the South Asia region (SAR), about 1.8 billion people are exposed to unsafe levels of air pollution – i.e. about 99 percent of the SAR population. In all other regions, the number of people exposed to high levels of air pollution account for a smaller share of the overall population. In the Middle East and North Africa (MENA), Sub-Saharan Africa (SSA), and Europe and Central Asia (ECA) regions and the United States and Canada, between 95 to 92 percent of the respective regional populations are exposed to unsafe PM2.5 concentrations. In Latin America and the Caribbean, exposure as a share of population is slightly lower (84 percent). Figures 3 and 4 provide a breakdown of regional exposure estimates in absolute and relative terms. Figure 3. Regional disaggregation of the global headcount estimates of people exposed to unsafe PM2.5 concentrations 8 Figure 4. Number of people exposed to air pollution by region (and as share of total regional population) 4.2. Countries with the largest air pollution exposed populations Several countries stand out with particularly large populations directly exposed to unsafe air pollution levels, thus corroborating the findings by Landigran et al (2018; Figures 5 and 6). The two most populous countries, China and India, also have the highest absolute population exposure to unsafe air pollution. About 36% of all people exposed to unsafe concentrations of PM2.5 air pollution globally reside in India or China. In India, 1.36 billion people (or 99 percent of the population) are exposed to unsafe PM2.5 concentrations (i.e. over 5μg/m3), of which 1.33 billion (96 percent) face hazardous levels (over 35 μg/m3). In China, 1.41 billion people (or 99 percent of the population) face unsafe PM2.5 concentrations (i.e. over 5μg/m3), of which 0.765 billion (53 percent) face hazardous levels (over 35 μg/m3). Figure 5. Absolute population exposure to high concentrations of air pollution at the country level (millions) 9 Figure 6. Countries with highest population exposure to unsafe PM2.5 levels a) Top 10 countries: Number of people exposed to unsafe concentrations of air pollution (and as share of total national population) b) Top 10 countries: Share of population exposed to hazardous concentrations of air pollution Figure 7 presents relative exposure estimates for all countries. It demonstrates that in large parts of the world and across all regions, the vast majority of the population are exposed to unsafe PM2.5 levels (i.e. over 5μg/m3; Figure 7a). Unlike for example flood hazards, which are highly localized, air pollution tends to cover and move across large areas – for instance blanketing entire cities or regions. As a consequence, if large proportions of a population live in densely populated areas, they tend to be collectively exposed to unsafe pollution levels. Considering a higher pollution threshold of 15 μg/m3 shows that populations in developing countries face heightened pollution (Figure 7b) – high exposure levels range from Central America, Western and Middle Africa, Eastern Europe, Middle East, as well as Central, South, and East Asia. The regions where large parts of the population even face hazardous PM2.5 concentrations (over 35 μg/m3) are Eastern China, the Indian subcontinent, and parts of Western Africa (Figure 7c). 10 Figure 7. Unsafe PM2.5 exposure as share of population a) Percentage of population exposed to PM2.5 over 5 μg/m3 b) Percentage of population exposed to PM2.5 over 15 μg/m3 c) Percentage of population exposed to PM2.5 over 35 μg/m3 11 4.3. Air pollution exposure at the subnational level Country-level estimates provided in the previous section can highlight countries in which air pollution risks are particularly prevalent. However, in most cases air pollution risks are not uniformly distributed across a country but concentrated in densely populated and urbanized areas. In this section we present PM2.5 exposure estimates disaggregated to subnational regions. World Bank household surveys are sampled to be statistically representative at different subnational levels – in this study we adopt these statistically representative subnational units which enable us to compare air pollution exposure estimates with socio- economic characteristics, such as income levels and poverty (Sections 4.4 &4.5). Unsafe air is almost universal among the 20 subnational regions with the largest number of people exposed to PM2.5 concentration over 5 μg/m3 (Figure 8). In these 20 subnational regions, 98.9 percent are exposed to unsafe PM2.5 concentrations (over 5 μg/m3), and 86.3 percent are exposed to hazardous levels (over 35 μg/m3). The most populous regions in China and India dominate the global ranking (Figure 8); of the top 20, ten are in India, nine in China, and one in Pakistan. These subnational regions have in common that they are home to large populations, are densely populated, and host to substantial industrial, agricultural and transport sector activities. For instance, the province of Punjab, Pakistan, with a population of 110 million people, contributes 57 percent of Pakistan’s GDP and hosts large urban centers such as Lahore. In our estimates we also identify 774 subnational areas in which over 99 percent of the population face PM2.5 concentrations over 5 μg/m3. In 33 subnational areas over 99 percent of the population face concentrations over 35 μg/m3. Figure 8. Top 20 regions: Number of people exposed to unsafe PM2.5 concentrations 4.4. Poverty and air pollution Considering the income levels of pollution exposed populations is crucial, as it is well documented that – given the same pollution levels – poor people tend to be more vulnerable to serious health impacts. Poorer people tend to be more exposed to air pollution because they are more likely to depend on jobs that require outdoor physical labor. When affected by pollution related diseases, they also tend to have more limited access to adequate and affordable health care, thus increasing mortality rates. In addition, 12 poorer countries tend to have less developed health care systems. In short, considering the interplay between pollution exposure and poverty can shed light on the vulnerability of affected populations. Poverty and air pollution exposure. By combining air pollution exposure estimates with survey-based sub-national data on poverty, we further estimate air pollution exposure among the world’s poor (Table 2). Our estimates show that 716 million people living in extreme poverty (living on less than $1.90 per day) are directly exposed to unsafe PM2.5 concentrations; of these, 405 million (or 57 percent) are in Sub-Saharan Africa (figure 9). Further, 275 million people living in extreme poverty are exposed to hazardous PM2.5 concentrations (over 35 μg/m3). Approximately one in ten people exposed to unsafe levels of air pollution live in extreme poverty. When poverty is defined using less extreme (i.e. higher) income thresholds, the number of air pollution- exposed poor people increases significantly. Around 1.8 billion are estimated to live in unsafe air pollution areas while living on less than $3.20 a day. The number increases to 2.9 billion when considering incomes below $5.50 a day. Increasing the poverty threshold from $1.90 to $5.50 doubles the number of poor people exposed to unsafe PM2.5 levels in Sub-Saharan Africa from 405 million to 877 million. 1 In South Asia (SAR) the number of the poor and pollution-exposed increases more than six-fold from 220 million to 1,426 million; in East Asia (EAP) the increase is six-fold from 38 million to 229 million. Overall, four of every ten people exposed to unsafe PM2.5 levels globally are living on less than $5.50 a day. Table 2. PM2.5 exposure headcounts at different poverty thresholds Poverty threshold (consumption in US $ per day) $1.90 $3.20 $5.50 Number of poor (millions) 768 1,853 3,034 Share of global population that is poor 9.9% 23.9% 39.2% Number of poor that are exposed to unsafe PM2.5 levels 716 1,752 2,870 (millions; over 5μg/m3) Share of population that is poor and exposed to unsafe 9.3% 22.7% 37.1% PM2.5 levels (over 5μg/m3) Number of poor that are exposed to hazardous PM2.5 levels 275 938 1,573 (millions; over 35μg/m3) Share of population that is poor and exposed to hazardous 3.5% 12.1% 20.3% PM2.5 levels (over 35μg/m3) Figure 9. Number of poor people with unsafe air pollution exposure, at different poverty thresholds and by regions (in billions) 1 In SSA, 39.3 percent of the region’s total population lives in extreme poverty ($1.90), and 91.82 percent of the region’s total population faces unsafe PM2.5 levels (over 5μg/m3). 13 Country-level poverty exposure. Of the 716 million people living in extreme poverty and exposure to unsafe concentrations of air pollution, almost half (48.6%) are located in just three countries. With over 202 million, India has the highest number of extremely poor people exposed to unsafe PM2.5 levels of any country in the world; corresponding to 14.7 percent of India’s overall population. The top 10 countries in terms of number of poor people exposed at $1.90 (figure 10) account for 67.8 percent of all poor people exposed to unsafe PM2.5 concentrations globally. Seven of the top ten countries are located in Sub-Saharan Africa. In the 10 countries with the highest share of the population being poor and exposed to high concentrations of air pollution, about 50 to 85 percent of the population is estimated to be exposed – most of which are in Sub- Saharan Africa (9 out of 10) (figure 11). Figure 12 further highlights that extreme poverty and exposure to unsafe PM2.5 concentrations coincide most acutely in Sub-Saharan Africa; though when considering higher poverty thresholds, exposure also becomes apparent in regions of the Middle East, South and East Asia, as well as Latin America. Figure 10. Top 10 countries - number of poor people (at $1.90/day) exposed to unsafe PM2.5 levels Figure 11. Top 10 countries – percentage of poor people (at $1.90/day) exposed to hazardous PM2.5 levels 14 Figure 12. Regional distribution of air pollution and poverty a) Share of population exposed to unsafe PM2.5 levels and living in poverty at $1.90/day b) Share of population exposed to unsafe PM2.5 levels and living in poverty at $3.20/day c) Share of population exposed to unsafe PM2.5 levels and living in poverty at $5.50/day 15 4.5. Relationship between income and air pollution concentrations The estimates above on the geographic distribution of PM2.5 exposure suggest that pollution levels differ according to the economic development and industrialization stages of countries. The notion of the Environmental Kuznets curve would suggest that air pollution levels would be highest in middle-income countries, where polluting activities like manufacturing dominate the economy while productive capital (e.g. technology) and regulations tend not to prioritize environmental quality. In low-income countries air pollution concentrations would be relatively low, as economic activities (e.g. agriculture) tend to rely less on fossil fuels and the consumption of polluting goods (e.g. high electricity use or private car ownership) is limited to small population groups. In high-income countries pollution would be low, as economic activity tends to be focused on less polluting sectors (e.g. services), polluting activities tend to be off- shored, while clean technologies are widely available and mandated by regulation. This also means that the pollution intensity along the economic development path is not set in stone. Whether today’s low-income countries indeed follow a Kuznets type curve – i.e. intensifying pollution as a byproduct of development – depends on the availability and affordability of clean technologies, as well as the incentive structure for their adoption. For instance, the provision of subsidies for fossil fuel consumption undermines the uptake of clean technologies. This entrenches high pollution levels in low- and middle-income countries, where such subsidies are particularly common (Parry et al., 2021). Figure 13. Mean concentrations of PM2.5 A Kuznets type relationship in air pollution exposure? The estimates from this study show that the vast majority of people breathing unsafe air are located in middle-income countries (Figure 14). The results show that of the 7.3 billion people exposed to unsafe concentrations of PM2.5 (over 5 μg/m3), 3.4 billion (47.3 percent) live in low- or lower-middle-income countries. In the middle-income countries 2.8 billion people are exposed to hazardous PM2.5 levels (over 35 μg/m3), compared to just 40.5 million in low- and high-income countries combined. In relative terms – i.e. as a share of the overall population – PM2.5 exposure is also highest in middle- income countries (Figure 15). About 64.5 percent of people in LMICs are exposed to PM2.5 levels over 35 μg/m3, compared to just 4.4 percent in LICs and 0.9 percent in HICs. The pattern holds regardless of which concentration threshold is considered – exposure to over 5 μg/m3, the safe threshold recommended by 16 the WHO, or 10 μg/m3, or 35 μg/m3 (Figure 15). The regional distribution of PM2.5 concentrations (Figure 3) illustrates that these high ambient air pollution levels in MICs can be explained to a large extent by the rapid economic growth and industrialization of South and East Asia. Figure 14. Number of people exposed to unsafe concentrations of air pollution, disaggregated by national income level classification Figure 15. Global population exposure to PM2.5 concentrations. Top row: Total headcounts in billions. Bottom row: Population share exposed a) concentration over 5 μg/m3 b) concentration over 10 μg/m3 c) concentration over 35 μg/m3 (4% increased mortality rate) (8% increased mortality rate) (over 24% increased mortality rate) Note: Red bars represent the population exposed to PM2.5 concentration levels over 5, 10, and 35 μg/m3 respectively. Gray bars represent the population exposed to levels below the threshold. 17 While this pattern is consistent with a Kuznets type relationship between pollution intensities and income levels, it is useful to consider the full range of PM2.5 concentrations (rather than a single threshold). We compute spatially averaged PM2.5 concentrations for each of the 2,183 subnational areas and compare these with population and income data. Figure 16 suggests support for two notions: areas with larger populations tend to have higher pollution levels and average pollution levels appear particularly high for areas in the middle-income category. Figure 16. Annual average PM2.5 concentrations in 2,183 subnational regions and the relationship with population (left) and mean income (right) We further examine this relationship statistically (Table 3) by considering both the full sample of subnational areas (models I to IV) and a subset of high-pollution areas with average concentrations over 10μg/m3 (models V to VIII). The role of population size is consistent across all specifications. Including population as an explanatory variable for pollution levels raises explanatory power and confirms that larger populations are associated with higher pollution. This is also due to the increased polluting activities that are correlated with larger populations, such as transport, electricity generation, or industrial production. In addition, income levels have some explanatory power. For the full sample (models I to IV), higher income levels are consistently associated with lower pollution (figure 17). Negative coefficients on first and second order polynomials suggest that the relationship between income and PM2.5 is both strictly monotonically decreasing (i.e. negative first order derivative of the functional form) and concave (i.e. non-positive second order derivative). Table 3. The role of income and population in explaining mean PM2.5 concentration levels Mean PM2.5 Full sample Average concentration over >10μg/m3 concentration (I) (II) (III) (IV) (V) (VI) (VII) (VIII) Log income -3.2670*** 0.3342 2.3043** -0.7977 -3.5551*** 2.3383 3.8596*** 1.1722 (0.281) (1.050) (0.997) (1.060) (0.408) (1.493) (1.402) (1.429) Log income ^2 -0.8223*** -1.4078*** -0.4958** -1.4683*** -1.9755*** -0.9029*** (0.231) (0.221) (0.231) (0.358) (0.337) (0.309) Log population 2.7994*** 2.3827*** 3.0789*** 2.6146*** (0.188) (0.160) (0.250) (0.211) Constant 22.3411*** 19.3178*** -20.4817*** -9.6542*** 28.0406*** 23.5092*** -20.2021*** -10.4571*** (0.658) (1.073) (2.851) (2.302) (0.876) (1.406) (3.784) (2.904) Region YES YES N 1707 1707 1707 1707 1049 1049 1049 1049 adj. R2 0.073 0.079 0.185 0.420 0.067 0.081 0.196 0.440 18 When focusing on areas with unsafe PM2.5 levels (here with average PM2.5 concentration over 10μg/m3; models V to VIII), a Kuznets type relationship is discernable with negative coefficients for squared income terms (Figure 17). In other words, among regions with unsafe air, PM2.5 levels increase along with income up to a level of about $9.5/day (or $3,478/year) after which they decrease with income. However, this threshold is low, meaning that for most of the income distribution the negative relationship dominates. 2 Polynomial regression specifications can in principle be prone to over-fitting, but a non-parametric kernel regression confirms the qualitative nature of the relationship between pollution and income as summarized in Figure 17 (for kernel regression results see Annex 2). Figure 17. Fitted lines for models (I) to (VIII) a) Full sample b) Average concentration over >10μg/m3 5. Conclusion This study offers a comprehensive account of the relationship between ambient (outdoor) air pollution exposure, economic development, and poverty in 211 countries and territories. Its global exposure estimates highlight that unsafe air quality is posing significant health risks to a vast majority of the global population. We find that 7.3 billion people, or 94 percent of the world population, live in areas that are exposed to unsafe levels of PM2.5 air pollution (PM2.5 concentration over 5 μg/m3). About 2.8 billion people, or 36 percent of the world population, are directly exposed to concentrations above 35 μg/m3 levels of air pollution, which increase mortality rates by over 24 percent. Yet the study also shows that pollution levels are particularly high in middle-income countries. Here, a wide range of factors are contributing to heightened concentration levels, including less stringent air quality regulations, the prevalence of older polluting machinery and vehicles, subsidized fossil fuels, congested urban transport systems, rapidly developing industrial sectors, and cut-and-burn practices in agriculture. Of the 7.28 billion people exposed to unsafe PM2.5 levels, 80 percent live in low- and middle- income countries. The rapidly growing economies in South and East Asia stand out in terms of absolute exposure, driven by decades of rapid economic growth and industrialization. China (1.41 billion people) and India (1.36 billion) alone account for 36 percent of global exposure to PM2.5 concentrations above WHO guidelines. 2 Note the switching of signs on first order polynomials between the full and restricted sample. Also note the limited increase in explanatory power (R2) from the inclusion of squared income terms. 19 Moreover, this study estimates that 716 million poor people (living on under $1.90 per day) live in areas with unsafe air pollution. At least 405 million people in Sub-Saharan Africa are estimated to live in both extreme poverty (using a $1.90 a day definition) and unsafe air pollution – thus making them particularly vulnerable to prolonged adverse impacts on livelihoods and well-being. For them, the same air pollution level is likely to imply more severe health risks than for higher income households. Among low-income population groups, high proportions of physical and outdoor labor mean that they are faced with heightened exposure and intake of pollutants. Constraints in terms of the accessibility, availability and quality of health care provision further increase air pollution related mortality among poor people. For instance, a study on air pollution and infant mortality suggests that mortality risks are 2 to 3 times as large in India compared to high-income countries. Moreover, exposure to indoor air pollution – not covered in this study – also affects poor people disproportionately, as they are more dependent on the use of polluting low-cost fuels such as charcoal, kerosene, or firewood for cooking and lighting. Air pollution is one of the leading causes of death worldwide, especially affecting poorer people who tend to be more exposed and vulnerable. The estimates provided in this study affirm the case for implementing targeted measures to reduce the pollution intensity of economic growth – for instance, supporting the uptake of less polluting technologies in industry and infrastructure, or facilitating the transition towards cleaner fuels (in particular electrification). In addition, measures are warranted to directly address the disproportionate exposure of poor people highlighted in this study. Expanding the provision of affordable and adequate health care in large urban centers in low- and middle-income countries can help reduce mortality, bringing it closer to the levels experienced in higher income countries. Mandating transparent accounting for environmental and health externalities in planning decisions can help to steer pollution sources (e.g. industrial zones or power plants) away from low-income communities. Finally, removing incentives that perpetuate the over-consumption of fossil fuels can yield a double dividend for poor people; e.g. fossil fuel subsidies are well documented to confer disproportionate monetary benefits to richer households, but the air pollution externalities associated with subsidized fossil fuel consumption are also a burden that is borne disproportionately by poorer households. Addressing these policy distortions can be pro-poor both in terms of fiscal and health benefits. References Adams, K., Greenbaum, D., Shaikh, R., van Erp, A., Russell, A. (2015). Particulate Matter Components, Sources, and Health: Systematic Approaches to Testing Effects. Journal of the Air & Waste Management Association. Vol. 65 (5): 544–58. Archsmith, J., Heyes, A., 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. Vol. 5 (4): 827–63. Barrows, G., Garg, T., Jha, A. (2019). The Health Costs of Coal-Fired Power Plants in India. IZA Discussion Paper Series. IZA DP No. 12838 Bell, M., Ebisu, K. (2012). Environmental inequality in exposures to airborne particulate matter components in the United States. Environmental Health Perspectives. Vol. 120: 1699–1704. Bell, M., Zanobetti, A., Dominici, F. (2013) Evidence on vulnerability and susceptibility to health risks associated with short-term exposure to particulate matter: a systematic review and meta-analysis. American Journal of Epidemiology. Vol. 178: 865–876. Bright, E., Rose, A., Urban, M. (2016) Landscan 2015 High-Resolution Global Population Data Set. Oak Ridge, TN: Oak Ridge National Laboratory (ORNL). Chay, K., Greenstone, M. (2005). Does air quality matter? Evidence from the housing market. Journal of Political Economy. Vol. 113(2): 376-424. 20 Chen, J., Hoek, G. (2020). Long-term exposure to PM and all-cause and cause-specific mortality: a systematic review and meta-analysis. Environment International. Vol. 143(105974) 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 III, A., Shin, H., Straif, K., Shaddick, G., Thomas, M., van Dingenen, R., van Donkelaar, A., Vos, T., Murray, C., Forouzanfar, M. (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. Colmer, J., Hardman, I., Shimshack, J., Voorheis, J. (2020). Disparities in PM2.5 air pollution in the United States. Science. Vol. 369: 575–578. Facebook Connectivity Lab and Center for International Earth Science Information Network - CIESIN. (2018). Available at: http://ciesin.columbia.edu/data/hrsl/#acknowledgements. (Accessed: 15th September 2021). Fann, N., Coffman, E., Timin, B., Kelly, J. T. (2018). The estimated change in the level and distribution of PM2.5- attributable health impacts in the United States: 2005–2014. Environmental Research. Vol. 167: 506–514. Freire, S., Schiavina, M., Florczyk, A., MacManusc, K., Pesaresi, M., Corbane, C., Borkovskac, O., Millsc, J., Pistolesic, L., Squiresc, J., Sliuzasb, R. (2020). Enhanced data and methods for improving open and free global population grids: putting ‘leaving no one behind’ into practice, International Journal for Digital Earth. Vol. 13(1): 61-77. Freire, S., MacManus, K., Pesaresi, M., Doxsey-Whitfield, E., Mills, J. (2016): Development of new open and free multi-temporal global population grids at 250 m resolution. Geospatial Data in a Changing World; Association of Geographic Information Laboratories in Europe (AGILE). 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. Lancet. Vol. 396(10258):1223-1249 Hajat, A., Hsia, C., O’Neill, M. (2015). Socioeconomic Disparities and Air Pollution Exposure: A Global Review. Current Environmental Health Reports. Vol. 2(4): 440–50. Harper, S., Ruder, E., Roman, H., Geggel, A., Nweke, O., Payne-Sturges, D., Levy, J. (2013). Using inequality measures to incorporate environmental justice into regulatory analyses. International Journal of Environmental Research and Public Health. Vol. 10: 4039–4059. Jbaily, A., Zhou, X., Liu, J., Lee, T., Kamareddine, L., Verguet, S., Dominici, F. (2022). Air pollution exposure disparities across US population and income groups. Nature. Vol. 601: 228–233. Kioumourtzoglou, M.A., Schwartz, J., James, P., Dominici, F., Zanobetti, A. (2016). PM2.5 and mortality in 207 US cities: modification by temperature and city. Epidemiology. Vol. 27: 221. La Nauze, A, Severnini, E. (2021). Air Pollution and Adult Cognition: Evidence from Brain Training. National Bureau of Economic Research (NBER). Working Paper Series. No. 28785 Landrigan, P., Fuller, R., Acosta, N., Adeyi, O., Arnold, R., Baldé, A., Bertollini, R., Bose-O'Reilly, S., Boufford, J., Breysse, P. Chiles, T., et al. (2018). The Lancet Commission on pollution and health. The Lancet, 391(10119): 462-512. Lall, S., Lundberg, M., Shalizi, Z. (2008). Implications of Alternate Policies on Welfare of Slum Dwellers: Evidence from Pune, India. Journal of Urban Economics. Vol. 63: 56–73. Lelieveld, J., Pozzer, A., Pöschl, U., Fnais, M., Haines, A., Münzel, T. (2020). Loss of Life Expectancy from Air Pollution Compared to Other Risk Factors: A Worldwide Perspective. Cardiovascular Research. Vol. 116 (11): 1910–17. Mikati, I., Benson, A. F., Luben, T. J., Sacks, J. D., Richmond-Bryant, J. (2018). Disparities in distribution of particulate matter emission sources by race and poverty status. American Journal of Public Health. Vol. 108: 480–485. Myllyvirta, L., Suarez, I., Uusivuori, E., Thieriot, H. (2020). Transboundary Air Pollution in the Jakarta, Banten, and West Java Provinces. Centre for Research on Energy and Clean Air. O’Neill, M., Jerrett, M., Kawachi, I., Levy, J., Cohen, A., Gouveia, N., Wilkinson, P., Fletcher, T., Cifuentes, L., Schwartz, J. (2003). Health, Wealth, and Air Pollution: Advancing Theory and Methods. Environmental Health Perspectives. Vol. 111(16): 1861–70. Parry, I., Black, S., Vernon, N. (2021). Still Not Getting Energy Prices Right: A Global and Country Update of Fossil Fuel Subsidies. IMF Working Paper. WP/21/236 Patel L., Friedman, E., Johannes, S., Lee, S., O'Brien, H., Schear, S. (2021). Air pollution as a social and structural determinant of health. The Journal of Climate Change and Health. Vol.3: 100035. 21 Peeples, L. (2020) News Feature: How air pollution threatens brain health. Proceedings of the National Academy of Sciences (PNAS). June 23, 2020. 117(25): 13856-13860 Peters, R., Ee, N., Peters, J., Booth, A., Mudway, I., Anstey, K. (2019). Air pollution and dementia: a systematic review. Journal of Alzheimer's Disease. Vol. 70. 145-63. Ramanathan, V., Feng, Y. (2009). Air Pollution, Greenhouse Gases and Climate Change: Global and Regional Perspectives. Atmospheric Environment. Vol. 43(1): 37–50. Reames, T., Bravo, M. (2019). People, Place and Pollution: Investigating Relationships between Air Quality Perceptions, Health Concerns, Exposure, and Individual- and Area-Level Characteristics. Environment International. Vol. 122(Jan.): 244–55. Rao, N.D., Kiesewetter, G., Min, J., Pachauri, S., Wagner, F. (2021). Household contributions to and impacts from air pollution in India. Nature Sustainability. Vol. 4: 859–867 Rentschler, J., Salhab, M. (2020). People in Harm’s Way Flood Exposure and Poverty in 189 Countries. World Bank Policy Research Paper Series. WPS 9447 Smith, A., Bates, P., Wing, O., Sampson, C., Quinn, N., Neal, J. (2019) New estimates of flood exposure in developing countries using high-resolution population data. Nature Communications. Vol. 10: 1814 Southerland, V., Brauer, M., Mohegh, A., Hammer, M., van Donkelaar, A., Martin, R., Apte, J., Anenberg, S. (2021) Global urban temporal trends in fine particulate matter (PM2.5) and attributable health burdens: estimates from global datasets. The Lancet Planetary Health, Vol. 6(2): 139-146 Tessum, C., Paolella, D., Chambliss, S., Apte, J., Hill, J., Marshall, J. (2021). PM2.5 polluters disproportionately and systemically affect people of color in the United States. Science Advances. Vol. 7(18). Thurston, G., Chen L.C., Campen, M. (2022) Particle toxicity’s role in air pollution. Science. 3 Feb 2022. Vol 375(6580): 506 van Donkelaar, A., Hammer, M., Bindle, L., Brauer, M., Brook, J., Garay, M., Hsu, C., Kalashnikova, O., Kahn, R., Lee, C., Levy, R., Lyapustin, A., Sayer, A., Martin, R. (2021). Monthly Global Estimates of Fine Particulate Matter and Their Uncertainty. Environmental Science & Technology. Vol. 55(22): 15287–15300 World Bank. (2021). Global Subnational Poverty Atlas (GSAP). Available at https://datacatalog.worldbank.org/int/search/dataset/0042041/international_poverty_line__subnational_p overty (Accessed: 5th November 2021). World Health Organization. (2005). Air Quality Guidelines. Global Update 2005. Particulate Matter, Ozone, Nitrogen Dioxide, and Sulfur Dioxide. Geneva: World Health Organization. World Health Organization. (2018). Burden of disease from ambient air pollution for 2016, version 2. April 2018. Geneva: World Health Organization. World Health Organization. (2021). WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. Geneva: World Health Organization. 22 Annex 1. Country level estimates Empty cells reflect the lack of input data. Poor ($1.9) Poor ($3.2) Poor ($5.5) Poor ($5.5) Safe Unsafe Hazardous and unsafe and unsafe and unsafe and hazardous (< 5 μg/m3) (> 5 μg/m3) (> 35 μg/m3) (> 5 μg/m3) (> 5 μg/m3) (>5 μg/m3) (> 35 μg/m3) Economy/ Pop. (m) (m) (%) (m) (%) (m) (%) (m) (%) (m) (%) (m) (%) (m) (%) territory 1 China 1,431.97 18.54 1.3 1,413.43 98.7 764.9 53.4 21.3 1.5 2 India 1,377.06 17.22 1.3 1,359.85 98.8 1,326.6 96.3 202.5 14.7 716.8 52.1 1,135.1 82.4 1,109.2 80.6 3 United States 329.86 24.98 7.6 304.88 92.4 0.0 0.0 3.0 0.9 3.8 1.1 5.3 1.6 0.0 0.0 4 Indonesia 269.68 18.28 6.8 251.40 93.2 84.1 31.2 9.0 3.4 53.8 19.9 133.2 49.4 41.9 15.5 5 Pakistan 220.52 1.87 0.8 218.66 99.2 205.2 93.1 3.3 1.5 50.5 22.9 139.2 63.1 130.6 59.2 6 Nigeria 205.63 2.47 1.2 203.16 98.8 116.9 56.9 79.7 38.7 141.0 68.6 181.4 88.2 103.3 50.3 7 Bangladesh 162.83 3.42 2.1 159.41 97.9 159.4 97.9 12.4 7.6 64.4 39.5 123.6 75.9 123.6 75.9 8 Brazil 212.30 54.14 25.5 158.15 74.5 0.0 0.0 5.0 2.3 10.7 5.0 25.1 11.8 0.0 0.0 9 Russia 145.60 8.59 5.9 137.00 94.1 0.0 0.0 0.0 0.0 0.5 0.3 4.9 3.4 0.0 0.0 10 Mexico 128.79 8.68 6.7 120.11 93.3 0.0 0.0 2.1 1.6 7.9 6.1 27.5 21.4 0.0 0.0 11 Japan 123.47 7.82 6.3 115.65 93.7 0.0 0.0 0.8 0.7 1.1 0.9 1.4 1.1 0.0 0.0 12 Ethiopia 114.83 0.78 0.7 114.05 99.3 0.0 0.0 27.0 23.5 68.4 59.6 100.3 87.3 0.0 0.0 13 Egypt, Arab Rep. 102.22 4.77 4.7 97.45 95.3 0.0 0.0 3.9 3.8 29.4 28.8 72.1 70.6 0.0 0.0 14 Vietnam 96.80 3.32 3.4 93.48 96.6 15.6 16.2 1.7 1.7 5.9 6.1 20.1 20.8 2.4 2.5 15 Congo, Dem. Rep. 89.41 0.83 0.9 88.57 99.1 11.5 12.9 65.9 73.7 80.1 89.6 86.3 96.6 10.8 12.1 16 Germany 83.70 0.61 0.7 83.10 99.3 0.0 0.0 0.1 0.1 0.2 0.2 0.4 0.5 0.0 0.0 17 Iran, Islamic Rep. 83.76 1.21 1.4 82.55 98.6 0.1 0.1 0.3 0.4 2.7 3.2 12.1 14.4 0.0 0.0 18 Turkey 82.46 4.22 5.1 78.24 94.9 0.0 0.0 0.0 0.0 1.1 1.3 6.7 8.1 0.0 0.0 19 Thailand 69.68 2.18 3.1 67.51 96.9 0.0 0.0 0.0 0.0 0.3 0.5 5.0 7.2 0.0 0.0 20 Philippines 104.57 41.07 39.3 63.50 60.7 0.0 0.0 2.3 2.2 12.4 11.8 31.6 30.3 0.0 0.0 21 United Kingdom 67.23 4.26 6.3 62.97 93.7 0.0 0.0 0.1 0.2 0.2 0.3 0.3 0.5 0.0 0.0 22 France 64.64 1.74 2.7 62.90 97.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 23 Italy 60.02 4.13 6.9 55.89 93.1 0.0 0.0 0.8 1.3 1.0 1.6 1.7 2.8 0.0 0.0 24 Tanzania 57.89 4.78 8.3 53.11 91.7 0.0 0.0 26.0 44.9 40.5 70.0 48.5 83.7 0.0 0.0 25 Myanmar 53.56 0.90 1.7 52.66 98.3 0.2 0.4 0.5 0.9 6.4 11.9 26.0 48.5 0.1 0.2 26 South Africa 59.29 9.43 15.9 49.85 84.1 0.0 0.0 9.1 15.4 18.3 30.8 28.1 47.4 0.0 0.0 27 Colombia 50.78 1.43 2.8 49.35 97.2 1.0 1.9 1.9 3.7 5.1 10.1 13.1 25.8 0.3 0.6 28 Korea, Rep. 50.01 1.30 2.6 48.71 97.4 0.0 0.0 0.1 0.2 0.2 0.5 0.5 1.0 0.0 0.0 29 Kenya 53.72 6.30 11.7 47.42 88.3 0.0 0.0 16.2 30.2 30.3 56.5 40.7 75.8 0.0 0.0 30 Uganda 45.60 0.88 1.9 44.72 98.1 0.0 0.0 17.8 39.1 30.7 67.3 38.9 85.4 0.0 0.0 31 Sudan 43.72 0.24 0.6 43.48 99.4 0.0 0.0 5.6 12.7 19.4 44.4 34.3 78.4 0.0 0.0 32 Ukraine 43.61 1.02 2.3 42.59 97.7 0.0 0.0 0.0 0.0 0.2 0.4 1.5 3.5 0.0 0.0 33 Algeria 43.75 3.81 8.7 39.94 91.3 0.0 0.0 0.1 0.3 0.9 2.1 8.4 19.1 0.0 0.0 34 Iraq 40.15 0.40 1.0 39.75 99.0 0.1 0.2 0.6 1.4 5.3 13.2 19.7 49.1 0.0 0.1 35 Afghanistan 38.82 0.14 0.3 38.69 99.7 11.5 29.5 36 Spain 45.70 7.62 16.7 38.09 83.3 0.0 0.0 0.3 0.6 0.4 0.9 0.8 1.7 0.0 0.0 37 Poland 37.81 0.26 0.7 37.55 99.3 0.0 0.0 0.1 0.3 0.2 0.5 0.4 1.1 0.0 0.0 38 Argentina 44.74 10.68 23.9 34.06 76.1 0.0 0.0 39 Saudi Arabia 34.60 0.71 2.0 33.90 98.0 1.1 3.2 40 Uzbekistan 33.44 0.36 1.1 33.07 98.9 1.4 4.2 4.0 12.1 14.2 42.5 26.3 78.5 1.1 3.2 41 Canada 36.84 4.46 12.1 32.39 87.9 0.0 0.0 0.1 0.2 0.2 0.4 0.2 0.6 0.0 0.0 42 Angola 32.67 2.02 6.2 30.66 93.8 0.0 0.0 13.6 41.7 20.7 63.5 26.6 81.5 0.0 0.0 43 Peru 32.82 2.40 7.3 30.42 92.7 8.5 25.9 0.8 2.5 2.6 7.9 6.8 20.9 0.6 1.7 44 Morocco 36.78 6.47 17.6 30.31 82.4 0.0 0.0 0.2 0.5 1.6 4.4 7.5 20.3 0.0 0.0 45 Ghana 30.99 1.20 3.9 29.79 96.1 19.3 62.3 3.5 11.5 8.3 26.8 15.8 51.1 8.4 27.2 46 Malaysia 31.81 2.02 6.4 29.79 93.6 2.7 8.5 0.0 0.0 0.0 0.1 0.5 1.5 0.0 0.0 47 Nepal 29.03 0.15 0.5 28.87 99.5 28.0 96.4 1.7 5.8 9.7 33.4 20.3 70.0 19.6 67.6 48 Yemen 29.68 1.03 3.5 28.65 96.5 0.0 0.2 14.9 50.3 23.3 78.5 27.0 91.1 0.0 0.2 49 Cameroon 26.38 0.30 1.1 26.09 98.9 4.0 15.2 5.8 22.0 11.1 42.1 17.4 66.0 2.4 9.1 50 Côte d'Ivoire 26.37 0.63 2.4 25.73 97.6 7.4 28.0 6.0 22.6 13.3 50.4 20.1 76.1 4.6 17.3 51 North Korea 25.40 0.80 3.2 24.59 96.8 0.4 1.6 52 Niger 24.10 0.68 2.8 23.42 97.2 0.0 0.0 10.1 42.1 17.9 74.1 21.9 90.8 0.0 0.0 53 Mozambique 31.09 9.06 29.1 22.03 70.9 0.0 0.0 13.3 42.8 17.7 56.9 20.1 64.7 0.0 0.0 54 Venezuela 28.06 6.20 22.1 21.86 77.9 0.0 0.0 55 Taiwan, China 23.68 1.85 7.8 21.83 92.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 23 56 Burkina Faso 20.86 0.09 0.4 20.77 99.6 0.0 0.0 6.9 32.9 14.2 68.3 18.4 88.3 0.0 0.0 57 Sri Lanka 21.16 1.94 9.2 19.22 90.8 0.0 0.0 0.1 0.7 1.9 8.9 7.6 36.0 0.0 0.0 58 Romania 19.21 0.05 0.3 19.16 99.7 0.0 0.0 0.4 2.3 0.9 4.9 2.0 10.2 0.0 0.0 59 Malawi 19.13 0.69 3.6 18.44 96.4 0.0 0.0 12.5 65.4 16.2 84.9 17.8 92.9 0.0 0.0 60 Zambia 18.37 0.18 1.0 18.18 99.0 0.0 0.0 10.6 57.8 13.6 74.1 15.9 86.8 0.0 0.0 61 Mali 20.23 2.19 10.8 18.04 89.2 0.0 0.0 7.7 38.2 13.3 65.8 16.7 82.5 0.0 0.0 62 Kazakhstan 18.72 0.94 5.0 17.78 95.0 0.0 0.0 0.0 0.0 0.1 0.3 1.0 5.4 0.0 0.0 63 Guatemala 17.89 0.30 1.7 17.59 98.3 0.0 0.0 1.3 7.0 3.8 21.0 7.8 43.8 0.0 0.0 64 Syria 17.30 0.30 1.7 17.01 98.3 0.0 0.0 7.5 43.6 13.0 74.9 15.7 91.0 0.0 0.0 65 Ecuador 17.58 0.59 3.4 16.99 96.6 0.0 0.0 0.5 3.0 1.5 8.8 3.9 22.4 0.0 0.0 66 Cambodia 16.62 0.25 1.5 16.37 98.5 0.0 0.0 67 Netherlands 16.98 0.65 3.8 16.33 96.2 0.0 0.0 0.0 0.2 0.0 0.3 0.1 0.3 0.0 0.0 68 Chad 16.41 0.23 1.4 16.18 98.6 0.0 0.0 6.4 39.0 10.9 66.3 14.1 85.7 0.0 0.0 69 Chile 18.98 3.63 19.2 15.34 80.8 0.0 0.0 0.0 0.2 0.1 0.5 0.5 2.5 0.0 0.0 70 Zimbabwe 14.86 0.12 0.8 14.74 99.2 0.0 0.0 4.6 30.6 8.6 57.9 11.7 78.5 0.0 0.0 71 Guinea 12.90 0.44 3.4 12.46 96.6 0.0 0.0 2.9 22.7 7.2 55.9 10.9 84.4 0.0 0.0 72 Rwanda 12.91 0.69 5.3 12.22 94.7 0.0 0.0 6.4 49.7 9.6 74.1 11.1 86.2 0.0 0.0 73 Senegal 16.41 4.56 27.8 11.84 72.2 0.0 0.0 4.1 25.2 7.7 46.7 10.3 62.6 0.0 0.0 74 Bolivia 11.65 0.07 0.6 11.58 99.4 0.0 0.0 0.5 4.3 1.2 10.0 2.5 21.4 0.0 0.0 75 Belgium 11.56 0.17 1.5 11.39 98.5 0.0 0.0 0.0 0.1 0.0 0.2 0.0 0.3 0.0 0.0 76 Burundi 11.86 0.68 5.7 11.18 94.3 0.0 0.0 8.9 74.9 10.4 87.4 11.0 92.3 0.0 0.0 77 South Sudan 11.18 0.04 0.3 11.14 99.7 0.0 0.0 9.5 84.8 10.6 94.8 11.0 98.7 0.0 0.0 78 Benin 12.06 0.99 8.2 11.06 91.8 6.8 56.3 5.3 43.9 8.3 69.2 10.2 84.5 6.1 50.8 79 Tunisia 11.75 0.96 8.1 10.80 91.9 0.0 0.0 0.0 0.4 0.5 4.3 2.6 22.2 0.0 0.0 80 Somalia 15.81 5.04 31.8 10.78 68.2 0.0 0.0 81 Czech Republic 10.70 0.04 0.4 10.66 99.6 0.0 0.0 0.0 0.1 0.0 0.1 0.0 0.4 0.0 0.0 82 Jordan 10.19 0.19 1.8 10.01 98.2 0.0 0.0 0.0 0.2 0.3 2.7 2.3 22.4 0.0 0.0 83 Cuba 11.25 1.27 11.3 9.98 88.7 0.0 0.0 84 Hungary 9.64 0.09 0.9 9.56 99.1 0.0 0.0 0.1 0.5 0.1 1.1 0.3 2.7 0.0 0.0 85 Tajikistan 9.53 0.11 1.2 9.42 98.8 0.2 2.6 0.2 2.3 1.2 12.2 3.9 40.8 0.1 1.1 86 Honduras 9.86 0.45 4.6 9.41 95.4 0.0 0.0 87 Belarus 9.44 0.14 1.5 9.30 98.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.4 0.0 0.0 88 Austria 8.98 0.05 0.5 8.94 99.5 0.0 0.0 0.0 0.3 0.0 0.4 0.1 0.7 0.0 0.0 89 United Arab Emirates 9.42 0.66 7.0 8.76 93.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 90 Serbia 8.74 0.01 0.2 8.72 99.8 0.0 0.0 0.0 0.0 0.1 0.7 0.5 5.7 0.0 0.0 91 Greece 9.94 1.28 12.9 8.66 87.1 0.0 0.0 0.1 0.8 0.1 1.4 0.4 3.9 0.0 0.0 92 Haiti 11.25 2.64 23.5 8.61 76.5 0.0 0.0 2.3 20.1 4.6 40.5 6.9 61.4 0.0 0.0 93 Dominican Republic 10.78 2.40 22.3 8.38 77.7 0.0 0.0 94 Azerbaijan 8.54 0.24 2.8 8.30 97.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 95 Sweden 9.92 1.66 16.7 8.27 83.3 0.0 0.0 0.0 0.2 0.0 0.2 0.0 0.5 0.0 0.0 96 Switzerland 8.61 0.36 4.2 8.25 95.8 0.0 0.0 97 Togo 8.13 0.29 3.6 7.84 96.4 5.5 67.4 3.5 43.3 5.6 68.5 7.0 85.6 4.7 57.6 98 Israel 8.34 1.01 12.1 7.33 87.9 0.0 0.0 0.0 0.2 0.0 0.6 0.2 2.2 0.0 0.0 99 Australia 24.72 17.42 70.5 7.30 29.5 0.0 0.0 0.0 0.2 0.1 0.2 0.1 0.2 0.0 0.0 100 Lao PDR 7.23 0.03 0.4 7.19 99.6 0.0 0.0 0.6 7.9 2.4 33.9 5.1 70.6 0.0 0.0 101 Paraguay 7.11 0.09 1.2 7.03 98.8 0.0 0.0 0.1 1.4 0.4 5.7 1.1 15.6 0.0 0.0 102 Sierra Leone 7.86 0.90 11.4 6.97 88.6 0.0 0.0 3.3 42.2 5.6 71.6 6.6 84.1 0.0 0.0 103 Bulgaria 6.89 0.08 1.2 6.81 98.8 0.0 0.0 0.1 1.3 0.2 3.0 0.5 7.3 0.0 0.0 104 Kyrgyz Republic 6.50 0.10 1.5 6.40 98.5 0.2 3.0 0.0 0.6 0.7 10.6 3.5 53.7 0.1 1.9 105 El Salvador 6.47 0.19 2.9 6.28 97.1 0.0 0.0 0.1 1.5 0.5 7.6 1.6 25.3 0.0 0.0 106 Nicaragua 6.62 0.35 5.3 6.27 94.7 0.0 0.0 0.2 2.8 0.7 11.2 2.1 31.5 0.0 0.0 107 Turkmenistan 6.02 0.24 4.1 5.78 95.9 0.0 0.0 0.1 1.0 0.6 10.1 1.9 31.9 0.0 0.0 108 Libya 6.84 1.13 16.5 5.71 83.5 0.0 0.0 109 Slovak Republic 5.45 0.01 0.3 5.44 99.7 0.0 0.0 110 Portugal 9.64 4.26 44.1 5.38 55.9 0.0 0.0 0.0 0.2 0.0 0.3 0.1 0.9 0.0 0.0 111 Denmark 5.75 0.44 7.6 5.31 92.4 0.0 0.0 112 Congo, Rep. 5.45 0.20 3.6 5.26 96.4 2.7 50.4 2.9 53.1 3.9 72.2 4.7 85.7 2.4 43.8 Hong Kong SAR, 113 6.04 0.93 15.4 5.11 84.6 0.0 0.0 China 114 Papua New Guinea 8.83 3.78 42.8 5.06 57.2 0.0 0.0 1.5 16.5 2.8 31.5 4.0 45.4 0.0 0.0 115 Singapore 5.71 0.85 14.9 4.86 85.1 4.9 85.1 Central African 116 4.79 0.06 1.2 4.73 98.8 0.7 15.6 3.4 71.7 4.1 85.8 4.5 93.7 0.7 13.6 Republic 117 Lebanon 6.33 1.66 26.2 4.67 73.8 0.0 0.0 0.0 0.0 0.0 0.0 0.1 1.3 0.0 0.0 West Bank and 118 4.92 0.33 6.7 4.59 93.3 0.0 0.0 0.0 0.8 0.2 4.1 1.0 20.0 0.0 0.0 Gaza 24 119 Finland 5.52 1.09 19.7 4.43 80.3 0.0 0.0 0.0 0.1 0.0 0.1 0.0 0.1 0.0 0.0 120 Costa Rica 5.08 0.67 13.3 4.40 86.7 0.0 0.0 0.1 1.2 0.1 2.8 0.5 8.9 0.0 0.0 121 Oman 5.00 0.64 12.9 4.36 87.1 0.2 4.5 122 Liberia 5.03 0.72 14.3 4.31 85.7 0.0 0.0 2.1 41.8 3.4 68.3 4.1 81.2 0.0 0.0 123 Moldova 4.01 0.05 1.2 3.96 98.8 0.0 0.0 0.0 0.0 0.0 0.8 0.5 11.3 0.0 0.0 124 Croatia 4.00 0.27 6.9 3.73 93.1 0.0 0.0 0.0 0.5 0.0 1.0 0.1 3.0 0.0 0.0 125 Georgia 3.98 0.37 9.3 3.61 90.7 0.0 0.0 0.2 3.8 0.5 13.3 1.5 36.6 0.0 0.0 126 Eritrea 3.52 0.05 1.5 3.47 98.5 0.0 0.0 127 Panama 4.26 0.94 22.0 3.33 78.0 0.0 0.0 0.0 0.8 0.1 2.9 0.3 7.7 0.0 0.0 128 Kosovo 3.35 0.04 1.1 3.32 98.9 0.0 0.0 0.0 0.3 0.1 2.6 0.7 20.6 0.0 0.0 129 Kuwait 4.14 0.87 21.0 3.27 79.0 2.0 47.7 130 Bosnia-Herzegovina 3.25 0.02 0.7 3.23 99.3 0.0 0.0 0.0 0.1 0.0 0.2 0.1 2.0 0.0 0.0 131 Armenia 2.96 0.05 1.8 2.91 98.2 0.0 0.0 0.0 1.3 0.3 8.9 1.2 41.0 0.0 0.0 132 Albania 2.87 0.14 5.0 2.73 95.0 0.0 0.0 0.0 1.0 0.2 6.6 0.8 28.9 0.0 0.0 133 Norway 5.27 2.55 48.4 2.72 51.6 0.0 0.0 0.0 0.1 0.0 0.2 0.0 0.2 0.0 0.0 134 Lithuania 2.72 0.03 1.2 2.68 98.8 0.0 0.0 135 Qatar 2.87 0.29 10.0 2.58 90.0 1.7 58.9 136 Jamaica 2.89 0.36 12.5 2.53 87.5 0.0 0.0 0.0 1.6 0.2 8.2 0.8 26.2 0.0 0.0 137 Mongolia 3.28 0.86 26.1 2.42 73.9 0.0 0.0 0.0 0.4 0.1 3.4 0.6 18.0 0.0 0.0 138 Botswana 2.35 0.05 2.3 2.29 97.7 0.0 0.0 0.3 13.7 0.8 35.0 1.4 58.1 0.0 0.0 139 Lesotho 2.14 0.07 3.3 2.07 96.7 0.0 0.0 0.6 27.1 1.0 48.7 1.5 71.4 0.0 0.0 140 Macedonia 2.08 0.02 1.2 2.06 98.8 0.0 0.0 0.1 4.1 0.2 8.0 0.4 17.8 0.0 0.0 141 Gambia, The 2.38 0.34 14.4 2.04 85.6 0.0 0.0 0.2 8.6 0.8 32.7 1.5 62.9 0.0 0.0 142 Slovenia 2.06 0.03 1.3 2.03 98.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.0 143 Gabon 2.17 0.15 7.1 2.02 92.9 0.0 0.2 0.1 3.9 0.3 13.7 0.8 36.7 0.0 0.1 144 Latvia 1.86 0.07 4.0 1.79 96.0 0.0 0.0 0.0 0.7 0.0 1.3 0.1 3.0 0.0 0.0 145 Guinea-Bissau 1.85 0.11 5.9 1.74 94.1 0.0 0.0 1.1 58.8 1.4 77.5 1.6 87.2 0.0 0.0 146 Namibia 2.54 0.87 34.2 1.67 65.8 0.0 0.0 0.3 13.3 0.7 26.6 1.1 41.9 0.0 0.0 147 Madagascar 27.42 26.17 95.5 1.25 4.5 0.0 0.0 0.9 3.1 1.1 3.9 1.2 4.3 0.0 0.0 148 Equatorial Guinea 1.39 0.16 11.2 1.23 88.8 0.0 0.0 149 Eswatini 1.16 0.01 0.7 1.15 99.3 0.0 0.0 0.3 28.2 0.6 51.3 0.8 70.8 0.0 0.0 150 Estonia 1.32 0.23 17.4 1.09 82.6 0.0 0.0 151 Bahrain 1.43 0.38 26.6 1.05 73.4 0.7 50.0 152 Timor-Leste 1.31 0.26 20.1 1.04 80.0 0.0 0.0 0.2 16.4 0.7 51.4 1.0 73.0 0.0 0.0 153 Cyprus 1.14 0.18 15.9 0.96 84.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 154 Ireland 4.91 4.03 81.9 0.89 18.1 0.0 0.0 155 Djibouti 0.97 0.22 22.2 0.76 77.8 0.0 0.0 0.2 20.8 0.4 38.0 0.6 58.3 0.0 0.0 156 Bhutan 0.75 0.01 1.2 0.74 98.8 0.4 47.2 0.0 0.7 0.1 8.0 0.2 31.3 0.1 19.9 157 Mauritania 4.62 4.00 86.4 0.63 13.6 0.0 0.0 0.1 1.9 0.2 5.1 0.5 10.4 0.0 0.0 158 Luxembourg 0.62 0.01 1.5 0.61 98.5 0.0 0.0 159 Montenegro 0.62 0.02 3.8 0.59 96.2 0.0 0.0 0.0 0.7 0.0 5.7 0.1 14.0 0.0 0.0 160 Suriname 0.58 0.02 4.0 0.56 96.0 0.0 0.0 0.1 18.6 0.2 28.7 0.3 44.5 0.0 0.0 161 Guyana 0.78 0.22 28.2 0.56 71.8 0.0 0.0 0.0 3.3 0.1 8.3 0.1 18.4 0.0 0.0 162 Belize 0.39 0.02 5.0 0.37 95.0 0.0 0.0 0.0 11.1 0.1 23.2 0.2 44.5 0.0 0.0 163 Brunei Darussalam 0.43 0.07 16.7 0.36 83.3 0.0 0.0 164 Uruguay 3.45 3.13 90.6 0.32 9.4 0.0 0.0 165 Malta 0.42 0.11 25.4 0.32 74.6 0.0 0.0 0.0 0.1 0.0 0.2 0.0 0.2 0.0 0.0 166 Macau, China 0.44 0.16 35.4 0.29 64.6 0.0 0.0 167 São Tomé & Príncipe 0.21 0.09 40.6 0.12 59.4 0.0 0.0 0.0 21.1 0.1 38.8 0.1 51.3 0.0 0.0 168 Andorra 0.08 0.01 9.2 0.07 90.8 0.0 0.0 169 New Zealand 2.63 2.57 97.6 0.06 2.4 0.0 0.0 170 Isle of Man 0.08 0.04 49.1 0.04 50.9 0.0 0.0 171 San Marino 0.03 0.00 1.1 0.03 98.9 0.0 0.0 172 Liechtenstein 0.03 0.01 19.7 0.03 80.3 0.0 0.0 173 Cayman Islands 0.06 0.06 94.2 0.00 5.8 0.0 0.0 174 Puerto Rico 2.86 2.86 100.0 0.00 0.0 0.0 0.0 175 Trinidad & Tobago 1.38 1.38 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 176 Mauritius 1.26 1.26 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 177 Comoros 0.82 0.82 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 178 Fiji 0.79 0.79 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 179 Solomon Islands 0.59 0.59 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 180 Cabo Verde 0.52 0.52 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 181 Bahamas 0.37 0.37 100.0 0.00 0.0 0.0 0.0 182 Iceland 0.33 0.33 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 183 Barbados 0.28 0.28 100.0 0.00 0.0 0.0 0.0 184 Vanuatu 0.27 0.27 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 25 185 French Polynesia 0.26 0.26 100.0 0.00 0.0 0.0 0.0 186 New Caledonia 0.26 0.26 100.0 0.00 0.0 0.0 0.0 187 Samoa 0.20 0.20 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 188 St. Lucia 0.18 0.18 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 189 Guam 0.17 0.17 100.0 0.00 0.0 0.0 0.0 190 Curacao 0.16 0.16 100.0 0.00 0.0 0.0 0.0 191 Micronesia, Fed. Sts. 0.11 0.11 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 192 Grenada 0.11 0.11 100.0 0.00 0.0 0.0 0.0 193 Aruba 0.11 0.11 100.0 0.00 0.0 0.0 0.0 194 St. Vincent & Grenadines 0.10 0.10 100.0 0.00 0.0 0.0 0.0 195 Tonga 0.10 0.10 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 196 Antigua & Barbuda 0.09 0.09 100.0 0.00 0.0 0.0 0.0 197 Seychelles 0.09 0.09 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 198 Dominica 0.07 0.07 100.0 0.00 0.0 0.0 0.0 199 Northern Mariana Is. 0.06 0.06 100.0 0.00 0.0 0.0 0.0 200 American Samoa 0.05 0.05 100.0 0.00 0.0 0.0 0.0 201 St. Kitts and Nevis 0.05 0.05 100.0 0.00 0.0 0.0 0.0 202 Faeroe Islands 0.05 0.05 100.0 0.00 0.0 0.0 0.0 203 Bermuda 0.04 0.04 100.0 0.00 0.0 0.0 0.0 204 Sint Maarten 0.04 0.04 100.0 0.00 0.0 0.0 0.0 205 Greenland 0.04 0.04 100.0 0.00 0.0 0.0 0.0 206 Turks & Caicos Is. 0.04 0.04 100.0 0.00 0.0 0.0 0.0 207 Marshall Islands 0.04 0.04 100.0 0.00 0.0 0.0 0.0 208 Saint-Martin 0.03 0.03 100.0 0.00 0.0 0.0 0.0 209 Maldives 0.02 0.02 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 210 Palau 0.02 0.02 100.0 0.00 0.0 0.0 0.0 211 Tuvalu 0.01 0.01 100.0 0.00 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Annex 2. Non-parametric kernel regression results Figure A.1. Non-parametric kernel regression results a) Full sample b) Average concentration over >10μg/m3 26