The mortality and economic benefits of achieving air pollution standards in India

Significance

Ambient PM_2.5 pollution poses a serious health threat throughout the world, but especially in India. We examine the mortality benefits of reducing annual ambient PM_2.5 to the Indian and World Health Organization (WHO) 25 μg/m³ and 15 μg/m³ interim standards. These estimates would be higher if household air pollution were zero, underscoring the importance of considering total exposures when estimating benefits of ambient standards. Pairing each standard with a 50% reduction in the percentage of households burning solid fuels yields mortality reductions that exceed the sum of reductions from implementing each policy separately. Implementing each standard paired with a 50% reduction in households using solid fuels yields economic benefits of 2.2%, 3.6%, and 4.9% of gross domestic product (GDP).

Abstract

The presence of high levels of PM_2.5 pollution is a major health threat in India. We evaluate the benefits of reducing ambient PM_2.5 and household air pollution from a 2019 baseline, when over 70% of the population of India was exposed to annual average PM_2.5 of more than 40 μg/m³ and 56% of households burned solid fuels for cooking. We estimate the health benefits from reduced mortality that arise when lowering ambient PM_2.5 to the India standard (40 μg/m³) and to two World Health Organization (WHO) interim standards (25 μg/m³ and 15 μg/m³). The benefits of the ambient standards are moderated by the high levels of household air pollution to which people are exposed due to the concavity of the exposure–response function. We therefore examine the benefits of achieving each standard paired with a 50% reduction in the percentage of households burning solid fuels. Annual mortality reductions from the combined policies exceed 300,000 lives saved for the India standard, 500,000 lives saved for the WHO 25 μg/m³ interim standard, and 675,000 lives saved for the WHO 15 μg/m³ interim standard. The economic benefits of each of these joint policies equal 2.2%, 3.6%, and 4.9% of Indian gross domestic product (GDP), using a Value per Statistical Life of 100 times per capita GDP. Results underscore the importance of considering household air pollution when evaluating ambient standards and the synergistic benefits of combined policies when the exposure–response function is concave.

Levels of ambient air pollution in Indian cities are among the highest in the world. In 2019, annual average ambient PM_2.5 pollution exceeded the World Health Organization (WHO) Interim Target 3 (15 μg/m³) (1) in virtually all cities in India and was over 75 μg/m³ in many cities in the Indo-Gangetic Plain (2). Exposure to PM_2.5 is widely viewed as contributing to premature mortality (3). Beginning in 2010, the Institute for Health Metrics and Evaluation has estimated deaths attributed to ambient PM_2.5, by region, in the Global Burden of Disease (GBD) studies (4–6). The 2019 GBD (7) estimates that approximately 980,000 people in India died because of ambient PM_2.5 exposure in 2019. There are, however, only a few studies (8–13) of the benefits of reducing ambient PM_2.5 exposure in India and none of which we are aware that address the benefits of reducing ambient PM_2.5 for all parts of India using the methodology we employ.

In this paper, we estimate the avoided mortality and associated economic benefits of lowering annual average PM_2.5 in all Indian districts to the national annual average standard of 40 μg/m³ (14) and to two interim WHO targets—25 μg/m³ (Interim Target 2) and 15 μg/m³ (Interim Target 3). The benefits of reducing annual average PM_2.5 to these targets are estimated for each district in India using baseline levels of PM_2.5 and baseline mortality for 2019. Reductions in mortality due to ischemic heart disease (IHD), stroke, lung cancer, type II diabetes, chronic obstructive pulmonary disease (COPD), and lower respiratory infection (LRI) are valued using the human capital approach (i.e., the present value of output lost due to premature mortality) and the Value per Statistical Life (VSL). Although the mortality analysis was conducted at the district level, results are summarized at the state level to facilitate ease of reference and interpretation.

Because the health impacts of ambient PM_2.5 depend on the total dose of PM_2.5 received, the benefits of reducing ambient levels of particulate matter depend on the household air pollution (HAP) to which people are exposed. We estimate the benefits of reducing ambient PM_2.5 controlling for HAP at the exposure levels estimated in the 2019 GBD study (7), using the concentration–response functions in the GBD (15). We estimate the benefits of reducing PM_2.5 levels in the year 2019, using population cohorts and mortality data for that year.^*

The concavity of exposure–response functions in the GBD implies that, for a given microgram reduction in ambient PM_2.5, the percentage reductions in mortality are more modest in states where total PM_2.5 exposure is high, a point noted by Sankar et al. (16) and Pope et al. (17). This implies that the benefits of measures to reduce ambient air pollution (AAP) will increase as HAP levels drop. We therefore estimate the benefits of programs to reduce ambient PM_2.5 with 50% reductions in the proportion of households burning solid fuels relative to a 2019 baseline.

We also examine separate policies to reduce ambient air pollution and HAP. Due to the concavity of exposure–response functions, the benefits of implementing these policies together are greater than the sum of the benefits from implementing each policy separately. This illustrates the synergies from reducing household and ambient air pollution—synergies that are even greater in areas where HAP is a source of ambient air pollution (2, 18).

We evaluate the benefits in 2019 of reducing ambient PM_2.5 and HAP from a 2019 baseline, when over 70% of the population of India was exposed to annual average PM_2.5 of more than 40 μg/m³ and 56% of households burned solid fuels for cooking (7). Fig. 1A shows population-weighted averages of annual average ambient PM_2.5 in 2019, by state, based on ref. 19. (See also SI Appendix, Table S1, Column 1.) As Fig. 1A shows, six states and Union Territories (UTs) in South India (Puducherry, Goa, Karnataka, Kerala, Andhra Pradesh, and Tamil Nadu) had on average achieved the 40 μg/m³ standard in 2019, as had four states in the Northeast (Manipur, Mizoram, Nagaland, and Sikkim). States in the Indo-Gangetic Plain, Central India, and Delhi would see the most difference in ambient PM_2.5 if the 40 μg/m³ standard were achieved. In contrast, all states would benefit from reducing ambient PM_2.5 to the WHO 25 μg/m³ or the WHO 15 μg/m³ interim standards.

Fig. 1.

(A) Annual average ambient PM_2.5 by state in 2019, population weighted and (B) average annual PM_2.5 exposure due to AAP and HAP, by state, in 2019.

Since the benefits of reducing ambient PM_2.5 also depend on HAP, SI Appendix, Fig. S1 maps the GBD estimates of the percentage of households burning solid fuels for cooking in 2019 (7). These are also listed in SI Appendix, Table S1 (Column 2), together with the additional PM_2.5 (Column 3) to which household members are exposed from burning solid fuels (7). When these exposure estimates are combined with estimates of ambient PM_2.5, average annual PM_2.5 exposure from all sources exceeds 100 μg/m³ in eleven states and UTs in India, as shown in Fig. 1B and SI Appendix, Table S1 (Column 4). It exceeds 125 μg/m³ in five states in the northern part of India, including Bihar and Uttar Pradesh, where population-weighted annual average PM_2.5 exposure from AAP and HAP exceeds 150 μg/m³.

We estimate the benefits of reducing ambient annual PM_2.5 exposure at the district level, using concentration–response functions from the GBD (15). SI Appendix, Fig. S2, which illustrates the relative risk of IHD for a person 70 y of age or older as a function of annual average PM_2.5 exposure, shows that for annual average exposures greater than 100 μg/m³, the reduction in risk for even a 50 μg/m³ reduction in PM_2.5 is small.^† This foreshadows the benefits of achieving the Indian standard of 40 μg/m³.

Because exposure to HAP constitutes over half of population-weighted annual average exposure to PM_2.5 in eight states^‡ and has been declining historically (20), we also pair reductions in AAP with 50% reductions in the proportion of households burning solid fuels relative to a 2019 baseline. The concavity of exposure–response functions implies that combining these policies yields greater benefits than the sum of the benefits from implementing each policy separately. Other things equal, the higher the proportion of households burning solid fuels, the greater the synergistic benefits.

To monetize the benefits of each policy, we use two approaches. The human capital approach estimates the gain in future output associated with deaths avoided by reducing PM_2.5 exposures. These benefits are estimated for people in each state as in ref. 7 using data for fiscal year 2019–2020 and are expressed in Purchasing Power Parity (PPP) dollars. We also value benefits using a VSL for India, which we measure, alternately, as 100 times Indian per capita income (21), and as 65 times Indian per capita income (22).

Results

Mortality Reductions Associated with Each Standard.

We calculate that 3.5 million deaths in India in 2019 were attributable to the six causes of death linked to PM_2.5 by the Global Burden of Disease Study (15). (SI Appendix, Table S2 presents these deaths for each of 28 states and 6 UTs in Column 2 and the state population counts in Column 1.) We describe the impact of achieving each of the six policies that we examine—lowering ambient PM_2.5 to 40, 25, and 15 μg/m³, respectively, and combining each of these policies with a 50% reduction in the percentage of households burning solid fuels for cooking—in terms of the percentage reduction in these deaths as well as in terms of lives saved.

The premature deaths averted by achieving an ambient standard depend on the magnitude of reduction in AAP, the size and demographics of the exposed population, and baseline exposure of the population to total PM_2.5. In Table 1 (Column 1), which lists the states with the highest mortality reductions associated with achieving the India standard, the first six states are among those that would experience the largest reductions in annual average AAP by achieving the 40 μg/m³ standard.^§ Annual average PM_2.5 would be reduced by 66.2 μg/m³ in the National Capital Territory (NCT), 45.5 μg/m³ in Uttar Pradesh, 37 μg/m³ in Haryana and Bihar, and more than 20 μg/m³ in Rajasthan and West Bengal. These states would also have the largest reductions in annual average PM_2.5 by achieving the WHO interim standard of 25 μg/m³. The top 12 states in Table 1 are also among the most populous states in India: Uttar Pradesh and Bihar together comprise over 25% of India’s population. The 12 states/UTs together make up 71% of the country’s population.

Table 1.

Mortality reductions in 2019 under four air quality programs for 12 states in India

State Name	India Standard (1)	India Standard & 50% of pH (2)	WHO 25 Interim Standard (3)	WHO 25 Interim Standard & 50% of pH (4)
Uttar Pradesh	41,000 (36,000 to 45,000)	77,000 (69,000 to 85,000)	59,000 (53,000 to 65,000)	100,000 (93,000 to 114,000)
Bihar	12,000 (11,000 to 14,000)	37,000 (33,000 to 42,000)	18,000 (16,000 to 20,000)	49,000 (44,000 to 54,000)
West Bengal	11,000 (9,500 to 12,000)	27,000 (24,000 to 30,000)	20,000 (18,000 to 22,000)	41,000 (37,000 to 45,000)
Rajasthan	7,700 (7,000 to 8,300)	21,000 (19,000 to 23,000)	14,000 (13,000 to 15,000)	31,000 (27,000 to 33,000)
NCT of Delhi	6,900 (6,100 to 7,600)	6,900 (6,200 to 7,700)	9,800 (8,700 to 11,000)	9,800 (8,800 to 11,000)
Haryana	6,600 (5,900 to 7,300)	9,300 (8,300 to 10,000)	10,000 (9,300 to 12,000)	14,000 (12,000 to 15,000)
Gujarat	6,400 (5,900 to 7,000)	13,000 (11,000 to 14,000)	16,000 (14,000 to 18,000)	24,000 (21,000 to 26,000)
Maharashtra	5,000 (4,600 to 5,600)	16,000 (14,000 to 18,000)	28,000 (24,000 to 30,000)	42,000 (36,000 to 45,000)
Punjab	4,100 (3,800 to 4,500)	6,700 (6,000 to 7,400)	10,000 (9,000 to 11,000)	13,000 (12,000 to 15,000)
Madhya Pradesh	3,900 (3,600 to 4,200)	20,000 (18,000 to 23,000)	11,000 (9,600 to 11,000)	31,000 (27,000 to 33,000)
Assam	2,700 (2,400 to 3,000)	7,800 (6,900 to 8,600)	5,100 (4,600 to 5,700)	12,000 (10,000 to 13,000)
Odisha	2,600 (2,300 to 3,000)	12,000 (11,000 to 14,000)	6,600 (5,900 to 7,400)	19,000 (17,000 to 20,000)

India Standard: 40 μg/m³; WHO 25: 25 μg/m³; p_H = fraction of households burning solid fuels. 95% Confidence Intervals in parentheses.

It is, however, also the case that in seven of the 12 states listed in Table 1, 60% or more of the population burn solid fuels for cooking (SI Appendix, Fig. S1).^¶ This implies that reductions in PM_2.5 due to the India standard achieve lower reductions in relative risk of death than would be the case if fewer households were exposed to HAP. To examine the magnitude of this effect, and to look at the benefits of reducing the percentage of households burning solid fuels, Table 1 (Column 2) also presents the reduction in deaths in 2019 from achieving the India standard together with a 50% reduction in households burning solid fuels for cooking. This more than doubles the estimated premature mortality avoided from the India standard alone. We found similar results when applying the benefits of achieving the WHO interim standard of 25 μg/m³ (Table 1, Columns 3 and 4).

To better understand the impact of combining the two policies, Fig. 2 shows the percentage reduction in baseline mortality from implementing the WHO 25 μg/m³ interim standard and a 50% reduction in the percentage of households burning solid fuels. The impact of implementing the two policies together exceeds the sum of implementing each policy separately due to the concavity of the concentration–response functions (SI Appendix, Fig. S2). The percentage reductions in mortality rates are greatest in North and Central India, where reductions in mortality rates corresponding to IHD, stroke, lung cancer, type II diabetes, COPD and LRI, range from 18.9% in Bihar to 14.3% in Chhattisgarh. The NCT achieves a 24.5% reduction in baseline mortality if it achieves the 25 μg/m³ standard.

Fig. 2.

Mortality reductions in 2019, by state, under WHO 25 standard, 50% reduction in percentage of households burning solid fuels ($p_H$), and the two policies combined.

Table 2 summarizes the mortality reductions associated with attaining the India standard and the two interim WHO standards, individually, and together with a 50% reduction in the percentage of households burning solid fuels. (Results for each policy and each state appear in SI Appendix, Tables S3–S5.) The total reduction in premature mortality achieved by each of the three policies doubles when going from the India standard (40 μg/m³)—118,000 deaths avoided—to the WHO 25 μg/m³ interim standard—264,000 deaths avoided. This is partly the result of states that have largely achieved the India standard, including Andhra Pradesh, Karnataka, Kerala, and Tamil Nadu, benefiting from the WHO 25 μg/m³ interim standard.^# It is also the case that the larger reduction in PM_2.5 under the WHO 25 μg/m³ interim standard moves states to the steeper portions of the concentration–response functions. When the WHO interim standard is lowered to 15 μg/m³, deaths avoided rise to 397,000.

Table 2.

Total mortality reductions in 2019 under six air quality programs in India

	India Standard (1)	WHO 25 Interim Standard (2)	WHO 15 Interim Standard (3)
Standard	118,000 (105,000 to 129,000)	264,000 (235,000 to 290,000)	397,000 (344,000 to 442,000)
Standard + 50% Reduction in HAP	310,000 (276,000 to 345,000)	501,000 (445,000 to 550,000)	679,000 (585,000 to 747,000)

India Standard: 40 μg/m³; WHO 25: 25 μg/m³; WHO 15: 15 μg/m³. 95% Confidence Intervals in parentheses.

The impact of lowering the percentage of households burning solid fuels also differs by standard. It raises deaths avoided by 192,000 in the case of the India standard, by 237,000 with the WHO 25 μg/m³ interim standard, and by 282,000 when combined with the WHO 15 μg/m³ interim standard. This reflects the fact that more stringent standards move population exposures to the steeper portions of concentration–response functions.

The Monetary Benefits of Achieving Each Standard.

We first value avoided mortality using the present discounted value of output gained when a person’s death is avoided, as in ref. 7. Deaths avoided are estimated by age group. The present value of output gained entails measuring average worker output by state, which is multiplied by the fraction of people in the person’s age group in the state who are working, to estimate the expected gain in market output for a person who is working. Thirty percent of average output is ascribed to the fraction of people who are not working to capture expected nonmarket output. Output and nonmarket output are assumed to grow at the real rate of growth in output per worker over the period 1990–2016 and are discounted to the present at a discount rate of 6%. All figures are in 2019 PPP dollars. (See SI Appendix for more details.)

Fig. 3 displays the present value of output gained from achieving the WHO interim standard of 25 μg/m³ and the WHO 25 μg/m³ interim standard combined with a 50% reduction in the percentage of households burning solid fuels. Output gained from the two programs is expressed as a fraction of gross state domestic product (GSDP). Achieving the WHO 25 μg/m³ interim standard combined with a 50% reduction in the percentage of households burning solid fuels is equivalent to a 0.5% increase in state GSDP in Bihar and increases of over 0.4% of GSDP in Delhi, Uttar Pradesh, and Haryana. Other states in India—Jharkhand, Rajasthan, Uttarakhand, West Bengal, Punjab, and Odisha, have increases in the present value of output more than 0.3% of GSDP. The value of productivity increases across India in 2019 PPP dollars associated with each of the policies we examine appear in Table 3 and by State in SI Appendix, Tables S6–S8. We spell out the human capital benefits associated with air pollution reductions because of their importance to policymakers (23).

Fig. 3.

Present value in 2019 of output gained from WHO 25 Interim Standard and WHO 25 Interim Standard combined with a 50% reduction in the percentage of households burning solid fuels ($p_H$) as a percent of GSDP.

Table 3.

Monetary benefits of mortality reductions in 2019 under six air quality programs in India (2019 PPP millions)

	India Standard (1)	WHO 25 InterimStandard (2)	WHO 15 InterimStandard (3)
	Standard
Present Value of Output Gained	7,000	17,000	26,000
Benefits Using VSL = India Per Capita GDP X 100	84,000	189,000	284,000
Benefits Using VSL = India Per Capita GDP X 65	55,000	123,000	185,000
	Standard + 50% Reduction in pH
Present Value of Output Gained	17,000	30,000	41,000
Benefits Using VSL = India Per Capita GDP X 100	221,000	358,000	485,000
Benefits Using VSL = India Per Capita GDP X 65	144,000	233,000	315,000

India Standard: 40 μg/m³; WHO 25: 25 μg/m³; WHO 15: 15 μg/m³; p_H = fraction of households burning solid fuels.

As an alternative, we estimate benefits using the VSL, which capture people’s willingness to pay for reductions in their risk of death.^|| This includes the value of output gained if a person does not die, but also the value of being alive. We transfer estimates of the VSL to India following procedures suggested by the Reference Case Guidelines for Benefit-Cost Analysis in Global Health and Development (21). Specifically, we estimate the VSL to be 100 times per capita income. Because the per capita income for India in 2019 was $7,151 PPP dollars (24), this implies a VSL of $715,000 2019 PPP dollars. As a sensitivity analysis, we use a VSL of 65 times per capita income, as assumed by the Lancet Commission on Pollution and Health (22) and the World Bank (25), which yields a VSL of $465,000 2019 PPP dollars. Results in Table 3 apply both VSLs to all persons in India.^**

Because per capita GSDP varies by a factor of 10 across the states of India, we also compute the VSL for each state in India. These results are reported in SI Appendix, Tables S6–S8. The aggregate benefits of mortality risk reductions are very similar whether computed using VSLs that vary by state or using the all-India VSLs discussed above.^††

The benefits of achieving the three ambient standards, individually, and in combination with a 50% reduction in the percentage of households burning solid fuels are shown in Table 3. Benefits, when valued using the India-wide VSL of $715,000 ($465,000), are substantial. They vary from 0.85% (0.55%) of gross domestic product (GDP) for achieving the India standard, to 1.9% (1.2%) of GDP for attaining the WHO 25 μg/m³ interim standard and 2.9% (1.9%) of GDP for meeting the WHO 15 μg/m³ interim standard. Benefits are much greater when each ambient standard is combined with a 50% reduction in the percentage of households burning solid fuels. Benefits increase to 2.2% (1.5%) of GDP for the India standard, 3.6% (2.3%) of GDP for the WHO 25 μg/m³ interim standard, and 4.9% (3.2%) of GDP for the WHO 15 μg/m³ interim standard. We emphasize that these benefits represent reductions in premature mortality for six causes of death: IHD, stroke, type II diabetes, lung cancer, COPD, and LRI. They do not capture the reduction in morbidity that would accompany each policy,^‡‡ nor impacts on agriculture (27) or other sectors due to reductions in ambient PM_2.5.

Discussion

We add to the literature on the benefits of reducing air pollution in India. Two recent papers, Adhikary et al. (10) and Kyrychenko (11), use regression methods to estimate the impact of PM_2.5 on mortality in India. Adhikary et al. (10) relate observed mortality rates in India over the period 2019–2021 to ambient PM_2.5 and HAP to see the difference in age-group mortality associated with different pollution parameters. Kyrychenko (11) estimates the reduction in infant mortality in India associated with the post-2010 economic slowdown.

Limaye et al. (12) and Manojkumar and Srimuruganandam (13) examine the impact of achieving air pollution standards in India using concentration–response functions—an approach similar to ours. Limaye et al. (12) examine the impact of the National Clean Air Program, National Air Quality Standards and WHO Air Quality Guidelines in Ahmedabad in 2030 relative to a Business-as-Usual baseline. Manojkumar and Srimuruganandam (13) calculate the benefits of achieving the India Standard of 40 μg/m³, a 30% reduction in AAP associated with the National Clean Air Program, and the WHO standard of 10 μg/m³ in 2019. Both papers use the US Environmental Protection Agency tool BenMAP-CE (28) to estimate health effects. BenMAP-CE treats AAP as the only source of exposure to PM_2.5. As noted above, when concentration–response functions are concave measuring total exposure to PM_2.5 is essential. Because the impact of HAP on total exposure to PM_2.5 is ignored, estimates of lives saved are much greater than if HAP were taken into account in measuring the benefits of reductions in AAP.

We view the incorporation of HAP into our analysis as a major contribution of the paper, both in measuring the benefits of ambient standards and in measuring the benefits of reducing HAP, in conjunction with achieving ambient standards. We acknowledge two limitations of our analysis. Our estimates of household exposure to burning solid fuels are at the state level, whereas ambient PM_2.5 is estimated at the district level. It is possible that households in urbanized districts are exposed to lower levels of HAP than households in more rural districts. District-level estimates of HAP exposure could therefore alter our estimates of the benefits of combining the two policies. We also acknowledge the limitations of measuring ambient PM_2.5 without distinguishing levels of nitrates and sulfates, which may be more damaging to human health than the total mass of PM_2.5 (29). Subject to these caveats, we summarize our main results.

Achieving the Indian annual average standard of 40 μg/m³ of PM_2.5 would confer large reductions in ambient PM_2.5 on states in the Indo-Gangetic Plain and Central India. As noted above, the NCT would experience a 66.2 μg/m³ reduction in annual average PM_2.5, followed by a 45.5 μg/m³ reduction in Uttar Pradesh, reductions of 37 μg/m³ in Haryana and Bihar, and over 20 μg/m³ in Rajasthan and West Bengal. These PM_2.5 reductions would lead to substantial mortality reductions in 2019—approximately 110,000 fewer deaths annually in the 12 states shown in Table 1—and 118,000 deaths annually when summed over the entire country. The mortality reductions associated with achieving the WHO 25 μg/m³ interim standard and the WHO 15 μg/m³ interim standard are even greater: 264,000 and 397,000 deaths avoided each year.

The benefits of these large reductions in ambient PM_2.5 in 2019 are, however, limited by the extent to which people in North and Central India are exposed to HAP. In seven of the states receiving the highest mortality reduction benefits from the Indian standard (Table 1),^§§ more than 60% of the population is exposed to high levels of indoor air pollution. If HAP exposure were zero in India, the benefits of the India standard in 2019 would increase from a reduction in 118,000 deaths to a reduction in 221,000.^¶¶ Attaining the WHO 25 μg/m³ interim standard would reduce mortality by 487,000 deaths annually and achieving the WHO 15 μg/m³ interim standard would reduce mortality by 716,000 deaths annually. This illustrates the importance of accounting for HAP exposures when estimating the benefits of ambient PM_2.5 reductions.

It is for this reason that we have combined policies to reduce HAP with the three air quality standards. As noted above, the concavity of concentration–response functions implies that the two policies in combination save more lives than either policy separately. The India standard plus a 50% reduction in HAP saves 310,000 lives annually, over 500,000 lives are saved when combining the WHO 25 μg/m³ interim standard plus the HAP reduction, and 679,000 lives are saved by the WHO 15 μg/m³ interim standard plus the HAP reduction.

Reductions in HAP would also improve ambient air quality, a benefit that we have not modeled. In the Indo-Gangetic Plain, emissions from household burning of solid fuels are a primary source of ambient air pollution. Chowdhury et al. (18) estimate that in 2015, over 30% of ambient PM_2.5 in the Indo-Gangetic Plain and Northeast India was attributable to HAP. They estimate eliminating HAP due to the burning of solid fuels and kerosene lighting in 2015 would bring 103 districts (187 million people) into compliance with the India standard. These estimates of the impact of HAP on AAP agree with a recent World Bank study (2). The study finds that, overall, approximately 25% of the ambient PM_2.5 in the Indo-Gangetic Plain in 2018 was attributed to residential and commercial burning (2). The study also estimates that 50% of the ambient PM_2.5 in Patna, Bihar in 2018 was the result of residential burning of solid fuels, as was one-third of ambient PM_2.5 in Chennai (2). The contribution of HAP to AAP has not been incorporated in our analysis and therefore represents additional benefits of the 50% reductions in the percentage of households burning solid fuels; however, quantifying these would require air quality modeling, which is beyond the scope of this study. In summary, the policies we consider, if implemented, would yield large health benefits and significant economic benefits.

Materials and Methods

Estimating Deaths Avoided by Reducing Ambient PM and HAP.

Using data on estimates of ambient PM_2.5 by district, the percentage of households burning solid fuels and the additional PM_2.5 generated by burning these fuels, we calculated baseline deaths for IHD, stroke, type II diabetes, lung cancer, COPD and LRI for people exposed to both ambient PM_2.5 and HAP and for people exposed only to ambient PM_2.5 following methods in refs. 7 and 30. A detailed description of data sources and processing to obtain district-level estimates is included in SI Appendix.

Formulas for Baseline Deaths.

Let M represent total deaths (for some cause of death) in a district. Then,

$$M={\lambda }_T\ast \mathrm{R}\mathrm{R}\left(P{M}_A+P{M}_H\right)\ast \mathrm{pop}\ast p_H+{\lambda }_T\ast \mathrm{R}\mathrm{R}\left(P{M}_A\right)\ast \mathrm{pop}\ast \left(1-p_H\right),$$ [1]

where λ_T denotes the death rate at the background level of PM_2.5, RR(z) is the relative risk of death at exposure level z, PM_A denotes the ambient PM_2.5 level to which people in the district are exposed and PM_H represents the additional PM level to which households burning solid fuels in the district are exposed. Relative risks are obtained from the MR-BRT (Meta-Regression with Bayesian priors, Regularization and Trimming), functions from the IHME (15). Pop is the population of the district, and p_H is the fraction of population in the district exposed to both HAP and AAP. Baseline deaths for each subgroup are given by

$$M_{AAP+HAP}^0={\lambda }_T\ast \mathrm{R}\mathrm{R}\left(P{M}_A+P{M}_H\right)\ast \text{pop}\ast p_H,$$ [2]

$$M_{\mathit{AAP}}^0={\lambda }_T\ast \text{RR}\left(P{M}_A\right)\ast \text{pop}\ast \left(1-p_H\right).$$ [3]

Eq. 1 can be solved as follows

$${\lambda }_T=\frac{M}{pop}*\frac{1}{RR\left(P{M}_A+P{M}_H\right)*p_H+RR\left(P{M}_A\right)*\left(1-p_H\right)},$$ [4]

and the result substituted into Eqs. (4) and (3) to solve for $M_{AAP+HAP}^0$ and $M_{\mathit{AAP}}^0$.

Estimating Deaths Avoided by Reducing PM_A and p_H.

The deaths avoidable by reducing PM_A from $P{M}_A^0$ to $P{M}_A^1$ due to air pollution control strategies are measured by the reduction in risk of death from moving from $P{M}_A^0$ to $P{M}_A^1$ multiplied by baseline deaths:

$$\mathrm{\Delta }M=M_{AAP+HAP}^0\left(\frac{RR\left(P{M}_A^1+P{M}_H\right)}{RR\left(P{M}_A^0+P{M}_H\right)}-1\right)+M_{\mathit{AAP}}^0\left(\frac{RR\left(P{M}_A^1\right)}{RR\left(P{M}_A^0\right)}-1\right).$$ [5]

This formula assumes that neither PM_H nor p_H are affected by the policy. When either PM_H or p_H is altered, the change in deaths due to the policy is given by

$$\begin{array}{c}\mathrm{\Delta }M=M_{AAP+HAP}^0\left(\frac{RR\left(P{M}_A^1+P{M}_H^1\right)}{RR\left(P{M}_A^0+P{M}_H^0\right)}\ast \frac{p_H^{\prime }}{p_H}-1\right)\hfill \\ +M_{\mathit{AAP}}^0\left(\frac{RR\left(P{M}_A^1\right)}{RR\left(P{M}_A^0\right)}\ast \frac{1-p_H^{\prime }}{1-p_H}-1\right).\hfill \end{array}$$ [6]

The CI for mortality reductions were estimated using the CI of the estimated MR-BRT functions from the IHME (8).

Valuing Deaths Avoided.

Deaths avoided were valued at the state level, using the present value of output gained if a person survived as a result of the policy (the human capital approach) and also using the VSL (the welfare approach).

The present discounted value of the gain in GDP attributable to reduced mortality associated with PM_2.5 in 2019 was estimated as described in ref. 7 and in SI Appendix.

The VSL represents the amount that people are willing and able to pay for reductions in their risk of death. It should therefore vary with per capita income, as well as with risk preferences. There is a large empirical literature that has examined how the VSL varies across countries. These studies suggest that the ratio of the VSL to per capita income, which represents the percentage of income that a person would spend to reduce their risk of dying by 0.0001 annually, declines as income declines (31). This ratio is approximately 170:1 in the United States and 100:1 in Organisation for Economic Co-operation and Development (OECD) countries. The Reference Case Guidelines for Benefit-Cost Analysis in Global Health and Development (21) suggest a default ratio of 100:1. The Lancet Commission on Pollution and Health (22) and the World Bank study on the Cost of Pollution (25) used ratios of the VSL to per capita income of 65 to 1 for low–middle-income countries, which we use in this study in addition to the 100 to 1 ratio.

Disclaimer.

The views expressed in this article are those of the author(s) and do not necessarily represent the views or the policies of the U.S. Department of State.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data are included in the article and/or SI Appendix.