Annex B. Methodology for calculating and valuing mortality and morbidity impacts

This annex provides a detailed overview of the methodology used to create projections of health impacts of air pollution, including both mortality and morbidity, as well as their valuation in monetary terms. The first part explains the methodology used to calculate air pollution-related deaths linked with high concentrations of fine particulate matter, while the second part focuses on morbidity impacts. Finally, the last part presents the methodology used for the monetary valuation of mortality and morbidity impacts.

Following the Global Burden of Disease (GBD, 2018[11]), the total amount of air pollution-related deaths attributable to outdoor air pollution corresponds to the sum of the deaths due to each disease for which there is an increased risk due to outdoor air pollution of fine particulate matter and ground-level ozone. For PM2.5, these illnesses are: ischemic heart disease (IHD), strokes, chronic obstructive pulmonary disease (COPD), lung cancer (LC), lower respiratory infection (LRI) and diabetes mellitus type 2 (DM). For ground-level ozone, the Global Burden of Disease (GBD, 2018[11]) indicates that exposure to increases the risk of deaths due to chronic obstructive pulmonary disease (COPD).

The mortality calculations for each disease is based on this formula:

Dtr= AFBDrt



Where deaths related to air pollution (D) are derived as the product between baseline deaths (BD) for each disease and the attributable fraction (AF), namely the fraction of baseline mortalities that can be associated with air pollution. The attributable fraction (AF) is derived as (1-1/RF), where RF is a disease-specific risk factor, which reflects how, for each disease, the risk of dying because of air pollution increases with higher concentrations of pollutants (PM2.5 and ground-level ozone).

The calculations of the health risks (RF) linked with exposure to PM2.5 used in this report rely on the GBD’s Integrated Exposure-Response (IER) functions (Cohen et al., 2018[12]; Burnett et al., 2014[13]). These functions are non-linear and become flat at higher exposures. The formula contains various parameters, one of which is the zero risk threshold, which is set at 2.5 μg/m3 concentrations of PM2.5.

For ground-level ozone, the RF is based on the following formula:

RF=elnRR10*Conc-Conc thr

Where Conc is the concentration of ground-level ozone measure in part per billions (ppb), Conc thr is the zero risk threshold of ozone concentration and RR is the Relative Risk associated to ground-level ozone.

Following GBD (2018[11]), this study uses the seasonal average of daily maximum eight hours mean as the metric for ground-level ozone, the concentration threshold is flat at 29.1 ppm and at a relative risk of 1.06.

Baseline mortalities are obtained from the GBD results tool (GBD, 2018[11]). To create the projection for 2020-40 we use GBD foresight, which relies on GBD 2016 data (Institute for Health Metrics and Evaluation (IHME), 2018[14]). To avoid discontinuities between GBD 2016 foresight and GBD 2017 data after 2017, the foresight data were scaled with their respective GBD 2016 value in 2017; this correction factor was applied on all years beyond 2017. Therefore, the final foresight data in the current set differ slightly from the GBD 2016 foresight data because they were tuned to match the 2017 data from GBD 2017. For 2050, base mortalities are assumed equal to 2040 levels.

The morbidity impacts of PM2.5 exposure that are quantified in this report are: the effect of chronic exposure on adult and childhood bronchitis, the effect of acute exposure on hospital admissions for respiratory and cardiovascular illness, restricted activity days, lost working days and asthma symptom days for children. The morbidity impacts for ground-level ozone are: the effect of acute exposure on hospital admissions for respiratory and cardiovascular illness and minor restricted activity days.

Quantifying morbidity effects requires detailed data, including the concentration response relationship, the size of population risk, and the prevalence of morbidity. However, this level of information is only available for a small number of countries. To obtain estimates at the global level, the morbidity impacts are extrapolated as a multiplier on mortality from air pollution exposure, based on the EU Clean Air Policy Package studies (Holland, 2014[7]; European Commission, 2013[15]). The advantage of assuming a linear relation between mortality and morbidity is that the calculation of morbidity automatically factors in the non-linearity in response functions that is accounted for in the mortality calculations. The drawback is that non-linearities are missed and that this approach cannot fully capture the connection between exposure to air pollution and illness.

The mortality-to-morbidity ratios are taken from the European Commission’s Clean Air Policy Package studies (Holland, 2014[7]; European Commission, 2013[15]). The study by Holland (2014[7]) supplies region-specific morbidity-to-mortality ratios for the 28 European countries in which the package was implemented. To calculate morbidity impacts at the global level, for countries not covered by the Clean Air Policy Package studies, the average of the ratios of the European countries is used. While this extrapolation is not ideal, no data are available at the global level. This assumption is limiting, as it assumes that mortality-to-morbidity ratios throughout the world are similar to those of European countries. Furthermore, it implicitly assumes that healthcare provision is similar in all countries. For hospital admissions, it implies that European admission rates are typical of all other countries, when there is substantial variation around the world with respect to access to healthcare systems. This problem is particularly serious for developing countries, where access to healthcare is much lower. A similar issue arises with respect to lost working days. The European results are based on European rates of absenteeism, reflecting specific social welfare and employment conditions.

There are other limitations of the methodology used to calculate morbidity impacts in this report. Ideally, changes in behaviour (e.g. in diet, smoking habits, etc.), social changes (e.g. healthcare and employment) and medical changes (e.g. changes in healthcare systems and in treatment of diseases) over time and in different world regions should be explicitly factored into the analysis, but this is not possible owing to lack of data at the global level.

The valuation of the welfare costs of the health impacts of outdoor air pollution includes both mortality and morbidity. The total welfare costs are calculated by multiplying each impact considered (e.g. number of hospital admissions, cases of illness, and mortality) by estimates of the unit welfare cost of each impact (e.g. the welfare cost of a hospital admission, a case of illness, and a mortality).

The welfare costs of air pollution-related mortality are obtained from a meta-analysis of a large number of studies of individuals’ willingness to pay (WTP) for a marginal reduction in their risk of mortality over time. Aggregating the individual results of the various WTPs in the meta-analysis allows us to quantify the so-called Value of a Statistical Life (VSL), a long-established metric that attributes a monetary value to life and, as a consequence, can be used to estimate the welfare costs of mortality (OECD, 2014[16]; OECD, 2012[17]). As a result of this meta-analysis, the base VSL in OECD countries is USD 3 million (2005 PPP) per life lost in 2005.

As this report has global coverage, it was necessary to calculate VSL values for countries outside the OECD. This report relies on the OECD database “Mortality and welfare cost from exposure to environmental risks” (OECD, 2020[18]) for the base year value for each region of the study, since it provides country-specific VSL for OECD countries and emerging economies. Welfare costs in this database are calculated using a methodology adapted from Roy and Braathen (2017[19]).

Furthermore, since the report also considers economic projections, the VSL values need to be adapted over time. A previous OECD study (OECD, 2014[16]) provides a benefit transfer methodology to determine country-specific VSL values from an OECD reference value, based on country income differentials. The benefit transfer methodology is used to adapt VSL to individual countries, but also to estimate its growth over time, as income rises. As argued in OECD (2006[20]), income should be used as the reference variable to adapt WTP over time, so as to avoid situations in which the WTP to save a statistical life rises faster over time than the rate of inflation. Existing studies – such as Costa and Kahn (2004[21]), who calculate the VSL changes in the United States for the period 1940-80 – find that VSL rises over time as income rises. The country-specific income levels over time that are necessary for the calculations are obtained from the International Monetary Fund until 2017 (IMF, 2019[22]) and from the economic projections of the OECD’s ENV-Growth model, which are also used for the calibration of the ENV-Linkages model.

The formula used to calculate the VSL is:


Where Y is the average income (GDP per capita) of country r in year t expressed in 2017 USD PPP; and β is the income elasticity of VSL. The income elasticity measures the percentage increase in VSL for a percentage increase in income.

The income elasticity used to calculate the country-specific VSL values is a key parameter; choosing different values can alter the results for welfare costs. The income elasticity variable assumes that as incomes rise, the WTP for a marginal reduction in the risk of death also rises, but not quite in proportion to the rise in incomes. The meta-analysis (OECD, 2012[17]) finds that the income elasticity is in the range of 0.7-0.9 for OECD countries, with significantly higher income elasticities for countries in the bottom 40th percentile of income. However the range proposed in OECD (OECD, 2012[17]) was considered to be too low for low-income countries as using such values would imply unrealistically high WTP values for these countries. Existing work on VSL (Hammitt and Robinson, 2011[23]; Roy and Braathen, 2017[19]) supports the assumption that the impact of income elasticity on the WTP does not necessarily hold true for emerging economies. Thus, following previous OECD work (OECD, 2016[5]) this report differentiates elasticity values by income group and uses a slightly higher elasticity for low-income countries. Specifically the income elasticities used are: 0.8 for high-income countries, 0.9 for middle-income countries and 1 for low-income countries (where country groups are distinguished using the World Bank income thresholds).

Given the difficulty in establishing the WTP to reduce the risks of mortality and the high dependency of the results on the key parameter value of income elasticity, the welfare costs results need to be interpreted in the context of the uncertainty surrounding the VSL values. An uncertainty analysis on the parameter values is provided in (OECD, 2016[5]).

While the VLS values are surrounded by uncertainty, a change in methodology would not affect the overall policy results of the analysis, which show high welfare costs associated with the deaths caused by outdoor air pollution.

The analysis of the health impacts of air pollution in this report distinguishes between two types of costs related to illness, as outlined in OECD (2016[5]):

  1. 1. The healthcare costs that are used to calculate healthcare expenditures as input to calculate the macroeconomic consequences of air pollution (see Annex A). Healthcare costs reflect the expenditures linked with each case of illness (e.g. the costs of hospital admissions, of going to the doctors or of buying medicines).

  2. 2. Welfare costs of morbidity, which reflect the pain and suffering of each case of illness. In other words, welfare costs of morbidity reflect the disutility of illness.

The welfare costs of morbidity used here rely on previous work by the European Commission (Holland, 2014[7]), which provides unit values for the welfare costs of the morbidity impacts (Table B.1). Morbidity welfare costs are adjusted to specific countries based on income, using the benefit transfer methodology used for mortality. Although there is a bias in transferring estimates of the disutility of morbidity from existing studies, mostly developed in Europe, to the global context, the benefit transfer method is the only available technique in this context, since valuation studies on the welfare impacts of air pollution-related illnesses only exist for a few areas in the world.


[13] Burnett, R. et al. (2014), “An Integrated Risk Function for Estimating the Global Burden of Disease Attributable to Ambient Fine Particulate Matter Exposure”, Environmental Health Perspectives, Vol. 122/4, pp. 397-403,

[1] Chateau, J., R. Dellink and E. Lanzi (2014), “An Overview of the OECD ENV-Linkages Model: Version 3”, OECD Environment Working Papers, No. 65, OECD Publishing, Paris,

[12] Cohen, A. et al. (2018), “Global estimates of mortality associated with long-term exposure to outdoor fine particulate matter”, Proceedings of the National Academy of Sciences, Vol. 115/38, pp. 9592-9597,

[21] Costa, D. and M. Kahn (2004), “Changes in the Value of Life, 1940–1980”, Journal of Risk and Uncertainty, Vol. 29/2, pp. 159-180,

[15] European Commission (2013), The Clean Air Policy Package, pp. No. SWD(2013)531, European Commission, Brussels,

[11] GBD (2018), “Global Burden of Disease Study 2017: All cause Mortality and Life Expectancy 1950-2017, Global Burden of Disease Collaborative Network.”, Seattle, United States: Institute for Health Metrics and Evaluation (IHME).

[3] Guillemette, Y. and D. Turner (2018), “The Long View: Scenarios for the World Economy to 2060”, OECD Economic Policy Papers, No. 22, OECD Publishing, Paris,

[23] Hammitt, J. and L. Robinson (2011), “The Income Elasticity of the Value per Statistical Life: Transferring Estimates between High and Low Income Populations”, Journal of Benefit-Cost Analysis, Vol. 2/1, pp. 1-29,

[7] Holland, M. (2014), Cost-benefit Analysis of Final Policy Scenarios for the EU Clean Air Package, Corresponding to IIASA TSAP Report No. 11, International Institute for Applied Systems Analysis (IIASA), Laxenburg, (accessed on 9 March 2021).

[4] IEA (2018), World Energy Outlook 2018, International Energy Agency, Paris,

[22] IMF (2019), World Economic Outlook Database, International Monetary Fund,

[14] Institute for Health Metrics and Evaluation (IHME) (2018), “Global Life Expectancy, All-Cause Mortality, and Cause-Specific Mortality Forecasts 2016-2040.”, Seattle, United States: Institute for Health Metrics and Evaluation (IHME)., (accessed on 2 December 19).

[24] Klimont, Z. et al. (forthcoming), “Global scenarios of anthropogenic emissions of air pollutants: ECLIPSE”.

[18] OECD (2020), Air quality and health: Mortality and welfare cost from exposure to air pollution (database), Statistics, OECD Environment, (accessed on 3 November 2020).

[2] OECD (2019), Global Material Resources Outlook to 2060: Economic Drivers and Environmental Consequences, OECD Publishing, Paris,

[5] OECD (2016), The Economic Consequences of Outdoor Air Pollution, OECD Publishing, Paris,

[10] OECD (2015), The Economic Consequences of Climate Change, OECD Publishing, Paris,

[16] OECD (2014), The Cost of Air Pollution: Health Impacts of Road Transport, OECD Publishing, Paris,

[17] OECD (2012), Mortality Risk Valuation in Environment, Health and Transport Policies, OECD Publishing, Paris,

[20] OECD (2006), Cost-Benefit Analysis and the Environment: Recent Developments, OECD Publishing, Paris,

[19] Roy, R. and N. Braathen (2017), “The Rising Cost of Ambient Air Pollution Thus Far in the 21st Century: Results from the BRIICS and the OECD Countries”, OECD Environment Working Papers, No. 124, OECD Publishing, Paris, (accessed on 26 April 2019).

[8] Vrontisi, Z. et al. (2016), “Economic impacts of EU clean air policies assessed in a CGE framework”, Environmental Science & Policy, Vol. 55, pp. 54-64,

[6] World Bank (2015), World Development Indicators, Washington, DC.,

[9] World Bank (2014), Doing Business 2015: Going Beyond Efficiency, The World Bank,

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2021

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at