Economic evaluation of health losses from air pollution in Beijing, China


Aggravated air pollution in Beijing, China has caused serious health concern. This paper comprehensively evaluates the health losses from illness and premature death caused by air pollution in monetary terms. We use the concentration of PM10 as an indicator of the pollution since it constitutes the primary pollutant in Beijing. By our estimation, air pollution in Beijing caused a health loss equivalent to Ұ583.02 million or 0.03 % of its GDP. Most of the losses took the form of depreciation in human capital that resulted from premature death. The losses from premature deaths were most salient for people of either old or young ages, with the former group suffering from the highest mortality rates and the latter group the highest per capital losses of human capitals from premature death. Policies that target on PM10 emission reduction, urban vegetation expansion, and protection of vulnerable groups are all proposed as possible solutions to air pollution risks in Beijing.


As Chinese people enjoyed an economic miracle during the past 30 years, they also unfortunately witnessed some by-products of the miracle, such as energy resource deprivation, serious air pollution, other kinds of environmental degradation, as well as the resultant increase in health risks and mortality rates. In 2001, about one third of China’s air quality monitored cities failed to attain the lowest level of National Ambient Air Quality Standard (NAAQS).Footnote 1 Annual mean concentrations of major pollutants, including sulfur dioxide (SO2), nitrogen dioxide (NO2), and particulate matter smaller than 10 μm (PM10)Footnote 2 in those cities, all exceeded the poorest standards in the NAAQS. Another one third of Chinese cities controlled their air quality within the range defined by grade III, but exceeded grade II. Only one third of cities kept their PM10 and NO2 pollution level below the nationally recognized standard, which is 40 μg/m3. Although China has made impressive progress in air quality management in the past decade (2001–2009),Footnote 3 its air condition would still be considered hazardous compared with that in the developed world (Fig. 1), and its pollution concentrations are far above the levels that are considered safe by the US, the European countries, and the World Health Organization (WHO) committee.

Fig. 1

Comparison of annual average PM10 concentration trend of China, USA, France, and Germany (1990–2011). Data source: The World Bank World Development Indicators (WDI) Database

Air pollution definitely leads to health risks. It is estimated by the World Health Organization (WHO) that about 7 million people around the world die of air pollution related diseases each year, which accounts for 12.5 % of the total deaths (Kuehn 2014). The report from WHO (2013) addressed that air pollution is most closely associated with two families of symptoms, i.e., respiratory and cardiovascular illness, which could lead to premature death. Given the well-known aggravated air pollution in China, its health damage would be more serious compared to other parts of the world. According to a rough estimation by Reuters’ staff, air pollution caused 350,000 to 500,000 premature deaths in China each year (Reuters Staff 2014).

The health impacts of air pollution have drawn attention from pathologists and have been widely investigated since the 1950s. These studies confirm the correlation between severe air pollution and increased risk of respiratory symptoms and death,Footnote 4 and also indicate that such impacts may vary among human groups from different regions, of different races, and at different ages. For example, a long-run investigation (May 1996–December 2010) showed that for an interquartile range increase in NO2 concentration, the morbidity rate of respiratory diseases increased by 2.43 % in São Paulo, Brazil (Bravo et al. 2016). As to SO2, the same increase in pollutant concentration would raise the morbidity rate of respiratory diseases by 1.6 % in Switzerland, and the hospitalization rate by 0.8–2.2 % in Brisbane, Australia, with adults relatively less affected and juveniles more affected (Holland et al. 2005; Samakovlis et al. 2005). Such impact was more precisely measured in the European Commission’s ExternE Project, which applied a dose-response model to estimate the health impact of major air pollutants, with the potential influence from weather condition, sunshine level, and seasonal variation all controlled. The Commission reported that 0.06 % of people exposed to PM10 pollution would die from related diseases if its concentration increased by 1 μg/m3.

The health impacts of these air pollutants, including NO2, SO2, and PM10, have also been evaluated in some parts of China. For example, NO2 pollution in Shanghai, China has raised the morbidity rate of respiratory diseases by 1.50 %, and for every additional 10 μg/m3 increase in PM10 concentration, the city’s mortality rate increased by 0.84 % (Chen et al. 2002; Li et al. 2004). As an effort to fight against air pollution, the city of Chengde in Hebei province adopted an integrated heating system in 1997, which successfully reduced local SO2 concentration from 3.86 to 0.264 μg/m3. Along with the improvement, mortality rate of respiratory diseases decreased from 8.1 to 2.3 ‰ (Zhang et al. 2000). A 10 μg/m3 increase in PM10 concentration in Beijing, China was associated with an increase of 0.33 % in daily ischemic heart disease (IHD) mortality (Xu et al. 2014). It is also found that the impact of PM10 is more salient in summer than in other seasons. Males and older people are more sensitive to PM10 pollution (Xu et al. 2014).

Yang and Pan (2008) evaluated the impacts of individual pollutants in Beijing and concluded that for every additional 10 μg/m3 increase in the concentration of SO2, NO2, and PM10, the death risks of cardiovascular and cerebrovascular diseases would increase by 0.40, 1.30, and 0.40 %, respectively. Zhang et al. (1999) comprehensively evaluated the impacts of all the three pollutants on human health. They found that the morbidity rate of respiratory diseases among juveniles was four times as high in heavily polluted zones in Tianjin compared to that in air-clean zones.

Health impact of air pollution has been separately related to individual pollutants in some existing studies; these pollutants, i.e., SO2, NO2, and PM10, however, are correlated, and their impacts are in fact overlapped (Künzli et al. 2000). A summation of the health impacts of individual pollutants would overestimate losses from air pollution. Hence, we evaluate air pollution in this study with PM10 since both SO2 and NO2 are important precursors of PM10 and their impacts are part of that of PM10 (Pope and Dockery 2006). Moreover, in the study year of 2012, PM10 constituted the primary pollutantFootnote 5 on 337 days of the 366 days in that year.

After the health impacts have been sufficiently assessed in pathological studies, more attention now is directed to economic valuation of these impacts. It is estimated that, each year, 1.2–2.0 % of the newly generated wealth around the world is lost due to health deterioration from air pollution (WHO 2006). The costs consist of two parts: (1) increase in protection cost and medical expenditure spent on treating air-pollution-related diseases, and (2) human capital depreciation, i.e., the decline and incapacity of labor force caused by physical condition deterioration, fatigue, diseases, and death. As a pioneer scholar in this field, Ostro (1983) delicately analyzed the US Health Interview Survey and found that a 10 % decrease in ambient levels of total suspended particular (TSP) was related to a 4.4 % decrease in average working loss days and 3.1 % in restricted activity days. The statistically significant and quantitatively important effect was later confirmed by Hausman et al. (1984). In China, the effect is also monetized. It is estimated that health loss caused by PM10 pollution accounts for 0.53 % of the gross domestic production (GDP) in Beijing (Cao and Han 2014; Zhang et al. 2007a, b) and 1.03 % in Shanghai (Kan and Chen 2004).Footnote 6

While those aforementioned studies provided a useful perspective in detecting air pollution’s health impact in an era characterized by pollution, the studies concerning China still have some shortages. In the works with focus on a Chinese city, many studies used dose-response coefficients cited from other studies (the mostly cited work includes Xu et al. 2014 and Aunan and Pan 2004) in assessing air pollution’s health impact and the associated economics losses; meanwhile, the most fundamental work of estimating the dose-response coefficients with field data is quite limited. Even among empirical studies that did make first-hand estimation of the dose-response coefficients, research attention was mostly restricted to the association between air pollution and mortality, leaving the investigation of air-pollution-induced morbidity quite sparse. This is because data of mortality was relatively easy to access, but data necessary for evaluating disease-specific morbidity are hardly available in China. According to a 2004 survey (Aunan and Pan 2004), only three studies, one for Beijing and two for Hong Kong, used outpatient visit data to estimate simple association between air pollution and hospital admission among Chinese populations, and they became the primary citing sources in following studies. However, those studies are quite old and may not be able to properly reflect the impact of current air pollution. For example, the study focusing on Beijing was published far back to 1995, over 20 years ago (Xu et al. 2014).

Our primary contribution lies in the most updated empirical estimation about how air pollution is associated with both death and illness among Chinese people, with supporting morbidity data collected from three local hospitals in Beijing in the year 2012. Possible human symptoms associated with air pollution consist of circulation system diseases, respiratory system diseases, and immune function deterioration (An et al. 2006; Dong et al. 2007). Since immune disadvantage may cause circulation diseases, and the impact of air pollution on them may overlap, we do not separately consider the health impact channeled through immune function deterioration in this study.

The second contribution is that we separate the impacts among subgroups defined by among populations with varied age, residence location, and etiology, with the potential influence from dietary habit and weather condition all controlled.

The evaluation of economic losses caused by diseases and premature death due to air pollution in Beijing constitutes our third contribution. It is estimated in the study that the city of Beijing lost Ұ583.02 million in 2102 due to air pollution. Of the losses, 18.3 % was spent to treat cardiovascular diseases that resulted from air pollution and 12.6 % to treat respiratory diseases. The other 69.1 % of the loss resulted from depreciation of human capital due to premature death. As confirmed by most previous studies, people of old age are more vulnerable to air pollution than their younger counterparts since they are most likely to be infected with cardiovascular diseases and respiratory diseases. The rules are especially true for groups older than 35. People under this age are less likely to be affected by air pollution, except for infants less than 5 years old, who are another major group of victims of Beijing’s air pollution.

The remainder of this paper is organized as follows: “Air pollution in Beijing” section introduces current air pollution conditions in Beijing. “Methodology and data” section builds a dose-response model and an economic loss assessment model to evaluate the pathological and economic impact of air pollution in Beijing. “Results and discussion” section discusses the results and “Concluding remarks” section concludes by offering policy suggestions.

Air pollution in Beijing

As indicated in the foregoing discussion, when discussing air pollution in Beijing, SO2, NO2, and PM10 are the three most referenced pollutants.Footnote 7 Figure 2 shows air pollution level in Beijing, as well as the respiratory disease mortality rates during 1994 to 2012. The whole period can be roughly divided into three stages. The first one, from 1994 to 1998, observed an increasing trend of air pollutant concentration along with an ascending mortality rate, which peaked in 1998. The second stage started from 1998 and lasted until 2008. It was characterized by intensive pollution control policies and air condition improvement. During this period, the concentrations of the three major pollutants all declined by 50 %. The remarkable improvement in Beijing’s air quality over this period may be attributed to the 160 air pollution control measures adopted by the Beijing municipal government, which aimed for greeting the 2008 summer Olympic Games. To impress the world as an excellent host of the 29th summer Olympic Games, the municipal government of Beijing also cleansed 16,000 small coal-fired boilers, removed more than 140 polluters, shut down all intensively polluting factories, and even cut off half of the traffic flow with an odd-even number-plate restriction measure in the year 2008. These practices inevitably led to human health benefits. Figure 2 also shows that the mortality rate of respiratory diseases plummeted in 2004, which reached 0.43 ‰. This is because SARS broke out in 2003 and many vulnerable people died of it. In order to control this epidemic disease, industrial production, business activities, and transportation volume were all vastly reduced. As a result of both air pollution mitigation and burst death of respiratory disease vulnerable people in 2003, the mortality rate of respiratory diseases plummeted in 2004. Air pollution concentration after 2008 was kept quite stable, and so was the mortality rate. This is the third stage.

Fig. 2

Air pollutant concentration and respiratory disease mortality rate in Beijing (1994–2012). Data source: 19942012 Beijing environment bulletin, 19952012 Beijing statistical yearbook


In addition to the individual air pollutant concentration standard (Table 1), China also employed the comprehensive air pollution index (API) to evaluate its air quality. The index gauges pollution condition by comprehensively including the three major pollutants and measures air pollution with the concentration of the worst pollutant (which is also called primary pollutant) on each day. In this way, API can better reflect air condition improvement than other indicators, such as average pollution level or the best level. The primary pollutant may vary day to day, but the score is kept consistently.

Table 1 The comparison of air quality standard between China and WHO(2005)

API in Beijing had been kept decreasing from 1994 to 2012, with three sharp declines in 2000, 2003, and 2008 (Fig. 3). This trend was at least correlated with, if not driven by, Beijing’s air pollution control policies (Table 2). Also, financial investment contributed to this improvement. From 1994 to 2004, the Beijing municipal government had been consistently raising its investment in environmental protection, which peaked twice, respectively, in 1999 and 2004. They may explain pollution mitigation in the following years.

Fig. 3

Annual average API and environmental investment in Beijing (1994–2012). Data source: Beijing Statistical Yearbook, Beijing environmental protection bureau

Table 2 Air pollution regulations implemented in Beijing, 2000–2008

While air pollution in Beijing, as indicated by both the pollutant concentration (Fig. 2) and API index (Fig. 3), had been kept relatively improved during the past decade, it should be noted that the city’s annual mean concentration of NO2 and PM10 were still 52 and 109 μg/m3, well above the WHO standard of 40 and 20 μg/m3, respectively, for the two pollutants and also above the grade I standard in NAAQS of 40 μg/m3 for both pollutants. The severe air pollution in China, especially in Beijing, causes human health damage.

Methodology and data

Economic losses that resulted from air-pollution-related illness consist of two parts: (1) medical costs, i.e., excessive protection cost and medical expenditure resulted from diseases caused by air pollution, and (2) loss of premature death, i.e., the decline and incapacity of labor force caused by premature death. The two parts are estimated separately and then added up together for a general estimation.

Medical costs

It is assumed that medical costs spent on each patient with air-pollution-related diseases, mainly cardiovascular diseases and diseases, are constant across all age groups, which are denoted by C car and C res respectively. Thus, the total air-pollution-related medical and protection costs imposed on a whole society (C med) equals individual costs multiplied by the numbers of patients who are affected by cardiovascular diseases and respiratory diseases, i.e., Z car and Z res

$$ {C}_{\mathrm{med}}={C}_{\mathrm{car}}\times {Z}_{\mathrm{car}}+{C}_{\mathrm{res}}\times {Z}_{\mathrm{res}} $$

The data we used to evaluate C car and C res are collected from Beijing Short Compile for Health Statistics in 2012.

Both Z car and Z res are assumed to relate with air pollutant concentration, and the relationships are estimated with a dose-response model developed by the U.S. Environmental Protection Agency (EPA):

$$ \begin{array}{l} \ln \left[E\left({Z}_{ij}\right)\right]=\hfill \\ {}\alpha +{\beta}_js\left({X}_i\right)+DOW+H+s\left( time,df=7\right)+s\left( temperature,df=6\right)+\hfill \\ {}s\left( humidity,df=6\right)+s\left( price,df=6\right)\hfill \end{array} $$

where Z ij is the amount of people going to see a doctor (PSD) on the ith day in the jth sample hospital due to either cardiovascular or respiratory diseases. β j represents the correlation between air pollutant concentration and the PSD in the district where the jth sample hospital locates. X ij is air pollutant concentration on the ith day at the location of the jth sample hospital. DOW is a dummy variable indicating whether the ith day is a working day or not. DOW = 1 represents working days and DOW = 0 represents non-working days. This variable is included because the pattern of hospital visits may vary under the two conditions. For a similar reason, we also include in the model the dummy variable of H, with H = 1 indicating heating period and H = 0 indicating non-heating period. s is the denotation of cubic spline function. The model explicitly considers the spline functions of air pollution, time, temperature, humidity, and price levels of meat and eggsFootnote 8 since these factors are all assumed to be non-linearly related to the incident of cardiovascular and respiratory diseases (Dusseldorp et al. 1995; Kwiterovich 1997; Chen et al. 2012).

The EPA dose-response model is open in terms of selection of pollution index. In this study, we use the concentration of PM10 as the indicator of air pollution concentration (X ij ) on each dayFootnote 9. We refer to the grade I standard in NAAQS when setting up the thresholdsFootnote 10 in this study: 40 μg/m3 for PM10.

This model is estimated with disease-specific PSD data (Jan 1st–Dec 31st 2012) collected from three localFootnote 11 hospitals respectively located in the districts of Shijingshan, Fengtai, and Daxing of Beijing. The three districts locate in the central and southern parts of Beijing, as shown in Fig. 4. The reason we use the local hospital visiting data is that such data can help eliminate potential sample contamination arising from non-local visits on our estimation of Beijing residents’ health conditions. Since air pollution in the central and southern parts of Beijing is a little more severe than that in the north, this would make simulation from our study slightly exaggerating the health impacts of air pollution in Beijing (Fig. 5).

Fig. 4

The map of Beijing

Fig. 5

The amount of premature deaths caused by air pollution by diseases

Air pollution data in the three districts were collected from the air quality monitoring stations in each area, one in Shijingshan, two in Fengtai, and two in Daxing. For the districts with more than one monitoring station, air pollutant concentration is equaled with the average of readings from each station. For the city of Beijing as a whole, air pollutant concentration is average among all pollution monitoring stations. All air quality monitoring stations involved in this study are run by the government, thus all the readings of pollutant concentration used are official data. Meteorology data, including temperature, air humidity, and rainfall in Beijing, were collected from Beijing Meteorological Bureau. The data of PSD associated with each disease in Beijing is collected from 2012 Beijing Health Statistics Yearbook, and the food price data is collected from Beijing Xinfadi Agricultural E-Trading Center.

Based on Eq. (2), for each sample hospital, we estimate its β j and derive the dose-response coefficient b j following Eq. (3), which measures the percentage increase in PSD associated with 10 μg/m3 increase in air pollutant concentration.

$$ {b}_j=\frac{ \ln {\beta}_j}{\varDelta D} $$

ΔD is the gap between annual average air pollutant concentration and its threshold. For a further step, b j is averaged among the three sample hospitals to eliminate potential bias that resulted from hospital scale and instability of daily PSD. With the correlation between air pollution and disease-specific PSD, we further estimate the number of people whose health is affected by air pollution in the target city, Beijing, with Eq. (4):

$$ {Z}_{\mathrm{res}/\mathrm{car}}=\varDelta Z={Z}_p-{Z}_c={Z}_p\left[1-\frac{1}{{\left(1+\overline{b}\right)}^{\varDelta D}}\right] $$

where Z p is the overall PSD amount for the cardiovascular diseases and respiratory diseases, Z c is the PSD amount without pollution,Footnote 12 ΔZ is the excessive PSD attributed to air pollution, and \( \overline{b} \) is the weighted average of b j Footnote 13.

Loss in premature death

Similar to the estimation of pollution-illness relationship, the daily number of disease-specific premature death is estimated with the EPA’s dose-response model:

$$ \begin{array}{l} \ln \left[\mathrm{E}\left({P}_{ik}\right)\right]\hfill \\ {}\alpha +{\beta}_ks\left({X}_i\right)+DOW+H+s\left( time,df=7\right)+s\left( temperature,df=6\right)+\hfill \\ {}s\left( humidity,df=6\right)+s\left( price,df=6\right)\hfill \end{array} $$

where P ik represents the number of premature death on the ith day caused by the kth kind of diseases (k = 1 means respiratory diseases and k = 2 means cardiovascular diseases), and β k represents the correlation between the number of premature death caused by disease k.

Disease-specific death toll for each day is necessary to estimate β k in Eq. (5). However, such data are neither available from the city of Beijing as a whole nor from the sample hospitals. As an alternative, we use data collected from Hong Kong in 2012 to estimate age-specific β k . Those data include air pollutant concentration data collected from Hong Kong Environmental Protection Bureau, humidity data from Hong Kong observatory, food price data from Hong Kong Census and Statistics Department, and daily death toll from the Center for Health Protection. Thus, the premature death dose-response coefficient we used in this study is derived from the Hong Kong population. Although it may not perfectly resemble the dose-response relationship among Beijing residents, we think they are currently the best available substitution for two reasons. First, it has been shown in various pathological studies that the dose-response relationships are quite similar among groups with similar races (HEI 2003; Dong and Zhi 2010; Daniel and Petrolia 2011). Second, weights of most chemical components of PM10 in the air (with similar degree of pollution) collected from Beijing and Hong Kong are not statistically different from each other (see Table 3). Thus, substituting Beijing’s air pollution dose-response coefficients with those of Hong Kong would at least provide us a reasonable estimation of the impact of air pollution on premature death in Beijing.

Table 3 A comparison of PM10’s chemical composition in Beijing and Hong Kong

Once we obtain the estimation for β k , we can calculate the dose-response coefficient b k (the percentage increase in premature death caused by 10 μg/m3 increase in air pollutant) with Eq. (6).

$$ {b}_k=\frac{ \ln {\beta}_k}{\varDelta D} $$

where ΔD is the gap between annual average concentration of PM10 and its threshold.

The number of disease-specific daily premature death resulted from air pollution in Beijing is estimated with the dose-response model, and a whole year’s premature death is estimated by totally summing the number of death on each day and in each district, as shown in Eq. (7):

$$ {P}_k={\sum}_{i-1}^{365}{\sum}_{n-1}^{17}{P}_{kin}={\sum}_{i=1}^{365}{\sum}_{n-1}^{17}{P}_n\times \varDelta {D}_{in}\times {b}_k $$

where P k is number of premature death caused by air pollution and channeled through disease k (cardiovascular disease and respiratory disease, respectively) in Beijing in the study year. We calculate it by summing the number of disease-specific death on each day (i) and in each district (n) that are attributable to air pollution, which is P kin . P kin is estimated by multiplying P n , the total number of premature death in district n, with ΔD in (the district-specific gap between daily mean of air pollution concentration and its threshold) and the dose-response coefficient b k . To estimate the economic loss caused by premature death, we need to calculate the number of premature deaths for various age strata to explore the economic loss with Eq. (8):

$$ {P}_{km}={\lambda}_{km}{P}_k $$

where P km is the number of premature deaths associated with illness k in the nth age strata; λ km is the proportion of the number of the premature death associated with illness k for the mth age stage to the total premature death associated with this disease. The total number of premature deaths caused by air pollution is partitioned into 16 age strata (with five ages as one group) in this study, as shown in Table 4.

Table 4 Mortality rates of cardiovascular and respiratory diseases across age cohorts, 2012 (unit = 1/100,000)

The loss varies along with the age at which a patient dies according to the human capital method. It would be reasonable to assume that the younger the patient dies, the greater the loss of human capital in premature death. Specifically, the per capita loss in premature death cost is calculated with a revised human capital method, which takes the loss mainly as the decrease in human capital gain resulted from life expectancy reduction.

$$ {c}_t^{\mathrm{death}}={\sum}_{i=1}^{80-t}\frac{gdp{\left(1+g\right)}^i}{{\left(1+r\right)}^i} $$

Equation (9) describes how to calculate the premature death loss for an individual at age t, which is denoted as c death t . In this calculation, we take 80 years, the average lifespan for Beijing residents in 2012, as life expectancy.Footnote 14 Any death that occurs before that would be considered as premature death, and the resultant losses in potential human capital gains from all following years till 80 are first converted to their present values in 2012 and then added up together as estimation for the per capita premature death loss. Human capital gains in each year are estimated with the per capita GDP of that year, which is assumed to increase at a rate of g. Per capita GDP in 2012 is gdp and the discount rate is r.

The annual average per capita GDP increase rate is estimated with the average increasing rate from 2001 to 2011, with the growth data collected from the Beijing Statistical Yearbook. The annual average discount rate data in Beijing is collected from World Economic Outlook Databases (IMF 2012).

The total economic loss in premature death is calculated with Eq. (10).

$$ T{C}^{\mathrm{death}}={\displaystyle \sum_{k=1}^2{\displaystyle \sum_{m=1}^{13}{P}_{km}{c}_t^{\mathrm{death}}}} $$

where TC death is the total economic loss caused by the premature death of all individuals because of cardiovascular diseases and respiratory diseases. P km is the number of premature deaths associated with illness k in the mth age strata. c death t is the premature death loss for an individual at age t, which is denoted in Eq. (9).

Results and discussion

In 2012, PM10 was the major pollutant in Beijing. On 337 days of the 366 days in 2012, it constituted the primary pollutant for API calculation. Its annual average concentration was far above the NAAQS grade I standard (40 μg/m3) in all districts of Beijing (Table 5).

Table 5 Annual mean pollutant concentration (μg/m3) and population (10,000) of districts in Beijing

Air pollution does lead to significant health impact. Its impact on morbidity rate of cardiovascular diseases and respiratory diseases may be inferred from Table 6. The positive correlation coefficients in the table indicate that, as the concentration of the pollutant increases, the morbidity of the both diseases will increase, with the incidence of respiratory diseases more sensitive to air pollution. To a further step, the dose-response coefficient for each disease in each sample hospital is estimated with the dose-response model and shown in Table 7.

Table 6 Correlation between air pollution and the morbidity rates of respiratory and cardiovascular diseases in sample districts
Table 7 Morbidity dose-response coefficients of respiratory and cardiovascular diseases

According to the estimation (Table 7), PSD amount of respiratory diseases increases by 0.71, 0.96, and 0.60 %, respectively, for every additional 10 μg/m3 increase in the concentration of air pollutant. We derive the city’s dose-response coefficient by averaging the correlations for both kinds of diseases among the three sample hospitals, with the annual PSD amount for each kind as the average weight. Taking the city as a whole, the dose-response coefficient is 0.42 % for cardiovascular diseases and 0.72 % for respiratory diseases. In other words, for every additional 10 μg/m3 increase in PM10 concentration, the PSD amount of cardiovascular diseases will increase by 0.42 % and that of respiratory diseases will increase by 0.72 % in Beijing.

Plugging our estimation of the dose-response coefficients into Eq. (3), we estimated that the PSD associated with cardiovascular diseases caused by air pollution was 278,000 and that of respiratory diseases was 190,700 in Beijing in 2012. Multiplying the two numbers with the per capital medical cost, we conclude that in the year 2012, air pollution in Beijing resulted in an economic loss of Ұ106.8632 million due to an aggravated risk of cardiovascular diseases and Ұ73.3051 million of cardiovascular diseases.

Similarly, the impact on mortality rate of cardiovascular diseases and respiratory diseases due to air pollution may be inferred from Table 6.

It is indicated in Table 6 that for every additional 10 μg/m3 increase in annual average concentration of PM10, the mortality rate of cardiovascular diseases rises by 0.2 % and that of respiratory diseases rises by 0.48 %.

Next, the rates of premature death, respectively channeled through cardiovascular diseases and respiratory diseases, are calculated for each district in Beijing (Table 8). Overall, air pollution in Beijing, as indicated by PM10 concentration, leads to an economic loss of Ұ402.8535 million in 2012. Also, the health threats diverge markedly among people of different ages. Generally speaking, the probability of air pollution induced death increases with age. Among the cohorts of 5 to 35 years old, very few people would die from air pollution. Beyond 35, noticeably more people are subject to premature death from air pollution. The mortality rate starts to moderately increase since the age of 35, and the increase is accelerated beyond the age of 65. For people older than 65, the risk of death from air pollution is substantially raised. It is estimated for the year 2012 that less than ten adults under age 35 would die of air pollution in Beijing, whereas about 200 adults older than 75 would be killed by the pollution. Finally, it should be noted that the cohort of children under 5 years old is an exception from the increasing trend of death risk with age. They are even more susceptible to air-pollution-induced death compared to juveniles (Table 9).

Table 8 Mortality dose-response coefficients of respiratory and cardiovascular diseases
Table 9 Mortality rates of respiratory and cardiovascular diseases by air pollutant

The health impacts of air pollution among age cohorts are further evaluated in terms of economic losses. It is shown in Fig. 6 that the loss is relatively higher among old age groups, although reduction in human capital for each individual death in these groups is low. Obviously, the size effect (the amounts of premature death in old groups) outweighs the price effect (the economic cost for individual death). Taking premature death channeled through cardiovascular diseases as an example, the death resulted in a loss of about half million among the group with age 0–4, which increased to Ұ6.2 million among people above 70 years old. Compared to that, premature death that resulted from respiratory diseases is generally less harmful for old people. If evaluated in monetary terms, it would only cause an economic loss of Ұ2.7 million in that cohort. However, respiratory diseases would sound terrifying for 0–4-year-old children since deaths caused by respiratory diseases are more than five times higher than those caused by cardiovascular diseases.

Fig. 6

Economic losses from premature death across age cohorts and by diseases

According to our estimation, 32.1 % of respiratory-disease-induced depreciation in human capital could be attributed to young child death, which accounted to Ұ3.9 million. The cost is so high because (1) newborns’ immunity systems are weak and they are more likely to die if infected by air pollution, and (2) individual death in the youngest cohort results in the highest loss since they potentially have the most future incomes.

Finally, we evaluate the loss of premature death from each district of Beijing. As shown in Fig. 7, the three districts with the highest risks of death from air pollution are Chaoyang, Haidian, and Fengtai, with 20.5, 16.9, and 11.3 % of air-pollution-induced premature deaths located in each district.

Fig. 7

Economic losses from premature death by diseases and districts

Population size is the primary reason for regional differences in air-pollution-induced premature death and economic losses. The districts that suffered the most from such death are those with the largest populations, including Chaoyang (population of 3.745 million), Haidian (population of 3.484 million), and Fengtai (population of 2.214 million). In addition, regional emissions are also responsible for regional variation of premature death due to air pollution. As Kenneth et al. (1993) and Tolley et al. (1994) analyzed, vehicle exhaust is one of the major sources of urban emissions. Thus, the costs of air pollution, either in the form of a higher mortality rate or as the associated economic loss, should concentrate around areas with high volume of traffic. In the city of Beijing, two regions are most susceptible to traffic jams: one covers the eastern parts of the 2nd and 3rd Ring Road, both of which are located in Chaoyang District, and the other covers the western part of the 2nd Ring Road, which is located at the intersection of Haidian and Xicheng districts (Public Security Traffic Management Bureau, Beijing Municipal Government). Chaoyang and Haidian are the two districts that suffered most economic losses from air pollution, as estimated in our study.

Another major source of urban emission is industry. Pollution concentration tends to be upheaved in regions with extensive industrial activities, such as Chaoyang, Haidian, Fengtai, Tongzhou, and Xicheng. The five districts together produced 38 % of total industrial added value of Beijing, and they also paid the most losses from air pollution (Fig. 8).

Fig. 8

Distributions of industrial production (left plot) and construction (right plot) in districts of Beijing, 2012. Data source: Beijing Statistical Yearbook 2012

Finally, lack of vegetation coverage may aggravate adverse health impact of air pollution. As shown in Fig. 9, vegetation coverage in the five districts that incurred the most economic loss from air-pollution-induced premature death are all below the city’s average level, i.e., 46.2 %.

Fig. 9

Average vegetation coverage of Beijing and its districts, 2012. The red line indicates average coverage in Beijing. Data source: Beijing Statistical Yearbook 2012

Concluding remarks

We provide the most updated estimation of how air pollution influences human health in Beijing with field data collected in three local hospitals in 2012. We also estimate the resultant economic losses. According to our estimation, 278,000 patients with cardiovascular diseases and 190,700 patients with respiratory diseases were victims of the city’s air pollution. Among them, 346 people died of cardiovascular diseases and 204 died of respiratory diseases, respectively accounting for 1.27 and 1.64 % of the total deaths from the two kinds of diseases. Considering both the medical costs spent on disease treatment and depreciation of human capital from premature death, the economic losses that resulted from air pollution in Beijing is Ұ583.0218 million, accounting for 0.03 % of the city’s GDP in 2012. The total is composed of four parts: (1) medical cost of Ұ106.8632 million spent on cardiovascular disease treatment, (2) medical cost of Ұ73.3051 million spent on respiratory disease treatment, (3) human capital depreciation of Ұ284.5620 million due to premature death that resulted from cardiovascular diseases, and (4) human capital depreciation of Ұ118.2915 million due to premature death that resulted from respiratory diseases.

Our estimation should be used with caution. On the one hand, we select sample hospitals from more polluted districts in Beijing, and when estimating the mortality rate of premature death from air pollution, we substitute the dose-response coefficient calculated based on the data from Hong Kong, where polluted air contains a higher percentage of hazardous chemical components. Both facts may exaggerate the loss from air pollution in our study. However, on the other hand, we do not include the potential loss from infrastructure and building accelerated aging, damaging the city image, reduced investment opportunity, tourism slipping (tourist arrivals, especially inbound tourists, have decreased from 2009), and people’s negative emotion that stemmed from air pollution, all of which would make this study a conservative one.

Air pollution’s health impacts are significantly varied along the dimensions of age and region in Beijing. Facing air pollution, people of old age and children generally suffer more than the other groups. Senior people over 70 years old and children under 5 years old are the major victims of Beijing’s air pollution. As to regional difference, health conditions of people living Chaoyang, Haidian, Fengtai, Tongzhou, and Daxing are more likely to be adversely affected by air pollution, with the first three districts facing the highest risks of death from air pollution.

In order to mitigate the negative impacts on human health of air pollution in Beijing, the following policy interventions are possibly effective. First, as the most fatal pollutant in Beijing, emission of PM10 (of course including PM2.5) should be strictly controlled, possibly in ways including but not limited to shifting Beijing’s industrial structure towards a highly technical and environmental friendly one, curbing private vehicle ownership growth, promoting the development of electric vehicles, reducing traffic jam with improved traffic management system, and cleaning power generation. Second, steps should be taken to green the city, especially the most industrialized parts of Chaoyang, Haidian, and Fengtai, through afforestation. Third, special protective measures should be applied to the elderly and children, as they are most vulnerable air pollution harms.

The main contribution of the paper is to empirically investigate the impact of air pollution on people’s health and accordingly economic loss in Beijing, China. Due to the availability of data, our results may be biased because we only used data collected from three sample hospitals, and we have to use the correlation between the number of premature deaths caused by diseases and air pollutants in Hong Kong to roughly estimate that in Beijing.


  1. 1.

    The National Ambient Air Quality Standard specifies three levels of compliance (NAAQS I-III). Grade I indicates that annual mean concentrations of NO2 and PM10 are lower than 40 μg/m3 and that of SO2 is lower than 20 μg/m3; grade II in the range of 40–70 μg/m3 for PM10 and 20–60 μg/m3 for SO2; and annual mean concentrations of SO2, NO2, and PM10 exceeding 60, 40, and 70 μg/m3, respectively, reflect air quality worse than the grade III standard (see GB 3095-2012).

  2. 2.

    By definition PM10 consists of particular matters with diameters less than 10 micrometer, definitely including those less than 2.5 micrometer, which are PM2.5.

  3. 3.

    According to a recent survey of air quality in China by the World Bank, almost 80 % cities were in grade I or II, and only a few left above grade III (China Ministry of Environmental Protection and World Bank 2012).

  4. 4.

    By comparing the health impacts of air pollution in coal-combustion and non-coal-combustion regions, Jin and his colleagues (2002) conclude that an adult is 1.7 times more likely to incur respiratory diseases in regions severely polluted by PM10, NO2, and SO2 than a counterpart in a relatively clean area, and the chance is 2.23 times higher for juveniles.

  5. 5.

    The concept of “primary pollutant” is derived from the definition of Air Pollution Index (API, GB3095-96). It was a comprehensive air quality standard used in China between 1996 and 2012 (Air Quality Index, AQI, replaced API in 2013). The index gauges pollution condition by comprehensively considering pollution caused by the three major pollutants, PM10, NO2, and SO2, and measures air pollution with the concentration of the worst pollutant, which is also called primary pollutant, on each day. In this way, API can better reflect air condition improvement than other indicators, such as average pollution level or the best level. The primary pollutant may vary day to day, but the score is kept consistently.

  6. 6.

    Since the record of PM2.5 concentration was not available before 2009 in China, the cited studies all focus on inhalable particulate matter (PM10).

  7. 7.

    This is especially true before the year 2012, when China employed the air pollution index (API) to evaluate its air pollution condition. API consists of only the three pollutants: SO2, NO2, and PM10. In later 2012, however, the API system was replaced with the air quality index (AQI) system which considers a broader range of pollutants. Since the study evaluated the health impact of air pollution for the whole year of 2012, we specifically focus on pollutants covered by API, which are SO2, NO2, and PM10.

  8. 8.

    Consumption of meat and eggs is an indicator of residential dietary pattern, which has potential influence on morbidity rate. We assume that supply of meat and eggs in the study period was stable and that price level was positively related to the demand as well as total assumption. Thus, the price index of meat and eggs could be used as an indicator of residential dietary habit.

  9. 9.

    The study focuses the impact of PM10, without explicit discussion about that of PM2.5, which is stemmed from the following reason: PM10 is referred as the inhalable particles with diameters less than 10 μm. It harms the human respiratory system and causes illness. The difference between PM10 and PM2.5 is that the particles with a diameter of 10 μm deposit generally at upper respiratory tracts, while the particles with diameters less than 2.5 μm can enter into deeper place of respiratory tracts.

  10. 10.

    Thresholds refer to a pollutant concentration, below which we believe air pollution would not harm human health. It is a key factor in almost all evaluations of the health impact of air pollution.

  11. 11.

    By “local,” we mean hospitals that mainly serve local residents. As the national capital, there are also many national hospitals located in Beijing, which serve both patients from and outside Beijing. These national hospitals are deliberately excluded from our sample.

  12. 12.

    We define pollution in this study with concentration of a pollutant higher than its threshold.

  13. 13.

    The weight is identified by the proportion of the PSD for the cardiovascular disease or respiratory disease in a specific hospital to the total PSD for the two kinds of diseases in the three hospitals.

  14. 14.

    According to the Chinese National Statistical Report (2013), life expectancy in Beijing is 80.18. We take the integer part in our calculation.


  1. An AP, Ke LF, Dong HQ (2006) The effect of atmosphere environment upon human health. J Environ Occup Med 3(22):78–81 (in Chinese)

    Google Scholar 

  2. Aunan K, Pan X (2004) Exposure-response functions for health effects of ambient air pollution applicable for China: a meta-analysis. Science of the Total Environ 2004(3):3–16

    Article  Google Scholar 

  3. Bravo MA, Son J, de Freitas CU, Gouveia N, Bell ML (2016) Air pollution and mortality in São Paulo, Brazil: Effects of multiple pollutants and analysis of susceptible populations. J Expo Sci Environ Epidemiol, 26(2):150–61

  4. Cao, C. H., Han, L. Y., 2014. Evaluation of health costs by PM10 pollution in Beijing. Working paper. Available at (in Chinese)

  5. Chen BH, Hong CJ, Zhu HG et al (2002) Quantitative evaluation of the impact of air sulfur dioxide on human health in the urban districts of Shanghai. J Environ Health 11(1):56–59 (in Chinese)

    Google Scholar 

  6. Chen R, Kan H, Chen B et al (2012) Association of particulate air pollution with daily mortality: the China Air Pollution and Health Effects Study. Am J Epidemiol 175:1173–1181

    Article  Google Scholar 

  7. Daniel R, Petrolia TK (2011) Preventing land loss in coastal Louisiana: estimates of WTP and WTA. J Environ Manag 92(3):859–865

    Article  Google Scholar 

  8. Dong XW, Zhi RZ (2010) Valuing recreational benefits of environmental amenity based on contingent valuation method: a case study of Jiuzhai Gou. Conf Environ Sci Inf Appl Technol 2:91–94

    Google Scholar 

  9. Dong SZ, Zhang ZY, Cai CK et al (2007) Effects of air pollution on pulmonary function of children in industrial area. J Environ Health 15(2):71–73 (in Chinese)

    Google Scholar 

  10. Dusseldorp A, Kruize H, Brunekreef B et al (1995) Association of PM10 and airborne iron with respiratory health of adults living near a steel factory. Am J Respir Crit Care Med 152:1032–1039

    Article  Google Scholar 

  11. Hausman A, Ostro B, Wise DA (1984) Air pollution and lost work. NBER. 31, 1263

  12. HEI Accountability Working Group (2003) Assessing health impact of air quality regulations: concepts and methods for accountability research. HEI communication

  13. Ho KF, Lee SC, Chan CK, Yu JC, Chow JC, Yao XH (2003) Characterization of chemical species in PM2.5 and PM10 aerosols in Hong Kong. Atmos Environ 37:31–39

    CAS  Article  Google Scholar 

  14. Holland M, Watkiss P, Pye S, Oliveira AD, Regemorter DV (2005) Cost-benefit analysis of policy option scenarios for the Clean Air for Europe Programme. AEA Technology Environment

  15. International Monetary Fund (2012) World Economic Outlook Database-April 2012 Edition. Available at

  16. Kan HD, Chen BH (2004) Particular air pollution in urban areas of Shanghai, China: health-based economic assessment. Sci Total Environ 322(1):71–79

    CAS  Article  Google Scholar 

  17. Kuehn BM (2014) WHO: more than 7 million air pollution deaths each year. J Am Med Assoc 311(15):1486

    CAS  Article  Google Scholar 

  18. Künzli N, Kaiser R, Medina S, Studnicka M, Chanel O, Filliger P, Herry M, Horak F Jr, Puybonnieux-Texier V, Quénel P, Schneider J, Seethaler R, Vergnaud J-C, Sommer H (2000) Public-health impact of outdoor and traffic-related air pollution: a European assessment. Lancet 356:795–801

    Article  Google Scholar 

  19. Kwiterovich P (1997) The effect of dietary fat, antioxidants, and pro-oxidants on blood lipids, lipoproteins, and atherosclerosis. Am Diet Assoc 97:31–41

    Article  Google Scholar 

  20. Li J, Guttikunda SK, Carmichael GR et al (2004) Quantifying the human health benefits of curbing air pollution in Shanghai. J Environ Manag 70(1):49–62

    Article  Google Scholar 

  21. Ostro BD (1983) The effects of air pollution on work loss and morbidity. J Environ Econ Manag 10(4):371–382

    Article  Google Scholar 

  22. Pope CA III, Dockery DW (2006) Health effects of fine particulate air pollution: lines that connect. J Air Waste Manag Assoc 56:709–742

    CAS  Article  Google Scholar 

  23. Reuters Staff BEIJING (2014) China sets targets for curbing air pollution, Reuters Health Medical News. January 9

  24. Samakovlis E, Huhtale A, Bellander T, Svartengren M (2005) Valuing health effects of air pollution focus on concentration-response functions. J Urban Econ 58(2):230–249

    Article  Google Scholar 

  25. Sun Y, Zhuang G, Tang A, Wang Y, An Z (2006) Chemical characteristics of PM2.5 and PM10 in haze-fog episodes in Beijing. Environ Sci Technol 40:3148–3155

    CAS  Article  Google Scholar 

  26. WHO (2006) Air Quality Guidelines, Global Update 2005, particulate matter, ozone, nitrogen dioxide and sulphur dioxide. WHO Regional Publications, WHO Regional Office for Europe, Copenhagen

    Google Scholar 

  27. WHO (2013) Review of Evidence on Health Aspects of Air Pollution - REVIHAAP Project. WHO Regional Publications, WHO Regional Office for Europe, Copenhagen

  28. Xu MM, Guo YM, Zhang YJ, Westerdahl D et al (2014) Spatiotemporal analysis of particulate air pollution and ischemic heart disease mortality in Beijing, China. Environ Heal 13(1):1–12

    Article  Google Scholar 

  29. Yang MJ, Pan XC (2008) Time-series analysis of air pollution and cardiovascular mortality in Beijing, China. J Environ Health 25(4):294–297

    CAS  Google Scholar 

  30. Zhang LY, Liu FZ, Jin L (1999) The study on the effects of air pollution on the children’s health in an industrial area in Tianjin. J Environ Health 16(2):93 (in Chinese)

    Google Scholar 

  31. Zhang XM, Pei XK, Wang JH (2000) Epidemiological study on air pollution. J Environ Health 17(1):9

    CAS  Google Scholar 

  32. Zhang M, Song Y, Cai X (2007a) A health-based assessment of particulate air pollution in urban areas of Beijing in 2000–2004. Sci Total Environ 376(1–3):100–108

    CAS  Article  Google Scholar 

  33. Zhang R, Cao J, Lee S, Shen Z, Ho K (2007b) Carbonaceous aerosols in PM10 and pollution gases in winter in Beijing. J Environ Sci 19:564–571

    CAS  Article  Google Scholar 

Download references


This research is supported by grants from the National Science Foundation of China (71373078, 71573273, 71403011, and 7140030534) and awarded by the hit-haunting for talents project in China University of Petroleum–Beijing (2462014YJRC052).

Author information



Corresponding author

Correspondence to Xueying Yu.

Additional information

Responsible editor: Philippe Garrigues

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Zhao, X., Yu, X., Wang, Y. et al. Economic evaluation of health losses from air pollution in Beijing, China. Environ Sci Pollut Res 23, 11716–11728 (2016).

Download citation


  • Air pollution
  • Health damage
  • Corrected human capital method