Study of Economic Aspects of Ambient Air Pollution on Health Effects


Study of Economic Aspects of
Ambient Air Pollution on Health Effects

Tender ref. AS 97-079

Reference : R0036-3.98

Client : Environmental Protection Department

Air Services Group Date : April 1998

For and on behalf of EHS Consultants Limited : Prepared by  :  Dr. Lai Wan Yee, Consultant

Checked by : Dr. Sarah Liao, Managing Director


Executive Summary











Table 1: Relative Risks (RRs) and 95% Confidence Interval (95% CI) for 50 ug/m3 increase Air Pollutants

Table 2: Results Of Economic Valuation Of Morbidity Using COI Estimates:

Table 3: Summary Of WTP Estimates From The Contingent Valuation On Morbidity

Table 4: Relative Risks (RRs) and 95% CI for 50 ug/m3 increase in the Concentration of Air Pollutants

Table 5: Results Of Economic Valuation Using COI And WTP Approaches:

Table 6: Results Of Economic Valuation Using COI And WTP Approaches:

Table 7: Relative Risks (With 95% CI for Composite Score) of 50 ug/m3 increase in Air Pollutants

Table 8: Summary Of Economic Cost Using Composite Score

Table 9: Summary Of Predicted Health Impacts For 1996

1. Introduction

1.1 This study is an economic valuation study to present the monetary value estimates for the adverse human health effects resulted from ambient air pollution. The relationships between ambient air pollution and morbidity and mortality are based on the study by the Chinese University (CUHK, 1997) and a follow-up study by the University of Hong Kong (HKU, 1998), both commissioned by the Hong Kong Environmental Protection Department (EPD).
1.2 Both the studies on short-term effects of ambient air pollution on public health revealed that, similar to other foreign studies, there were significant correlation between the concentrations of air pollutants and the mortality and morbidity rates of certain types of respiratory and cardiovascular diseases. It is assumed that the impacts of the four selected criterion pollutants, namely nitrogen dioxide (NO2), sulphur dioxide (SO2), respirable suspended particulates (RSP), and ozone (O3), on diseases under the broad categories of respiratory and circulatory diseases have already been ascertained by the previous studies.
1.3 This report presents the economic costs of the effects of ambient air pollution on health by the cost of illness (COI) and willingness-to-pay (WTP) approaches. The available quantitative data was fitted into a mathematical model constructed to represent the economic impact of air pollution in Hong Kong with respect to morbidity and mortality. For the COI approach, data available on lost of earnings and medical costs were used for the measurements of the value of morbidity and mortality; while cost estimates from the contingent valuation were used for the WTP approach.
1.4 The concept of COI and WTP had been adopted in a number of studies for the economic evaluation of health impacts associated with ambient air pollution. For example, one study showed that there is an association between the US 1980 mortality rates and respirable particulates and/or toxic fraction of the aerosols(Ozkaynak and Thurston, 1987). In another study, it was concluded that the measurable costs of air pollution are high enough to justify substantial expenditures to control vehicle emission rates (Small and Kazimi, 1995). There are also several studies relating to the cost of disease and premature death due to air pollution using the concept of willingness to pay (WTP) and actual cost calculation (Tolley et al., 1994; HMSO, 1996).

2. Objective of the study

2.1 The objective of this study is to evaluate the monetary values associated with the morbidity and mortality caused by the rise of individual criteria ambient air pollutant based on the relative risk factors provided by EPD.

3. Methodolgy

3.1 Disaggregated data was used for estimating COI for the individuals (for details on COI and WTP, see Appendix 1a). For this study, hospital admission (morbidity) and mortality data provided by the Hospital Authority, and out-patient data provided by the Department of Health were used in the COI evaluation.
3.2 The estimates were then used to extrapolate the total amount of illness caused by air pollution and here we shall assume that the population is subject to the same exposure. The data on relative risk which link individual health effects to changes in air pollution were combined with the COT estimates.
3.3 For WTP valuation, the study team undertook questionnaire design with the view of applying the focus group (FG) technique to generate individual and group responses in the survey (see Appendix 1b and 1c). This was to ensure good responses as well as reliable data. The FG technique has been regarded as a stand alone procedure for obtaining data on any social context that is being studied in an ethnographic manner (Morgan, 1993). Through group interaction, the shared experiences can maximise exploration of different perspectives within a group setting. Group and individual responses can be obtained and the moderator can lead the discussion on the issues before presenting the more formal quantitative questions.

4. Data/statistics used for estimation of economic cost


Data Obtained From Focus Group Survey:

4.1.1 Details containing the dataset obtained from the FG Survey may be cross reference to the Report on Focus Group Survey Data (R1921.97).

4.1.2 While the main focus of the questionnaire was to determine individual?s willingness to pay (WTP) to avoid the disease occurrence, WTA is used to determine how much compensation is required by the respondent to give up the non market good in question. The dataset obtained were compiled (see Appendix 2 showing a summary of the FG Survey Data) and cross reference may also be made to the Report on Focus Group Survey Data, R1921.97, November, 1997.


Non Survey Data:

4.2.1 Apart from the survey, data were also collected from other sources, for example, the Hospital Authority (HA), Department of Health, Census and Statistics Department and Government Gazette.
4.2.2 The dataset from HA is a similar set as that used by CUHK in the statistical analysis for 1994 and 1995. The dataset was made available to EHS through EPD. The information includes both morbidity and mortality by the diseases identified with breakdown by age groups.
4.2.3 Since the CUHK study, EPD has also commissioned a follow-up study on the health impact of air pollution. The relative risks estimated by the Department of Community Medicine at the University of Hong Kong (HKU) are made available to EHS by EPD. While the CUHK study used data of 1994-95, the HKU study used data of 1995-96 and they included different ICD9 codes for circulatory and respiratory diseases. As such, the results of the CUHK study may not be comparable to those from the HKU study (HKU, Jan., 1998, p. 5 & p.17)
4.2.4 The data provided by the Department of Health show the number of attendance at the general out-patient clinics over a one-week period in August, 1995. Breakdown by age groups and by diseases of first diagnosis and cases referred to emergency were indicated (see Appendix 3 for details).

5. Models For Health Evaluation

This mathematical model for evaluation of the economic aspects of ambient air pollution on public health comprises two main modules as follows:

i) COI estimates
ii) WTP estimates


Model to Estimate Economic Cost of Morbidity


Using COI Estimates The dollar value of morbidity is calculated from the data collected. For COI estimates, they are based on values associated with cost of illness, i.e., direct and indirect cost, such as, medical expenses (including self-medication and doctor consultation) and productivity loss, compensation schemes and awards. For this study, the economic costs of health effects due to air pollution will be limited to hospital treatment as in-patients and mortality (upon which CUHK preformed the statistical analysis of association between health effects and air pollution). The models with specific air pollutants for hospital admissions indicating the relative risks (RR) with 95% confidence intervals (95% CI) (by CUHK/HKU) is the basis for estimating the monetary value of the health outcome. Since the actual number of admission (as health outcome) resulting from pollution effect is known (from the available model), the cost of illness can be estimated by considering the factors concerned.

The COI estimated is then used as an input for estimating the economic cost as presented in the following model:

E = b(dj) * COI(dj) * { RR(pi, dj) -1 }

E = economic cost of morbidity by hospital admission 
b(dj) = actual hospital admission and out-patient doctor consultation due to a particular disease dj
dj = diseases (e.g., respiratory or circulatory diseases)
COI = estimated cost of illness related to hospital admissions and out-patient treatment due to dj
RR = Relative risk for morbidity - pi, dj (air pollutant & disease specific in respective order)
pi = air pollutants (e.g. NO2, SO2, RSP, and O3) In practice, COI is first computed based on the statistics on hospital admission and the result obtained is subsequently applied in the model. In this way, the economic cost associated with hospital admission is the product of the total cost of illness and the relative risk (RR) of admission. RR has already been estimated in a separate study on the short-term health impact of air pollution in Hong Kong (see CUHK, 1997; HKU, 1998). It is obtained by multiplying the exponential of the regression coefficient by the defined increase in the pollutant.

For the calculation of COI for morbidity (where hospital admission is expected to be the key cost component), we have identify the cost components (direct and indirect cost related to the illness) as follows:

Cost components (COI estimates):

Direct cost Indirect cost

Self medication:

Buying medication over the counter 

Defensive expenditure e.g., special diet, equipment

Output (earning) loss:

Off-work day (not serious enough for admission)

Treated as Out-patient: 

Doctor consultation - GOPC (12% of all cases)

GOPC charges (registration fee at HK$34) 

Subsidies for GOPC cases (HK$141 per case) 

Doctor consultation - private (88% of all cases)

Off-work day (average no. of days hospitalised for admission cases and 1 day off for non hospitalised out-patient cases) 
  1. assume no accompanying cost
  2. assume accompanying cost 

Hospital Admission:

Average length of hospital stay 
Room rate per day 
Subsidies involved The 1995 dataset has been used instead of 1994 dataset. COI for respiratory diseases and cardiovascular diseases are calculated separately. For the statistics on the number of hospital admissions in 1995, distribution by age groups are used and together with the average length of hospital stay, both direct and indirect cost of illness are estimated. The use of the HA statistics is consistent with the CUHK study. The overall employment earnings at HK$10,000 per month has been used to estimate indirect cost. This statistics is provided by the Census and Statistics Department. This is the latest figure available (1996 and 1997) and it is compiled from the General Household Survey by occupation of main employment. Since the very young and the elderly are more likely to require someone else to assist them especially when their symptoms are severe enough to warrant hospital admission, estimates with accompanying cost are more representative than those without accompany cost. Direct cost is largely the inpatient charges which include maintenance fee. According to the Hong Kong Government's charge rate, the entitled person is required to pay a sum of HK$68 as the maintenance fee (includes charges for clinical, biochemical and pathological investigations, vaccines and general nursing but excludes consultation, diagnostic imaging etc.). For the same health care services, the non entitled person will have to pay HK$3,130 (Special Supplement No.4, 1996)8. We use the difference between this amount (HK$3,130) and the daily room charge of HK$68 as the subsidy to the entitled person (i.e., HK$3,062). The sum of HK$3,130 would therefore be the public ward maintenance fee which includes charges for clinical, biochemical and pathological investigations (including consultation, diagnostic imaging and other examinations), vaccines and general nursing, where such examination or treatment are necessary, and medicines within the scale provided at the hospitals (S. S. No.4 to Gazette No. 44/1996, Para.2.0). Indirect cost is made up of earnings loss. For the calculation, we assume that the average length of hospital admission is also the number of days work has been affected due to sickness. In this case, we assume employment outside of the house is affected only when individual is being admitted to hospital. In addition, there is a subdivision for this particular cost area. For some persons, we assume the individuals do not require the assistance of others, however, some categories, often the young and old tend to require accompanying cost. We treat those aged 14 and below and those above 65 to be those where there will be accompanying. The cost is estimated on a per day basis from the known monthly overall wage for Hong Kong (statistics obtained from the Census and Statistics Department). Medical insurance payments can be treated as subsidies on medical expenses. Even if the victim need not pay the full medial bill, the payment by insurance company constitutes a cost to the society because resources are limited and once it is applied in one area, other area would have a lesser share. Since detailed information on individuals'medical insurance coverage is not readily available, and so long as we avoid double counting of the cost variables, we deem it acceptable to skip the differences due to insurance payment by employing companies. The key results of the COI estimates for respiratory and cardiovascular diseases are presented in Appendix 4a and 4b. As the economic evaluation of morbidity based on hospital admission alone is likely to underestimate the health impact of air pollution, we have taken into account the total number of out-patient attendances. Doctor consultation fee includes both public GOPCs and General Practitioner (GP) categories. The data for GOPC attendances are available from the Department of Health Survey (DH, 1997). The information provided by the DH also specified that the GOPC attendances at public out-patient clinics only made up 12% of the market share. Hence, we have used the statistics given to estimate the remaining 88% GOPC attendances. The statistics on GOPC are the survey findings on "Disease Surveillance at General Out-patient Clinics" conducted by the Department of Health over one week (18 - 24 August) in 1995. These findings could be subject to seasonal effect and hence might not represent the characteristics of the patient attending GOPC in a longer period. Also, the survey covered patients attending public day GOPCs for doctor consultation only. Patients attending non-public, evening, Sunday and public holiday and mobile clinics were not included in the survey. The economic evaluation of the GOPC attendances offered only broad estimates since it is difficult if not impossible to determine that the pre-disposing factor of the ailment is an associated increase in air pollution. The symptoms of illnesses treated at GOPCs tend to be less severe. Had the two previous statistical analysis also considered the impact of the criterion air pollutants on out-patient cases, we would have a RR for the GOPCs. In this study, we assume all the out-patient cases treated for respiratory diseases and cardiovascular diseases suffered the impact of air pollution. We also assume that every clinic visit will incur an average of one day of work loss (based on the average wage (i.e., HK$120,000/365). The results of COI (Appendix 4) are applied to the valuation of health impacts together with the relative risks estimated previously by CUHK (September, 1997) and HKU (Jan., 1998). It is considered necessary to take into consideration both hospital admissions and GOPCs attendances in the broad category of morbidity. Details of the workings on economic valuation of morbidity using the relative risks for 50ug/m3 increase in the levels of NO2, SO2, RSP and O 3 are shown in Appendix 5a (by respiratory and cardiovascular diseases) and Appendix 5b (combined respiratory and cardiovascular diseases).

The set of RR ( Table 1) is taken from the follow-up study of health impacts by HKU (Jan., 1998).

Table 1: Relative risks (RR) and 95% confidence interval (95% CI) for 50 ug/m3 increase in the concentration of air pollutants for hospital admissions of respiratory and cardiovascular diseases:
Diseases/Pollutants Respiratory Circulatory Respiratory & Circulatory
NO2 1.08 1.08 1.07
SO2 1.03 1.05 1.04
RSP 1.05 1.03 1.04
O3 1.10 1.07 1.07

Source: HKU (Jan., 1998) 

The results of the valuation on morbidity based on respiratory and cardiovascular diseases and combined respiratory and cardiovascular diseases are summarised below (Table 2):

Table 2: Results of Economic Valuation of Morbidity Using COI Estimates:
Economic cost (in HK$ million)
Relative Risk 50ug /m3 Respiratory Diseases Cardiovascular Diseases Combined Respiratory & Cardiovascular 
NO2 522.20 451.90 852.34
SO2 195.82 282.44 487.05
RSP 326.37 169.46 487.05
O3 652.75 395.41 852.34 In comparing the economic cost of the impact of air pollution on respiratory and cardiovascular disease, RSP accounted for a higher economic cost in respiratory diseases than in cardiovascular diseases. This is probably the result of a higher incidence of respiratory illnesses caused by RSP. Comparison is also made between direct cost of COI estimated and public expenditure on medical and health services. For the purpose of comparing public expenditure on health, especially health expenses incurred by HA and the Department of Health, the direct cost component of COI estimates is extracted from this study.

We shall present the economic valuation of direct cost as follows:

Category I:   Category II:

By type of diseases:

  • respiratory diseases
  • cardiovascular diseases
  By combined respiratory and cardiovascular diseases:
1. Hospital admissions   1. Hospital admissions 
2. GOPC and private doctor consultation   2. GOPC and private doctor consultation

The results of direct cost estimated are presented in Appendix 5c(i) and 5c(ii).

The direct cost estimated based on respiratory and cardiovascular diseases in separate categories (see Appendix 5c(i)) and as combined diseases (Appendix 5c(ii)) were presented but the estimates in the latter were used for the purpose of comparison. The direct cost on an annual basis ranged from HK$255.98 million to HK$447.97 million. The annual expenditure of the Department of Health (DH) and HA were HK$2,283 and HK$18,922 respectively for 1995/1996 (see Department of Health, Annual Report 1995/96; Census and Statistics Department, 1997). The direct cost calculated is equivalent to almost 20% of DH's expenditure for 1995/1996 (at the upper limit) whereas it is only equivalent to 2.37% of HA's expenditure over the same period. This can be explained by the relative size of HA expenditure compared to that of DH (an eight-fold increase). Details are shown in Appendix 5d.


Economic Cost of Morbidity Calculated Using WTP Estimates

WTP estimates go beyond the lost of earnings and medical costs. However, WTP values are mainly obtained from contingent valuation and they are grouped as follows:

WTP to avoid the symptoms of the diseases concerned:

Light symptoms without the need for doctor consultation

One symptom day suffered (descriptions of symptoms provided)

Seven symptom day suffered

WTP to avoid hospitalisation specific to the diseases:

One day of hospitalisation 

One week of hospitalisation

The appropriate value from the FG data set is summarised in Table 3 below:

Table 3 Summary of WTP Estimates From the Contingent Valuation on Morbidity
  WTP 1 week symptom but no hospital admission 1day hospitalisation
Respiratory diseases Min 30 20
Max 10,000 20,000
SD 1,825.19 3,664
Average 735.52 1,137
Cardiovascular diseases Min 150 50
Max 15,000 20,000
SD 3,170.09 3,981
Average 1,517.59 1,736

In substituting the WTP values to estimate the economic cost of morbidity, we use a similar equation to the one used for COI estimates for hospital admissions. COI input is substituted with WTP. The model is shown as follows:

E = [ b(dj) * ( RR(pi, dj) -1) ] * WTP(dj)

= economic cost of morbidity (by hospital admission & out-patient attendances)
= actual hospital admission and out-patient doctor consultation of a particular illness d
dj = illness (e.g., respiratory, circulatory diseases) 
RR  = Relative risk for morbidity due to (pi, dj) (available from study by HKU)
pi = air pollutants (e.g. NO2, SO2, RSP, and O3)
WTP = estimated monetary value WTP to avoid dj The "average" WTP value to avoid a day of hospital admission is taken into account. In addition, the average willingness to pay by individuals'to avoid the symptoms during the disease occurrence (non hospitalised cases) is also considered in the calculation. This average bid for 7-symptom days for cases without hospital admission is obtained from the FG survey. Details of the calculations are presented in Appendix 6(i) and Appendix 6(ii). The former shows estimates of the costs based on respiratory diseases and the latter shows estimates for cardiovascular diseases. According to Chestnut (1995), subsidy may be considered under certain circumstances 1. In this study, we have estimated the economic cost with an element of subsidy in the willingness to pay estimates for morbidity. We considered WTP values by the individuals in the FG group understated and this is probably for reason that they left out the subsidy by the public sector. The results of the economic cost for total morbidity (respiratory and cardiovascular diseases) are presented in Appendix 7. Appendix 7 presents the calculation of the total economic cost with the relative risks available from HKU study. The working is similar to the calculation using the COI estimates. The results of monetary valuation by respiratory diseases and cardiovascular diseases (as two separate diseases and as combined respiratory and cardiovascular diseases) are presented in Appendix 7.

1Chestnut recommends that for a comprehensive measure of WTP, the "share cost" should be added to individual WTP. He argues that financial costs of health effects not always borne by the individual but are shared through health insurance and public health care subsidies. In some instances therefore empirical estimates of WTP to avoid or reduce health effects may not fully reflect these shared costs. The incidence Chestnut is referring to is probably valid in the case of Hong Kong where the room rate for hospital admission is relatively low and a large proportion of the focus group members indicated that they were willing to pay round about the amount of this fee rather than the subsidy which is as much as 40 times the fee payment.


Model to Estimate Economic Cost of Mortality


Cost of Illness (COI) Approach to Value Life (Mortality)  Cost of illness approach has been the prevailing method used to value life up until the recent years although in practice, it is the economists who tend to favour the WTP method whereas non economists (e.g., engineers and health professionals) may continue to use the human capital calculations (using COI approach) (Haight, 1994)9. The human capital approach is age-specific and it measures the value of individual's life to society, i.e., future production potential. It is usually calculated as the present discounted value of expected labour earnings. The standard approach assumes a zero value for persons without labour income. However, the choice of an appropriate social discount rate to convert future earnings into present values is still unresolved. Different discount rates will provide different present value of future earnings. The value is significantly bigger for low discount rate. The rate of discount recommended by the US Office of Management and Budget is 10% (Landefeld & Seskin, 1982)10. This represents an estimate of the average rate of return on private investment before taxes and after inflation. In order to calculate the value of a statistical life (VSL), we need to know the life expectancy, labour force participation and projected earnings. The statistical value of a life year is the aggregate of all the monthly earnings for the year (monthly earnings multiply by 12). The estimated capitalised lifetime earnings is thus the VSL. For example, if a fresh graduate were to enter the labour market at the age of 23, his/her capitalised value would be calculated based on the average earnings over his paid employment period assuming the person would work until his/her retirement age. For example, a in the US, a 39-year period is used. We shall use the net present value model for calculating the capitalised earnings. We assume the individual is healthy and earning an income (here, income is derived from labour of the individual before tax payments). For the calculation, we are required to consider the net present value of a stream of earnings received over the employment period, assuming retirement is at age 65, thus, a period of about 40 years may be considered. The earnings takes into account tax payment. The reported earnings is generally made in constant or base year dollars. It is necessary to convert the nominal discount rate to a real rate by an adjustment for inflation. For example, if we were to estimate the net present value (NPV) of capitalised earnings for a fresh graduate, we may use the real dollar value (at constant price) instead of the nominal value and apply a discount rate.

WTP Approach to Value Loss of Life For this, we use a change in risk that will affect loss of life. The revealed-preference through the WTP method is an alternative to human capital (COI) approach. It is used to measure the private valuation individuals place on small reductions in the risk (probability) of death (i.e., m/^ p where, m is the WTP for the benefit of living and ^ p is change in risk). The result obtained indicates the value per statistical life. The median value indicated by individuals' willingness to pay (from the FG questionnaire survey) will be retrieved from the dataset compiled (see questionnaire which refers to a small change in mortality - risk reduction). The results obtained for the VSL using COI and WTP approaches will be discussed here. We shall begin with the COI estimates of the VSL. The overall earning for the Hong Kong population is estimated at HK$10,000 (in 1997 prices) (see Census & Statistics Department (C&SD), correspondence where C&SD provided EHS statistics from Household Survey by major employment categories, 1998). For this purpose, we shall use a discount rate which reflects the individual rate of time preference. The "Yield of Exchange Fund Bills and Notes" for a 10-year period at around 7% is used (see HKMA, Monthly Statistical Bulletin, July, 1997). The calculation performed is based on the overall employment earnings for Hong Kong 1996/1997 at HK$10,000 per month or HK$120,000 per annum (Census and Statistics Department, 1998). The stream of earnings discounted over 40 years period is HK$1,597,579. This means that the value of a statistical life is about HK$1.60 million. Details of the calculations performed on the stream of income at a discount rate of 7.0% are shown in Appendix 8a. According to the data by Department of Health, the total number of deaths for all age groups due to respiratory diseases and cardiovascular diseases were 5,821 and 8,748 respectively for 1995/96. The total number of deaths (for 1995/96) is taken into account in the calculation. The capitalised earnings over a span of 40 years, based on a 7.0% discount rate (as the opportunity cost of capital) is used in the calculations. Estimation of capitalised earnings are shown in Appendix 8(i) and Appendix 8(ii). The statistics show that the largest number of deaths occurred in the age group "65 and above" and it is true for both respiratory diseases and circulatory diseases and they accounted for 86.75% and 83.04% respectively. Had the calculations been limited to only those from age groups within the categories of working age population (inclusive of age groups "15-24" and "55-64"), the number of deaths accounted for about 13% and 16% of deaths from respiratory and circulatory diseases respectively. Details of the calculations are presented in Appendix 8c(i) and Appendix 8c(ii) (which show the VSL estimated by deaths from respiratory and cardiovascular diseases). For the WTP estimates, we use the dataset from the focus group survey. For a risk reduction of death from respiratory diseases by 0.01%, the median value individuals were prepared to pay is HK$500. Thus, the implied value per statistical life (VSL) is HK$5 million. The median of individuals' WTP to reduce the risk of cardiovascular diseases by the same amount (i.e. 0.01%) is also HK$500. The VSL is therefore the same as that for respiratory diseases (i.e., HK$ 5 million). This median value is then used to estimate VSL based on the actual number of deaths. In the calculation of VSL using WTP estimates, we take into account age factor. For working purposes, we estimated the mean age of all deaths by the diseases concerned. For simplicity, the victims of air pollution are divided into two broad categories (age below/ age above 76 and 75 for respiratory and cardiovascular diseases respectively). For the group below the mean age, we estimated the VSL of premature deaths by discounting the values based on WTP estimates over a 40-year period (similar to the estimation applied to COI). However, for age group beyond the mean age, their willingness to pay to change the risk of death would not be discounted to reflect that unlike the younger population, they can realise the value in the immediate future. Details of the estimations are shown in Appendix 8b.

Substituting VSL Obtained Into the Valuation Model


For the cost calculation for mortality, we are required to refer to the relative risks and the following is the Relative Risks (RRs) with 95% CI for every 50 ug/m3 increase in air pollutants (available from the report by HKU, 1998). The matrix is as follows (Table 4):

Table 4: Relative risks (RR) and 95% confidence interval (95% CI) for 50 ug/m3 increase in the concentration of air pollutants for hospital deaths of respiratory and cardiovascular diseases:
Diseases/Pollutants Respiratory Circulatory Respiratory & Circulatory
NO2 1.14 1.10 1.12
SO2 1.08 0.89 0.93
RSP 1.05 1.05 1.04
O3 1.22 1.15 1.07

Source: HKU (Jan., 1998)


Since we have estimated the VSL, the value can be substituted into the model below to value the economic cost of mortality.

E' = [ b'dj) * ( RR' (pi, dj)  -1 ) * VSL(dj) ]

E' = economic cost related to mortality 
b' = mortality rate due to a particular illness dj
RR' = Relative Risk for mortality due to (pi, dj)
pi = air pollutants (e.g. NO2, SO2, RSP, and O3)
dj = illness (e.g. respiratory or circulatory diseases),
VSL = estimated value of statistical life of one additional mortality due to dj
5.3.3 We shall input the VSL calculated (both the human capital and WTP estimates) to the model to calculate the economic cost of mortality for these two approaches.
5.3.4 Details of economic cost calculated using human capital estimates are shown in Appendix 9a and Appendix 9b (i.e., as separate-individual diseases and combined - respiratory and cardiovascular diseases) per 50 ug/m3 increase in the pollutants.
5.3.5 Economic cost of mortality calculated using WTP estimates is shown in Appendix 10a and Appendix 10b (same format as Appendix 9a and 9b).

The following presents a summary of the economic costs estimated with the relative risks per 50 ug/m3 increase in the air pollutants. Details of the calculations have already been presented as follows:

Appendix 4 and Appendix 5 - COI estimates and economic valuation for morbidity and calculations specifically on the Direct Cost component; Appendix 6 and Appendix 7 - WTP estimates and economic valuation for morbidity; Appendices 8, 9 and 10 - VSL using COI & WTP estimates and the economic valuation of mortality


The following is a summary of the economic cost of health impacts (see Table 5 and Table 6):


Table 5: Results of Economic Valuation Using COI and WTP Approaches:
Relative Risk
50 ug/m3

Respiratory Diseases
(HK$ million)

(HK$ million)
Combined of Respiratory and circulatory Diseases 
(HK$ million)
NO2 COI 522.20 451.90 852.34
D.C. 269.21 239.57 445.18
WTP 626.97 1,044.42 1,462.46
SO2 COI 195.82 282.44 487.05
D.C. 100.95 149.73 254.39
WTP 235.11 652.76 835.69
RSP COI 326.37 169.46 487.05
D.C. 168.26 89.84 254.39
WTP 391.85 391.66 835.69
O3 COI 652.75 395.41 852.34
D.C. 336.51 209.62 445.18
WTP 783.71 913.87 1,462.46

Note: * D.C. = Direct Cost


Table 6: Results of Economic Valuation Using COI and WTP Approaches:
Relative Risk
50 ug/m3

Respiratory Diseases
(HK$ million)

(HK$ million)
Combined of Respiratory and circulatory Diseases 
(HK$ million)
NO2 COI 1301.93 1,397.56 2,793.02
WTP 1805.84 1,868.12 3,789.61
SO2 COI 743.96 - -
WTP 1031.91 - -
RSP COI 464.98 698.78 931.01
WTP 644.94 934.06 1,263.20
O3 COI 2,045.89 2,096.34 1,629.26
WTP 2,837.75 2,802.19 2,210.60
Note: There is probably an inconsistent relationship between cardiovascular diseases/ combined respiratory and cardiovascular diseases and the pollutant SO2, thus the RR is less than one. Under such circumstances, it would not be appropriate to estimate the economic cost.
5.3.8 Details indicating the economic cost per ug/m3 increase in the pollution level is shown in Appendix 11. The results can be used to quantify the economic benefit of the reduction in air pollution.

6.  Health Effects using Composite Score of Pollutants

6.1 The composite score has also been applied to estimate the economic cost.

RR for composite score as shown in Table 7 will be used to estimate the economic cost in the following section.

Table 7 Relative Risks (with 95% CI for composite score of air pollutant) for 50 ug/m3 Increase in the Concentration of Air Pollutants for Hospital Admissions of Respiratory and Cardiovascular Diseases

Hosiptal Admissions
For all ages

Hospital Deaths
For all ages

Respiratory & Circulatory Diseases 1.06 1.07
Respiratory 1.08 1.09
Circulatory 1.04 1.05

Source: HKU (Mar., 1998)

Note: The composite was generated (see HKU, 1998, p.21) from the four pollutants considered elsewhere in this report, i.e., (NO2, SO2, RSP and O3)by principal components analysis and the first principal component was used which explained 68% of the variance with loading of 0.491, 0.105, 0.726 and 0.469 respectively).

Table 7 presents the RR is for all ages (see HKU Report, 1998). We use the same approach to estimate the economic cost for the composite score and the results are shown in Appendix 12a and 12b. The following (Table 8) presents a summary of the economic cost using the Composite Score as shown in Appendix 12a:

Table 8: Summary of Economic Cost Using Composite Score
Morbidity/Mortality by
COI/WTP estimate
Respiratory Diseases
(HK$ million)
Cardiovascular Diseases
(HK$ million)
Combined of Respiratory and circulatory Diseases
(HK$ million)
Using COI estimates  522.20 225.95 730.57
Using WTP estimates  626.97 522.21 1,253.54
Using COI estimates  836.96 698.78 1,629.26
Using WTP estimates  934.06 1,160.90 2,210.61

Appendix 12b presents the economic cost of morbidity and mortality calculated based on the composite score where pollution concentration is based on the concentration levels of the individual criterion air pollutants (i.e., NO2, SO2, RSP and O3) at the respective annual averages. Here, the composite score at 79.8 ug/m3 is obtained by the summation of the two factors (i.e., the change in the criterion air pollutants and the weighting factors.


[(52.3*0.491) + (18*0.015) + (54*0.726) + (27.7*0.469)]

= 79.8 ug/m3

7.  Results discussion


Results of Health Effects

7.11 In this study, we have basically used two key economic measurements to come up with the monetary values for the health effects. They both account for the morbidity and mortality effects by specific disease categories calculated by COI and WTP estimates.
7.1.2 We presented the economic valuation according to the two models used by HKU (Jan., 1998) in its study on the statistical relationship between ambient air pollution and its effects on human health. The first set of results is based on the single pollutant model and the second is that based on the composite score. In the latter, the expressed health impact of air pollutants took into account multiple pollutants individuals are exposed to in the real situation.

For the first (single pollutant model), the monetary value of the adverse health effects in terms of disease occurrence specifically combine respiratory and cardiovascular diseases for a whole year (estimated by 50ug/m3 increase in the pollutants concerned - namely NO2, SO2, RSP and O3) by COI estimates are as follows:

NO2 HK$852.34 million
SO2 HK$487.05 million
RSP HK$487.05 million
O3 HK$852.34 million

Whereas by using WTP approach, the same health effects by disease occurrence (by 50ug/m3 increase in the pollutants concerned) reveal a totally different set of economic cost as shown below:

NO2 HK$1,462.46 million
SO2 HK$835.69 million
RSP HK$835.69 million
O3 HK$1,462.46 million

In terms of premature death, the economic cost estimated by COI estimates by the pollutants concerned are as follows:

NO2 HK$2,793.02 million
SO2 -
RSP HK$ 931.01 million
O3 HK$ 1,629.26 million

No economic cost is indicated by the effect of SO2 is explained by the absence of any significant relation between cardiovascular diseases and respiratory and cardiovascular diseases. However, there exists a relation between SO2 and respiratory diseases and the economic cost of pre-mature death estimated by COI is HK$743.96 million.


The calculation of mortality by WTP estimates are as follows:

NO2 HK$ 3,789.61 million
SO2 -
RSP HK$ 1,263.20 million
O3 HK$ 2,210.60 million
7.17 We have explained why the absence of an economic cost on the effect of SO2 . However, for respiratory diseases resulted from SO2 , the economic cost of pre-mature death estimated by WTP is HK$1,031.91 million.

The total economic cost of the health effects in a year for combined respiratory and cardiovascular diseases by pollutants (i.e., the four criterion pollutants) using COI and WTP estimates for morbidity and mortality are as follows:

By COI estimate:
NO2 HK$3,645.36 million
SO2 HK$487.05 million
RSP HK$1,418.06 million
O3 HK$2,481.60 million
By WTP estimate:
NO2 HK$5,252.07 million
SO2 HK$835.69 million
RSP HK$2,098.89 million
O3 HK$3,673.06 million
7.1.9 The relatively low economic cost attributed to SO2 in both COI and WTP estimates is the result that its effect is only measured in respiratory diseases as presented by HKU study.
7.1.10 The results discussed so far are those derived from the relation determined by single pollutant model. As such, it would not be possible to sum the monetary values across the pollutants.
7.1.11 The gap between monetary values estimated by COI and WTP methods is mainly attributable to the factors considered in each of the approach. WTP estimates take into consideration non market values such as pain and suffering over and above direct health related expenditure and earning loss. Hence, WTP approach is expected to reveal a higher economic cost.

In terms of economic valuation based on composite score by COI and WTP estimates, the economic costs estimated for morbidity and mortality for combined respiratory and cardiovascular diseases are as follows:

By COI estimate:
Morbidity  HK$1,186.68 million
Mortality HK$2,654.04 million
Total  HK$3,840.72 million
By WTP estimate:
Morbidity  HK$2,036.14 million
Mortality HK$3,601.05 million
Total  HK$5,637.19 million
7.1.13 The calculations in this study may be further fine-tuned in the future when more specific data are made available. Nevertheless, the monetary values obtained provide indication to the health effect benefits which could be derived if we can cut back on air pollution.

Health Effect Expressed as Percentage of GDP

7.2.1 The economic costs of health effects as a result of air pollution are compared to Gross Domestic Product (GDP) which are important economic indicators. We use production-based GDP associated with the contribution of individual economic activities to overall GDP (Census and Statistics Department, 199711, 12). In 1996, Hong Kong's production-based GDP was HK$1,108,600 million (at current prices).
7.2.2 For reason that it would be inappropriate to add up the different economic costs across the board, no attempt is made to sum the economic cost in terms of GDP.

In the composite score approach which reflects the real situation better compared to the single pollutant approach, the results reveal a relatively higher percentage of GDP as shown below:

  By COI estimate: As % GDP
Morbidity and Mortality HK$3,840.72 million 0.35 %
  By WTP estimate: As % GDP
Morbidity and Mortality HK$5,637.19 million 0.51 %

Economic Cost Per Unit Increase in Criterion Pollutants


Comparison in monetary terms for each ug/m3 increase in the concentration level of each of the pollutants) is shown as follows:

By COI estimate:

  Per ug/ m3 increase in pollutants

Combined of Respiratory and Circulatory Diseases
(HK$ million)

Morbidity NO2 17.05
SO2 9.74
RSP 9.74
O3 17.05
Mortality NO2 55.86
SO2 -
RSP 18.62
O3 32.59

By WTP estimate:

  Per ug/m3 increase in pollutants

Combined of Respiratory and Circulatory Diseases
(HK$ million)

Morbidity NO2 29.25
SO2 16.71
RSP 16.71
O3 29.25
Mortality NO2 75.79
SO2 -
RSP 25.26
O3 44.21

Estimation of Economic Cost Based on the Available Unit Cost Value

7.4.1 The economic costs obtained in the single pollutant model is used to find the monetary value of the health effects for 1996 using air pollution data from the Environmental Protection Department (EPD, 1997)13.
7.4.2 Having established the unit increase in air pollution, we then multiply this set of results by the unit cost per ug/m3 increase for each pollutant. Details of the calculation are in Appendix 13.

Table 9 presents a summary of the predicted economic cost on combined respiratory and cardiovascular diseases for 1996, assuming that the health impacts elucidated in the previous studies are still valid.

Table 9: Summary of Predicted Health Impacts for 1996
By COI estimate: As % GDP 
NO2 HK$1,250.33 million 0.34
SO2 HK$137.45 million 0.02
RSP HK$388.25 million 0.13
O3 HK$248.12 million 0.12
By WTP estimate: As % GDP 
NO2 HK$5,493.66 million 0.50
SO2 HK$300.85 million 0.03
RSP HK$2,271.00million 0.20
O3 HK$2043.87 million 0.18

8.  Recommendation

8.1 This pilot study has concluded that costs attributable to health effects caused directly by air pollution account for 0.35% of Hong Kong's GDP, which is less significant than expected. This is only a preliminary study based on limited data but there is a definite identification of disease related air pollution episodes as well as a response from the public (through focus group study) that they do attach significant economic values to such disease avoidance.
8.2 We recommend that a more indepth study should be carried out with categorised statistics and data to produce a economic-health-air pollution model for Hong Kong. This model will allow decision makers to allocate appropriate resources to reduce the ill effects due to air pollution and to establish a comprehensive database for managing air pollution related health issues.

We suggest that it would be necessary to start collecting data to assist in the indepth study. These include:


Categorisation of number of general out-patient clinic visits by disease types

  • data for government clinics
  • data from private doctors

Direct medical costs data

  • doctor consultation fee
  • self-medication and any other related expenses
  • length of symptom days etc.
  • frequency of doctor consultation prior to hospital admission

Cost of health services available

  • health delivery cost including subsidies for each category of the diseases treated either as in-patient or out-patient

Data on earnings

  • personal income earnings which is representative of the general population, medical insurance etc.

Relative risk (RR) - statistical analysis data

  • report of actual number of "extra admissions" by age groups
  • to make sure that the sub-grouping of disease used for statistical analysis to establish the RRs for the various period (for each year) is consistent
  • to extend coverage of diseases - should not be limited to acute state of the illness but also cover chronic cases turn acute (specify in details the grouping of diseases)

9. Conclusions

From the pilot study based on one-year's data on air pollution, hospital admissions and out-patient doctor consultation related to respiratory and cardiovascular diseases, the cost of illness due to air pollution amounts to 0.35% of Hong Kong's GDP.

On the issue of willingness-to-pay, for mortality and morbidity based on a 29-person focus group study amounts to 0.51% of Hong Kong's GDP.


Appendix 1a: Cost Of Illness (COI) And Willingness-to-Pay (WTP)
Appendix 1b: Focus Group Survey
Appendix 1c: Focus Group Questionnaire Survey
Appendix 2: Summary Of Focus Group Survey Data
Appendix 3: Health Statistics - Department Of Health
Appendix 4a: COI : Respiratory Diseases
Appendix 4b: COI : Cardiovascular Diseases
Appendix 5a: Economic Valuation Using COI Estimates (Respiratory & Cardiovascular Diseases)
Appendix 5b: Economic Valuation Using COI Estimates (Combined Respiratory & Cardiovascular Diseases)
Appendix 5c(i): Economic Valuation Using COI Estimates: Direct Cost (Respiratory & Cardiovascular Diseases)
Appendix 5c(ii): Economic Valuation Using COI Estimates: Direct Cost (Combined Respiratory & Cardiovascular Diseases)
Appendix 5d: Compare Expenditure On The Medical And Health Services With Direct Expenditure Related To Hospital Admissions And Clinic Attendances
Appendix 6(i): WTP (Respiratory Diseases)
Appendix 6(ii): WTP (Cardiovascular Diseases)
Appendix 7: Economic Valuation Using WTP Estimates
Appendix 8a: Employment Earnings : By Stream Return
Appendix 8(i): VSL (COI) Deaths From Respiratory Diseases
Appendix 8(ii): VSL (COI) Deaths From Cardiovascular Diseases
Appendix 8b: WTP (Deaths From Respiratory Diseases)
Appendix 9a: Economic Valuation Using VSL (COI) Estimate (Mortality) - Respiratory & Cardiovascular Diseases
Appendix 9b: Economic Valuation Using VSL (COI) Estimate (Mortality) - Combined Respiratory & Cardiovascular Diseases
Appendix 10a: Economic Valuation Using VSL (WTP) Estimate (Mortality) - Respiratory & Cardiovascular Diseases
Appendix 10b: Economic Valuation Using VSL (WTP) Estimate (Mortality) - Combined Respiratory & Cardiovascular Diseases
Appendix 11: Results Of Economic Valuation Using COI And WTP Estimates
Appendix 12a: Economic Valuation Using Composite Score
Appendix 12b: Economic Valuation Using Composite Score At 79.8 ug/m3
Appendix 13: Economic Cost of Health Impact For Air Pollution, 1996



Back to top dot_clear.gif Table of Content