Analysis of residents' willingness to pay to reduce air pollution to improve children's health in community and hospital settings in Shanghai, China

Science of the Total Environment 533 (2015) 283–289

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Analysis of residents' willingness to pay to reduce air pollution to improve children's health in community and hospital settings in Shanghai, China Keran Wang a,1, Jinyi Wu a,1, Rui Wang a, Yingying Yang a, Renjie Chen a, Jay E. Maddock b, Yuanan Lu b,⁎ a b

School of Public Health, Fudan University, Shanghai, China Department of Public Health Sciences, University of Hawaii at Manoa, Honolulu, HI 96822, USA

H I G H L I G H T S • • • •

Contingent Valuation Method (CVM) was used to measure willingness to pay (WTP). Most respondents are willing to make a contribution to air quality improvement. Respondents from hospital settings are willing to pay more for improved air quality. This study concludes both education and income are important predictors of WTP.

a r t i c l e

i n f o

Article history: Received 10 September 2014 Received in revised form 28 January 2015 Accepted 28 June 2015 Available online 11 July 2015 Editor: D. Barcelo Keywords: Willingness to pay Child health Air pollution Shanghai

a b s t r a c t Background: Shanghai, along with many major cities in China, faces deterioration of air quality and increases in air pollution-related respiratory diseases (RDs) in children due to rapid industrialization and urbanization. The Contingent Valuation Method (CVM) was used to qualitatively and quantitatively measure the willingness to pay (WTP) for reducing children's RDs through air quality improvement. Methods: Between April and May, 2014, 975 face-to-face interviews were collected from parents in a community-based and a hospital-setting in Shanghai. Multiple imputation and the Probit model were used to determine the relationship between the WTP and the related environmental factors, child health factors and the socio-economic status. Results: Most respondents reported being willing to make a financial contribution to improve air quality in both the community (52.6%) and hospital (70.2%) samples. Those in the hospital setting were willing to pay significantly more ¥504 (USD$80.7) compared to the community sample ¥428 ($68.5) as expected. Reasons for those not being willing to pay included lack of disposable income and believing that responsibility of the air quality was a community issue. These did not differ by sample. Annual household income and education were related to WTP. Conclusion: This study indicated that parents in Shanghai would be willing to pay for improved air quality. Children's health can be the incentive for the citizens' participation and support in the air quality improvement, therefore, hospital settings may present unique places to improve education about air quality and enhance advocacy efforts. This study also suggested that future environmental policies be addressed more rigorously for targeted populations. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Air pollution has become a persistent concern among large cities in developing countries over the past several decades (Li et al., 2008; Ma et al., 2013). Shanghai, as one of the largest and most developed cities in China, has been aware of the issue of air pollution for many years. According to the Shanghai Environment Bulletin in 2013, there was a total of 124 days in which the Air Quality Index (AQI) exceeded 100. Based on ⁎ Corresponding author. E-mail address: [email protected] (Y. Lu). 1 These authors contributed equally to this manuscript.

http://dx.doi.org/10.1016/j.scitotenv.2015.06.140 0048-9697/© 2015 Elsevier B.V. All rights reserved.

a study of respiratory diseases among children in Shanghai, which includes conditions of the upper respiratory tract, trachea, bronchi, bronchioles, alveoli, pleura and pleural cavity the prevalence of respiratory illness increased from 1.5% in 1988 to 3.3% in 2000, and 4.5% in 2009. These results may be attributed to increased air pollution and lack of exercise (Niu et al., 2010; Voorhees et al., 2014). A recent study estimated that 346,000 respiratory hospital admissions could be avoided if China meets class 2 air quality standards in urban areas (Shang et al., 2012). The deterioration of air quality appears to be a major risk factor leading to children's RDs. The Contingent Valuation Method (CVM) is used widely in environmental health to assess the amount that people are willing to pay for

284

K. Wang et al. / Science of the Total Environment 533 (2015) 283–289

environmental services (Fox, 2013; Kairu-Wanyoike et al., 2014; Tambor et al., 2014; Wang and Song, 2009). CVM is based on constructing a hypothetical scenario, so that one can investigate how people react to the changes in environmental quality and how much they may want to pay for that. In this way, air quality, which is an intangible good but does not have a market price, can have a potential monetary value (Marella and Raga, 2014). Instead of direct market methods, CVM can solve practical problems in environmental economic analysis. This method has three formats, including open-ended, biding games, and dichotomous choices. Dichotomous choices can reflect the more authentic thoughts of the respondents' willing to pay (WTP) than the other two methods (Cai et al., 2007; Wang and Mullahy, 2006). With air pollution and economic growth serving as opposing forces in Shanghai, the use of CVM can improve the understanding of the amount of money the city could potentially raise to combat the growing problem of air quality degradation. In this study, we assessed WTP among caregivers with children in a community and a hospital respectively. Using CVM, we compared the WTP in these two areas to assess if there were any differences. We hypothesized that parents with sick children in the hospital setting would have a higher WTP compared with parents with healthy children in community. The findings from this research may help provide some practical advice and ideas for the government in their approach to improve air quality in Shanghai and other cities in developing nations. 2. Materials and methods 2.1. Study site Shanghai is a rapidly developing city in China as well as one of the most polluted cities (Cai et al., 2014; Yang et al., 2002). To get a

preliminary understanding of residents' willingness to pay (WTP) to reduce air pollution in Shanghai using contingent valuation method (CVM), we conducted a survey in two city districts (Fig. 1). Shanghai is located in the eastern part of China and has a population of over 25 million. According to Shanghai Environment Bulletin in 2013, there were 124 days were the air quality was in the unhealthy range (AQI N 100) including 23 days where air quality was unsafe for all residents (Shanghai Environmental Protection Bureau, 2013; Yang et al., 2014). 2.2. Sampling and data collection Between April and May, 2014, surveys were collected in a community setting in the Jiading district and in a hospital in the Minhang district (Fig. 1), both of which are rapidly developing, formally rural districts in Shanghai. Pretest surveys (n = 30) were conducted to evaluate the comprehension of the target population and revealed that the participants could understand all the questions, thus, the original pretest questionnaire was retained for this study. Clustering sampling using a randomized number method was adopted to choose the survey districts in this study. Based on previous air quality studies conducted in China, a sample size of 1000 was selected, with 500 in the community sample and 500 in the hospital setting (Zhang et al., 2014). Adults aged between 18 and 65 years old who have at least one child aged between 1 and 12 years old were included in the study. If there is more than one child in a family who meet our inclusion criteria, all of the children could be assumed as study samples. Each questionnaire was collected on face-to-face basis by trained interviewers, who described the meaning of each question and available choices to participants in order to avoid the response bias. Our survey was conducted from 9 A.M. to 4 P.M. each day for a total of 45 days.

Fig. 1. Geographical location of study area in China. Shanghai comprises 16 urban areas. A community in the Jiading district and a hospital in the Minhang district have been covered in this study, both of which are rapidly developing rural districts in Shanghai.

K. Wang et al. / Science of the Total Environment 533 (2015) 283–289

2.3. Questionnaire In order to assess WTP, demographics, social-economic variables, subjective perception of local air quality, belief in a relationship between children's RDs and air pollution and contingent valuation were assessed (Zhang et al., 2014). Respondents were asked about signs and symptoms of their children which were not life-threating and needed to seek treatment in the hospital. They were also asked to provide their home address at the beginning of the questionnaire in order to avoid repeat participation. In the design of our study, several measurements were considered to make sure that confounders are neutralized, such as constituting inclusion criteria, training interviewers, synchronizing the survey time of different population, and using the multivariate statistic method. The first part of the survey contains demographic questions, including parents' education, children and parents' age and their gender. The second section concerns family socio-economic status, such as annual household income, family size, and expenditures towards air pollution related illnesses in their children. The third section is about respondents' perception of the current air pollution level in their residence area, including subjective air quality, air pollution influence on children, and relationship between air pollution and health questions. The final section contained the contingent valuation survey, including willingness to pay and the amount of the payment. The core questions were as follows: in order to keep children from RDs, a series of protective measurements need to be implemented, which incur cost paid by residents. Considering respondents' household income and expenditure on RDs, (1) are you willing to pay? and (2) how much are you willing to pay to achieve this goal? (Note: no money is actually paid from participants in this hypothetical question study).

285

method endeavors to keep the advantages of ML estimates with permission of uncertainty due to imputation. Uncertainty is neglected in single imputation, and it should be considered in the completed-data analysis. The MI process involves creating more than one set of replacements for the missing values based on plausible models for data, and then producing multiple completed datasets for analysis (Butler and Robert, 1982; He, 2010). The statistical reasoning behind multiple imputations is that the observed-data likelihood can be estimated by the average of the completed-data likelihood over unknown missing values. Therefore multiple imputation analysis that combines the likelihood-based analysis from each completed dataset equals to the analysis based on the observed-data likelihood, while the imputation uncertainty is reflected by the variation across the multiple completed datasets. 3.2. Probit model Probit model is a classic model, which belongs to the Generalized Linear Model (Russo et al., 2014; Ye et al., 2014). With this model, the relationship between probability of an incident and influencing variables is assessed. The Probit model can be defined as the following form: Prob ðWTPN0Þ ¼ f ðAge; Sex; Education; Income; …Þ Positive WTP is the dependent variable, which is a binary factor. Age, sex, education, and income are considered to be the independent variables. 4. Results

2.4. Statistical analysis

4.1. Descriptions of variables

Using Epidata 3.1 (EpiData Association, USA, http://epidata.dk), questionnaires were imported into the computer, Excel 2010 (Microsoft Ltd, USA, http://www.microsoftstore.com/store/msusa/en_US/html/ pbpage.OfficeCompare) was applied to manage data and Statistical package IBM SPSS 19.0 (IBM Corporation USA, http://www-01.ibm. com/software/analytics/spss/) was used to analyze the data. Since the rate of missing values varied widely (e.g. age of parents had about 30% missing data), multiple imputations were used to fill in missing values. After filling the data, all the variables met the requirements had less than 10% missing data. There were four parts of the analysis. First, initial descriptive analysis for both the community and hospital samples was conducted. Then, Chi-square tests were used to examine the difference in WTP and the variation of the amount of WTP between the community and hospital groups. Next, we used chi-square test to analyze the univariate factors in the two samples. Finally, the Probit model was used to identify variables that affected the respondents' decision on WTP. In order to find out the different questions influence on WTP, the variables were entered into the model in three steps. Demographical and social economic status were entered in the first step, attitudes towards air pollution were entered in the second one, and the relationship between children health and air quality was entered last.

4.1.1. Descriptions of independent variables In our study, 495 (99%) of 500 questionnaires collected from the community setting were retrieved during the survey period, while 479 (96%) of 500 interviewees were completed. Table 1 presents the independent variables of the respondents in both settings. The independent variables included interviewees' and their children's demographic characteristics, their socio-economic status, their perception of local air quality, and the relationship between the children RDs and the air pollution. The ChiSquare test between child's current health state in the community and hospital was statistically significant (χ2 = 136.877, P b 0.001). The expenditures on the children's RDs between the two groups were also significantly different (χ2 = 122.613, P b 0.001), indicating a higher WTP for the hospital group than that of community group.

3. Theory 3.1. Multiple imputation A variety of weighting and imputation procedures are recommended for contingent valuation to help correct for the inevitable deviations from the desired sample and there are statistical methods to help correct for sample selection bias (Carson and Richard, 1999). Under the Missing at Random (MAR) situation, the multiple imputation (MI)

4.1.2. Descriptions of dependent variables In the community sample, 242 (50.5%) of the 479 valid participants reported being willing to pay for air quality improvement. Among the 242 group, only 202 responded to the question regarding the amount of WTP. The percentages of the amount of WTP for ￥1–200 ($32), ￥200–1000 ($160), ￥1000–5000 ($800), and more than ￥5000 accounted for 64.9%, 29.7%, 4.5%, and 1.0%, respectively (Table 2). A little less than a half of the sample population (47.4%) expressed their unwillingness to pay for air quality improvement. Lack of disposable income (23.2%) was the most common reason for not being willing to pay. The second reason for unwillingness to pay was the belief that improving air quality is the government's responsibility. Only 3.5% of participants thought that the air quality was good enough and did not need to be improved (Table 3). In the hospital setting, of 433 valid responses, 304 (70.2%) respondents were willing to pay for air improvement, which is significantly higher (P b 0.05) than in the community (50.5%). While over a third of respondents (33.7%) were willing to pay ￥200–1000 for improving air quality, around 59.2% respondents would like to pay less than

286

K. Wang et al. / Science of the Total Environment 533 (2015) 283–289

Table 1 Descriptions of the independent variables. Metric variables

Age 1 Age 2 Family size Categorical variables

1

Gender 1

2

Gender 2

Education

Average annual household income (Unit:￥103)

Expenditure3 (Unit: ￥)

Pollution4

Health state5

Correlation6

Influence7

Description

Respondents' current age Children's current age Number of family size Description

Male Female Male Female Below primary school Junior middle school Senior middle school or technical middle school College University Graduate school b20 20–50 50–100 110–200 N200 b100 100–500 501–1000 1001–2000 2001–5000 N5000 Perfect Relatively good Fair Relatively polluted Severely polluted Healthy Relatively healthy Fair Relatively poor Poor Closely related Generally related Little related No related Greatly influenced Generally influenced Little influenced No influence

Community

Hospital

Mean

S.D.

Mean

S.D.

34.17 8.26 4.29

6.773 2.397 1.249

32.18 5.41 4.13

7.195 2.793 1.233

Community

Hospital

No.

Percentage⁎(%)

No.

Percentage⁎(%)

249 211 254 227 65 199 87

54.1 45.9 52.8 47.2 14.4 44.2 19.3

176 230 255 199 12 50 58

43.3 56.7 56.2 43.8 3.2 13.2 15.3

45 45 9 0 5 133 109 150 66 154 114 62 49 21 52 43 41 161 162 98 173 166 32 4 160 161 106 15 169 172 128 12

10.0 10.0 2.0 0 1.3 33.5 27.5 37.8 14.2 33.0 24.5 13.3 10.5 4.5 10.6 8.7 8.3 32.7 32.9 20.7 36.6 35.1 6.8 0.8 36.2 36.4 24.0 3.4 35.1 35.8 26.6 2.5

98 137 24 1 5 136 141 145 77 140 114 58 37 26 21 69 55 162 135 106 161 164 22 3 164 120 132 19 35 162 202 49

25.9 36.1 6.3 0.2 1.2 31.8 32.9 33.9 17.0 31.0 25.2 12.8 8.2 5.8 4.5 14.6 11.7 34.4 28.7 23.2 35.3 36.0 4.8 0.7 37.7 27.6 30.3 4.4 7.8 36.2 45.1 10.9

1

Respondents' gender. Children's gender. 3 Expenditure on kids' RDs. 4 Self-reported air quality. 5 Kids' current health status. 6 Self-reported air pollution and health having correlation. 7 Self-reported air pollution influencing respiratory system. ⁎ Valid percentage. 2

￥200 for air quality improvement (Table 2). A small portion of respondents provided negative answer to this question since they believed that improvement of air quality is the responsibility of the government. Lack of disposable income was another important reason of negative response (Table 3). 4.2. Difference of WTP between the community and the hospital The WTP (yes/no) between respondents in community and hospital setting shows a statistically significant difference (χ2 = 36.682, P b 0.001; OR = 0.433), with the responders from the hospital being more willing to pay money to improve air quality. However, the result shows no statistical significance (χ2 = 2.821, P = 0.615, OR = 0.433) regarding the amount of WTP between the two settings.

4.3. Univariate determinants of the positive WTP The Chi–square test was used separately to check the distribution of independent variables for WTP. In the community, WTP was related with respondents' education (χ2 = 13.790, P = 0.015) and their perceived level of air pollution (χ2 = 16.095, P = 0.006) (Table 4). Significant differences were not present in the hospital sample (Table 4).

4.4. Multivariate determinants of the positive WTP The generalized linear model was used to determine the combined influence of the independent variables. The Probit model is utilized separately in the community and hospital samples, and the combined

K. Wang et al. / Science of the Total Environment 533 (2015) 283–289

287

Table 2 Comparison of amount of WTP (per year) for air pollution improvement between community and hospital groups⁎. WTP (￥)

1–200 200–999 1000–4999 ≥5000

Community

Hospital

Total

Frequency

Percentage (%)

Frequency

Percentage (%)

Frequency

Percentage (%)

131 60 9 2

64.9 29.7 4.5 1.0

158 90 15 4

59.2 33.7 5.6 1.5

289 150 24 6

61.6 32.0 5.1 1.3

⁎ Does not include 0.

model is also executed. The detailed parameters are presented in Tables 5, 6 and 7. 4.4.1. Multivariate determinants of the positive WTP in community In the first model of the community (Table 5), only respondents' and their children's demographic characteristics and their social-economic factors were included. Education was shown to have a positive relationship with the probability of positive WTP (P b 0.05). In the second model, health status and perceived recognition of pollution were added into the model in addition of expenditure, education was the only variable that had a statistically significant correlation with the positive WTP. Perceived recognition of the relationship between the children's RDs and air pollution besides the effect of the air quality has been put into the third model, and the respondents' gender shows negative relationship with the probability of the positive WTP. 4.4.2. Multivariate determinants of the positive WTP in hospital The Probit model of the hospital sample (Table 6) shows different significant determinants compared to the community model. In the second and the third models, the respondents' perceived views on current air pollution had a negative relationship with WTP. 4.4.3. Multi-variable-based determinants of the positive WTP in total The Probit model was also used to assess the combined samples. At the first step, education and annual household income had a positive relationship with the probability of WTP. In the second model and third models, annual household income had strong positive correlation with the positive WTP. 5. Discussion Air pollution has become an increasingly important governmental issue in Shanghai. Several medical studies have demonstrated the link between children's RDs and air pollution (Ma et al., 2013; Niu et al., 2010). However, prior to this study little was known about residents, especially parents' attitudes towards air pollution and their children's health. Meanwhile, a couple of recent studies from Nanchang, China have shown that nearly all (98%) residents think improving air pollution is responsibility of every resident, which indicates that residents in China are willing to cooperate with and support governmental actions Table 3 Reasons for the respondents' negative WTP in community⁎. Reasons for rejection

Income is too low to afford it. The current air quality is good enough. Related expense is the accountability of government, but not the residents. Related expense is the accountability of pollution department, not the residents. This issue is impossible Not interested in WTP related research ⁎ These are from multiple choices question.

Community

Hospital

Percentage (%)

Percentage (%)

23.2 3.5 18.9

31.3 1.6 55.7

18.6

56.2

3.7 5.1

7.1 0.8

to alleviate air pollution and improve air quality (Lee et al., 2014; Zhang et al., 2014). Community support and WTP to improve air quality can inform policymakers about changes to current governmental and taxation policies. Due to the hypothetical approach of CVM, it is necessary to conduct the test on the validity of related studies (Venkatachalam, 2004; Wang and Song, 2009; Wang et al., 2008). Validity was generally high in this study. The pretest showed a high level of face validity and comprehension in the target population. The response rate in both sample groups exceeded 90%, indicating a minimal level of response bias in the community. One of the ways to test the validity of WTP is to compare the results with other study's outcomes (Wang and Mao, 2006). Due to the designed question format, the specific amount of WTP has been calculated by average value of each choice and the corresponding weight. In this study, the highest choice (equal or more than ￥5000) was considered as the “average”. Amounts higher than ￥5000 were not allowed. The average amount of WTP in total was ￥471.6 ($75.5) per year, which is consistent with recent study in Qingdao, China (Wang et al., 2008). The significant relationship between WTP and income also can be used to test the validity (Johannesson and Jonsson, 1991). All of the three Probit models showed that the probability of positive response of WTP had a statistically significant relationship with the annual household income, which also can prove the validity of this study. There are some CVM related studies on children's health, such as in the field of child maltreatment death (Corso et al., 2013), childhood asthma (Brandt et al., 2012), the child safety seats (Jarahi et al., 2011), and childhood diarrhea (Akhter and Larson, 2010). Like the studies above, this study focuses on the WTP for the reduction of air pollution to reduce children's RDs. In order to check the difference between the parent's WTP of the healthy and relatively unhealthy children, two different populations were assessed in our study. As expected, children in the community sample tended to be healthier than that in hospital group.

Table 4 Chi square test between the WTP and the dependent variables in Shanghai. Factors

Gender 1 Gender 2 Education Average annual household income Family size Expenditure Pollution Health state correlation Influence

WTP (yes/no) in community

WTP (yes/no) in hospital

χ2

P

χ2

P

1.576 0.780 13.790a 7.940

0.209 0.377 0.015⁎ 0.094

0.123 0.005 3.392b 6.679

0.725 0.943 0.641 0.154

4.877a 9.229 16.095 2.839a 1.938 1.108

0.807 0.102 0.006⁎ 0.588 0.585 0.777

5.776 3.406 8.576b 8.756b 4.190b 1.294b

0.569 0.638 0.116 0.061 0.229 0.759

a In the community: education/family size/health state: use the fisher exact test, the additional use the Chi square test. b In the hospital: education/pollution/ health state/ correlation/ influence: use the fisher exact test, the additional use the Chi square test. ⁎ P b 0.05.

288

K. Wang et al. / Science of the Total Environment 533 (2015) 283–289

Table 5 Probit model for the probability of positive WTP in community, N = 377. Variable

Age 1 Age 2 Gender 1 Gender 2 Education Annual household income Family size Expenditure Health state Pollution Correlation Influence

Model 1

Model 2

Table 7 Probit model for the probability of WTP in Total, N = 679.

Model 3

Variable

B

S.E.

B

S.E.

B

S.E.

0.012 −0.011 −0.190 0.083 0.119⁎ 0.110

0.0104 0.0354 0.1366 0.1343 0.0613 0.0694

0.012 −0.000 −0.225 0.034 0.106⁎ 0.098

0.0107 0.0376 0.1419 0.1395 0.0641 0.0718

0.015 0.001 −0.252⁎ 0.078 0.066 0.093

0.0109 0.0392 0.1463 0.1442 0.0672 0.0737

0.0546

0.048 0.058 −0.017 0.043

0.0563 0.0605 0.0856 0.0504

0.046 0.062 −0.036 0.028 0.092 −0.089

0.0572 0.0629 0.0891 0.0518 0.0820 0.0935

0.043

⁎ P b 0.05.

Table 6 Probit model for the probability of positive WTP in hospital, N = 302.

Age 1 Age 2 Gender 1 Gender 2 Education Annual household income Family size Expenditure Health state Pollution Correlation Influence ⁎ P b 0.05.

Model 2

Model 3

S.E.

B

S.E.

B

S.E.

0.010 −0.036 −0.039 0.033 0.105⁎ 0.120⁎

0.0076 0.0201 0.1019 0.1016 0.446 0.0503

0.008 −0.033 −0.047 0.003 0.089 0.107⁎

0.0079 0.0217 0.1054 0.1049 0.0478 0.0519

0.009 −0.034 −0.054 0.015 0.078 0.110⁎

0.0080 0.0223 0.1079 0.1071 0.0492 0.0528

0.0427 0.0441 0.0659 0.0414

0.050 0.069 0.009 −0.013 0.072 −0.112

0.0432 0.0459 0.0658 0.0425 0.0658 0.0808

0.041

0.0411

0.046 0.066 0.024 0.001

⁎ P b 0.05.

As we hypothesized, participants in the hospital sample had a significantly higher rate of WTP than those in the community, which verified that the parents of relatively unhealthy children have more desire to improve the air quality and more of them would like to pay the money for air pollution control. The amount of WTP offered by the parents of relatively unhealthy children is higher than that of the other ones. Using the approach described above, people would like to pay an average of ￥428.1 ($68.5) in the community samples, while the respondents in the hospital averaged ￥504.4 ($80.7) to pay for the air quality improvement. This amount is larger than a previous study by Peng and Wang (2003). This indicates that hospital setting may be an effective place for education and advocacy around cleaner air. The reasons for not being WTP also varied a lot, and can provide information on how to establish policies on air quality management. The most common reasons in both community and hospital groups are low income and the recognition of the responsibility for pollution. Lack of income is one of the reasons that the WTP is lower than the actual value of the air improvement, especially in developing countries (Timothy et al., 1999; Yeung and Dong, 2005; Zhang et al., 2012). A recent study indicated the Gini coefficient in China in 2010 was between 0.53 and 0.55 making it one of the highest in the world (Xie and Zhou, 2014). This indicates a large gap between the rich and poor in China. The sixth census of Shanghai in 2010 stated that the floating (transient) population accounts for 39% the total population of Shanghai, indicating Shanghai as one of the top cities with a large number of floating people in China. Shanghai with its large floating population could have an even larger Gini coefficient making across the board taxes or fees very difficult to implement. The latter assumption suggests that it requires more public education on environmental health issues with the local population.

Variable

Age 1 Age 2 Gender 1 Gender 2 Education Annual household income Family size Expenditure Health state Pollution Correlation Influence

Model 1 B

Model 1

Model 2

Model 3

B

S.E.

B

S.E.

B

S.E.

0.009 −0.008 0.128 0.010 0.002 0.112

0.0116 0.0297 0.1608 0.1622 0.0719 0.0768

0.001 −0.007 0.187 0.031 0.012 0.107

0.0120 0.0314 0.1680 0.1684 0.0784 0.0799

0.004 −0.004 0.184 0.008 0.014 0.109

0.0122 0.0319 0.1725 0.1712 0.0797 0.0810

0.017

0.0656

0.039 0.019 −0.007 −0.145⁎

0.0694 0.0705 0.1132 0.0816

0.051 0.024 −0.008 −0.148⁎ 0.056 0.064

0.0701 0.0731 0.1167 0.0829 0.1176 0.1918

Education and the annual household income are the two factors, which influence respondents' positive response of WTP. This results is consistent with related studies carried in Jinan, Chongqing Beijing and Qingdao (Wang et al., 2006), which means that residents in Shanghai who had a higher education and more income tend to have better awareness of air pollution and the related accountability, and simultaneously, higher income gives people the ability to pay for environmental issues. Within the total sample, the majority (69.6%) of the participants who have a bachelor degree or higher were WTP, while the percentage of the respondents who were WTP but did not have bachelor degree were only 57.6%. However, the variables effecting WTP in the two samples were different from each other. In the community, education and being male were related to WTP. These results are consistent with studies in Qingdao. In the hospital, the perceived level of air pollution was significantly related to WTP, which indicates that residents who are not satisfied with air quality are more willing to make a change. The difference in predictors between the two setting shows that different populations may be influenced by different factors to some extent, while education and income are the general factors which both influence the WTP response. To parents of relatively unhealthy children, issues affecting their child's health may be more salient, even if they are being treated for non-RD related issues. While other CVM studies in China have focused on the air improvements or the natural landscapes (Wang et al., 2006; Wang et al., 2003), this study has more specific objectives and pays specific attention to children's health. The two samples assessed in this study also show the variation between different populations, which may suggest the different policies or approaches to working with these populations. Due to lack of recent studies in CVM in Shanghai, the amount of WTP has not been assessed to figure out whether the WTP on the children issue will impact the participants' response of WTP on air quality. Further studies on this issue are needed to provide more evidences on the environment policy. 6. Conclusions This study showed a high level of willingness to pay to improve air quality in China to protect children's health. Respondents in the hospital setting were more willing to pay and pay more than those in a community setting. Education and income were important predictors of WTP. This study extended previous research in China by focusing specifically on children's health. Results indicate that it may be necessary to carry out different strategies or policies with different populations, which can lead to a more influential change, not only in the air quality issue but towards residents' attitudes as well. Similar studies on WTP for air improvement on children's health in China have not been conducted before, and further studies are needed to confirm our conclusions.

K. Wang et al. / Science of the Total Environment 533 (2015) 283–289

Acknowledgments Authors would like to thank Dr. Yingyao Chen, Professor of School of Public Health of Fudan University and Thomas Lee of Office for Public Health Studies of University of Hawaii at Manoa, for their coordination and assistance in this study; and this study was supported in part from two scholarship awards to Dr. Yuanan Lu — Fudan Visiting Scholarship Award and Nanchang Gainjiang Scholarship Award. References Akhter, S., Larson, C.P., 2010. Willingness to pay for zinc treatment of childhood diarrhoea in a rural population of Bangladesh. Health Policy Plan. 25, 230–236. Brandt, S., Vasquez, L., Hanemann, S., 2012. Contingent valuation scenarios for chronic illnesses: the case of childhood asthma. Value Health 15, 1077–1083. Butler, J.S., Robert, M., 1982. A computationally efficient quadrature procedure for the one-factor multinomial probit model. Econometric 1999, 761–764. Cai, C.G., Chen, G., Qiao, X., 2007. Application of contingent valuation method to estimate health economic loss caused by environmental pollution. J. Environ. Health 4, 265–267 (in Chinese). Cai, J., Zhao, A., Zhao, J., Chen, R., Wang, W., Ha, S., Xu, X., Kan, H., 2014. Acute effects of air pollution on asthma hospitalization in Shanghai, China. Environ. Pollut. 191, 139–144. Carson, R.T., Richard, T., 1999. Contingent Valuation: A User's Guide. University of California, San Diego. Corso, P.S., Ingels, J., Roldos, B., 2013. A comparison of willingness to pay to prevent child maltreatment deaths in Ecuador and the United States. Int. J. Environ. Res. Public Health 10, 1342–1355. Fox, M.S., 2013. A Foundation Ontology for Global City Indicators. University of Toronto, Toronto, Global Cities Institute. He, Y., 2010. Missing data analysis using multiple imputation: getting to the heart of the matter. Circ. Cardiovasc. Qual. Outcomes 3 (1), 98–105. Jarahi, L., Karbakhsh, L., Rashidian, M., 2011. Parental willingness to pay for child safety seats in Mashad, Iran. BMC Public Health 11, 281. Johannesson, M., Jonsson, B., 1991. Willingness to pay for antihypertensive therapy— results of a Swedish pilot study. J. Health Econ. 10, 461–473. Kairu-Wanyoike, S.W., Kaitibie, S., Heffernan, C., Taylor, N.M., Gitau, G.K., Kiara, H., McKeever, D., 2014. Willingness to pay for contagious bovine pleuropneumonia vaccination in Narok South District of Kenya. Prev. Vet. Med. 115, 130–142. Lee, T.H., Maddock, J.E., Zhang, P., Jiang, Z.Q., Zhang, L.J., Li, D.M., Zhou, X.J., Yuan, Z.K., Lu, Y., 2014. Caretakers' perceptions of air quality and its effects on their children in Nanchang, China. Glob. Health J. 1 (2), 37–43. Li, G.L., Zhou, M., Chen, C.H., Wang, H.L., Wang, Q., Lou, S.R., Qiao, L.P., Tang, X.B., Li, L., Huang, H.Y., Chen, M.H., Luo, H., Su, D., Ding, D., 2008. Assessment of willingness to pay for potential meteorological risk resources. Meteorol. Mon. 34, 79–83 (in Chinese). Ma, J., Chen, Z., Wu, M., Feng, J., Horii, Y., Ohura, T., Kannan, K., 2013. Airborne PM2.5/ PM10-associated chlorinated polycyclic aromatic hydrocarbons and their parent compounds in a suburban area in Shanghai, China. Environ. Sci. Technol. 16, 7615–7623 (in Chinese). Marella, G., Raga, R., 2014. Use of the contingent valuation method in the assessment of a landfill mining project. Waste Manag. 34, 1199–1205. Niu, C.J., Wang, J., Zhao, Z., 2010. Prevalence of respiratory symptoms and diseases among children and adolescent in urban Shanghai. Chin. J. Sci. Health 6, 708–710 (in Chinese).

289

Peng, X.Z., Wang, T., 2003. The payment of the economic loss of the diseases caused by air pollution in Shanghai. World Econ. Forum 12, 32–44 (in Chinese). Russo, B.J., Kay, J.J., Savolainen, P.T., Gates, T.J., 2014. Assessing characteristics related to the use of seatbelts and cell phones by drivers: application of a bivariate probit model. J. Saf. Res. 49, 137–142. Shang, Z., Chen, Y., Yang, K., 2012. Assessing local communities' willingness to pay for river network protection: a contingent valuation study of Shanghai, China. Int. J. Environ. Res. Public Health 9, 3866–3882. Shanghai Environmental Protection Bureau, 2013. Shanghai environment bulletin. http://www.sepb.gov.cn/fa/cms/shhj/shhj2272/shhj2159/2014/06/86604.htm. Tambor, M., Pavlova, M., Rechel, B., Golinowska, S., Sowada, C., Groot, W., 2014. Willingness to pay for publicly financed health care services in Central and Eastern Europe: evidence from six countries based on a contingent valuation method. Soc. Sci. Med. 116, 193–201. Timothy, S., Brett, D., Susana, M., 1999. Valuing water quality in China: purpose, approach and policy. J. Environ. Health 11, 309–315. Venkatachalam, L., 2004. The contingent valuation method: a review. Environ. Impact Assess. Rev. 24, 89–124. Voorhees, A.S., Wang, J.D., Wang, C.C., Kan, H.D., 2014. Public health benefits of reducing air pollution in Shanghai: a proof-of-concept methodology with application to BenMAP. Sci. Total Environ. 485, 396–405. Wang, H., Mao, J., 2006. The economic value of air-pollution-related health risks in China: a contingent valuation study. Sci. Total Environ. 367, 50–57. Wang, H., Mullahy, J., 2006. Willingness to pay for reducing fatal risk by improving air quality: a contingent valuation study in Chongqing, China. Sci. Total Environ. 367, 50–57. Wang, Y., Song, Z.Y., 2009. Air quality assessment by contingent valuation in Ji'nan, China. J. Environ. Manag. 90, 1022–1029. Wang, S.B., Wang, P.J., Hu, Z.P., 2003. Assessing the landscape services value with contingent valuation method: a case study on Suzhou Brook, Shanghai City. J. Fudan Nat. Sci. 42, 463–467 (in Chinese). Wang, X.J., Zhang, W., Li, Y., Yang, K.Z., Bai, M., 2006. Air quality improvement estimation and assessment using contingent valuation method, a case study in Beijing. Environ. Monit. Assess. 120, 153–168. Wang, Y.Y., Zhang, B., Liu, B., 2008. Analysis of residents' willingness to pay for air pollution related diseases in Qingdao. J. Environ. Health 4, 326–330 (in Chinese). Xie, Y., Zhou, X., 2014. Income inequality in today's China. Proc. Natl. Acad. Sci. 111, 6928–6933. Yang, K.Z., Meng, B., Yang, L., 2002. The feasibility of contingent valuation method in economic valuation of environment in China: a case study of the residents' willingness to pay in Beijing. Adv. Earth Sci. 17, 420–425 (in Chinese). Yang, N., Damgaard, A., Lu, F., Shao, L.M., Brogaard, L.K., He, P.J., 2014. Environmental impact assessment on the construction and operation of municipal solid waste sanitary landfills in developing countries: China case study. Waste Manag. 34, 929–937. Ye, C.Y., Chen, L.J., Ku, P.W., Chen, C.M., 2014. Modelling the vicious circle between obesity and physical activity in children and adolescents using a bivariate probit model with endogenous regressors. J. Biosoc. Sci. 15, 1–14. Yeung, R.Y., Dong, S.R., 2005. Can we use contingent valuation to assess the demand for childhood immunisation in developing countries: a systematic review of the literature. Appl. Health Econ. Health Policy 4, 165–173. Zhang, Y.F., Zhang, L., Zhou, J., 2012. Comparison study on WTP for river environmental restoration of urban river in Shanghai: case studies in 2006 and 2010. Northwest Popul. 12, 41–45 (in Chinese). Zhang, L.J., Yuan, Z.K., Maddock, J.E., Zhang, P., Jiang, Z.Q., Lee, T., Zou, J.J., Lu, Y.A., 2014. Air quality and environmental protection concerns among residents in Nanchang, China. Air Qual. Atmos. Health 12, 100–108.