The monetary valuation of acute respiratory illness from air pollution in Turkey

Atmospheric Pollution Research xxx (2015) 1e10

Contents lists available at ScienceDirect

H O S T E D BY

Atmospheric Pollution Research journal homepage: http://www.journals.elsevier.com/locate/apr

Original article

The monetary valuation of acute respiratory illness from air pollution in Turkey Shihomi Ara a, *, Cem Tekes¸in b a b

Department of Economics, Hacettepe University, Ankara, 06800, Turkey Department of Agricultural, Food and Resource Economics, Michigan State University, East Lansing, MI, 48824, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 8 March 2015 Received in revised form 22 July 2015 Accepted 23 July 2015 Available online xxx

A contingent valuation study is conducted in three locations in Turkey using a total of 1362 observations to elicit willingness to pay (WTP) to avoid the acute respiratory illness (three days of severe coughing and throat pain). This is the first research of this kind in Turkey, and also one of a few studies conducted in developing countries. Median WTPs are estimated as 65, 51, and 83 PPP-adjusted 2012 USD for AfsinElbistan, Kutahya-Tavsanli and Ankara, respectively. Income elasticities of WTP are derived as 0.8 for Afsin-Elbistan and tested statistically indifferent from 1.0 for Kutahya-Tavsanli, and Ankara, and are found to be greater than the cases in existing studies conducted both in developed and developing countries. We also find that 60 to 90 percent of WTP are devoted for avoiding pain and discomfort/ restricted activity days, and much less weights are given for avoiding possible financial losses. As for the determinants of WTP, university graduates, those who have experienced coughing within one month, have spent out-of-pocket medical expenses and actually lost some part of their income due the last experienced minor symptoms are willing to pay more to avoid the future acute respiratory illness while women and the household using coal as the main source of home-heating are willing to pay less in one or more study areas. New air quality standard for PM10 (transition from 150 mg/m3 to 40 mg/m3 by 2019) causes the reductions in minor respiratory symptoms by 11, 8, and 4.4 per person, and the resulting welfare gains are calculated as 157 million, 123 million, and 1464 million PPP-adjusted 2012 USD for Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively. Copyright © 2015 Turkish National Committee for Air Pollution Research and Control. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Air pollution Respiratory illness Contingent valuation method Willingness to pay Turkey

1. Introduction The ambient air quality standard in Turkey is in transition and the European Union standard will be fully adapted by 2019. Yet, there are some places where air quality is the major cause of health damages, especially acute and chronic respiratory illnesses in this country. Our initial hypothesis was that the major source of air pollution in the cities with high concentration of major pollutants is coal-fired electric power plants and the surrounding coal-mines nearby. Although we observed severe health issues due to a such ulhan and Alemdar, that are located less than power plant in Çog 1 km to an old Afs¸in-Elbistan coal-fired thermal power plant (A plant), the source of severe air quality problem especially in winter

* Corresponding author. Tel.: þ90 312 297 8650x117; fax: þ90 312 299 2003. E-mail address: [email protected] (S. Ara). Peer review under responsibility of Turkish National Committee for Air Pollution Research and Control.

are found to be mainly due to the low quality coal based house heating in the area. In Afsin-Elbistan, one of our study areas, the natural gas network has not reached the region, and the winter time air quality has been very bad. Our interviewers reported many encounters with families having multiple family members, including small children, with respiratory illnesses. In Kutahya-Tavsanli, people have been experiencing serious air pollution due to the same reason as in Afsin-Elbistan. However, after the introduction of natural gas in the region, many citizens witnessed the improvement of air quality. Therefore, Kutahya-Tavsanli is an interesting case going through a rapid improvement in air quality. In Ankara, most of the houses and apartments are equipped with central heating. However, due to the increase in the natural gas price and the distribution of free coal to low income houses by the municipality, the use of coalfired heating at home has been increasing and contributing to the reduction in winter-time air quality. Since the adoption of EU air quality standard requires necessary investments sufficient enough to lower air pollution, for example a

http://dx.doi.org/10.1016/j.apr.2015.07.008 1309-1042/Copyright © 2015 Turkish National Committee for Air Pollution Research and Control. Production and hosting by Elsevier B.V. All rights reserved.

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

2

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

reduction of the particulate matter less than 10 mm in diameters (PM10) from 150 mg/m3 to 40 mg/m3 in 5 years, country/case specific benefit-cost analyses of such policy changes have to be conducted to observe the balance between various costs of pollution control and benefits of improved environmental quality and human health condition. Prior to our project which aimed to measure the value of statistical life (VSL), value of a life-year (VOLY) and welfare gain from reduced minor respiratory illness, there was no value of such measures available for Turkey. In this article, we attempt to estimate the welfare gain from the reduction of minor respiratory symptoms (3-days of severe coughing and throat pain) due to improved air quality in Turkey. We also attempt to decompose respondents' willingness to pay (WTP) into three categories, among which two of them are related to economic/financial concerns such as avoidance of (1) medical expenses and (2) lost-income, and the other is a health related quality of life consideration which is avoidance of (3) pain and discomfort. We found that the main component of WTP is to avoid pain and discomfort due to acute respiratory symptoms. In other words, purely market oriented economic valuations such as cost-of-illness approach merely provide the lower bound of WTP and is leaving out the potentially large fraction of welfare losses due to pain and discomfort suffered by individuals. Our finding confirms the results in earlier studies conducted in USA (Rowe and Chestnut, 1985; Harrington and Portney, 1987; Chestnut et al., 1988; Dickie and Gerking, 1991; Chestnut et al., 1997) and in Taiwan (Alberini and Krupnick, 2000). We also estimate the income elasticity of WTP for avoiding minor respiratory symptoms. The estimation of the elasticity is critical for two reasons. The first is to be used in benefit transfer and the second is to judge if the improvement in health condition is considered as either necessity or luxury in Turkey. While a large amount of stated-preference based studies are conducted in developed countries, significantly smaller amount of studies can be found for developing countries. In most of developing countries, the monetary values of health end-points are largely missing, and they are usually borrowed from the studies conducted in developed countries as needed in benefit-cost analyses and environmental impact assessments. There are mainly two problems in doing so. Firstly, the levels of pollution are quite different between developed and developing countries. For example, the ambient air quality annual standard value of PM10 prior to January, 2014 for Turkey was 150 mg/m3 while it has been 40 mg/m3 for EU countries. As discussed later in this article, the three-year (2009e2011) average in annual PM10 level is 100 mg/m3 for Afsin-Elbistan, for example. It is generally the case that PM10 levels are higher for developing countries and significantly lower for the developed world. Observing the public perceptions given the existing “mid to high-end” of air pollution level add important contribution to the existing literature. Secondly, in developing countries, WTP is constrained more by “ability to pay” rather than “willingness to pay” comparing to the cases in developed countries where “willingness to pay” itself is the determinant. Our interviewers reported that some respondents stated that they were willing to pay, but could not afford and answered NoeNo to our double-bounded dichotomous choice contingent valuation question. Stated-preference studies conducted in the developing countries are valuable since they could provide additional perspective to the existing studies from the developed world. The paper is organized as follows. The second section reviews existing literature on WTP estimation of acute respiratory symptoms. Section 3 contains the methodology, survey design, implementation and estimation models of our CVM study, followed by the data analysis in Section 4. In Section 5, we discuss the application of the found WTP to the new ambient air standard, and we conclude the article by conclusion.

2. Literature review In this section, we review the existing studies on revealing WTP for avoiding minor respiratory symptoms conducted in USA (Tolley et al., 1986; Chestnut et al., 1997), Canada (Johnson et al., 2000), Taiwan (Alberini et al., 1997; Alberini and Krupnick, 1998; Liu et al., 2000), China (Hammitt and Zhou, 2006) and Thailand (Chestnut et al., 1997). Tolley et al. (1986) estimated WTP for seven minor symptoms including coughing spells and irritating throat separately in two cities in USA (Chicago and Denver) and found that mean WTP per month to eliminate 3 symptom days per month are $8.87 for coughing spells and $8.42 for irritating throat. Chestnut et al. (1997) used three-month daily symptom diary study filled by 141 respondents in Bangkok, Thailand and estimated that median WTP for preventing a future day with respiratory symptoms with three different levels ((a) symptom day with no activity restriction, (b) reduced-activity day and (c) work loss day) were $4, $12 and $24, respectively for Thailand. Given the reported median annual household income of $8 000, each WTP corresponds to 0.6%, 1.2% and 3.6% of monthly household income. They concluded, in comparison to the WTP estimates in US, that the proportions of the income the respondents in Bangkok were willing to spend were equal to or higher than the estimates in US, and that the improvement of health condition was seen as a basic necessity in Thailand. They also confirmed that discomfort and the restricted activity on the days with symptoms bother respondents more than the financial losses such as medical expense or lost income. Johnson et al. (2000) conducted choice experiments with four attributes (types of minor symptoms, duration, daily activity levels and annual costs) to estimate WTP for various symptoms for 1, 5 and 10 days time periods with different daily activity levels. For example, willingness to pay to avoid one-day coughing, wheezing or shortness of breath with the condition that one has to stay at home was estimated as CAN$ 158 (1997 CAN$), while it rose to CAN$ 435 for five-day symptoms. Alberini et al. (1997) conducted an in-person survey using CVM in Taiwan in 1992 to estimate WTP to avoid acute respiratory illness which respondents had experienced most recently. Average duration of the illness was 5.3 days and average number of symptoms experienced was 2.2 among 789 respondents. They found that WTP to avoid entire episode with 2.2 symptoms when the episode was a cold was $20.45 (in 1992 USD) for 1-day episode, $34.62 for 5-day episode, and if the illness was not a cold, WTP to avoid 1-day episode was $30.73 while it was $52.01 for 5-day episode. They found that among individual characteristics, monthly household income, educational attainment, the number of sick-leave days afforded by one's job, being married, and health history of serious respiratory illness or chronic illness positively affected WTP. The income elasticity of WTP was 0.45. Alberini and Krupnick (1998) conducted CVM to estimate WTP for avoiding acute respiratory illness due to air pollution in Taiwan in 1991 and 1992 and found that WTP to avoid 1-day illness with 2.2 symptoms is $26 and 2-day symptom is $32.34. For the avoidance of 3-day symptoms which is used in our study, WTP is approximately $37 in 1992 USD, which corresponds to about 1.5% of household income. The income elasticity of WTP estimated was 0.41, statistically different from 0. Hammitt and Zhou (2006) estimated WTP for preventing an episode of minor illness similar to each respondents' most recent cold as 3 to 6 dollars in Beijing, Anquing and rural areas in China. Durations of the recent cold are reported between 4.7 and 7.4 days on average, and the number of symptoms is on average 3.4. The number of days the respondents missed work are 1.2 in rural areas, 0.9 in Beijing and 0.2 in Anqing on average. The income elasticities were estimated as 0.01, 0.14 and 0.09 in the rural areas, Anquing

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

and Beijing, respectively, while the ones for Beijing and the rural areas were found to be insignificantly different from zero. Liu et al. (2000) estimated WTP to avoid the recurrence of the cold episode (average duration of 6.5 days) each respondent described as $37 (in 1995 USD) in Taiwan. They reported that 54% of the respondents visited a doctor and the mean out-of-pocket cost was $20.11. The estimated income elasticity of WTP was 0.4. The common observations across these studies are that WTP is significantly greater than the sum of the potential financial loss due to medical cost or loss-in-income due to the illnesses, indicating that the discomfort and reduced ability to handle daily activities are the larger components of WTP in comparison to financial losses. Interestingly, many representative articles estimating WTP for minor (respiratory) illnesses were published in 1996 and 1997 and for Taiwan. Yet, the number of available studies in other developing countries, especially in non-Asian countries is very limited.

3

Elbistan (n ¼ 119) by ten trained interviewers under our direct supervision in all locations. Face-to-Face interviews are conducted based on the random home addresses obtained from the Turkish Statistical Institute. After the modification of WTP for each location and wordings for the better clarity of the survey, the main survey was conducted during June 4e11 in Afsin and Elbistan, June 23 e July 2 in Central Kutahya and Tavsanli, and June 14e21 in Ankara. Initially a total of 1390 surveys (492, 562 and 352 for Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively) are collected. After eliminating coding errors and observations with critical missing values and protest votes, we used 480, 532 and 350 observations for Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively, making a total of 1362 observations. 3.3. Estimation models

3. Contingent valuation method

The definitions of all variables are listed in Table 1. Model 1 includes constant, monthly household income and bid:

3.1. Survey design

V ¼ b0 þ b1 LNHHINC þ b2 BID

ðModel 1Þ

Akaike Information Criterion (AIC) values are compared for linear, log-linear and log-logistic form, and log-logistic form are found to have the lowest AIC for all cases. Therefore, all models are estimated in log-logistic form. Model 2 attempts to reveal the relationship between the individual characteristics and WTP.

The formal theoretical model of CVM for estimating WTP for avoiding illnesses is well established in the literature. See Berger et al. (1988), Cropper and Freeman (1991), Harrington and Portney (1987) or Haab and McConnell (2003), for example. The survey contains the contingent valuation question which is intended to reveal people's willingness to pay value for minor symptom (throat pain and severe coughing for three days). This symptom was selected to represent the minor respiratory symptom which could be caused by air pollution. Once we identify the WTP for each episode of such symptom, we can calculate the monetary value of the benefit due to air pollution reduction policies in terms of minor symptom. Ostro (1994) derived the coefficient for respiratory symptoms, including both lower and upper respiratory symptoms, per year per person in relation with PM10. Symptoms of lower respiratory illness include shortness of breath, weakness, high fever, coughing and fatigue, and of upper respiratory illness include cough, sore throat, runny nose, nasal congestion, headache, low grade fever, facial pressure and sneezing. According to their study, the coefficient is 0.183 * (change in annual PM10), meaning a reduction of PM10 by 10 mg/m3 will reduce 1.83 respiratory symptoms per year per person. If the government policy could reduce PM10 by 50 mg/m3, then a resident of the city could reduce the number of respiratory symptoms by 9.15 in the year. It means that 50 mg/m3 reduction in PM10 in a given year in Elbistan, for example, could reduce 9.15 respiratory symptoms per person, and for 139 855 people in Elbistan in 2012, results in a total of 1 279 673 reduced number of respiratory symptoms. By using estimated coefficient of Ostro (1994), we can derive the value of avoided respiratory symptoms as [Estimated WTP per symptom] *9.15 symptoms * [Population of the city] as the gained welfare due to pollution reduction within the city per year. The hypothetical question is reported in the Appendix. In the question, each respondent is assigned randomly one of 5 or 6 initial WTP values (20, 40, 60, 100 and 150 for Afsin-Elbistan and KutahyaTavsanli, and 20, 40, 60, 100, 150 and 300 for Ankara). Depending on the initial answers (yes/no), the higher (if “yes” for the initial value) or the lower (if “no”) values are asked as the follow-up questions.

4. Results

3.2. Data collection

4.1. Descriptive statistics

Followed by focus group discussions and testing of the survey versions with university students, pre-tests were conducted during September 21e25, in 2011 in Central Kutahya (n ¼ 95) and Tavsanli (n ¼ 85), May 13e14, 2012 in Ankara (n ¼ 122) and June 2, 2012 in

In this section, we report the estimated results to revealed WTP for minor respiratory symptom, three days of throat pain and severe coughing. The estimation of double-bounded dichotomous choice (DBDC) question was conducted with the NLOGIT software.

V ¼ b0 þ b1 LNHHINC þ b2 GENDER þ b3 AGE þ b4 UNIV þ b5 SMOKER þ b6 SPORT þ b7 COUGHT þ b8 RESP þ b9 MEDCOST þ b10 LOSTINC þ b11 HLTHBAD þ b12 HEATCOAL þ b13 BID

ðModel 2Þ

The expected signs for the coefficients are positive for b1, b4, b6, b7, b8, b9, and b10, negative for b12 and b13 and uncertain for b0, b2,b3, b5, and b11. In addition to the estimation of models for each city, we have estimated each model by pooling all the observations from three study areas (the pooled case). For the pooled case, the 3-year average of PM10 is included in order to measure the relationship between air quality and WTP in each area. Since we do not have the location-specific PM10 values for each survey respondent, as an available alternative, we used 2009e2011 average of PM10 and they are 100, 83 and 64 for Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively. The median WTP from log-logistic model for Model 1 is

Median WTP ¼ exp

b þ b1 HHINC  0 b2

!

and for Model 2 is

0 Median WTP ¼ exp@ 

b0 þ

P i

b2

bi zi

1 A

where zi is the average of the ith individual characteristics zi.

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

4

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

Table 1 Variable descriptions. Variable

Description

Attribute variables BID See Table 2 for the bid structure LNHHINC log (monthly household income/1000) GENDER 1 if the respondent is female, 0 otherwise AGE Age of the respondent UNIV 1 if having university or higher degree, 0 otherwise SMOKER 1 if the respondent smokes regularly, 0 otherwise SPORT 1 if the respondent exercises regularly (once a week or more), 0 otherwise COUGH 1 if the respondent has experienced (experiencing) coughing in one month, 0 otherwise RESP 1 if the respondent has experienced (experiencing) chronic respiratory diseases including Asthma, Chronic Bronchitis, and Emphysema in last three years, 0 otherwise. MEDCOST 1 if the respondent has paid out of pocket for her hospital visit and/or medication for the minor symptom(s) experienced in one month, 0 otherwise. LOSTINC 1 if the respondent has lost some part of her income due to the minor symptom(s) experienced in one month, 0 otherwise. HLTHBAD 1 if the respondent considers she is in bad health condition for her age, 0 otherwise HEATCOAL 1 if coal is used as the main source of household heating, 0 otherwise PM10 2009e2011 Average PM10 (100, 83, 64 mg/m3 for Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively).

Each respondent is assigned one WTP value first, then asked if she is willing to pay the initial amount. If the answer is yes, the second amount, the higher than the first offer is asked. If the answer to the first offer is no, the lower amount is asked. DBDC question is the most common method used in contingent valuation method. In Table 2, yes/yes means yes answer to both questions, yes/no is for yes to the first offer and no to the higher offer, no/yes is no to the first offer and agree to pay the lower offer, and no/no is not agreeing to pay any offered amount. The bid structures for each city and the responses are reported in Table 2. All the bid responses follow the monotonicity of proportions of accepting the first bids in all cities, except for the fourth bids in Afsin-Elbistan. The highest bid of (300, 500, and 150) is added only for Ankara due to the higher household income observed in Ankara. Table 3 includes the summary statistics for the variables included in the estimated models and respondents' opinion on air quality and own health. The monthly household income is approximately 1700 TL ($1123) for Afsin-Elbistan and KutahyaTavsanli, 2600 TL ($1717) for Ankara and 1971 TL ($1301) for the pooled data. Women are slightly under-represented in AfsinElbistan, Ankara and the pooled case. The average ages are around 40e42. While university graduates are 30% of respondents in Ankara, it is about 10% in other two areas. The occurrences of coughing and chronic respiratory illnesses are observed to be relatively higher in Afsin-Elbistan, indicating the existence of higher health risk factors in the area. As for the health status, only 38% of respondents from Afsin-Elbistan consider having good health for

their ages and 20% consider that their health is worse than it should be. For Ankara, half of the respondents consider that they are in good health condition and only 12% consider they are in bad condition. 38% of observations from Ankara do exercise regularly, while it is 29% for other cities. One of the reasons for this high health risk conditions in Afsin-Elbistan is the intensive use of low-quality coal for heating during winter due to lack of natural gas provision in the area. 71% of respondents use coal as the main source of househeating while the proportion is lower in other study areas (33% in Kutahya-Tavsanli, 4% in Ankara). For non-respiratory illnesses such as cardio-vascular illnesses, cancer, diabetes, we did not observe any regional differences in their occurrences. 34, 23, and 31% of respondents in Afsin-Elbistan, KutahyaTavsanli, and Ankara, respectively, have paid out-of-pocket expenses for hospital visit and/or medication to cure the minor symptom(s) they have experienced within a month. Small number of respondents has lost their monthly income due to the minor symptom (6% in Afsin-Elbistan, 0.6% in Kutahya-Tavsanli, and 2% in Ankara). The smoking rates are relatively similar in all three areas among our respondents and 46% (Afsin-Elbistan), 38% (Kutahya-Tavsanli) and 42% (Ankara). However, the percentage of those who have quit smoking is significantly higher in AfsinElbistan (41%) comparing to other cities (22% in Kutahya and 19% in Ankara). Problems in air quality are well perceived among respondents. AQYR is based on a question “What is your opinion about air quality in your city? 1. very good, 2. good, 3. normal, 4. bad, 5. very bad”, AQWIN is based on a similar question, asking

Table 2 Bid structure and response ratios. Version

City

Bid

Bid high

Bid low

Yes/yes

Yes/no

No/yes

No/no

n

Minor Symptom

Afsin-Elbistan

20 40 60 100 150 20 40 60 100 150 20 40 60 100 150 300

40 60 100 150 300 40 60 100 150 300 40 60 100 150 300 500

10 20 40 60 100 10 20 40 60 100 10 20 40 60 100 150

66 (63%) 72 (67%) 59 (45%) 35 (52%) 12 (17%) 74 (69%) 62 (58%) 51 (48%) 24 (23%) 10 (9%) 3 (100%) 57 (77%) 43 (57%) 30 (41%) 17 (26%) 6 (10%)

21 (20%) 7 (7%) 22 (17%) 9 (13%) 15 (21%) 13 (12%) 13 (12%) 13 (12%) 21 (20%) 21 (20%) 0 (0%) 8 (11%) 10 (13%) 14 (19%) 17 (26%) 7 (12%)

6 (6%) 8 (7%) 8 (6%) 7 (10%) 7 (10%) 4 (4%) 11 (10%) 9 (8%) 16 (15%) 17 (16%) 0 (0%) 1 (1%) 3 (4%) 6 (8%) 4 (6%) 9 (16%)

11 (10%) 19 (18%) 42 (32%) 15 (22%) 38 (53%) 16 (15%) 20 (19%) 33 (31%) 45 (42%) 59 (55%) 0 (0%) 8 (11%) 20 (26%) 22 (30%) 28 (42%) 36 (62%)

104 107 131 67 72 107 106 106 106 107 3 74 76 73 66 58

Kutahya-Tavsanli

Ankara

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

5

Table 3 Descriptive statistics of variables. Variable

HHINC GENDER AGE UNIV SMOKER SPORT COUGH RESP MEDCOST LOSTINC HLTHBAD HEATCOAL AQYR AQWIN AQHIST HLTHGOOD

Afsin-Elbistan

Kutahya-Tavsanli

Ankara

Pooled

Mean

Std. dev

Mean

Std. dev

Mean

Std. dev

Mean

Std. dev

1711.61 0.44 40.26 0.12 0.46 0.29 0.37 0.29 0.34 0.06 0.20 0.71 3.83 4.71 2.84 0.38

1269.35 1.61 12.86 0.32 0.58 0.45 0.84 0.45 0.48 0.23 0.40 0.45 0.84 0.61 0.51 0.49

1764.17 0.50 42.83 0.11 0.38 0.29 0.16 0.22 0.23 0.01 0.09 0.33 2.77 3.62 1.67 0.47

1036.65 0.50 13.16 0.34 0.49 0.45 0.37 0.42 0.42 0.08 0.29 0.47 0.83 1.01 0.94 0.50

2640.08 0.45 42.63 0.30 0.42 0.38 0.19 0.18 0.31 0.02 0.12 0.04 2.81 3.41 2.16 0.50

1457.95 0.50 12.67 0.48 0.49 0.52 0.39 0.38 0.46 0.13 0.32 0.22 0.82 1.01 0.93 0.50

1970.74 0.46 41.87 0.16 0.42 0.31 0.23 0.23 0.29 0.03 0.13 0.39 3.15 3.95 2.21 0.45

1298.84 1.04 12.98 0.38 0.53 0.46 0.42 0.42 0.45 0.16 0.34 0.49 0.97 1.06 0.96 0.50

specifically for the air quality in winter, and AQHIST is the answer of the question “How is the current air quality comparing to the past? 1. getting better, 2. same as 5 years ago, 3. getting worse.” It is revealed that Afsin-Elbistan respondents perceive annual air quality as bad and winter air quality as very bad on average and they consider that the situation is getting worse. On the other hand, Kutahya respondents consider that the air quality is somewhat improving over time. This is due to an introduction of natural gas in the area in 2005. Although 33% of respondents still rely on coal-heating, the subscription to natural gas has been increasing over years. At least one of the minor respiratory symptoms (headache, eye irritation, coughing, throat ache, shortness of breath, chest pain and allergies) has been experienced within one month of our survey by 79, 60, and 66% of respondents in Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, respectively. This indicates that the majority of our respondents could respond the asked contingent question based on their actual experience of the recent past. Out of these respondents who have experienced at least one of the minor symptoms, on average 42% visited hospitals/clinics and 55% took medication for the symptom. The discomfort levels associated with these symptoms are on average 2.81. It is surprising to observe that more than 20% of those who have experienced a minor symptom reported the discomfort level as 4 or 5 (Table 4). It is possible that the WTPs revealed by these respondents may overestimate WTP for minor respiratory symptoms with level 2 discomfort level (Fig. 1). The out of pocket medical expenses (including both the cost of hospital visit and medication) are 21, 11, and 11 TL as the average among those who have experienced one or more minor symptoms, including 56, 62, and 55% of respondents with zero expenditure for Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, respectively. The average medical costs among those who have paid positive medical cost are 48, 66, and 34 TL for Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, respectively. As for the lost income, out of the total of 931 respondents who have experienced minor symptom within one month, 55 (5.9%) respondents missed work due to the symptom and 36 (3.7%) respondents have experienced loss in income. The rate is the highest in Afsin-Elbistan where self-employed workers are found to be the most frequent compared to the other study areas (Table 4). The average loss of incomes among those who have suffered from one or more of minor symptoms are 20.5 TL, 0.53 TL, and 1.16 TL, and for those who have lost some part of their income, the average losses are 277 TL, 57 TL, and 54 TL for Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively.

Table 4 Out-of-pocket medical expenses, lost income and discomfort level. Count (%) Afsin-Elbistan

Kutahya-Tavsanli

Out of pocket medical expenses(TL) 0 213 56% 197 1e50 138 37% 109 51e100 12 3% 4 101e200 9 2% 6 >200 6 2% 1 Lost income (TL) 0 350 93% 314 1e100 14 4% 3 101e200 6 2% 0 201e300 4 1% 0 301> 4 1% 0 Discomfort level 0 13 3% 0 1 23 6% 15 2 107 28% 91 3 125 33% 136 4 81 21% 61 5 29 8% 14

Ankara

62% 34% 1% 2% 0%

127 94 8 3 0

55% 41% 3% 1% 0%

99% 1% 0% 0% 0%

227 4 1 0 0

98% 2% 0% 0% 0%

0% 5% 29% 43% 19% 4%

6 12 80 78 48 8

3% 5% 34% 34% 21% 3%

Percentages are calculated based on those who have experienced at least one minor symptom within a month (378, 317 and 232 observations in Afsin-Elbistan, Kutahya-Tavsanli and Ankara, respectively).

4.2. Estimated results The estimated results for Model 1 and Model 2 for each studyarea and the pooled data are listed in Table 5. All estimated coefficients for Model 1 are statistically significant at one percent level with expected signs. Median WTP for avoiding “throat pain and severe coughing which last for three days” based on mean monthly household income for each sample are 98 TL (65 PPPadjusted USD), 77 TL ($51), 126 TL ($83) and 94 TL ($62) and correspond to 5.7%, 4.4%, 4.8% and 4.8% of the monthly household income for Afsin-Elbistan, Kutahya-Tavsanli, Ankara and the pooled case, respectively. Although the value of WTP is not the highest in Afsin-Elbistan where the air pollution is the serious environmental and health issue, it is the highest in percentage in monthly household income among the three study areas. The mean WTPs calculated by truncating at the maximum bids (300 TL for AfsinElbistan and Kutahya-Tavsanli, and 500 TL for Ankara and the pooled case) are 88 TL ($58), 84 TL ($55), 137 TL ($90), and 117 TL ($62), respectively. Median WTPs are greater than mean WTPs for all cases except for Afsin-Elbistan.

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

6

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

Fig. 1. Discomfort scale (translated in Turkish in the survey).

Table 5 Estimated results for minor symptom. Afsin-Elbistan MODEL 1 CONSTANT BID LNHHINC

4.798 (0.363) 1.107 (0.085) 0.887 (0.103)

Kutahya-Tavsanli MODEL 2

*** *** ***

GENDER AGE UNIV SMOKER SPORT COUGH RESP MEDICCOST LOSTINC HLTHBAD HEATCOAL

4.839 (0.571) 1.139 (0.089) 0.810 (0.112) 0.022 (0.122) 0.008 (0.008) 0.082 (0.346) 0.187 (0.172) 0.126 (0.219) 0.647 (0.216) 0.185 (0.219) 0.091 (0.211) 0.996 (0.534) 0.192 (0.251) 0.561 (0.230)

MODEL 1 *** *** ***

5.327 (0.355) 1.373 (0.087) 1.517 (0.167)

Ankara MODEL 2

*** *** ***

***

*

**

6.401 (0.576) 1.427 (0.091) 1.262 (0.199) 0.691 (0.213) 0.004 (0.007) 0.100 (0.334) 0.221 (0.214) 0.007 (0.215) 0.082 (0.262) 0.061 (0.237) 0.490 (0.242) 0.766 (1.918) 0.251 (0.407) 0.718 (0.199)

Pooled

MODEL 1 *** *** ***

5.307 (0.516) 1.269 (0.111) 1.024 (0.191)

MODEL 2 *** *** ***

***

**

***

5.066 (0.702) 1.329 (0.122) 0.839 (0.213) 0.059 (0.234) 0.003 (0.009) 0.917 (0.274) 0.090 (0.233) 0.161 (0.226) 0.788 (0.334) 0.258 (0.340) 0.117 (0.251) 0.352 (1.074) 0.715 (0.384) 0.356 (0.503)

MODEL 1 *** *** ***

5.109 (0.223) 1.242 (0.052) 1.104 (0.076)

MODEL 2 *** *** ***

***

**

*

PM10

3.966 (0.536) 1.256 (0.054) 0.951 (0.087) 0.044 (0.042) 0.002 (0.005) 0.410 (0.170) 0.031 (0.108) 0.124 (0.126) 0.482 (0.144) 0.042 (0.140) 0.173 (0.129) 0.701 (0.413) 0.182 (0.175) 0.548 (0.138) 0.014 (0.005)

N. Observations LogL AIC

480 522 1049

480 509 1045

532 580 1166

532 566 1160

350 370 746

350 358 743

1362 1481 2969

1362 1456 2943

WTP (median) WTP (mean) 95% Lower CI 95% Upper CI

98 88 82 117

100 90 84 120

77 84 67 88

76 84 67 87

126 137 106 150

129 140 108 153

94 117 86 103

95 118 87 105

Income elasticity of WTP

0.8 (0.109)

***

1.1 (0.132)

***

0.8 (0.156)

***

0.9 (0.067)

*** *** ***

**

***

*

*** ***

***

*, **, *** corresponds to statistically significance levels of 10%, 5% and 1%, respectively. Standard Errors in parentheses. Mean WTPs are calculated by truncating at the maximum bids (300 for Afsin-Elbistan and Kutahya-Tavsanli and 500 TL for Ankara and Pooled Data). 95% Confidence Intervals (CI) are for median WTP. Income Elasticities of WTP are tested statistically indifferent from 1.0 for all cases, except for Afsin-Elbistan.

WTPs for avoiding “3-days of throat pain and severe coughing” we found for Turkey are similar to the WTP of studies conducted in Taiwan (Alberini and Krupnick, 1998) and Thailand (Chestnut et al., 1997) once adjusted to inflation, significantly higher than the case in China (Hammitt and Zhou, 2006) and lower than the WTP that reported most of the symptoms in USA (Liu et al., 2000) in which WTP for one-day avoidance of various symptoms are in the range between $10 and $150 in 1995 USD.

The factors positively affecting the WTP values are LNHHINC (all cases), COUGH (Afsin-Elbistan, Ankara), LOSTINC (Afsin-Elbistan), MEDCOST (Kutahya-Tavsanli), UNIV (Ankara), HLTHBAD (Ankara) and PM10 (Pooled), while HEATCOAL (Afsin-Elbistan, Kutahya), and GENDER (Kutahya) are negative determinants of WTP. In other words, those who have higher household income, have experienced coughing within one month, have lost part of their income due to one or more of the minor respiratory symptoms, have paid

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

out-of-pocket medical cost, are university graduates, or consider the current health status as bad are willing to pay more, while those who use coal as the main heating source or are woman are willing to pay less. It is interesting to observe the regional differences in the influencing determinants. Gender, for example is a significant factor in Kutahya, but not in other regions. UNIV is relevant only in Ankara (30% of the respondents are university graduates in Ankara, while it is 12 and 11% in Afsin-Elbistan and Kutahya, respectively). Lost income is not an important factor in Kutahya or Ankara (8% missed work in Afsin-Elbistan while only 2 and 3% missed work in Kutahya and Ankara) since much smaller proportion of respondents actually have missed work and even fewer respondents have lost their partial income due to the missed - work - days. Statistically significant COUGH coefficients indicate that the experienced symptom which is still new in their memory (within one month) could influence WTP significantly. SMOKER and SPORT are found not to have statistically significant impacts on WTP although our a priori expectations are negative for SMOKER and positive for SPORT. Since smoking is often associated with severe coughing, it is interesting to find it insignificantly affecting the WTP for all study areas. AGE as well does not have any statistically explanatory power in the model, revealing that WTP does not vary according to the respondents' age, unlike for the case of the value of statistical life (VSL) (Tekesin and Ara, 2014) and for the value of life year (VOLY) (Tekesin and Ara, 2015). The coefficient for PM10 included in Model 2 of the pooled case is positive and significant at one-percent level, indicating that the higher the PM10 level of the area, the higher the WTP of the respondents who reside in the area becomes. Therefore, it is confirmed that worse back-ground air pollution actually increases WTP. The negative and statistically significant HEATCOAL estimates, even after controlling for household income, provide what may seem to be a counter-intuitive result. Those who are more at risk of suffering from minor respiratory symptoms are willing to pay less for the reduction of such risk. There are three possible explanations for this observation. The first explanation could be that they may be used to the symptom described, and did not feel the need to pay and reduce the symptoms. The second could be that they may be possessing the sense of “control” over their current health and air quality situation. If they stop using coal-based heating, they know that they are able to decrease the symptom days. Therefore, they are not willing to pay extra money for other measures. The third interpretation is that there may be unobserved factors connecting coal users and their WTP. The alternative to coal-heating in AfsinElbistan is central heating of the building, and it is either central heating or natural-gas based heating in Kutahya-Tavsanli. It is possible that the unobserved variables are related to the attributes determining their housing decisions. Hedonic price analysis may be necessary in the future to actually reveal the underlying preference. In order to further investigate the health-pollution relationship, we regress our health indicators (COUGH, RESP and HLTHBAD) on LNHHINC and PM10 using the logit model. We find that PM10 is positive and is statistically significantly associated with all health indicators while LNHHINC is negatively related with RESP (10% significance level) and HLTHBAD (1% significance level). When other possible determinants of these health indicators, such as the time spent outdoor each day (OUTDOOR), SMOKER and SPORT are included, we find that SMOKER has positive and significant relationship only with COUGH, OUTDOOR does not have any influence on our health indicators, and SPORT is negative and significant at 1% level for all indicators. However, since the explanatory power of these logistic models are very low due to the obvious missing variables, the results of these models attempting to reveal the health-pollution relationship should be considered as referencepurpose only.

7

The differences between WTP calculated based on Model 1 and Model 2 are very small for all study areas (e.g. 98 TL for Model 1 and 100 TL for Model 2 for Afsin-Elbistan). Although the mean monthly household income level is the lowest in Afsin-Elbistan and the highest in Ankara, our results reveal that the WTP estimates do not follow the same order, the lowest in Kutahya - Tavsanli and the highest in Ankara. The higher WTP in Afsin-Elbistan is most probably due to the higher rate of experienced minor respiratory symptoms. As we discussed earlier, significantly more respondents in Afsin-Elbistan have experienced minor respiratory symptoms such as headache, eye irritation, coughing, throat pain, and chest pain (Table 3) and as a result 20 percent of the respondents consider themselves in a bad health condition (9 and 12% in Kutahya and Ankara). This result implies the importance of background risk and health condition when we attempt a benefit transfer to other regions in Turkey. Marginal willingness to pay (MWTP) derived from statistically significant estimates are reported in Table 6. If the monthly household income increases by 100 TL, it is expected to increase WTP by 3e5 TL. Females are willing to pay 37 TL less for avoiding three-day respiratory symptoms in Kutahya-Tavsanli, university graduates are willing to pay 104 TL more compared to nongraduate in Ankara, those that have experienced coughing within one month of the survey are willing to pay 63 TL (Afsin-Elbistan) and 93 TL (Ankara) more compared to those who did not have the symptom, having spent out-of-pocket medical cost caused Kutahya-Tavsanli respondents to willing to pay 29 TL more than those who did not pay, having lost partial income due to the minor respiratory symptoms lead to pay 133 TL more than those who did not lose any income (Afsin-Elbistan), and those who perceive their health as bad for their age are willing to pay 86 TL more for avoidance of the minor respiratory symptoms. Those who use coal as the main house-heating source are willing to pay 56 TL (AfsinElbistan) and 35 TL (Kutahya-Tavsanli) less. If PM10 increases (decreases) by 10 mg/m3, WTP is expected to increase (decrease) by 11 TL. It is also important to realize the composition of WTP. Our assumption is that we can decompose an individual's WTP into three categories: (i) out-of-pocket medical expenses, (ii) lost income due to the illness, and (iii) discomfort and restrictions in daily activities. Majority of respondents (79%, 61%, and 66% of AfsinElbistan, Kutahya-Tavsanli, and Ankara respondents, respectively) have spent some amount for hospital visit and/or medicine for the experienced minor symptom(s). The average expenditures are 21, 11, and 11 TL for those who have suffered the minor symptom(s) within one month of the survey and are 48, 66, and 34 TL for those who have paid a positive medical cost in Afsin-Elbistan, KutahyaTavsanli, and Ankara, respectively. Here, medical cost measure is based on self-reported out-of-pocket medical cost paid by the respondent for the minor symptom experienced within one month. If the respondents visit public hospitals or public health centers, they are charged a small amount of co-payment. The other possible out-of-pocket expenses are for obtaining drug prescriptions and for purchasing medicines. Out-of-pocket expenses vary greatly if the treatment is received at private hospitals and the actual expenses depend on the private insurance coverage. We have to note that out-of-pocket expenses reflect only a fraction of the actual medical cost of curing the illness. In addition, only a few respondents have actually lost the part of their income due to the minor symptom(s) and the total average of the lost income is 20.5 TL for Afsin-Elbistan where the LOSTINC variable is one of the statistically significant determinant of WTP, but for other two study areas, the lost income is negligible and the total average is 0.53 TL and 1.16 TL. Since self-employed, part time or daily-contracted workers, not fully-employed workers could lose

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

8

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

Table 6 Marginal Willingness to Pay based on Model 2 Results.

HHINC GENDER UNIV COUGH MEDCOST LOSTINC HLTHBAD HEATCOAL PM10

Description

Afsin-Elbistan

Kutahya-Tavsanli

Ankara

Pooled

HHINC þ100 TL Female University graduates Have experienced coughing within one month Have spent out-of-pocket medical cost Have lost partial income Perceives that health condition is bad for her age. Use coal as main source of household heating PM10 þ 10 mg/m3

5.18 [ 3.06 7.30] 1.94 [19.19 23.08] 7.02 [63.58 49.54] 62.66 [13.04 112.29] 8.10 [29.41 45.62] 133.25 [¡72.46 338.97] 17.82 [30.29 65.92] ¡55.49 [¡107.05e3.93]

4.46 [2.78 6.15] ¡37.06 [¡60.92e13.21] 5.18 [38.07 27.70] 4.46 [23.95 32.87] 28.76 [¡2.64 60.17] 31.61 [149.39 86.17] 12.42 [49.01 24.17] ¡35.42 [¡53.72e17.13]

3.54 [1.55 5.52] 5.73 [49.89 38.44] 104.15 [28.17 180.13] 93.20 [¡2.94 189.34] 11.11 [57.26 35.03] 38.84 [225.74 303.43] 86.04 [¡27.49 199.56] 30.58 [106.47 45.31]

4.42 [3.38 5.46] 3.32 [9.53 2.90] 34.84 [2.94 66.75] 41.09 [13.97 68.22] 13.75 [6.73 34.24] 69.59 [¡34.77 173.95] 14.35 [14.59 43.28] ¡39.75 [¡59.32e20.17] 10.97 [2.45 19.49]

Bold values are based on the estimates which are statistically significant at least at 10% level in Model 2. All values are in 2012 TL [95% Confidence Interval].

their partial monthly income due to their absences, LOSTINC variable specifically measures the effect of the visible lost-income cost of minor symptom to such workers on WTP. Considering these facts, and MWTP estimates for MEDCOST and LOSTINC (the respondents who have paid out-of-pocket medical cost in KutahyaTavsanli are willing to pay 33 TL more in addition to the estimated WTP of 76 TL and the respondents who have lost a partial income in Afsin-Elbistan are willing to pay 158 TL more while the median WTP is 100 TL), it is possible to conclude that respondents are willing to pay to avoid 3-day of minor respiratory symptoms not only to avoid the possible medical expenses and potential loss in their income, but also to avoid discomfort, restrictions in daily activities and possibly other factors which cannot be specified by the survey we used. The estimated results of Model 2 suggest that WTP could be decomposed into 21, 14, and 8.6% of medical costs, 20.3, 0.3, and 1% of lost income and 59, 86, and 90% for avoiding discomfort due to the set of minor symptoms for Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, respectively, using the average medical expenditures and lost income among those who have suffered minor symptom within one month. For those who actually spent positive medical costs and lost some part of income, the composition of WTP is 61% to avoid loss in income and 39% to avoid discomfort/restricted daily activities/other non-monetary cause for Afsin-Elbistan respondents and for Kutahya-Tavsanli, the composition is 33% to avoid medical cost and 67% to avoid discomfort and other reasons, if we can assume that respondents who have spent medical costs and lost partial income during the most recently experienced minor symptoms consider the same kind of financial loss could occur for the hypothetical symptoms they are asked to evaluate. For Ankara respondents, 100% of WTP is for nonmonetary causes. The income elasticities of WTP are estimated as 0.80, 1.10, 0.81 and 0.9 for Afsin-Elbistan, Kutahya-Tavsanli, Ankara and the pooled case, respectively. However, for Kutahya-Tavsanli, Ankara and the pooled case, the elasticities are tested statistically insignificantly different from 1.0 by setting the null hypothesis as H0:(bLNHHINC/bBID)1 ¼ 0 and for Afsin-Elbistan, the elasticity is statistically significant at the 10% level. Therefore, we can conclude that the good “avoiding 3-day respiratory symptoms” is an income-neutral good in Kutahya-Tavsanli and Ankara and is a necessity in Afsin-Elbistan. These elasticities found are higher than the ones in existing studies such as Liu et al. (2000) which found the income elasticity of 0.4 for avoiding the recurrence of the own cold which was reported by the respondents in Taiwan, and Alberini et al. (1997) that reported 0.3 for avoiding an acute illness episode. 5. An application to air pollution policy evaluation The Turkish ambient air quality standard is in transition. It was 150 mg/m3 for PM10 by the end of 2013, but starting from 2014, the standard level is targeted to be lowered to 60 mg/m3, and

furthermore, it will be the same as the EU standard of 40 mg/m3 by year 2019. Given that the three year average (2009e2011) of PM10 levels for Afsin-Elbistan, Kutahya-Tavsanli and Ankara are 100, 83 and 64 mg/m3, respectively. The expected reduction in PM10 levels by 2019 are 60, 43 and 24 in the corresponding cities. By using the WTP for minor respiratory symptom estimated in this study and Ostro (1994)'s dose-response coefficient for “Respiratory symptoms/person” (0.183 respiratory symptom reduced for 1 mg/m3 reduction of PM10 for each individual), we attempt to derive the welfare gains from the reductions in minor respiratory symptoms due to this policy change. We adopt the estimated WTP for avoiding one case of minor respiratory symptom (throat pain and severe coughing for three days) from Model 1. Although we did not present this symptom as the result of air pollution in our hypothetical question in order to avoid protest responses against the local government, the description is possibly considered as one of the representative symptoms of acute respiratory symptom from air pollution. WTP for avoiding one case of minor respiratory symptom for the next one month are estimated as 98, 77, and 126 TL for AfsinElbistan, Kutahya-Tavsanli, and Ankara, respectively. We expect reductions of approximately 11, 8, and 4 respiratory symptoms per person due to the adaptation of new policy, and the improvement causes welfare gains of 1 076, 606, and 553 TL per person in AfsinElbistan, Kutahya-Tavsanli, and Ankara, respectively. Given the target PM10 emission reduction levels and population for each study area, we derived the welfare gains from PM10 emission reduction by 60, 43 and 24 mg/m3 in terms of reduction of minor respiratory symptom are 238 million TL ($157 million in PPPadjusted 2012 USD), 186 million TL ($123 million), and 2216 million TL ($1464 million) for Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, respectively. The derivation method is summarized in Table 7. The total WTP or welfare gain from reducing 11, 8 and 4 respiratory symptoms correspond to 1 076, 606 and 553 TL for AfsinElbistan, Kutahya-Tavsanli and Ankara, respectively. Considering that the annual household income are 20 539, 21 170 and 31 680 TL for each area, the total WTP for reducing 11, 8 and 4 respiratory episodes annually account for 5.2, 2.9 and 1.7% of the annual household income. This seems high considering that the acute respiratory symptoms are one of many other possible health effects due to air pollution. However, for example, Chestnut et al. (1997) report median WTP of $4 and mean WTP of $16 for avoiding one respiratory symptom day. They estimate the occurrence of the minor respiratory symptom to be 2.5e8.3 cases per person for an increase in PM10 by 10 mg/m3. It is 15e50 cases for 60 mg/m3 if we adjust the change of PM10 for Afsin-Elbistan. Given that the annual household income reported is $8 000, it corresponds to 0.75%e2.5% of the household income if we use median WTP and 3e10% of the household income if mean WTP is used. Considering their findings, it seems that our finding is within a reasonable range for KutahyaTavsanli and Ankara, although the rate is slightly high in AfsinElbistan due to the high PM10 level.

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

9

Table 7 Estimated WTP application to air pollution policy evaluation.

[A] 2009e2011 Average of PM10 (mg/m3) [B] Target reduction by 2019 (40 mg/m3) [C] Respiratory symptom/person [D] Reduced respiratory symptoms/PERSON [E] WTP for avoiding a minor respiratory symptom (TL) [F] Welfare gain from PM10 reduction due to reduced minor respiratory symptoms per person (TL) [95% CI] [G] Population (2000) [H] Reduced number of respiratory symptoms for each study area [I] Welfare gain from PM10 reduction from reduced minor respiratory symptoms (in million TL) [95% CI. in million TL] [J] [I] in PPP-adjusted 2012 USD in million [95% CI. in million $]

Source/derivation

Afsin-Elbistan

PM10 emission data ¼ [A] - 40 Ostro (1994) ¼ [B]*[C] Estimated median WTP (model 1)

100 83 60 43 0.183 per mg/m3 PM10 10.98 7.87 98 77

¼ [D]*[E]

1076 [900e1285]

2000 Census ¼ [D]*[G] ¼ [F]*[G]

¼

½I1:189 1:8

6. Discussion In this study, we derived individual WTP for avoiding minor respiratory illness (three days of severe coughing and throat pain) for one month. We did not adapt to estimate WTP for avoiding “the episode of acute respiratory illness most recently experienced by the respondent” as used in Alberini et al. (1997) in order minimize the uncertainty in respondents' memory. An additional survey version to estimate the WTP for minor respiratory illness with different severity/duration was not conducted due to the budget constraint of the project. Since it is typical the case that WTP for 6day symptom is not twice as much as (usually less than) the WTP for 3-day symptom due to diminishing marginal utility and/or substitution effect, caution should be taken when applied to any policy evaluation. In order to calculate the welfare gains from the new ambient air quality standard in terms of minor respiratory illness, we adopted the exposure-response function of PM10 obtained by Ostro (1994). Once this function is available for our study locations or for Turkey, we can update our welfare calculation based on a more locationspecific value. 7. Conclusion The health benefits from air quality improvements should be included in the policy assessments on pollution reduction, and the estimation of monetized health benefits makes it possible for health benefits to be included in benefit-cost analyses of such policies and projects. This study is a case study in Turkey, one of the fast-growing developing countries. Our results provide not only the country-specific WTP and welfare gains for Turkey but also the basis for the differences of evaluated welfares between developed and developing countries and even among developing countries as well. In order to measure the monetized health benefits from the compliance with the EU air quality standard in terms of PM10 (40 mg/m3) from existing emission levels (2009e2011 average of 100, 83, and 64 mg/m3 for Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, PM10 standard level until 2013: 150 mg/m3), we conducted a contingent valuation survey in JuneeJuly 2012 to estimate the WTP for avoiding minor-respiratory illnesses (throat pain and severe coughing for three days) within a month with a total of 1362 respondents. The median WTP for avoiding one episode of the described minor respiratory symptom within one month lies in the 77e126 TL (51e83 PPP-adjusted 2012 USD) range. In addition to household

Kutahya-Tavsanli

Ankara

Pooled

64 24

82 42

4.39 126

7.69 94

606 [527e683]

553 [465e639]

723 [661e792]

220 985 2 426 415

307 680 2421 134

4 007 860 17 602 521

4536 525 34 885 877

238

186

2216

3280

[199e284] 157 [131e188]

[162e210] 123 [107e139]

[1864e2561] 1464 [1231e1692]

[3000e3593] 2167 [1982e2373]

income, the experience of missing work due to the minor symptoms and having coughing episode within one month affect WTP positively, while using coal as the main winter heat source negatively affect WTP in one or more study areas. Health benefits in terms of reduction of the cases of minor respiratory illnesses as the result of PM10 reduction to the EU air quality standard level are found to be the reductions of 4e11 cases of minor respiratory illnesses per person. WTPs for avoiding a case of minor respiratory illness with the discomfort level of “2” are estimated as 98 (Afsin-Elbistan), 77 (Kutahya-Tavsanli), and 126 TL (Ankara), and the individual welfare gains based on the estimated WTP and the reduced cases of respiratory illnesses are derived as 553e1076 TL per person depending on the location. Considering that the average medical costs actually spent for the minor symptom experienced within a month and lost income due to the illness are 21, 11 and 11 TL and 20.5, 0.53 and 1.16 TL for AfsinElbistan, Kutahya-Tavsanli and Ankara, respectively, together with our estimated WTP for avoiding one acute respiratory episode of 98, 77 and 126 TL for each area, we can conclude that WTP could be decomposed into 21, 14, and 8.6% of medical costs, 20.3, 0.3, and 1% of lost income and 59, 86, and 90% for avoiding discomfort due to the set of minor symptoms for Afsin-Elbistan, Kutahya-Tavsanli, and Ankara, respectively. Moreover, the result of Model 2 provides only weak evidence of the influence of medical costs (statistically significant at 5% level only in Kutahya) and lost income (statistically significant at 10% level only in Afsin-Elbistan) on WTP. In other words, large proportions of WTP for avoiding minor respiratory symptoms account for non-financial losses, and it proves that the calculation of health benefits for the evaluation of new environmental policies should include not only avoided financial losses such as medical cost and lost income but also the value of avoided pain, discomfort and restricted daily activity.

Competing interests The authors declare that they have no competing interests.

Authors' contributions CT helped designing the questionnaire, translated the questionnaire, organized the survey versions and interviewers, and implemented the initial data analysis. SA designed the questionnaire, supervised data collection, conducted data analysis and prepared the manuscript.

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008

10

S. Ara, C. Tekes¸in / Atmospheric Pollution Research xxx (2015) 1e10

Conflict of interest The authors declare that they have no conflict of interest. Acknowledgments This work was supported in part by the Scientific and Technological Research Council of Turkey with the grant 108K474 under the National Young Researchers Career Development Program. Appendix A. Hypothetical question for minor symptom Suppose you were told that, within one month, you would experience “three days of throat pain with serious coughing”. These symptoms may restrict your daily activities such as going to work/ school/shopping/leisure or house works such as cleaning/cooking/ taking care of children. Imagine the level of pain and discomfort is “2” in the pain scale below.

Q1. Have you experienced the symptoms with similar level of discomfort before? [ yes/no] If yes, when was the last time you had the symptom?

1. within a week

2. within a month

3. within three months

4. within a year

5. other

Q2. What do you do when you experience such symptom?

1. 4. 5. 6.

Continue with normal life 2. Take medication 3. Visit a doctor/hospital Get sick days from your work and stay at home (if you are working) Stay at home without going out (if you are not working) Other

Q3. If you were told that by paying certain amount of money, you can avoid having the symptom [three days of throat pain with serious coughing], how much would you pay to avoid the illness episode entirely for the coming one month? By agreeing to pay, you are avoiding all of your pain, suffering and discomfort all together with all the expenses paid for medication and seeing a doctor and the time you

spent visiting the doctor. You also do not have to miss work (therefore, you can avoid the possible reduction in your income), leisure or daily activities. Remember that if you pay to completely avoid being sick this time, you have to give up some other use of this money. Would you pay [ ] TL to avoid getting the symptom this month? Please remember that if you agree to pay, you may have to give up some of the planned expenditure for the activities such as going to watch a movie, eating out at a restaurant, buying new clothes.

1. Yes

2. No

3. Don't know

References Alberini, A., Krupnick, A., 1998. Air quality and episodes of acute respiratory illness in Taiwan cities: evidence from survey data. J. Urban Econ. 44, 68e92. Alberini, A., Krupnick, A., 2000. Cost-of-illness and willingness-to-pay estimates of the benefits of improved air quality: evidence from Taiwan. Land Econ. 76 (1), 37e53. Alberini, A., Cropper, M., Fu, T.T., Krupnick, A., Liu, J.T., Shaw, D., Harrington, W., 1997. Valuing health effects of air pollution in developing countries: the case of Taiwan. J. Environ. Econ. Manag. 34, 107e126. Berger, M.C., Bloquist, G.C., Kenkel, D., Tolley, G.S., 1988. Valuing changes in health risks: a comparison of alternative measures. South. Econ. J. 53, 963e984. Chestnut, L.G., Colome, S., Keller, L.R., Lampert, W., Rowe, R., Wojciechowski, S.L., 1988. Heart Disease Patients' Averting Behavior, Costs of Illness, and Willingness to Pay to Avoid Angina Episodes. EPA Report230-10-88-042. EPA, Office of Policy Analysis, Washington, p. 164. Chestnut, L.G., Ostro, B.D., Vichit-Vadakan, N., 1997. Transferability of air pollution control health benefits estimates from the United States to developing countries: evidence from the Bangkok study. Am. J. Agric. Econ. 79 (5), 1630e1635. Cropper, M.L., Freeman, M.A., 1991. In: Braden, J.B., Kolstad, C.D. (Eds.), Valuing Environmental Health Effects, in Measuring the Demand for Environmental Quality. Elsevier, Amsterdam, pp. 165e212. Dickie, M., Gerking, S., 1991. Valuing reduced morbidity: a household production approach. South. Econ. J. 57 (3), 690e702. Haab, T.C., McConnell, K.E., 2003. Valuing environmental and natural resources: the econometrics of non-market valuation. Edward Elgar Publishing, MA, USA, pp. 23e58. Hammitt, J.K., Zhou, Y., 2006. The economic value of air-pollution-related health risks in China: a contingent valuation study. Environ. Resour. Econ. 33, 399e423. Harrington, W., Portney, P.R., 1987. Valuing the benefits of health and safety regulation. J. Urban Econ. 22, 101e112. Johnson, F.R., Banzhaf, M.R., Desvousges, W.H., 2000. Willingness to pay for improved respiratory and cardiovascular health: a multiple-format, statedpreference approach. Health Econ. 9, 295e317. Liu, J.-T., Hammitt, J.K., Wang, J.-D., Liu, J.-L., 2000. Mother's willingness to pay for her own and her child's health: a contingent valuation study in Taiwan. Health Econ. 9, 319e326. Ostro, B., 1994. Estimating the Health Effects of Air Pollutants: a Method with an Application to Jakarta. Policy Research Working Paper 1301. The World Bank, Washington, USA, p. 70. Rowe, R.D., Chestnut, L.G., 1985. Valuing Changes in Morbidity: WTP versus COI Measures. Energy and Resource Consultants, Inc, Boulder, Colorado, p. 27. Tolley, G., Babcock, L., Berger, M., Bilotti, A., Blomquist, G., Brien, M., Fabian, R., Fishelson, G., Kahn, C., Kelly, A., Kenkel, D., Krumm, R., Miller, T., Oshfeldt, R., Rosen, S., Webb, W., Wilson, W., Zelder, M., 1986. Valuation of Reductions in Human Health Symptoms and Risks. Report for U.S.EPA grant ACR-811053-01-0. U.S. Environmental Protection Agency, Washington, DC, p. 36.

Please cite this article in press as: Ara, S., Tekes¸in, C., The monetary valuation of acute respiratory illness from air pollution in Turkey, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/j.apr.2015.07.008