Ecology, Economy and Society–the INSEE Journal 1 (1): 43–65, April 2018 RESEARCH PAPER Health Damages from Air Pollution: Evidence from Opencast Coal Mining Region of Odisha, India Tapaswini Nayak  and Indrani Roy Chowdhury  Abstract: The present study examines the productive day‘s loss of local communities in the opencast coal mining region of Angul (Talcher) district in Odisha, due to respiratory illness, by using the dose-response function model. The productive day‘s loss is estimated in terms of restricted activities or work days lost, due to severe respiratory illness (RI), induced by air pollution. Health diaries are analyzed through the seasonal household survey to predict the likelihood of RI- related sickness (in terms of the restricted days) of the residents of the mining region (due to air pollution). Poisson and negative binomial regression are fitted for the purpose of count data analysis. The regression result confirms that there is a positive and significant relationship between the level of air pollution (respiratory suspended particulate matter (RSPM)/particulate matter less than 10 g/m3(PM10) and RI-related sick days, depicting that a reduction in air pollution level (PM10) may cause a reduction in expected number of RI-related sick days in the coal mining region. Further monetary welfare gain from avoiding the RI-related sick days are estimated for the population of Talcher coal mining area, Odisha. Keywords: Air pollution; respiratory illness; health diaries; dose-response function; count data regression. 1. INTRODUCTION Many developing economies with abundant mineral reserves are trapped in a low-level equilibrium, with abject poverty, unemployment, ill health, and  Department of Economics, Kendrapara Autonomous College, Odisha. India 754240; tapaswininayak82@gmail.com   Centre for the Regional Development, School of Social Sciences, Jawaharlal Nehru University (JNU), New Delhi, India-110067; indranirc1@gmail.com Copyright © Nayak and Roy Chowdhury 2018. Released under Creative Commons Attribution-NonCommercial 4.0 International licence (CC BY-NC 4.0) by the authors. Published by Indian Society for Ecological Economics (INSEE), c/o Institute of Economic Growth, University Enclave, North Campus, Delhi 110007. ISSN: 2581-6152 (print); 2581-6101 (web). DOI: https://doi.org/10.37773/ees.v1i1.9 https://doi.org/10.37773/ees.v1i1.9 Ecology, Economy and Society–the INSEE Journal [44] Figure 1: Mineral Growth Trend in Odisha (in last 10 years) (Production in million tons) Source: Odisha Economy survey, 2014-15 poor education. This is the well-documented ‘resource curse‘ or ‗paradox of plenty‘ (Sachs and Warner 1997, 2001). Balancing efficient utilization of mineral resources with ensuring equitable and sustainable development and safeguarding the local environment and societal well-being is often a challenge. In the post-liberalization period, many Indian states are encountering a voracious appetite for energy and infrastructure. The intense interstate competitions trigger the demand for energy and infrastructure, which is needed to attract investment to boost growth. This appetite has led to the rampant extraction and utilization of mineral resources in these states. Mining activities are often accompanied by a drastic jump in mineral rich pollution load (both air and water pollution), widespread soil degradation and forest cover loss (State of Environment Report 2008; David Stuligross 1999). The state of Odisha is mineral-rich. In the post-reform period, Odisha and other mineral-rich states have been struggling to answer a difficult question: should they intensify their extraction of mineral resources in their quest for a high rate of growth at the cost of their environment? Odisha is blessed with an abundance of mineral reserves — 24 per cent of India‘s coal reserves, 17 per cent of iron ore, 98 per cent of chromite, 51 per cent of bauxite, 35 per cent of manganese, and 92 per cent of nickel ore (Odisha Economic Survey 2014–15). The state has the second largest reserve of coal in India. The share of mining in Odisha‘s gross state domestic product (GSDP) has escalated significantly to 7 per cent of GSDP in 2014–15 (Odisha Economic Survey 2014–15). The state ranks first in [45] Tapaswini Nayak and Indrani Roy Chowdhury Figure 2: Production of Major Minerals in Odisha, 2013-14 (in Rs. lakh) Source: Odisha Economic survey, 2014-15 country in terms of value of coal reserves and extraction. The state has witnessed an unprecedented increase in coal production from 8 million tons in 2003–04 to 60 million tons in 2013–14. Figures 1 and 2 depict the production recent growth trends of major mineral resources in Odisha. Coal alone accounted for 70 per cent of the mineral royalty received by the state for each of the past four years (Odisha Economic Survey 2014–15). Odisha is one of the richest biodiversity regions in South-East Asia (State of the Environment Report Odisha 2007). In terms of mineral deposits and production, the state occupies a vital position in the country (Murthy and Rao 2006; Economic Survey of Odisha 2014–15). The strong performance of the mining sector in terms of both production and value has thus resulted in a significant growth of the state‘s average GSDP (Economic Survey of Odisha 2014–15). The investment share of the mining sector in total outstanding investments makes Odisha the fifteenth highest recipient of foreign direct investment (FDI) among all Indian states. The mining revenue receipt has been steadily increasing over the past decade (Odisha Economic Survey 2014–15). A report by the State of Environment (2008) has declared all the coal belts of the state most unsafe for human health and living conditions. The neighbourhood of the coal belt witnesses alarmingly high concentration of suspended particulate matter (SPM) and respirable particulate matter (RPM) exceeding the national standard level by several folds. Moreover, WHO (2002) suggests that coal mining process which release toxic pollutants can adversely affect the air quality in the surrounding environment. Coal is an important source of fuel and power and is therefore critically important for the growth and development of a transition economy like Odisha‘s. Moreover, coal has the highest forward linkage with a number of other industries. Therefore, extraction of coal is an attractive option for a state for the creation of direct and indirect employment opportunities, foreign exchange earnings through FDI, generation of mineral royalties, tax revenues, etc. However, coal production is treated as ‗dirty industry‘ as it is the most polluting natural resource (Elizabeth and York 2012; Michelle 2014). Apart from the direct impact of the occupational hazards of the coal miners, coal mining activities, particularly open cast coal mining, imposes Ecology, Economy and Society–the INSEE Journal [46] highly negative environmental externalities (through air pollution, water pollution, contamination and loss of fertility of land, and forest loss and degradation) and health externalities (in terms of poor quality of health among the local communities, living in the proximity of opencast coal mining region). Mining and extraction activities create irreversible damages to the local environment, having long-run future consequences. From mining to combustion of coal, all the intervening processes release various toxic pollutants (such as RSPM (PM10)), carbon dioxide sulfur dioxide (SO2) which adversely affect the air quality in the surrounding environment (WHO 2002). These huge amounts of wastages and pollutants (along with fly ash, coal dust, and heat) have dangerous health impact for the human beings living in the neighbourhood. Thus, from the policy perspective, or for state planning, it poses a serious challenge, in terms of a trade-off between the overall growth and development of the state vis-a-vis environmental degradation and health concerns of the local communities. Given the economic importance of coal, for a state like Odisha, forgoing the mineral extraction activities implies forgoing immediate opportunities for the entire state seems exorbitantly costly at the present stage of its development. As a consequence, the environment and health concerns of local communities in the mining region remain on the back burner. Opencast coal mining leads to more severe air pollution than underground mining (Ghose and Banerjee 1995). All the mining processes, starting from drilling to transportation and screening, are the major sources of such emissions (Pathak et al. 2007). Air pollution in opencast coal mines is mainly due to the toxic emission of particulate matter and gases including methane, sulfur dioxide, and oxides of nitrogen. These air pollutants reduce air quality and ultimately affect people around mining areas (Nanda and Tiwari 2001). The major air pollutants produced by opencast coal mining are SPM and RPM (Sinha and Banerjee 1997; CMRI 1998). Mine fire is also considered as a major source of air pollution in some of the coalfields. The problems may have local, regional, and global impacts. Dust generated by opencast coal mining also affects agricultural productivity (Mishra 2008). Several research findings (Gupta 2006; Chowdhury and Imran 2010) show high concentrations of lower atmospheric pollution, especially RSPM(PM10), adversely contribute to human morbidity, increase respiratory syndromes, and reduce lung functions. Longer term exposure to air pollution (PM10) may lead to irritation, asthma, high blood pressure and heart diseases (Pope et al. 1995, 2007). High concentration of RSPM increase respiratory diseases such as chronic bronchitis and bronchial cases, while gaseous emissions contribute towards global warming, acid rain, and ground-level ozone besides causing health hazards to the exposed [47] Tapaswini Nayak and Indrani Roy Chowdhury population (Pathak et al. 2007). Therefore, health risk and morbidity due to air pollution is high where opencast coal mining is predominant. Pattanayak et al. (2011) also observe that in Keonjhar, Odisha, local communities are vulnerable to diseases such as acute respiratory illness (RI), gastric, tuberculosis, and skin diseases, due to high exposure to toxic elements. All these health problems no doubt have economic costs arising from the expenses incurred in treating the diseases and productive days lost (Ostro 1994; Banerjee 2001). Traditionally, opportunity costs of human health was interpreted solely on account of the individual‘s physical well-being without considering the environmental and social backgrounds around him (Rapport and Mergler 2004). From the 1970s, there has been a departure from the traditional belief, and ‗health‘ is recognized as a precious asset in itself and an absolutely essential ingredient for human development and productivity (Grossman 1972). Good health with mental well-being is termed both a ‗resource‘ and ‗means‘ for stimulating economic development, and an outcome of economic development (Brundtland 2002); the WHO (2005) argues that ‗quality of health‘ is related not only to freedom from diseases but also associated with a healthy and hygienic environment. Therefore, this paper attempts to explore the linkages between health and the environment in the opencast coal mining area in Angul district of Odisha. This linkage paves the pathway for the significance of environmental quality in determining the quality of health and well-being. Given the enormous economic cost involved in terms of productive days lost and treatment of RI-related diseases caused by coal-mining activity, it is critically important to estimate the health impact of air pollution in the opencast coal mining region. This paper is organized as follows. Section 2 reviews the literature. Section 3 explains the dose-response methodology that depicts the association between human health and air pollution. Section 4 presents the household survey design. Section 5 describes the econometric methodology. Section 6 presents the empirical estimation of RI-related sick days and discusses the regression results. Section 7 discusses the evaluation of the welfare gain of air pollution reduction. Finally, Section 8 presents the main findings and conclusions. Ecology, Economy and Society–the INSEE Journal [48] 2. LITERATURE REVIEW The seminal work by Grossman (1972) pioneered the formulation of a theoretical model of household production function. He analyses the demand for health by developing human capital model, where he suggests ‗health‘ as a durable human capital stock and medical care as an important factor in its production. Cropper (1981) added pollution as a crucial ingredient and calculated the health damage reduction from the reduction in the air pollution level. Ostro (1994) suggests the a three-step method of calculating the health benefits of reducing air pollution. The first is to find out the dose-response relationship that shows the health outcome of air pollution. The second is to set the target group population, who are victimized, and exposed to the specific air pollution effect. The third is to calculate the change in air pollution level by implementing a specific pollution-reducing technique. The more recent studies have attempted to explore a relationship between air pollution and the occurrence of illness by using dose-response and damage functions, while some economists have calculated the health damage cost of air pollution levels using one of three methods: cost-of- illness (COI), dose-response, and contingency valuation (Alberini et al. 1997; Ostro 1994; Krupanik 2000). In the case of South Asia, Murty et al. (2003) and Gupta (2008) calculated the health benefits from the reduction in urban air pollution level in the Indian states. Mishra (2012) estimated the damage to human health and agricultural productivity loss from pollution induced by coal-mining activities in Odisha. Chaulya (2003) shows that the annual average SPM and RSPM (PM10) concentration in the Ib valley coalfield of Odisha exceeds the standards set in the Indian National Ambient Air Quality Standards (NAAQS) protocol for both residential and industrial areas. The problem is not limited to these areas alone; air ambient quality is critical in the state‘s entire mining belt. The literature on estimating the morbidity and health impact costs are mostly confined to urban air pollution, vehicular pollution, etc. There are some studies on the health impact of air pollution in the cement industry (Bogahawatte 2008) and iron mining (Pattanayak et al. 2011), and of the occupational hazards (i.e., direct impact) of coal mining (Sarkar et al. 2004), but there are only a few studies to the best of our knowledge on the health impact of air pollution induced by opencast coal mining in Odisha (Mishra 2012; Singh et al. 2010; Hota and Mishra 2010). Most studies address the health concerns from air pollution and the pollution load in coal mining areas, but they do not address many important questions. [49] Tapaswini Nayak and Indrani Roy Chowdhury Table 1: Categorisation of Ambient Air Quality Status with respect to criteria pollutants based on prescribed standard and exceedance factor Pollution level Residential/Industrial RSPM(PM10) (µg/m3) PM2.5 (µg/m3 ) Low (L) 0-30 0-20 Moderate (M) 30-60 20-40 High (H) 60-90 40-60 Critical (C) >90 >60 Source: State Pollution Control Board, Bhubaneswar Therefore, this study tries to bridge the research gap by examining the association between the air pollution level and negative health outcomes (we consider RI for this study)1 in the opencast coal mining region (Angul- Talcher) of Odisha. This study also tries to analyse the economic impact of negative health effects by considering the loss of wages of productive days for the study region. We raise some unanswered questions. (1) To what extent does the coal mining activities contributed to the health-related well-being of the local community in the mining region and to their environment? (2) Does the critical ambient air quality near the opencast coal belt matter to the health of the neighbouring community? (3) How do the socio-economic factors affected the health stock of the local communities in the neighbourhood of coal mining areas? (4) How many productive days are lost by the local communities due to RI related diseases on account of such negative externalities ? Against this backdrop, the study area is located in a coal mining district Angul of Odisha, in the eastern part of India. The state has two coalfields namely- Talcher coalfield (in Angul district) and lb River coalfield (in Jharsuguda district).The coalfield of Ib valley is comparatively new compared to that of Talcher coal field(Directorate of Mines, Government of Odisha, Bhubaneswar). The older the coal mines project, the more is its environmental impact. The numbers of active opencast coal mine projects are also larger in Talcher coal field area than the IB valley coal field. As per the latest report by Central Pollution Control Board (CPCB 2009) on Comprehensive Environment Assessment of Industrial Clusters, Talcher coal filed region of Odisha has secured seventh position as one of the most 1 RI includes all the common acute syndromes of respiratory diseases such as: bronchial asthma, cough with mucus, and sputum production, respiration allergy to dust, allergic cough, running nose, headache, flu, and fever, chest tightness, pain, and acute bronchitis. These symptoms are identified as commonly reported respiratory illness after consulting the specialist doctors from the Mahanadi Coal Field Hospital, located in the study region. Ecology, Economy and Society–the INSEE Journal [50] critically polluted industrial cluster in India.2 Keeping in mind of all these perspectives we select the Angul-Talcher coal mining region as our study region3. Table 2: Ambient Air Quality of the study region 2015-16 Month Station 1 (MCL) Station 2 (TTPS) Station 3 (NALCO Nagar, Angul) Station 4 (Industrial Estate, Angul) Average Average Average Average RSPM (PM10) PM2.5 RSPM (PM10) PM2.5 RSPM (PM10) PM2.5 RSPM (PM10) PM2.5 January 171 130 154 122 137 98 137 94 February 186 99 165 115 117 80 108 80 March 161 87 136 76 129 65 118 74 April 177 105 91 53 108 57 92 51 May 149 93 123 79 97 80 95 50 June 140 87 71 44 78 38 84 47 July 126 72 56 34 80 47 88 53 August 100 57 61 34 79 39 65 49 September 87 39 70 51 100 49 96 57 October 133 48 95 75 91 52 98 54 November 135 75 126 68 105 57 105 54 December 134 79 164 86 107 54 105 56 Average Annual 140 81 111.08 69.75 102 57 99.25 60 Source: CPCB (2009) 3. DEFINING THE RELATIONSHIP BETWEEN HUMAN RESPIRATORY HEALTH AND AIR POLLUTION: DOSE- RESPONSE FUNCTION Often, the relationship between air pollution and premature mortality is studied by using time-series analysis of daily observations of number of deaths and the pollution level. These studies capture the effects of the short-term exposure to pollution on the probability of dying. But it is expected that people, who are in a weak physical state or who have a history of chronic diseases are more susceptible to contact recurring diseases due to the high concentration of pollution in the environment and are therefore supposed to be the most vulnerable. These types of exposure can be captured through the ‘dose-response function‘. If there is a cause of high 2 The concentration of PM10 in air is very high for the study region which often exceeds the prescribed level (by Central Pollution Control Board) by two folds. See the table no (1) and (2) for detail about the severity of air pollution. 3 See Figure 3 for the district wise extraction and value of coal in Odisha in recent year. [51] Tapaswini Nayak and Indrani Roy Chowdhury Figure 3: District wise break-up of extraction and value of coal in Odisha, 2013-14 (Quantity in lakh MT and value in crore of rupees) Source: Odisha Economy survey, 2014-15 concentration of pollutants in the surrounding environment, then it may trigger some negative health outcome (i.e., negative externalities of environmental pollution) as its effect. This association between cause and effect has been known as ‗dose-response function‘.4 This function captures a range of responses towards the respiratory illness of different individuals due to different degree of outdoor pollutants, when other confounding socio-economic factors and the level of indoor pollution are controlled. This function therefore relates the importance of a ‗Dose‘ (that is the intensity of air pollution to which the local people are exposed) to the ‗responsiveness‘ of the individuals (the response may be any kind of morbidity due to the respiratory diseases, stress, and even death or mortality). The degree of impact can be captured as the proportion or the days of sickness or restricted activity days due to respiratory illness syndromes (H).This method estimates the partial derivative (or the slope of b) of ‗Dose-Response‘ function as given in the following equation (1). The purpose is to calculate the marginal effect of a change in occurrence of a given health outcome related to a change in air ambient quality (A) with other confounding factors. The next step by Ostro (1994) is to calculate the numbers of people who are supposed to be exposed to the particular air pollution effect (i.e. POPi).Then how many productive days (restricted activity days or sick days), they are losing because of the exposure. Finally, the last step is to 4See Ostro (1983, 1987, 1994), who estimated the ‗Dose-Response Function‘ to find out the impact of air pollution on morbidity and found that PM10, affect both the restricted activity days and work days lost. Ecology, Economy and Society–the INSEE Journal [52] calculate the welfare gain due to a unit change or reduction in ambient air quality (dA). Thus the dose-response function can be expressed as (following Ostro 1994): dAPOPbdH ii * * (1) Where, i dH = change in individual‘s risk of health outcome i b = slope from the ‗dose-response‘ function. i POP = targeted population at risk of health impact i dA = change in ambient air quality under consideration 4. SURVEY DESIGN The present study is based on the primary survey at the household level, though pollution data are collected from the secondary sources such as State Pollution Control Board, Odisha and Regional Pollution Control Board, Angul. To understand the severity of air pollution on the health outcome of the local residents we first conducted a pilot survey followed by five rounds of group discussions during September 2015. Baseline information was collected from that pilot survey from some randomly selected households in the vicinity of Mahanadi Coal Field Limited. The baseline information did confirm the severity of health problem in the region. To capture the exposure of air pollution on human health (for those who are residing near the coal field area) through ‗dose-response‘ method, we required background information on the health status of individuals residing in the study region along with other environmental and socio- economic information. The pilot survey helped us to design a well- structured questionnaire for the final survey. We conducted the survey during September 2015 to January 2016, involving (210) households of the 10 neighbouring villages, of Angul-Talcher opencast coal mining region of Mahanadi Coal Field Limited (MCL). Following Krupanick (2000), Gupta (2008), Murty et al. (2003), Adhikari (2012) and Imran et al. (2010), the survey kept track of weekly health diary of the selected households for six weeks extended over two seasons. We followed two stage stratification, for selecting sample (following Gupta 2006 and Adhikari 2012). The reason for following a two stage stratification sampling procedure is to capture precisely the local people‘s exposure to air pollution and the consequent adverse health impact. In the first stage of stratification, the villages within a four-kilometre radius from the air pollution monitoring station were identified (and a total 10 villages were selected. We then randomly selected [53] Tapaswini Nayak and Indrani Roy Chowdhury 21 households from each village. Thus a total of (21 x 10 = 210) households from 10 villages and an aggregate of 855 members were surveyed5. Out of 855 household members 254 members were suffering from RI symptoms during the first round of the survey. Given the seasonal differentiates in air quality levels, we collected health diary information on 254 household members repeatedly for six weeks during two different seasons (i.e. October or post-monsoon and December or winter). The health diaries covered the RI-related health status, of mitigating expenditure, chronic illness etc. Therefore the survey resulted in a pooled data, containing 1524 (= 254 x 6) observations. But by using the Stata software 12.0, we found only 932(restrictive sample) observations having RI diseases, sick days, and mitigating expenses over the entire 6 weeks. Given that the present study is focused on the number of RI-related sick days, so a total of 932 observations on the basis of availability of full information could only be taken for empirical investigation. 5. EMPIRICAL ESTIMATION OF RI-RELATED SICK DAYS Our goal is to estimate the number of sick days or restricted activity days of a person suffering from respiratory diseases. This dependent variable counts the duration of sickness episode which is related to the particular respiratory syndromes (associated with air pollution) based on one week recall period6 .The dependent variable counts how many days in a week (based on one week recall period), an individual is unable to work or restricted from other activities due to respiratory illness. As our dependent variable i.e., ‗sick days‘ is finite number of non-negative integers (small discrete values, say 0 1 2, 3, 4, …) or count values, assuming 0 values in several observations, we need a probability distribution that takes care of this count data. We used the Poisson and Negative Binomial regression to model the count outcomes i.e., the negative health impacts of air pollution (captured by the number of restricted activity days) as experienced by the respondent in a given time period. 5 The sampling framework takes into account the randomly selected households from each of the 10 villages, which are located within a stipulated area i.e., 4 km radius surrounding the station. So the selection bias is likely to be minimum and therefore can be comfortably ignored. Here proximity is used to identify the neighbourhood areas which are very close within the radius of 4 km from the opencast coal mines. 6 For the primary survey we collected information regarding the prevalence of RI-related diseases and the mitigating expenditure incurred by the household and therefore we set the recall period of one week. This is deliberately done to extract correct information on the mitigating expenditure data. Ecology, Economy and Society–the INSEE Journal [54] Given that Poisson regression is the starting point of count data analysis, we initially fitted the Poisson regression to the number of sick days by using the method of Pseudo Maximum Likelihood Estimation or Quasi- Maximum likelihood estimation. It is possible to write the probability of number of sick days(Y) for a respective individual, using the Poisson regression model as: Prob. ....7,...23,1,0, !           it it itit it it it y y ey x y Y it (2) where x i e    Here it y denotes the RI-related sick days or restricted activity days (related to air pollution level) for an individual i at time period t. ‗e‘ is the exponential function of itit y  , the mean value of the number of sick days due to pollution.  is the vector of regression coefficients while itx is the vector of independent variables. And it u is the unobserved error term that may affect the individual health stock. The estimated mean value is therefore: x t e   ˆˆ  . A log transformation gives the equation: x i  ˆˆln  As the count data variance usually exceeds the mean, the Poisson model has its limitation arising from the property of equi-dispersion. Alternatively we also run the negative binomial model (NBM) as an alternative model which accounts for the overdispersion in count data, when the conditional variance surpasses the conditional mean. The only difference between two is that NBM has an ancillary parameter (α) to model the overdispersion. 5.1. Choice Variables The vector of explanatory variables comprises of daily PM10 data obtained from the nearest air pollution monitoring station from the respondents' residence. The study has narrowed down the ‗daily (PM10)‘ as the single- most important indicator for air pollution in the coal belt, as the exposure to PM10 is considered to be most damaging impact on respiratory health (Gupta 2006; Chowdhury and Imran 2010).7 7 Several research findings show high concentrations of lower atmospheric pollution especially RSPM (PM10) contribute to human morbidity, increase respiratory syndromes, and reduced lung functions (Gupta 2006; Chowdhury and Imran 2010). The secondary data on air pollution shows that in open cast coal mining areas in Odisha, the PM10 level is [55] Tapaswini Nayak and Indrani Roy Chowdhury Table 3: Example of Variables and Categories Used Coal mining externalities (Due to air pollution) Defining variables Original Questions Health Adult outdoor exposure and workplace information Based on the questions regarding workplace location, number of working hours/day, affected by coal fumes/dust, loss of income/day if work lost due to RI related sickness Children outdoor exposure Based on questions regarding number of hours spent daily in school/college/other places and therefore whether exposed to coal fumes/dust or not. Occurrence of respiratory illness To capture the respiratory health status- different types of RI related diseases and its frequency of occurrence in daily and weekly basis, based on the recall period in the previous week and in last year were asked. Health effects Status of health with the level of air pollution Health cost Income lost due to loss of productive/restricted workdays Annual income Income calculated from different sources Per capita income Households total income from different sources divided by the number of household members Indoor pollution Captured by the questions on the nature of fuel use, use of chimney/exhaust fan in kitchen, disposal of solid waste? Ownership status Different types of assets from different sources Source: Author‘s own survey schedule We control for a bunch of variables 8 including pre-existing illness (from previous week), chronic diseases,9 exposure to outdoor pollution or coal alarmingly high and often exceeds the national prescribed level by Central Pollution Board by several fold. 8 Table 3 defines the main variables and its original question which were asked at the time of survey. 9 When one individual has been suffering from the diseases like asthma, blood pressure, heart diseases, and tuberculosis for more than 1 year, then that person is to be treated to have a chronic illness. The chronic disease of an individual describes his health stock. If an individual has been suffering from a chronic disease then it is expected that he is more Ecology, Economy and Society–the INSEE Journal [56] Table 4: Descriptive Statistics of Independent Variables Variable Description Obs. Mean Standard Dev. Minima Maxima PM10 Average Weekly PM10 (ug/m3) 6 148.68 13.79 132.5 167.5 Age Age of Household Members in Years 932 38.40 20.36 1 82 Edu_inyears Education in Years of Schooling (for illiterate) 932 6.58 4.08 0 15 HH_size Household Size 210 4.06 1.15 1 8 HH-Income Household Income in Rupees. 932 19062 12564 1500 80000 PCI Per capita Income in Rupees. (Household Income divided by Number of household Members) 932 4806 3094.07 700 20000 Source: Primary Survey, 2015-2016 dust, 10 smoking etc., to avoid overestimation of pollution exposure on health effects of the respondents. Besides these individual characteristics vulnerable to air pollution exposure and is therefore supposed to have higher number of illness related sick days. This is used as control variable in our econometric analysis, which captures the occurrence of chronic illness among the individuals. It assigns a value ‗1‘ if the respondent reports a chronic diseases (as mentioned above) otherwise it assigns a value ‗0‘. 10As discussed, coal is one of the most polluting natural resources (because it emits toxic pollutants such as carbon dioxide, sulfur dioxide, respiratory suspended particulate matters etc. to atmosphere). The brittle nature of coal creates coal dust in the coal mining activities including loading unloading, transportation etc. In the open cast coal mining neighbourhood villages, the suspended coal dusts often create haze in the atmosphere which often interrupts visibility. Given these facts it is evident that regular exposure of the air borne contaminants due to coal and coal dust may trigger respiratory illness. It is anticipated that more outdoor exposure may increase the probability of illness. The exposure to coal dust is a dummy variable which assigns 1 if the representative individual is exposed to outdoor coal dust (or stay in outdoor more than 1 hour a day), otherwise it is 0 (zero). [57] Tapaswini Nayak and Indrani Roy Chowdhury Table 5 Poisson Regression Model Dependent Variable (sick days) Poisson Independent variables Coefficient(P-value) Standard Error PM10 0.0059(0.001)*** 0.1907 Age -0.0004(0.887) 0.0033 Age square 6.2307(0.989) 0.0004 Gender -0.061(0.178) 0.0454 HH size 0.021(0.191) 0.0156 Exposed_coaldust -0.057(0.178) 0.0426 Awar_polutnhelth -0.139(0.001)*** 0.0436 Asthma_cronic 0.123(0.001)*** 0.0378 B.P -0.004(0,383) 0.0478 Indiv_smoking 0.105(0.079)* 0.0604 Fuel_coaldumy 0.071(0.081)* 0.0413 Prevsweek_ill 0.123(0.001)*** 0.0370 Prevsweek_expend. -0.052(0.146) 0.0361 Exercise -0.028(0.539) 0.0466 PCI 0.042(0.215) 0.0343 Constant_ -3.260(0.001)*** 0.9934 Log pseudo likelihood -1914.7761 Wald ch2(df) (15) 95.28 Prob.>chi2 0.000 Pseudo R2 0.024 No of observation 932 Source: Authors Note: ***, ** and * indicate significance at 1%, 5% and 10% level respectively. (like age, gender, education, awareness about adverse health effect of pollution, per capita income level, and exercise habits) and household characteristics (household size and medium of cooking, fuel or indoor pollution), that may affect the health outcomes are controlled in the regression analysis.11 11 We have presented the descriptive tables of independent variables in table 4, which includes only the continuous and categorized independent variable‘s descriptive statistics. We have not included some independent variables viz., gender, chronic disease, exposure to coal dust, or outdoor pollution, smoking habit, awareness regarding effect of pollution on health, coal as a fuel, exercise and previous week illness and illness related expenditure‘s descriptive statistics as these variables has taken as dummy variable in regression model. Ecology, Economy and Society–the INSEE Journal [58] 6. REGRESSION ANALYSIS The regression result of the Poisson and NBMs are given in Tables 5 and 6, which display the regression coefficients, standard errors, and probability values (z>p) for each explanatory variable. In table (5) and (6), both the regression model (Poisson and NB) present approximately similar results, because the ‗α‘ value (the parameter which accounts for overdispersion) is very small (i.e., - 0.0103) in NBM. The only difference is that the Poisson model gives the robust standard error (as it is based on pseudo maximum likelihood method and the NBM accounts the overdispersion as it has one ancillary parameter i.e., α. The regression results confirms our hypothesis that there is a positive and significant relationship between the level of air pollution (RSPM/PM10) and RI-related sick days. The coefficient of PM10 (0.0059) is positive and statistically significant at 1 per cent level of significance (p-value is 0.001 or equivalently t-value is 4.22) in the Poisson model. The positive coefficient of pollution variable depicts that a reduction in air pollution level (PM10 level) will cause a reduction in expected number of RI-related sick days. The coefficient of awareness regarding pollution related health impact is negative and significant at 1 per cent level; showing a decrease in sick days due to the rise in awareness level. The coefficient of chronic diseases asthma is positive and significant at 1 per cent level, thereby presenting an increase in sick days as the chronic asthma increases. The other socio- economic and control variables such as individual smoking habits, coal- fuel12 and illness in previous week have positive coefficients as expected and are also statistically significant.13 The coal as a daily household fuel has been controlled for indoor air pollution. This variable is statistically significant and positive impact on aggravating the respiratory illness and its related sick days. The individuals who have active smoking habits are more susceptible to the exposure of air pollution and thus respiratory illness related sick days increases for them. 12 The frequency of using ‗coal‘ as a cooking fuel contributes to the level of indoor air pollution. The study assumes it as a dummy variable giving the value 1 if a representative household uses coal for their cooking regularly. Here, we consider if a household uses coal as a fuel for cooking more than 20 times in a month, then we take the value 1 and it takes 0 otherwise. 13 We have checked the multicollinearity and heteroscedasticity test for the model after running the regression and found these problems among the independent variables are not a serious concern for the model. The Wald chi2 (15) is 95.28 with p-value 0.000, which indicates the model is significant as a whole. [59] Tapaswini Nayak and Indrani Roy Chowdhury Table 6 Negative Binomial Regression Result Dependent Variable (sick days) Negative Binomial Independent variables Coefficient(P-value) Standard Error PM10 0.0059 (0.000)*** 0.1942 Age -0.0004(0.891) 0.0034 Age square 4.5107(0.992) 0.0004 Gender -0.061(0.186) 0.0463 HH size 0.021(0.200) 0.1595 Exposed_coaldust -0.057(0.185) 0.4341 Awar_polutnhelth 0.139(0.001)*** 0.0444 Asthma_cronic 0.123(0.001)*** 0.3855 B.P -0.004(0,390) 0.4875 Indiv_smoking 0.105(0.084)* 0.6159 Fuel_coaldumy 0.071(0.091)* 0.0421 Prevsweek_ill 0.123(0.001)*** 0.0377 Prevsweek_expend. -0.052(0.151) 0.3686 Exercise -0.028(0.555) 0.0475 PCI 0.042(0.220) 0.0349 Constant_ 3.266(0.001)*** 1.0110 Lnalpha : -4.567SE(1.4013) Alpha: 0.0103SE (0.1455) LR test of alpha=0 Chibar2(01) Prob.>chibar2 54.43 (0.000) Log likelihood : -1914.5084 LRchi2(df): (15) 85.98 Prob.>chi2 0.000 Pseudo R2 0.022 No of observation 932 Source: Authors Note: ***, ** and * indicate significance at 1%, 5% and 10% level respectively The presence of respiratory illness in the previous week among the individual increases the risk of experiencing more RI-related sick days. 7. WELFARE GAIN (IN MONETARY TERMS) OF THE REDUCTION IN RI-RELATED SICK DAYS To estimate the monetary benefits from the reduction in RI-related sick days, we calculate the marginal effect from the Poisson regression. Marginal effect defines the reduction in RI-related sick days due to the fall in PM10 levels in air. Ecology, Economy and Society–the INSEE Journal [60] Following Gupta (2006), Chowdhury and Imran (2010) and Adhikari (2012), monetary benefits from reduction in RI-related sick days can be expressed as: 7 365 10  PM (3) Where  is the coefficient value of PM10 from Poisson regression model (β = 0.0059) λ is the predicted value of (H) or RI sick days from the Poisson model (λ = 0.0205) (λ= 0.0205, is the predicted value of sick days from the Poisson model, which is calculated by the help of post-regression analysis command in Stata-12.0) ∆PM10 is the change in the level of PM10 from the current to the safe standard prescribed level, (the average current level PM10 in two seasons is 159 µg /m3 and the prescribed standard level by Odisha Pollution Control Board (OPCB) is 90.4 µg /m3 for residential area.) (β * λ) is the marginal gain from Poisson estimation or (0.0059 * 0.0205 = 0.00012) The Poisson regression result shows that 1 µg /m3 reduction in PM10 results in a marginal benefit of 0.00012(days) for a representative household member in a week. In order to calculate the welfare gain in sick days for a representative household member by reduction in PM10 to a safe level, we have to multiply the marginal benefit value by ΔPM10 i.e., 0.00012 * 68.6 = 0.00823 as gain in reduction in sick days for a representative household member. The annual benefit in monetary terms or welfare gain can be calculated by multiplying this value 0.00823 with the 7 365 i.e., 0.4292 as annual monetary benefit for reduction in sick days due to reduction in PM10 level or improvement of PM10. Putting the values in equation (3) 0.0059 * 0.0205 * 68.6 * 52.14 = (0.43) days per year for a representative household member. Here: β = 0.0059, λ = 0.0205, ΔPM10 = 159 µg /m3 - 90.4 µg /m3 = 68.6 µg /m3 and 14.52 7 365  . [61] Tapaswini Nayak and Indrani Roy Chowdhury The calculated average wage (per monthly) of working individual is Rs. 275 from working group sample from the survey data. Thus the annual monetary gain for a representative household member is 0.43 * 275= Rs.118.25 by avoiding the RI-related sick days due to reduction in PM10 level. By extrapolating for the total population of Talcher coal mining area, the monetary welfare gain from avoiding the RI-related sick days are estimated as Rs. 64, 95,51,5.66 per annum. However our study presents lower bound estimation of monetary welfare gain (of health improvement) from the reduction in air pollution in the opencast coal mining of MCL of Talcher area. The study has the limitation of not capturing the averting expenditure of air pollution and the opportunity cost of time spent for seeking care. It poses a serious constraint to gather information on these variables from the survey because of the heterogeneity of the respondents (covering children, women, and old persons other than the working people). Moreover the region is endowed with some active public health centres run by MCL, offering some basic health services to the local community. The out of pocket expenditure which was captured during the survey was mainly the expense incurred by the residents from the private consultation over and above the free services they can avail from primary healthcare centres (PHC). The compulsion to visit the private practitioner was mainly due to congestion, long waiting time, recurrence, lack of doctor, non-availability of medicine etc., as reported in the survey. 8. CONCLUSION For developmental processes, the economic importance of coal is undoubtedly very high, being the cheapest source of energy. However coal is the most polluting natural resource; from production to combustion it leads to the emissions of various harmful gases including methane (CH4), sulfur dioxide (SO2), and nitrogen oxides (NOx), carbon monoxide (CO), particulate matter, various toxins, etc. Besides coal dust, coal wastes, and coal particles are produced during the mining process, soot is released during the transportation of coal. It is evident from various studies that all these pollutants can cause severe and potentially deadly respiratory and lung problems, and cardiovascular diseases. Thus, apart from the direct impact of the occupational hazards of the coal miners, coal mining activities (particularly opencast coal mining), imposes highly negative environmental externalities (through air pollution, water pollution, soil contamination and fertility loss and forest loss and degradation) and health externalities (in terms of poor quality of health among the local communities, living in the Ecology, Economy and Society–the INSEE Journal [62] proximity of opencast coal mining region). Although there is huge economic cost involved in tackling these environmental and health externalities,14 it is seldom addressed seriously at the policy level or in the implementation stage. Thus the pricing of coal involves huge cross- subsidies in India from the unaccounted environmental and health damages. Even the conservative attempt of imputing these negative externalities in the pricing of coal would have increased the price of coal several folds. This primary concern of this paper has been on the impact of air pollution on respiratory health (in terms of their number of sick days or restricted days due to RI) among the residents of some coal mining regions of Odisha. The attempt is to capture the partial impact of air pollution (triggered by coal production) on respiratory health damages. Given that the level of PM10 is always at an alarming level in the Angul-Talcher coal mining region, it is natural to expect that the respiratory health outcome would be very pervasive and critical in the proximity of the coal field. It is evident from the literature that the critical air pollution level (particularly concentration of PM10 in air) affects the respiratory health status (lung and chest problems, cough, bronchial asthma, bronchitis, TB and other lung related diseases) of the residents in the proximity of the coal field. Our empirical exercise predict the likelihood of RI-related sickness (in terms of their restricted days/sick days) of the residents near the open cast coal mines (due to air pollution). We find there is a strong and direct relation between acute respiratory disease related sick days or restricted activity days and the level PM10 (air pollution) in Angul-Talcher coal mining area. The positive value of the coefficient of pollution variable depicts that a reduction in air pollution level (PM10 level) causes a reduction in expected number of RI-related sick days. We find that a representative individual in the Angul-Talcher region can save 0.43 days or Rs. 118. 25 by avoiding the 14 See Epstein et al. (2011, 92). At the Center for Health and the Global Environment, Harvard Medical School, Dr. Paul Epstein and others found that accounting for the full costs of coal would double or triple its price. The study, released in the Annals of the New York Academy of Sciences, tallied the economic, health, and environmental costs associated with each stage in the life cycle of coal—extraction, transportation, processing, and combustion—and estimated those costs, which are borne by the public at large, at USD 175–500 billion annually. Air pollutants from combustion accounted for USD 187.5 billion, mercury impacts as much as USD 29.3 billion. The report estimated the public health burden in Appalachian communities at USD 74.6 billion a year, and that death, injury, and increased healthcare costs constituted much of the impact. The study discussed several other impacts that are not easily quantified: coal mining and processing releases heavy metal toxins and carcinogens into water supplies; coal miners die at work or are injured; and coal mining societies are impacted. [63] Tapaswini Nayak and Indrani Roy Chowdhury RI-related sick days, if there is a reduction of 68.6 µg /m3 of PM10 per year. This is calculated by taking the difference between the average current level of PM10in the two seasons (i.e. 159 µg /m3) and the standard level (i.e., 90.4 µg /m3) as prescribed by the OPCB. Thus the monetary welfare gain from avoiding the RI-related sick days for the entire population of Talcher coal mining area are estimated to be Rs. 64, 95,515. 66 per annum. Although the study derives the lower bound estimates of monetary gain due to the reduction in air pollution, the figures bears some serious policy concerns. REFERENCES Adhikari, N. 2012. ―Measuring the Health Benefits from Reducing Air Pollution in Kathmandu Valley.‖ Working Paper no. WP 69–12, South Asian Network for Development and Environmental Economics, Kathmandu. Alberini, A. and A. Krupnick. 2000. ―Cost of Illness and Willingness to Pay Estimates of the Benefits of Improved Air Quality: Evidence from Taiwan.‖ Land Economics 76: 37-53. https://doi.org/10.2307/3147256 Alberini, Anna, Maureen Cropper, Tsu-Tan Fu, Alan Krupnick, Jin-Tan Liu, Daigee Shaw and Winston Harrington. 1997. ―Valuing Health Effects of Air Pollution in Developing Countries: The Taiwan Experience.‖ Journal of Environmental Economics and Management. 34: 107-26. https://doi.org/10.1006/jeem.1997.1007 Banerjee, S. 2001. ―Economic Valuation of Environmental Benefits and Costs.‖ In Environmental Economics: An Indian Perspective edited by R.N. Bhattacharya, 125-159. New Delhi: Oxford University Press. Bell, S. Elizabeth and Richard York. 2012. ―Coal, Injustice and Environmental Destruction: Introduction to the Special Issue on Coal and the Environment.‖ Organization and Environment. 25(4): 359-367. https://doi.org/10.1177/1086026612468138 Bresnahan, B.W., M. Dickie and S. Gerking. 1997. ―Averting Behaviour and Urban Air Pollution.‖ Land Economics 73: 340-357. https://doi.org/10.2307/3147172 Cameron, A.C. and P.K. Trivedi. 1986. ―Econometric Models Based on Count Data: Comparisons and Applications of Some Estimators and Tests.‖ Journal of Applied Econometrics 1(1): 29-53. https://doi.org/10.1002/jae.3950010104 Cameron, A.C. and P.K. Trivedi. 1998. Regression Analysis of Count Data. New York, USA: Cambridge University Press. https://doi.org/10.1017/CBO9780511814365 Centre for Science and Environment. 2008. Rich Land Poor People- Is Sustainable Mining Possible? New Delhi: Centre for Science and Environment. Chaudhary, T. and M. Imran. 2010. ―Morbidity Costs of Vehicular Air Pollution: Examining Dhaka City in Bangladesh‖. Working paper no. 47-10. South Asian Network for Development and Environmental Economics, Kathmandu. https://doi.org/10.2307/3147256 https://doi.org/10.1006/jeem.1997.1007 https://doi.org/10.1177/1086026612468138 https://doi.org/10.2307/3147172 https://doi.org/10.1002/jae.3950010104 https://doi.org/10.1017/CBO9780511814365 Ecology, Economy and Society–the INSEE Journal [64] Chaulya, S.K. 2003. ―Air quality standard exceedance and management in an industrial mining area‖. Environmental Conservation 30 (3): 266-273. https://doi.org/10.1017/S0376892903000262 Das Vidhya. 1995. ―Development or Destruction? New Mining Projects in Orissa.‖ Economic and Political Weekly. 30 (22): 1281-1282 Dickie, M. And S. Gerking. 1991. ―Valuing Reduced Morbidity: A Household Production Approach.‖ Southern Economic Journal 57 (3): 690-702. https://doi.org/10.2307/1059783 ENVIS. 2015. ―Air Pollution.‖ ENVIS Centre of Odisha‘s State of Environment, Forest and Environment Department, Orissa, Accessed online at http://orienvis.nic.in/index1.aspx?lid=22&mid=1&langid=1&linkid=20 on 4th January 2017. Freeman, A.M. 1993. The Measurement of Environmental and Resource Values. Washington DC: Resources for the Future. Ghose, M.K. and S.K. Banerjee. 1995. ―Status of Air Pollution Caused by Coal Washery Projects in India.‖ Environmental Monetary and Assessment. 38: 97-105. https://doi.org/10.1007/BF00547129 Government of Odisha. 2008. State of Environment Report. Bhubaneswar: Government of Odisha. Government of Odisha. 2015. Economic Survey (2014-15). Bhubaneswar: Planning and Coordination Department, Government of Odisha. Green, W.H. 2003. Econometric Analysis. Fourth Edition. New Delhi: Pearson Education. Grossman, M. 1972. ―On the concept of health capital and the demand for health.‖ The Journal of Political Economy 80 (2): 223-255. https://doi.org/10.1086/259880 Gupta, U. 2006. ―Valuation of Urban Air Pollution: A Case Study of Kanpur City in India.‖ Working Paper no. 17-06. South Asian Network for Development and Environmental Economics, Kathmandu. Khatua, Sanjay and William Stanley. 2006. ―Ecological debt: A case study from Odisha, India: integrated rural development of weaker sections in India.‖ In . Ecological Debt: the People of the South are the creditors: cases from Ecuador, Mozambique, Brazil and India edited by A.K. Peralta 125-165. Geneva: World Council of Churches. Mishra, P.P. 2009. ―Coal Mining and Rural Livelihoods: Case of the Ib Valley Coalfield, Orissa.‖ Economic and Political Weekly 44 (44): 117-123. Mishra, P.P. and Ayan Kumar Pujari. 2008. ―Impact of Mining on Agricultural Productivity: A case study of the Indian state of Odisha.‖ South Asia Economic Journal 9(2): 337-350. https://doi.org/10.1177/139156140800900204 Murty, M. N. S.C. Gulati and A. Banerjee. 2003. ―Health Benefits from Urban Air Pollution Abatement in the Indian Subcontinent‖. Working Paper no. 236 , Institute of Economic Growth, Delhi. Accessed online at http://iegindia.org/upload/pdf/wp236.pdf on 4th January 2017 Nanda, S.N. and S.N. Tiwari. 2001. ―Concentration of SPM in the Burla-Hirakud- Sambalpur region (Odisha).‖ Indian Journal of Environmental Protection 21(3): 193-202 https://doi.org/10.1017/S0376892903000262 https://doi.org/10.2307/1059783 https://doi.org/10.1007/BF00547129 https://doi.org/10.1086/259880 https://doi.org/10.1177/139156140800900204 [65] Tapaswini Nayak and Indrani Roy Chowdhury Nijhuis, Michelle. 2014. ―Can Coal Ever be Clean?‖ National Geographic Magazine April: 29-40. Ostro, B. 1994. ―Estimating the Health Effects of Air Pollutants: A Method with to Jakarta.‖ Working Paper No. 301, Policy Research Department, World Bank, Washington DC. Accessed online at http://documents.worldbank.org/curated/en/355391468752348015/pdf/multi0p age.pdf on 4th January 2017. Ostro, B., J.M. Sanchez, C. Aranda and GS. Eskeland. 1995. ―Air Pollution and Mortality: Results from Santiago, Chile.‖ Working Paper no. 1453, Policy Research Department, World Bank, Washington DC. Pathak, Gopal, Kirti Avishek and A.K. Gorai. 2007. ―Environmental and Social Impacts of Coal Mining and its Management Practices in India.‖ First International Conference on Managing the Social and Environmental Consequences of Coal Mining in India, Organized by Australian National University, New Delhi, November, 19-21. Pope III, C. A. 2007. ―Mortality effects of longer term exposures to fine particulate air pollution: Review of recent epidemiological evidence.‖ Inhalation Toxicology 19 (Supplement 1): 33-38. https://doi.org/10.1080/08958370701492961 Pope III, C.A., David V. Bates and Mark E. Raizenne. 1995. ―Health Effects of Particulate air pollution: time for reassessment‖ Environmental Health Prospective 103 (5): 472-480. https://doi.org/10.1289/ehp.95103472 Reza, Rizwan, M.K. Jain and Gurdeep Singh. 2009. ―Impact of mining activities on surface water quality in Angul-Talcher region of Odisha, India.‖ Mining Engineers Journal 10(11): 22-28. Sachs, J.D. and A.M. Warner. 1995. ―Natural Resource Abundance and Economic Growth.‖ Working paper no. 5398. National Bureau of Economic Research. https://doi.org/10.3386/w5398 Sachs, J.D. and A.M. Warner. 2001. ―Natural Resources and Economic Development: The Curse of Natural Resources.‖ European Economic Review 45: 827- 838. https://doi.org/10.1016/S0014-2921(01)00125-8 Saha, Shubhayu, Subhrendu K. Pattanayak, Erin O. Sills and Ashok K. Singha.2011. ―Under-mining health: Environmental justice and mining in India.‖ Health & Place 17 (1): 140–148. https://doi.org/10.1016/j.healthplace.2010.09.007 Singh, Prasoon Kumar, Ruchi Singh, and Gurdeep Singh. 2010. ―Impact of Coal Mining and Industrial Activities on Land use pattern in Angul-Talcher region of Odisha, India.‖ International Journal of Engineering Science and Technology 2 (12): 7771- 7784 Sinha, S. and S.P. Banerjee. 1997. ―Characterization of haul road in Indian opencast iron ore mine.‖ Atmospheric Environment 31 (17): 2809-2814. https://doi.org/10.1016/S1352-2310(97)00095-2 WHO. 2010. ―Quantifying the Burden of Disease from Mortality and Morbidity‖. World Health Organisation. Accessed online at http://www.who.int/healthinfo/global_burden_disease/metrics_daly/en/index.ht ml on 4th January 2017. https://doi.org/10.1080/08958370701492961 https://doi.org/10.1289/ehp.95103472 https://doi.org/10.3386/w5398 https://doi.org/10.1016/S0014-2921(01)00125-8 https://doi.org/10.1016/j.healthplace.2010.09.007 https://doi.org/10.1016/S1352-2310(97)00095-2