Article pubs.acs.org/est
Integrating Input Output Analysis with Risk Assessment to Evaluate the Population Risk of Arsenic Hwong-wen Ma,* Hsiu-Ching Shih, Ming-Lung Hung, Chia-Wei Chao, and Pei-Chiun Li Graduate Institute of Environmental Engineering, National Taiwan University, 71 Chou-Shan Rd., Taipei, Taiwan 106 S Supporting Information *
ABSTRACT: Multimedia and site-specific risk assessments (RA) of major sources releasing arsenic (As) were converted into sector-based risk coefficients, which were integrated with the Input Output Table (IO) to analyze the association between sector activities and health risks. The developed IO-RA framework is a valuable tool for unfolding the risk chain linking the receptors, exposure pathways, emission sources, and production and consumption activities associated with various industrial sectors. The enlarged decision space along the chain can then be considered in planning risk management strategies. This case study estimates that air emissions of As result in 1.54 carcinogenic cases. Export is the primary driving force and accounts for approximately 48% of the final demand that leads to population risks of As. The ranking of the contribution of the five sectors in terms of total population risks is as follows: electricity supply (1.06E+00), steelmaking (2.2 × 10−1), cement kilns (1.50 × 10−1), semiconductor manufacturing (6.34 × 10−2) and incinerators (4.31 × 10−2). The electricity supply, steelmaking industry, and cement kilns are the major sectors, not only because their emissions directly cause risk but also because they have a stronger influence on the risk generated by other sectors.
1. INTRODUCTION Risk assessment (RA) concerns the identification of causal links between adverse health effects and various types of hazardous activities. RA is used to support decision making in environmental management,1 and it has been used for increased emission control, pollution reduction, 2,3 and total quantity control.4 Source-oriented thinking is popular in RA, which links sources to receptors and identifies the points of intervention on the risk chain, from the sources of pollutants to the consequence receptors. However, the release of pollutants is influenced by industrial production activities and, in turn, consumption behaviors. With an enlarged system and decision space, more upstream options of the release, i.e., industrial linkage and consumption, could also be considered in risk management of a substance. Concerning industrial linkage and consumption, input− output (IO) analysis can be combined with RA to quantify the risk of a substance. IO analysis was developed by Leontief in 1936;5 it is a tool for delineating the relationship among various industrial sectors along the supply chain from production to consumption. Its most basic form consists of a system of linear equations, each describing the distribution of an industry’s product throughout the economy.6,7 Since the late 1960s, the IO framework has been extended to account for environmental pollution generation and abatement associated with interindustrial activities.8 IO analysis has been coupled with life cycle assessment (LCA) 9−12 and material flow analysis (MFA) 13,14 © 2011 American Chemical Society
to reveal the comprehensive environmental impact of releases along the supply and demand chain of related industries. Moreover, IO analysis was combined with RA to emphasize the importance of the releasing hotspot of substances, e.g., the chemicals released from products and industrial sectors in the decabromodiphenyl ether production and manufacturing stages.15 The spatial distribution of the health risk of Cr was estimated by the sector-specific economic activities in the U.S.16 However, multimedia, multipathway exposure, and the effects of final demand have not been thoroughly discussed in the existing literature. Focusing on the role of industrial sectors in risks, this study linked IO and site-specific risk assessment to quantify the risk of a substance released from industrial sectors by converting site-specific risks resulting from air pollution sources into sector-based coefficients of IO. The multimedia and sitespecific risks associated with major releasing sources were estimated and then combined with production values of relevant sectors to calculate the risk coefficient that reflect risk generated with unit production for individual sectors; these risk coefficients can then be used in the IO table. The IO table with a risk extension (referred to as IO-RA, the integration of input−output analysis and risk assessment) can identify the Received: Revised: Accepted: Published: 1104
November 20, 2010 December 3, 2011 December 14, 2011 December 14, 2011 dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
Environmental Science & Technology
Article
environmental parameters were also needed for the modeling. (As a manuscript presenting a methodology of integrating risk information into IO analysis, it does not list all of the parameters and values used in the risk assessment practice in detail; the introduction of the models and the associated parameters are included in the Supporting Information, parts C and D.) Finally, exposure factors relating to the rate of contact with the contact media were incorporated to obtain the receptor exposure dose at various locations. Average daily intake (ADI) from various exposure scenarios consisting of environmental media, contact media, and intake routes was calculated as eq 1.23 Using a linear cancer slope factor, the exposure would produce excess carcinogenic risk at various locations associated with individual stacks. The risks from individual exposure pathways, including air inhalation, ingestion of contaminated soil, drinking water, and food (nine items), were summed to provide the distribution of individual cancer risk in a county. The individual risk was then multiplied by the population to derive the population risks in a county due to a particular source of release. The population risks attributable to the sources that belong to an industrial sector were summed respectively, and then divided by the sector’s production value to obtain the risk coefficient attached to each sector. Equation 2 summarizes the calculation of risk coefficient for a sector.
contributions of sector activities and consumer demand to risks. Throughout the assessment, the release and flow of individual substances and the associated risks as a result of consumption and production activities can be quantified to facilitate the planning of substance management strategies. Here, we use IO-RA to simulate the risks of arsenic (As) released from air emission and interpret how risks vary across sensitive industrial sectors and are driven by final demand. Exposure to arsenic compounds through ingestion, inhalation and dermal contact with contaminated media may lead to adverse health effects, including skin lesions, peripheral neuropathy, anemia, and cancer of the skin, lungs, liver, and bladder.17−19 Electricity supply, cement kiln, steelmaking, incinerator, and semiconductor manufacturing industries were identified as sensitive industrial sectors of As emissions in this study.20 The risks of As will decrease with reduced emissions by the industrial sectors and final demand. The relations among risk, industrial sectors, and final demand will be emphasized.
2. MATERIALS AND METHODS 2.1. Data Sources. This study used air emissions of As in Taiwan as a case study. Taiwan releases 5.0 tons As annually into the air as estimated in the year 2009, mostly from coalfired power plants and cement kilns.20 The major sources of As are the following five economic sectors: electricity supply, cement kiln, steelmaking, semiconductor manufacturing industries, and incinerator. According to the Taiwan Emission Data System 5.0 (TEDS),21 which is the most comprehensive database maintained by the Taiwan EPA, and includes the coordinates, emission rate, height, emission velocity, temperature and diameter for all sources of air emissions, a total of 121 points of direct emission were identified by the above five sectors. The relative emission contributions of the abovementioned sectors are 73.34%, 21.03%, 3.68%, 1.68%, and 0.27%, respectively. Figure S1 of the Supporting Information shows the distribution and annual emissions of these sources. The risks of As decrease with reduced direct emissions by these five industrial sectors and indirect emissions from other sectors varied with these five sectors. All industrial sectors are classified into 49 sectors by the Directorate-General of Budget, Accounting and Statistics, Executive Yuan, ROC (Taiwan), and the Transactions Table of Domestic Goods and Services of these 49 sectors (the sector’s production value) constitutes the basic data. Final demand is categorized into household consumption, government consumption, capital formation, change in inventory and export.22 2.2. Methods. On the basis of emission characteristics in TEDS, an environmental transport modeling was conducted to estimate the spatial distributions of As in various environmental media resulting from the emissions of individual stacks. In this study, the American Meteorological Society-Environmental Protection Agency Regulatory Model (AERMOD), a steadystate atmospheric dispersion model, was employed.23 Stack characteristics, meteorological data, and terrain conditions were combined with the model to obtain air concentrations and depositions of As that were released from the stacks, leading to the exposure of nearby residents. Then, the multimedia and multiple pathway exposure assessment in the Human Health Risk Assessment Protocol for Hazardous Waste Combustion Facilities (HHRAP) was used to calculate the spatial distributions of As concentrations in soil and water, along with the contamination of contact media, including drinking water and food items.23 Physicochemical properties and
ADI ijkm = C ijkm = ×
R=
IUjk BW
×
EF × ED AT
(1)
∑ ∑ (Pk × CSF × ∑ ∑ ADIijkm)/PV k
m
i
j
(2)
ADIijkm = Average daily intake in county k from exposure to contact medium j contaminated by environmental medium i impacted by emission source m (mgkg-d−1) Cijkm = The concentration of the contaminant in the contact medium j affected by environmental medium i in county k resulting from emission source m (mgm−3 for air; mgL−1 for water; mgkg−1 for food) IUjk = Contact rate of contact medium j in county k (m3d−1 for inhalation; Ld−1 for drinking water; kgd−1 for ingestion of food); BW = Average body weight (kg); EF = Exposure frequency; ED = Exposure duration (d); AT = Averaging time (d); R = The risk coefficient of direct As release for NT$ 1 million dollars production value of a sector (cases/NT$); Pk = Population in administrative area k (person); CSF = Cancer Slope Factor (kgdmg−1); PV = Production value of a sector (New Taiwan Dollar(NT$)). The risk multiplier (∂L)/(∂F), i.e., Leontief inverse, can then be calculated to estimate the total cancer risks caused by a unit of final demand in a sector8 (eq 3 and eq 4). The risk multiplier makes clear the dependence of each sector’s emissions and resultant risks on the values of each of the final demands. Table S3 of the Supporting Information summarizes the risk multiplier between sectors. The IO framework has been extended with the risk multiplier to account for health risks associated with interindustrial activities. 1105
dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
Environmental Science & Technology
Article
Figure 1. Spatial distribution of population risk of As resulting from 121 sources of air emissions in Taiwan; the major contributive sectors are also shown for the three areas with the greatest risk.
∂L = R(1 − D)−1 ∂F
(3)
L = RX = R(1 − D)−1F
(4)
produce inputs necessary for the activities. IO-RA used this concept coupled with population risk assessment to define forward linkage as an industry affecting the recipient sectors of its product in terms of risk and backward linkage as an industry placing demand on the sectors with contributions to risk. The two indexes that correspond to forward and backward linkages, i.e., sensibility and dispersion, were calculated in the relation analysis of population risk as eq 5 and eq 6,24 respectively.
L = The total risk of As, represented by a 1 × n vector, n being 49 because there are 49 sectors (cases of cancer); (∂L)/(∂F) = Risk multiplier (cases/NT$); R = The risk coefficient of direct As release for NT$ 1 million dollars of production value, represented by a 1 × n vector (cases/NT$); X = n × 1 total production output vector (NT$); I = n × n identity matrix; D = n × n technical coefficient matrix; F = 1 × n total demand vector (NT$). Additionally, to interpret interindustry relations and risk of As, forward linkage and backward linkage were explored to examine the roles of industries in the economy. Forward linkage is the interindustrial relation in which the production activity of an industry triggers the utilization of its products by other sectors, and backward linkage refers to nonprimitive production activities that are likely to drive other sectors to
∂Lij ⎛ ⎞ ∑nj = 1 ⎜ 1 ⎟ ∂Fij ⎟ Ui = log⎜ ⎜⎜ 1/n ∑n ∑n ∂Lij ⎟⎟ i = 1 j = 1 ∂Fij ⎠ ⎝
(5)
∂Lij ⎛ ⎞ ∑in= 1 ⎜ 1 ⎟ ∂Fij ⎟ Uj = log⎜ ⎜⎜ 1/n ∑n ∑n ∂Lij ⎟⎟ i = 1 j = 1 ∂Fij ⎠ ⎝
(6)
Ui = Sensibility of the ith sector; Uj = Dispersion of the jth sector; 1106
dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
Environmental Science & Technology
Article
Figure 2. Population risk distribution among various administrative areas caused by emissions of key industries.
Figure 3. The contributions of various exposure pathways to the resulting risks from the five key industries.
risks were experienced in southern and central Taiwan. In southern Taiwan, the population risk in KH City is 3.04 × 10−1 cases, and risk from electricity supply industry and steelmaking industry are the greatest: 1.27 × 10−1 cases and 1.28 × 10−1 cases, respectively. The population risk of As in Taichung (in central Taiwan), the second severely polluted region, is 2.43 × 10−1 cases, and the main contributive industry is the electricity supply industry (94%). Taoyuan (1.67 × 10−1 cases), and Changhua (1.39 × 10−1 cases) are also severely polluted cities. A list of the most hazardous counties is as follows: KH City, Taichung, Taoyuan and Changhua. Hence, the building of new plants should be avoided in these counties. The ranking of the contribution of the five sectors in terms of total population risks is as follows: electricity supply, steelmaking, cement kilns, semiconductor manufacturing, and incinerators. The total population risks caused by the As emissions from these five industries are 1.06, 2.20 × 10−1, 1.50 × 10−1, 6.34 × 10−2, and 4.31 × 10−2, respectively. The ranking of risk agrees well with the ranking of As emissions from the five sectors. However, although the emissions from the steelmaking industry (3.68%) are less than those of cement kilns (21.03%), the total population risk of As from the steelmaking industry is slightly higher than that of cement kilns. This discrepancy of ranking between emissions and risk reveals the roles played by the spatial distribution of sources and sitespecific environmental conditions. Figure 2 shows the risk distribution among various administrative areas caused by the emissions of each industry. For example, the electricity supply contributes to the population risks ranging from 0 cases (Nanto County) to
(∂L)/(∂F)ij = Risk multiplier of input from sector i to sector j (cases/NT$). The two indexes of any sectors can be plotted onto a graph with dispersion as the x axis and sensibility as the y axis. The sectors falling within the first and fourth quadrants have larger environmental impact on the supply side, whereas the sectors falling within the first and second quadrants have larger environmental impact on the demand side. The intersection, the first quarter, has crucial impact on both the supply and demand sides.
3. RESULTS AND DISCUSSION The results of the risk are presented from the perspectives of industry, final demand, and the relation between industries. The first perspective identifies the As risk resulting from direct release of stationary sources; the second perspective discusses the driver behind the source of risk, i.e., the final demand; and the third perspective investigates the links among industries to examine sensibility and dispersion in the “supply chain” of As risk. 3.1. Population Risk of Industry. In this study, the risks of 121 As-emission sources corresponding to the five industries were assessed with site-specific exposure scenarios. The population risks to the residents in various counties/cities are shown in Figure 1; they range from 4.62 × 10−8 (Nantou) to 3.04 × 10−1 (KH City). Total population risk of As is 1.54 (under 30 years of exposure duration and a population size of 2.3 × 107), meaning residents suffering from cancer as a result of the As emitted from the 121 sources were expected to be 1.54 cases (Table S1 of the Supporting Information). Greater 1107
dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
Environmental Science & Technology
Article
Figure 4. The population risks resulting from various exposure pathways in different administrative areas.
Table 1. Population Risk of As Caused by Each Category of Final Demand item contribution of population risk
household consumption (%)
government consumption (%)
capital formation (%)
change in inventory (%)
export (%)
final demand (cases)
36.10
2.20
13.05
0.52
48.13
1.54
Figure 5. Sensibility and dispersion of key sectors on population risk of As.
2.29 × 10−1 cases (Taichung County). It can be seen that the risk is not distributed evenly among various administrative areas. In particular, power need is supplied by only a few power plants, which leads to greater diversity of population risks among the administrative areas. In contrast, the steelmaking industry is concentrated in KH City, which has the highest population density, and the resulting distribution of population risk is quite diverse. Regarding characteristic exposure pathways of As, inhalation is the primary exposure pathway, closely followed by vegetable ingestion and then drinking water. Figure 3 shows the
contributions of various exposure pathways to the population risks caused by individual key sectors. In addition, inhalation risk and vegetable ingestion risk are associated with the electricity supply, cement, and steelmaking industries, whereas drinking water ingestion risk is mostly associated with the electricity supply and semiconductor industries. Figure 4 shows the contribution of various exposure pathways in different cities and counties. Inhalation is often a major pathway in urban counties, whereas vegetable ingestion is a major pathway in rural ones. The risk values are provided in Tables S2 of the Supporting Information. 1108
dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
Environmental Science & Technology
Article
3.2. Population Risk of Final Demand. The IO-RA approach is well-suited for analyzing the driver of the resulting risk from a demand perspective. The calculation of risk multipliers found that the population risks per unit of final demand in a particular sector are 2.67 × 10−6, 7.30 × 10−7, 3.93 × 10−7, 1.21 × 10−7, and 1.13 × 10−7 for the electricity supply industry, cement kilns, steelmaking industry, incinerators, and semiconductor industry, respectively. Table 1 summarizes the contributions from the final demand categories: the population risk due to household consumption is 0.56, government consumption is 0.03, capital formation is 0.20, change in inventory is 0.01 and export is 0.74. Export is the primary driving force; because of the production for export, 0.74 cases are likely to suffer from cancer due to As, which accounts for about 48.13% of the risks caused by the final demand. With further analysis of the risk multiplier (Table S3 of the Supporting Information), the major industries that contribute to population risks due to the demand for export are as follows: the electricity supply industry (15.25%), semiconductor manufacturing (12.69%), the steelmaking industry (6.21%), commodity trading (5.72%), and public and other construction (5.60%). In addition to exports, household consumption also accounts for a large portion (36.1%) of the demand driving the activities that contribute to population risks of As. This understanding will facilitate an examination of the structure of supply and demand of economic activities in terms of both health impacts and traditional economic gains. 3.3. Relation between Industry and Final Demand. The analysis of forward and backward linkage informs the interaction between sectors of the generation of population risk of As. A sector with forward linkage indicates that it provides inputs to respond to the demand of the other industries and thus generates an As risk, whereas a sector with backward linkage indicates that its demand drives the activities of the other industries to provide inputs to the sector and thus produces As risks. In relation to the health risks of As, the industry with the highest forward linkage (sensibility) is the electricity supply industry, followed by cement kilns, and steelmaking industries. Backward linkage (dispersion) follows the same sequence. Figure 5 shows the results of the forward and backward linkage analysis for the top five sectors. The sectors located within the first quadrant are the key industries for generating As risks because they have greater forward and backward linkage, meaning that these industries are not only able to promote the activities of other industries, but are also indispensable in cooperation with other industries regarding the increase and decrease of risk related to As emissions. Therefore, if the structure or activities of these sectors changes, the impact on the population risks of As resulting from other industries is considerable. On the contrary, incinerator and semiconductor manufacturing industries are located in the third quadrant, meaning the interaction of the two sectors with other industries is less intense. IO-RA is a valuable tool to assess the consequences of policy scenarios in terms of health risks. The policy scenarios may include changes in final demand, advancement of technological efficiency, modification of industrial structure, and strategies of exposure reduction. It is an informative assessment tool because it describes how a policy, strategy, or program influences various sectors’ activities, such as the emissions of chemicals, and the subsequent site-specific health risks. To provide a more complete picture, future research should (1) incorporate uncertainty analysis that provides uncertainty information
associated with the assessment results; because of the complexity involved in the integration of economic and environmental factors, an efficient uncertainty analysis framework that deals with IO and RA is needed; (2) include discharges from other media, such as wastewater and solid wastes. Furthermore, the nonpoint or dissipative emissions could be considered, and (3) combine IO-RA with economic cost and benefit analysis to facilitate a more balanced evaluation of an industry or sector in terms of the positive and negative roles it plays in society.
■
ASSOCIATED CONTENT
S Supporting Information *
Two figures and three data tables provide additional information and detail results of population risks in parts A and B. The models used and the associated parameters are elaborated in parts C and D. This information is available free of charge via the Internet at http://pubs.acs.org/.
■
AUTHOR INFORMATION
Corresponding Author
*Tel.: +886-2-23630406; fax: +886-2-23928830; e-mail:
[email protected].
■
REFERENCES
(1) Robson, M. G. Toscano, W. A. Risk Assessment for Environmental Health; John Wiley & Sons, Inc.: New York, 2007. (2) Liptak, J. F.; Lombardo, G. The development of chemicalspecific, risk-based soil cleanup guidelines results in timely and costeffective remediation. J. Soil Contam. 1996, 5, 83−94. (3) Meneses, M.; Schuhmacher, M.; Domingo, J. L. Health risk of emissions of dioxin and furans from a municipal waste incinerator: comparison with other emission sources. Environ. Int. 2004, 30, 481− 489. (4) Kao, W. Y.; Ma, H. W.; Wang, L. C.; Chang-Chien, G. P. Sitespecific health risk assessment of dioxins and furans in a industrial region with numerous emission sources. J. Hazard. Mater. 2007, 145, 471−481. (5) Leontief, W. Quantitative input-output relations in the economic system of the United States. Rev. Econ. Stat. 1936, 18, 105−125. (6) Leontief, W. Environmental repercussions and the economic structure: an input-output approach. Rev. Econ. Stat. 1970, 52, 262− 271. (7) Miller, I.; Shelly, M.; Jonmaire, P.; Lee, R. V.; Harbison, R. D. Assessing contributory risk using economic input-output life-cycle analysis. Int. J. Environ. Health R. 2005, 15, 107−115. (8) Miller, R. E.; Blair, P. D. Input-Output Analysis: Foundations and Extensions; Prentice-Hall, Inc.: Englewood Cliffs, NJ, 1985. (9) Hendrickson, C. T.; Lave, L. B.; Matthews, H. S. Environmental Life Cycle Assessment of Goods and Services: An Input-Output Approach; Resources for the Future, 2006. (10) Hendrickson, C.; Horvath, A.; Joshi, S.; Lave, L. Economic input-output models for environmental life cycle assessment. Policy Anal. 1998, 32, 184A−191A. (11) Matthews, H. S.; Small, M. J. Extending the boundaries of life cycle assessment through environmental economic input-output models. J. Ind. Ecol. 2001, 4, 7−10. (12) Mattila, T. J.; Pakarinen, S.; Sokka, L. Quantifying the total environmental impacts of an industrial symbiosis a comparison of process-, hybrid and input output life cycle assessment. Environ. Sci. Technol. 2010, 44, 4309−4314. (13) Hawkins, T.; Hendrickson, C.; Higgins, C.; Matthews, H. S.; Suh, S. A mixed-unit input-output model for environmental life-cycle assessment and material flow analysis. Environ. Sci. Technol. 2007, 41, 1024−1031.
1109
dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
Environmental Science & Technology
Article
(14) Bailey, R.; Bras, B.; Allen, J. K. Applying ecological input-output flow analysis to material flows in industrial system. J. Ind. Ecol. 2004, 8, 45−91. (15) Wright, H. E.; Zhang, Q.; Mihelcic, J. R. Integrating economic input-output life cycle assessment with risk assessment for a screeninglevel analysis. Int. J. Life Cycle Assess. 2008, 13, 412−420. (16) Rehr, A. P.; Small, M. J.; Matthews, H. S.; Hendrickson, C. T. Economic sources and spatial distribution of airborne chromium risks in the U.S. Environ. Sci. Technol. 2010, 4 (6), 2131−2137. (17) Goering, P. L.; Aposhian, H. V.; Mass, M. J.; Cebrian, M.; Beck, B. D.; Waalkes, M. P. The enigma of arsenic carcinogenesis: Role of metabolism. Toxicol. Sci. 1999, 49, 5−14. (18) Tchounwou, P. B.; Centeno, J. A.; Patlolla, A. K. Arsenic toxicity, mutagenesis, and carcinogenesisA health risk assessment and management approach. Mol. Cell. Biochem. 2004, 255, 47−55. (19) Agency for Toxic Substance and Disease Registry; http://www. atsdr.cdc.gov/ (20) The project of emission investigation and health risk assessment of Dioxins and heavy metals from the stationary sources; Environmental Protection Administration, Executive Yuan: Taipei, 2009. (21) Taiwan Emission Data System; Taiwan Environmental Protection Administration, Executive Yuan: Taipei, 2000. (22) Transactions Table of Domestic Goods and Services; Taiwan Directorate-General of Budget, Accounting and Statistics, Executive Yuan: Taipei, 2004. (23) Human Health Risk Assessment Protocol for Hazardous Waste Combustion Facilities; U.S. EPA, Office of Solid Waste and Emergency Response: Washington, DC, 2005. (24) Baer, W.; Kerstenetzky, I. Import substitution and industrialization in Brazil. The American Economic Review. 1964, 54, 411−425.
1110
dx.doi.org/10.1021/es203036r | Environ. Sci. Technol. 2012, 46, 1104−1110
