WPS5420 Policy Research Working Paper 5420 Environmental Performance Rating and Disclosure An Empirical Investigation of China's Green Watch Program Yanhong Jin Hua Wang David Wheeler The World Bank Development Research Group Environment and Energy Team September 2010 Policy Research Working Paper 5420 Abstract Environmental performance rating and disclosure has locations, time trend, and initial level of environmental emerged as an alternative or complementary approach performance, the analysis finds that firms covered by to conventional pollution regulation, especially in Green Watch improve their environmental performance developing countries. However, little systematic research more than non-covered firms. Bad performers improve has been conducted on the effectiveness of this emerging more than good performers, and moderately non- policy instrument. This paper investigates the impact of compliant firms improve more than firms that are a Chinese performance rating and disclosure program, significantly out of compliance. The reasons for these Green Watch, which has been operating for 10 years. different responses seem to be that the strengths of To assess the impact of Green Watch, the authors use incentives that the disclosure program provides to the panel data on pollution emissions from rated and polluters at different levels of compliance are different unrated firms, before and after implementation of the and the abatement costs of achieving desired levels of program. Controlling for the characteristics of firms and ratings are different for different firms. This paper--a product of the Environment and Energy Team, Development Research Group--is part of a larger effort in the department to understand and improve environmental governance in developing countries.. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at hwang1@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Environmental Performance Rating and Disclosure: An Empirical Investigation of China's Green Watch Program Yanhong Jin1 Department of Agricultural, Food and Resource Economics Rutgers University, New Brunswick, NJ 08850, USA Hua Wang Development Research Group, World Bank, MC3-331, 1818 H. St., N.W., Washington, DC 20433, USA David Wheeler Center for Global Development, Washington, DC Key words: environmental performance rating, public disclosure, developing country, China, Green Watch, program evaluation 1 The authors are Assistant Professor of Rutgers University, Senior Economist of the World Bank, and Senior Fellow of the Center for Global Development, respectively. The authors gratefully acknowledge support for this research from Profs. Genfa Lu, Yuan Wang and Gangxi Xu of Nanjing University and would also like to thank seminar participants at the World Bank for their comments and suggestions. The views expressed in this paper are solely of those authors and do not necessarily represent the views of the World Bank, its Board of Executive Directors or the countries they represent. Correspondence: Hua Wang at hwang1@worldbank.org. 1. Introduction Environmental performance rating and disclosure (PRD) has emerged as a substitute or complement for traditional pollution regulation, especially in developing countries [6][28][31]. Indonesia's PROPER (Program for Pollution Control, Evaluation and Rating), initiated in June 1995, was the first PRD program in developing countries. Because of its perceived overall success, as measured by reduced emissions at a lower regulatory cost, many countries have established similar programs for a variety of industry sectors and pollutants in diverse economic, institutional and cultural settings. These programs include the Philippines' EcoWatch, India's Green Rating Project, China's Green Watch, Vietnam's Green Bamboo, Ghana's EPRD, and Ukraine's PRIDE. PRD programs are particularly attractive for developing countries because institutional weaknesses hinder conventional monitoring and enforcement of environmental laws, regulations, and standards [8], and because PRD programs have lower regulatory costs [6]. The literature on the effectiveness of PRD programs is very limited and falls into two groups. The first compares the environmental performance ratings of firms before and after a program is implemented, and ascribes any ratings improvements to the program [1]. However, this approach may be confounded by time-varying factors such as technology improvements. The second group compares polluting emissions from rated and unrated firms, and credits performance improvements by rated firms to the program. However, this approach may be confounded by selection bias (e.g., firms with better environmental performance may be more likely to be rated). It is rare to have pollution data for both rated and unrated firms before and after implementation of a PRD program. Garcia et al. [10][11] assess the effectiveness of Indonesia's PROPER using measured pollution from rated and unrated firms, both ex ante and ex post. Their 2 2007 study suggests that PROPER has reduced emissions intensity, with a particularly rapid and strong impact on firms that have poor initial compliance records. Their 2009 study finds a strong reactive response during the first six months of disclosure, followed by a more moderate, but still significant, longer-run response as management adjusts to the new regime. This study extends PRD assessment to China, using panel data on pollution from rated and unrated firms, before and after implementation of the Green Watch program. Our work offers two main contributions to the literature. First, we exploit the panel structure of the data to control for confounding factors such as time-variant technology improvement and selection bias between rated and unrated firms. Second, we go beyond a single measure of environmental performance to consider the impact of ratings disclosure on several measures, including emissions intensity and effluent concentrations for a variety of air and water pollutants. The remainder of the paper is organized as follows. Section 2 reviews the relevant literature, focusing on the role of PRD programs in developing countries. Section 3 describes China's Green Watch program, while Section 4 describes our survey instrument and provides descriptive statistics for major variables. Section 5 presents our estimation model and results, and Section 6 summarizes and concludes the paper. 2. Previous research The literature on pollution control policies includes extensive work on command-and- control, market-based and information-based instruments [6]]. Command-and-control instruments are often inefficient and ineffective in developing countries, because firms may fail to report adequately, regulators may lack the technical and administrative capacity for effective monitoring and enforcement, and judicial systems may be weak and/or corrupt. These same weaknesses limit regulators' ability to employ market-based instruments, which also work less effectively in countries where market failures are common and legal and institutional supports for formal market activities are weak. Information-based instruments can be effective in developing countries where strong regulatory institutions and/or well-developed markets are absent, but where enough information can be reliably obtained to provide credible performance ratings. In practice, diverse information programs have served as complements to command-and-control and market-based instruments [21]. Information programs reduce the information asymmetry between polluters and environmental stakeholders (consumers, communities, NGOs, investors), empowering these stakeholders to pressure polluters for improved environmental performance [5][17][26]. When implemented correctly, information instruments promote better interaction and dialogue among firms, stakeholders and regulators [10]. Information instruments also leverage markets in significant ways. An extensive empirical literature suggests that disclosure of firms' bad environmental performance reduces their stock prices both in developed countries [8][14][19][23],[24] and developing countries such as Argentina, Chile, Mexico, and the Philippines [7]. Jackson [16] and Boyle and Kiel [4] review the impacts of disclosure on housing prices in the US, which are found to be lower near Superfund sites [22][29], hazardous waste sites [30], non-hazardous landfills [25], nuclear radiation sources [9], and polluting manufacturing plants [11]. Housing prices also respond to publicized environmental contamination incidents [19][20]. Information instruments have diverse forms, including reports of measured pollution, environmental accident reports, and environmental performance ratings. In the US, for example, the Toxics Release Inventory (TRI) discloses toxic chemical releases and waste management activities by significant toxic polluters and federal facilities. In developing countries, however, weak regulatory institutions may have difficulty in implementing such emissions inventories. In addition, despite an emerging literature on stakeholders' role in improving firms' environmental performance [2][3][27][33], concerns remain about the public's ability to understand and utilize complex emissions reports. For example, Bui and Mayer [5] find that the release of TRI's highly-detailed information on facilities' toxic emissions has virtually no effect on housing prices in neighboring areas, even when the release of such information is unexpected. The dual problems of emissions inventories in developing countries ­ technical feasibility and public understanding ­ have led to a preference for programs that condense complex information into environmental performance ratings that are disclosed to the public. Research on the effectiveness of performance rating and disclosure (PRD) programs suggests that that have a significant, positive impact on regulatory compliance [1][6][10][11][32]. Dasgupta et al. [6] summarize the changes in compliance rates for several PRD programs in Asia. During the first and second years after inception, compliance rates among covered firms increased from 37% to 61% in Indonesia, 8% to 58% in the Philippines, 10% to 24% in Vietnam, 75% to 85% in Zhenjiang, China and 23% to 62% in Hohhot, China. Several empirical studies also find that PRD programs have improved firms' environmental performance in Indonesia [1][10][11] and China [32]. However, data constraints generally limit these studies to comparisons of environmental ratings before and after program implementation, or comparisons of compliance status between rated and unrated firms. Unfortunately, intertemporal rating comparisons are subject to confounding effects from time-varying factors such as technology change, while cross-sectional comparisons can be subject to significant selection bias. 3. China's Green Watch Program Despite long-standing efforts to control pollution with traditional regulatory instruments, China continues to have severe pollution problems. This has led China's State Environmental Protection Administration (SEPA) to test the effectiveness of environmental performance rating and disclosure in a program supported by the World Bank. In 1999, SEPA launched its Green Watch program in Zhenjiang City, Jiangsu Province and Hohhot City, Inner Mongolia Autonomous District. Zhenjiang implemented a relatively complex rating system, as shown in Figure 1, while Hohhot used a simpler rating system that was suited to its lower level of economic and institutional development (Wang et al., 2004). As shown in Figure 1, Green Watch in Jiangsu rates firms' environmental performance from best to worst in five colors ­ green for superior performance; blue for full compliance; yellow for meeting major compliance standards but violating some minor requirements; red for violating important standards; and black for more extreme non-compliance. Green Watch ratings provide incentives for firms to improve their environmental performance in a comprehensive way. The primary benchmarks for ratings are China's emission and discharge standards that specify effluent concentration limits. Firms violating any of these standards are rated red, and firms violating standards in more than 60% of inspections are rated black. The secondary benchmarks are China's load-based emission and discharge standards. Firms that satisfy the primary benchmarks but violate the secondary standards are rated yellow. The ratings system also incorporates other performance indicators, including hazardous waste disposal practices, solid waste recycling, pollution accidents, public complaints, internal management requirements, China cleaner production certificates, ISO 14000 certificates, administrative penalties, and other citations for illegal activity. For each indicator, the system specifies a link to ratings that is clear, unambiguous and publicly available. The first Green Watch ratings were disclosed through the media in 1999. The program was extended from Zhenjiang to all of Jiangsu Province in 2001, and to eight other provinces during 2003-2005. Nationwide implementation of Green Watch has been promoted since 2005. Overall, the available evidence suggests a positive impact for the program. Table 1 shows that in Zhenjiang, the percentage of firms with positive ratings (green, blue and yellow) increased from 75% in 1999 to 85% in 2000. The most significant changes were in the extremely-noncompliant black group, whose percentage dropped from 11% in 1999 to 2% in 2000, and a major shift from the partially-compliant yellow group (44% to 22%) to the fully-compliant blue group (27% to 61%). Evidence for the Green Watch program in Jiangsu Province indicates both increasing participation by firms and improvement in their compliance rates. As shown in Table 1, the number of rated firms increased more than tenfold, from 1,059 in 2001 to 11,215 in 2006; and the percentage of firms with positive ratings (green, blue, and yellow) increased from 83% in 2001 to 90% in 2006. Furthermore, Table 1 suggests that Green Watch ratings provide a strong improvement incentive for noncompliant (red and black) firms, with stronger effects on firms with red ratings (moderate noncompliance) than those with black ratings (extreme noncompliance). 4. Data This study utilizes a pollution dataset for both rated and unrated firms during the period 1996-2001 in four cities of Jiangsu province (Huaian, Wuxi, Yangzhou, and Zhenjiang). Following the success of the pilot program in Zhenjiang, Huaian, Wuxi and Yangzhou adopted the same program in 2001. Table 2 provides information on socioeconomic and environmental conditions in the four cities, as well as polluting emissions in 2001. Wuxi has the largest population as well as the highest GDP per capita, while Huaian is the poorest. Wuxi and Yangzhou have the lowest readings for air quality, measured by SO2 (sulfur dioxide) and NO2 (nitrogen dioxide), and water quality measured by TSS (total suspended solids) and regulatory compliance percentage. The dataset includes detailed information on the firms' characteristics, pollution, and environmental performance ratings. We obtained this information from the municipal environmental protection bureaus of the four cities. Their pollution monitoring, inspection and environmental information systems are well-developed and well-managed, primarily because of their long-standing experience with pollution registration requirements and China's pollution charge system.2 Table 3 shows that 36.7% of the firms in the sample were rated by Green Watch. The majority of rated firms were assigned blue (60.38%) and yellow (22.37%); only a few earned the best (green) rating (2.96 %) or the worst (black - 2.96%). The distributions are similar across cities, with the majority of firms rated blue and yellow, and very few green and black. 5. Multivariate analysis Our pollution data are sufficiently detailed to permit assessment of Green Watch for both water and air pollution, measured by intensity and effluent concentration. Pollution intensity is total emissions divided by the gross value of output. We use total suspended solids (TSS), chemical oxygen demand (COD) and generated waste water to measure water pollution, and sulfur dioxide (SO2), waste gas, and dust and smoke to measure air pollution. 2 For discussion of firm-level pollution data in China, see Wang and Wheeler (2006). The dependent variables in our multivariate analyses are changes in pollution intensity and concentration for different pollutants. Let pollution intensity be specified as Yi,t for firm i in year t. The dependent variable for the intensity equation is the first difference, Yi,t - Yi,t-1. The reduced-form fixed effects model for Yi,t - Yi,t-1 is (1) Yit-Yit-1 = 0 + 1 Fit + 2 Cit + 3 Rit + 1 t + µit + it. where Fit and Cit are vectors of characteristics of the firm and the city; Rit is a vector that incorporates both rating status (rated or unrated) and colorcategory assignments for rated firms; t is a time trend; µi represents unobserved firm effects; and it is a random error term. Endogeneity is not a serious problem in this case, because ratings released in year t are based on multi-dimensional performance observations during year t ­ 1. If sample firms were randomly assigned to rated and unrated groups, we would not expect a statistical difference in intergroup pollution at t -1, before the first Green Watch disclosure in period t. Assessing prior randomness is complicated in this case by the distributions of pollution intensity and effluent concentration. Both are highly skewed, with skewness coefficients ranging from 3 to 9. In this case, the traditional student t test for equality of pre-rating group means is not appropriate. We employ the nonparametric Wilcoxon-Mann-Whitney test for equal means and the K-sample test for equal medians. Our results, reported in Table 4, show that significant differences in means and medians are common in the sample. In Zhenjiang, where Green Watch began in 1999, we find significant differences in mean and/or median pollution intensities for waste water, COD and dust and smoke, and significant differences in mean and/or median effluent concentrations for TSS, COD and dust and smoke. Table 4 reports similar findings for the other three sample cities (Huaian, Wuxi and Yangzhou), where the first public disclosure of ratings occurred in early 2001.3 In light of these results, it is appropriate to introduce controls for pre-program pollution in our estimating equation: (2) Yit-Yit-1 = 0 + 1 Fit + 2 Cit + 3 Rit + 1 t + 2 Yit-1 + µit + it To determine the appropriate estimator, we employ Breusch and Pagan Lagrangian multiplier (BPLM) tests for random effects. We reject the null hypothesis in favor of the random effects model for air pollution intensities, and for air and water effluent concentrations. We assume that it is correlated across firms within a city but uncorrelated across firms in different cities. Table 5/6 and 7/8 present estimation results for changes in pollution intensity and effluent concentration, respectively. In Tables 5 and 7, we test whether a firm reduces pollution simply because it is rated. A priori, it is possible that self-scrutiny by a rated firm results in better environmental management and reduced pollution, even if the firm's rating is good. Our results for the regression variable PRD are consistent with this hypothesis: PRD rating has a negative impact on pollution for all equations in Tables 5 and 7, and a statistically significant impact on TSS and SO2 for pollution intensity, and dust and smoke for effluent concentration. Tables 6 and 8 provide more insight, by identifying specific color ratings for firms. Here we find very strong results for water pollution intensity (TSS and COD) and dust-and-smoke intensity in Table 6, with highly-significant reductions for poorly-rated firms that are much larger than reductions for firms with better ratings. Intensities generally decline more among 3 We also conducted the equal mean and median tests for each of the three cities. The results are qualitatively similar. rated firms for the other pollutants as well, but without the striking differential for poorly-rated firms. The same general pattern holds in Table 8, with generally-declining effluent concentrations across all rated firms and the largest impacts among poorly-rated firms. Although some concentration results are highly significant, the overall significance level is somewhat lower than for pollution intensities. Across both tables, red-rated firms exhibit stronger responses than black-rated firms. 6. Summary and conclusions This study has employed a new panel data set to test the impact of environmental performance rating and disclosure (PRD) on polluting firms in China. The data include ex ante and ex post pollution measures for both rated and unrated firms, enabling us to control for confounding factors such as time-variant technology improvement and selection bias. Our results strongly suggest that Green Watch has significantly reduced pollution from rated firms, with particularly strong impacts on firms with poor ratings. Among poorly-rated red and black firms, the impact is generally greater on red-rated firms that are closer to compliance with regulations. The reasons for these responses can be that the incentive for improvement that the Green Watch generates is stronger for firms with poor ratings than those with good ratings, and that the abatement costs for the red-rated firms to achieve compliance are lower than those black rated firms, even though the pressure for improvement can be stronger with the black-rated firms than the red-rated firms. This research also adds some insights to the growing comparative literature on PRD's. After studying PRD experiences in Indonesia (PROPER) and the Philippines (EcoWatch), Dasgupta et al. [6] argue that PRD programs are most effective in moving moderately non-compliant firms into compliance with regulations, but may provide insufficient incentives to induce significant improvements by the worst performers or firms with good ratings. However, our results for Green Watch indicate significant impacts for firms with good (green and blue) ratings. The stronger result for our four cities in Jiangsu Province may stem from two additional benefits for green-rated firms: (a) Enterprises awarded green in a particular year can be given priority consideration in the selection of enterprises with the best economic and social performance records; and (b) an enterprise that has won green for three consecutive years is given preferential status by provincial environmental regulators. The Jiangsu experience suggests that stronger results can be produced by PRD programs that target highly-rated firms for benefits beyond reputational improvement. References [1] Afsah, S., B. Laplante, and D. Wheeler, Regulation in the information age: Indonesian public information program for environmental management, Research Paper, World Bank, Washington, DC. 1997. [2] Arora, S. and T. Cason, Do community characteristics influence environmental outcomes? evidence from the toxics release inventory, J. Appl. Econ. 1:413-453 (1998). [3] Blackman, A. and G.J. Banniseter, Community pressure and clean technology in the informal sector: an econometric analysis of the adoption of propane by traditional Mexican brick makers, J. Environ. Econ. Manage. 35:1-21 (1998). [4] Boyle, M.A. and K. A. Kiel, A survey of house price hedonic studies of the impact of environmental externalities, J. Real Estate Lit. 9:117-144 (2001). [5] Bui, L.T. and C.J. Mayer, Regulation and capitalization of environmental amenities from the toxic release inventory in Massachusetts, Rev. Econ. Stat. 85:693-708 (2003). [6] Dasgupta, S., H. Wang, and D. Wheeler, Disclosure strategies for pollution control, In The International Yearbook of Environmental and Resource Economics 2006/2007: A Survey of Current Issues. Tietenberg, T. and Folmer, H. (Ed.), Northampton, MA: Edward Elgar Publishers, pp. 93-119 (2006). [7] Dasgupta, S., B. Laplante, and N. Mamingi, Pollution and capital markets in developing countries, J. Environ. Econ. Manage. 42:310-335 (2001). [8] Foulon, J., P. Lanoie, and B. Laplante, Incentives for pollution control: regulation or information? J. Environ. Econ. Manage. 44:169-187 (2002). [9] Gamble, H.B. and R.H., Downing effects of nuclear power plants on residential property values, J. Reg. Sci. 22:457-478 (1982). [10] Garcia J.H., T. Sterner, and S. Afsah, Public disclosure of industrial pollution: the PROPER approach for Indonesia?" Environ. Dev. Econ. 12:739-756 (2007). [11] Garcial, J.H, S. Afsah and T. Sterner, Which Firms are more sensitive to public disclosure schemes for pollution control? evidence from Indonesia's PROPER program, Environ. Res. Econ. 42:151-168 (2009). [12] Hanna, B.G, Home values, incomes, and industrial pollution, J. Environ. Econ. Manage. 54:100-112 (2007). [13] Hanley, N., J.F. Shogren, and B. White, Environmental Economics in Theory and Practice. Oxford University Press, New York (1997). [14] Hamilton, J., Pollution as news: media and stock market reaction to the toxics release inventory Data, J. Environ. Econ. Manage. 28:98-113 (1995). [15] Imbens, G.W., Nonparametric estimation of average treatment effects under exogeneity: a review, Rev. Econ. Stat. 86:4-29 (2004). [16] Jackson, T.O., The effects of environmental contamination on real estate: a literature review, J. Real Estate Lit. 9:93-116 (2001). [17] Kennedy, P.W., B. Laplante, and J. Maxwell, Pollution policy: the role for publicly provided information, J. Environ. Econ. Manage. 26: 31-43 (1994). [18] Khanna, M., W. Rose, H. Quimio, and D. Bojlova., Toxic release information: a policy tool for environmental protection, J. of Environ. Econ. Manage. 36:243-266 (1998). [19] Kiel, K.A., Measuring the impact of the discovery and cleaning of identified hazardous waste sites on house value, Land Econ. 71:428-435 (1995). [20] Kiel, K.A. and K.T. McClain, House prices during sitting decision stages: the case of an incinerator from rumor through operation, J. Environ. Econ. Manage. 28:241-255 (1995). [21] Kleindorfer, P.R. and E.W. Orts, Information regulation of environmental risks, Risk Anal. 18:155-170 (1998). [22] Kohlase, J.E., The impact of toxic waste site on housing values, J. Urban Econ. 30:1-26. (1991). [23] Konar, S. and M. Cohen, Information as regulation: the effect of community right to know laws on toxic emissions, J. of Environ. Econ. Manage. 32:109-124 (1997). [24] Lonoie, P., B. Laplante, and M. Roy, Can capital markets create incentives for pollution control? Ecol. Econ. 26:31-41 (1998). [25] Michaels, P.G. and V.K. Simth, Market segmentation and value amenities with hedonic models: the case of hazardous waste sites, J. Urban Econ. 28:223-242 (1990). [26] Oberholzer-Gee, F. and M. Mitsunari, Information regulation: do the victims of externalities pay attention? J. Regulatory Econ. 30:141-158 (2006). [27] Pargal, S. and D. Wheeler, Informal regulation of industrial pollution in developing countries: evidence from Indonesia, J. Political Economy 104: 1314-1327 (1996). [28] Portney, P., Environmental problems and policy 2000-2050, J. Econ. Perspect. 14:199- 206 (2000) [29] Reichert, A.K., Impact of a toxic waste superfund site on property values, Appraisal J. 65:381-392 (1997). [30] Thayer, M., H. Albers and M. Rahmatian, The benefits of reducing exposure to waste disposal sites: a hedonic housing value approach." J. Real Estate Res. 7:265-282 (1992). [31] Tietenberg, T., Disclosure strategies for pollution control, Environ. Res. Econ. 11:587- 602 (1998). [32] Wang, H., J. Bi, D. Wheeler, J. Wang, D. Cao, G. Lu, and Y. Wang, Environmental performance rating and disclosure: China's Green Watch program, J. Environ. Manage. 71:123-133 (2004). [33] Wheeler, D., H. Hettige, S. Pargal, and M. Singh, Formal and informal regulation of industrial pollution: evidence from Indonesia and the U.S." World Bank Econ. Rev. 11:433- 450 (1997). Table 1. Firms' Environmental Performance Ratings by the Green Watch Program in Jiangsu Province, China (% Representation in Parentheses) Year Green Blue Yellow Red Black Total Pilot program 1999 3 25 40 13 10 91 in Zhenjiang (3.30) (27.47) (43.96) (14.29) (10.99) City 2000 2 58 21 12 2 95 (2.10) (61.05) (22.11) (12.63) (2.10) Province-wide 2001 77 512 288 141 41 1059 program (7.27) (48.35) (27.20) (13.31) (3.87) 2002 182 1196 655 398 77 2508 (7.26) (47.69) (26.12) (15.87) (3.07) 2003 267 1545 789 367 106 3074 (8.69) (50.26) (25.67) (11.94) (3.44) 2004 329 2659 1467 525 114 5094 (6.46) (52.20) (28.80) (10.31) (2.24) 2005 530 4016 2614 702 143 8005 (6.62) (50.17) (32.65) (8.77) (1.79) 2006 702 5414 3944 1000 155 11215 (6.26) (48.27) (35.17) (8.92) (1.38) Sources: Pilot program in Zhenjiang: Wang et al. (2004). Province-wide program: the Legislative Affairs Office of the China State Council (2007) (available at http://www.chinalaw.gov.cn/article/dfxx/dffzxx/js/200706/20070600021431.shtml; last assessed on May 19, 2009). Table 2. City Comparisons for Industiral Pollution and Socioeconomic and Environmental Conditions, 2001 Huaian Wuxi Yangzhou Zhenjiang Socioeconomic conditions GDP per capita (Yuan) 14,359 37,700 21,311 18,852 Economic growth rate (%) 11.05 12.20 7.30 11.10 Unemployment rate (%) 3.84 3.62 3.60 2.30 Population (1,000) 558 2,131 1,097 628 Environmental conditions TSS: Total suspend solids (mg /m3) 0.158 0.144 0.237 0.105 SO2: sulfur dioxide (mg /m3) 0.037 0.056 0.023 0.024 NO2: nitrogen dioxide (mg /m3) 0.027 0.034 0.035 0.038 % of drinking water meeting standards 93.00 97.96 98.80 96.43 % of surface water meeting standards 83.00 91.67 62.00 88.89 Noise (dB(A)) 55.80 56.90 53.20 55.50 Total industrial pollution emissions Waste water (10,000 tons) 1,674 14,010 3,774 4,544 COD: chemical oxygen demand (tons) 1,708 N.A. 6,787 25,200 3 Waste gas (100 million m ) 129 471 461 1,895 Smoke (tons) 11,063 8,611 5,385 47,421 SO2 (tons) 8,863 21,492 35,765 96,377 Solid waste (10,000 tons) 1 8 N.A. 265 Sources: Municipal governments of the four cities. Table 3. Distributions of Sample Firms During 1997-2001 Total number of rated (R) and nonrated (NR) firms by city and year Year Status Huaian Wuxi Yangzhou Zhenjiang Total 1997 a Unrated 42 33 64 89 228 1998 a Unrated 46 26 71 81 224 1999 a Unrated 46 32 76 12 166 Rated 57 57 2000 Unrated 54 43 68 13 178 Rated 78 78 2001 Unrated 16 1 13 131 161 Rated 39 69 59 91 258 1997-2001 Unrated 204 135 292 326 957 Rated 39 69 59 226 393 Total 243 204 351 552 1350 Distribution of rated firms by rating colors* Green Blue Yellow Red Black Rated ZhenJiang 1 34 17 4 1 57 (1999) (1.75) (59.65) (29.82) (7.02) (1.75) Zhenjiang 2 48 19 8 1 78 (2000) (2.56) (61.54) (24.36) (10.26) (1.28) Huai'an 2 29 4 3 1 39 (2001) (5.13) (74.36) (10.26) (7.69) (2.56) Wuxi 15 20 20 8 6 69 (2001) (21.74) (28.99) (28.99) (10.59) (8.70) Yangzhou 2 52 5 0 0 59 (2001) (3.39) (88.14) (8.47) (0.00) (0.00) Zhenjiang 2 60 20 7 2 91 (2001) (2.20) (65.93) (21.98) (7.69) (2.20) All cities 24 243 85 30 11 393 (1998-2001) (6.11) (61.83) (21.63) (7.63) (2.80) * Figures in parentheses represent the percent of firms by rating colors. a Green Watch began in Zhenjiang in 1999, and in the other three cities in 2001. Thus, no firms were rated in 1997 and 1998, and only some firms in Zhenjiang were rated in 1999 and 2000. Table 4. Pollution Intensities and Effluent Concentrations for Unrated Firms and Firms Rated for the First Time Pollution Intensity Effluent Concentration Water TSS COD SO2 Dust/Smoke Gas TSS COD SO2 Dust/Smoke Zhenjiang launched its pilot Green Watch Program in 1999 Unrated 50.06 68.05 63.44 0.58 0.38 6.38 115.44 170.43 183.52 199.82 (103.13) (190.49) (177.37) (1.70) (0.81) (13.39) (51.90) (97.95) (343.55) (166.58) Rated 6.93 26.71 23.82 0.04 0.01 2.30 93.54 179.51 413.45 485.60 (19.94) (63.85) (67.41) (0.11) (0.06) (5.68) (79.60) (227.12) (522.20) (709.19) Equal mean test 3.64* 0.13 0.82 0.42 3.21* 0.52 3.77** 1.30 0.73 0.67 Equal median test 5.14** 0.15 2.73* 0.17 0.48 0.10 6.71*** 3.25* 1.99 2.58* Three cities (Huaian, Wuxi and Yangzhou) launched Green Watch programs in 2001 Unrated 76.74 8.79 1.89 0.16 0.11 24.17 95.83 278.24 967.66 253.96 (245.71) (28.67) (6.67) (0.49) (0.31) (102.11) (56.86) (304.49) (999.83) (218.00) Rated 143.92 30.34 119.72 0.13 0.06 28.81 120.27 229.84 1284.55 295.01 (492.41) (94.04) (556.27) (0.68) (0.28) (148.89) (335.57) (650.73) (2476.77) (784.40) Equal mean test 2.93** 0.06 6.83*** 1.67 1.25 2.42 5.06** 8.71*** 0.01 0.22 Equal median test 7.54*** 1.23 2.58* 0.05 0.83 1.52 7.02*** 10.57*** 0.14 0.23 Table 5. Estimation Results for Pollution Intensity Increases: Rated vs. Unrated Firms Water Pollution Air Pollution Waste Dust & Water TSS COD SO2 Waste Gas Smoke PRD -41.62 -15.47** -18.23 -0.05 -11.06 -0.02 (29.35) (7.82) (21.85) (0.06) (12.96) (0.03) Lagged pollution -0.56* -0.94*** -1.01*** -1.00*** -0.75*** -0.80*** intensity (0.31) (0.15) (0.17) 0.00 (0.27) (0.16) City dummies (base = Wuxi) Huanan -277.03*** -11.29*** -25.94 -0.02 -6.27* 0.03*** (26.99) (3.79) (23.11) (0.01) (3.24) (0.01) Yangzhou -259.84*** -6.37 -24.28 -0.09*** -14.56* -0.02*** (24.50) (8.66) (42.11) (0.02) (7.48) (0.01) Zhengjiang -287.32*** 0.07 -6.66 -0.07*** -11.41* 0.00 (36.43) (5.31) (37.96) (0.01) (6.51) (0.01) Firm size (base = small) Large 16.31** -14.18 -29.13 -0.07*** -6.53*** -0.02 (7.20) (10.29) (37.82) (0.02) (1.50) (0.02) Medium 30.02 -14.73** -22.42 -0.02 -0.01 -0.01 (30.30) (6.01) (25.97) (0.03) (6.48) (0.02) Ownership structure (base = private) State-owned -114.86** 14.66* 23.14 (0.03) (9.15) (0.02) (52.68) (8.27) (19.18) (0.09) (6.44) (0.05) Collectively- -113.49* 12.91*** 1.66 -0.06 -16.09** -0.02 owned (61.29) (3.53) (19.67) (0.08) (6.73) (0.05) HK, Macao, & -169.5 46.81*** -10.71 -0.04 -13.75* -0.04 Taiwan investor (158.32) (17.24) (19.96) (0.08) (7.60) (0.04) Foreign investor -17.23 9.4 -2.85 -0.06 -21.77*** 0.01 (46.86) (12.93) (5.59) (0.07) (8.24) (0.05) Companies with -113.59* 23.82* 51.00 (0.06) -17.30*** (0.03) limited shares (62.37) (13.86) (45.32) (0.07) (6.11) (0.04) Others -110.31*** 7.27 2.67 (0.08) -19.37** (0.04) (33.46) (7.43) (7.98) (0.07) (9.22) (0.04) Firm age (years) 0.36* -0.21** -0.12** 0.00 (0.08) 0.00 (0.21) (0.10) (0.05) 0.00 (0.13) 0.00 Industry (base = mining) Food & beverages 41.43 18.64 17.37 -0.26*** 4.98 -0.32*** (84.23) (16.99) (10.80) (0.03) (7.14) (0.07) Textiles and leather -57.54 11.97 25.70* -0.19** 3.85 -0.30*** (48.65) (9.28) (14.86) (0.08) (3.96) (0.06) Pulp & paper -39.94 47.03** 95.62 -0.26*** 3.27 -0.33*** (69.45) (19.59) (72.80) (0.04) (3.70) (0.07) Chemicals -18.07 27.17 86.95** -0.15** 8.61* -0.26*** (55.47) (18.02) (43.88) (0.07) (4.93) (0.07) Medical -28.76 7.18 46.21** -0.28*** 2.59 -0.33*** (50.67) (8.67) (18.42) (0.02) (3.73) (0.06) Fiber, rubber & -33.66 19.83 9.34 -0.26*** 20 -0.33*** plastic (65.93) (26.68) (15.45) (0.04) (12.85) (0.07) Smelting (52.41) 4.56 11.52 -0.29*** 0.81 -0.33*** (66.49) (12.51) (14.89) (0.03) (5.51) (0.07) Machinery (100.65) 11.87 12.53** -0.28*** 0.50 -0.33*** manufacture (89.26) (11.87) (6.02) (0.02) (1.63) (0.06) Utilities 145.69* 5.53 13.66 -0.12*** 33.50 -0.27*** (74.49) (16.52) (39.68) (0.03) (35.80) (0.08) Transportation (58.63) 20.67* (2.78) -0.30*** (1.72) -0.32*** (67.32) (11.46) (8.49) (0.06) (3.95) (0.08) Others -24.54 9.74 12.69 -0.24*** 1.12 -0.31*** (25.03) (16.18) (8.75) (0.04) (3.19) (0.06) Time trend 13.04 1.67 2.46 0.00 3.28 0.00 (8.84) (2.50) (6.27) (0.02) (4.18) (0.01) Constant 377.03*** (0.18) 2.11 0.43*** 18.97*** 0.36*** -107.85 -20.45 -34.02 -0.08 -6.75 -0.07 No. of obs 1320 1128 1296 1158 1229 1104 2 within R 0.15 0.61 0.73 0.99 0.49 0.82 2 between R 0.16 0.52 0.84 0.97 0.33 0.61 2 overall R 0.14 0.39 0.77 0.99 0.33 0.62 Breusch and Pagan Lagrangian Multiplier test for random effects Test statistics: 2(1) 0.04 0.19 0.05 75.13*** 5.27** 21.68*** Numbers in parentheses are standard errors of the estimated coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. Table 6. Estimation Results for Pollution Intensity Increases: Five-Color Ratings Waste Water Air Water TSS COD SO2 Waste Gas Dust/Smoke Rating dummies (base = not rated) Green -46.24 -4.37 -0.51 -0.05** -6.06** 0.01 (53.17) (6.63) (4.01) (0.02) (2.45) (0.01) Blue -12.35 -9.18*** -5.1 -0.03** -4.26 0.00 (8.56) (2.39) (4.58) (0.02) (6.20) (0.01) Yellow -16.32 -13.69* -30.51 -0.04* -5.84 -0.03** (18.19) (7.32) (23.78) (0.02) (3.88) (0.01) Red 24.03 -35.14*** -44.40*** -0.05** -5.48* -0.03* (30.88) (5.76) (10.95) (0.02) (2.87) (0.01) Black -11.4 -19.72*** -25.06*** -0.15 -5.3 -0.16* (32.89) (2.67) (8.34) (0.12) (5.81) (0.09) Lagged pollution -0.56* -0.94*** -1.00*** -1.00*** -0.75*** -0.80*** intensity (0.31) (0.15) (0.16) 0.00 (0.27) (0.16) City dummies (base = Wuxi) Huanan -270.10*** -11.20*** -27.55 -0.02* -5.79** 0.02** (23.91) (3.13) (22.27) (0.01) (2.32) (0.01) Yangzhou -251.97*** -6.64 -26.67 -0.10*** -14.27** -0.03*** (23.46) (7.87) (40.37) (0.02) (6.37) (0.01) Zhengjiang -275.58*** -0.24 -6.82 -0.07*** -12.54* 0.00 (30.74) (4.41) (33.79) (0.01) (7.21) (0.00) Firm size (base = small) Large 13.66** -15.44 -30.23 -0.07*** -8.09*** -0.02* (6.36) (11.08) (39.11) (0.02) (2.57) (0.01) Medium 24.35 -16.17** -23.81 -0.03 -2.02 -0.01 (30.51) (7.55) (29.53) (0.02) (4.53) (0.01) Ownership structure (base = private) State-owned -113.99** 15.31* 23.5 -0.03 -8.77 -0.02 (47.38) (9.17) (20.86) (0.09) (7.29) (0.05) Collectively owned -106.28** 14.33*** 2.7 -0.06 -15.24** -0.02 (47.70) (3.97) (17.08) (0.09) (6.85) (0.05) HK, Macao & -146.61 48.76*** -8.14 -0.04 -10.38* -0.04 Taiwan investor (178.46) (17.67) (27.85) (0.07) (5.78) (0.04) Foreign investor 19.3 11.12 -0.84 -0.06 -19.02*** 0.01 (56.03) (11.37) (7.19) (0.07) (6.22) (0.05) Companies with -99.66** 25.36* 52.34 -0.05 -16.26** -0.03 limited shares (48.54) (14.97) (50.35) (0.07) (6.31) (0.04) Others -92.33*** 11.51* 10.45 -0.07 -17.10** -0.03 (29.16) (6.84) (16.47) (0.06) (7.77) (0.04) Firm age (years) 0.32*** -0.21** -0.07 0 -0.1 0 (0.11) (0.10) (0.06) 0.00 (0.14) 0.00 Industry (base = mining) Food & beverages 50.26 19.02 16.87* -0.26*** 3.8 -0.31*** (89.18) (16.31) (9.34) (0.03) (5.06) (0.08) Textiles and leather -64 13.86* 27.70* -0.19** 3.33 -0.29*** (49.92) (7.39) (14.67) (0.09) (2.74) (0.07) Pulp & paper -45.68 47.79** 94.92 -0.27*** 2.27 -0.33*** (77.88) (20.04) (69.00) (0.04) (2.70) (0.07) Chemicals -25.34 27.33 85.24* -0.15** 7.33* -0.26*** (62.05) (18.48) (46.95) (0.07) (4.23) (0.07) Medical -32.38 8.39 47.90*** -0.28*** -0.05 -0.33*** (49.98) (8.36) (17.31) (0.02) (2.53) (0.07) Fiber, rubber & -38.3 19.67 5.93 -0.26*** 18.71 -0.32*** plastic (65.87) (26.08) (20.49) (0.04) 13.91) (0.07) Smelting -49.93 4.12 10.07 -0.29*** 0.03 -0.33*** (62.06) (11.75) (14.23) (0.03) (4.38) (0.07) Machinery -102.24 11.2 9.34*** -0.29*** -0.87 -0.32*** manufacture (87.87) (12.63) (3.44) (0.02) (1.05) (0.06) Utilities 154.48* 4.44 10.57 -0.12*** 31.06 -0.27*** (90.40) (16.83) (36.76) (0.04) 32.86) (0.08) Transportation -42.63 23.17*** -0.19 -0.29*** 3.27 -0.32*** (60.82) (6.49) (10.26) (0.04) (2.84) (0.07) Others -22.62 10.17 13.15* -0.24*** 0.85 -0.30*** (32.06) (15.22) (7.11) (0.03) (2.27) (0.06) Time trend 12.82 1.67 2.52 0.00 3.28 0.00 (8.56) (2.52) (6.32) (0.02) (4.22) (0.01) Constant 404.84*** 4.3 10.24 0.45*** 30.09*** 0.37*** (114.78) (25.46) (48.67) (0.09) (9.50) (0.09) No. of obs 1320 (4) 1128 (4) 1296(4) 1158 (4) 1229 (4) 1104 (4) 2 within R 0.15 0.61 0.73 0.99 0.49 0.82 2 Between R 0.16 0.52 0.84 0.97 0.33 0.61 2 overall R 0.14 0.39 0.77 0.99 0.33 0.62 Breusch and Pagan Lagrangian Multiplier test for random effects Test statistics: 2(1) 0.06 0.14 0.05 71.64*** 5.17** 23.77*** Numbers in parentheses are standard errors of the estimated coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. Table 7. Estimation Results for Pollution Concentration Increase: Rated vs. Unrated firms Water Air TSS COD SO2 Dust/Smoke PRD -21.76 -31.45 -32.72** -29.94** (17.67) (28.59) (14.69) (14.63) Lagged pollution intensity -0.50* -0.57*** -1.02*** -1.00*** (0.28) (0.03) (0.00) (0.00) City dummies (base = Wuxi) Huanan -44.52 -32.72*** -136.48* -133.30* (37.69) (8.98) (78.80) (72.49) Yangzhou 10.99 -85.95** -3 -29.81 (29.41) (36.22) (21.63) (29.67) Zhengjiang 9.11 -50.03* -40.24 -44.08 (37.23) (25.89) (32.61) (28.44) Firm size (base = small) Large 46.17 -11.1 91.68 89.42 (30.65) (25.74) (115.56) (105.23) Medium 31.62** 38.21 68.88** 74.29*** (16.11) (36.70) (33.95) (23.99) Ownership structure (base = private) State-owned -7.97 -48.77** 2.79 -10.12 (31.43) (22.30) (81.34) (99.86) Collectively-owned -24.28 -81.77 -75.84*** -78.17 (31.73) (50.92) (29.33) (51.18) HK, Macao & Taiwan investor -71.65** -123.66*** -150.52 -146.23 (35.61) (40.66) (95.66) (90.76) Foreign investor -23.82 -78.31** -70.96 -74.38 (41.25) (35.54) (48.88) (52.14) Companies with limited shares -30.64 -68.33 -104.19*** -87.09** (23.55) (48.15) (40.15) (43.20) Others -46.89** -103.96** -33.63 -31.56 (18.73) (47.75) (55.95) (69.48) Firm age (years) 0.17 0.7 0.43 0.38 (0.42) (0.87) (0.64) (0.68) Industry (base = mining) Food & beverages 31.87 -100.11* 70.79 3.7 (27.39) (51.64) (47.65) (26.30) Textiles and leather 14.72* -71.75 18.11 -25.07 (8.15) (60.39) (22.11) (23.43) Pulp, paper & print 41.38* -48.51 53.41* 6.48 (23.19) (30.64) (31.95) (43.91) Chemicals 68.33*** -24.58 155.20*** 96.44*** (22.19) (60.20) (37.76) (35.55) Medical 272.51 79.85 681.55* 571.31* (190.54) (108.69) (364.86) (345.14) Fiber, rubber & plastic 28.79 -54.47* 41.76 -18.84 (28.23) (28.32) (73.15) (17.63) Smelting 15.55 -28.35* 41.75 -21.98 (21.69) (15.67) (32.05) (30.01) Machinery manufacture 6.16 -54.18 -26.72 -59.85 (12.24) (47.44) (21.91) (41.71) Utilities 20.87** 2.2 0.62 -46.04 (10.09) (63.66) (48.36) (70.90) Transportation 144.51* 95.16*** 439.38*** 379.11*** (84.83) (10.56) (94.84) (58.23) Others 40.96*** -10.94 29.16 -20.65 (15.68) (44.81) (70.07) (38.23) Time 3.84 6.78 3.14 1.20 (8.03) (7.90) (18.50) (17.48) Constant -3.13 129.43** 70.99 130.67 (73.45) (51.03) (124.84) (103.53) No. of obs 967 914 659 664 2 within R 0.53 0.01 0.14 0.96 2 between R 0.11 0.76 0.98 0.95 2 overall R 0.24 0.57 0.96 0.96 Breusch and Pagan Lagrangian Multiplier test for random effects Test statistics: 2(1) 9.05*** 8.91*** 8.61** 8.52*** Numbers in parentheses are standard errors of the estimated coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. Table 8. Estimation Results for Pollution Concentration Increase: Five-Color Ratings Water pollution Air pollution TSS COD SO2 Dust/Smoke Rating dummies (base = not rated) Green -38.95 -55.95*** -46.42 -40.46 (25.97) (9.49) (70.40) (75.08) Blue -7.09 -3.83 -12.14 -7.70 (18.04) (19.48) (23.17) (23.33) Yellow -41.46 -63.43 -54.54*** -53.79*** (26.54) (39.15) (12.15) (10.35) Red -68.47** -77.8 -101.34* -100.56 (34.07) (59.15) (59.88) (62.64) Black 9.02 -109.21** 33.02 23.29 (11.58) (48.53) (28.47) (19.68) Lagged pollution intensity -0.50* -0.57*** -1.01*** -1.00*** (0.28) (0.03) (0.00) (0.00) City dummies (base = Wuxi) Huanan -47.82 -40.70*** -137.09* -134.01* (38.77) (8.29) (79.30) (73.40) Yangzhou 8.14 -93.54*** -4.91 -31.37 (30.45) (35.12) (24.20) (31.73) Zhengjiang 8.39 -54.57** -39.73 -43.68 (37.45) (24.41) (33.35) (29.12) Firm size (base = small) Large 47.88 -9.47 92.21 90.07 (29.90) (24.87) (116.47) (105.35) Medium 31.91** 37.46 69.04** 74.09*** (15.63) (37.44) (34.67) (24.31) Ownership structure (base = private) State-owned -7.00 -45.87** 4.27 -7.18 (32.37) (19.54) (84.70) (102.78) Collectively-owned -21.71 -77.1 -73.43** -74.79 (31.05) (47.51) (32.93) (52.46) HK, Macao & Taiwan investor -65.87* -116.33*** -140.2 -135.37 (35.41) (37.48) (96.19) (90.21) Foreign investor -17.34 -67.09** -59.55 -59.53 (38.43) (32.29) (46.26) (51.74) Companies with limited shares -28.13 -62.52 -101.31** -83.70* (23.66) (48.72) (41.87) (45.07) Others -42.68** -88.13** -30.49 -26.83 (21.08) (43.05) (65.66) (79.01) Firm age (years) 0.2 0.71 0.46 0.42 (0.46) (0.95) (0.67) (0.70) Industry (base = mining) Food & beverages 33.51 -100.96* 70.56 1.46 (30.03) (55.52) (49.40) (26.47) Textiles and leather 18.65** -67.9 22.38 -22.09 (8.78) (61.76) (18.45) (17.09) Pulp & paper 50.24* -44.35 69.28* 20.58 (27.29) (31.55) (36.85) (49.40) Chemicals 72.18*** -27.88 160.67*** 100.26*** (24.02) (60.28) (36.49) (31.95) Medical 279.24 79.76 686.83* 573.04* (194.64) (111.22) (366.71) (346.37) Fiber, rubber & plastic 29.2 -59.85* 40.74 -22.71 (28.77) (32.48) (74.32) (17.33) Smelting 18.62 -30.11* 46.8 -20.37 (23.73) (17.80) (34.14) (29.70) Machinery manufacture 7.39 -60.81 -24.94 -61.55 (12.60) (50.83) (19.93) (41.75) Utility 21.58** -4.32 0.34 -49.59 (9.66) (68.11) (47.24) (71.06) Transportation 147.44* 93.82*** 442.96*** 380.16*** (86.75) (11.30) (95.04) (57.75) Others 46.08*** -11.19 36.24 -16.45 (17.25) (47.13) (71.49) (39.33) Time trend 3.5 5.87 2.79 0.95 (8.26) (7.28) (18.68) (17.70) Constant -6.66 134.60*** 65.27 126.36 (70.41) (46.86) (122.78) (99.61) No. of obs (clusters) 967 (3) 914 (4) 659 (4) 664 (4) within R2 0.53 0.01 0.14 0.96 between R2 0.12 0.77 0.98 0.95 overall R2 0.24 0.57 0.96 0.96 Breusch and Pagan Lagrangian Multiplier test for random effects Test statistics: 2(1) 9.05*** 8.91*** 8.61** 8.52*** Numbers in parentheses are standard errors of the estimated coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. Figure 1. Rating Criteria of the Green Watch Program in Jiangsu No Comply with concentration standards > 60% Yes No Hazardous waste disposal = 100% Yes No Yes No Comply with load-based standards Yes Yes Illegal behaviors No Yes Administrative penalty No Yes Penalty > 50,000 Pollution accidents Yes No No Yes Serious pollution accidents Internal management requirements Yes No No Utilization of solid waste > 80% No Yes Public complaints Yes No Cleaner production No Yes ISO 14000 No Yes GREEN BLUE YELLOW RED BLACK Source: Revised based on Figure 1 in Wang et al. (2004).