Impact of Informal Regulation of Pollution on Water Quality in Rivers in India Bishwanath Goldar* and Nandini Banerjee** Abstract This paper presents an econometric analysis of determinants of water quality in Indian rivers. Water quality (water class) data for 106 monitoring points on 10 important rivers for five years, 1995 to 1999, are used for the analysis. To explain variations in water quality, an Ordered Probit model is estimated, in which industrialization, urbanization, irrigation and fertilizer use in agriculture, and poll percentage (a proxy for informal regulation) are taken as the main explanatory variables. A significant negative relationship is found between the level of industrialization of a district and the water quality in rivers at monitoring point(s) falling in the district. Similarly, irrigation and fertilizer consumption in agriculture are found to bear a significant negative relationship with river water quality. A significant positive relationship is found between poll percentage and water quality, and also between the proportion of people who have completed school education in a state and the water quality in rivers flowing through the state. These results point to a significant favorable effect of informal regulation of pollution on water quality in rivers in India. JEL: Q 25 [Paper presented at the Second World Congress of Environmental and Resource Economists, Monterey, June 24-27, 2002] ________________ * ** Institute of Economic Growth, University Enclave, Delhi – 110007, India National Council of Applied Economic Research, New Delhi, India Impact of Informal Regulation of Pollution on Water Quality in Rivers in India Bishwanath Goldar and Nandini Banerjee 1. Introduction Water pollution is a major environmental concern in India. The main sources of water pollution are: (1) discharge of domestic sewage and industrial effluents, which contain organic pollutants, chemicals and heavy metals, and (2) run-off from land-based activities such as agriculture. Agricultural development is causing deterioration in water quality in Indian rivers in two ways. First, with increasing use of fertilizers and pesticides in agriculture, the run-off from irrigated lands is polluting the water bodies. Secondly, because of the growing irrigation intensity and the high rates of abstraction of ground and surface water for this purpose, rivers at many places do not have sufficient water for dilution of industrial effluents/ domestic sewage1, aggravating thereby the problem of water pollution. It is known that the present formal environmental regulation system in India, based on ‘command and control,’ has not performed well in controlling pollution of Indian rivers. Despite a strong legal framework and the existence of a large bureaucracy for dealing with environmental regulation, the public perception is that implementation remains weak (Pargal, Mani and Huq, 1997; Murty, 1999; Murty and Prashad, 1999). Water quality monitoring data (discussed later) reveal that at many monitoring points on India rivers, the water quality is poor. Also, there is a good deal of anecdotal evidence on high levels of pollution of Indian rivers, especially in relatively more industrialized states and in the vicinity of big cities.2 The object of this paper is to analyze variations in water quality across different monitoring points on Indian rivers with a view to identifying determinants of water quality. For this purpose, an econometric model is estimated. Since industrialization, urbanization, and irrigation and fertilizer consumption in agriculture are expected to be important factors affecting water quality, these variables are included in the model as determinants of water quality. An important focus of the study is on the impact of informal regulation of pollution on ambient water quality, motivated by a growing body of literature on the role of informal regulation in controlling pollution. Accordingly, some proxy variables for informal regulation (reflecting the strength of informal regulation) are included in the econometric model for capturing the effect of informal regulation on water quality. 1 Chopra and Goldar (2000); Central Pollution Control Board, Ministry of Environment and Forests, Government of India, Annual Report, 2000-01. 2 See, for example, State of India’s Environment, Centre for Science and Environment, New Delhi. 1 The significance of informal regulation for achieving environmental goals is well recognized in the literature.3 It is known that when formal regulation is weak or absent, informal regulation through local community participation can force the polluter to abate pollution. Informal regulation takes many forms, including demand for compensation by community groups, social ostracism of the polluting firm’s employees, the threat of physical violence, and efforts to monitor and publicize the firm’s emissions (Pargal, Hettige, Singh and Wheeler, 1997). Two “formal” channels of informal regulation are (1) to report violation of legal standards to the regulatory institutions (where such standards and institutions exist), and (2) to put pressure on regulators (politicians and administrators) to tighten their monitoring and enforcement. Education, degree of political organization and environmental awareness are considered to be important factors determining the strength of informal regulation, which is influenced also by information, legal or political recourse, media coverage, the presence of non-government organizations (NGOs), and the efficiency of existing formal regulation. Many of these factors are correlated with the community income levels. Therefore, in several studies (for example, Pargal, Hettige, Singh and Wheeler, 1997), the mean community income level (or the development level) has been taken as a proxy variable in econometric analysis to capture the effect of informal regulation. There have been two earlier studies on informal regulation of water pollution in India, one by Pargal, Mani and Huq (1997) and the other by Murty and Prashad (1999) (discussed later). Both studies have considered the discharge of effluents by large and medium scale industries and examined how this is influenced by the characteristics of local communities. The present study is different from the studies of Pargal-Mani-Huq and Murty-Prashad in that it is concerned with ambient water quality, not industrial discharge of effluents. Hence, it has a more comprehensive coverage of the sources of water pollution. The rest of the paper is organized as follows. Sections 2 and 3 discuss the system of formal regulation of water pollution in India and the existing water quality monitoring network, providing thereby background information for the study. Section 4 quickly reviews the two earlier studies undertaken in the Indian context on informal regulation of industrial water pollution. Section 5 discusses the data and methodology used for this study. Section 6 presents the econometric results. Section 7 summarizes and concludes. 2. Regulation of Water Pollution in India The legal provisions that empower the Indian government to enforce environmental regulations are the Water (Prevention and Control of Pollution) Act (1974) and the Environment (Protection) Act (1986).4 The Water Act prescribes both general and 3 See, for example, Pargal and Wheeler (1996), Afsah, Laplante and Wheeler (1996), Pargal, Hettige, Singh and Wheeler (1997) and Pargal, Mani and Huq (1997). 4 For discussion on formal regulation of water pollution in India, see Pargal, Mani and Huq (1997), Kuik et al. (1997, Chapter 4), Murty (1999), Sankar (1999) and Goldar and Pandey (2001), among others. 2 industry specific standards for the discharge of wastewater into water bodies. Discharge of wastewater, carrying pollutant concentrations beyond the specified standards, into surface waters, public sewers, on land for irrigation and marine coastal waters is prohibited. The Act lays down penalties for non-compliance. These standards uniformly apply to all firms within an industry, or to all firms in general (where specific standards do not exist). The standards differ according to the class of water bodies into which the wastewater is discharged (for example, the standards are most strict for discharge into surface water bodies and relatively less strict for disposal on land for irrigation). The pollution standards are concentration based, i.e. they are specified as milligrams (mg) of pollutant per liter of wastewater discharged. The Environment Act provides the Central Government with greater powers to set and enforce environmental standards than what was provided in the Water Act. However, the basic features pertaining to industrial pollution abatement remain the same. There is a basic division of power between the center and the states in India in regard to environmental regulation, reflecting the federal nature of the Indian Constitution. The mandate of the Central Pollution Control Board (CPCB) is to set environmental standards for all plants in India, lay down ambient standards, and coordinate the activities of the State Pollution Control Boards (SPCBs). The implementation of environmental laws and their enforcement, however, are decentralized, and are the responsibility of the SPCBs. Anecdotal evidence suggests wide variations in enforcement across the states (Pargal, Mani and Huq, 1997). SPCBs have the legal authority to conduct periodic inspections of plants to check whether they have the appropriate consent to operate, whether they have effluent treatment plants, take samples for analysis, etc. Some of these inspections are also programmed in response to public requests and litigation. There are penalties for noncompliance. Until 1988, the enforcement authority of the SPCBs was very weak. But, now, the SPCBs have the power to close non-compliant factories or cut-off their water and electricity by administrative orders. Highly Polluting Industries In 1992, the CPCB identified 1551 large and medium industrial units in 17 categories of highly polluting industries, contributing the major part of the industrial pollution load. The industrial units were given a time schedule to install necessary pollution control equipment to comply with the prescribed standards. In 1993, out of the 1551 identified industrial units, 540 were defaulters, i.e. these did not have pollution control equipment to comply with the standards. Over time, more and more of the defaulting industries have installed the necessary pollution control equipment or have closed down. By December 2000, out of 1551 industrial units, 1350 had installed the necessary pollution control facilities, 177 had been closed down and the remaining 24 industrial units were defaulting. A state-wise summary status of the pollution control in the 17 categories of highly polluting industries is given in Annex I. 3 Industrial Pollution Control along the Rivers and Lakes A program was initiated in 1993-94 to identify polluting industries along the Indian rivers for priority action for control of industrial discharge into rivers. In July 1997, the National River Conservation Authority decided that the polluting industries which were discharging their effluents into rivers and lakes should be directed to install requisite effluent treatment systems within three months failing which closure notices should be issued. A total of 851 industrial units were identified in 1997 which were discharging 100 kg/day or more of BOD (Biological Oxygen Demand). In the course of the next few years, most of these units installed the requisite effluent treatment systems. The number of defaulters declined from 851 in August 1997, to 574 in June 1998, 514 in March 1999 and 22 in December 2000. As of December 31, 2000, 233 out of the 851 identified polluting industrial units had been closed, 596 had installed the requisite effluent treatment systems, and 22 remained defaulters. The state-wise break-up of identified industrial units and defaulting units is shown in Annex II. It is evident from the above that during the second half of the 1990s the pollution control authorities at the central and state level had made most of the water polluting large and medium scale industries install the requisite wastewater treatment facilities. It is, however, difficult to say whether this has had a large effect in terms of reducing discharge of pollutant by industries into the rivers. Doubts arise on this point because even if an effluent treatment plant is installed, it may not be operated regularly. Indeed, there is a perception that the wastewater treatment plants in many industrial units in India are activated only at the time inspections are scheduled to occur (Pargal, Mani and Huq, 1997). Urban Wastewater It should be mentioned here that disposal of untreated wastewater from the cities and towns is a major cause of water pollution of rivers in India. It is known that the cities and town are generating large volumes of wastewater, of which only a small part is treated. To give some facts, a study undertaken by the CPCB has brought out that in 1994-95 class-I cities in India (299 cities) generated 16,663 mld of wastewater of which only about 24% was treated.5 6 The rest was disposed without treatment. A similar study of the class-II towns (345 towns) done by the CPCB has brought out that in 1995 the volume of wastewater generated in class-II towns was 1649 mld of which only about 4% was treated.7 5 Status of water supply and wastewater generation, collection, treatment and disposal in class-I cities, Central Pollution Control Board, Delhi, February 2000. 6 The proportion of wastewater treated was 20.5% in 1989-90. 7 Status of water supply and wastewater generation, collection, treatment and disposal in class-II towns, Central Pollution Control Board, Delhi, March 2000. 4 The present legislation gives the CPCB sufficient power for control of water pollution. For example, the CPCB is empowered to lay down and maintain ambient water standards, to demand information regarding effluent emissions, to shut down polluting activities and prevent new discharges of effluent and sewage. However, the legislation has had limited success in checking water pollution arising from discharge of untreated sewage. The rise of public opinion against polluting industries has found expression in Public Interest Litigation (PIL). But, this has not been matched by a similar challenge to municipal pollution. A few cases have been filed against municipalities for their failure to keep the cities clean and maintain environmental standards (for example, the Ratlam case). These cases, filed by citizen, manifest that there is some degree of informal pressure on municipalities (at least on some of them) to perform their duties for the control of pollution. The media has probably also played a role in making the municipalities pay some attention to the environmental problems. It must be admitted, however, that neither formal nor informal pressure has been very successful in making the municipalities undertake collection and treatment of urban wastewater. 3. Water Quality Monitoring Water quality monitoring program was started by the CPCB in 1976 with 18 stations on river Yamuna. The program was gradually extended over time. In 1989, there were 324 monitoring stations. At present, there are 507 monitoring stations in the country spread over all important water bodies. Out of 507 stations, 414 stations are on rivers, 25 on ground water, 38 on lakes and 30 on canals, creeks, drains, ponds etc. The quality of water is monitored for 25 physico-chemical and biological parameters. The monitoring network covers 126 rivers (including the tributaries), wells, lakes, creeks, ponds, tanks, drains and canals. Five classes are used for water quality, A to E. A is the best quality, and E is the worst. At certain stations, the water quality is found to be below E. The criteria for the water classes are shown in Table 1. The distribution of river water quality at the monitoring stations (in terms of water class) during 1999 is shown in Table 2. At 4.7% of the monitoring stations, the water class was A. At about 60% of the monitoring stations, the water class was D, E or “below E”. On the basis of last 10 years’ water quality monitoring results, the CPCB has estimated riverine length having different level of pollution. This is presented in Table 3. The estimates indicate that about 14% of the length of river stretches are highly polluted (BOD level above 6 mg/l), while another 19% are moderately polluted (BOD level in the range of 3 to 6 mg/l). Thus, about two-thirds of the length of river stretches are relatively clean (BOD level less than 3 mg/l). 5 Table 1: Primary water quality criteria for designated-best-use-classes .-----------------------.----------.------------------------------. | Designated-Best-Use | Class of | Criteria | | | water | | |-----------------------|----------|------------------------------| | Drinking Water | A | 1. Total Coliforms Organism | | Source without | | MPN/100ml shall be 50 or | | conventional | | less | | treatment but after | | 2. pH between 6.5 and 8.5 | | disinfection | | 3. Dissolved Oxygen 6mg/l or | | | | more | | | | 4. Biochemical Oxygen Demand | | | | 5 days 20oC 2mg/l or less | |-----------------------|----------|------------------------------| | Outdoor bathing | B | 1. Total Coliforms Organism | | (Organised) | | MPN/100ml shall be 500 or | | | | less | | | | 2. pH between 6.5 and 8.5 | | | | 3. Dissolved Oxygen 5mg/l or | | | | more | | | | 4. Biochemical Oxygen Demand | | | | 5 days 20oC 3mg/l or less | |-----------------------|----------|------------------------------| | Drinking water | C | 1. Total Coliforms Organism | | source after | | MPN/100ml shall be 5000 | | conventional | | or less | | treatment and | | 2. pH between 6 to 9 | | disinfection | | 3. Dissolved Oxygen 4mg/l or | | | | more | | | | 4. Biochemical Oxygen Demand | | | | 5 days 20oC 3mg/l or less | |-----------------------|----------|------------------------------| | Propagation of Wild | D | 1. pH between 6.5 to 8.5 | | life and Fisheries | | 2. Dissolved Oxygen 4mg/l or | | | | more | | | | 3. Free Ammonia (as N) 1.2 | | | | mg/l or less | |-----------------------|----------|------------------------------| | Irrigation, | E | 1. pH between 6.0 to 8.5 | | Industrial Cooling, | | 2. Electrical Conductivity | | Controlled Waste | | at 25oC micro mhos/cm Max.| | disposal | | 2250 | | | | 3. Sodium absorption Ratio | | | | Max. 26 | | | | 4. Boron Max. 2mg/l | '-----------------------'----------'------------------------------' Source: Central Pollution Control Board, Ministry of Environment and Forests, Government of India 6 Table 2: Distribution of Monitoring Points according to Water Class, 1999 Class A B C D E Below E All % of monitoring points 4.7 14.1 19.7 54.6 4.7 2.2 100.0 Source: Computed from Water Quality data of the CPCB, available at their website. Table 3: Riverine length according to pollution level Pollution level Definition used High pollution Moderate pollution Relatively clean All BOD >6 mg/l BOD= 3-6 mg/l BOD<3 mg/l length of river stretches (km) 6086 8691 30242 45019 Per cent 14 19 67 100 Source: Prepared from data provided in the Annual Report of the CPCB, 2000-01. The water quality monitoring results obtained in recent years indicate that the critical source of pollution of Indian aquatic resources lies in organic and bacterial contamination.8 This is attributable mainly to discharge of domestic wastewater in untreated form from the urban centers, which may be explained by the fact that the municipal corporations do not have adequate resources for treating the ever-increasing load of municipal sewage. A related factor, aggravating the pollution problem, is that the receiving water bodies do not have sufficient water for dilution. It goes without saying that the problem of pollution of rivers caused by untreated urban sewage is compounded by the discharge of industrial effluents. 4. Earlier Studies on Informal Regulation of Pollution in India There have been two earlier studies on informal regulation of water pollution in India, one by Pargal, Mani and Huq (1997) and the other by Murty and Prashad (1999). The studies are briefly discussed below. Pargal, Mani and Huq (1997) have used survey data for 250 industrial plants 9 to examine regulatory inspections and water pollution emissions in India, and to check whether the monitoring and enforcement efforts of provincial pollution control authorities 8 9 CPCB, Annual Report, 2000-01. Data were collected in a survey of industrial plants in eight states of India, conducted in early 1996. 7 are affected by local community characteristics that act as proxies for political power. They have also tested for informal pressure on plants that would result in negotiated lower emissions. Two main plant level variables used in the study are emissions and inspections. The former is measured by the BOD load, and the latter by the total number of inspections that each plant had been subject to between 1990 and 1994. Pargal et al. note that in their sample of 250 plants, 51 plants indicated that they had undertaken abatement in response to NGO pressure and 102 said they had done so in response to complaints from neighboring communities. This brings out the importance of NGOs and community pressure in making industrial plant undertake control of industrial pollution An important finding of the study is that high levels of pollution elicit a formal regulatory response in the form of inspections in India. However, no significant impact of inspections is found on emissions, which is interpreted by the authors as reflective of bureaucratic or other problems in following through. As regards informal sources of pressure on industrial plants, the results of the analysis indicate a positive relationship between district development index (taken as a proxy for informal regulation) and the number of inspections, but no significant negative relationship is found between district development index and BOD emissions. This, according to the authors, could be due to community activism being unrelated to levels of urbanization, income, and education so that dirty plants are targeted, irrespective of where they are located, or these findings could mean that direct community pressure on plants is not a major determinant of emissions reduction in the sample of Indian firms covered in the study. Another inference drawn by the authors is that the community pressure that exists is probably channeled through the formal mechanism rather than through direct negotiation with plants. All in all, the econometric analysis undertaken by Pargal, Mani and Huq provides no evidence of successful informal pressure on plants in India, even though many firms covered in the survey said that they had undertaken pollution abatement in response to NGO pressure or complaints of neighboring communities. Murty and Prashad (1999) have carried out for Indian industry an analysis similar to that of Pragal, Mani and Huq (1997). They, however, find evidence of significant informal pressure on firms to control pollution. Murty and Prashad have used cross-section data for a sample of 100 factories belonging to 11 highly water polluting industries in 13 states, collected in a survey done in 1994/95. The ratio of BOD concentration in effluent to that in influent is taken as the dependent variable, an index of the reduction in pollution due to efforts made by the factory. In the regression equation estimated, they use a number of explanatory variables, including community specific variables like district development index and the rate of participation in the previous parliamentary elections. 8 The results of the regression analysis show a significant negative relationship between the index of development and the BOD ratio, implying thereby that higher levels of development are associated with greater abatement of pollution by factories. Similarly, a significant negative relationship is found between political participation of local communities as indicated by their rate of participation in recent elections and the BOD ratio. The implication is that the more active the local people are politically, the higher is the extent of pollution abatement done by the factories located in that area. The main conclusion of the study of Murty and Prashad is that the affluence of local communities and the level of political activity reflected in their rate of participation in elections play important roles in determining the degree of compliance of industries with water pollution standards. Needless to say, this is indicative of significant informal pressure on firms to curb water pollution. 5. Data Sources and Methodology of the Present Study As mentioned earlier, this study is concerned with ambient water quality in India. For the analysis, data on water quality for 10 rivers 10 (106 monitoring points) for five years, 1995 to 1999 are used. These data, collected under the water-quality monitoring program of the government, have been taken from the website of the CPCB. The timeseries and cross-section data are pooled to estimate an econometric model for explaining variations in water quality. Since the available water quality data are in terms in water class at different monitoring points, an ordinary regression analysis cannot be applied. An Ordered Probit model has therefore been applied for the econometric analysis of water quality. One would expect water quality in rivers to be affected by industrialization, urbanization, irrigation and fertilizer use in agriculture. These variables have therefore been included in the econometric model as explanatory variables. To obtain data on these variables, the water quality monitoring points have been mapped into districts (since the data on these explanatory variables could be obtained only at that level). Data on industrialization, urbanization, irrigation and fertilizer use have been taken from a publication of the CMIE (Center for Monitoring Indian Economy), entitled ‘District Profile 2000’.11 An important explanatory variable used in the model is the poll percentage (as an indicator of informal regulation). In this case, it has been necessary to map the water 10 The ten rivers (including tributaries) covered in the study are Brahmani, Cauvery, Chambal, Ganga, Godavari, Mahanadi, Mahi, Sabarmati, Tapi and Yamuna. 11 The ratio of workers engaged in manufacturing to all workers (full-time workers or “main workers” as defined by the population census) in a district is taken as an indicator of the level of industrialization. These data relate to 1991 (not available for the mid-1990s). An indicator for the level of urbanization is formed by taking the ratio of urban households to total households in a district. In this case, again, the data relate to 1991. Indicators of irrigation intensity (gross irrigated area as % of gross cropped area) and fertilizer use (kg of fertilizer consumption per hectare) are based on data for 1995. 9 quality monitoring points into the parliamentary constituencies. Poll percentage data for different parliamentary constituencies of the 11th Loksabha (1996) have been collected from Election Commission of India office. In addition to the variables listed above, two other variables have been included in the model for which data could be obtained only at the state level. Thus, in these cases, the figure obtained for a state has been applied to all the observations belonging to the state. These two variables are (1) Per capita gross domestic product (in 1997-98), data taken from Economic Survey, Government of India, and (2) Percentage of people who have completed school education (higher secondary) (among people aged six years or above), data taken from National Family Health Survey II (1997-98). 6. Empirical Results It would be useful to present first some summary statistics on water quality for the 106 monitoring stations covered in the study. Table 4 shows the distribution of these monitoring points according to water quality (water class) for three years, 1995, 1997 and 1999. It is seen from the table that during the period 1995 to 1999, there has been, on average, an improvement in river water quality. The shares of D and E class have declined while the shares of A and B class have gone up between 1995 and 1999. Table 4: Distribution of Monitoring Points according to Water Class Water class A B C D E or lower All Percentage of monitoring points falling in the water class in the year: 1995 1997 1999 0 16 14 62 8 100 4 19 14 56 7 100 4 27 19 45 5 100 Note: The distribution shown is for the 106 monitoring points covered in the study. From the popular writings on India’s environment one would get the impression that the water quality in Indian rivers has been going down. The finding of an improvement in average water quality for the sample of 106 monitoring points therefore calls for a closer scrutiny. Accordingly, a more detailed examination of water quality information was undertaken for 13 major rivers (including the ones covered in the study) for the years 1995 (277 points) and 1999 (325 points). The comparison did not reveal any significant improvement (or deterioration) in the average water quality between 1995 and 1999. It was realized, however, that the data for 1999 contain water quality information in respect of many new monitoring points, and this may affect inter-temporal 10 comparability of the results. When the new points were excluded, the water quality monitoring results indicated a slight improvement in average water quality in rivers between 1995 and 1999. Estimates of the Ordered Probit Model Table 5 presents the estimates of the Ordered Probit model. Data for all the five years are used for the estimation of the model. It should be pointed out here that the data on each of the explanatory variables relate to a specific year. For example, the poll percentage data relate to 1996, and the same set of figures is used for five years, 1995 to 1999. The implication is that the explanatory variables are able to explain only the crosssectional variation in water quality, not inter-temporal changes. Table 5: Determinants of Water Quality in Indian Rivers: Estimates of the Ordered Probit Model Dependent Variable: Water Class Period: 1995-1999 Explanatory Estimates of the Model variables (1) (2) (3) Poll percentage 0.0285 (6.3)*** 0.0228 (5.4)*** 0.0212 (5.2)*** Urbanization 0.0086 (2.5)** -0.0000 (-0.01) 0.0087 (2.6)*** Industrialization -0.0859 (-4.1)*** -0.0655 (-3.2)*** -0.0841 (-4.3)*** Per capita GDP in -0.0997 (-5.1)*** -0.1418 (-6.9)*** the state % population 0.2644 (4.6)*** 0.2671 (4.9)*** completed school education (higher secondary) Fertilizer use -0.0036 (-2.6)*** -0.0028(-2.4)** intensity Irrigation intensity -0.0136 (-5.1)*** Year dummy ‘96 Year dummy ‘97 Year dummy ‘98 Year dummy ‘99 -0.0931 (-0.6) 0.2113 (1.3) 0.4253 (2.7)*** 0.5241 (3.3)*** -0.1173 (-0.8) 0.2018 (1.3) 0.3950 (2.6)*** 0.4832 (3.1)*** -0.0913 (-0.6) 0.2067 (1.3) 0.4150 (2.7)*** 0.5087 (3.3)*** LR chi-square (df) No. of observations 97.0 (10) 511 129.1 (10) 521 69.2 (8) 511 Note: Figures in parentheses give t-ratios. Due to gaps in data, all the 530 observations could not be used for the estimation of the model. * statistically significant at ten per cent level of significance. ** statistically significant at five per cent level of significance. *** statistically significant at one per cent level of significance. 11 Since water quality data for different years are pooled for the estimation of the model, dummy variables for years have been included. The purpose is to capture the effect of year-specific factors. This will also capture any trend effect on water quality. It is seen from Table 5 that the coefficient of the poll percentage variable (relating to the 1996 parliamentary elections) is positive and statistically significant at one per cent level (see also Figure 1 which reveals a cross-district positive association between poll percentage and water quality in rivers). This finding is in agreement with the results of Murty and Prashad (1999). The finding of a significant positive relationship between poll percentage and ambient water quality points to the existence of successful informal regulation of pollution in India with a significant favorable effect on water quality in rivers. This conclusion is reinforced by the positive relationship found between water quality and school education – the coefficient of the education variable is positive and statistically significant. It is needless to say that a high proportion of educated people in the community should render informal regulation strong, and therefore a positive relationship between water quality and education may be interpreted as a reflection of the informal regulation of pollution.12 Fig. 1: Poll Percentage in Constituencies and Water Quality at Corresponding Monitoring Points 100% 80% A+B 60% C 40% D+E+Below E 20% 0% low poll percentage high poll percentage Note: A comparison of water quality is presented in the figure between the bottom one-third and top one-third of the locations in terms of poll percentage. The distribution of monitoring points according to water class is shown in the figure. The poll percentage data are for 1996. The water quality observations are for five years, 1995 to 1999. See table 1 for the definition of water classes. 12 It may be pointed out here that literacy was tried as an explanatory variable in the model. But, the results were not found satisfactory. 12 Turning to other explanatory variables, a significant negative relationship is found between water quality and industrialization (measured by the ratio of workers engaged in manufacturing activities out of total workers in a district). Such a relationship is expected. Given the weakness of the formal regulation system, a high level of industrialization should result in higher pollution load and hence lower water quality in rivers. Further, if industries give high employment benefits (reflected in the proportion of workers engaged in industries), it would be difficult for the local community to put much pressure on industries to curb pollution, a point already recognized in the literature. The results in respect of the irrigation (gross irrigated area as % to gross cropped area) and fertilizer use (kg per hectare) variables indicate that the higher the extent of irrigation and fertilizer use in a district, the lower is the water quality in rivers (other things remaining the same). This provides empirical support to a point made in the introductory section of the paper that agricultural development is adding to water quality problems in rivers in India, because of run-off from irrigated lands and because the water available in rivers is reduced to levels insufficient for dilution of pollutants. Since fertilizer use and irrigation are highly correlated, both cannot be included in the same equation. This is the reason why in equation (2), fertilizer has been replaced by irrigation. The coefficient of GDP per capita is negative and statistically significant. This may be interpreted as showing the adverse effect of economic development on environment (in the initial stages). It may be mentioned here that Murty and Prashad (1999) have used this variable to explain the extent of pollution abatement in Indian factories. They found an inverse relationship between GDP per capita and the extent of pollution abatement done. As regards urbanization, the coefficient is found to be positive and statistically significant in some of the equations estimated. Inasmuch as the discharge of urban wastewater is a major cause of pollution of Indian rivers, one should expect a negative relationship to arise between urbanization and water quality. It needs be noted, however, that the income level or development level of communities is often taken as a variable to represent informal regulation. Urbanization is obviously an important component of development and is correlated with development. Accordingly, the finding of a positive coefficient of urbanization in the estimated model may be treated as a reflection of the favorable effect of informal regulation on water quality. Turing to of the year dummies, the coefficients for 1998 and 1999 are found to be positive and statistically significant (1995 is the excluded category). This probably captures the effects of certain factors which contributed to improvement in water quality in the late 1990s. This is obviously consistent with the water quality data reported in Table 4. 13 Since the coefficients for the year dummies for 1998 and 1999 are found to be statistically significant, the Ordered Probit model has been estimated separately for the periods 1995-97 and 1998-99. The results are reported in Table 6. In respect of poll percentage and the proportion of people having higher secondary school education, the results reported in Table 6 are by and large similar to those reported in Table 5. Thus, there is evidence of effective informal regulation of pollution both in the period 1995-97 and in the period 1998-99. It is interesting to note, however, that the coefficient of the industrialization variable is statistically significant in the estimates for 1995-97 but not for 1998-99. On the other hand, the coefficient of irrigation is higher in numerical value and statistical significance in the estimates for 1998-99 as compared to the estimates for 1995-97. It seems therefore that the adverse effect of industrialization on water quality in rivers became weaker, while the adverse effect of irrigation became stronger, during the late 1990s as compared to the mid-1990s. Table 6: Determinants of Water Quality in Indian Rivers: Estimates of the Ordered Probit Model, 1995-97 and 1998-99 (Dependent Variable: Water Class) Explanatory variables Poll percentage Urbanization Industrialization Per capita GDP in the state % population completed school education (higher secondary) Irrigation intensity Year dummy ‘96 Year dummy ‘97 Year dummy ‘99 Estimates of the Model For 1995-97 0.0212 (3.8)*** -0.0000 (-0.01) -0.1045 (-3.9)*** -0.1696 (-6.2)*** For 1998-99 0.0246 (3.8)*** -0.0003 (-0.06) -0.0139 (-0.5) -0.1224 (-3.8)*** 0.2962 (4.2)*** 0.2640 (3.1)*** -0.0083 (-2.4)** -0.0213 (-5.0)*** -0.1314 (-0.8) 0.2172 (1.4) 0.0745 (0.5) LR chi-square (df) 73.3 (8) 57.7 (7) No. of observations 313 208 Note: Figures in parentheses give t-ratios. ** statistically significant at five per cent level of significance. *** statistically significant at one per cent level of significance. 14 The above finding that the level of industrialization bore a significant negative relationship with water quality in rivers in the mid-1990s, but not in the late 1990s finds corroboration when trends in water quality are analyzed for the category of relatively more industrialized districts. Figure 2 presents a comparison of water quality in such districts in 1995/96 with that in 1998/99. The figure brings out clearly that there was a marked improvement in water quality in these districts during the second half of the 1990s. This must have narrowed the gap in water quality between more industrialized and less industrialized districts, and therefore made the inverse relationship between industrialization and river water quality weaker. Fig. 2: Distribution of Monitoring Points in terms of Water quality (class) in relatively more industrialised districts 100% 80% B C D+E 60% 40% 20% 0% 1995/96 1998/99 Note: Distribution of monitoring points in 1995/96 and 1998/99 by water quality (class) is shown in the graph for the relatively more industrialized districts (top 25% districts in terms of the level of industrialization). For the definition of water classes, see Table 1. What caused the marked improvement in water quality in relatively more industrialized districts during the second half of the 1990s? It seems the credit should partly go to the pollution control boards. It was noted in Section 2 above (see Annex I and II) that during this period, the boards compelled most water polluting large and medium scale industries to install requisite wastewater treatment facilities. Further, informal pressure (particularly, public interest litigation) must have also played an important role in making industries undertake pollution abatement. 15 7. Conclusion The paper investigated determinants of water quality in rivers in India. Data on water quality (water class) for 106 monitoring points for five years 1995 to 1999 were used for the analysis. Given the nature of the water quality data, an ordinary regression analysis could not be applied. The econometric analysis was therefore undertaken by using an Ordered Probit model. A significant negative relationship was found between the level of industrialization, irrigation and fertilizer consumption intensity in a district and the water quality in the monitoring points falling in the district. A significant positive relationship was been found between poll percentage and water quality. The results also indicated a positive relationship between the proportion of population having completed school education (higher secondary) in a state and water quality in rivers flowing through the state. These results may be taken as indicative of a significant favorable effect of informal regulation on water quality in rivers in India. Econometric analysis brought out that the adverse effect of industrialization on water quality in rivers declined in the late 1990s compared to the mid-1990s, while the adverse effect of irrigation on water quality in rivers intensified in this period. The lessening of the adverse effect of industrialization on river water quality is perhaps a consequence of the increased efforts of the pollution control authorities directed at making industries undertake pollution abatement. Informal pressure on industrial firms must have also played an important role. 16 References Afsah, S., B. Laplante and D. Wheeler (1996), “Controlling Industrial Pollution: A New Paradigm”, Policy Research Working Paper #1672, Policy Research Department, Environment, Infrastructure and Agriculture Division, World Bank, Washington, DC. Chopra, K. and B. Goldar (2000), ‘Sustainable Development Framework for India: The Case of Water Resources’, Report Submitted to the United Nations University, Tokyo, as a part of the UN University Project on ‘Sustainable Development Framework for India’, October. Goldar, B. and R. Pandey (2001), “Water Pricing and Abatement of Industrial Water Pollution: A Study of Distilleries in India”, Environmental Economics and Policy Studies, 4(2). Kuik, O.J., M.V. Nadkarni, F.H. Oosterhuis, G.S. Sastry and A.E. Akkerman (1997), Pollution Control in the South and North: A comparative assessment of environment policy approaches in India and the Netherlands, Indo-Dutch Studies on Development Alternatives – 21, New Delhi: Sage Publications. Murty, M.N. (1999), “Role of Government in Environmental Management”, in M.N. Murty, A.J. James and Smita Misra (eds.), Economics of Industrial Pollution Abatement: Theory and Empirical Evidence from the Indian Experience, Delhi: Oxford University Press. Murty, M.N. and U.R. Prashad (1999), “Emissions Reduction and Influence of Local Communities in India”, in M.N. Murty, A.J. James and Smita Misra (eds.), Economics of Industrial Pollution Abatement: Theory and Empirical Evidence from the Indian Experience, Delhi: Oxford University Press. Pargal, S., H. Hettige, M. Singh, and D. Wheeler (1997), “Formal and Informal Regulation of Industrial Pollution: Comparative Evidence from Indonesia and the United States”, World Bank Economic Review, 11(5): 433-50. Pargal, S., M. Mani and M. Huq (1997), “Inspections and Emissions in India: Puzzling Survey Evidence on Industrial Water Pollution”, PRD Working Paper #1810, Development Research Group, World Bank, Washington, DC, August. Pargal, S. and D. Wheeler (1996). “Informal Regulation of Industrial Pollution in Developing Countries: Evidence From Indonesia”, Journal of Political Economy, 104(6): 1314-1327. Sankar, U. (1999), “Laws and Institutions Relating to Environmental Protection in India”, Occasional Paper no. 2, Madras School of Economics, Chennai. 17 Annex I: Status of Pollution Control in 17 Categories of Highly Polluting Industries, India, 1995 and 2000 State/ Union Number of No. of units not having adequate territory units facilities to comply with standards identified Andhra Pradesh Assam Bihar Goa Gujarat Haryana Himachal Pradesh Jammu and Kashmir Karnataka Kerala Madhya Pradesh Maharashtra Orissa Punjab Rajasthan Tamil Nadu Uttar Pradesh West Bengal Delhi Pondicherry Other states/UT Total Mar-95 Dec-2000 173 15 62 6 177 43 9 8 85 28 78 335 23 45 49 119 224 58 5 6 3 32 5 11 0 8 7 0 4 21 4 21 28 10 11 2 8 40 27 3 4 6 1 1 2 0 0 0 0 0 0 0 5 5 4 0 0 0 3 3 0 0 0 1551 252 24 Source: Central Pollution Control Board, Annual Report, 1994-95 and 2000-01 18 Annex II: Status of Defaulters under the Program of Industrial Pollution Control Along the Rivers and Lakes, India, 1997 and 2000 State/Union Territory Number of Closed defaulters subsequently in Aug. 97 Acquired requisite treatment/ disposal facilities Number of defaulters in Dec.2000 Andhra Pradesh Assam Bihar Goa Gujarat Haryana Himachal Pradesh Jammu and Kashmir Karnataka Kerala Madhya Pradesh Maharashtra Orissa Punjab Rajasthan Tamil Nadu Uttar Pradesh West Bengal Pondicherry 60 7 14 0 17 21 0 0 20 36 2 6 9 18 0 366 241 30 4 17 5 4 0 3 8 0 0 2 4 1 3 1 1 0 118 59 7 0 37 0 10 0 14 12 0 0 17 32 0 3 4 16 0 248 176 23 4 6 2 0 0 0 1 0 0 1 0 1 0 4 1 0 0 6 0 0 Total 851 233 596 22 Source: Central Pollution Control Board. 19