Analyzing the relationship between water pollution and economic activity for a more effective pollution control policy in Bali Province, Indonesia

16 An adequate water supply is essential for the and sustainable growth of the Balinese economy. In addition to mounting water demand, Bali’s water supply has been compromised by high levels of water pollution. Despite being paid great attention, Bali’s e arlier efforts to control water pollution yet to prove effective, mainly owing to their reliance on traditional methods and 20 regulations that focus on water pollution being linked to discrete sets of economic activity (e.g., 21 processing industries, livestock farming, and hotels). However, all economic sectors are 22 interconnected through supply chains; thus, water pollution is the combined result of all sectors’ 23 environmental performance. Therefore, determining the structural relationships between water 24 pollution and economic activity serves as an important basis for more effective forms of 25 pollution control for the Balinese economy. In this study, accordingly, we employed an 26 environmentally extended input – output model to establish the links between water pollution and 27 the production processes of the entire economy. Using biochemical oxygen demand (BOD) as a 28 proxy for water quality in our analysis, we estimated that 246,868 tons of BOD were produced 29 from Bali ’ s economic activity in 2007. Further, we identified the chief BOD-emitting sectors and 30 found that intermediate demand and household demand were the major causes of BOD discharge 31 in the economy. Utilizing supply chain relationships, we also accounted for the indirect role of 32 each sector in total BOD emissions. Moreover, we categorized the sectors into four groups based 33 on their direct and indirect BOD emission characteristics and offered appropriate policy 34 measures for each group. Managing demand (i.e., lowering household consumption and exports) 35 and shifting input suppliers (i.e., from polluters to non-polluters) are effective measures to 36 control pollution for Categories I and II, respectively; clean production and abatement is advised for Category III; and a hybrid approach (i.e., demand management and abatement technology) is 38 recommended for Category IV. 39


7
In 2018, the Balinese economy reportedly grew by 6.35%. The contribution of various 129 sectors to the GRDP for 2018 is demonstrated in Fig. 2 [6]. The accommodation and food service 130 sector's activities under tourism provide the dominant share of GRDP, followed by the 131 agricultural, forestry, and fishery sector. The contribution of the agricultural, forestry, and 132 fishery sector has declined in recent years. The Balinese economy has shifted from a primary to a 133 tertiary economy with the recent growth of the tourism sector [6]. Other sectors, such as the 134 transportation and construction industries, have a dominant share of the GRDP, accounting for 135 9.5% and 9.4%, respectively. These sectors have reported increasing growth in the GRDP owing 136 to the expansion of the tourism sector. In total, 382 companies representing large (100 or more 137 employees) and medium (20-99 employees) industries were in operation in Bali in 2017; of 138 these, the food and beverage industry had the greatest share of the GRDP [6]. 139

Conventional I-O table 142
In an economy's production process, each sector requires inputs from its own and other 143 sectors to produce goods and services. The concept of describing the inter-linkages between 144 sectors in an economic system was introduced by the economist Wassily Leontief. His I-O 145 model is conventionally used to describe the interconnections between sectors [30]. Table 1  146 presents an example of an economy's flow of goods and services, visualizing the 147 interdependence between the sectors [31]. The rows represent the proportion of output that each 148 sector sells (seller) to other sectors (purchaser); the columns represent the proportion of products 149 and services required (purchased) by each sector from other sectors as inputs to meet total output. 8 consumed by households, the government, and exports (final demand) as well as the amount of 152 products and services that sectors import (imports) and the compensation paid to labor (value 153 added). I-O tables are formulated based on the data for a particular economic area, nation, or 154 region [32]. 155 describes the inter-sectoral relationships in an economic system [33]. It depicts the fact that total 158 output is equal to the amount used internally used by the system as intermediate consumption 159 plus the amount consumed by the final customers (Leontief as cited in Nguyen [21]). 160 where is intermediate consumption [21], is the final demand for goods by each sector, and 161 is the technical coefficients of production, which are described as the amount of products that 162 the j th sector purchases from the i th sector to produce one unit of (j th ) output. 163

=
(2) Eq. (1) can be rewritten as follows after incorporating : 164 Converting the equation into matrix notation for the entire economy yields the following: 165 where I denotes an identity matrix and (I − A) - The release of water pollutants is usually expected to be linearly proportional to the size of 177 sectoral outputs [21,23]. We assume PI as pollution intensity and define it as the amount of 178 water pollutants released to produce one unit of output (in monetary terms) for a sector. Its 179 elements, , denote pollution intensity related to the y th water quality parameter for sector i in 180 a particular year. The y × n matrix, PI, is the pollution intensity matrix. The pollution load (PL) 181 can then be computed using Eq. (7). 182 The PL for the k th parameter for sector i, , is expressed as follows: 183 Substituting the value of xi from Eq. (6) into Eq. (8) yields the following: 184 The PL of sector i is the amount released to satisfy all production in this sector (xi), including 185 both the intermediate (∑ ) and final demand ( ). 186

Water pollutant emissions 187
Each economic sector acts as both a supplier and receiver of inputs in the economy's 188 production process. Based on their roles, the emissions of water pollutants from sectoral 189 activities can be distinguished into two categories: direct (i.e., a source of emissions) and indirect 190 (i.e., a cause of emissions An estimated total of 246,868 tons of BOD was released in the economy's production 245 process in 2007. Each sectoral direct BOD load is shown in Fig. 3. Three sectors-livestock and 246 poultry (Sector 2); agriculture, forestry, and fishery (Sector 1); and food, beverage, coffee, and 247 tobacco (Sector 4)-accounted for 99.5% of the total BOD emissions; of these, Sector 2 248 produced 96% of the total BOD emissions. The BOD amount varied among the sectors because 249 of differences in BOD emission intensity and total output. The results are consistent with earlier 250 findings that indicated livestock plays a major role in freshwater pollution in many parts of the 251 world [38, 44, 46]. The increasing population and the prevalence of many head of cattle-the 252 reported increase in cattle and pig numbers was 3.62% and 11.73%, respectively, from 2017 to 13 2018 [6]-positioned this sector at the top in terms of water pollution. Conversely, the 254 agricultural land under Sector 1 has recently been declining; however, the heavy use of fertilizers 255 in intensive agricultural farming has kept the sector a leading cause of freshwater pollution. 256 Followed by these two sectors, Sector 4 (food, beverage, coffee, and tobacco) emitted BOD to a 257 greater extent, with a value of 1,894 tons, which notably serves to meet household consumption. 258 The remaining 13 sectors discharged 1,049 tons of BOD, which accounted for only 0.5% of the 259 total BOD emissions. 260 BOD emitters (Sectors 1-3 and 12). This indicates that these sectors discharged major BOD in 270 producing their products or services for other economic sectors. However, exports were the 271 major cause of BOD emissions (more than 60% of BOD) in the sectors (Sectors 5-9 and 13) that 272 mainly comprise manufacturing industries. This is likely attributable to Bali's adoption of an 273 open economy as exports play an important role in the regional economy [6]. Household demand 274 was the primary cause of BOD (nearly 50%) for Sector 4 (food, beverage, coffee, and tobacco) 14 Gross stock and changes had negative BOD values for Sectors 4 and 5, indicating that this 277 portion (BOD) was not generated for that year but fulfilled by the stock. 278 Apart from the direct BOD discharge, we determined indirect BOD emissions by sector (Fig.  281 5). Sectors 2, 13, 1, and 4 produced as much BOD from other sectors or from themselves to 282 satisfy their inputs. Sector 2 was the chief indirect BOD emitter, accounting for 65% of total 283 BOD emissions; this is likely due to its reliance on its own sector, which is the top direct BOD 284 emitter, for inputs (e.g., baby chicks, calves, fingerlings, and animal feeds). Sector 13 (hotels and 285 restaurants) is responsible for emitting 30% of total BOD owing to its close connection with 286 other sectors-most likely the livestock and agricultural sectors, both major BOD emitters-to 287 operate their business. Importantly, other service and trade sectors show a noticeable 288 contribution to indirect BOD emissions as opposed to their role in direct BOD emissions. This is 289 particularly essential for pollution control planning because the direct emissions of these sectors 290 are often overlooked. 291

Disaggregation of sectoral BOD emissions 294
An analysis of the sectoral roles in indirect water pollutant emissions has added a new 295 perspective to the conventional approach of direct sectoral water pollution. However, although 296 these are good indicators, these aspects (i.e., direct and indirect pollution) remain unable to fully 297 depict the behavior of sectoral pollution in the economy. To enrich our analysis, we demonstrate 298 the flow of BOD throughout the entire economic sector (Table 3) and its various pollution 299 components (Table 4). In Table 3, the rows of the matrix indicate the BOD amount that sector i 300 produces to fulfill sector j's demand. The row sums represent the total BOD directly emitted by 301 sector i in producing products or services to fulfill all forms of the economy's demand (i.e., DiE). 302 The columns of matrix j indicate the purchases made by sector j from sector i during the 303 production process, and the column sums represent the total BOD indirectly emitted by sector j 304 from other sectors (i) in obtaining its input requirements (i.e., IDiE). The absence of row data for 305 Sectors 10, 14, and 15 indicates that these sectors do not directly emit BOD (row), while the 306 column values for the same sectors show their indirect role in producing BOD in the economy. 307 Subsequently, we disaggregate sectoral BOD loads into five components: (i) a sector's own 308 pollution ( ) (the amount of y pollutant emitted by sector i to produce its own input); (ii) 309 semi-own pollution ( ) (the amount of y pollutant generated by sector i to produce the 310 inputs for other sectors, which is required to produce inputs that sector i purchases to fulfill final  (Table 4). 316 The values vary among the sectors according to their different emission properties. For 317 instance, a high value of for Sectors 1 and 2, pollution seller sectors, indicates that these 318 sectors produce considerable BOD in satisfying intermediate demand. Similarly, Sector 2 is also 319 responsible for producing substantial BOD (24.7% of the total BOD) in fulfilling its own inputs, 320 highlighting the internal dependence of the sector (for their own inputs, i.e., ). Other than 321 Sectors 1, 2, and 3, a majority of sectors have a typically high range of , characterizing these 322 sectors as pollution inducers that cause other sectors to emit BOD to fulfill their input needs. 323 Among these, Sector 13 (hotels and restaurants) has an exceptionally high value, indicating a 324 close linkage between this sector and others and its responsibility for producing a significant 325 amount of the economy's water pollutants. Sector 4 (food, beverage, coffee, and tobacco) is a 326 major pollution inducer because of its processing of raw primary products (i.e., agricultural and 327 livestock products) into edible food items. 328 Table 3 Matrix of BOD (kg) flow in the economy. 329 Table 4 Disaggregated sectoral water pollution components with BOD loads (in tons). 330

Grouping of sectors 331
To distinguish the sectoral BOD emission behavior and appropriately identify the 332 management plan, we grouped sectors by plotting the percentage of BOD-true forward 333 pollution ( )/direct emissions (DiE) and true backward pollution ( )/ indirect emissions 334 (IDiE) (Fig. 6). DiE is defined as the amount of water pollutants (herein BOD) that are directly 335 discharged by a sector while producing the products required to satisfy all forms of demand (i.e., 336 intermediate demand and final demand) [22]. By contrast, IDiE is the amount of water pollutants 337 (herein BOD) that are discharged by a sector and other sectors to produce the inputs that it 338 requires. Generally, high /DiE values indicate that the sectors are liable to produce more 339 water pollutants for other sectors, so the proportion of water pollutants produced for their own 340 sectoral inputs is lower. Conversely, sectors with high /IDiE produce more water pollutants 341 from other sectors. 342 Figure 6 shows that most sectors fall under Categories I and II. In contrast, there is only one 343 sector in Category III, and none of the sectors belong to Category IV. Category I has less than 344 50% of both /DiE and /IDiE, indicating that the sectors depend heavily on their own 345 sectoral input and produce more than 50% of the total direct and indirect sectoral BOD discharge 346 in fulfilling their own sector's input demands. As major sources of pollution for their own input 347 requirements, these sectors are characterized as self-polluting sectors. The pollution under this 348 category could be better addressed by the product's final demand. In this regard, policy should 349 focus on measures to lower household consumption and reduce the exports of these sectors. 350 This possesses a high (over 50%) of /DiE but has a /IDiE value of less than 50%, 362 indicating that the sector emits considerable BOD to fulfill intermediate demand and that it does 363 not cause other sectors to produce BOD for its input requirements. This category of sectoral 364 pollution control will be better dealt with by implementing clean production technologies and 365 wastewater treatment practices. No sectors can be found under Category IV, which is 366 characterized by indirectly producing a huge amount of BOD from the other sectors and also 367 directly producing a significant amount of BOD to supply other sector demand. Although we did 368 not find any sectors in this group, two sectors (i.e., Sectors 10 and 12) were extremely close to 369 this category. 370

Policy implications and perspectives for pollution control 371
The analysis herein of sectoral water pollution behavior in the production process of the 372 economy offers vital policy directives for controlling water pollution. In this study, livestock and 373 poultry sectors were categorized as the chief BOD emitter; therefore, major attention should be 374 paid to this sector. The handling and treatment of livestock waste is not common despite such 375 waste containing valuable ingredients that can be processed for manure and biogas production. and wastewater treatment practices should focus on the safe discharge of the wastewater produced, especially from manufacturing industries such as Sectors 4 (food, beverage, coffee, 391 and tobacco) and 8 (fuels and chemical industry, rubber, plastics). 392 Decreasing the total output of each sector by cutting its demand is the best method to control 393 water pollution. The total output consists of the intermediate and final demands of the economy, 394 while the final demand is further directed by exports, households, and gross stock. These 395 components of demand should be carefully examined and considered in terms of pollution 396 reduction. For instance, exports are responsible for producing a significant amount of BOD from 397 major manufacturing industries such as Sectors 5 (textile, apparel, and leather goods), 6 (timber 398 industry and wood products), 7 (paper industry, paper, and cardboard goods), 8 (fuels and 399 chemical industry, rubber, plastics), and 9 (other processing industries). Curtailing the exports of 400 these sectors based on their pollution loads to minimize the water pollution load could be an 401 alternative. However, such policies should be examined from an economic and social perspective. 402 Household consumption is the primary reason for BOD emissions by the top BOD emitters-403 Sectors 4 (food, beverage, coffee, and tobacco) and 2 (livestock and poultry). Changes in 404 household consumption, including changes in dietary habits (e.g., shifting consumable food 405 items toward environmentally friendly products), could be the best approach to minimize the PLs 406 from these sectors. Although such changes may take time, starting initiatives in this area would 407 significantly impact pollution control in the long term. Considering indirect roles, Sector 13 408 (hotels and restaurants) demonstrates significant impact on water pollution by indirectly inducing 409 BOD. The indirect BOD discharge of Sector 13 is around 190 times higher than the sector's 410 direct BOD emissions. This sends an important message-aside from focusing on pollution 411 control practices within the premises of hotels and restaurants, policy should also seek and 412 prioritize supply-side pollution (from associated sectors). This provides new insights, such as 413 offsetting the high investment cost for Sector 2 to adopt a wastewater treatment plant, by 414 mobilizing the environmental fees/revenue collected from another sector, for example, the 415 tourism sector. Alternatively, the possibility of decreasing the livestock and poultry population 416 and increasing the imports of such products could be sought. Again, socioeconomic aspects 417 should be evaluated for such decision-making 418 The classification of the sectors (Fig. 6) serves as a useful tool in planning appropriate 419 pollution control policies. Managing demand by lowering household consumption, changing 420 food habits (toward environmentally friendly food), and curtailing exports are effective methods 421 to inhibit the water pollution of the sectors in Category I. Sectors under Category II are subject to 422 pollution control by seeking alternative input suppliers (from high-to low-polluting suppliers) 423 aside from the measures suggested for Category I. Sectors under Category III will be better 424 managed by implementing clean production technology and wastewater treatment practices. 425

Conclusions 426
This study is the first to analyze the relationship between economic activity and the potential 427 for water pollution in Bali, Indonesia. Going beyond conventional methods, potential water-428 polluting sectors were identified in this study based on their direct and indirect roles in BOD 429 emissions. This study recognizes significant BOD emission drivers, guiding policymakers and 430 practitioners to target initiatives that reduce Bali's water pollution. Certain sectors-namely, 431 Sector 2 (livestock and poultry), Sector 1 (agriculture, forestry, and fishery), and Sector 4 (food 432 and beverage)-accounted for 99.5% of the direct BOD discharge, among which Sector 2 433 accounted for 96% of the total BOD discharged. For direct BOD emissions, intermediate demand 434 was a major driver in Sectors 1-3 and Sector 12. Similarly, exports were the cause of more than demand dominated the major portion of BOD emissions in Sectors 4 (food, beverage, coffee, and 437 tobacco) and 10 (electricity, and drinking water). 438 In particular, we determined each sector's indirect BOD emissions. The livestock and poultry 439 sector, the top BOD emitter, produced more than 50% of its BOD for its own sector demand, 440 which was also recognized as the most self-polluting sector. Hotels and restaurants heavily rely 441 on other sectors and were responsible for indirectly emitting BOD from different sectors. Sectors 442 such as the trades, transportation, and service sectors, whose direct BOD emissions are limited, 443 still significantly contributed to indirect BOD emissions. 444 Moreover, we grouped the sectors into four categories. Those in Category I emit considerable 445 water pollutants for their own input requirements; these sectors are called self-polluters and are 446 better addressed by demand management and employing clean production technologies. Sectors 447 in Category II directly emit a significant portion of water pollutants for their own sector's input 448 demand but heavily rely upon the total indirect water pollutants from other sectors. This 449 category's pollution control will be better managed by shifting from high-to low-polluting      Table 1 Schematic representation of a conventional I-O table.  623  624  Sector  Intermediate sector  Final  demand Total output   --------------15  ----------------