Structural analysis of the interrelationship between economic activities and water pollution in Vietnam in the period of 2000–2011

Abstract

Rapid economic growth and poor waste management have left Vietnam with severe water pollution problems. It is thus important to develop a model to evaluate the relationship between economic activities and water pollution to identify water pollution mitigation strategies within the context of economic development. Recent works have demonstrated the effectiveness of the input–output model in analyzing the interplay between the economy and the environment. To comprehensively understand this relationship, the behavior and trend of water pollution during a specified period should be investigated on. The interaction of different economic sectors and its impacts on water pollution should also be analyzed. For the Vietnamese economy, such aspects have not been fully addressed in previous studies. This work thus examines the state of water pollution in Vietnam as indicated by water quality parameters, total suspended solids and biological oxygen demand, with particular attention to the individual contribution of various economic sectors. The period between the years 2000 and 2011 is taken into account in this work. Environmentally extended input–output analysis coupled with vertical integrated coefficient method is used to analyze the interindustry linkages of sectors and to classify sector role as either key sector or pollution puller or pusher. The pollution trend reveals the tremendous increase in total suspended solids from 345,000 tonnes in 2000 to 1,199,000 tonnes in 2011, while the total biological oxygen demand increased from 43,400 tonnes in 2000 to 123,000 tonnes in 2011. Results show that the basic metals industry was the major contributor of total suspended solids, while the food, beverage and tobacco and agriculture, fishery and forestry sectors contributed most to the biological oxygen demand. The results of sectoral linkage evaluation highlight that food, beverage and tobacco and agriculture, fishery and forestry sectors were key sectors for both water quality parameters. These results provide environmental managers and policy maker insights on how to prioritize economic sectors to achieve emission reduction targets.

Introduction

The comprehensive economic reforms from 1986, which have implemented open door policies and transformed Vietnam from a centrally planned economy into a market economy, have contributed significantly to the country’s economic development (Que and Thanh 2001). Since the year 2000, the Vietnamese economy has prospered with annual economic growth rate up to above 5% and has been forecasted to be the 20th largest economy in the world by 2050 (PricewaterhouseCoopers 2017). However, this rapid economic development has resulted in the degradation of the ecosystem which also affects the living conditions of humans (The World Bank et al. 2002). Water environmental pollution is considered as the most important concern for Vietnam (Global Environmental Forum 2002). The main reason for this is the huge amount of wastewater coming from household activities and the operation of factories which is discharged into water bodies without treatment. The rapid industrialization and urbanization accompanied with poor wastewater management resulted in the drastic degradation of water quality (WEPA 2011; WSSA 2012). Furthermore, rapid population and economic growth resulted in the increase in demand for freshwater use. Tran (2010) reported that the rate of water consumption in industries, households and agriculture is increasing annually by 7, 9 and 3.4%, respectively. Meanwhile, the available freshwater resources are insufficient to meet these increasing demands for water use (WEPA 2011; Hoa et al. 2013). Only 70% of the population is likely to have access to safe water and sanitation (Tran 2010). Traditionally, the rural population uses groundwater pumped from private tube wells. However, there is recent evidence of arsenic contamination in the groundwater in several regions in Vietnam. Arsenic contamination in groundwater is caused by several factors. This includes contamination from the presence of arsenic in geological structures, the use of plant protection drugs containing arsenic and the release of arsenic from chemical factories to name a few (WSSA 2011). This arsenic contamination may cause serious health issues such as cancer or skin problems to people in regions of exposure (Jessen 2009). With such serious issues on water pollution and freshwater resource scarcity, there is thus a need to develop effective strategies to avoid further pollution of freshwater regions and to improve water quality.

To address these problems, the Vietnamese government has employed various strategies, ranging from regulatory to economic tools, but these actions have not seemingly brought positive effect. The reason is that the interrelationship between the economy and the environment has not been well understood (Le et al. 2013). Understanding this relationship is vital in identifying the key sources of pollution and in characterizing wastewater releases. This helps classify sectors as either pollution sellers, pollution purchasers or both. The environmental performance of an economic sector is linked to all other sectors by virtue of its supply chain (Leontief and Ford 1972). Furthermore, studies on the effect of production activities on water quality and water pollution data in Vietnam are limited. Meanwhile, researches on these aspects are carried carefully in developed countries and many developing countries. For example, Wu et al. (2017) proposed a method for quantifying a water pollution ecological compensation standard for China by using emergy theory. An analysis of virtual water pollution transfer embodied in economic activities was employed for Beijing’s economy (Li et al. 2017), while Cazcarro et al. (2016) evaluated the relationship between structural production and environmental impact in the consideration of trade among countries. Sanchez-Choliz and Duarte (2005) analyzed the relationship between processes and water pollution for the Spanish economy to identify the roles of economic sector blocks as either generator or consumer of pollutants. Recently, several researchers made numerous efforts in proposing environmental friendly water pollution treatment methods. For example, Kamyab et al. (2015) looked at the efficiency of microalgae in the removal of pollutants from palm oil mill effluents. The wastewater was then investigated for its potential as biofuel substrate (Kamyab et al. 2016) and resource for fertilizer production (Kamyab et al. 2017). Microalgae can help significantly reduce water pollutants in wastewater as it utilizes the pollutants as nutrients for growth. Rezania et al. (2016) emphasized that the use of phytoremediation technology is likely to effectively treat heavy metal-polluted sites with low treatment cost. In Vietnam, there are only a few studies dealing with the interrelationship between economic activities and pollution flows so far. One of the most important references is the ICEM (2007) published by World Bank, which evaluated the national pollution flows by quantifying the different pollutants found in air, land and water media from the manufacturing industries. In addition, Hung et al. (2008) analyzed industrial pollution by focusing on the effect of trade liberalization on the pollution of sectors and subsectors. However, this only considered pollution from manufacturing sectors and did not consider pollution flows from other sectors such as agriculture, mining, services and consumer use. Furthermore, an analysis of the interrelationship among economic sectors and their individual pollution characteristics has not been discussed. A better understanding of these factors is useful in providing and selecting more environmentally friendly production supply chains such as the selection of less pollutive raw materials or inputs and the use of cleaner technologies.

According to Miller and Blair (2009), the best way to describe the interrelationship among economic sectors is to apply input–output (IO) analysis. IO analysis is an analytical framework first proposed by Professor Wassily Leontief in the 1930s and provides a quantitative description of how goods and services of an economy are related. In recognition of his work, he achieved the Nobel Prize in Economics in 1973 (Leontief 1936, 1970). IO analysis shows the interconnectedness of the economy’s various sectors by considering the product of each sector both as goods for final demand (i.e., household and government consumption) and as raw materials in the production of commodities in the same sector or in other sectors. The framework captures both direct and indirect effects which may result from an economic shock. The IO model can be extended to capture other characteristics of the economic system. Hoa et al. (2016), for example, utilized the inoperability input–output model (IIM) developed by Haimes and Jiang (2001) and the vulnerability index proposed by Yu et al. (2014) to develop a multicriteria model for disaster vulnerability due to implications of energy policy. The extension of IO analysis to account for environmental impacts is known as environmentally extended input–output (EEIO) analysis, and it has been employed in the accounting of carbon emissions in supply chains for sustainability assessment (Suh 2009; Murray and Wood 2010) and in optimizing multiregional bioethanol supply chains in the presence of multiple objectives (Tan et al. 2008, 2009). Furthermore, IO analysis can be used to study the linkages between economic sectors and to determine their roles (Rasmussen 1956; Yan and Ames 1965). These authors utilized the Leontief inverse or Gosh inverse (Gosh 1958) to categorize sectors as either forward linkage, backward linkage or key sectors (those that act both as forward and as backward linkage sectors) or nonsignificant sectors in relation to purely economic aspects. Then, Sanchez-Choliz and Duarte (2003, 2005) expanded the scope of these models to include resources and emissions through the vertical integration coefficient (VIC) methodology. This facilitated the identification of sector roles in the economy with respect to environmental impacts. The integration of EEIO analysis with VIC may thus create a powerful framework for analyzing the interrelationship between sectors from the point of view of pollution.

In Vietnam, the application of EEIO for environmental–economic impact assessment was performed by Trinh and Nguyen (2013) and Le et al. (2013). Trinh and Nguyen (2013) employed EEIO to develop the framework of an interregional IO model for the years 2000 and 2007. Meanwhile, Le et al. (2013) used the integration of EEIO and VIC for the year 2000 to identify and quantify the sources and causes of water pollution, indirect emissions and direct emissions. Although Le et al. (2013) were successful in quantifying the total pollution load of main sectors, the direct and indirect pollution, they failed to show which sectors are pollution sellers and which are pollution purchasers. Furthermore, both Trinh and Nguyen (2013) and Le et al. (2013) failed to provide a time series analysis of the total pollution load trend induced by economic sectors.

To supplement this research gap, the main objective of this work focuses on looking at the changes in water quality from the year 2000 to 2011. The characterization of economic sectors in relation to water pollution in 2011 is then employed using EEIO coupled with VIC to identify the role of each economic sector as either consumer/purchaser or/and a producer/seller of water pollution. Finally, the combination of water quality trend and characterization of sectoral pollution will give a comprehensive picture on how economic sectors contribute to the main water quality parameters. The rest of paper is organized as follows. “Problem statement” section formalizes the problem that this research intends to address, and the methods are then discussed in “Methods” section. “Data collection” section mentions the database used in this work. Results and discussion are then mentioned in this section. Finally, conclusions and recommendations for future work are discussed.

Problem statement

In this work, the evaluation of how the different economic sectors in an economy contribute toward water pollution relies on results from both the water quality trend and the characteristics of sectors in relation to water pollution. The water quality trend focuses on the annual total pollution load contributed by the different sectors during the observed period. In contrast, the characteristics of sectors are obtained by taking into account water pollution flows and sectoral linkages during a specific year. The formal problem statement can thus be stated as follows:

• Given an economy with n number of economic sectors, an n × n technical coefficient matrix, A, can be obtained for a specific year which describes the interactions between the economic sectors. There is a nx1 vector, x, which represents the gross domestic output (GDO) of sectors. This is collected from the statistical databases according to the identified time series range.

• There are environmental flows in the production of goods and services from each sector, which are quantified using k number of quality indicators. In this context, only two indicators (k = 2) of water quality are considered, biological oxygen demand (BOD) level and total suspended solid (TSS). Thus, a 2 × n vector of pollution intensity, PI, is estimated for each year in the observed period.

The problem then is to evaluate how each economic sector contributes to the yearly change in water quality and to identify whether an economic sector is a purchaser (has inputs from high polluting sectors) or seller (is an input to other economic sectors) of water pollution or both. Suggestions for policy makers and manufacturers on how to prioritize sectors to achieve emission reductions are given.

Methods

The framework developed for evaluating environmental quality parameters is shown in Fig. 1:

Fig. 1

Framework of water quality parameter assessment

Generic input–output model The IO framework has been traditionally applied to study the interconnectedness between various economic sectors in an economic system. Its basic formulation includes a system of linear equations in which total output is equal to the amount used internally by the system, as intermediate consumption, plus the amount consumed by final customers (Leontief 1985). For an economy with n sectors, vector x denotes the GDO vector, y is the final demand vector, and Z is the n × n input–output transaction matrix which can be obtained from statistical databases. In matrix notion, the IO model can be stated as follows (Miller and Blair 2009):

$$\varvec{x} = \varvec{EZ} + \varvec{y}$$

(1)

where E = [1, 1, …, 1]. Let A represent the n × n technical coefficient matrix with elements, a _ij (a _ij  = Z _ij /x _j ), indicating the amount of product from the ith sector needed to produce one unit of output from the jth sector. Based on this, Eq. (1) can be rewritten as follows:

$$\varvec{x} = \left( {\varvec{I} - \varvec{A}} \right)^{{ - {\mathbf{1}}}} \varvec{y}$$

(2)

where I denotes an identity matrix and (I − A)⁻¹ is the Leontief inverse matrix. Let α _ij represent the elements of the Leontief matrix. Then, the GDO of sector i, x _i , can be expressed as:

$$x_{i} = \mathop \sum \limits_{j} \alpha_{ij} y_{j}$$

(3)

The detailed discussion of Eqs. (2) and (3) can be found in Miller and Blair (2009).

EEIO model, total pollution load and embedded pollution load

The extension of the IO model known as an environmentally extended input–output (EEIO) model is considered a useful technique for quantifying resource use and emissions resulting from the production activities of each sector (Butnar and Llop 2007). The amount of resource use or emissions from each sector is assumed to be linearly proportional to the size or output of that sector. Let PI denote pollution intensity which is the amount of pollution load emitted to produce one unit of monetary output for each sector as indicated by water quality parameters. Its elements, PI ^k _i , represent pollution intensity corresponding to the kth water quality parameter type for sector i in a specific year. The k × n matrix PI is referred to as the pollution intensity matrix. The total pollution load (PL) can then be calculated using Eq. (4):

$${\mathbf{PL}} = {\mathbf{PI}} \, {\mathbf{x}}$$

(4)

And the total pollution load for the kth parameter for sector i, PL ^k _i is expressed as:

$${\text{PL}}_{i}^{k} = {\text{PI}}_{i}^{k} x_{i}$$

(5)

Substituting x _i from Eq. (3) into Eq. (5), an equation for calculating the total pollution load of sector i is given as follows:

$${\text{PL}}_{i}^{k} = {\text{PI}}_{i}^{k} \mathop \sum \limits_{j} \alpha_{ij} y_{j}$$

(6)

The total pollution load of sector i provides the total annual pollution that is released to satisfy the whole production of this sector $(x_{i} )$ which includes intermediate (∑ _j a _ij x _j ) and final demands (y _i ). The trend of the water quality parameters shows how the total pollution load varies in each sector during the observed period. Based on the water quality trend, the policy makers can identify which pollution mitigation methods should be implemented by focusing on sectors which contribute greatly toward increasing the BOD or TSS. Alternatively, strategies for managing sectors, which have significantly increasing potential for pollution generation, can be explored.

Besides sectoral total pollution load, there is embedded pollution load which is defined as the pollution load associated with the production of the final demand of a given sector. This is either emitted by the sector itself or attributed to the sector due to the purchase of inputs from other sectors (Sanchez-Choliz and Duarte 2003). It is different from the total pollution load of a sector which is associated with the pollution generated to obtain the GDO of that sector. Let DI ^k _i denote the embedded pollution load for the kth parameter resulting from the production of demand $y_{i}$ of sector i, which is given by:

$${\text{DI}}_{i}^{k} = \mathop \sum \limits_{j} {\text{PI}}_{j}^{k} \alpha_{ji} y_{i}$$

(7)

The detailed discussion of this formulation was given in Sanchez-Choliz and Duarte (2003). Equation (1) shows that when the final demand of sector i increases, there is a need to increase the production of that sector itself. In addition, there is also a need to increase the productivity of other sectors which are linked to it either directly (i.e., as a supplier) or indirectly (i.e., across the supply chain). As a result, the environmental burden as expressed in Eqs. 4 and 7 also increases. The evaluation of water pollution from economic sectors is clearly visualized when the total pollution load and embedded pollution load are decomposed into different components corresponding to product flows. The next section discusses this in more detail.

Vertical integration coefficient (VIC)

The VIC is employed by disaggregating each embedded pollution load from each sector into four components of pollution including true backward linkage (TB ^k _i ), semi-own pollution (SE ^k _i ), own pollution (OW ^k _i ) and final demand pollution (FI ^k _i ). The total pollution load of a sector is also decomposed into four components and includes true forward linkage (TF ^k _i ), SE ^k _i , OW ^k _i and FI ^k _i . These components are related to the flow of goods which are produced to provide for the needs of manufacturing processes required for economic activities (Fig. 2). Table 1 shows the description of pollution components. The detailed discussion of the VIC methodology as introduced by Sanchez-Choliz and Duarte (2003) is given in “Appendix”.

Fig. 2

Pollution components in response to flows of product

Table 1 Description of pollution components

After calculating the pollution components, the component indices are given to assess sectoral characterization in relation to pollution flows. Table 2 shows the relative indices for VIC evaluation together with their description, while details for index computation are shown in Appendix. Based on these indices, sectors can be classified as either key sectors, forward linkage, backward linkage or nonsignificant sectors by using the guide in Table 3. It can be said that the values are low when they are close to 0 and high when they are close to 1. Backward linkage sectors are sectors with high pollution amount by virtue of high inputs (purchases) from other sectors; thus, they are called pullers. In contrast, if a sector manufactures goods, then the pollution generated by this sector is virtually sold to other sectors without it finally returning to this sector, in which case this sector is considered as a forward linkage sector and called a pusher (Sanchez-Choliz and Duarte 2003, 2005). Key sectors are those that act both as backward and as forward linkage sectors. According to Shmelev (2010), key sectors have high propensity to cause pollution across the economy in comparison with other sector types. The term “self-polluter by demand” means that pollution is mainly emitted from the production of final demand. In contrast, sectors where most of the pollution is due to own pollution or semi-own pollution are considered as self-polluter by inputs.

Table 2 Description of indices for pollution components

Table 3 Sectoral characterization based on indices for pollution components

Data collection

Water quality parameters

In terms of harmful effects, there are many substances such as pesticides and fertilizers from agricultural activities, heavy metals from mining activities and chemicals from industries, which are considered as major risks to human health and the ecosystem. However, data on these problems are limited and the monitoring of hazardous substances has not been systematically conducted. Thus, recent researches have focused on main water quality parameters in Vietnam with available data collected from government databases or obtained from secondary data. Trinh and Nguyen (2013) and Nguyen (2015) indicated that total suspended solids (TSS), biochemical oxygen demand (BOD), chemical oxygen demand (COD) and ammonia–nitrogen (NH4–N) are important parameters for measuring water quality in Vietnam and are used as indicators of water pollution load. These parameters are used to determine the level of suspended solids, organic substances, nutrition and lubricant. The decomposition processes of these substances may consume dissolved oxygen which is needed for the growth of species and organisms inside bodies of water. As a consequence, when the amount of these substances inside water bodies increases, the amount of dissolved oxygen decreases. This may lead to the death of fish and other aquatic organisms. In other words, at high concentrations, these parameters indicate low water quality. However, data for these indicators for Vietnamese economic sectors are not available from government statistics. Furthermore, global data only allow for the calculation of TSS and BOD load. Thus, these two parameters are chosen for analysis in this paper. The method presented here can easily be augmented once more information becomes available. Details for data collection and data validation are mentioned in “Pollution intensity of water pollution load (PI) and validation process” section.

Gross domestic output (GDO) and IO table

To calculate the annual total pollution load of sectors in the period 2000–2011 using Eq. (4), the GDO of different sectors should be known. These data can be obtained from existing IO tables in the years 2000, 2007 and 2011 (GSO 2003, 2009, 2013). For other years, the GDO of sectors can be achieved from the General Statistics Office (GSO) Web site (GSO 2006, 2011). To analyze the interrelationship between environmental pollution and economic activities, the integration of EEIO and VIC approaches is applied. However, since these methods can be applied for specific years with the condition of IO table availability, the newest IO table of Vietnam in 2011 is employed for this purpose. The 2011 IO table contains 138 columns and rows with 6 final demand columns and 4 value-added rows. The framework of the 2011 IO table is shown in Table 4, while the complete IO table is available from the GSO Web site (GSO 2013).

Table 4 Framework of the Vietnamese IO table for 2011

The Vietnamese IO table was compiled relying on the Provisional Central Product Classification System, while the manufacturing industries in industrial pollution projection system (IPPS) (Hettige et al. 1995) were classified according to International Standard Industrial Classification version 2 (ISIC 2). The manufacturing sectors from the original IO table are thus reallocated to be consistent with IPPS data. The detailed procedure can be found in Le et al. (2013). After reallocating, the IO table with 138 sectors is aggregated and classified into 18 main sectors, which is more convenient for environmental pollution computation. This classification is compatible with the classifications found in the IO table from GSO (2013) and the classification according to the International Standard Industrial Classification version 3 (ISIC 3). The description of the 18 sectors and sectoral GDO in 2000 and 2011 is shown in Table 5. In this table, the proportion of sectoral GDO, which is calculated using the average GDO of years 2000 and 2011, is also mentioned. Based on Table 5, it is shown that sectors which contribute the highest amount of GDO to the economy include Sectors 1 (agriculture, fishery and forestry) (13.4%), Sector 10 (manufacturing of fabricated metal products, machinery and equipment) (12.3%) and Sector 3 (food, beverage and tobacco) (11.6%). Sectors which contribute low GDO include Sector 5 (manufacture of wood and wood products) (0.7%), Sector 6 (manufacture of paper and paper products, printing and publishing) (1.3%) and Sector 17 (government services) (1.9%). However, the total pollution load of sectors may be different from this ranking because they do not only depend on GDO, but also rely on pollution intensity (PI). Sectors with less GDO may have very high total pollution load if they have high PI values. The next part discusses the PI of the different economic sectors.

Table 5 Nomenclature of 18 economic sectors and their GDO for the years 2000 and 2011

Pollution intensity of water pollution load (PI) and validation process

To compute the total water pollution load contributed by each sector, there is a need to determine the pollution intensity (PI), which is the amount of pollution load generated in the production of one monetary unit output of each sector (kg/million VND in a specific year). The PI data of economic sectors in terms of water quality parameters in Vietnam are not available from government statistics; thus, secondary data are used to estimate the PI. Trinh and Nguyen (2013) estimated the PI of water quality indicators for the years 2000 and 2007 based on data published by government, survey results in enterprise, information on wastewater calculations employed by the National Environmental Agency and study results obtained by the Institute of Tropical Technology and Environmental Protection. Trinh and Nguyen (2013) used the IO tables of 2000 and 2007 which were disaggregated into 12 economic sectors. This included nine nonmanufacturing sectors (Sectors 1, 2, 12, 13, 14, 15, 16, 17 and 18, as listed in Table 5) and three manufacturing sectors. The classification of nine nonmanufacturing sectors in Trinh and Nguyen (2013) is compatible with the classification of nine nonmanufacturing sectors used in this study; thus, PI data of these nine sectors were obtained from Trinh and Nguyen (2013). However, the classification of manufacturing sectors in Trinh and Nguyen (2013) is not compatible with the classification used in this work. Therefore, PI of nine manufacturing sectors (Sectors 3, 4, 5, 6, 7, 8, 9, 10 and 11 as listed in Table 5) is obtained from another database—IPPS (Hettige et al. 1995), which was proposed by the Infrastructure and Environment Team of World Bank. IPPS combines the data of industrial activity with data on pollution release. IPPS relates the PI with respect to value added, GDO and employment (Hettige et al. 1995). In this context, PI with respect to GDO is used because of the availability of GDO data. The PI in the IPPS for the year 1997 (pound per million 1987 US dollars) is adjusted to the correlative year in Vietnamese currency (e.g., kg per million 2011 VND) using the deflation factor based on consumer price index (CPI), as outlined by Mani and Jha (2006).

IPPS has been used in many countries such as Nigeria (Oketola 2007; Odesanya et al. 2012), India (Pandey 2005), Lagos (Oketola and Osibanjo 2007), Brazil and Latin American (Aguayo et al. 2001). In Southeast Asia, the estimation of pollution flows for industrial sectors was also applied for Malaysia (Oketola and Osibanjo 2009), Lao PDR (GMS Environmental Operation Center 2014), Cambodia (San et al. 2018) and Thailand (Benoit and Craig 2001). In Vietnam, ICEM (2007) adapted the database from IPPS to compute the total pollution load of several pollutants for specific regions of the country in the year 2004. According to this report, many existing industrial operations in Vietnam now make use of technologies which are over 15–20 years old and are similar to technologies which operated in the USA during the late 1980s. Thus, the existing Vietnamese industrial technologies are well represented by the underlying technologies reflected in IPPS. Furthermore, ICEM (2007) also carried out the validation of IPPS data by comparing it with data from CTC Vietnam. Correlation coefficients were calculated using the pollution load parameters TSS and BOD between CTC data and IPPS data. This was performed at the provincial and sectoral levels. The results were then ranked in increasing order. The purpose of this process was to know whether high values of total pollution load estimated from the IPPS data well corresponded with high values of total pollution load obtained from CTC data. The results of this process are shown in Tables 6 and 7.

Table 6 Spearman’s rank correlation between estimated BOD and TSS at the provincial level

Table 7 Spearman’s rank correlation between estimated BOD and TSS at the sectoral level

Table 6 indicates high positive correlation for both TSS and BOD (0.71 and 0.69) which implies that there is a close match in terms of priority ranking between these two data sets. Table 7 gives the correlation coefficient at the sectoral level which resulted in relatively low values of correlation (TSS: 0.24 and BOD: 0.33). However, ICEM (2007) then performed a statistical test of the level of significance. The results show that differences in total pollution load between sectors of these two data sets are not statistically significant. This means that PI data from IPPS are suitable to use for the Vietnam scenario. After the PI of sectors is obtained, the extended IO table is generated and is found in Table 8.

Table 8 Water quality parameter-extended IO table in 2011

Results and discussion

Total pollution load trend

In the manufacture of goods and services, the products of each sector are distributed into intermediate demand (inputs to other sectors) and final demand (i.e., household and government consumption). Each process has an associated pollution load. Thus, the total pollution load of an economic sector is simply the sum of the pollution associated with the production of goods or services in the sector to serve all requirements needed by the economy to operate as indicated in Eq. (4). The trend of the total water pollution load as indicated by the water quality parameters represents the variations in total pollution load within the period of 2000–2011. This trend is given for two main water quality indicators—TSS and BOD.

Trend of total suspended solids (TSS)

The TSS trend is shown in Fig. 3. In detail, the line “sum” representing the TSS of the entire economy indicates that in the year 2000, TSS was approximately 300,000 tonnes, but this number increased to more than 1,000,000 tonnes by the year 2011. It means the TSS of the whole economy increased 4 times within 11 years. The reason is that the GDO of each sector increased quickly from 2000 to 2011 (except 2002) (Table 3) as a result of rapid economic growth in Vietnam. At the sectoral level, results also indicate the rapid increase in TSS of each sector. The variation of TSS among the top 6 most polluting sectors and the remaining sectors in the period is also shown in Fig. 3. It can be seen that Sector 9 (basic metal industries), despite comprising only a small proportion of GDO (2% as shown in Table 5), was a main contributor to TSS, accounting for 71% of TSS within the period of consideration. This results from the high PI of this sector (9.88 kg/million VND on average within the period considered) which was 10 times larger than that of the second highest source (Sector 6 with 0.98 kg/million VND). The second source of TSS was Sector 11 (other manufacturing sectors), accounting for 7% of average TSS load of the whole period. Following these sectors are Sector 6 (manufacture of paper and paper product, printing and publishing) and Sector 7 (manufacturing of industrial chemicals). These sectors had high TSS because many of the processes in these sectors are water intensive. In addition, wastewater treatment from these sectors was not well employed (ICEM 2007). Sector 3 (food, beverage and tobacco) and Sector 1 (agriculture, fishery and forestry) show mid-level of TSS. These sectors had lower TSS per unit of GDO, but they had the highest proportion of GDO contribution (see Table 5). The rest of the sectors had insignificant TSS contribution because they had low PI and low GDO.

Fig. 3

The trend of TSS pollution load

From this study, the total amount of TSS due to economic activities in the year 2000 was 344,942 tonnes. The results obtained by other researchers in Vietnam such as Trinh and Nguyen (2013) were 71,391 tonnes of TSS. Trinh and Nguyen (2013) made use of a PI factor which was estimated based on data published by the government as explained in previous section. The results indicate a large difference between the two studies with + 0.79 dissimilarity. The difference is attributed to the large difference in the pollution intensity (PI) of TSS used in these two studies. The average PI of TSS of all manufacturing sectors in this study is 1.7479 kg/million VND. This is almost the same as the result obtained by Le et al. (2013) with 1.8995 kg/million, but it is 28 times higher than in the study of Trinh and Nguyen (2013) which had an average PI of 0.06 kg/million VND. If the PI of Sector 9 (basic metal industries) (13.19 kg/million VND) was replaced by the PI of Sector 6 (manufacture of paper and paper products, printing and publishing) (1.307 kg/million VND), which had the second highest PI, the total TSS in the year 2000 will be 118,154 tonnes. Hence, the dissimilarity can be reduced to only + 0.4. Therefore, it can be said that the high value of PI of TSS in Sector 9 (basic metal industries) obtained from IPPS explains the main difference in TSS between the results obtained here and those obtained by Trinh and Nguyen (2013).

Trend of biological oxygen demand (BOD)

The results of total BOD pollution load indicate that BOD pollution load was 8 times less than TSS. However, the BOD of the entire economy also quickly increased from 42,600 tonnes in 2000 to 121,360 tonnes in 2011 (Fig. 4). Sector 3 (food, beverage and tobacco) was the main contributor of BOD with an average amount of 24,834 tonnes, accounting for 35% of total BOD generation. The next major source of BOD was Sector 1 (agriculture, fishery and forestry) with 19,649 tonnes accounting for 28% of the total average BOD in this period. Two of these sectors show high level of total BOD pollution load because of high sectoral activity as reflected by the high GDO. The mid-level contributor of BOD was Sector 6 (manufacture of paper and paper product, printing and publishing), Sector 9 (basic metal industries), Sector 2 (mining and quarrying) and Sector 7 (manufacturing of industrial chemicals). These sectors were water-intensive industries, but their GDO was low. The remaining sectors had low-level BOD because of very low PI values and sectoral GDO; thus, these sectors were considered as insignificant.

Fig. 4

The trend of BOD pollution load

The total amount of BOD in the year 2000 for the whole economy was 43,419 tonnes which is − 0.017 dissimilar to the work of Trinh and Nguyen (2013). This points out that there is a slight difference between estimating the PI of BOD obtained from the IPPS and that obtained from the government database. In other words, the IPPS data can be well applied for BOD calculations in Vietnam. The trend in performance of the water quality indicators highlights the variations in total pollution load generated by economic sectors to satisfy the demand for products and services. Meanwhile, these results do not reflect the role of sectors and their sectoral connections in terms of pollution. Thus, there is a need to decompose each water pollutant load and embedded pollution load into different components to characterize sectors according to their contribution to water pollution. This facilitates the classification of each sector as consumer/purchaser or producer/seller of water pollution. The next section explains this in more detail.

Role of sectors in relation to water pollution

This subsection uses formulas which are discussed in detail in Appendix to calculate for the components of TSS and BOD parameters in the year 2011. Table 9 in Appendix shows the total pollution load (PL ^k _i ), the embedded pollution load (DI ^k _i ) and pollution components in 2011. In Table 9, it can be seen that Sector 9 (basic metal industries) was responsible for majority of TSS. A total of 95% of TSS from Sector 9 was sold to other sectors of the Vietnamese economy. Thus, Sector 9 is considered a pollution producer or so-called pollution pusher. In contrast, Sector 10 (manufacturing of fabricated metal products, machinery and equipment) and Sector 13 (construction) are of the opposite extreme case and are considered as TSS purchasing sectors and called pollution consumers/pullers. Sector 10 (manufacturing of fabricated metal products, machinery and equipment) required inputs from other sectors which corresponded to purchase 47% of the TSS generated by other sectors of the economy. Similarly, Sector 13 (construction) purchased more than 39%.

Table 9 Sectoral performance in terms of pollution components in 2011

The largest purchaser of BOD was Sector 10 (manufacturing of fabricated metal products, machinery and equipment), which purchased 50,500 tonnes (34%) of total BOD from other sectors (Table 9). The second highest purchaser of BOD was Sector 13 (construction) purchasing 45,400 tonnes (31%). These sectors are thus pollution pullers again. Meanwhile, Sector 9 (basic metal industries) was again considered as a pollution pusher and was responsible for contributing the highest BOD, with 58% of its total pollution load distributed to other sectors.

From this result, pollution-pushing sectors and pollution-pulling sectors with respect to TSS and BOD can be identified. In order to identify which sectors should be prioritized when it comes to pollution reduction, the sectoral indices are calculated and characterized (Table 10). These indices are classified as high and low values as indicated in Table 3. The 18 sectors are then classified into 3 groups based on whether the index value score is high, intermediate or low. The top 6 sectors which received the highest index scores are classified as high, while the last 6 sectors with the smallest index scores are classified as low. Based on Table 3, the characteristic of each sector in relation to TSS and BOD for the year 2011 is given in Table 11. Table 3 provides the guideline on how sectors are classified. Sector 1 (agriculture, fishery and forestry) and Sector 3 (food, beverage and tobacco) are considered key sectors for both TSS and BOD. These sectors had the highest propensity to cause water pollution across the economy. Table 12 summarizes the key results on sector classification based on their TSS and BOD contribution.

Table 10 Sectoral indices

Table 11 Sectoral characteristic/role with respect to TSS and BOD in 2011

Table 12 The role of important sectors in relation to water pollution

Pollution reduction priority

Analyzing the total pollution load trend in the period 2000 to 2011 and evaluating the pollution components in the year 2011 provide insights for the development of waste management strategies. Pollution reduction strategies should focus on:

• (1) Sectors with high total pollution load during the period. This includes Sector 9 (basic metal industries) for TSS; Sector 1 (agriculture, fishery and forestry) and Sector 3 (food, beverage and tobacco) for BOD.

• (2) Sectors having high water pollution impact based on their interrelationship with other economic sectors. Sector 1 (agriculture, fishery and forestry) and Sector 3 (food, beverage and tobacco) have been identified as key sectors in terms of TSS and BOD. Sector 13 (construction) and Sector 10 (manufacturing of fabricated metal products, machinery and equipment) were considered pollution pullers, while Sector 9 (basic metal industries) was considered as pollution pusher.

To decrease water pollution, there are several suggestions to help government, policy makers, household sectors and industrial manufactures in the development and effective implementation of strategies. An example strategy for pollution-pulling sectors is to improve the efficiency of technologies employed in their processes. This can potentially reduce the required inputs of industries and sectors, thereby reducing the pollution associated with sector requirements.

On the side of government and policy makers, they need to devise appropriate regulations for wastewater management especially in sectors contributing to high pollution load and/or having high water pollution impact.

For manufacturers, the best way to reduce pollution is to lower the value of PI. In Sector 9 for example, improvements include strategies which reduce the TSS intensity of processes. This can be obtained by applying cleaner technologies or by improving wastewater treatment systems. In addition, the manufacturers can reduce pollution by shifting from highly pollutive raw materials to less pollutive inputs or by improving process efficiency.

For household sectors, there is a need to manage household consumption by encouraging the use of more environmentally friendly products. Furthermore, the decrease of consumption also helps.

Conclusions and recommendations

This study can be considered as a preliminary investigation of the total water pollution load resulting from economic activities in Vietnam, as indicated by water quality parameters. This can potentially help environmental managers and policy makers in monitoring water pollution in Vietnam. The methodology presented makes use of available IO tables in the years 2000, 2007 and 2011 and other economic data in Vietnam to calculate for two water quality indicators (e.g., TSS and BOD). The results show a tremendous increase in TSS, from 345,000 tonnes in 2000 to 1,199,000 tonnes in 2011, while the total BOD increased from 43,400 tonnes in 2000 to 123,000 tonnes in 2011. For TSS, Sector 9 (basic metal industry) was the main contributor, while Sector 3 (food, beverage and tobacco) and Sector 1 (agriculture, fishery and forestry) were the main contributors for BOD.

In taking account of the linkages among economic sectors and the interrelationship between economic sectors and water pollution, a hybrid approach which utilizes both EEIO and VIC is employed to identify the role of sectors. This is based on an evaluation of pollution component indices proposed by Sanchez-Choliz and Duarte (2003). These indices allow to quantify and disaggregate the pollution that is produced in all sectors. This methodology hence provides a more complete analysis for identifying the role of sectors as either producer/seller or consumer/purchaser of water pollution or both (key sectors). The IO table of 2011 is used for this purpose, and the results highlight that when TSS and BOD are both considered, Sector 1 (agriculture, fishery and forestry) and Sector 3 (food, beverage and tobacco) are identified as key sectors, Sector 9 (basic metal industry) is the highest pollution pusher, while Sector 10 (manufacturing of fabricated metal products, machinery and equipment) and Sector 13 (construction) are the highest pollution pullers. Finally, suggestions for prioritizing sectors to meet pollution reduction targets are presented to help government, policy makers, house hold sectors and industrial manufactures identify the proper water pollution mitigation strategies.

This research focused on the Vietnam case, but it can be easily adapted for use in other nations with existing IO tables and environmental data. This paper is used as another example of applying environmental data from IPPS to calculate environmental pollution. Other developing countries can make use of the methodology presented in this work to track pollution generation in economies of interest and to analyze the interrelationship between economic activities and environmental pollution. The weak point of this research is that the application of global data such as IPPS may result in some uncertainties and the focus has primarily been on the water quality indicators TSS and BOD. Future work can focus on the examination of other pollutants (e.g., chemical oxygen demand, ammonia–nitrogen and hazardous chemicals) or environmental quality indicators and consider different types of environmental media or pathways of pollution release (e.g., water, air and land). In addition, tools for handling data uncertainty can be integrated in the model to result in more robust analysis.

Appendix

The method of vertically integrated coefficient

The method of VCI is developed by Sanchez-Choliz and Duarte (2003) which is based on the Leontief equation:

$$\mathop {x_{i} }\limits^{ \wedge } = A\mathop {x_{i} }\limits^{ \wedge } + \mathop {y_{i} }\limits^{ \wedge }$$

(8)

Eq. (8) can be modified to be new equation:

$$\mathop {x_{i} }\limits^{ \wedge } = \left( {I{ - }A} \right)^{ - 1} \mathop {y_{i} }\limits^{ \wedge }$$

(9)

where y ^∧ _i vector of final demand, $\wedge$ symbol for diagonal matrix, ${\text{A}} = (a_{ij} )$ technical coefficient matrix with n × n dimension, I identity matrix. From Eqs. 8 and 9, it can be rewritten as:

$$\mathop {x_{i} }\limits^{ \wedge } = A\left( {I{ - }A} \right)^{ - 1} \mathop {y_{i} }\limits^{ \wedge } + \mathop {y_{i} }\limits^{ \wedge } = M\mathop {y_{i} }\limits^{ \wedge } + \mathop {y_{i} }\limits^{ \wedge }$$

(10)

where

$$M{ = }A\left( {I{ - }A} \right)^{ - 1} = (m_{{_{ij} }} )$$

(11)

$$(\alpha_{ij} ) = (I{ - }A )^{ - 1} \quad \left( {\text{Leontief inverse}} \right)$$

(12)

Then, if we multiply pollution intensity $PI_{j}^{k}$, which is the amount of pollution load of an activity j for kth-type environmental quality parameter emitted to produce one unit of monetary output for that activity, with final demand $y_{i}$, we obtain the total value of kth parameter:

$$\mathop {{\text{DI}}_{i}^{k} }\limits^{ \wedge } = \sum\limits_{j = 1}^{n} {{\text{PI}}_{j}^{k} } \alpha_{ji} \mathop {y_{i} }\limits^{ \wedge }$$

(13)

is the kth parameter that is associated with the production of the final demand $y_{i}$ of sector i.

Furthermore, pre-multiplying Eq. (10) by vector $\mathop {PI_{j}^{k} }\limits^{ \wedge }$, we obtain:

$$\mathop {{\text{PI}}_{j}^{k} }\limits^{ \wedge } \mathop {x_{i} }\limits^{ \wedge } = \mathop {{\text{PI}}_{j}^{k} }\limits^{ \wedge } {\text{M}}\mathop {y_{i} }\limits^{ \wedge } + \mathop {{\text{PI}}_{j}^{k} }\limits^{ \wedge } \mathop {y_{i} }\limits^{ \wedge }$$

(14)

It can be seen that (14) can be broken down into:

$${\text{PI}}_{j}^{k} \alpha_{ji} y_{i} = {\text{PI}}_{j}^{k} m_{ji} y_{i} \quad if\,i \, \# \, j$$

(15)

and

$${\text{PI}}_{i}^{k} \alpha_{ii} y_{i} = {\text{PI}}_{i}^{k} m_{ii} y_{i} + {\text{PI}}_{i}^{k} y_{i} \quad if\,i \, = \, j$$

(16)

From the combination of three equations (Eq. 13, 15 and 16), it can be written as follows:

$$\begin{aligned} {\text{DI}}_{i}^{k} = \sum\limits_{j = 1,j\# i}^{n} {{\text{PI}}_{j}^{k} } \alpha_{ji} y_{i} + {\text{PI}}_{i}^{k} \alpha_{ii} \mathop {y_{i} }\limits^{ \wedge } = \sum\limits_{j = 1,j\# i}^{n} {{\text{PI}}_{j}^{k} } \alpha_{ji} y_{i} + {\text{PI}}_{i}^{k} m_{ii} y_{i} + {\text{PI}}_{i}^{k} y_{i} \hfill \\ = \sum\limits_{j = 1,j\# i}^{n} {{\text{PI}}_{j}^{k} } \alpha_{ji} y_{i} + {\text{PI}}_{i}^{k} \left( {\sum\limits_{j = 1,j\# i}^{n} {a_{ij} \alpha_{ij} } } \right)y_{i} + {\text{PI}}_{i}^{k} a_{ii} \alpha_{ii} y_{i} + {\text{PI}}_{i}^{k} y_{i} \hfill \\ \end{aligned}$$

(17)

Each term of Eq. (17) is defined as follows:

$${\text{TB}}_{i}^{k} = \sum\limits_{j = 1,j\# i}^{n} {{\text{PI}}_{j}^{k} } \alpha_{ji} y_{i} \quad {\text{True backward linkage}}$$

(18)

$${\text{SE}}_{i}^{k} = {\text{PI}}_{i}^{k} \left( {\sum\limits_{j = 1,j\# i}^{n} {a_{ij} \alpha_{ij} } } \right)y_{i} \quad {\text{Semi}} - {\text{own pollution}}$$

(19)

$${\text{OW}}_{i}^{k} = {\text{PI}}_{i}^{k} a_{ii} \alpha_{ii} y_{i} \quad {\text{Own pollution}}$$

(20)

$${\text{FI}}_{i}^{k} = {\text{PI}}_{i}^{k} y_{i} \quad {\text{Final demand pollution}}$$

(21)

Forward linkage:${\text{PI}}_{i}^{k} x_{i}$ is total pollution load of sector i which consists of ${\text{TB}}_{i}^{k}$, ${\text{SE}}_{i}^{k}$, ${\text{OW}}_{i}^{k}$, ${\text{FI}}_{i}^{k}$ and the one emitted to produce inputs for other sectors without returning which is considered as true forward linkage:

$${\text{TF}}_{i}^{k} = {\text{PI}}_{i}^{k} x_{i} - {\text{FI}}_{i}^{k} - {\text{OW}}_{i}^{k} - {\text{SE}}_{i}^{k}$$

(22)

The difference between true backward linkage and true forward linkage is the following:

$$d_{i} = {\text{TB}}_{i}^{k} - {\text{TF}}_{i}^{k}$$

(23)

$d_{i}$ tells us whether true backward linkage (buying of pollution from other sectors) or true forward linkage (selling of production to other sectors) is the larger. If $d_{i} > 0$, then sector i is an overall polluting pullers and other sectors pollute for it. If d _i  < 0, then sector i is an overall polluting pusher and pollutes for other sectors to obtain their final demands.