Developing a Framework on Designing a Sustainable Supply Chain by Integrating Input-Output Analysis and DEMATEL Method: A Case Study on Textile Industry in Indonesia

Concern about a sustainable supply chain has grown significantly during the last two decades. One reason is the increased awareness among stakeholders of the impact of industry actions that cause environmental issues. A sustainable supply chain is usually justified in terms of its contribution to an expansion of economic activity, an improvement in environmental quality, and enhancing human well-being [1]. Nowadays, companies are trying to incorporate sustainability standards into their strategic design planning to reduce ecological and social risks and ensure profitability and growth [2]. Moreover, sustainable supply chain (SC) management becomes an important strategic decision for the manufacturing industries.

Stricter regulations by governments and pressure from various stakeholders concerning environmental issues have contributed to the rising importance of including sustainability in the supply chain design [2]. Consequently, environmental aspects of SC have been discussed in the literature [3]. Although some companies consider environmental improvement, the practice in a SC is not easy. One of the reasons is that SC consists of cross-function parties.

A practical, sustainable supply chain design requires developing analytical models and designing appropriate measurement tools. Therefore, it is important to understand the quantitative impact of environmental issues in supply chain decision-making. This research explores a two-dimensional sustainable supply chain design, which includes economic and environmental aspects. Determining the relationship between parties and quantifying its implication will help decision-makers identify strategies to manage and coordinate the relations within the SC [4].

A sustainable supply chain structure is a long and complex problem. Designing an effective, sustainable supply chain requires the development of analytical models and appropriate measurement tools. Currently, the majority of supply chain structure research uses a micro-level approach. One approach used is the development of mathematical models. A number of studies in the last decade have proposed various optimization models to overcome sustainable supply chain design problems [5-7]. Apart from that, some studies consider a combination of cost factors and emission reductions from production and transportation processes as criteria in decision-making [8, 9]. Zhang et al. [10] designed a network model that minimizes total costs, maximizes customer demand coverage and minimizes negative environmental impacts.

Previous research shows the use of IO analysis in the supply chain design of specific products/processes. IO analysis was initially only used to analyze changes in my economy. However, IO integration has been developed with other approaches, such as LCA, DEMATEL, and ANP. You et al. [7] developed a Multi-Objective (MO) model by integrating an LCA approach to design an optimal biofuel supply chain. On the same research object, Yue et al. [11] developed a model to minimize total costs and GHG (Green House Gases) emissions both directly/indirectly. The model combines MO-MILP (Multi-Objective – Multi Integer Linear Programming), Life Cycle Assessment (LCA), and IO analysis to provide the results of the biofuel supply chain's techno-economic, social and emissions analysis. This research focuses on the biofuel supply chain, and the social indicator is the number of local jobs gained.

Based on previous research, IO analysis has been used in macro-level research. However, research that uses IO analysis at the macro level generally looks at environmental impacts due to the dynamics of global economic change [12-14]. Feng et al. [15] used IO analysis to evaluate the remanufacturing industrial sector, focusing on the impact on other industrial sectors, emission reduction, and the national economy. Research on supply chain design at the macro level is still rare, even though supply chain design requires a comprehensive, holistic analysis of the supply chain.

Several previous research studies have used the integration of IO analysis with the DEMATEL method for decision-making regarding supplier selection [16], sharing economy [17], and supply chain design, which considers environmental impacts in the steel industry [4]. However, the relationship between sectors has not yet been deeply discussed. Therefore, this research adopts an IO analysis approach and integrates it with the DEMATEL to understand the close relationship between sectors. DEMATEL was selected over other MCDM techniques due to its ability to analyze the relationship between sectors based on the relation's intensity, not only similarity as other techniques provide.

 The textile industry in Indonesia is a case study in this research. The textile industry is one of the manufacturing industries experiencing rapid growth. The Indonesian Ministry of Industry has made the textile industry one of the development priorities in the Making Indonesia 4.0 Roadmap program. The development of the textile industry through Industry 4.0 aims to increase the domestic textile industry's competitiveness by utilizing technology capable of producing clothing and textiles for more specific needs [18]. Unfortunately, the economic contribution of Indonesia's fashion or textile industry is not directly proportional to its impact on the environment. Among the G20 countries, the textile industry of Indonesia ranks 2^nd as the cause of water pollution [19]. Until now, the process of creating an environmentally friendly textile industry has faced many challenges.

The growth of the textile industry indicates the magnitude of the impact of the textile industry supply chain from an economic and environmental perspective. The textile industry supply chain begins with fiber production using raw materials from various types of plants, followed by the yarn and fabric production process. The final stage is producing consumer products, such as clothing and garment products. Apart from that, many other industries indirectly form the textile industry supply chain, which indirectly becomes part of the environmental impact of textile industry activities. This is why analyzing the supply chain structure at the macro level is important. Thus, the research question is how the supply chain structure would explain the economic and environmental impact in the textile industry.

This study develops an approach that uses Input Output (IO) data taken from the World Input-Output Database (WIOD) to measure environmental impacts at the economic sector level. The purpose of this study is to design the supply chain structure of the textile industry and identify the relationships and linkages between sectors in the supply chain to analyze the industry with the most significant environmental impact. The method combines DEMATEL (Decision-Making Trial and Evaluation Laboratory) and the IO model. This macro-level analysis approach uses Indonesia's macroeconomic data as a case study based on the textile industry supply chain in three tiers. The integration model is expected to produce a quantitative analysis of economic and environmental impacts on designing the supply chain structure of the textile industry.

This research uses the world I/O table as the primary matrix M. The data released in 2016 leads to a 56×56 matrix demonstrating direct and indirect relationships between industries. Table 1 presents the names of sectors in WIOD, while Table 2 illustrates the matrix M (in millions of US$).

The normal matrix N aims to calculate the ratio of data using Eq. (1), with the value of $\delta$ (Eq. (2)) as follows:

$\delta=\left(\frac{1}{392,813}\right)=2.546 \times 10^{-6}$                      (7)

The next step is calculating the total relation matrix (T) using Eq. (3) and determining the vectors R and J. The vector R is the summation of rows (Eq. (4)), while the vector J is the sum of columns (Eq. (5)). The matrix T with vectors R and J is illustrated in Table 3. Matrix T presents the degree of influence of each industry. The analysis's degree of influence shows industries' superiority in the supply chain. The superiority vector determines sectors that have a dominant influence on other industries.

Table 1. Sectors in WIOD

Table 2. Matrix M

Table 3. Total relation matrix (S)

1.png

Figure 1. The superiority vector (R+J) diagram

Figure 1 illustrates the R+J diagram of the top 20 industries with the highest degree, which shows r5 and r27, namely the Manufacture of food products, beverages, and tobacco products and the construction sector. Both sectors play an important role in the relationship between industries. Figure 1 shows the industry strength of influence based on the highest to lowest according to the R+J value of the economic transactions. The highest industry is r5 (Manufacture of food products, beverages and tobacco products). This indicates that the food industry is the most significant. The second highest influence is the construction industry (r27). The five highest influences are r4, r1, r29, r10, and r24.

Meanwhile, Figure 2 illustrates the relation (R-C) diagram of 20 industries. In other literature, the relation vector is also known as the net effect. It shows that sector r1 (Crop and animal production, hunting, and related service activities) has the highest positive degree. It is followed by r4 (Mining and quarrying) in second place. This indicates they are the two industries with the most influence in Indonesia. The graph's positive side (right side) is called the causal group, which requires a large amount of input from other sectors [17]. It consists of sectors that have a significant influence on the economic relationship. The three highest sectors are r1, r4 and r29.

On the other hand, the diagrams' negative side shows that sectors with input-oriented strength are more significant than output-oriented strength. In other words, sectors with a negative R-J indicate they are network suppliers. The three main suppliers are r27 (Construction), r5 (Manufacture of food products, beverages, and tobacco products), and r36 (Accommodation and food service activities).

2.png

Figure 2. The relation vector (R-J) diagram

The MICMAC diagram (Figure 3) shows a Cartesian diagram that is divided into four quadrants, namely autonomous (I), dependent (II), linkage (III), and independent (IV).

3.png

Figure 3. MICMAC diagram

Sectors r27 and r5 are in the quadrant IV. This indicates that sectors r27 and r5 significantly influence other industries but are unaffected by them. This implies the effectiveness of sectors such as construction and food products, beverages, and tobacco in increasing the overall economic system performance [23, 24]. The MICMAC diagram also shows that none of the sectors are in quadrant I, which means none had low driving and independence power. This also implies all sectors have their roles in Indonesia's economic growth. Most sectors have weak driving power but strong dependence power, thus located in quadrant II (dependent). This indicates the dependency of sectors to each other. The growth of one sector will have a positive and/or negative impact on different sectors. Finally, three sectors (r4, r1, r29) are in quadrant III, which indicates a mutual relationship among these sectors. These three sectors are both driving and dependent and are affected by their actions, thus making them unstable and difficult to address.

There are five industrial sectors in quadrants 3 and 4, while the other sectors are spread across quadrants 1 and 2. The industrial sectors in quadrants 3 and 4 indicate that these sectors act as industries influenced by demand [8]. The five sectors are r27, r5, r4, r1, r29. Meanwhile, other sectors can be classified as those influenced by production. This also implies a supply chain strategy for the sectors. Where sectors in Quadrant 1 are suitable with a pull system strategy. Meanwhile, the push strategy suits quadrant 1 and 2 sectors more.

To develop the supply chain network, we need to determine industries in tier 2 and tier 3 of the textile supply chain. Table 2 shows the three highest industries in column c6 (Textile industries) from matrix T. The industries ranked the first three are r29, r11, and r1. Table 4 shows the industries in tier 2.

Table 4. Three main tier 2 suppliers of textile industries

The same method was applied to determine Tier 3 industries. Figure 4 shows the economic supply chain network of Textile Industries in Indonesia. This supply chain network helps identify market segments more intuitively, which should be prioritized in resource efficiency efforts. Tier 2 and Tier 3 represent the intermediate demands and describe the resources used by other sectors to produce other products and services that are ultimately used in the textile industry.

The first objective of this research is to design the supply chain structure of textile industries and understand the criticality of resources. The result shows that chemical manufacturing, wholesale trade, and crop and animal sectors are the three main sectors in Tier 1. Moreover, the results also imply the importance of considered sectors in the second and third tiers. Interestingly, most sectors categorized as prominent or influential (based on the MICMAC diagram) are considered second and third tiers in the supply chain. This highlights the importance of considering the upstream supply chain to improve overall performance.

The next step is to design the supply chain network by considering the environmental aspects. The purpose is to examine resources with significant contributions to sustainable footprint in the textile supply chain. This paper used CO₂ intensity as the indicator of impact on the environment. First, the intensity matrix of CO₂ (M*) due to economic transaction using Eq. (6). The diagonal matrix of CO₂ production is taken from the WIOD database. As mentioned previously, only 18 sectors would be considered in this analysis. The 18 highest sectors are selected based on the most increased economic interaction with the textile industries. Thus, M* is an 18×18 matrix.

4.png

Figure 4. Indonesia's textile industry supply chain network based on economic transactions 2014

By following the DEMATEL methods, the total relation matrix for CO₂ (T*) was calculated. The T* was used to design the sustainable supply chain network for the textile industry. Table 5 shows the three industries with the highest contributions to producing CO₂ in the textile sectors, based on the relative values of CO₂ emission of each sector.

The electricity and gas industry is the most polluted sector in the textile supply chain. The chemical manufacturing and crop and animal production sectors follow this. This differs from the result of the most essential sectors based on economic transactions (Table 4), where wholesale trade is first—followed by chemical manufacturing and crop and animal product sectors. Interestingly, electricity and gas (r24) is in the 5^th position as the sector that contributes economically to development but ranks first as a polluted sector in the textile industries. Moreover, this implies the contribution of tier 2 and tier 3 sectors toward environmental impact.

Table 5. The first three most polluting industries in Indonesia's textile supply chain in 2014

Next, the supply chain network's second tiers and the third tier sectors based on CO₂ emission are found. Figure 5 presents the relative intensity of CO₂ emission relation within the SC of textiles industries in Indonesia.

5.png

Figure 5. The CO₂ emission interaction among sectors

Incorporating environmental consideration into the supply chain is well recognized, particularly in the sector where enterprises rely on natural resources as role material [25]. By analyzing the contribution of resources/suppliers in the supply chain, enterprises can focus on sectors that need sustainability improvement. Hence, a new sustainability supply chain for the textile industry should be developed.