Integrated ANP and TOPSIS Method for Supplier Performance Assessment

Christine Nataliaa*, Ita Primsa Surbaktia, Chendrasari Wahyu Oktaviab, Andre Sugiokoa a Industrial Engineering Department, Atma Jaya Catholic University of Indonesia Jalan Raya Cisauk, BSD City, Tangerang, Banten, 15345, Indonesia b Industrial Engineering Department, Wijaya Putra University Benowo Campus, Jalan Raya Benowo 1-3, Surabaya, Jawa Timur, 60197, Indonesia * Corresponding author: chrisnatalia@atmajaya.ac.id

A supplier has become one of the main factors that influence a company's supply chain activities. Supplier assessment is vital as suppliers have different performance. This study aims at assessing supplier performance using the integration of ANP and TOPSIS methods. Supplier performance assessment was based on supplier criteria indicators. The weighting criteria used ANP was used to determine the most significant influence criteria of supplier performance. Furthermore, TOPSIS was also employed to obtain supplier preference. Eight criteria and twenty-five subcriteria were used for the supplier performance assessment. The three highest sub-criteria were specification of quality, the flexibility of order changes, and production capacity. The priority results for suppliers were sorted from the highest to lowest ratio values. consume 40 to 80 percent of the total production costs, it affects company performance [7]. Choosing the right supplier can reduce purchasing costs, increase profit, reduce lead time, upgrade consumer satisfaction, and intensify company performance [8]. It also can improve company performance [3]. Supplier evaluation and selection is a complicated activity as it involves many criteria [9] [10] [11]. Various criteria must be considered in the process of decision-making [12].
Several studies have been carried out in supplier selection/evaluation. Generally, evaluation and supplier selection use qualitative and quantitative approaches [13]. Simić, et al. [12] reviewed the fuzzy set theory and models for supplier performance. Yusuf et al. [14] proposed integrating the Analytical Hierarchy Process (AHP) and TOPSIS. They offered six criteria and fourteen sub-criteria. Sarkar and Mohapatra [15] used a fuzzy set approach. Twenty-three criteria were proposed for supplier selection, such as price, capability, and quality. Pangestu [16] offered the Fuzzy Analytical Hierarchy Process. He employed some criteria such as rejection rate, payment level, delivery, and price. Azwir and Pasaribu [17] used the ANP method with four criteria: rejected rate, payment transaction, delivery, and price. Khoiro [18] proposed AHP and Taguchi Loss function. This study used the criteria, including price, quality, delivery, underweight, availability, payment system, guarantee policy, and repair service. Lin, et al. [19] developed methods of Analytic Network Process (ANP), Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), and Linear Programming (LP). Twenty-one criteria were used as the basis for supplier assessment [19].
Based on previous studies, several studies have systematically investigated supplier selection/evaluation. One famous study was conducted by Yuliandono et al. [20]. They considered fourteen criteria for supplier selection. FAHP and TOPSIS were proposed to the evaluation/selection suppliers. However, this research has not considered relationships between the criteria. Very little research has addressed the selection of suppliers to examine the network model between the criteria and sub-criteria. It is deemed essential to find out the relationships between criteria and sub-criteria. Criteria and subcriteria are categorized as influential and influenced, respectively. This study aims to integrate the ANP and TOPSIS methods into supplier performance assessment. Several methods can be combined with the quality of decision-making [15]. Hence, the integration between ANP and TOPSIS is proposed to evaluate supplier performance to improve decision-making quality. Supplier selection and assessment may use several factors [21]. ANP is applied to criteria weighting due to its superiority in obtaining dependency and interdependence interactions between criteria and sub-criteria [22]. Weights of criteria and sub-criteria in ANP are obtained through TOPSIS in evaluating supplier performance. TOPSIS is an effective method to find the most efficient alternative [23]. Hence, It is expected that this research will contribute to a deeper understanding of supplier performance assessment, specifically on the ANP and TOPSIS methods.

Methods
In this section, we proposed a framework for supplier performance assessment. It can be seen in Fig. 1. This study suggested the integration of ANP and TOPSIS to evaluate supplier performance. The ANP was used to weight the criteria and sub-criteria. The weights of criteria and sub-criteria in ANP were then used in TOPSIS to assess supplier performance. The steps of ANP and TOPSIS are explained below.

Identification of criteria and sub-criteria
The first step was identifying the criteria and sub-criteria used in selecting suppliers. The criteria for supplier assessment depend on the company's supply chain strategy [24]. The criteria were determined from the literature review and discussion with experts [13]. In this study, ten criteria and twenty-nine sub-criteria from the literature review are shown in Table 1. Table 1. Criteria and sub-criteria

Criteria
Sub-criteria References

Assessment of criteria and sub-criteria
All criteria and sub-criteria were assessed according to Table 1. The assessment was to select critical criteria and sub-criteria on the supplier performance assessment. The results of the assessment of criteria and sub-criteria were weighted based on the ANP procedures. We proposed assessing criteria and sub-criteria using five levels of importance scale, ranging from 1 (not important) to 5 (very important). The sub-criteria with an average value below 3 indicated it was not important.

Analytical Network Process (ANP)
As stated earlier, ANP was used to assess the weights of criteria and sub-criteria. This study employed ANP modified from Saaty and Vargas [27]. The steps of ANP are as follows:

Determining the Network Relationship
Based on the assessed criteria and sub-criteria, this study developed interdependencies between the criteria. At this stage, questionnaires were used to determine the relationship between the criteria. Experts filled out the questionnaires. The relationship between the criteria was described by arrows [24]. The ANP model was developed by Saaty [28] based on the relation of the interdependencies between the subcriteria [29]. We created a relationship between the criteria, in which the arrow indicated dependency between the networks.

Determining the Weights
At the last stage, ANP was employed to determine the criteria and sub-criteria weights. In this stage, a pairwise comparison was used based on the interdependence criteria in the previous step. Super Decision Software [25] was made use to determine the weights of the criteria and sub-criteria. The criteria and sub-criteria weights were utilized to assess supplier performance through the TOPSIS method. The TOPSIS procedure is described in the next section.

TOPSIS Method
This study employed the TOPSIS method for the supplier performance assessment. The weights of criteria and sub-criteria at the TOPSIS stage were obtained from the ANP method. The principle of the TOPSIS method was chosen based on the distance from the ideal solution. The positive ideal solution is defined as the sum of the best values that can be achieved for each attribute. On the other hand, the negative ideal solution consists of all the lowest values that can be achieved for each quality [22]. The stages of the TOPSIS method in this study were based on [30].

Data Collection
A case study was conducted in manufacturing companies in the automotive industry. Three experts were involved as respondents in this study. Respondents were chosen based on their responsibilities in the procurement department. The three respondents were the procurement and finance department manager, head of the procurement department, and procurement department staff. The respondents filled out the questionnaires about supplier performance assessment. The supplier performance assessment itself was conducted in five suppliers: Supplier 1, Supplier 2, Supplier 3, Supplier 4, and Supplier 5.
Four questionnaires were used to assess supplier performance. The first questionnaire was to evaluate the criteria and sub-criteria. It used five levels of importance scale, ranging from 1 (not important) to 5 (very important). The second questionnaire was to determine the interdependence (network relations) of the criteria. It used a checklist. The third questionnaire was the pairwise comparison questionnaire (ANP procedure). Three questions in the pairwise comparison questionnaire were the criteria question between the inner cluster sub-criteria questions and the outer cluster sub-criteria question. These questions used nine levels of importance ranging from 1 (equally important) to 9 (very important). The fourth questionnaire was supplier performance assessment using TOPSIS. It used five levels of supplier assessment scale sub-criteria, such as 1 (very poor) to 5 (very high).

Results of Evaluation criteria and sub-criteria and interdependence
The average importance level of the criteria and sub-criteria (questionnaire 1) is presented in Table 2. It shows the criteria and sub-criteria with the average values based on the questionnaires filled out by three respondents. This result shows eight criteria and twenty-five sub-criteria as the indicators in the assessment of supplier performance. On the other hand, two criteria and four sub-criteria were not selected as indicators of the assessment.
The relationship between the sub-criteria (the second questionnaire) shows that dependency between the elements occurred. In Table 3, the yellow cell box shows that more than one respondent assessed the influence of one sub-criteria toward another subcriteria. However, the orange cell box describes that only one respondent evaluated the impact-for example, three respondents assessed the influence of B1 criteria on C2 criteria. Based on Table 3, the quality criterion was a criterion that affected other criteria.
Furthermore, This result also describes interdependence between the criteria. Fig.  2 shows a network relation model between the criteria based on the results of the second questionnaire. The reputation and performance were influenced by the design and development, quality, price, delivery, responsive, and flexibility clusters. However, reputation and performance influenced the quality and production facility and capacity.

ANP Model Based on Network
Three types of networks in ANP included emotional dependence, external dependence, and feedback (influence between groups). Fig. 2 shows that each cluster had an inner dependency relationship. It occurred when elements in the same cluster had a meaningful relationship. The example of the inner dependency network was quality criteria. Respondents assessed the quality according to the specifications influenced by the percentage of rejected raw materials and packaging accuracy. Meanwhile, when the packaging is not proper, it also affected the quality of the raw materials. It is in line with the statement in the research by Kurniawati et al. [26].
The outer dependence relationship in the quality criteria was also present. The sophistication of the production equipment affected the quality according to the specifications. Besides, the quality criteria, such as the quality according to specifications and quality consistency, influenced price consistency. The findings are consistent with Pujotomo, et al. [22] as they showed the effect of quality consistency toward price consistency.
The feedback network was the cluster element that affected the elements in other clusters and vice versa. Based on the results in Fig. 3, the relationship in the feedbacks between the criteria occurred. The quality criteria had feedback toward delivery. This study confirms the results of the study by Kurniati et al. [26]. They found out the interrelationship between quality criteria and delivery. The results of the study are also in line with Ekawati et al. [31]. ISSN : 1978- Table 4 shows the results of the weighting of the criteria and sub-criteria utilizing ANP. It shows the weights of the criteria and sub-criteria of each cluster. The criteria with the highest weights were quality, facility and production capacity, and requirement. Furthermore, the five most essential sub-criteria were quality according to specifications, changes in the number of orders, production capacity, percentage of rejected raw materials, and production lead time. Quality, according to the specifications, was the most crucial sub-criteria in supplier performance assessment. This result is by Pujotomo, et al [22], Taherdoost and Brard [5], Kurniawati, et al. [26], and Rashidi and Cullinane [25].
The weight sub-criteria assessment of supplier performance shows that the company attached great importance to the specifications and four other sub-criteria. Furthermore, the lowest weight is in the responsiveness sub-criteria, both product-related information and claims. In other words, responsiveness was the little significant indicator.

Conclusions
This study aims to integrate the ANP method and the TOPSIS method into supplier performance assessment. The eight criteria used in the assessment consisted of quality criteria, shipping criteria, price criteria, reputation and performance criteria, flexibility criteria, responsive criteria, product development and design criteria, and facility and production capacity criteria.
Twenty-five sub-criteria were selected for supplier performance assessment.
The result shows the main criteria in supplier performance assessment were quality criteria, facilities and production capacity criteria, flexibility criteria, price criteria, reputation and performance, delivery, design and development, and responsiveness. The most considerable sub-criteria was quality according to specifications, flexibility in changing the number of orders, production capacity, and percentage of 73% 59% 37% 48% 50% rejected raw material. These results indicate the effective integration of ANP and TOPSIS used in supplier performance assessment.
This study has limitations in the scope of criteria and sub-criteria chosen by the company. Hence, the Focus Group Discussion (FGD) process may give better results. Besides, integrating the ANP-TOPSIS methods with other procedures can be developed to provide a better solution.