Credit Risk Comprehensive Evaluation Method for Online Trading Company 1 *1, Corresponding Author School of Economics and Management, Beijing Forestry University, fankun@bjfu.edu.cn Abstract A new comprehensive evaluation method is proposed to assess the credit risk of online trading company. Sometimes many companies have good online credit, but they maybe face credit crisis because of their serious financial problems. So we need to consider both online and offline credit risk. In this paper, credit risk evaluation model consisting of offline and online credit evaluation index systems is presented to determine the credibility of the e-commerce participants. The offline index includes the financial situation, internal characteristic and external condition these three major indicators, while online index consists of product, service, transport and overall credit evaluation these four major indexes. Expert assessment method and the analytic hierarchy process (AHP) are used to determine the indexes weights and the multi-hierarchy fuzzy comprehensive evaluation method is used to calculate the credit evaluation score. Based on this credit score, we can obtain the online trading company's standard credit rating. Finally, case analysis illustrates our credit risk comprehensive evaluation method is effective and comprehensive. Keywords: E-commerce, Credit Risk, Comprehensive Evaluation, Online Trading Company 1. Introduction With the rapid development of e-commerce, enterprises are facing an increasing number of opportunities and challenges. Because of the openness and anonymity of online transactions, the sellers and the buyers are both faced with credit risk, and credit problems become obstacles to the development of e-commerce. Therefore, the invention of an effective credit risk comprehensive evaluation method for online trading companies has important practical significance. The world-wide scholars have performed lots of researches on credit evaluation. Chen and Chiou [1] presented a fuzzy credit-rating approach to deal with the problem arisen from the credit-rating table currently used in Taiwan. Jiao et al. [2] used fuzzy adaptive network (FAN) to model the credit rating of small financial enterprises. Min and Lee [3] proposed a DEA-based approach to credit scoring. Hajek [4] presented the modeling possibilities of neural networks on a complex real-world problem, i.e. municipal credit rating modeling. Besides, numerous studies have been developed to discuss the bank credit evaluation problems [5,6]. However, all of the researches about general credit evaluation methods have their different application and are not applicable for e-commerce field. Therefore, researchers have to develop new alternative methods. Guo and Zheng [7] proposed a model to evaluation the trust of transactions in the C2C e-commerce business. Xu and Zhang [8] presented a new credit evaluation method to determine the credibility of the electronic commerce participants. They invented a multi-indicator method, which used the analytic hierarchy process (AHP) determine the indicator s weight and used the set pair analysis (SPA) to achieve a overall credit score of the evaluated. However, this multi-indicator only included purchase times, purchase amount, monetary value, product quality, product price and time liness of delivery these six online indicators, while the offline credit indicators of sellers were not considered. In this paper, we try to invent a credit risk comprehensive evaluation method for online trading companies, which includes offline and online credit risk index systems. Here, online trading companies refer to Internet companies, commercial enterprises or manufacturers which mainly engaged in the online sale businesses. The purpose of this article is three-fold: (1) to highlight the offline credit that also exists in the online trading companies; (2) to find the intersection between business performance evaluation and credit evaluation; and (3) to build the credit risk comprehensive evaluation system and propose the fuzzy evaluation method for Chinese e-commerce businesses. Advances in information Sciences and Service Sciences(AISS) Volume4, Number6, April 212 doi: 1.4156/AISS.vol4.issue6.12 12
2. The evaluation index system design for online trading company The design of the index system is fundamental and specific in the evaluation process so that it needs extensive investigation, careful analysis and synthesis. Specifically, to construct an index system, we have to induce and synthesize the main factors affecting the credit risk into a series of indexes that should be clear in connotation and denotation, and can be cope easily. Furthermore, the indexes should be organized by their internal connections and hierarchical relationship. It is very important to establish a set of scientific and reasonable index systems [9]. Through expert consultation, we list all selected indicators in Table 1 and Table 2, and group them into two categories, i.e. offline credit risk evaluation index system and online credit risk evaluation index system. As can be seen from Table 1, offline credit risk evaluation index U (in Grade one) of online trading company includes the Financial situation U 1, Internal characteristic U 2 and External condition U 3. And these three major indexes have 4, 3 and 2 sub-indexes in grade three respectively. The financial situation indexes in grade four are 9 specific financial indicators, such as Rate of Return on Common Stockholders Equity (U 111 ), Return on Sales (U 112 ), Asset-liability ratio (U 121), Quick ratio (U 122 ), Interest coverage (U 123 ), Turnover of total capital (U 131 ), Turnover ratio of receivable (U 132 ), Sales growth rate (U 141 ) and Rate of capital accumulation (U 142 ). The values of these nine financial indicators can be obtained from the company. In addition, we can get the evaluations of internal characteristic U 211 -U 232 and external condition U 311 -U 322 by expert assessment. In Table 2, online credit risk evaluation index R (in Grade one) of online trading company includes product (R 1 ), Service (R 2 ), Transport (R 3 ) and Overall credit evaluation (R 4 ) these four major indexes. However, online indicators in [8] include purchase times, purchase amount, monetary value, product quality, product price and time liness of delivery. It is not difficult to find that we use cost performance instead of product quality and product price. Meanwhile, transaction amount and purchase amount should be the same meaning, and this situation also occurs between transaction times and purchase times. In addition, transportation time is also as same as the time liness of delivery. Thus, it is obviously that our online credit risk index system includes all online indicators proposed in [8] except monetary value, and we expand the online indexes in order to get more accurate evaluation. Table 1. Offline credit risk evaluation index system and weight Grade two Grade three Grade four Financial situation (U 1 ), (.57) Profitability(U 11 ), (.45) Internal characteristic (U 2 ), (.29) External condition(u 3 ), (.14) Debt-paying ability(u 12 ), (.32) Operating status (U 13 ), (.13) Dynamic prospect (U 14 ), (.1) Market quotation (U 21 ), (.29) The strength of the buyer (U 213 ), (.26) Business management(u 22 ), (.5) The manager s ability(u 221 ), (.4) Business standing (U 23 ), (.21) Industry conditions (U 31 ), (.5) Industry outlook (U 32 ), (.5) Rate of Return on Common Stockholders Equity (U 111 ), (.4) Return on Sales (U 112 ), (.6) Asset-liability ratio (U 121), (.41) Quick ratio (U 122 ), (.33) Interest coverage (U 123 ), (.26) Turnover of total capital (U 131 ), (.5) Turnover ratio of receivable (U 132 ), (.5) Sales growth rate (U 141 ), (.6) Rate of capital accumulation (U 142 ), (.4) Marketing strategy (U 211 ), (.41) Market share (U 212 ), (.33) Enterprise organizational structure (U 222 ), (.4) Production control (U 223 ), (.2) Contract performance (U 231 ), (.5) The use of capital and loans (U 232 ), (.5) Stage of development of the industry(u 311 ), (.4) The difficulty of entering the industry(u 312 ), (.2) Competition in the industry (U 313 ), (.4) Policy support (U 321 ), (.5) Outlook (U 322 ), (.5) 13
Table 2. Online credit risk evaluation index system and weight Grade two Grade three Product (R1), (.35) Service (R2), (.25) Transport (R3), (.23) Overall credit evaluation (R4), (.17) Cost performance (R11), (.45) Transaction amount (R12), (.13) Transaction times (R13), (.1) Consistency of product with purchase intention (R14), (.32) Pre-sale services (R21), (.3) After-sales service (R22), (.7) Transportation time (R31), (.4) Transportation cost (R32), (.4) Damage (R33), (.2) Long-term credit (R41), (.4) The current credit (R42), (.6) The Analytical Hierarchy Process (AHP), developed by Saaty and Vargas [1], is one of the methods used in multi-criteria decision-making and can be employed to assist individuals and groups in ranking the credit risk attributes. By incorporating both subjective and objective data into a logical hierarchical framework, AHP provides decision-makers with an intuitive approach to evaluating the importance of every element of a decision through pairwise comparison. The AHP is best suited for multi-criteria problems for which it is not possible to accurately quantify the impact of each of the alternatives [9]. For this reason, we use AHP to calculate the weight of each index. The results are shown as Table 1 and Table 2. 3. Fuzzy comprehensive evaluation model for credit risk of online trading company The evaluation of credit risk involves many factors. What is more, there are abounding uncertainty factors and dynamic variables with high fuzziness. An assessment model for credit risk of online trading company is established in this article by applying fuzzy mathematics theory. 3.1. Fuzzy comprehensive evaluation model Multi-hierarchy fuzzy comprehensive evaluation is implemented in the following steps: (1) Determine the Evaluation set P = {p1, p2,, pm}, m is the number of evaluations. In this article, m = 5, p1 = very high, p2 = high, p2 = moderate, p2 = low, p2 = very low. And provide that P1 = 1, P2 = 8, P3 = 6, P4 = 4, P5 = 2. (2) Let the Index set U (R in the online evaluation index system) in grade one be expressed as U = {U1, U2,, Uk}, k is the number of indexes in grade two which are included in U. Then separate Ui (i = 1, 2,, k) into n sub-evaluation indexes, Ui = {Ui1, Ui2,, Uin}, n is the number of indexes in grade three which are included in Ui. Similarly, Uij (j = 1, 2,, n) can be further separated into l subsets: Uij = {Uij1, Uij2,, Uijl}, l is the number of indexes in grade four which are included in Uij. (3) Determine the weight of evaluation indexes. Define the Weight set W as W = {W1, W2,, Wk}. Similarly, Wi and Wij can be expressed as: Wi = {Wi1, Wi2,, Win}, i = 1, 2,, k. Wij = {Wij1, Wij2,, Wijl}, j = 1, 2,, n. which meet the conditions of k W i 1 i n W j 1 ij l W t 1 ijt 1 1 1 (4) Build the fuzzy evaluation matrix of indexes. Constructing the membership function for quantitative index and using the expert assessment method for qualitative index to determine the 14
membership. For example, if the index set U has four grades, the membership of the index Uijt relative to P is: fijt1 fijt 2 fijtm Fijt p1 p2 pm Therefore, the fuzzy evaluation matrix of Uij is: Fij1 Fij 2 Fij Fijt fij11 f ij 21 f ijt1 f ij12 f ij 22 fijt 2 f ij1m fij 2 m fijtm (5) Use the fuzzy synthesis algorithm. Firstly, fuzzy transformation is made to get the result of index Uij relative to P: Bij = Wij Fij = (bij1, bij2,, bijm) Then, the fuzzy evaluation matrix of Ui is: Bi = (Bi1, Bi2,, Bij)T Similarly, the result of index Ui relative to P is: Ai = Wi Bi = (ai1, ai2,, aim) The fuzzy evaluation matrix of U is: A = (A1, A2,, Ai)T Lastly, the result of index U relative to P (i.e. the result of fuzzy comprehensive evaluation) is: Z= W A= (z1, z2,, zm) (6) The credit evaluation score is: P = Z PT 3.2. The determination of the weight Expert Assessment Method and Analytical Hierarchy Process (AHP) are used to get the weight of the index. Firstly, the experts score the relative importance of each index belonging to the same upper index to construct the judgment matrix, and then determine the weight of each index. For example, the business management index in offline evaluation index system includes the manager s ability, enterprise organizational structure and production control these three sub-indexes. Experts gave the scores: (1) importance of the manager s ability on enterprise organizational structure is 1; (2) importance of the manager s ability on production control is 2; (3) importance of enterprise organizational structure on production control is 2. The judgment matrix is: 1 2 1 2 / 5 2 / 5 2 / 5 6 / 5.4 Normalized Add the numbers of each line respectively, Normalized 6 / 5.4 1 2 2 / 5 2 / 5 2 / 5 obtain the weight of the indexes 1 1/ 2 1/ 2 1 1/ 5 1/ 5 1/ 5 6 / 5.2 Then, the weight of the manager s ability, enterprise organizational structure and production control is.4,.4 and.2 respectively. 3.3. The determination of the membership In the offline credit risk index system (see Table 1), financial position index includes nine subindexes in grade four. These financial indexes are quantitative and can be obtained from the company. Table 3 shows the trade industry standard which specifies the performance evaluation of China s enterprises. There are five standard values corresponding to five evaluations. Obviously, these five evaluations are as same as the Evaluation set P defined in this article. Therefore, we can use the membership calculation method of the minimum-optimum index proposed by Zadeh [11] to get the membership. 15
Table 3. A part of the performance Evaluation of China's enterprises- Trade Industry Standard [12] Indexes Very high High Moderate Low Very low Rate of Return on Common Stockholders Equity Return on Sales Asset-liability ratio Quick ratio Interest coverage Turnover of total capital Turnover ratio of receivable Sales growth rate Rate of capital accumulation 16.2 23.3 39.4 138.4 1.7 3.1 12.7 24.9 26.7 12.2 18.4 51.9 114.3 7.7 2.4 8 14.7 17.5 8.4 13.9 61.3 93.3 4.8 1.9 6 5.6 9.2 4.6 9.4 68 69.6 2.2 1.5 4.4-4.3.3 1.6 5.8 78.2 51.7 1.1 3.2-11.9-7.6 Define x1 as the standard value of Very high, and x2, x3, x4, x5 are the standard values of High, Moderate, Low, Very low respectively. Assume x is the value of a quantitative index. We can get the value of fijtw (1 w 5) according to Formula (1). x xw 1 fijtw x x, w w 1 xw x fijt ( w 1) x x, w w 1 fijt1 1, fijt 5 1, xw 1 x xw, 1 w 5 (1) xw 1 x xw, 1 w 5 x x1 x x5 For example, the Return on Sales of a company is 15.6, i.e. x = 15.6. Obviously, x 3 x x2 (see Table 3). By use of Formula (1) to get f ij22 =.38 and fij23 =.62. Except financial position index, the other indexes in the offline evaluation index system are qualitative indexes. Expert Assessment Method is used to find the membership of the index relative to the Evaluation set P. In the online credit risk evaluation index system, transaction amount R12, transaction times R13, transportation time R31 and transportation cost R32 are quantitative, and the value of these indexes can be got by the trading system. By setting the range of Pi (i = 1, 2,, 5) for different index and using statistical method, membership will be gained. For example (see Table 4), assume that 2 buyers buy goods from the company, there are 14 buyers receive goods in 24 hours, and then the membership of p1 is.7. Indexes (unit) Table 4. The range of Pi for transportation time (R31) Very high(p1) High(P2) Moderate(P3) Low(P4) Transportation time (R31), (hour) (,24] (24,72] (72,12] (12,168] Very low(p5) (168, + ) Except above four indexes, the other seven indexes are qualitative. After the transaction, the buyers give the evaluation of these indexes to the company, and then the membership will be obtained by means of fuzzy membership function which determined by statistical method. 3.4. Credit risk comprehensive evaluation As mentioned above, offline and online index system are constructed respectively, thus we can obtain two evaluation results. In order to get the comprehensive evaluation score of the online trading company, we use Formula (2): (2) P P ' P '' where in: P refers to the evaluation score calculated by offline indexes, and P refers to the evaluation score calculated by online indexes. αand βare the weight of offline index U and online index R respectively. 16
4. Representation of online trading company s credit rating According to Implementation Measures of Enterprise Credit Rating Evaluation promulgated by China Electronic Commerce Association, the enterprise credit rating is divided into three grades and five levels, i.e. AAA, AA, A, B, C. Based on the detailed description of every credit status, this article proposes the precise range of credit scores (see Table 5). Table 5. Representation of online trading companies credit rating Notation Credit score Credit status AAA AA A B C 9-1 8-89 7-79 6-69 <6 Very good Good Less good General Poor Obviously, if a company has gained 85 point after credit risk comprehensive evaluation, then this company s credit rating is AA. 5. Case analysis We choose a garment trading company X selling clothes on China garment network (www.efu.com.cn) which is a platform devoted to the garment industry B2B e-commerce services. Through data collection, the financial indexes values of the online trading company X were obtained (see Table 6). Using the method in Section 3.2., the single-factor evaluation matrixes of U 11 -U 14 are obtained and represented in Table 7. Table 6. Financial indexes value of the online trading company X Quantitative indexes Quantitative indexes Value (Grade three) (Grade three) Rate of Return on Common Stockholders Equity 11.7 Turnover of total capital Return on Sales 15.6 Turnover ratio of receivable Asset-liability ratio 46.8 Sales growth rate Quick ratio 119.22 Rate of capital accumulation Interest coverage 7.6 Value 2.2 7.5 1.3 16.5 Provided that 1 experts are invited to participate in the comprehensive evaluation, the single-factor evaluation matrixes of U 21 -U 32 are represented in Table 7. In addition, 2 buyers are invited to evaluate the online indexes, and the single-factor evaluation matrixes of R 1 -R 4 are listed in Table 8. 17
Table 7. Single-factor evaluation matrixes of U11-U32 (Offline credit risk evaluation index system) Evaluation set Grade two Grade three Grade four Very high high moderate low Very low Financial position (U1),(.57) Internal features (U2), (.29) External characteristics (U3), (.14) U11, (.45) U111, (.4) U112, (.6).87.38.13.62 U12, (.32) U121, (.41) U122, (.33) U123, (.26).26.2.74.8.97.3 U13, (.13) U131, (.5) U132, (.5).6.75.4.25 U14, (.1) U141, (.6) U142, (.4).52.88.48.12 U21, (.29) U211, (.41) U212, (.33) U213, (.26).2.6.5.4.3.2.5.1.1.1 U22, (.5) U221, (.4) U222, (.4) U223, (.2).1.6.6.2.3.3.2.7 U23, (.21) U231, (.5) U232, (.5).5.4.5.1.3.1.1 U31, (.5) U311, (.4) U312, (.2) U313, (.4).6.4.2.3.3.3.1.3.3.2 U32, (.5) U321, (.5) U322, (.5).1.1.4.4.3.3.2.1.1 Table 8. Single-factor evaluation matrixes of R1-R4 (Online credit risk evaluation index system) Evaluation set Grade two Grade three Very high high moderate low Very low Product (R1), (.35) R11, (.45) R12, (.13) R13, (.1) R14, (.32).75.5.9.15.5.1.8.1.2 Service (R2), (.25) R21, (.3) R22, (.7).9.5.5.8.2 Transport (R3), (.23) R31, (.4) R32, (.4) R33, (.2).7.95.3.5.4.6 Overall credit evaluation (R4), (.17) R41, (.4) R42, (.6).2.5.8.75.2 According to the fuzzy comprehensive evaluation model introduced in Section 3.1, the result of index U relative to P: Z= W A.5523.71972.2255 = (.57.29.14).26152.4485.29117.3196.15.25.35.26.115.25 = (.14232.57665.24912.2537.655) In Section 2, we defined: P1=1, P2 =8, P3=6, P4 =4, P5 =2, then the credit evaluation score of offline indexes is: P = Z PT= 76.46 In the light of Table 4, the offline credit rating of company X is A. 18
At the same time, the result of index R relative to P: Z= W A.235.58.159.26 = (.35.25.13.17).27.15.575.14.66.14.8.12.11.77.12 = (.25425.35585.2198.545.156) The credit evaluation score of offline indexes is: P = Z PT= 69.57 Thus the online credit rating of company X is also B. Assume that α=.5, β=.5, then the comprehensive evaluation score of online trading company X is: P P ' P ''.5 76.46.5 69.57 73.2 Finally, the comprehensive evaluation rating of online trading company X is A. Obviously, the offline credit of company X is not as same as online credit, and through this comprehensive evaluation method company X gains a higher credit rating than only using online evaluation. This credit rating is more in line with the real situation of company X. 6. Conclusions Previous studies usually ignored important offline credit of the company when they did ecommerce credit evaluation. Sometimes many companies have good online credit, but they maybe face credit crisis because of their serious financial problems. So we need to consider both online and offline credit risk. The comprehensive evaluation model presented in this article can objectively assess the credit risk of online trading company, because the model consists of offline and online credit evaluation index systems. We use expert assessment method and AHP to determine the indexes weights and use the multi-hierarchy fuzzy comprehensive evaluation method to calculate the credit evaluation score. Based on the credit score, we can obtain the online trading company's standard credit rating. Finally, case analysis illustrates our credit risk comprehensive evaluation method is more effective and comprehensive than [8]. Surely, the reasonability of the indexes and their weight in the offline and online evaluation systems has to continually face up to tests and judgments during application. Further study should be made on how to automatically identify online malicious evaluation. 7. Acknowledgement The research is supported by the Beijing Forestry University Young Scientist Fund under Project No. YSE211-8. 8. References [1] Liang-Hsuan Chen, Tai-Wei Chiou, A Fuzzy Credit-rating Approach for Commercial Loans: A Taiwan Case, Omega, vol. 27, pp.47-419, 1999. [2] Yue Jiao, Yu-Ru Syau, E.Stanley Lee, Modelling Credit Rating by Fuzzy Adaptive network, Mathematical and Computer Modelling, vol. 45, pp.717-731, 27. [3] Jae H. Min, Young-Chan Lee, A Practical Approach to Credit Scoring, Expert Systems with Applications, vol. 35, pp.1762-177, 28. [4] Petr Hajek, Municipal Credit Rating Modelling by Neural Networks, Decision Support Systems, vol. 51, pp. 18-118, 211. [5] Wang Yi, Xia Huo-Song, Liu Jian, Commercial Credit Difference Evaluation and Prediction Model: Based on Neural Network, JCIT: Journal of Convergence Information Technology, vol. 5, no. 9, pp. 257-266, 21. 19
[6] Chen Lin, Zhou ZongFang, A Measure on Joint Default Risk Based Credit Rating Information and Combinatorial Copula Function, JCIT: Journal of Convergence Information Technology, vol. 4, no. 1, pp. 39-46, 29. [7] Guo Yi-han, Zheng Zhi, Research on Credit Evaluation Model for C2C E-commerce Website, Journal of Beijing University of Posts and Telecommunications (Social Sciences Edition), vol. 13, no. 4, pp. 21-25, 211. [8] Yingtao Xu, Ying Zhang, A Online Credit Evaluation Method Based on AHP and SPA, Communications in Nonlinear Science and Numerical Simulation, vol. 14, pp.331-336, 29. [9] Lizhi Wu, Aizhu Ren, Research on Urban Fire Risk Comprehensive Evaluation and Its Applications in China, Human and Ecological Risk Assessment, vol. 15, pp. 778-788, 29. [1] T. L. Saaty, L. G. Vargas, Models, Methods, Concepts & Applications of the Analytic Hierarchy Process, Kluwer Academic Publishers, Boston, MA, USA, pp. 1-13, 21. [11] L. A. Zadeh, Quantitative Fuzzy Semantics, Information Sciences, vol.3, pp.159-176, 1971. [12] State-Bureau of financial supervision and appraisal (S-BFSA). Performance Evaluation Standard of China's enterprises, Economic Science Press, Beijing, pp.387, 211 11