关键词: 供应链金融, 违约风险, 信用风险评价, 神经网络

Abstract:

Interest margin, the traditional source of income for commercial banks, has become less and less as the competition environment becomes more severe, and the supply chain financial model emerges as the times require. Develop a supply chain financial model that allows banks to more directly serve small and medium-sized enterprises(SMEs) with difficulty in financing in the supply chain, drive the flow of products, funds, and information throughout the supply chain, promote the interactive development of capital and industry, and build an industrial ecology of mutually beneficial, sustainable development, and healthy interaction among banks, enterprises, and commodity supply chains. However, as banks join the supply chain, the credit status of core enterprises and SMEs will be enlarged due to the connection of the supply chain, and the credit risk will also be expanded, and even be transmitted to the entire supply chain. Therefore, the key to the success of supply chain financial business is to further establish the credit risk evaluation system of SMEs in line with the characteristics of supply chain finance, strengthen the corresponding risk management in combination with the sources of risk, and effectively control the risk.Based on the generally recognized 5C principles in the industry and the well-organized domestic and foreign classic literature, combined with the actual characteristics of the supply chain business, this paper establishes a credit risk evaluation system that includes both the credit risk of the credit subject and the debt credit risk. On the one hand, in view of the fact that the existing supply chain financial evaluation indicator system rarely involves the credit risk factors in the supply chain, this paper makes a comprehensive analysis of the cooperation, collateral and overall operation of the supply chain of SMEs, which is more suitable for the supply chain. On the other hand, the previous literature paid too much attention to the impact of quantitative financial indicators on credit risk, ignoring the impact of other qualitative indicators and non-financial indicators. This paper has a good combination of the two types, while retaining certain financial indicators, other qualitative indicators and non-financial indicators that may have impact are added to make the indicator system more complete. This paper finally establishes a selection system of credit risk evaluation indicators for SMEs in supply chain finance. This system consists of 4 criteria levels: applicant qualification, counterparty qualification, asset status under financing and supply chain operation, 14 secondary indicators and 127 tertiary indicators.Taking into account factors such as the limitation of supply chain development level in China, and the development degree of supply chain in different industries, this paper selects the 2014-2018 years with comprehensive and complete effective data disclosure, the equipment manufacturing industry with obvious supply chain characteristics, long business development time and relatively complete and mature overall service as the empirical objects. After screening, 940 supply chain samples are formed for equipment manufacturing enterprises listed on SME Board of Shenzhen Stock Exchange in 2014-2018. This paper defines the risk factor by comparing the with-interest debt rate with the lower value of the Standard Value of Corporate Performance Evaluation. Enterprises that have higher with-interest debt rate than the lower value are identified as risky enterprises. The proportion of enterprises with credit risk is 22.45%.According to the idea of eliminating redundant information, the first screening is performed using partial correlation-variance analysis. Under the same criteria level, the indicators reflecting duplicate data information and economic meaning information with less ability to distinguish credit risk are deleted. A total of 64 indicators are deleted for indicators with a partial correlation coefficient greater than 0.6 and a smaller F value. According to the principle of overall risk factor identification optimization, the indicators group with the greatest ability of default identification is selected by gradually deleting the indicator with the greatest reduction of overall accuracy of the model through stepwise neural network, and a total of 15 indicators are deleted. In the end, a credit risk evaluation indicator system for supply chain finance with 48 indicators that significantly distinguished SME risk factors is established. The empirical results show that compared with the one-time deletion of neural network, the stepwise neural network deletion method has a great improvement in the accuracy rate of enterprises with no credit risk, enterprises with credit risk and all enterprises. Especially for the enterprises with credit risk that are more concerned by the banking industry, the accuracy rate has increased by 14.21%, reaching 84.83%. For banks, they can more accurately and comprehensively grasp the credit risk of SMEs, reduce the uncertainty of loans, improve their own profits, and control the loss of default. For SMEs, accurately identifying their own risks under supply chain finance is also conducive to giving full play to the advantages of the supply chain. With the help of the characteristics of supply chain finance, they can obtain better loans for themselves and more reasonable loan pricing and loan limit, so as to alleviate their "financing difficulties".The study found that the credit risk evaluation indicator system constructed by taking into account factors such as the profitability and solvency of the core enterprises in the supply chain, the degree of transaction between SMEs and the core enterprises, the external environment faced by the supply chain is more representative. The credit risk evaluation indicator system for SMEs in the supply chain financial environment established through the screening of the optimal principle of risk factor identification has an accuracy of 90.53% in discriminating enterprise risk factors, which is significantly higher than that of other models, and the discrimination effect is better. The credit risk evaluation indicator system constructed in this paper reflects the characteristics of credit risk of SMEs in supply chain finance, and can be applied in the process of bank evaluation of credit risk of SMEs.

Key words: supply chain finance, default risk, credit risk evaluation system, neural network