基于组合赋权的供应链应收账款融资BP风险评价模型

发布时间：2018-04-20 07:42

本文选题：供应链融资 + 应收账款融资　；参考：《安徽工业大学》2017年硕士论文

【摘要】：供应链金融是以解决中小企业融资难为目标的创新性金融产品,有效缓解了中小企业融资难融资贵。供应链金融的核心内容供应链融资发展非常迅速,供应链融资主要包括应收账款融资、融通仓融资和保兑仓融资三种模式。时常发生的供应链融资骗贷事件不仅造成难以估量的经济损失和社会影响,也给供应链融资业务带来巨大的负面效应,同时在供应链融资的发展过程中存在实践应用领先于理论研究的情形,供应链融资风险研究较为薄弱。因此,本文以供应链应收账款融资为研究对象,主要从风险评价指标体系建立、评价指标权重衡量以及评价模型推广应用三个方面进行研究。具体内容如下:(1)建立供应链应收账款融资风险指标体系。通过广泛阅读相关文献和咨询行业专家初步得到5个一级指标,11个二级指标,42个具体指标的指标体系;基于Wind中国企业库和咨询企业管理人员收集51条供应链应收账款融资数据;利用stata软件对51条样本数据进行相关性分析和敏感性分析分别删除9个和7个具体指标,最终得到5个一级指标和26个具体指标的指标体系。(2)确定供应链应收账款融资风险指标的组合权重。为保证指标权重的优越性,本文采用主客观组合赋权法确定融资风险指标权重。主观赋权法收集七位行业专家的指标权重打分,利用yaahp软件计算专家的AHP权重,matlab计算FAHP权重,对比结果认为FAHP的计算结果优于AHP;客观赋权法介绍熵权法及部分改进的熵权法,通过文献分析认为本文的数据预处理方法和计算较多指标权重能够克服熵值微小差异造成权重成倍波动这一问题,利用熵权法可得到较为理想的指标客观权重;主客观赋权法组合比重的确定是以风险评价值最小和最大熵原理为目标建立的比重计算模型,利用FAHP和熵权法的权重计算结果求出两种赋权法的比重并最终求得组合权重,实例验证组合权重优于赋权法单独使用;最后利用组合权重加权相乘得到51个样本的风险值。(3)建立基于BP训练权重和阈值的风险评估模型。为提高本文组合权重评估模型的便捷性,建立基于BP训练权重和阈值的风险评估模型实现优势互补。采用实验的方法确定BP训练模型的网络层次、输入层和输出层节点数、各层传递函数、训练函数、隐含层节点数、学习速率等,建立基于BP训练权重和阈值的正向计算评估模型,以组合权重风险评价值为期望输出对评估模型进行检验,检验结果表明BP评价模型完全能够达到供应链应收账款融资风险评定的误差要求,可在今后评价中小企业应收账款融资的过程中推广应用。本文所构建的供应链应收账款BP风险评价模型能够较好地衡量供应链应收账款融资行为的风险水平,为金融机构开展此项业务提供一定的参考依据。
[Abstract]:Supply chain finance is an innovative financial product aimed at solving the financing difficulties of SMEs, which effectively alleviates the difficulty in financing SMEs. The core content of supply chain finance is the rapid development of supply chain financing. Supply chain financing mainly includes accounts receivable financing, finance warehouse financing and confirmed warehouse financing. The frequent supply chain financing fraud not only causes incalculable economic loss and social impact, but also brings huge negative effects to the supply chain financing business. At the same time, in the process of the development of supply chain financing, the practical application is ahead of the theoretical research, and the research of supply chain financing risk is relatively weak. Therefore, this paper takes the supply chain accounts receivable financing as the research object, mainly from the risk evaluation index system establishment, the evaluation index weight measurement and the evaluation model promotion application three aspects carries on the research. The concrete contents are as follows: 1) establishing the index system of supply chain accounts receivable financing risk. Through extensive reading of relevant literature and consulting industry experts get 5 primary indicators, 11 secondary indicators, 42 specific indicators of the index system; Based on the Wind Chinese enterprise bank and consulting enterprise managers collect 51 supply chain accounts receivable financing data, using stata software to carry out correlation analysis and sensitivity analysis of 51 sample data to delete 9 and 7 specific indicators, respectively. Finally, the index system of 5 primary indexes and 26 specific indexes is obtained) to determine the combined weight of supply chain accounts receivable financing risk index. In order to ensure the superiority of index weight, this paper adopts subjective and objective combination weighting method to determine the index weight of financing risk. The subjective weighting method collects the index weights of seven industry experts and calculates the AHP weights of the experts by using yaahp software. The comparison results show that the calculation results of FAHP are superior to those of AHP, and the objective weighting method introduces the entropy weight method and the partially improved entropy weight method. Through the literature analysis, it is concluded that the method of data preprocessing and the calculation of the weight of more indexes can overcome the problem of multiplying the weight caused by the slight difference in entropy value, and the objective weight of the index can be obtained by using the entropy weight method. The determination of combined weight of subjective and objective weighting method is based on the principle of minimum and maximum entropy of risk evaluation. By using the weight calculation results of FAHP and entropy weight method, the weight of two weighting methods is obtained and the combined weight is finally obtained. Finally, a risk assessment model based on BP training weight and threshold is established by multiplying the combined weights to get the risk value of 51 samples. In order to improve the convenience of the combined weight evaluation model, a risk assessment model based on BP training weight and threshold is established to achieve complementary advantages. The network layer, input layer and output layer node number, transfer function, training function, hidden layer node number, learning rate of BP training model are determined by experimental method. A forward calculation evaluation model based on BP training weight and threshold is established, and the risk evaluation value of combined weight is used as the expected output to test the evaluation model. The test results show that the BP evaluation model can completely meet the error requirements of the risk assessment of the supply chain accounts receivable financing, and can be widely used in the process of evaluating the accounts receivable financing of small and medium-sized enterprises in the future. The BP risk evaluation model of supply chain accounts receivable constructed in this paper can better measure the risk level of supply chain accounts receivable financing behavior and provide a certain reference for financial institutions to carry out this business.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;F275

【参考文献】