我国农村中小银行信用评级模型的评价与构建

发布时间：2018-07-21 13:20

【摘要】：随着我国金融体制改革的深化及市场化程度的加深,农村中小银行作为县域广大小微客户的金融服务提供者,必然在经济转型过程中扮演重要角色。然而,与之相对的是目前农村中小银行不良贷款率偏高,资产质量不高,同时缺乏对客户违约风险进行有效识别和预测的方法和工具,本文以农村中小银行信用风险测度及评级体系建立为切入点,探索我国农村中小银行信用风险管理方法。首先,本文在系统梳理实践和理论中涉及的各类小微客户信用评价指标的基础上,建立科学完善的小微客户信用指标体系库,并基于此构建农村商业银行个体工商户信用评价指标体系。其次,本文通过对我国农村中小银行信用风险度量以及信用评级方法进行研究,结合目前国内外研究前沿,在对现有模型进行适用性分析的基础上探索适合我国农村中小银行的信用风险度量模型,并在此基础上尝试构造信用风险度量的基于神经网络的logistic模型的混合模型。根据本文构建的个体工商户信用评价指标体系,采用农商行实际小微客户信贷交易及授信资料数据对logistic模型、神经网络方法以及混合模型进行检验,证实了创新构造的混合模型具有预测精确性高以及预测稳定性强两方面的优越性。最后,本文基于神经网络的logistic模型的混合模型的预测结果,开发并设计了相应的农村商业银行个体工商户信用评分卡,同时从预测精度以及稳健性两个方面验证了信用评分卡的效力。本文的研究不仅为开发适用于我国农村商业银行的违约风险估计工具和信用评级体系提供了理论依据,更具有很强的实践意义,将有助于指导我国农村商业银行提升信用风险的管控能力和水平。
[Abstract]:With the deepening of China's financial system reform and the deepening of marketization, small and medium-sized rural banks, as the financial service providers of the vast number of small and micro customers in the county area, are bound to play an important role in the process of economic transformation. However, at present, the non-performing loan ratio of small and medium-sized rural banks is on the high side, the quality of assets is not high, and there is a lack of methods and tools to effectively identify and predict the default risk of customers. Based on the establishment of credit risk measurement and rating system of rural small and medium-sized banks, this paper explores the methods of credit risk management in rural small and medium-sized banks in China. First of all, on the basis of systematically combing all kinds of small and micro customer credit evaluation indexes involved in practice and theory, this paper establishes a scientific and perfect small and micro customer credit index system database. Based on this, the index system of individual commercial bank credit evaluation is constructed. Secondly, this paper studies the credit risk measurement and credit rating methods of rural small and medium-sized banks in China, combining with the current research frontier at home and abroad. Based on the analysis of the applicability of the existing models, this paper explores the credit risk measurement model suitable for the rural small and medium-sized banks in China, and then attempts to construct a hybrid model of logistic model based on neural network for credit risk measurement. According to the credit evaluation index system of individual industrial and commercial households in this paper, the logistic model, neural network method and mixed model are tested by using the data of credit transaction and credit data of small and micro customers of agricultural and commercial banks. It is proved that the hybrid model with innovative structure has the advantages of high prediction accuracy and strong predictive stability. Finally, based on the prediction results of the mixed model of logistic model based on neural network, the corresponding credit score card of individual industrial and commercial households in rural commercial banks is developed and designed. At the same time, the effectiveness of credit scoring card is verified from two aspects: prediction accuracy and robustness. The research in this paper not only provides a theoretical basis for the development of default risk estimation tools and credit rating system for rural commercial banks in China, but also has a strong practical significance. It will be helpful to guide the rural commercial banks to improve their credit risk control ability and level.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.35;F224

【参考文献】