粗糙集理论在银行信贷风险评估中的应用研究

发布时间：2018-03-26 13:12

本文选题：德尔菲法　切入点：粗糙集理论　出处：《华南理工大学》2015年硕士论文

【摘要】：作为拥有巨额资本、海量客户的银行业来说,为社会上主要资金的流通起着重要作用,是现代金融体系中重要的组成部分之一。对我国银行现状来说,贷款业务是其最重要的业务之一,银行的利润大部分来自贷款业务。近几年在政策大力支持和市场巨大需求下银行信贷投放量不断上升,信贷风险也不断攀高。实际上,对信贷风险处理如稍有不慎,便有可能造成银行经营危机甚至破产,多米诺骨牌效应的影响下信贷风险可能越演越烈,从而造成整个金融体系的瘫痪。因此,在贷款过程中如何加强对信贷风险进行管控不仅成为竞争日趋激烈下银行经营管理的重要组成部分,而且也关系到我国整个金融市场的发展。信贷风险评估属于银行信贷风险管控体系中的一部分,如果信贷风险不能得到科学合理的评估,那么就谈不上随之的风险应对、监控等。通常银行在过去实践当中,对信贷风险评估过多的依据借款人的财务因素,而忽视了非财务因素的影响。本文首先在前人研究的基础上归纳了信贷风险评估指标,通过德尔菲法对初选指标进行统计分析,从而构建了较为完善的指标体系,为后面的评估研究作了良好的铺垫。接下来运用粗糙集理论建立了信贷风险评估模型,在分析比较每个步骤通用方法的基础上选用了适合信贷风险评估的方法,针对属性约简,介绍了常用的约简算法,鉴于信贷风险评估对象数据量较大且指标较多的情况,引入了遗传算法进行属性约简,在结合粗糙集理论基础上利用改进的遗传算法进行研究,通过实例分析结果显示改进启发式遗传算法能够加快收敛速度并且相对于一般属性约简遗传算法能够节省运行时间,能够较好地应用于信贷风险评估时的属性约简。最后本文选取了我国沪深市场的上市公司数据进行实证分析,实证结果显示基于粗糙集理论的信贷风险评估准确度较高,能够在一定程度上为银行信贷人员在进行信贷决策过程中提供参考。
[Abstract]:As a banking industry with huge capital and huge customers, it plays an important role in the circulation of major funds in the society and is one of the important parts of the modern financial system. Loan business is one of its most important business, and the profits of banks come mostly from loan business. In recent years, under the strong policy support and the huge market demand, the amount of bank credit has been increasing, and the credit risk is also rising. In fact, If there is a slight carelessness in the handling of credit risk, it may result in a banking crisis or even bankruptcy. Under the influence of the domino effect, the credit risk may become more and more severe, resulting in the paralysis of the entire financial system. How to strengthen the credit risk control in the process of loan is not only an important part of the bank management under the increasingly fierce competition. It is also related to the development of the whole financial market in China. Credit risk assessment is a part of the credit risk control system of banks. If the credit risk cannot be scientifically and reasonably assessed, then there is no corresponding risk response. In the past, banks usually overevaluate the credit risk based on the borrower's financial factors, but ignore the influence of non-financial factors. Firstly, this paper summarizes the credit risk assessment indicators based on the previous studies. Through the statistical analysis of the primary index by Delphi method, a perfect index system is constructed, which makes a good foreshadowing for the later evaluation research. Then, the credit risk assessment model is established by using rough set theory. On the basis of analyzing and comparing the general methods of each step, the suitable method for credit risk assessment is selected. Aiming at attribute reduction, the commonly used reduction algorithms are introduced. In view of the large amount of data and the large index of credit risk assessment object, In this paper, genetic algorithm is introduced to reduce attributes. Based on rough set theory, improved genetic algorithm is used to study. The results of example analysis show that the improved heuristic genetic algorithm can accelerate the convergence speed and save the running time compared with the general attribute reduction genetic algorithm. Finally, this paper selects the data of listed companies in Shanghai and Shenzhen market for empirical analysis. The empirical results show that the accuracy of credit risk assessment based on rough set theory is high. To a certain extent, can provide a reference for bank credit personnel in the credit decision-making process.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F832.4

【参考文献】