基于违约风险判别的小型工业企业信用评级研究

发布时间：2018-03-18 08:15

本文选题：信用评级　切入点：违约风险　出处：《大连理工大学》2016年博士论文　论文类型：学位论文

【摘要】：信用评级的本质是揭示贷款数据与违约风险的关系与规律,确定一笔贷款或债务违约的可能性。由于小型工业企业规模小,财务信息不完善,很难找到经典的指标和信用评价理论进行评价,因此建立一套适用于小型工业企业的信用风险评级体系是金融机构亟需解决的问题。基于违约风险判别的小型工业企业信用风险评级研究包括小型工业企业信用评级指标体系的构建、小型工业企业信用评分模型的构建、以及小型工业企业信用等级划分模型的构建三部分内容。一是小型工业企业信用评级指标体系的构建是指根据指标对违约状态鉴别精度的影响程度对指标进行遴选,构建既能显著区分小型工业企业违约状态、又反映小型工业企业客户清偿能力的信用风险评级指标体系。二是小型工业企业信用评分模型的构建是指根据非违约企业的评价得分越高、违约企业的评价得分越低的基本评价思路,建立多目标非线性规划模型对遴选出的评价指标进行组合赋权,建立信用评分模型,求解不同小型工业企业的信用评分。三是小型工业企业信用等级划分模型的构建是指以信用差异度和违约金字塔为标准,构建非线性规划模型划分信用等级,使信用等级划分结果不仅能满足信用等级越高、违约损失率越低的违约金字塔标准,还能保证信用状况差异大的客户划分为不同信用等级。本论文共分为六章。第一章是绪论；第二章是基于违约风险判别的信用评级理论基础；第三章是基于Fisher判别的小型工业企业信用评级指标体系的构建；第四章是基于最大违约鉴别能力组合赋权的信用评分模型的构建；第五章是基于信用差异度最大的信用等级划分模型研究；第六章是结论及展望。本论文的主要工作及创新如下：(1)信用等级划分方面的工作及创新：一是根据第七个信用等级中最后一个样本的信用评分P朋,与第n1个信用等级中第一个样本的信用评分P1k+1确定相邻两个等级的信用评分差值,以所有信用等级的评分差值之和∑(Pmkk-P1k+1)最大为目标函数,保证信用状况差异大的客户划分为不同信用等级。二是以信用等级由高到低的违约损失率严格递增为约束条件建立信用等级划分模型,保证信用等级划分结果满足信用等级越高、违约损失率越低的违约金字塔标准,避免出现信用等级很高、违约损失率反而不低的不合理现象。(2)指标组合赋权方面的工作及创新：通过以非违约企业的指标加权数据到正理想点的距离代数和最小为第一个目标函数,以违约企业的指标加权数据到负理想点的距离代数和最小为第二个目标函数,构建多目标非线性规划模型进行组合赋权,在满足了“非违约企业的评价得分越高、违约企业的评价得分越低”要求的目标下得到最优的组合赋权的权重系数,使赋权结果保证了评级模型能够将违约企业与非违约企业最大地区分开。改变了现有研究的组合赋权脱离评价目的的弊端,改变了现有研究中违约与非违约企业的评价得分存在大量重叠、对两类企业的区分能力低的弊端。(3)指标遴选方面的工作及创新：根据有、无特定指标两种状态下、Fisher判别对违约状态鉴别精度的提高或降低,反映特定指标对违约状态的影响程度,剔除对违约状态的判别精度没有影响或有降低影响的指标,保留可以显著提高违约状态判别精度的指标,完善了现有研究遴选指标的标准与违约状态无关的不足。
[Abstract]:The nature of credit rating is to reveal the loan default risk and the relationship between data and rules, to determine the possibility of a loan or debt default. Because the scale of small and medium sized industrial enterprises, financial information is not perfect, it is difficult to find the index and credit evaluation of the classical theory of evaluation, therefore to establish a suitable for small and medium-sized industrial enterprises credit rating the system of financial institutions is an urgent problem to be solved. The construction of the credit risk rating of small and medium sized industrial enterprises default risk identification including small and medium sized industrial enterprises credit rating index system based on the construction of evaluation model for small and medium sized industrial enterprises, and small industrial enterprises credit rating classification model is constructed of three parts. One is the construction of small and medium sized industrial enterprises credit rating the index system is defined according to the indicators of the index of the influence degree of default identification accuracy of selection, which can not only significantly The distinction between small and medium sized industrial enterprises default, credit risk rating index system to reflect the small industrial enterprises customer solvency. The two is to build a small industrial enterprises credit scoring model refers to the non default enterprise evaluation scores higher default evaluation score lower basic evaluation idea, evaluation index system of multiobjective nonlinear programming the model of selected combination weighting, establish credit scoring model, credit for small industrial enterprises score. Three is to build a classification model of credit rating for small industrial enterprises refers to the letter of the difference and default of Pyramid as a standard nonlinear programming model is constructed based on division of credit rating, the credit rating classification results can not only meet the credit the higher the level, the default loss rate lower default Pyramid standards, but also ensure the credit status of customers is not the division of major differences with the letter Use level. This paper is divided into six chapters. The first chapter is the introduction; the second chapter is the default risk discrimination of credit rating based on the theoretical basis of; the third chapter is the construction of credit rating index system of small and medium sized industrial enterprises based on Fisher discriminant; the fourth chapter is based on the maximum default identification capability of combination weighting of the credit scoring model; the fifth chapter is the research on credit rating classification model based on the maximum credit difference; the sixth chapter is the conclusion and prospect. The main work and innovation are as follows: (1) the work and innovation of credit rating Division: one is according to a sample of seventh credit rating in credit scoring P friends, with the first sample N1 credit rating in credit scoring to determine the difference of P1k+1 credit score of two adjacent level, with all the credit rating score and sigma (Pmkk-P1k+1) as the objective function, Ensure the credit status of customers for different division of major differences in credit rating. Two is a credit rating from high to low LGD strictly increasing established credit rating classification model as constraint conditions, guarantee the credit rating classification results meet the higher the credit rating, default loss rate is lower in Pyramid to avoid the default standard, credit rating is high default, unreasonable loss rate but not low. (2) index combination weighting work and Innovation: the distance algebra positive ideal point to non default weighted index data of the enterprise to minimum as the first objective function, the default weighted index data of the enterprise to the distance from the negative ideal point and minimum algebra for the second objective functions, construct the combination weighting multi-objective nonlinear programming model, to meet the "non default enterprise evaluation score higher, lower default enterprise evaluation score "To get the optimal weighting coefficient of combination weighting requirements under the target of the weighted results ensure that the rating model can separate default and non default maximum enterprise enterprise area. Changing the combination of existing research from the drawbacks of weighting the evaluation objective, change the existing research in default and non default valuation there is a lot of overlap score. The ability to distinguish defects on two types of enterprises is low. (3) index selection work and innovation aspects: according to the index of two, no specific condition, Fisher discriminant of default state identification accuracy increase or decrease, reflecting the specific impact on the default index, excluding the default state without discrimination accuracy influence or reduce the effect of retention index, can significantly improve the classification accuracy of default index, improve the existing research on the selection index of the standard has nothing to do with the default is not enough.

【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：F832.4;F270;F425

【相似文献】