信用评价中区分度问题与对策研究

发布时间：2018-09-08 12:48

【摘要】：信用评价本质上应该是帮助投资者消除或者减少不对称信息的机制,从而为其提供具有投资价值和风险预警价值的参考信息。但是从近些年来的金融危机和欧洲债务危机来看,信用风险评估已极大地引起人们重视,人们越来越质疑信用评级机构所提供的信用信息的前瞻性与参考价值。信用评价不仅应如实反映受评对象信用状况,还应方便市场参与者进行区分各个受评对象,为其提供有效的参考信息,帮助其做出正确的投资决策,这就要求信用评价具有一定的区分性。如何实现有效区分将是本文主要的研究问题。本文是在深入剖析现有信用评价体系基础上,发现了两类区分度问题,因此本文的主要研究内容如下：首先,当区分多个信用受评对象时,尤其是对信用水平近似的多个受评对象进行排序或者抉择时,评价过程中存在的指标对于区分受评对象评价结果时可能被认为是不起作用的,或者说是其不具有一定的区分性。为了很好的区分受评对象,我们将对该缺乏区分性的指标进行筛选,从而起到准确区分受评对象的作用。如何实现有效筛选过程,将是本文研究的主要问题之一。文章运用了熵理论的指标筛选方法,并引入指标区分度的概念,从定量分析的角度对指标体系进行了优化,从而起到了区分受评对象的作用。文中通过高新技术企业信用评价为例,对高新技术企业指标体系进行了优化与筛选,从而为区分所选高新技术企业信用水平起到了一定的优化作用。其次,由于评级机构通常采用线性方法处理原始指标数据,这常常会导致指标得分区分度不足的问题,受评对象具有区分性的指标得分能在很大程度上影响最终评级结果的区分性。对此本文提出运用S形曲线模型处理的方法,并通过实证与常用的线性方法进行比较分析,S形曲线法比传统线性处理方法能更好体现各受评对象在指标值上的区分度,同时,扩大了最终评价结果的差异度,使各等级的受评对象得到合理的信用得分。另外,S形曲线处理指标的方法也是对现有处理指标得分方法的一个重要补充。
[Abstract]:Credit evaluation should essentially be a mechanism to help investors eliminate or reduce asymmetric information, so as to provide reference information with investment value and risk early warning value. Credit evaluation should not only reflect the credit status of the assessed objects truthfully, but also facilitate market participants to distinguish the assessed objects, provide effective reference information for them, and help them make correct investment decisions, which requires credit evaluation to have a certain distinction. How to achieve effective distinction will be the main research issue in this paper. This paper is based on an in-depth analysis of the existing credit evaluation system, found two types of discrimination problems, so the main research contents of this paper are as follows:
Firstly, when differentiating multiple credit evaluation objects, especially when sorting or choosing multiple evaluation objects with similar credit level, the existing indicators in the evaluation process may be considered ineffective or indistinguishable for distinguishing the evaluation results of the evaluated objects. As an example, we will screen the indices which lack distinction, so as to distinguish the evaluated objects accurately. How to realize the effective screening process will be one of the main problems in this paper. In this paper, the index system of high-tech enterprises is optimized and screened by taking credit evaluation of high-tech enterprises as an example, so as to play a certain role in optimizing the credit level of selected high-tech enterprises.
Secondly, because the rating agencies usually use linear method to deal with the original index data, this often leads to the problem of insufficient score differentiation. The score of the target with distinction can greatly affect the distinction of the final rating results. Compared with the traditional linear method, the S-curve method can better reflect the distinction of the target value of each evaluation object. At the same time, it enlarges the difference of the final evaluation result, and makes the evaluation object of each grade get a reasonable credit score. An important supplement to the method of scoring indicators.
【学位授予单位】：北京化工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F830.59

【参考文献】