【作者】 韦宁；

【导师】 戴晓凤；

【作者基本信息】 湖南大学 ， 金融学， 2008， 硕士

【摘要】 度量信用风险的模型根据其信息源的不同分为两类——会计信息信用风险模型和市场信息信用风险模型。虽然两类模型的信息源不同,评价方法也大相径庭,但是它们都是针对相同的评价主体进行信用状态的辨识,因此两者应当以相似或相近的结论共同反映出评价主体的信用状态变化。它是模型间比较研究的基础。然而,近来国内许多相关研究中出现了相互矛盾的结论,这直接挑战了上述两类模型间的根本联系。为了证明这一内在联系的现实性,本文首先系统的讨论了两类信用风险模型一致性问题的内涵及分析框架。其次,分别利用代表两类模型的logistic和KMV对288家上市公司的进行信用状态的度量与排序。最后,利用两个模型评价的序数结果在面板数据多维信息空间中进行层级性和趋势性上的一致性判研。结果显示,时序上两类模型序数结果一致,都能反映出评价主体自身信用状态的变化趋势;而截面上的层级序数结果则无法证明两者相关性的存在。可见,评价主体间信用状态评价结果的差异不仅仅只是评价效率上的区别,还存在着其他限制性因素破坏了内在的联系。更多还原

【Abstract】 According to various information sources, there are two different models measuring credit risks, accounting credit risk model and marketing credit risk model. They are entire diverse models using different information sources and estimation approaches; however, they are identification models of risks based on the same estimation object, which is the fundamental of comparative study among models. Nevertheless, those emerging antinomies from domestic research are directly challenged the internal logic between the two models foresaid.In order to prove the consistency of internal logic connection of accounting credit risk model and marketing credit risk model, this dissertation systematically discussed the definitions and analysis framework of consistency problem of two kinds of credit risk models at first. Then, to measure and rank the credit estate of 288 listing companies, it adopted LOGISTIC and KMV model, which are dealt with different credit risks. Lastly, we focused on distinguishing the consistency of ranking and trend by the ordinal results in the pool data of models.The conclusion reveals that the two kinds of credit risk models have the same ordinal results from time series perspective, as well as they reflect the changing trends of estimation object’s credit statement, but the consistency from cross-section. Due to the various models’efficiencies and other restrict factors existing, the different credit risk models led to diversity occurring between credit estates of different estimation objects.更多还原