基于GMDH模型和主成分logistic模型的信用风险评估方法研究

【作者】 郭富霞；

【导师】 高齐圣；

【作者基本信息】 青岛大学 ， 金融学， 2012， 硕士

【摘要】 信用风险是银行面临的最主要风险,是银行风险管理的核心内容。准确度量、有效管理信用风险,是商业银行稳定运行的重要保障。GMDH是一种比较新颖的建模方法,适应于复杂非线性系统辨识及预测。采用GMDH方法建立的预测模型能较好地描述各因素间的复杂非线性关系,在建模过程中变量的层层筛选保证了计算的收敛速度,使得主观因素影响减少,增强了客观性：并且采用数据分组,避免了传统方法中用同一组数据建模和检验所产生的检验精度高但适应性差的不良现象。本文以上市公司违约概率作为商业银行信用风险高低的衡量标准,利用我国上市公司的财务数据,结合GMDH方法和主成分LOGISTIC模型构造了商业银行信用风险评估模型。实证研究结果表明,把GMDH方法引入银行信用风险评价问题当中,将一个多元判别模型通过简单的二元多项式模型自组织生成,大大减小了对样本数据量的要求。相对主成分LOGISTIC模型,GMDH模型预测风险能力和推广能力较强,能够为商业银行信用风险程度的判定提供客观依据。更多还原

【Abstract】 Credit risk in the financial markets is the most ancient kind of risk.lt is also one of the basic risk which the bank faces. The bank plays an important part in the field of finance of China. How to measure the commercial bank credit risk in an effective and scientific way is the most important problems.The GMDH method is adopted to establish the prediction model, which can describe the complex non-linear relationship among the various factors.To avoid the traditional method using the same set of data modeling and verification of generated testing high precision,we use GMDH model.Aimed at the credit risk of commercial banks, taking the listed companies’default rate as a measure, a model for credit risk evaluation was built combined with principal component analysis and Logistic model.We also use GMDH model to solve the promble. Results show that GMDH model can effective predicts credit risks of the listed company better than Logistic model.更多还原