摘要
商业银行信用评级(Bank's Credit Rating)是对一家银行当前偿付其金融债务的总体金融能力的评价。对商业银行进行信用评级意义重大：第一,商业银行信用的评级结果是各国金融监管当局进行监测和控制银行业风险维护金融体系安全的根据。第二,商业银行信用评级也是银行进行自身风险管理的基础。第三,金融机构间的业务往来、合作关系的建立需要以商业银行信用评级状况作为基础。第四,工商企业和社会公众可以根据各商业银行的信用评级状况选择与其业务往来的商业银行。
本论文共分五章。第一章分析了论文的选题依据、相关研究进展、研究方法、研究的技术路线和研究内容。第二章基于非线性映射的商业银行信用评价指标体系的构建。第三章关于最优赋权方法的商业银行信用评价模型研究。第四章关于小样本问题的信用评级研究。第五章为结论与展望。论文的主要工作如下：
(1)引入非线性映射原理,构建了商业银行信用评级指标体系
根据有、无特定指标两种状态的非线性映射结果的欧氏距离,反映特定指标对评价结果的影响程度,解决了特定指标对评价结果影响程度的可量化问题,为非线性映射删除指标奠定基础。对全部指标做非线性映射,删掉特定指标再做非线性映射。求两个非线性映射结果的欧氏距离,欧氏距离无变化或变化很小表示该指标对评价结果影响小。通过设定复相关系数和相关系数同时超过阈值删除指标,避免了单一标准导致指标的误删,保证了删除指标后信息含量损失少。建立了6个准则层19个指标的商业银行信用风险评价指标体系,用17%的指标反映了91%的原始信息。实证研究表明,本研究的评价结果与穆迪、大公国际的评级结果序关系一致。保持与穆迪、大公国际的评级结果序关系一致有两个原因：一是穆迪、大公国际等权威机构的核心评级指标、评价方法以及赋权方法是不对外披露的。因此人们无法通过其方法对权威机构没有评级的商业银行进行评级。二是保证评价结果与权威机构序关系一致,既保证了评价结果的合理性又解决了可以对所有商业银行进行评级的问题。
(2)利用改进Spearman检验,建立了最优赋权方法的商业银行信用评价模型
根据权威评级机构公布的商业银行信用评级结果,利用改进的Spearman秩相关检验对不同赋权方法得到的评价结果进行检验,选择与权威机构评级结果最接近的赋权方法作为最优赋权方法,解决了现有研究无法解决的最优赋权方法确定的问题。采用了主观赋权的AHP、G1法和客观赋权的离差最大化法、变异系数法、熵值法等五种赋权方法。与权威评级机构穆迪和大公国际的评级结果进行对比检验,解决了现有研究无法解决的最优赋权方法确定的问题。实证研究表明,对于商业银行的信用评价问题,使用熵值法赋权最合适。
(3)建立了基于小样本检验——模拟的商业银行评级模型
通过对评价得分的分布进行检验,找到评价得分的分布规律,为评价得分的数据扩充提供依据,这正是本研究区别于现有研究的不同之处。根据通过分布检验的分布参数模拟生成与评价得分同分布的随机数据,扩充样本数量,使得扩充后的样本数据与原始数据具有相同的分布特征,解决小样本无法划分等级的问题。实证研究表明,中国的商业银行的评价得分既不是正态分布,也不是指数正态分布和对数正态分布,而是服从一种特殊分布,评价得分自然对数的平方服从正态分布的。通过这一分布规律,可以对商业银行的评价得分数据进行模拟扩充达到大样本要求,避免了小样本划分评价等级不准确的问题。
The commercial bank credit rating is one kind of assessment of the total financial capacity that some bank can pay its financial debts currently. The importance of commercial bank credit rating is as follows:Firstly, the rating results contribute to surveying and controlling banking risks and maintaining the financial system safety for the financial regulation authorities in various countries. Secondly, their own risk managements of commercial banks are on the basis of credit rating. Thirdly, the commercial bank credit rating is taken into consideration seriously during the business transaction and the buildup of cooperation relationship among the financial institutions. Fourthly, the enterprises and the public can choose the commercial banks that provide the corresponding service and operations according to the credit rating.
This dissertation is divided into five chapters. In the first chapter, the author analyzes the grounds of selected topic, relative research progress and methods, technique routes and research content. In the second one, the index system of commercial bank credit rating based on nonlinear mapping is established. In the third one, the commercial bank credit rating model of the optimal weighting is expounded. In the fourth one, small sample credit rating research is conducted. In the last one, we draw some conclusions and do some expectations. The main content is as follows:
(1) The nonlinear mapping theory is introduced and the index system of commercial bank credit rating is established. According to Euclidean distances of two different states of nonlinear mapping results with and without specific indexes, it shows how much the specific indexes have impacted on the rating results. The measurable method can contribute to nonlinear mapping'deleting indexes. All indexes are mapped nonlinearly. After the specific indexes are deleted, the nonlinear mapping is conducted. We try to obtain Euclidean distances of two nonlinear mapping results. If there is no change or little change of Euclidean distances, the indexes will have little impact on the rating results. We set that the indexes are deleted when the multiple correlation coefficient and the correlation coefficient exceed the threshold at the same time, which avoids deleting indexes mistakenly with single standard and ensures less information loss. The commercial bank credit risk evaluation index system with6criteria layers and19indexes is established, which reflects91%of original information with17%of indexes. The empirical research shows that the rating results of research have the same order relationship with ones of Moody's and Dagong Global. There are two causes of the same order relationship:Firstly, the core rating indexes, the rating methods and weighting methods aren't announced by the authorities such as Moody's and Dagong Global and so on. So it can't be rated the commercial banks which the autorities haven't rated with their methods. Secondly, this kind of same order relationship not only ensures the rationality of rating results but also rates all commercial banks.
(2) The optimal weighting commercial bank credit rating model is established with the improved Spearman test. According to the commercial bank credit rating results issued by the authoritative rating institutions, the rating results of different weighting methods are tested with the improved Spearman. We choose the weighting method which brings rating results very close to the ones of authoritative institutions as the optimal weighting. So this can solve the problem that the current researches can't fix the optimal weighting. There are five weighting methods analyzed in this research such as AMP and G1of subjective weighting and maximum deviation, variation coefficient method and entropy method of objective weighting. The rating results of Moody's and Dagong Global Credit Rating Co. are contrasted and proved; we can fix the optimal weighting problem. The empirical research shows that the entropy weighting is optimum when rating the commercial bank credit.
(3) The commercial bank rating model based on small sample testing-simulating is established. Through testing the distribution of rating scores, we can find the distribution law of rating scores and provide the basis of data amplification, which differentiate this research with current other researches. According to the distribution parameters that passed the distribution tests, we can simulate and generate the random data in step with the rating score distribution. The number of samples is expanded. So the expanded sample data and the original data have the same distribution characteristic, which makes small samples be rated. The empirical research shows that the rating scores of the commercial banks in China follow neither the normal distribution nor the exponential normal distribution and the logarithmic normal distribution, but follow a special distribution, that is, the natural logarithms' squares of the rating scores follow the normal distribution. And then we can simulate and amplify the rating score data to meet the requirements of large samples, which avoids the inaccuracy of small sample rating.
引文
[1]Derviz A., Podpiera J.. Predicting Bank CAMELS and S&P Ratings:The Case of the Czech Republic[J]. Emerging Markets Finance & Trade,2008,44(1):117-130.
[2]中国银行业监督管理委员会.股份制商业银行风险评级体系[R].2004.2.
[3]Moody's Investors Service. Global Credit Research [R].2005.6.
[4]Standard & Poor's. China Top 50 Banks [R].2007.6.
[5]大公国际资信评估有限公司.大公商业银行信用评级方法框架[R].2005.6.
[6]中国诚信国际.商业银行信用评价方法[R].2004.3.
[7]信用中国.穆迪公司对银行信用评级的财务比率指标[EB/OL]. http://www.ccn86.com/news/e school/20050923/12782.shtml,2004-9-23.
[8]Poon W. P. H., Firth M., Fung H. A multivariate analysis of the determinants of Moody's bank financial strength ratings[J]. Journal of International Financial Markets, Institutions and Money, 1999,9(3):267-283.
[9]Staub R. B., Souza G. S., Tabak B. M. Evolution of bank efficiency in Brazil:A DEA approach [J]. European Journal of Operational Research,2010,202(1):204-213.
[10]周春喜,任佳慧.商业银行信用风险综合评价研究[J].科研管理,2004,25(2)：53-58.
[11]王犁.我国商业银行信用指标体系及综合评价[J].河北工程大学学报(社会科学版),2009,6(1)：24-27.
[12]迟国泰,王际科,杜娟.基于灰色系统理论的商业银行竞争力评价模型[J].拧制与决策,2006,21(3)：346-350.
[13]迟国泰,郑杏果,杨中原.基于主成分分析的国有商业银行竞争力评价研究[J].管理学报,2009,6(2)：228-233.
[14]李建军.商业银行绩效国际比较评价体系的设计及其实证[J].金融论坛,2004(9)：34-40.
[15]杜娟.商业银行竞争力评价指标体系和方法研究[D].大连理工大学,2005年3月.
[16]张留成,王思薇.我国商业银行竞争力评价指标体系[J].黑龙江科技信息,2007(2)：84.
[17]李文军.商业银行竞争力评价模型[J].河南金融管理干部学院学报,2006(2)：70-74.
[18]段愿,宋熠东.商业银行竞争力评价指标实证研究[J].湖南税务高等专科学校学报.2007,20(1)：41-44.
[19]涂莹莹,陈玉菁.中小银行竞争力评价体系构建及实证分析[J].浙江金融,2007(6)：27-28.
[20]尹宗成,丁目佳.基于财务视角的上市商业银行竞争力评价研究[J].统计应用,2008(2)：33-35.
[21]祁建熹.银行竞争力评价方法及其指标体系的构建[J].现代商业,2008(35)：8-9.
[22]黄颖君,刘潋.我国境内上市商业银行竞争力评价及实证研究[J].现代企业教育.2007(11)：71-74.
[23]蔡亚蓉.中外商业银行竞争力评价分析[J].中央财经大学学报,2007(7)：26-30.
[24]宣丹旎.沪深上市银行竞争力评价[J].农村金融研究,2004(9)：47-50.
[25]丁欢新.商业银行竞争力评价的现状及评价指标构建[J].商业经济与管理,2003(11)：56-59.
[26]韩俊梅,周高宾.我国商业银行的风险管理-基于CAMELS(?)标的实证分析[J].浙江金融,2007(6)：4-26.
[27]金秀,靳冬利.基于骆驼评价指标的中国银行经营效率实证研究[J].经济研究导刊,2007(5)：69-71.
[28]渤海证券研究所.2003年上市银行评级报告[R].2004.04.26.
[29]叶家声.借鉴“穆迪方法”探索我国商业银行信用等级评估方案[J].华南金融研究,1998,13(2)：37-39.
[30]贺书婕.穆迪公司及信用评级制度(上)[J].城市金融论坛.2000(8)：46-52.
[31]苏娜.我国商业银行信用评级研究.首都经济贸易大学[D].2007年5月.
[32]牛源.中国商业银行风险预警系统的构建及其实证研究[J].北方经济,2007(5)：93-95.
[33]贾曼莉.我国商业银行信用评级方法研究[D].哈尔滨工程大学,2007年6月.
[34]葛兆强.国际评级机构的银行信用评级原理、方法及其局限[J].华南金融研究.2001,16(1)：30-33.
[35]葛兆强.银行个体的信用评级及银行债务评级(上)[J].农村金融研究,2001(4)：4-9.
[36]沈忠刚,刘庆福,杨文武.商业银行风险预警系统的构建[J].现代管理科学,2004(9)：98-99.
[37]刘竹音.基于格金公式和层次分析法建立商业银行风险评价体系的研究[D].山东大学,2008年3月.
[38]黄传照.我国商业银行分支机构风险评价体系构建的探讨[J].商业银行经营与管理.2004(10)：51-53.
[39]辛玲.我国上市银行风险的模糊综合评价[J].中国管理信息化.2009,12(12)：54-56.
[40]刘建辉.如何评价银行信用风险[J].中国国情国力,2000(03)：16-17.
[41]谭燕芝,张运东.信用风险水平与宏观经济变量的实证研究——基于中国、美国、日本部分银行的比较分析[J].国际金融研究,2009(4)：48-56.
[42]翟淑萍.我国商业银行信用管理水平综合评价的因子分析[J].改革与战略,2007(5)：54-57.
[43]郭茵,陆亚琴.中西方商业银行资本金管理比较及其对我国银行的借鉴[J].2005,20(4)：83-85.
[44]范柏乃.我国城市居民生活质量评价体系的构建与实际测度[J].浙江大学学报,2006,36(4)：122-131.
[45]耿金花,高齐圣,张嗣赢.基于层次分析法和因子分析的社区满意度评价体系[J].系统管理学报,2007,16(6)：673-677.
[46]范柏乃,单世涛,陆长生.城市技术创新能力评价指标筛选方法研究[J].科学学研究,2002(6)：661-668.
[47]周立斌,李刚,迟国泰.基与R聚类-变异系数分析的人的全面发展评价指标体系构建[J].系统上程,2010,28(12)：56-63.
[48]顾雪松.基于科学发展观的科学技术评价研究[D].大连理工大学,2009年3月.
[49]Altman E. L.. Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy [J].Journal of Finance.1968,23(9):189-209.
[50]Altman E. L., Haldeman, Narayanan. ZETA Analysis:A New Model to Identify Bankruptcy Risk of Corporations [J]. Journal of Banking and Finance,1977(1):29-54.
[51]Scott E. The Probability of Bankruptcy:A Comparison of Empirical Predictions and Theoretical Models [J]. Journal of Banking and finance,1981(9):317-344.
[52]陈静.市公司财务恶化预测的实证分析[J].会计研究,1999(4)：31-38.
[53]张玲.财务危机预警分析判别模型[J].数量经济技术经济研究,2000(3)：49-51.
[54]陈瑜.对我国证券市场ST公司预测的实证研究[J].经济科学,2000(6)：57-67.
[55]施锡铃,邹新月.典型判别分析在企业信用风险评估中的应用[J].财经研究,2001,27(10)：53-57.
[56]张爱民,祝春山,许丹健.上市公司财务失败的主成分预测模型及其实证研究[J].金融研究,2001(3)：10-25.
[57]王建成.企业信贷能力的因子分析模糊综合评价及应用[J].系统工程,2002，20(3)：26-31.
[58]梁琪.企业信用风险的主成分判别模型及其实证研究[J].财经研究,2003,29(5)：52-56.
[59]布慧敏.我国商业银行信用风险量化度量方法研究[J].统计与决策,2005(1)：99-100.
[60]Martin D. Early Warning of Bank Failure:A logit Regression Approach [J]. Journal of Banking and Finance.1977(2):249-276.
[61]Press S. J., Wilson S.. Choosing between logistic regression and discriminant analysis[J]. America Statistics Association,1978(73):699-705.
[62]Ohlson J.Financial Rations and the Probabilistic Prediction of Bankruptcy[J]. Journal of Accounting Research,1980(2):109-130.
[63]Lawrence E. L.. An analysis of default risk in mobile home credit [J]. Journal of Banking and Finance,1992(6):299-312.
[64]Smith, Lawrence. Forecasting losses on a liquidating long-term loan portfolio [J]. Journal of Banking and Finance,1995(22):959-985.
[65]王春峰,万海晖.商业银行信用风险评价及其实证研究[J].管理科学学报,1998(1)：68-72.
[66]吴世农,卢贤义.我国上市公司财务困境的预测模型研究[J].经济研究,2001(6)：46-55.
[67]于立勇,詹捷辉.基于Logistic回归分析的违约概率预测研究[J].财经研究,2004(9)：15-23.
[68]方洪全,曾勇.银行信用风险评价方法实证研究及比较分析[J].金融研究,2004(1)：41-49.
[69]李志辉,李萌.我国商业银行信用风险识别模型及其实证研究[J].广东社会科学,2006(2)：28-31.
[70]李萌.Logit模型在商业银行信用风险评估中的应用研究[J].管理科学,2005,18(2)：33-38.
[71]Odom M.D., Sharda R.. A neural network for bankruptcy prediction[A], International Joint Conference on Neural Network[C], New York:New York University Press,1990,163-168.
[72]Desai V. S., Crook J. N., Overstreet G. A.. A comparison of neural networks and linear scoring models in the credit union environment [J]. European Journal of Operational Research,1996(95):24-37.
[73]Altman E., Marco G., et al. Corporate Distress Diagnosis:Comparisons Using Linear Discriminant Analysis and Neural Networks (the Italian Experience) [J]. Journal of Banking and Finance,1994(18): 505-529.
[74]West D. Neural network credit scoring models [J]. Computers & Operations Research,2000(27): 362-385.
[75]Tsai C., Wu J.. Using neural network ensembles for bankruptcy prediction and credit scoring[J]. Expert Systems with Applications,2008,34(4):2639-2649.
[76]Angelini E., Tollo G., Roli A.. A Neural Network Approach for Credit Risk Evaluation[J]. The Quarterly Review of Economics and Finance,2008(11):733-755.
[77]Michael B., Gordy A.. Comparative Anatomy of Credit Risk Models[J]. Journal of Banking and Finance,2008 (24):119-149.
[78]Crouhy M., Galai D., Mark R.. A Comparative Analysis of Current Credit Risk Models[J]. Journal of Banking and Finance,2008(24):59-117.
[79]Anderson R., Sundaresan S.. A Comparative Study of Structural Models of Corporate Bond Yields:An Exploratory Investigation [J]. Journal of Banking and Finance,2008(24):255-269.
[80]工春峰,康莉.基于神经网络技术的商业银行信用风险评估[J].系统工程理论与实践,2001(2)：73-79.
[81]陈雄华,林成德,叶武.基于神经网络的企业信用等级评估[J].系统工程学报,2002,17(6)：570-575.
[82]章忠志,符林,唐焕文.基于人工神经网络的商业银行信用风险模型[J].经济数学,2003,20(3)：42-47.
[83]吴德胜,梁木梁.基于V-fold Cross-validation和Elman神经网络的信用评价研究[J].系统工程理论与实践,2004(4)：92-98.
[84]吴冲,吕静杰,潘启树,刘云焘.基于模糊神经网络的商业银行信用风险评估模型研究[J].系统工程理论与实践.2004(11)：1-8.
[85]李萌,陈柳钦.基于BP神经网络的商业银行信用风险识别实证分析[J].经济学研究.2007(1)：18-29.
[86]Lin S., Shiue Y., Chen S., Cheng H.. Applying enhanced data mining approaches in predicting bank performance:A case of Taiwanese commercial banks [J]. Expert Systems with Applications,2009, 36(9):11543-11551.
[87]吴冲,夏唅.基于五级分类支持向量机集成的商业银行信用风险评模型研究[J].预测,2009(4)：57-61.
[88]Bellotti T., Crook J.. Support Vector Machines for Credit Scoring and Discovery of Significant Features[J]. Expert Systems with Applications,2009(3):3302-3308.
[89]Huang Cheng-Lung, Chen Mu-Chen, Wang Chieh-Jen. Credit Scoring With A Data Mining Approach Based On Support Vector Machines [J]. Expert Systems with Applications,2007(11):847-856.
[90]Zhou L., Lai K. K., Yu L.. Least Squares Support Vector Machines Ensemble Models Forcreditscoring[J]. Expert Systems with Applications,2010(1):127-133.
[91]刘云焘,吴冲,上敏,乔木.基于支持向量机的商业银行信用风险评估模型研究[J].预测,2005(1)：52-55.
[92]肖文兵,费奇,万虎.基于支持向量机的信用评估模型及风险评价[J].华中科技大学学报(自然科学版),2007(5)：23-26.
[93]王春峰,康莉.基于遗传规划方法的商业银行信用风险评估模型[J].系统工程理论与实践,2001(2)：73-79.
[94]王春峰,赵欣,韩冬.基于改进蚁群算法的商业银行信用风险评估方法[J].天津大学学报(社会科学版),2005,7(2)：81-85.
[95]郭亚军.综合评价理论与方法[M].北京：科学出版社,2002年.
[96]吕香亭.综合评价指标筛选方法综述[J].合作经济与发展,2009(365)：54.
[97]唐启义,冯明兴.使用统计分析及其DPS数据处理系统[M].科学出版社,2002年：349-352.
[98]高惠璇.应用多元统计分析[M].北京：北京大学出版社,2005年：264-290.
[99]中国银行、中国工商银行、中国建设银行等41家商业银行年报[R].2009年.
[100]陈伟,夏健华.综合t、客观权重信息的最优组合方法[J].数学的实践与认识,2007,37(1)：17-22.
[101]Moody's Investors Service. Financial institutions[EB/OL]. http://www.moodys.com/,2010-03-2 0.
[102]大公国际资信评估有限公司.大公国际资信评估有限公司发布金融危机下的中国银行业最新信用评级与展望[EB/OL].http://www.fzpf.gov.cn/hxjrw/news.asp?id=194000,2010-04-20.
[103]茆诗松.概率论与数理统计[M].北京：中国统计出版社,2000第二版：331-334,410-415.