Design And Implementation Of Credit Risk Internal Rating System In Domestic Commercial Bank

The Interna l Ratings-Based Approach (IRB) is origina lly proposed in BaselAccord II in2004to achie ve the se lf-disc ip line regulation of commercia l banks and toassure that the ir capita l adequacy ratio can meet the require me nt of eterna lsupervis ion. The Interna l Ratings-Based System is a platform to realize the IRBprinc ip le of commerc ial banks. It can be cons idered as a genera l term of vario us typesof methods, imp le mentation processes, controlling modules, data collecting,informatio n techno lo gy, IRB rules and the default loss measure ment. Based on thelong-term application and feedbacks, the va lue of IRB is fully re flected. The industryhas formed a consens us that the Interna l Ratings-Based System is adopted for dailycredit risk manage me nt. It can meet the requireme nts, which inc lude theself-disc ip line required by the regulatory authorities and the reduction of eterna lcredit rating dependences. Furthermore, it a lso matc hes the law of deve loping for thebank ing industry. Nowadays, credit risk re ma ins the ma in risk faced by the Chinesebank ing industry. Neverthe less, the techno logy and measureme nt of risk mana geme ntin most commerc ia l banks in China are relative ly backward. There is still a widedeve lopme nt space for the Interna l Ratings-Based Syste m in the industry.Based on the reality and practice of credit operations in the domestic commerc ialbank, this paper is focused on ana lyzing the common require ments of Interna l RatingSystem to propose the reasonable IRB a nd pattern in the industry. The key of Interna lRating is establis hing the relatio nship between credit rating and defa ult probability.This paper introduces the general process of data mining and statistica l analys isestablis hed by one of the interna l credit grading system, whic h is called Mark ing Ca rdModel. The methodolo gy o f establis hing the Marking Card Model is emp loyed toanalys is the data collection, bus iness definitions, data scrubbing, model packet,variab le constructio n, variable analys is, model confirmatio n, mode l validation, andmodel ca libratio n. Meanwhile, based on the data collected in a German bank, anemp irical research is cond ucted to analys is the private reta il sector in the MarkingCard Model. According to the result o f Lo gistic regressio n by using SPSS, it showsthat the mode l is variab ility, which enab les to get applicab le prediction.Fina lly, accord ing to the curre nt s ituation in the industry, inc luding lack of initia l data, defic ienc y of Rating methodology, low coverage rate of rating assets,underutilized default probability model for assess ment, application o f Interna lRatings-Based Syste m in the large domestic co mmercia l banks is ana lyzed toelaborate a feas ible program. The progra m argues that the short-term comb ination ofMarking Card Mode l and Default Probability Model can be e mplo yed to assessdefa ult information. Then, a trans ition is created from this to the lo ng-term Defa ultProbability Model.