Retail Banking Credit Risk Modeling Study

To improve their competence, the commercial banks in our county arepaying more attention to retail credit. The retail credit mainly means loanto private. In the trend of the development of finance in the world, theincome from retail credit becomes more important for the banks. Thoughour banks have invested lots in the retail exposures, we can see traditionalmethod popularly employed in analysis of retail credit risk. Measurementby quantity in risk is the base of the exactitude management of credit risk,which is still the weakness of our banks.Characteristic of retail credit need new technique in analysis of credit.The use of credit model can enhance the efficiency of retail operation andpredict the credit lose better. Though attaching importance to the creditrisk of retail exposures, our banks have not taken enough notes to creditmodel of retail exposures. This makes the development of credit analysisdrop behind the expanding of retail credit. Though we have not had toomuch lost in retail credit, we should not ignore the risk in this field.Though some scholars in our country have some research about creditmodels, but most of them focus on the positive analysis about idiographicmodels on some samples,lack of a systemic discuss on making modelsabout retail credit risk.This thesis is composed of three parts.The first part includes chapter 1 and 2. The first chapter expatiates onthe risk of retail credit of banks. Firstly, a definitude definition is made;secondly content about the credit risk is given; then this chapter studiesthe particularity about retail credit risk from a few aspects.The second chapter introduces models of retail credit. The thesisdescribes the individual analysis for traditional means and models ofretail credit and means of the portfolio credit analysis. Though thetraditional means have their advantage, their shortcomings are becomingsharp. The management of retail exposures shall be accurate statistics andquantitative analysis. The enormous quantity of retail credit needs theirrisk analysis based on more effective way. So the models of retail creditrisk shall be used more widely. This chapter introduces some types ofretail credit models, such as judgmental systems, quantitative creditscoring models, decision tree models, neural network models, dynamiccredit models,and some models of credit portfolio, and then gives somecommentary.The second part consists of chapter 3 to 6. This part covers the keypoints of building credit model for retail credit of banks.The third chapter discusses the way to employee models of retail creditrisk. The key to build credit models is not only a statistics job but also acomprehensive systematic work. In this job lots of details must beconsidered besides significance in statistics. This chapter studies somefacts besides mathematics that have to be consider in models and thebottleneck of spreading the models in retail credit. The most successfulorganizations do not always have better models but they always haveright persons with knowledge, experience and latest technology, they alsopay attention to the quality of the data sample that will be used to createthe model.The fourth chapter studies the base of making credit models, managementinformation system (MIS) of credit. For our banks, perfect data is muchless than perfect expert, which is our weakness. So here comes somesuggestion about building our own MIS. The system needs to describe notonly the customers' data but also the alternative strategies applied tospecific risk categories. The management information system of creditusually contents three parts: data stock; interim layer to dispose data andanalysis part. This part gives some details of the three parts.The fifth part considers the sample rejection fact in credit models.Sampling bias is an issue that creates problems for the application ofcredit risk models. A very important example of sampling bias arises incredit risk models trained on only those loans that have been accepted inthe past when the models are intended to estimate risk on all...