Statistical Methods In Credit Risk Management In A Number Of Applications

Credit card as the electronic means of credit trade is the most important means of payment and credit tool.Credit card business improves propensity to consume and boosts economy.However,credit card is a consumer credit product with high financial risk.Due to impossible to completely hedge clients' credit risk,we must evaluate client credit risk through scientific way,take effective measures such as pricing and granting credit line to match risk and income,and then meet credit risk management objective to ensure issuers and industry healthy development.Credit card risk management includes many aspects.This paper studied validation of credit risk evaluation model and credit card pricing.Firstly,the deficiency of two popular methods,measure of ordinal association and ROC analysis, for validation of binary regression models was illustrated by a designed numerical example.An improved method based on permutation test was proposed,whose performance was investigated by two real examples.In addition,paper studied credit card pricing based on portfolio theory.Computed customers' risk coefficient and set APRs by CAPM.Paper compared normal and CAPM credit card pricing from aspect of risks and income and found that CAPM credit card pricing had significant improvement in risk control.At last,innovation and insufficiency of this paper were summarized.