Research On Credit Card Credit Risk Assessment Of Commercial Banks In China

Credit risk management is a core challenge for all type of business, and specifically crucial to credit card business. In the past thirty years, the credit card business has evolved tremendously in China benefiting from the nation-wide resident income increase and the change of consumer behavior. While various advanced mathematical models has been widely applied to control credit risk among international banks, in China, due to relatively short history in adopting real credit card business, the Chinese commercial banks are still suffering from lack of sophisticated analytical skills to avoid risk. Compare to their international industry-peers that scaled credit card business as core profit source, the Chinese commercial banks has just stepped into this game and would require extraordinary effort to catch up. In this paper, we will discuss how to improve the quality of credit card risk control, and how to accelerate the process of emerging to advanced risk management tools for the Chinese commercial banks.This paper first introduces the credit card industry development mode and business characteristics, followed by discussion to the three major risks, then presents the statement that the credit risk has become the primary risk of this business and the primary subject to be controlled, as well as the evidence that supports it. After introducing the current status of credit risk, the paper furthers down to the root cause of such risk in China with detailed analysis. Among the three current credit risk assessment models:applicant's credit score, behavior scores and profitability score, the paper determined the behavior scores approach combined with logistic regression and decision tree methods would carry the best suitability to the Chinese market. To validate this point, extensive analysis has been performed with real-life examples. The validation focuses on the history of applicant's past overdue records, withdrawal records, balance ratio, transaction types, transaction time, frequency to use card, payment time and payment amount. Binary logistic regression models and decision tree models have been established respectively to predict credit card default rate on each case.In the end, the paper reaches a few key recommendations for the Chinese commercial banks to improve overall credit risk control processes based on most recent industry research and the previous analysis:first, establish a dynamic individual credit evaluation system for internal use of the bank second, enhance retail-time monitoring on card activities by improving data integrity and analytical efficiencythird, make clear short term and long term strategy in regards of risk control fourth, provide tiered training programs to fit different internal needs, meanwhile, actively seeking external resources that can lead business transformation at different levels.