存活分析模型之应用-财团法人金融联合徵信中心.PDF

Review of Financial Risk Management 1 96 1-30 Credit Analysis of Credit Card Holders The Application of Survival Model Yeh, C lover M C hang, C hing-Y i Ph.D. Candidate, Graduate Institute of Finance, Financial Specialist, Cheng Shin Rubber Ind. National Taiwan University Co., Ltd. L iao, H sien-H sing J ou, David G Associate Professor, Department and Graduate Institute of Finance, National Taiwan University Chairman, Hontai Life Insurance Co., Ltd. and Adjunct Associate Professor, Department and Graduate Institute of Finance, National Taiwan University Logit Probit (Survival Analysis) J E L G21 * ** E-mail: ntu.edu.tw 2 96 6 Abstract Most current methods on credit analysis of credit card holders are only able to predict the default (such as the discriminant analysis) or default probability (such as the Logit or the Probit model) in a period time of future time at a specific point of time. These methods cannot provide the information on the default probability in different time points in the future. Employing the survival analysis that is extensively used in medical and actuarial science, a pre-warning model and a predictable survival table for different groups of cardholders are established. Banks can use the pre-warning model o