信贷信息不对称下的信用卡信用风险研究

Forecasting of Credit Card Credit Risk Under Asymmetric Information Based on Nonparametric Random Forests

信用卡市场具有信贷信息不对称的特征,是信用卡风险产生的主要原因之一。在信贷信息不对称条件下,如何利用统计分析、数据挖掘等高新技术,建立可靠的分析模型,对信用卡用户的行为进行风险识别和预测,具有非常重要的意义。本文首次把改进后的非参数随机森林分类(RFC)方法应用到信用卡信用风险的评估中,并和其他模型进行比较,发现非参数随机森林方法往往要优于基准的Logitic模型和SVC模型。实证发现职业、年龄、家庭人口数、月刷卡额、学历、家庭月收入对信用风险有显著影响,而性别、婚姻状况等对信用风险影响不显著。

Information asymmetry is one of the major reason for credit risk in the credit card market.In this case,it is importance to establish a reliable model for risk identification by statistical analysis,data mining and other high-tech. This is the first application of the RFC in the credit risk assessment,and the empirical evidence reveals that RFC does outperform the SVC and Logistic model.the result has important implication for the enriching credit risk assessment system and enhancing risk management. Evidence found that variables of occupation,age,population in family,the amount of credit card,education,family income have a significant effect on credit card default risk,but variables of gender,marital status has not significant effect on the credit risk.