摘要
为描绘商业银行大型集团客户在压力情景下的信用风险情况,本文以某银行过去12年的不良贷款数据为对象,进行了压力测试实证研究。在研究中引入还原不良贷款率作为中间变量,使用Logit回归建立宏观经济数据和还原不良贷款率之间的量化模型,进而通过商业银行内部评级数据调整得到不同情景下每个集团成员的违约概率;在此基础上,通过蒙特卡洛模拟计算损失分布情况,并估算集团关联风险所致额外损失额。研究结果显示,大型集团客户的压力损失分布,呈现明显的"厚尾"特征,进出口、质押式回购利率、工业生产者出厂价格指数、广义货币、工业增加值对其的影响较大。据此,本文建议监管部门应引导商业银行加强宏观经济和政策研究,扶优控劣优化用信结构,控制用信集中度。
In order to study the credit risk of large conglomerate clients under stressed scenarios, this paper conducts an empirical study of stress testing based on the non-performing loan data of a commercial bank within a period of 12 years. The accumulated non-performing loan ratio is first introduced as an intermediate variable, then a logit regression model is constructed to study the impact of macro-economic data on the accumulated non-performing loan ratio. Based on the model and the internal rating data, the probability of default of each conglomerate member under different scenarios is estimated respectively. We compute loss distributions of the whole portfolio under Monte-Carlo simulations and then measure the extra loss caused by the correlation introduced by conglomerates. The result shows a significant fat-tail in the loss distribution curve, which is impacted by foreign trade, pledged repo interest rate, PPI, M2 and industrial added value. Accordingly, several suggestions are proposed, including more researches on macroeconomics and policies, optimizing credit portfolio, and controlling credit concentration ratio.
出处
《金融监管研究》
CSSCI
北大核心
2019年第5期18-29,共12页
Financial Regulation Research
基金
中国农业银行大客户部课题(编号:2018DK02)
