摘要
如何精确计量信用风险一直是理论界和实务部门的难点和热点问题。本文使用广义线性混合模型对信用风险进行建模分析,将影响违约概率的可观测因素和不可观测因素分别用固定效应和随机效应表示,根据需要随机效应可扩展为多个因子。研究表明,模型具有较好的延展性,宏观经济变量作为可观测变量无法全部解释违约率的异质性,随机效应可以更好地捕捉违约率的异质性,行业因素对违约概率的影响比宏观经济变量显著。
How to accurately measure the credit risk is always a hard and hot issue in the research and administrative institutions. This paper is to model and analyze the credit risks by the generalized linear mixed models （GLMMs）. The observable and unobservable factors, which affect credit default probability, are expressed as fixed effects and random effects respectively, and the random effects can be expanded into multiple variables as required. The result shows that model can be extended with more variables and the heterogeneity of default probability can＇ t be totally explained by the macroeconomic covariates, but can be well captured by the random effects. It is also found that the impacts from industrial sectors on the credit default probability are more significant than those from the macroeconomic variables.
出处
《统计研究》
CSSCI
北大核心
2017年第8期53-60,共8页
Statistical Research
关键词
信用风险
GLMMs
违约概率
Credit Risk
GLMMs （ Generalized Linear Mixed Models）
PD （ Probability of Default）
