贝叶斯方法在信用风险度量中的应用研究综述  

Bayesian Methods in Credit Risk Measurement:A Survey

作　　者：丁东洋[1] 周丽莉[1] 刘乐平[2] 

机构地区：[1]南昌大学,江西南昌330031 [2]天津财经大学,天津300222

出　　处：《数理统计与管理》2013年第1期42-56,共15页Journal of Applied Statistics and Management

基　　金：天津社科规划项目"宏观统计数据可靠性评估方法研究"(TJTJ10-651)资助;全国统计科研计划项目"小域估计理论及其在我国统计调查中的应用"(2009LZ020)资助

摘　　要：贝叶斯方法可以有效的处理信用风险度量中常见的数据缺失问题,而且为科学使用专家意见等主观经验提供了有效途径,已被广泛应用于信用风险度量领域。本文从模型构建、估计方法及模型比较三个方面对应用贝叶斯方法度量信用风险的重要文献进行综述,重点关注信用风险的违约相关性和风险蔓延性等最新研究热点,为深入研究信用风险度量问题提供参考,并引起国内风险分析人员对贝叶斯方法的兴趣。Bayesian methods can effectively deal with the common problem of missing data, and provide a formal way for the scientific use of subjective experience, has been widely used in credit risk measurement. This paper presents a survey of important literature of credit risk measurement using Bayesian methods in three aspects of modeling, estimation methods and model comparison, and focus on the hot issues of default correlation and risk contagion of the latest credit risk research, in order to provide reference for further research in the credit risk measurement, and raise the domestic risk analyzer interest in Bayesian methods.

关 键 词：贝叶斯方法 信用风险度量 违约依赖性 

分 类 号：O212[理学—概率论与数理统计]