基于完全信息的测算信用等级违约概率的新方法

A New Method of Measuring Default Probability with Complete Information in Credit Ratings

研究目标:为充分利用历史数据的信息测算违约概率提供一种新方法。研究方法:从违约强度与违约概率的关系出发,提出计算违约概率的新思想。以银行贷款为例,在分析贷款的起止时间与违约概率考察期间的时序关系之后,通过创新性地构造特征函数的方式,描述了考察期间贷款数目的时变特征,提出了利用历史数据测算违约概率的公式。研究发现:传统的测算方法对违约概率存在一定程度的低估。研究创新:采用该方法测算违约概率时,考察期间的任何一笔贷款都是潜在的测算样本的数据源。相比于传统方法中"只选择在考察期初存在的贷款作为测算样本"的做法,本文的方法利用历史违约数据的信息更完全。研究价值:为基于历史数据测算违约概率提供了新思路,研究成果可广泛用于银行贷款的违约概率的测算。

Present a new method of calculating default probability by making a full use of historical data.Research Methods:After analyzing relationship between default intensity and default probability,this paper provides a new way of estimating default probability.Specifically,this paper discusses the temporal relation between the beginning and ending time of loan and inspection horizon of the measuring PD,then proposes a characteristic function to construct a function for describing the time-varied number the loan,and presents a new formula of estimating default probability.Research Findings:By an empirical study,we find there is an underestimation of default probability in traditional methods.Research Innovations:Using this method,any loan appeared in the inspection horizon can be sampled from historical data and used for calculating default probability.Compared with the traditional methods,this method can make full use of the information implied in historical data and has more practical application value.Research Value:This paper gives a new way of calculating probability of default.It can be widely used in the practice of risk management of commercial banks.Key Words:Default Probabili