基于评级模型的违约率估计

Estimation of Probability of Default by Using Ratings-based Models

违约率是信用风险建模的核心输入变量,文章基于评级模型对违约率进行估计。估计违约数据很少的低违约组合的违约率是一个比较困难的问题,用最大谨慎原则方法解决这类问题时结果偏大,过于保守。文章将最大谨慎原则的思想与极大似然方法相结合,估计了低违约组合的违约率。估计的结果比仅用最大谨慎原则得到的结果很大程度上降低了保守度。

Probability of default （PD） is a core input variable in the credit risk model. In this paper, PD is estimated by using ratings-based models. How to estimate PD of the low default portfolios which have few default data is a difficult problem. When using the most prudent estimation principle to solve this problem, the result is too large and over-conservative. Combining the thought of the most prudent estimation principle with the maximum likelihood approach, it is estimated that PD is of the low default portfolios in this paper. The estimating results are much less conservative than the ones which are obtained only by using the most prudent estimation principle.