中文摘要：

银行信用组合违约风险的度量和计算对银行监管有着重要的意义.使蒙特卡洛研究信用组合违约概率时,为提高模拟效率,越来越多的学者采用了重要性抽样技术来实现.它主要通过条件独立性和＂均值移动＂两个步骤实现.本文基于前人研究结果的基础之上,提出了一种基于违约相关性矩阵的多因子变方差的重要性抽样算法.该算法通过主成分分析选择违约结构中的占优成分并扩大其方差来实现.数值算例证明了该方法在信用组合遭遇极值事件时,能够提高模拟效率及计算精度,具有一定的计算优势。

英文摘要：

The bank credit portfolio risk measurement has great significance to bank supervision.One of the most popular methods to estimate the default probability of credit asset is Monte Carlo simulation.In order to improve the simulation efficiency,more and more studies have adopted the important sampling technique to deal with it.In this paper,we propose an importance sampling procedure which does not need the conditional independence which previous studies had to base on.The procedure we provide uses principal component analysis to choose dominant factors.Numerical experiments are provided and the results show that our approach when a credit portfolio confronts extreme events,offers substantial variance reduction and outperforms plain Monte Carlo algorithm and Morokoff IS algorithm.