本文介绍的违约概率模型是建立在企业信用等级转移概率矩阵受宏观经济条件以及企业个体因素共同影响这一基本假设的基础之上。从历史上已知的企业信用等级转移概率矩阵提取出反映宏观经济条件的因子称为??-????????????. 给定宏观经济相关系数后，每一个??-???????????? 的数值都可以反解出信用等级转移概率矩阵。因此，为了预测信用等级的变化，该模型又建立了??-???????????? 与宏观经济变量间的线性回归模型。进而，可以通过对选定的宏观经济变量的预测来实现对??-???????????? 和信用等级转移概率矩阵的预测。特别地，可以预测出不同等级企业的违约概率。

This work introduces a default probability model, m-factor model, which is based on the credit rating transition probability matrix data. This model assumes the credit rating transition of a counterparty is determined by two types of factors: macroeconomic factor and idiosyncratic factor. From historically observed credit rating transition probability matrix, we can extract the reflection factor of macroeconomic conditions, called m-factor. The value of every m-factor can construct a credit rating transition probability matrix, when the macroeconomic correlation coefficient is given. Therefore, in order to predict the change of credit rating, we establish a linear regression model between extracted m-factor series and macroeconomic variables. In turn, from the forecasted values of the selected macroeconomic variables we can estimate the value of m-factor and credit rating transition probability matrix. In particular, this model can provide estimated default probability of each rating grade.