我国债券违约风险评估模型的构建与应用

Construction and Application of Bond Default Risk Evaluation Model in China

自2014年我国首只债券违约以来,信用债违约数量与金额呈现双增长趋势并在近两年呈现爆发式上涨。本文聚焦债券违约风险的产生原因与前瞻性识别,通过分析内外部违约风险因素及其外在表现形式总结出风险识别指标体系。之后基于Logistic回归构建债券违约概率识别模型并对市场样本进行预测精度与时效性检验,结果表明模型对违约债券的识别率高于非违约债券,可认为识别结果偏谨慎有助于违约风险的规避。最后基于系统聚类分析对模型进行改进,将单纯评判违约与否修正为债券违约风险等级的归类,使风险评价结果更符合市场真实情况。

Since the default of China’s first bond in 2014,the number and amount of credit bond defaults have shown a double-growth trend and have exploded in the past two years,which means bond defaults have entered a normalized era.This paper focuses on the causes and identification of bond default risk,and firstly summarizes the risk index system by analyzing the risk factors and their external manifestations.Then,based on Logistic regression,a bond default probability identification model was built and the market samples were tested for prediction accuracy and timeliness.The results showed the model’s recognition rate of defaulted bonds was higher than that of non-defaulted ones,which means the results are more effective for avoiding the default risk.Later,based on the systematic cluster analysis,the model was improved to the classification of bond default risk,which made the evaluation more in line with real situations of the market.