Z-score模型最优截断值确定方法比较——基于违约风险预测能力角度

Z-score模型在对企业进行财务困境和违约风险判别方面具有重要的应用价值,最优截断值的确定方法对于提高模型的违约风险判别能力至关重要.文章以医药生物行业上市公司作为样本,运用Fisher逐步判别法从15类财务比率中筛选出判别能力较强的7个指标构建了Z-Score模型,并尝试用加权平均法和考虑先验概率和误判成本的ZETAc模型法分别确定最优截断值,结果发现运用ZETAc模型法能够提高模型整体预测的准确率,而且能够降低第Ⅰ类错误分类的成本,其预测违约风险的能力明显优于加权平均法.

A Comparison of Approaches to Determine Optimal Cut-off Score from the Perspective of Default Risk Prediction

The Z-score model has important application value in the aspects of predicting bankruptcy default risk of enterprises. Methods to determine the optimal cut-off score are crucial to improve model＇s default risk predicting ability. This paper built the Z-Score model by utilizing a sample of and the bio- pharmaceutical listed companies and adopting Stepwise method to select 7 distinguishes ratios from 15 financial ratios. Then the default risk prediction results by using Zetac cut-off score was compared to the default risk prediction results by using weighted average cut-off score. The comparison results showed that prediction by using Zetac cut-off score is more accurate and is able to minimize cost incurred from type I classi cation error, its forecasting ability of default risk is superior to the weighted average method.