基于变量聚类和COX比例风险模型的企业财务预警研究

A Study of Financial Distress Warning based on R-variance Clustering and COX Proportional Hazards Model

以沪深两市A股上市的制造企业为样本,运用变量聚类方法,从32个财务与非财务指标中选取11个指标作为财务预警典型指标,在此基础上,建立COX比例风险模型。通过COX比例风险模型实证研究发现:速动比率、股东权益相对年初增长率、资产负债率为危险因素,增加了发生财务困境的危险性;总资产周转率、独立董事比例为保护因素,降低了发生财务困境的危险性。通过COX比例风险模型的判别能力检验,COX模型针对训练样本和测试样本的综合识别准确率均达到80%以上。研究结果证实,COX比例风险模型具有较好的财务预警判别能力。

The manufacturing enterprises listed in Shanghai and Shen Stock Markets are selected as samples for empirical research.We use cluster method to reduce financial and non-financial variables from32 to 11as the early warning indicators,and build the COX proportional hazards models.Through this study,we find that quick ratio,the beginning of shareholders＇ rights and interests of the relative growth rate,ratio of liabilities to assets are risk factors which may increase the risk financial distress;while total assets turnover,the proportion of independent directors are the protect factors which may reduce the risk of financial distress.Through training of the COX proportional hazards model,the recognition accuracy rate reach more than 80%,showing that the COX proportional hazards models have good financial warning discriminant ability.