摘要
国内上市公司发生ST可以看作违约,现有的研究基本基于静态的判别分析。本文提出了基于Cox模型的动态违约预报模型,考虑了多阶段财务状态变量对违约强度的影响。选取1998年以前国内A股市场上市的公司作为样本,发现流动比率、其它应收款/总资产和资产周转率等财务指标对公司是否发生ST有显著影响,且其它应收款占比影响较大,投资者应当提高警惕。同时还发现,我国的证券市场还处于初期的不断调试完善阶段,模型存在微小的结构性变化。本文还针对训练样本外的测试样本做了预报检验,并以Logit模型作为对比,发现二者均可以进行违约预报,且基于Cox的动态模型优于传统的Logit模型。
It can be regarded as default if domestic listed company turn into ST. Most existing research are based on static discriminant analysis. In this paper we propose dynamic default prediction mode] based on Cox model, considering multi-period financial state variables＇ effect on hazard intensity. We select the domestic A-share market companies listed prior to 1998 as samples, and find that current ratio, other receivables/total assets and asset turnover have great effect on companies＇ turning into ST. The proportion of other receivalbes is a important cause of ST, and investors should be alert. At the same time, we find that our securities market is still in the early stage of improvement, and the model has little structural change. We also do the out of sample test, comparing with Logit model, and show that both of them can predict default. The dynamic model based on Cox precedes traditional Logit model.
出处
《数理统计与管理》
CSSCI
北大核心
2014年第2期317-328,共12页
Journal of Applied Statistics and Management
基金
国家杰出青年基金项目(编号:70825004)
国家自然科学基金重点资助项目(编号:10731010)
国家自然科学基金委创新研究群体科学基金项目(编号:10721101)
国家973项目子项目(编号:2007CB814902)
上海财经大学"211工程"三期重点学科建设项目
上海市重点学科建设项目(编号:B803)
