摘要
文章尝试从多维视角下运用实证方法对信用风险度量模型的绩效与稳定性进行研究,利用上市公司2001~2004年连续4年的财务数据检验模型对新数据的稳健性、模型校准的时间跨度要求以及模型度量受行业因素影响的程度。实证结果发现:(1)运用跨时间数据对同一模型和同一系列数据对不同模型的稳定性进行检验,模型表现出较好的稳定性;(2)模型的预测能力受行业数据因素的影响而降低,但仍具有较好的稳定性;(3)模型具有较强的预测能力,当用所建立模型对2005年的部分上市公司进行风险预测时,其准确率均在90%以上;(4)跨年度数据建立模型预测的稳定性较年度数据建立模型预测的稳定性略高。
The paper attempts to empirically investigate the efficiency and reliability of the models from multi-angles by employing the data of Chinese listed firms from 2001 to 2004. The result reveals that: (1) the models show the stronger reliability when they are examined by cross-year data and the same data for different models; (2) the efficiency of the models is affected by the factor of industry when examined by cross-industry data; (3) the result of the prediction is well by employing the data in 2005; (4) the reliability of the models structured by multi-year data is better than the one of the models by single year data.
出处
《证券市场导报》
CSSCI
北大核心
2008年第6期41-47,共7页
Securities Market Herald