中国上市公司未来收益预测的实证研究——贝叶斯动态模型及其预测  

On earnings forecast of Chinese listed companies——Bayesian Dynamic model and forecasting

作　　者：陆璇[1] 刘慧霞[1] 陈晓[2] 

机构地区：[1]清华大学数学科学系,北京100084 [2]清华大学经济管理学院,北京100084

出　　处：《中南大学学报（社会科学版）》2003年第2期223-226,共4页Journal of Central South University：Social Sciences

基　　金：清华大学基础研究基金(JC2000049).

摘　　要：预测上市公司的未来收益是投资者、证券商、债权人和管理层所关注的问题。国内外的一些实证研究结果表明:上市公司定期公布的财务报告中包含关于公司未来收益变化的信息。基于此,利用中国上市公司年报中的信息,采用贝叶斯动态回归模型对公司未来收益的变化进行预测,并将其结果与静态回归模型的预测结果进行比较,认为贝叶斯动态回归模型的预测效果在一般情况下优于静态回归模型,但要取得更好的预测效果还有待于更多的数据积累,同时还应考虑宏观经济环境变化的影响。Forecasting of listed companies' future earnings is an important issue for investors, stockbrokers, creditors and administrators. Studies by foreign and domestic researchers show that financial reports disclosed by the listed companies contain information about the earnings variability in the future. On the basis of these studies, the authors use Bayesian dynamic regression model to forecast the variation of the company′s prospective earnings, using the data in the annual financial reports of Chinese listed companies, and compare the empirical analysis result under Bayesian dynamic regression model with the result under ordinary static regression model. The authors′ conclusion is that Bayesian dynamic regression model generally offers better forecast than static regression model and that more empirical data and information about the macroeconomic circumstances must be obtained in order to have more satisfactory forecast.

关 键 词：中国 上市公司 收益预测 实证研究 贝叶斯动态回归模型 静态回归模型 预测效果 宏观经济 

分 类 号：F279.246[经济管理—企业管理] F832.5[经济管理—国民经济]