摘要
中国股市因法制建设、市场机制不完善以及投资者心理不成熟等原因,易受外界各种因素的影响而产生突变。本文引入基于贝叶斯推断的突变点判断方法,通过构造似然函数,同时利用先验信息和样本信息,并采用吉布斯抽样完成突变点位置的判断。此外,本文将突变点个数的判断问题作为模型的选择问题对待,采用后验优比进行抉择。引入该方法后,本文对我国上证指数序列的突变点个数及突变位置进行判断。研究发现,在1993年1月至2008年5月,上证指数存在四个突变点,分别为:1994年12月前后,2000年1月(或2001年7月)前后,2006年4月前后,2007年11月前后。
Subjected to a variety of outside factors, change-point appears easily due to the imperfect legal system and market mechanism as well as psychologically immature investors and other reasons in the stock market of China. This paper determineds the locations of change-points by introducing the method which is based on the theory of Bayesian inference, constructing a likelihood function, while using prior information and sample information, as well as adopting Gibbs sampling. In addition, Which considers the problem for the number of change-points as a problem of selection for the model, is determined by utilizing the posterior odds. After the introduction of the method, determined the number and locations of change-points in the sequence of Shanghai Composite Index. It was found in the study that, from January 1993 to May 2008, Shanghai Composite Index had four change-points as follows： about December 1994, January 2000 （or July 2001）, April 2006 and November 2007.
出处
《中国管理科学》
CSSCI
北大核心
2009年第3期8-17,共10页
Chinese Journal of Management Science
基金
辽宁省高等学校创新团队支持计划资助项目
关键词
贝叶斯推断
突变点
上证指数
GIBBS抽样
bayesian inference
change-points
shanghai composite index
gibbs sampling
