摘要
现有文献主要讨论单个市场收益的长记忆性质或者市场间收益的相依性,但都没有将上述两个问题联系起来进行研究.与以往研究不同,文章将研究多个国家的股票市场收益与美国股票市场收益之间尾部相依性的长记忆问题.这不仅能检验市场的有效性,还有助于刻画国际市场间的相依结构,观察危机传染的特征并做出预警.为了刻画市场间尾部相依长记忆性的动态变化结构,文章对现有模型加以改进,提出了长记忆性的变点检测方法.从变点的角度实证分析了中国、日本、德国和法国股市与美国股市之间尾部相依的长记忆性,进而研究了次贷危机的传染现象.实证结果表明,文章提出的模型在分析金融传染方面是有效的,在不同的股市之间的实证结果具有相似性.第一变点和第二变点的时刻分别与次贷危机的结束和开始时间相对应,基于两个变点将数据分为三段对估计结果进行分析发现,下尾相依系数在危机传染期间明显变大,而且相依性的长记忆性也明显增强,说明次贷危机对所分析的国家具有传染效应.
Most existing literatures study long memory properties in the return of one single market or the dependence structure between two markets,but none of them studies the two problems together.Unlike them,this paper focuses on the long memory effects in the tail dependence of different international stock markets.The topic is important because the existence of long memory effects invalidates market efficiency hypothesis.And understanding the long memory dynamics in tail dependence is useful to capture cross-market dependence and alert for the crisis contagion.We improve the existing model and use this new model to analyze the dependence structure between Chinese,Japanese,German,French stock markets and American stock market from the change-point view individually to capture the long memory property of the tail dependence.Our results show that the improved model is stable for different stock markets,and the first change-point and the second change-point are in accordance with the ending time and the beginning time of the subprime crisis.Tail correlation increases significantly during the crisis contagion period,and its long memory is stronger than that of the other two periods.
出处
《系统科学与数学》
CSCD
北大核心
2016年第12期2282-2293,共12页
Journal of Systems Science and Mathematical Sciences
基金
国家自然科学基金青年-面上连续项目(71371007)
面上项目(71671171)
重点项目(71631006)资助课题
