Wang Xudong, Hui Xiaofeng
School of Management, Harbin Institute of Technology, Harbin 150001, China.
Entropy (Basel). 2018 Sep 3;20(9):663. doi: 10.3390/e20090663.
This paper applies effective transfer entropy to research the information transfer in the Chinese stock market around its crash in 2015. According to the market states, the entire period is divided into four sub-phases: the tranquil, bull, crash, and post-crash periods. Kernel density estimation is used to calculate the effective transfer entropy. Then, the information transfer network is constructed. Nodes' centralities and the directed maximum spanning trees of the networks are analyzed. The results show that, in the tranquil period, the information transfer is weak in the market. In the bull period, the strength and scope of the information transfer increases. The utility sector outputs a great deal of information and is the hub node for the information flow. In the crash period, the information transfer grows further. The market efficiency in this period is worse than that in the other three sub-periods. The information technology sector is the biggest information source, while the consumer staples sector receives the most information. The interactions of the sectors become more direct. In the post-crash period, information transfer declines but is still stronger than the tranquil time. The financial sector receives the largest amount of information and is the pivot node.
本文应用有效转移熵来研究2015年中国股市暴跌前后的信息传递。根据市场状态，整个时期分为四个子阶段：平静期、牛市期、暴跌期和暴跌后时期。采用核密度估计来计算有效转移熵。然后构建信息传递网络。分析了网络中节点的中心性和有向最大生成树。结果表明，在平静期，市场中的信息传递较弱。在牛市期，信息传递的强度和范围增加。公用事业部门输出大量信息，是信息流的枢纽节点。在暴跌期，信息传递进一步增强。这一时期的市场效率低于其他三个子时期。信息技术部门是最大的信息源，而日常消费品部门接收的信息最多。各部门之间的相互作用变得更加直接。在暴跌后时期，信息传递有所下降，但仍强于平静期。金融部门接收的信息量最大，是关键节点。
