主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2019, Vol. 27 ›› Issue (5): 42-49.doi: 10.16381/j.cnki.issn1003-207x.2019.05.005

• 论文 • 上一篇    下一篇

基于状态空间模型的宏观经济因素对股市流动性的建模分析

何迪1, 周勇2,3   

  1. 1. 南京大学经济学院, 江苏 南京 210093;
    2. 统计与数据科学前沿理论及应用教育部重点实验室, 上海 200062;
    3. 华东师范大学统计交叉科学研究院和统计学院, 上海 200062
  • 收稿日期:2017-10-24 修回日期:2018-02-05 出版日期:2019-05-20 发布日期:2019-05-25
  • 通讯作者: 周勇(1964-),男(汉族),广西人,中国科学院数学与系统科学研究院,研究员,华东师范大学统计交叉科学研究院和统计学院教授,研究方向:统计学、数量金融与风险管理、生存分析、计量经济,E-mail:yzhou@amss.ac.cn E-mail:yzhou@amss.ac.cn
  • 基金资助:

    国家自然科学基金委重点资助项目(71331006);国家自然科学重大研究计划重点资助项目(91546202)

Modeling and Analyzing Liquidity in Stock Market Using Macroeconomic Factors Based on State Space Model

HE Di1, ZHOU Yong2,3   

  1. 1. School of Economics, Nanjing University, Nanjing 210093, China;
    2. Key Laboratory of Advanced Theory and Application in Statistics and Data Science, MOE, Shanghai 200062, China;
    3. Academy of Statistics and Interdisciplinary Sciences, East China Normal University, Shanghai 200062, China
  • Received:2017-10-24 Revised:2018-02-05 Online:2019-05-20 Published:2019-05-25

摘要: 证券市场的流动性是评估一国证券市场是否健康运转的重要指标。关于流动性的各种学术研究也是金融市场微观结构理论中的热点问题。本文主要考察宏观经济因素对时变的股市流动性的影响,选取度量股票流动性的指标以及宏观经济变量,针对行业分类的面板数据提出三个动态因子模型,并引入跨行业的能够捕捉风险聚集现象的潜在因子进行建模,利用状态空间模型与Kalman滤波进行分析和样本外流动性风险预测。我们提出的新颖的带回归效应的高斯面板数据时间序列模型,结合了用主成分的方法从大量宏观经济协变量中提取出来的因子,用来分析和预测股市的流动性风险,具有较强的信息挖掘作用。在对上证A 股的实证研究中,我们发现动态的潜在因子的引入对于防止流动性估计的偏差是需要的。

关键词: 流动性, 宏观经济因素, 状态空间模型, Kalman滤波, 潜在因子

Abstract: Liquidity in stock market is a crucial indicator to assess whether a country's stock market is in healthy operation. The academic research about liquidity is a hot issue on financial market microstructure theory. The influence of macroeconomic factors to time-varying liquidity in Chinese stock market is considered in this paper. First the indexes are selected to measure stock liquidity and macroeconomic covariates, and then three dynamic factor models are proposed for the panel data of industry sectors, with introducing a latent factor that captures the risk cross-sectionally clustering phenomenon to modeling. Moreover, liquidity is analyzed using the tools of state space model and Kalman filter, and the out-of-cample liquidity risk prediction of all the models is implemented. The novel Gaussian panel data time series model with regression effects is presented, for the analysis and forecast of stock liquidity risk, containing the principal components from a large number of macroeconomic covariates, which has a strong information mining effect. In an empirical application to stock data from Shanghai stock exchange A share market, it is found that a dynamic latent component or frailty factor is needed to prevent a downward bias in the estimation of liquidity.

Key words: liquidity, macroeconomic factors, state space model, Kalman filtering, latent factor

中图分类号: