We build a multivariate heterogeneous autoregressive model(DIHM-MHAR)with double independent infinite hidden Markov regime switching structure to study the dependence and structural characteristics of high-frequency risks across Chinese financial markets.Further,we use the forecast variance decomposition approach proposed by Diebold and Yilmaz[1,2]to analyze the dynamic transmission mechanism of the high-frequency systemic risk.The five-minute high-frequency data of Chinese stock market,stock index futures market and treasury futures market from September 6,2013 to July 21,2023 are selected as samples for empirical research.It is found that(1)the estimated high-frequency realize volatility estimator can better reflect the occurrence of risk events in China's financial markets and is a good proxy of systematic risk;(2)the risks of stock market and stock index futures market are significantly driven by other markets,while the risk of treasury futures market is not affected by other markets;(3)the coefficients and variances of variables that affect the risk of each financial market have the significant time-varying characteristics,and the coefficient shows an obvious feature of regime switching;(4)on the whole,the risk spillover degree of Chinese financial markets is relatively large;(5)the stock market and stock index futures market are the net suppliers of the risk of the financial markets,while the treasury futures market is the net demander;(6)the stock market majorly supplies the risks to the stock index futures market and the treasury futures market,and treasury futures market majorly supplies the risk to stock index fu-tures market.This study is of great practical significance to guide financial supervision and investment in financial assets.
关键词
中国金融市场/高频风险/无限隐马尔可夫状态结构/DIHM-MHAR模型/预测方差分解法
Key words
Chinese Financial Markets/High-frequency Systemic Risks/Infinitely Markov State Structure/DIHM-MHAR Model/Forecast Error Decomposition Approach
维普
万方数据