关键词:

风险传染, 动态条件相关性模型-混频数据抽样模型, 长/短期关联, 动态相关, 系统性风险

Abstract: Exploring excessive institutional dependences is the key to preventing and controlling large-scale outbreaks of systemic risks, but long-term correlations between the institutions are inevitable, and financial crises can also trigger short-term excessive correlations between institutions. This paper, employing the DCC-MIDAS model and dynamic conditional correlation coefficient decomposition approach, studies the long-term and short-term correlations within a unified framework. Through the analysis of the time-varying and persistent differences of short-term correlations, as well as the sensitive factors of long-term correlations, it explores the changes in the scope and duration of risk contagion between institutions in China’s financial system during different periods of structural changes, as well as the macro and micro characteristics of long-term correlations between financial institutions. It is found that the contagion during the shocks is temporary, and with the diversification of China’s financial institutions, the magnitude and persistence of the contagion have decreased. The size differences can reduce the long-term correlation and indirectly reduce the contagion. However, with the increasing marketization, the influence of market factors on long-term correlation increases in China’s system. It is suggested to use institutional differentiation as a reference factor for risk control strategies and to monitor and regulate the impact of market indicators on the degree of association to prevent institutions from forming excessively high long-term associations, in order to control potential contagion in the future.

Key words:

 risk contagion, DCC-MIDAS, long/short-term correlation, dynamic correlation, systemic risk