Abstract: With the widespread financial market risk contagion or cross-market risk contagion, the multi-market volatility caused by risk contagion tends to be convergent with multiple outbreaks, profound impact, wide diffusion, and long duration. Based on the concern and focus of the market and the industry on this phenomenon, theoretical and empirical studies on the risk contagion phenomenon have become a popular topic in the field of finance. Due to the pervasive and disruptive cross-market risk contagion in financial markets, which may further develop into a national or regional systemic financial risk and financial crisis. Therefore, theoretical, methodological, and empirical studies on the cross-market risk contagion phenomenon are not only of great theoretical value but also of significant practical significance. Considering the correlation and directness of inter-market risk contagion, this paper focuses on China's commodity futures markets, where cross-market risk contagion is more significant and intuitive, and systematically studies the cross-market risk contagion via comprehensive measurement, network analysis, and trend evolution. The possible contributions are as follows: This paper focuses on the cross-market risk contagion phenomenon of commodity futures, including both methodological comprehensive measurement index and contagion network analysis, as well as empirical phenomenon summary and regular overview. The methodology and conclusions help investors, managers, and regulators to understand and grasp the cross-market risk contagion phenomenon and overall characteristics, the magnitude and direction of systemic risk, the transmission paths and impact so that they can formulate reasonable portfolio strategies, production, and operation strategies, and industry market regulation strategies, to reduce the adverse impact of cross-market risk contagion effectively, and provide suggestions for systemic risk management.
This paper's research logic is as follows: Firstly, this paper designs the contagion model from the perspective of methodological synthesis, fusing the multivariate heteroskedasticity model with the Diebold-Yilmaz method to design a new cross-market risk contagion measure index, and proposes a comprehensive cross-market risk contagion measure model based on the trend classification. Secondly, this paper uses five types of network identification methods to expand the analysis of the cross-market risk contagion network. Finally, this paper explores the trend evolution of the cross-market risk contagion phenomenon from the perspectives of the cross-market risk contagion index, contagion capacity, and dynamics of contagion factions. Meanwhile, this paper selects nine commodity futures markets to demonstrate the feasibility and effectiveness of the methodology, theory, and countermeasures.
The study finds that: (1)The cross-market risk contagion in commodity futures markets is interconnecting and time-varying. The paper constructs contagion networks for five categories of trends, and the network line color, thickness, and arrows point to the significance and directionality of the risk contagion. In addition, the upward, downward, and leveling trends of the comprehensive measure index and the dynamic evolutions of the faction category all reflect the time-varying characteristics of cross-market risk contagion from different perspectives. (2)The cross-market risk contagion in commodity futures markets presents market differentiation and impact asymmetry. Among them, the risk contagion and risk being contagion phenomenon are the most significant in metals futures, followed by chemical and agricultural futures. In addition, the risk being contagion of crude oil futures is significantly stronger than its risk contagion, and the risk contagion of copper futures and chemical futures within the downward trend is significantly stronger than the upward trend. (3)The cliques of commodity futures markets can be characterized as overall consistency and individual intersectionality. In the faction category, chemical, metal, and agricultural futures are classified into three different factions. In addition, the risk contagion and risk being contagion are significant for metals futures and chemical futures under different trends, which indicates that they belong to the key nodes and contagion centers of the whole market. (4)The cross-market risks in commodity futures markets are spread through nodes and cliques. The nodes contagion of commodity futures risk cross-market mainly revolves around silver, gold, copper, crude oil, asphalt, and natural rubber futures, which means that the above nodes have greater risk impact and impacted ability within the risk contagion network. At the same time, the results of the contagion faction category show that cross-market risk contagion emerge earlier and is more pronounced within the same faction.

Key words: cross-market risk contagion, commodity futures, comprehensive measurement, network analysis, trend evolution

摘要： 在关联金融市场价格波动的相互溢出影响下,由风险传染而形成的多市场波动趋同现象正呈现多爆发、强影响、广扩散与久持续等系列特征,为了结合上述特征合理测度并有效应对这类风险跨市场传染现象,文章以我国商品期货为具体实证对象,分别从综合测度、网络分析与趋势演化三个角度对风险跨市场传染现象进行了系统研究。具体研究中,先立足Diebold-Yilmaz公式构建了风险跨市场传染测度指数,并结合网络分析、派系类分及趋势演化等方法,提出了风险跨市场传染现象的综合测度模型。在此基础上,以我国具有代表性的九个商品期货为对象,进行了新模型、新方法和新过程下的系列实证,并得出结论:商品期货风险跨市场传染现象具备网络性与时变性,并呈现市场差异性和影响不对称性;商品期货风险跨市场传染现象派系类分具备总体一致性与个体交叉性;商品期货风险跨市场传染主要通过点传染及派系传染进行。

关键词: 风险跨市场传染, 商品期货, 综合测度, 网络分析, 趋势演化