Dynamic risk spillover networks and causal paths of steel product prices in China

Abstract

[Objective] Exploring the dynamic risk spillover characteristics and causal paths between steel products in China is of practical significance for understanding the market dynamics of the steel industry and formulating effective risk management strategies. [Methods] This study mapped the dynamic risk spillover networks of steel product prices in China from March 2018 to March 2024, analyzed the dynamic characteristics of network structures and steel products by using the quadratic assignment procedure (QAP) correlation and node metrics, and revealed the causal paths of risk spillover by using the streaming algorithm. [Results] (1) There were obvious time-series correlation characteristics between the dynamic network structures, the Diebold and Yilmaz spillover network had higher time-series correlation than the Granger causal network, and the correlation of the dynamic risk spillover network in the period of April 2020 to April 2023 was relatively stable. (2) Cold rolled was the main risk spillover product, with a substantial increase in spillover strength during COVID-19. Shipbuilding plate and seamless pipe were the main risk spill-in products. Connectivity between products in the network was very strong, but risk propagation was not fast. (3) Long-term and multi-step causal paths dominated the market, while short-term and direct causal paths accounted for a relatively low proportion. Cold rolling → seamless pipe, cold rolling → seamless pipe → cold rolling non-oriented silicon steel, and cold rolling → rebar → seamless pipe → cold rolling non-oriented silicon steel are important causal paths. [Conclusion] To cope with the risk spillover between steel products, we should strengthen the tracking and monitoring of key steel products; develop a long-term risk management strategy; and enhance market transparency and industrial chain coordination to reduce market volatility caused by risk spillover.

Keywords

• steel products|risk spillover network|causal paths|streaming algorithm|dynamic characteristics