Assessment and evolution analysis of the global wood pulp trade network resilience based on underload cascading failure

This study aims to assess the resilience of the global pulp trade network by applying complex network modeling methods. A directed weighted network model was constructed based on UN commodity trade data from 2002 to 2023, and a dynamic resilience evaluation framework based on an underload cascading failure model was proposed, systematically revealing both static and dynamic resilience characteristics during the network's spatiotemporal evolution. Findings show that the global pulp trade network exhibits significant spatiotemporal evolution: while the number of nodes has decreased, the number of edges continues to increase, reinforcing the dominance of core countries such as China and Brazil. The multipolar trend has led to the formation of five major communities. Although total trade volume remains stable - reflecting the industrial necessity of pulp - different external shocks have distinct impacts: the financial crisis triggered delayed effects through market adjustments, whereas the outbreak of the pandemic initiated immediate institutional interventions. Static resilience analysis reveals significantly improved network transmission efficiency, with high-trade-volume countries becoming less clustered. Assortativity patterns indicate continued dependency of smaller countries on central nodes, while power-law fitting suggests weakened hierarchy and increased robustness against targeted attacks due to diversified structures. Notably, China's role as a hub has grown rapidly, with its incoming strength increasing by 5.14 times and weighted betweenness centrality rising by 3.54 times. Cascading failure simulation further demonstrates that the failure of key nodes, such as Brazil, may trigger large-scale chain reactions, directly causing trade reductions in 32 countries. Under targeted attack scenarios, network performance loss far exceeds random disruptions, confirming the structural risks associated with supply concentration. Dynamic resilience optimization shows that reducing failure thresholds can enhance global efficiency redundancy by 5.57 times and increase the strength of the largest connected subgraph by 3.59 times, offering quantitative support for resilience-enhancing strategies. Moreover, external shocks exhibit two distinct modes of impact on network resilience: the financial crisis induced gradual adaptation via market mechanisms, while the pandemic triggered emergency response mechanisms, highlighting the critical role of institutional intervention in supply chain stability and revealing fundamental differences between market-driven and policy-driven resilience pathways. The innovations of this research lie in the first application of an underload cascading failure model to pulp trade network resilience assessment, breaking through the limitations of traditional static analyses. It also constructs a dynamic resilience evaluation framework integrated with this model, enabling end-to-end analysis from failure simulation to resilience quantification. By transforming time-series responses into quantifiable metrics through cumulative integration, this work addresses the lack of standardized resilience indicators in forest product trade networks. The findings provide scientific support for policies related to supply chain security, sustainable forest management, and cleaner production practices.